Yhd-33 3620 The Jorc Code - 2.6.6

Fourth International Mining
Geology Conference
14 - 17 May 2000
Coolum, Queensland
The Australasian Institute of Mining and Metallurgy

Publication Series No 3/2000
Published by
THE AUSTRALASIAN INSTITUTE OF MINING AND METALLURGY
Level 3, 15 - 31 Pelham Street, Carlton Victoria 3053 Australia
The Australasian Institute of Mining and Metallurgy 2000
The Institute is not responsible as a body for the facts and opinions advanced in any of its public ations.
ISBN 1 875776 75 3
Desktop published by:

Penelope Griffiths and Angie Spry for The Australasian Institute of Mining and Metallurgy
Printed by:
New Generation Print and Copy
12 Barkly Street
Brunswick East VIC 3057
Organising Committee
Andrew Vigar (Chairman)
Robin Vigar
Mark Berry
Ron Cunneen
Roussos Dimitrakopoulos
Ian Kelso
Foy Leckie
Kevin Lines
Brice Mutton
Graham Pope
Andrew Scott
Andrew Waltho
AusIMM Central Services

Miriam Way (Events Manager)
Pauline Weaver (Events Assistant)
Penelope Griffiths (Publications Manager)
Angie Spry (Publications Assistant)
Foreword
We all appreciate the current difficult conditions within the global mining industry and the great
efforts being made within Australia, in particular, to compete. I believe we can be justifiably proud
of the high standards of the professionals within our industry. Conferences like the Fourth
International Mining Geology Conference, organised by the bodies that represent the professionals
within the industry, are one of the keys to continuing professional education and maintaining this
edge.
This Conference follows on from successful events staged at Launceston in 1997, Kalgoorlie in
1993 and Mt Isa, Queensland in 1990. It represents one of the few events in Australia aimed
directly at mining geologists and has been designed to review the advances in knowledge since the
first conference ten years ago. A joint committee drawn from members of The Australasian
Institute of Mining and Metallurgy and the Australian Institute of Geoscientists in Southern
Queensland was formed to organise this conference.
We believe that this is the premier forum for mine geologists from throughout Australia and
overseas to meet and exchange information and ideas on their operations and the latest
developments and innovations in key areas that directly impact on their day-to-day work. We hope
that the proceedings, together with the hands-on discussions with the presenters and the forum
session, will provide new insights to challenge the way we all do business, both now and in the
future.
The organisation of a major event such as the Fourth International Mining Geology Conference
requires the support of a large group of people. I would like to take this opportunity to thank
everyone who has helped on the organising committee, for without their dedication and enthusiasm
the conference could not have taken place.
On behalf of the organising committee, I would like to thank the authors of all technical papers,
and their companies, for their excellent and thought provoking contributions. Presentations are the
core of a conference and the quality of the contributions to this conference is the equal of in the
past.
Finally I would like to acknowledge the contribution provided by industry organisations and
companies throughout Australia and overseas, for sponsorship, a stimulating trade exhibition and
for supporting delegates to attend this conference.
At the dawn of the new Millennium, the mining industry continues to maintain its critical
importance to the health and well-being of our planet and all its inhabitants. Mine geologists
continue to sit at the forefront of innovation and improvement in the industry and add to their
community as a whole. We hope that this event, and these proceedings, helps in your endeavours.
Andrew J Vigar
Conference Committee Chairman
Contents
Role of the Mine Geologist
The Mine Geologist in a Business Perspective
D Head
Dynamic Links Between Geology and the Mining Process
T C McCuaig, J Vann and

C Seymour
Sampling and Databases

Systems and Protocols of Geological Information Management at Telfer
Gold Mine, Western Australia
R D Carlson, G R Howard
and G Back
21
Technical Data Management at Porgera
A W Burgess, J R Foley,
R J Henham and
A G Shellshear
27
Sampling Practice at the Vera Nancy Gold Mine
D Sims
35
Leachwell Versus Fire Assay: Comments from Assay Quality Assurance

at Tarmoola Gold Mine
D J Elder
43
Controls on High-Grade Gold Distribution at Vera Nancy Mine
D Sims
55
Geological Modelling and Grade Control in a Narrow Vein, High-Grade

Gold Mine
D Sims
65
Mine Geology Practices at the Sunrise Open Pit
E Haren and P Williams
77
The Spatial Distribution of Grade
J R Vearncombe and
S Vearncombe
87
The Bunyip Lateritic Nickel-Cobalt Deposit, Cawse Nickel-Cobalt

Operations, Kalgoorlie, Western Australia
A Bywater and S M Denn
95
Cadia Hill Gold Mine One Year Down the Track
C F Moorhead and
Cadia Geology Team
105
The Geita and Kukuluma Mineralised Trends, Lake Victoria Goldfield,

Tanzania Orebody Characteristics and Project Planning
D Bansah, R Chase,
A Davidson, H Michael,
M Skead and H Stuart
115
Kunwarara Magnesite Deposit
S Wilcock
129
Mine Geology
Current Geological Understanding of Telfer Gold Mine
G R Howard, T Hansen,
C Moore, P J Moffitt,
R J Inglis, R D Carlson,
I Kirchner, D Coupland,
S Leary and A Tomsett
135
Geology and Structure of the Morning Star Mine, Mt Magnet, WA
R Mason, N Archibald,
D Holden, T Blyth,
S Huffadine, R Bradey,
A Jones and P Androvich
143
The 1999 JORC Code What Does it Mean for Todays Mining
Geologist?
P R Stephenson
157
Resource Evaluation of Nuggety Slate-Hosted Gold-Quartz Reefs
S C Dominy, B W Cuffley,
G F Johansen, A E Annels
and I M Platten
169
Multiple Indicator Kriging Is it Suited to My Deposit?
J Vann, D Guibal and

M Harley
187
Mining Bench Height Evaluation for the Wallaby Resource

A Conditional Simulation Case Study
I M Glacken, M Nopp and

M Titley
195
Ore Definition at the Henty Gold Mine Tasmania
N Schofield
207
Computer-Based Resource Estimation in Accordance with the

1999 JORC Code
J Duke and P Hanna
215
Mining Grade Control Past, Present and Future
W J Shaw
223
Evolution of Grade Control at KCGM
V M OBrien and T Cutts
229
The Use of Magnesium Oxide and Iron to Predict Host Units in Grade
Control and Exploration at Bulong Nickel Operation
W J Bollenhagen
239
Ore Characterisation for Mine to Mill Fragmentation
A Scott, J Segui and

S Kanchibotla
247
Upgrade Ability and Geology of Cawse Nickel Ore
S M Denn, C G Ferguson
and S L Makin
255
The Interaction Between Geology, Mining and Metallurgy at Stawell

Gold Mines Pty Ltd
D Fredericksen
263
Resource and Reserves
Grade Control
Mine to Mill
Geotechnical and Structural

The Regional Controls Exerted on Rock Discontinuities at the Iron Duke
Deposit, South Australia and Extrapolation Throughout the Rock Mass
C N Winsor
271
Pasminco Century Mine Open Pit Slope Design A Geotechnical

Perspective
A J Dutton
283
Modelling Bulk Density The Importance of Getting it Right
I T Lipton
291
The Expanding Role of Mine Geophysics
P K Fullagar
301
Detailed Orebody Mapping Using Borehole Radar
A Wellington, G Turner,
I Mason and J Hargreaves
315
Application of the SIROLOG Downhole Geophysical Tool at Callide

Coalfields East Central Queensland
W Nichols
321
Microseismic Monitoring of Shear Zones and Related Seismic Activity at

Broken Hill
A J Morley, J M Murray
and G C Reed
331
Finding More Ore, Further From the Drill Hole, With DHMMR
J Bishop and R Lewis
337
Nuclear Borehole Logging Techniques Developed by CSIRO Exploration

and Mining for the Metalliferous Mining Industry
M Borsaru and
J Charbucinski
347
New Technologies
Role of the Mine Geologist

D Head
Dynamic Links Between Geology and the

Mining Process
T C McCuaig, J Vann and

C Seymour

D Head1
ABSTRACT
The role of the mine geologist in a business perspective is self-evident to
some, but to others it may be a new concept. All mine geologists need to
know exactly how they add value to the business if they are to be
successful. Alternatively, if they do not add value, then they will become
redundant.
This paper seeks to understand some of the contributing problems that
hinder the mine geologist from operating in a business perspective. In
doing so, a new context (for some) is proposed to aid the mine geologist
and the operation gain greater value from geology. The proposal suggests
how to ensure that the activities of our geologists are intrinsically linked
to the business process that we are involved in.
The mine geologist can and must play a pivotal role in the success of
an operation. To do this, they must understand some principles of
business such as:
1.
process,
2.
customer/supplier relationships, and
3.
organisational structure.
Our geologists must know the business objectives, understand the
product that the customer needs and use their geological skills to deliver
the product safely, on time and to specification.
As geologists, we can be instrumental in ensuring that the operational
environment is structured in such a way that maximum value is realised,
not only from the mine geologists but the entire site team.
In doing so, mine geologists will realise their full potential in
delivering the outcomes the business demand, and therefore be
recognised as strong contributors to business success.
INTRODUCTION
Mine geology is an intrinsic component of the mining process.
There are few operations that cannot benefit from the input of
carefully considered and applied geology. From the relatively
simple massive coal seams of Victorias Latrobe Valley, to some
of the complex and narrow gold orebodies of the Yilgarn of
Western Australia, each operation will benefit from focussed
mine geology, driven by a business perspective.
The notion of a mine geologist in a business perspective seems
intuitive, however, stopping to clarify what is meant by the
concept, one realises how difficult it is to define. It is therefore a
worthwhile exercise to consider some of the problems that the
mine geologist is often faced with in trying to operate in a
business perspective.
Problems such as:
having to operate in an environment that has not clearly
defined the process and stated the outcomes that the
operation expects from its geologists;
poor communication between departments for effective
information flow; and
suboptimal site departmental structure.
These are some of the key problems facing mine geologists.
The author wishes to state that the ideas expressed in this
paper have developed over many years. They incorporate
concepts based primarily on personal experience that has been
developed as a result of in-house training at various operations
and many hours of discussion with colleagues. They therefore
reflect the authors personal opinion.
1.
MAusIMM, General Manager Resource Development, Normandy

Yandal, 100 Hutt Street, Adelaide SA 5000.
4th International Mining Geology Conference
On reflection, the concept of a business perspective for a mine

geologist assumes that mine geology is part of a process. A
process is a system that combines a range of inputs, and converts
them into outputs or outcomes. For the mine geologist, the
process refers to that of mining. Discovering, estimating and
adding value, so that the metal, mineral or compound can be
mined and processed to supply a product to a customer that can
be assumed to generate wealth.
In many ways, this process is nothing more than a glorified
factory. We take raw materials, process them and market a
product.
Now, I appreciate that some will take offence at this analogy,
however, it is a useful one to consider, because it allows us to
visualise the process in a different perspective and enables us to
define the mine geologists role in the process.
The issues to be addressed in this paper are:
to understand the process;
to determine who the customers and suppliers are; and
to know how to structure our relationships and communicate
effectively within them.
We can then ensure that we maximise our geological skills in
order to deliver the product our customer demands. That is, to
operate in a business perspective.
To begin, try to visualise the activities of the mine geologist in
terms of this context. We must ask ourselves, What are the
critical relationships that are needed for the mine geologist to be
effective in the process? Who are the customers? What do
they want and when?
We will also need to review how the organisational structure of
a site can either support or hinder the success of the mine
geologist in fulfilling the expectations of the business.
Finally, this paper looks at two areas where an improved
understanding of the process, with better customer focus and
more effective relationships can assist the mine geologist in their
role. Firstly, in grade control and then, how an integrated process
for estimating Resources and Reserves is established across a
business.
Essentially, this paper deals with change. Recognising that
there are alternative and often-improved ways of operating is
important. How to get there is far more difficult. Managing the
change process will be more effective if it is driven from within
and I hope that as mine geologists we can be part of that change
process.
THE PROCESS
Most mining operations have traditional departments based on
professional skills. These departments (engineering, geology,
accounting, and metallurgy) exist to support the process and the
business and they do, to a certain degree. Generally, they fulfill
their part of the process as a professional silo where each silo
does its best to support the business. Of course, the underlying
process does not recognise arbitrary silos because the process
requires a seamless flow of knowledge, materials and
information through time to deliver a product. The process
expects co-operation and teamwork, but ironically, sometimes
the professional silo inadvertently hinders the flow of the
process. Poor structure, poor communication or both can
compound the hindering effect.
Coolum, Qld, 14 - 17 May 2000
D HEAD
In a mine, many people visualise the process of mining

differently and this is an important point. If we all visualise a
different process for the one site, how will we work as a team
and maintain a business perspective?
Each department manager is likely to have his or her own
vision of the process. The following section examines three
generalised models based on personal experience.
Model 1 Centralist
Model one is what is referred to as the centralist model. The
primary understanding of this model and those who work by it, is
that the department in question sees itself as the centre of the
process and that all other departments provide inputs to it. Figure
1 illustrates this.
In this model, Department A is larger than all the others and

is clearly the most important. The department seems to be more
important than the customer is!
Model 2 Linear
In model two, departments are arranged linearly (Figure 2). The
departments accept that there is some flow to the process, but
resist mixing so as to ensure the longevity and security of each
department (silo). Based on personal experience, this seems to be
the favoured model by most Australian mining operations as it
appears to be most common.
A
C
D
Product
FIG 2 - Schematic representation of the Linear Model.
Department
A
Model 3 Functional Teams
Customer/Product
FIG 1 - Schematic representation of the Centralist Model.
The preferred model (three) suggests that functional teams and

not departments carry out the steps or stages of the process.
Functional teams will contain a blend of skilled professionals,
organised so as to maximise their involvement in the process to
ensure that the product is delivered on specification and meets
the quality and timeliness requirements. It is possible that the
team may become a silo in this model, however, this is less
likely because the emphasis is on the team rather than a
professional department. The issue with this model (for the
siloists) is that the professional identity can be lost. Individual
identities as part of a department may be lost, but a new identity
as part of an effective team should be gained.
Figure 3 illustrates how it may look.
FIG 3 - Schematic representation of how the Functional Team approach may look.
THE MINE GEOLOGIST IN A BUSINESS PERSPECTIVE
If we accept that this is how the underlying process is working,

then as mine geologists we can begin to understand where we
can fit into the process. What information do we need in order to
deliver a product that our customer requires? Who are our
customers? And what relationships do we require in order to
perform in a business perspective?
CUSTOMER/SUPPLIER RELATIONSHIPS
Understanding the concept of a customer/supplier relationship is
critical in the success of a business. It seems that this is often an
area that is poorly managed and understood, particularly by mine
geologists. There is often a perception amongst mine geologists
that operations carry out geological activities simply because
they must. An opposing view is to ask, why do operations carry
out geological activities? The answer is that, someone wants to
be supplied with geological information. Someone has valued the
information and deemed that it shall be part of the process.
In the event that geological information is not sought or
required in the process and yet is still collected, then the
operation may well ask, It may be good geology, but is it good
business? In some instances, there will be operations that have
geologists but do not know why they carry out geological
activities at all.
These two examples are difficult sites to operate in a business
perspective. This is a result of the team not establishing the
customer/supplier relationships.
A group of geologists, engineers and metallurgists have been
assembled together, to mine and process an orebody but each
group operates in isolation and does not understand the operation
of, nor necessity for the other groups.
Leadership is also a critical issue here but that topic is outside
the scope of this paper.
Defining the customer/supplier relationships is a key step in
operating in a business perspective. By understanding the overall
process and therefore appreciating the critical areas that mine
geologists add value, the mine geologists can begin to understand
whom their customers and suppliers are. Armed with this
knowledge, it is possible to assess exactly whom the geologist
needs to communicate with, why and when. Each operation
needs to carry out this assessment, as there is no strict blue print
for specific customer/supplier relationships. Once the
relationship is defined, the process of setting the deliverables and
hence accountabilities can begin.
Establishing these relationships is important and ensures that
the activities of the mine geologist within the process are helping
support the business. Having established that the relationship is
important, we must then communicate through effective
organisational structure. Relationships require effective
communication and organisational structure provides the
framework or rules for this communication to occur.
department, the resource department and so on. As each

department divides itself into smaller and smaller
subdepartments, so the difficulty of operating in a business
perspective increases.
Overriding the effect of departmentalisation, is the process.
No matter how hard we try to departmentalise ourselves into
secure silos, the process (in order to succeed) demands
interaction, communication and teamwork.
The corollary is failure.
For example, the strategic planning engineer needs to know
about the tonnes and grades to be scheduled in year five of the
life of mine plan. He or she needs to know the value of the
estimate, the confidence in the estimate and the risk in the
estimate. It would make sense then, that an organisational
structure that enables the effective communication between the
key members of the process at any given point in time would be
advantageous. We need a functional team and not a traditional
department.
The following organisational structure attempts to solve the
problem of professional silos by developing functional teams.
The teams are staffed by personnel who have the requisite skills
to add value at a particular point in the process.
The weighting of particular disciplines within the functional
teams changes throughout the process.
Operational leverage is gained by individuals realising how
their particular contribution (specialty) assists their customer and
hence appreciates how they support the overall process.
The teams shown in Figure 4 could be assembled to manage
the process, (skills, weighting, activities are examples only).
Resource team
Activities
Skills required and weighting
Safety, Strategic Analysis,

Strategic Planning, Contract
Management, Project Evaluation,
Mine Design
Geology
60
Engineering
Metallurgy
Finance
Computing
25
Planning team
Activities
Safety, Strategic Analysis,

Strategic Planning, Contract
Management, Project Evaluation,
Mine Design
Engineering
30
Financial
20
Geological
20
Metallurgical
20
Environment
10
Production team
ORGANISATIONAL STRUCTURE
Having now described the process that the mine geologists are
involved in and established the customer/supplier relationships
required to effectively deliver the product, the organisational
structure can now be reviewed.
As mentioned earlier, the most common organisational
structure of the mine is one of professional silos. That is a
geology department, an engineering department, a metallurgy
department and so on. The result of this type of structure can
often stifle the communications between suppliers and
customers.
Even within a department, we can see professional subsilos.
For example within the traditional geology department, we may
have; the exploration department, the open pit geology
Activities
Safety, Drill and Blast, Survey,

Grade Control, Processing,
Crushing and Grinding, Short
Term Planning and Design,
Contract Management,
Geotechnical, Mine Design, Cost
Control, Tailings Disposal,
Personnel Management
Engineering
30
Metallurgy
30
Geology
10
Human
Resources
Environment
Financial
10
Survey
FIG 4 - Organisational structure for Functional Teams.
D HEAD
The examples show that the professional silos or traditional

departments are not necessary. The emphasis must be on
organising a team that is balanced with the appropriate skills and
arranged so as to offer the best probability of delivering the
business outcomes.
TWO EXAMPLES
Two traditional geological activities that are in reality
multi-disciplinary processes (grade control and Ore Reserves)
can be viewed in the context of the previous discussions. What is
required of these processes? Who needs the information? What
quality and format is expected? What inputs are required?
Grade control process

The mine geologist is often chosen to manage grade control. So,
what is required to perform the function well and what
relationships, within the process are required? The following
outline simplifies the process but will illustrate the issues being
addressed.
1.
Planning engineers need to produce a plan and schedule

detailing when an area will be available for grade control.
2.
A grade control design plan is required to provide

information to the surveyors, detailing when and where
holes are to be drilled.
3.
The assay laboratory needs to be aware of how many

samples will be arriving and when.
4.
Drilling programs and drilling contractors need to be

co-ordinated.
5.
Access to the area needs to be approved by the production

engineer.
6.
Faces need to be mapped and data collected by a geologist.
7.
Drill holes need to be logged by a geologist.
8.
Data needs to be entered in to a database.
9.
Quality control checks are required on an ongoing basis.
10. Data needs to be interpreted by the geologists and

engineers (and possibly metallurgists if appropriate).
11. An estimate of the tonnes and grade needs to be provided to
the planning and production engineers.
The number of relationships and interactions required to
successfully perform the grade control process has been
simplified here. However the example illustrates the complexity
and the need for the mine geologist to develop and maintain
strong internal and external relationships so that the product (the
tonnes and grade estimate) can be safely provided to the
customer on time and to specification.
grade and will be based on geotechnical data, mineralogical data,

lithological data, grade and metallurgical data, estimation
confidence, bulk density and structural data. At this stage the
Mineral Inventory Model does not presuppose mining cut-off
grade, marginal cut-off grade, minimum mining widths, dilution,
or any other criteria or assumptions that must be made in order to
estimate Resources and Reserves. All of these assumptions are
made at the planning stage and hence the Mineral Inventory
Model cannot be used for direct external reporting of Resources
and Reserves.
The Mineral Inventory Model is the outcome of detailed
exploration efforts in the first instance, but for any operating
mine, it is the outcome of on-going mine geological project
work, production data, drilling and detailed reconciliation
processes. It then becomes a significant input into the strategic
planning process.
The Mineral Inventory Model can be used as the basis for
many mining scenarios carried out by a team of engineers,
financiers, metallurgists, geologists, environmental, legal and
environmental personnel from which one is chosen (a corporate
decision) to become the Life of Mine plan. As a result of this
process, the Ore Reserve and Mineral Resources are delivered as
outcomes (based on a number of corporate and operational
assumptions) for external publication and internal design and
scheduling (see Figure 5). Those areas of the Mineral Inventory
Model that do not have sufficient confidence to be regarded as
Ore can then be targeted for follow up exploration.
In this process, the mine geologists are accountable for
delivering the Mineral Inventory Model at a pre-determined date
each year. Knowing this, and having the detailed parameters
required (as outlined by the customer) within the model, we can
then set clear, reasonable and achievable individual
accountabilities for the outcome required. This process then
drives improvement into areas that are definable and which
demonstrably support the business process.
Reflecting on the understandings of the process as discussed
above, there is nothing uniquely geological about the grade
control or Ore Reserve activity. They are examples of activities
undertaken by functional teams rather than professional silos so
as to produce an outcome. However, because of history, the
professional silo mentality often blocks the interactions and
relationships necessary to achieve the result. The blocking is
ironic as it only serves to perpetuate and justify the professional
silo. That is, the engineering department may say the geologists
are late again with the dig plan , and so it goes on.
As professionals, we need to understand the process,
understand the customer/supplier relationships and then ensure
that a structure is put in place to support the process and assist in
effective communication.
CONCLUSION
Resource/Reserve process
The Resource/Reserve process has long been considered the
domain of the geologist. However, as the industry becomes more
sophisticated, more and more companies have realised that
Resources and Reserves are the outcomes of a highly disciplined
process resulting from the interaction of specialist teams of
people, each with their own skills, knowledge and data. When
combined into a coherent process, they deliver a product that we
term Mineral Resources and Ore Reserves (as defined by the
JORC code).
It is suggested here that the first step in estimating Ore
Reserves and Mineral Resources be for the geologists to prepare
what may be termed the Mineral Inventory Model. The Mineral
Inventory Model is a precursor to Resource and Reserve
estimation. At a given point in time, the Model will represent the
geologists best estimate on the spatial distribution of tonnes and
I started out by stating that all mines will benefit from

appropriately applied mining geology.
This will only occur when the mining geology is driven in a
business perspective.
To achieve the business objectives, the issue is not so much our
individual skills in a specific area of geology (although this is
important) but rather, how we use our skills to deliver the
outcomes the business demands.
We must understand the process that we are involved in, we
must understand who our customers are and what they want and
we must then develop relationships that facilitate the
communication of the required information.
If we can do this, then as mine geologists, we will be operating
in a business perspective and we will be recognised as positive
contributors to business success.
THE MINE GEOLOGIST IN A BUSINESS PERSPECTIVE
The Mineral Inventory Model

Geology
Data
Knowledge
The Mineral
Inventory Model
- Classified, Measured,
Indicated, Inferred
Reconciliation
Resources
Strategic
Planning
Reserves
Filter on remaining
Mineral Inventory
Matrix of
Potential
Reserves
YES/NO
YES
Treasury Analysis
Of Matrix of
Potential Reserves
Production
Reserve
One option
is selected for
Life of Mine Plan
NO
Public
Resource
incl.
Reserves
To
Management
for Review
Mineral
Inventory
(Not Public)
FIG 5 - Schematic outline of the Resource/Reserve process.
ACKNOWLEDGEMENTS
I would like to thank the management of Normandy Yandal
Limited for permission to publish this paper.
I wish to acknowledge the assistance and motivation provided
by D Ryan and C Gee for the numerous drafts reviewed. Thank
you also to N Phillips and J Hergt for their comments.
To those of you who I have had the pleasure to work with,
thank you for the feedback and experience that has assisted in the
development of this paper.
There are two mining engineers that I must thank for their
continued support and belief over many years. You will know
who you are. Thank you.
REFERENCES
The Joint Ore Reserves Committee, 1999. Australian Code for Reporting
of Mineral Resources and Ore Reserves (The JORC Code).
Coolum, Qld, 14 - 17 Mayl 2000
Dynamic Links Between Geology and the Mining Process

T C McCuaig1, J Vann2 and C Seymour2
ABSTRACT
3.
Geology is generally treated as a front end step in mine development.

After infill drilling, the majority of resource estimation, mine-planning,
development, mineral processing strategies and operational procedures
are planned and implemented with little geological input, particularly in
open pit operations. However, significant benefits in overall costs, cash
flow, and technical risk reduction can be realised when sound geology is
utilised more rigorously throughout the mining process.
This under-utilisation of geology in the mining process has arisen due
to a combination of factors:
1.
A lack of interest in, or knowledge of the full mining process by

geologists. Few geologists fully understand where their inputs
make critical impacts on downstream users.
2.
As a result of this, geologists generally do not focus their inputs

towards areas of greatest value added, ie addressing issues critical
for other members of the mining team.
3.
Most importantly, few geologists effectively communicate their

results to the downstream users of their geology models. The
results are that the potential benefits of geology, and the
appropriate status of geology in the mining industry, are not fully
realised.
It is fundamental to realise exactly what geologists do. Geologists
constrain geological processes and geometry. All geoscientific tools at
the geologists disposal are geared towards one of these goals. An
emphasis on geological processes is present throughout the exploration
process, whereas in the resource estimation and mine development
process an emphasis on geometry is required.
The mining process is a web of disciplines that are attempting to optimise
development of a resource to provide value to shareholders. Yet few
people within these disciplines have a full appreciation of this web, and
where sound geology inputs can add value through various feedback
loops. Geology inputs do not end at the resource estimation stage, nor are
they relegated to grade control implementation. If geologists have an
appreciation of the mining process and close liaison with the other
disciplines such as geostatistics, geotechnical engineering, hydrogeology,
and mining engineering, they can significantly benefit project costs,
timelines, and constrain technical risk by addressing their issues at the
earliest possible stage. Furthermore, the mining process is not a single
pass timeline, but an iterative process where the exploration, resource
definition, mine planning and mining loop is continually repeated over
the life of a project. Therefore, the potential inputs of geology are critical
throughout the mining process.
Effective communication of the impacts of geology inputs on
downstream users is often lacking on mining projects. The most effective
way to communicate geological results is through where it impacts most
project risk. Focussing all geology inputs and communication in terms
of potential material risk allows the geologist to be more effective in
contributing value to the mining process, and allows the downstream
users of geology inputs to better constrain the risks associated with their
aspects of the mining project.
Through a program of geologists:
1.
taking an interest in the entire mining process and identifying

critical areas for geology inputs to downstream users (the
geologists clients);
2.
focussing geology inputs to address the potentially material issues

on mining projects; and
1.
Senior Associate, SRK Consulting, PO Box 943, West Perth WA

6872.
2.
MAusIMM, Technical Director, SRK Consulting, PO Box 943, West

Perth WA 6872.
communicating their inputs to the clients in terms of technical

risk, geology will be much better integrated into the mining
process.
As a result, the potential benefits of geology throughout the mining
process will be realised, and the status of geologists within the mining
process will be elevated to a more pivotal position.
INTRODUCTION UNDER-UTILISATION OF
GEOLOGY IN THE MINING PROCESS
Geology is generally treated as a front-end step in the mine
development (Figure 1). After infill drilling, the majority of
resource estimation, mine-planning, development, mineral
processing strategies and operational procedures are planned and
implemented with little geological input, particularly in open pit
operations. Although geologists are employed throughout the
mining operation, they are often fully occupied with grade
control and production issues with little time allocated to
assessing potential geology inputs to mining operations. This is
particularly true with the recent trend towards reducing
geological staff in the interests of reducing operational costs.
However, significant benefits in overall costs, cash flow, and
technical risk reduction can be realised when sound geology is
utilised more rigorously throughout the mining process.
For example, rarely does a project finish infill drilling without
requiring significant additional work to be done at the
prefeasibility, feasibility or mining stages. While fully addressing
all possible issues on projects is not a practically achievable goal,
it is the authors contention that such additional work, and
resultant delays or fatal (material) flaws in project development,
can largely be avoided. This can be achieved if geologists and
management possess an acute awareness of the integration of
geology in the mining process, and by directing data collection
and synthesis at early project stages towards end-user
requirements, eg resource estimation (domain definition issues;
drill orientation, density and data quality issues impacting on
resource classification), geotechnical (geometry and structure
issues early characterisation of structures and rock mass),
metallurgical (recovery issues early petrography of ore, assay
for potential penalty elements), hydrogeological (water problems
installation of piezometers in exploration/infill drill holes), to
name but a few.
This under-utilisation of geology in the mining process is
rarely due to incompetence, but rather to mutual ignorance and
lack of communication between various disciplines or stages in
the mining process stream. Geologists often do not fully
appreciate the requirements of their downstream users, and the
downstream users do not fully appreciate what geologists can do
for them.
Geologists who invest time and effort in understanding the
potential downstream application to their inputs, and with the
ability to communicate their potential contributions to mine
planners and management will not only be more effective, but
will assist in increasing the awareness, status and benefits of
geology in the mining industry.
Similarly, resource estimation specialists, mining engineers,
mine planners and metallurgists who seek to capitalise on the
geologists potential inputs will be better able to evaluate,
mitigate and manage project risk in their respective technical
fields.
T C McCUAIG, J VANN and C SEYMOUR
GEOLOGY
ONE WAY TRAFFIC
NO FEEDBACK
?
Resource
ONE WAY TRAFFIC
NO FEEDBACK
Reserve
ONE WAY TRAFFIC
NO FEEDBACK
Mine Planning
TWO WAY TRAFFIC
Startup
KKG/bw2006/28March2000
FIG 1 - Traditional view of the role of geology in the mining process.
10
DYNAMIC LINKS BETWEEN GEOLOGY AND THE MINING PROCESS
Geological processes and genesis of orebodies
THE GEOLOGISTS TOOLKIT WHERE IS IT

THAT GEOLOGISTS ADD VALUE TO THE
MINING PROCESS?
Geological processes that are investigated generally address

controls on mineralisation, but may also involve other issues
such as development of weathering profiles, or decay of
mineralised waste at surface (acid mine drainage). In these
studies, the geologists continually formulate hypotheses, as to the
location of mineralisation for example, and determine what types
of data they will require to test the hypothesis or better predict
where mineralisation may occur. We emphasise the importance
of focussing on process versus models in these investigations.
Genetic models are essential in that they form a framework in
which we can visualise and understand geological processes.
However, models tend to take on a life of their own and tend to
drive data collection in self-fulfilling directions (McCuaig and
Hronsky, submitted). It is an appreciation of geological processes
that advance our understanding, and result in the refinement of
geological models or the generation of new geological models.
The effective geologist must be a model-builder, not a model
pusher. The search for evidence that mineralisation processes
have occurred based on our understanding of ore genesis
dominates in the exploration phase (Figure 3).
What is it that geologists can add to the mining process? What

are the real issues that they address? Todays geologists have a
plethora of geoscientific investigation techniques and datasets at
their disposal (Figure 2). The challenge is to recognise what lines
of investigation will add the most value, and thus where to direct
data collection effort, interpretation effort, and dollars.
What geologists do geometry vs genesis

In the opinion of the authors, geological investigations using the
myriad tools listed in Figure 2 are all ultimately aimed at
constraining either geological processes (eg genesis of ore) or
geometry. For example, regional aeromagnetic data is used to
determine the geometry of lithology and structure, or evidence
that alteration processes have taken place (magnetite-producing
versus magnetite-destructive), in order to make predictions as to
the potential location of mineralisation (ore genesis). Structural
analysis of orebodies is aimed at determining the geometry of
ore, whereas kinematic analysis of oriented core is focussed on
understanding structural processes that controlled the genesis of
ore. Analytical geochemical studies on mineral deposits (SEM
analyses of minerals, stable isotopes, fluid inclusions, radiogenic
isotopes) are aimed at constraining geological processes that
have operated in a region, and those which controlled the genesis
of mineralisation. Gold assays are generally aimed at
determining the presence or geometry of mineralisation during
exploration phases, in addition to their use to quantify resources
at a later stage.
Determining orebody geometry

By the infill drilling stage, the empirical determination of
lithology, structure and mineralisation distribution, albeit
constrained by our understanding of processes, has begun to
dominate the exploration psychology (Figure 3). During this
exploration phase the geologists draws upon their toolkit to
constrain these geometrical issues, as they are the critical issues
from this point through to the mine development stage. While
genetic models and understanding of mineralisation processes
GEOMETRY MODEL
BASE
DATA
SOURCES
Lithology
Distribution
Grade
Distribution
GEOPHYSICS
magnetics
gravity
seismic
geostats
metals
lithogeochem
ALTERATION
GEOCHEMISTRY
MAPPING
STRUCTURE
petrography
stable
isotopes
DRILLING
Structure
Distribution
radiogenic
isotopes
fluid/
solute
sources
wholerock
assays
geothermometry
mineral
and
stability
and composition geobarometry
fluid
inclusions
cross-cutting
relationships
kinematic
analysis
geochronology
P-T-X
PARAGENESIS
Alteration
processes/
petrogenesis
Controls on Ore
(metal deposition)
GENETIC MODEL
(process)
FIG 2 - Examples of the multiple geoscience subdisciplines and analytical techniques that modern geologists have at their disposal.
11
high
RELATIVE CONTRIBUTION/VALUE
OF GEOLOGICAL INPUTS
Major Financial
Consequences
Profit Margin
Maximised
Opportunity Costs
RISK ASSESSMENT
UNDERSTANDING OF 3D GEOMETRY
UNDERSTANDING OF GEOLOGICAL PROCESSES
Geologist must
begin considering
end user inputs
very
low
Grass Roots
Exploration
Advanced
Development
Prospect
Drilling
(realisation
that a significant
resource is present)
Prefeasibility
Bankable
Feasibility
(Financing
Due
Diligence)
Startup
Operations
PHASE OF MINING PROCESS

FIG 3 - Process, geometry and risk inputs through the exploration to operation phases of mine development.
are important to delineation of extensions to ore and near-mine to

far-field exploration efforts, they are not as critical to resource
estimation and optimisation. For example whether a deposit is
intrusion-related or not is usually immaterial to its successful
development in fact, full genetic understanding of mineral
deposits is rarely reached even after the resource has been fully
exploited. Nevertheless, an understanding of processes is still
important (eg understanding structural processes and kinematics
as an aid to interpreting correct geometry of mineralisation).
Risk analysis and geology

A third issue that geologists address is technical risk (Figure 3).
Although often undertaken in an ad hoc manner, such assessment
by geologists determines the risk (uncertainty x consequence)
associated with both the empirical determinations of geometry
and the understanding of genetic processes discussed previously.
In assessing geological processes, models, and geometric
interpretations, an objective rather than subjective approach is the
key. Believing that a geological interpretation of geometry is true
will not make it so, and can lead to material errors in resource
delineation and estimation. Such objective analysis is critical for
effective risk evaluation of resources.
Geologists can significantly contribute to the risk assessment
process through an understanding of the issues facing the
downstream users of geology inputs. For example, inadequate
understanding of the geometry of mineralisation is one of the
critical contributors to resource risk, as it results in improper
stationarity decisions for resource estimation. Many of such high
risk issues in mining projects can be significantly ameliorated
through the focussed application of sound geology, as we outline
below.
So, having outlined the issues that geologists can potentially
constrain, and the tools at their disposal, how exactly does the
geologist fit into the mining process?
12
THE MINING PROCESS

Components of the mining process
The mining process comprises a number of subdisciplines and
stages, which are summarised in Figure 4.
Figure 4 emphasises the role of the geologists in the mining
process, and how their understanding of geometry and geological
processes directly or indirectly impacts on the various mining
subdisciplines. The fundamental contribution of geologists is to
contribute a 3D geological model, utilising the tools at their
disposal in the geologists toolkit (Figure 2, Figure 4).
Examples of key issues facing downstream users that can have
material effects on projects, and into which geologists can have
significant inputs, are listed in Table 1.
The ore reserve estimation process as an example

Each aspect of the mining process has dynamic links with
geology. To illustrate some of the dynamic links between
geology and the mining process, we use the example of the
resource estimation process as outlined in Figure 5. This reserve
estimation process runs in parallel with the stages in the mining
process outlined in Figures 3 and 4. Note that the inputs of
various subdisciplines overlap, and have strong links in these
areas. Geology is the one discipline that links strongly
throughout the mining process.
Furthermore, Figure 5 emphasises that other subdisciplines of
the mining process must be considered throughout the reserve
estimation process. While examination of Figures 3 through 5
may suggest that this is obvious, it is remarkable in the authors
collective experience how often this cross-fertilisation between
the various disciplines is not accomplished, with the tasks of
geology interpretation, resource modelling, geotechnical
engineering and mine planning and metallurgy being carried out
petrography
geophysic
mapping
drilling
lithology
distribution
structure
distribution
geochemistry
grade/alteration
distribution
Structural
Analysis
3D GEOLOGICAL MODEL
Target
Generation/
ranking
Genetic Model
Alteration
Processes
Extensions
to ore
sterilisation
Exploration
Strategies
KPI
3D Structure
and Lithology
Stationarity
Decisions
Geostatistics
variography
re
Metallurgy
(petrography)
controls on
ore minerals
co
ve
ry
Mineral
Processing
Strategy
Water Levels
and Structures
3D distribution
estimation
rock mass
character
stress
model
water
chemistry
Geotechnical
Domain
Model
3D Resource
Model
resource
classification
Waste
Character
and
Management
Reservoirs
smu
mining method
cut-off grade
water
amount
Hydrogeology
Model
water
balance
Environmental
Strategy
mine
closure
mining
footprint
MINERAL
BALANCE
SHEET
(reserves, schedule, costs, etc)
MINING STRATEGY
KPI
Comparison
Option
Decision to
spend $
RETURN TO SHAREHOLDERS
FIG 4 - Disciplines and stages in the mining process, emphasising major linkages.
13
TABLE 1
Techical subdiscliplines of the mining process amnd where geology impacts upon them. Items in bold type represent
potentially material risks.
Sub-discipline/aspect
Key contribution
Examples of where geologists can make a difference
Geology/resource delineation
- 3D geological model for inputs into resource and Fully utilise geologist's toolkit to constrain both geometry and
genesis of ore. Aim is to provide best geometric + genetic
mine planning
model possible
- Understanding of mineralisation processes for
inputs to near mine exploration/sterilisation.
Geostatistics/resource estimation
3D grade distribution estimation block model
Well constrained 3D geometry model to aid in stationarity

decisions (correct shapes, position and classification of
boundaries)
Close liaison throughout resource estimation, particularly in
variography stages to clearly communicate geometric controls
on ore
Proofing of resultant block model to verify that it is
geologically sensible
Sound geology can severely impact on resource classification
and reduce technical risk
Geotechnical engineering/rock
mass characterisation, slope and
support considerations
Geotechnical domain model
Knowledge of geotechnical domain logging

Attention to nature, density and distribution of structures and
lithology AWAY from mineralisation (eg pit walls)
Identification of high-risk geotechnical domains
Attention to geotechnical data collation during infill drilling
can significantly reduce costs
Hydrogeology/water balance
Hydrogeological model
Attention to collation of water data during exploration.

Installation of piezometers (relatively cheap) for water level
information over as wide an area as possible from early
exploration stages can save significant dollars at the feasibility
stage by not having to redrill hydrogeology-specific holes
Accurate delineation of potential aquifers/aquacludes
(structural, lithological)
Metallurgy and process

engineering/ore recoveries,
processing technology and
procedures
Recovery model and processing strategy
Environmental
Environmental strategy
Petrography of mineralisation from early stages to help

constrain mineralogical nature and controls on distribution of
any refractory ore
Assay for more than primary commodity (penalty elements,
environmentally sensitive elements, Co, S for Cu deposits) helps to get a handle on their spatial distribution before
feasibility
Early identification and distribution of potentially
acid-forming material, acid neutralising material for design of
waste management strategies
Identification of any high concentrations of environmentally
sensitive elements
Mine planning
Mineral balance sheet

Life of mine plan
Robust geometric model and genetic model to prioritise

conversion of resources and reserves, identify potential
extensions to ore, properly assess sterilisation of
infrastructure sites
Delineation of critical areas of risk in geometry
interpretation that could affect scheduling/reserve
conversion
Assistance in defining potential selectivity of mining
Definition and subtleties of weathering profile
Overall risk assessments
Operations
Effective execution and optimisation of mine plan Design of adequate grade control procedures in conjunction
with geostatisticians, particularly noting where geology can
help constrain ore/waste blocks (eg highly visible alteration
assemblages, prominent ore-controlling structures)
Continued improvement of geological model for multiple
iterations and continued optimisation of resources and
mine plans.
14
Iterations of Technical, Financial and Risk Assessments
Activity
Exploration
and Data
Collection
Modifying
Factors
Statistical
Analysis,
Mineral
Resource
Estimation,
Classification
& Reporting
Geological
Interpretation
and Modelling
Mining,
Metallurgical,
Economic,
Marketing,
Legal, Environmental,
Social & Govt.
Risk and Sensitivity

Assessments
Ore Reserve
Estimation,
Classification
& Reporting
Monitoring the
Resources
and Reserves
Grade Control,
Production &
Reconciliation
Input
Laboratory
Geology
Geostatistics
Geotechnical
Hydrogeology
Mining Engineering
(Mine Planning)
Metallurgy
Marketing
Environmental
Financial
Legal, Social
& Government
FIG 5 - The ore reserve estimation process.
in isolation from each other. Such insular approaches minimise

the benefits of cross-fertilisation and significantly increases the
technical risk and probability of fatal flaws in mineral projects.
It is also imperative to recognise that the mining process is
iterative right through to the operational phase. As more
information is gained during mining, geology models (geometric
and genetic) are improved, resources recalculated and
reclassified, and mine designs altered to optimise the project.
Thus, since projects are not a single pass through Figures 3 - 5,
there is even more opportunity for cross-fertilisation and geology
inputs are critical throughout the life of mine.
RISK ASSESSMENT AND GEOLOGY

Given the keystone aspect of geology in the mining process
outlined in the previous sections, it is natural that geologists
should be instrumental in the evaluation of technical risk on
projects, yet this is rarely the case. Geologists often tend to focus
their investigations into areas that do not maximise value
received versus expenditure of time and money. This arises
because geologists by nature tend to be more interested in the
genetic and theoretical side of mineral deposits (most geologists
did not enter geology specifically to become mining geologists,
or even to be involved in the mining industry), and because few
geologists have a full appreciation of the concepts of how
geology fits within the mining process as outlined above.
However, with an appreciation of where geology inputs impact
on the mining process, a geologist can provide quality inputs to
the above categories to evaluate, manage, and hopefully mitigate
risk in all aspects of mining much more tightly. A geologist
should focus inputs to those issues that could have a material
impact on projects. In the current economic climate with low
commodity prices, such input can make or break projects.
Effective risk management focuses on issues that could lead to

material errors in mineral projects (high uncertainty and high
financial consequence). Results of geological assessment of
project risk must then be effectively communicated to other
disciplines and management. Risk matrices are useful ways to
accomplish this (Table 2). Presentation in this format clearly
communicates where the risks lie and allows management to
quickly comprehend the issues and decide where to commit
resources to add value to projects. In the example presented in
Table 2, it is clear that the geological inputs to the resource
estimation are rated at moderate to high risk, and that more
emphasis on constraining the structural controls on ore and the
refractory nature of ore are required. Such tables rating the
quality of geological inputs can be constructed for any or all
aspects of the mining process.
GEOLOGISTS SHOULD LEAD THE WAY

The authors contention in this paper has been that geology is
presently under-utilised in the mining process. This situation has
arisen through a combination of some or all of the following
factors:
1.
most geologists do not fully understand the mining process,

and therefore where they can contribute more effectively to
downstream users in mineral project development;
2.
geologists do fully utilise the toolkit at their disposal to

address these issues;
3.
geologists do not focus their inputs to prioritise issues with

potential material effects on projects; and
4.
geologists do not effectively communicate their results to

the rest of the mining community.
15
TABLE 2
Example of a risk matrix for geological inputs into a resource estimation. Risks are classified as low = little risk to project, medium =
moderate, but non-material risk to project and high = potentially material risk to project.
Aspect of data/interpretation
Survey data
Contribution to Drill campaign 1 Drill campaign 2

overall project risk
Development
Comments
low
low
low
low
Survey data is of high quality
Sampling procedures
medium
medium
low
low
Sampling procedures are adequate for

resource estimation purposes.
Development sampling is inadequate
for estimation, but has not been used
in estimation
Specific gravity data
low
low
low
n/a
Specific gravity data is of acceptable

quality
Assay data
low
low
low
low
Assay data is of acceptable quality
low
high
medium
medium
low
medium
low
medium
low
low
low
low
Structural logging is of poor quality

due to poor orientation quality and
logging techniques. Weathering has
not been consistently logged.
Mapping/logging quality
Lithology
Structure1
Alteration
Weathering
Geology interpretation
Lithology
Structure2
Controls on mineralisation/
alteration3
Weathering surfaces
Understanding of nature of
refractory ore4
Stationarity decisions
Position of domain
boundaries5
Nature of boundaries
low
medium
low
medium
Structure interpretation is poorly

constrained, and controls on
oreshoots are unconstrained.
Weathering surfaces have not been
adequately defined, but contributes
only moderate risk as most of
resource is in fresh rock.
low
high
high
medium
high
high
medium
low
Insufficient analysis of controls on

refractory ore. Requires petrography
on representative samples.
high
medium
low
Geological domains for resource

estimation purposes are only loosely
defined due to uncertainty in
structural interpretation.
medium-high
medium
1. Structural logging is inadequate due to poor core orientation. Drill campaign 1 also employed incorrect structural logging techniques.
2. The structural interpretation is poorly constrained, due to structural information quality, and misinterpretation of shallow mineralised links
between steeper major shear zones. This is viewed as a high-risk issue that may have a material effect on the resource estimation.
3. The controls on mineralisation, oreshoots, and variation in mineralogy with host rock is poorly understood. Given the variably refractory nature
of the ore, this could potentially have a material effect on reserve estimation.
4. The ore is variably refractory, with recoveries increasing with degree of weathering. However, the controls on refractory ore are poorly
understood in the fresh rock (the bulk of the resource), and contributes high and potentially material risk to the project. However, it will not
materially effect initial cash flows due to high Au recoveries in the weathered zone.
5. Due to the uncertainty in thestructural interpretation, the position of major boundaries for purposes of resource estimation may be in error,
contributing high and potentially material error to the project.
The end result of this, and especially the latter point, is that the
other mining disciplines do not fully appreciate where sound
geology can assist them in their respective tasks.
The recent economic climate has resulted in a severe downturn
in the geology profession. While the lack of appreciation of the
potential applications of geology is widespread throughout the
industry, the onus is on geologists to lead the way towards more
fully integrating geology with the mining process by increasing
the awareness, benefits and ultimately status of geology in
mining.
The geologists challenge is to utilise an appreciation and
understanding of the complete mining process as outlined in
Figure 2 through Figure 5 in a five-fold approach:
1.
Become familiar with the geologists toolkit. Fully

understand what investigative techniques are at your
disposal and what technical issues they can possibly
constrain.
2.
Take an avid interest in, and educate themselves in, these

various aspects of the mining process.
16
3.
Identify the major issues facing the downstream

processes/users, and treat them as your clients. Realise
which will be potentially material in nature (prioritisation
of high risk issues).
4.
Focus geology data collection, interpretation and inputs to

address these high-risk issues.
5.
Communicate the results to clients or project team

members (downstream users) in a practical format (eg risk
matrix).
In addition, it is necessary to adopt a mining process mindset
and address these issues at the earliest possible stage of resource
delineation. By being aware of and addressing issues such as
those outlined in Table 1 early in the process, significant cost
benefits, shortening of timelines to start-up, and technical risk
reduction can be achieved. We emphasise that this can only be
achieved through close liaison between all mining subdisciplines,
each with a mutual appreciation of the others abilities,
limitations and required inputs.
ACKNOWLEDGEMENTS
REFERENCES
The authors wish to thank their many colleagues within SRK and
clients worldwide who have contributed to their understanding of
the role of geology inputs to the mining process. S Jackson and
an anonymous reviewer are thanked for critically reviewing the
manuscript. J Mower and P Sansom are thanked for assistance in
assembling this document. Drafting of figures was provided by
Kerry King Graphics.
McCuaig, T C and Hronsky, J M A, 2000. The Exploration Industry

Economic Geology Research interface: A review of its current status
and vision for the future. SEG workshop, Reno, Nevada, September
2000.
Stephenson, P R and Vann, J, 1998. Common sense and good
communication in mineral resource and ore reserve estimation, in
Proceedings PACRIM 99, pp 435-441 (The Australasian Institute of
Mining and Metallurgy: Melbourne).
17
Sampling and Databases

Systems and Protocols of Geological
Information Management at Telfer Gold Mine,
Western Australia
R D Carlson, G R Howard
and G Back
21
A W Burgess, J R Foley,
R J Henham and
A G Shellshear
27
Sampling Practice at the Vera Nancy Gold

Mine
D Sims
35
Leachwell Versus Fire Assay: Comments from

Assay Quality Assurance at Tarmoola Gold
Mine
D J Elder
43
Systems and Protocols of Geological Information Management

at Telfer Gold Mine, Western Australia
R D Carlson1, G R Howard2 and G Back3
ABSTRACT
Management of information systems for data collection is essential to
maintain data integrity and continuity, and manage loss control on a large
world-class mine such as Telfer Gold Mine. This paper aims to describe
the individual systems and protocols of geological data capture at Telfer,
then focus on Quality Assurance Quality Control data as an example of
loss control in practice. Loss control systems are required to provide
capacity, flexibility, structure and management tools to enhance data
interrogation. The data that have been collected can then provide
geological, financial, environmental, and contractual information to
optimise management of business best practice.
Telfer Gold Mine is a wholly owned and operated subsidiary of
Newcrest Mining Limited. The mine is located in a remote area of NW
Western Australia, and has been operating for 20 years. Geological data
have been collected at Telfer since 1973. The data include drilling, assay,
geological, survey, structural and geophysical data. Data management has
evolved through many systems with data stored on many different
hardware and software platforms. Commencing in 1995, collation,
standardisation and validation of the historic data was completed. This
has resulted in the first Access based drill hole database incorporating all
available drilling data. This drill hole database, now hosted in SQL
Server, is the hub for all geology related data management systems.
The Telfer Gold Mine Geology Department has developed a series of
inter-related database systems in order to manage the core geology
business issues. Databases relating to drill hole planning, drill contract
management, geological logging, accounting (standard costing,
forecasting, invoice reconciliation), quality control (QAQC), and
environment rehabilitation have all been developed to manage loss
control.
Management effort to provide adequate data systems and protocols has
driven the system from concept to utilisation. The improved productivity,
data integrity and data analysis capabilities provided by the new database
systems at Telfer have allowed for greater confidence in decision making.
Systems such as the drill hole and QAQC databases, when utilised in the
proper manner provide up-to-date data analysis for resource audits and
contract negotiations.
The loss control systems are essential to maintain confidence in data
integrity and hence directly effect the future of mining at Telfer.
INTRODUCTION
Control of avoidable loss in the management
system is one of the few remaining areas of
business with major cost-reduction potential
DNV, 1999.
The Telfer Gold Mine is a world-class gold/copper mine in the
remote northwest of Western Australia. Originally discovered in
1971 and mined since 1975, Telfer is a mature mine that has
produced in excess of 5.7 million ounces of gold (Howard et al,
2000). The mine is currently undergoing a major review to define
extensions to the existing open pittable and underground
resources. This paper describes the individual systems and
protocols of geological data capture at Telfer, with focus on
specific Quality Assurance Quality Control data as an example of
loss control in practice.
1.
Newcrest Mining Ltd, Geology Department, Telfer Gold Mine,

PO Box 6380, East Perth WA 6004.
2.
MAusIMM, Newcrest Mining Ltd, Geology Department, Telfer Gold

Mine, PO Box 6380, East Perth WA 6004.
3.
Back Enterprises Pty Ltd, 10 James Street, Guildford WA 6055.
The database established to store current and historic drilling

records, is the single most valuable resource held by Telfer. The
system currently stores in excess of two million assay records
and 1.5 million drill hole geological records.
The revised protocols established in the past two years have
been focussed on providing a modern, robust, flexible and cost
efficient data storage system capable of minimising business risk
through appropriate loss control systems.
SYSTEMS ARCHITECTURE
Network
A PC network was established at Telfer in the late-1980s, and
has since been progressively upgraded on an as needs basis. It
currently uses NT servers and Windows based client machines.
All client machines were recently upgraded to 300 Mhz Celeron
processors or better and NT4 operating systems during the Y2K
upgrade project.
Principal accounting and operations management systems such
as payroll, purchasing, and maintenance planning are based on
the MIMS application system over an Oracle database.
Microsoft Access
Microsoft Access was adopted as the front end development tool
for the databases because of the low technological demands and
likely future development life cycle ensuring the database will
have a long shelf life. The initial systems were developed in
Access 2. For Y2K compliance and 32-bit capability, the
front-end databases are now fully converted from Access 2 to
Access 97. All laptops including rugged logging computers used
for field based logging utilise Access 97.
SQL Server
The existing data were converted to a standardised database over
several years. Data were loaded, reformatted, re-coded, and
validated and now have a high level of confidence. Owing to
performance, security and reliability reasons the database was
converted to Microsoft SQL Server version 6.5. Version 6.5 now
forms the datastore for all databases with shared data. The data
reside in seven distinct databases with some local data remaining
in Access. A part-time specialist SQL Server Database
Administrator supports SQL Server. SQL Server will be
upgraded to Version 7 in the near future.
Communications
A recent upgrade of the site cabling from coaxial to fibre optic
has increased communication speeds from 10 Mb/s to 100 Mb/s.
This has substantially improved reliability and performance.
Enhancement of hardware at both client and server ends of the
process has improved productivity and reduced frustration
arising from response delays and failed processes.
Client Server
The move to SQL Server was the result of the data storage and
retrieval needs of Telfer exceeding Microsoft Accesss design
limitations. Update and retrieval times were becoming too long
and data security was becoming a concern with nearly two
21
R D CARLSON, G R HOWARD and G BACK
million records in the database. With a separation of roles

between Access, as the front end, and SQL Server as an
intelligent back end, the security, performance and network
traffic outcomes have been significantly improved.
Replication
The Drill Hole Database is replicated between Perth and Telfer
and is equally available for review and extraction at either end.
The replication is managed by SQL Server and is reliable,
reasonably quick and does not appear to place undue traffic on
Perth-Telfer Communications.
ODBC
Online Data Base Connectivity is a standard initiated by
Microsoft and now supported by many products including SQL
Server, Access, Surpac, MapInfo, etc. Integration of data
generation by modelling products such as Surpac, and database
management by custom written Access applications such as the
Face Sampling database via ODBC, provides time saving and
reduces error.
Imaging
All core photographs are now captured in JPEG (a compressed
image format) files. The images are filed electronically and are
available for on-line review on the network. The drill hole
database links drill samples to the relevant image file. Core tray
image processing is now a routine part of the data management
process. All historic colour slide images are being scanned for
ease of access in data validation and geological interpretation
Integrated security
NT network security has been implemented so that SQL Server
will permit only appropriate access levels based on the
permissions assigned to the network logon. Data are assigned
for update by authorised staff, with read only for other staff.
Sensitive data can be restricted by the network logon.
DATABASES
Planning index
The planning index is utilised as a front end to the data handling
process for all drill holes planned at Telfer (Figure 1). The
planning index records all details of drill hole planning including
collar coordinates, dip, azimuth, predicted depth and accounting
cost centres. drill hole locations determine hole numbering with
the database assigning sequential numbers to holes drilled on the
same section northing. Accounting cost centres are defined
automatically from grid coordinate data. Drilling type such as
reverse circulation (RC) pre-collared diamond drill holes are
defined so that the holes can be scheduled. All planning
information is recorded in the drill hole database to validate the
pick up of the surveyed collar and end of hole information. If
data vary from planned details then senior staff are required to
sign off on changes, agreeing to the disparity. This is part of the
validation process.
Forecasting
The forecasting database is utilised to estimate and schedule the
drilling programs. Details of drill hole sequence by drill rig are
entered. Forecasting is then estimated using predefined drilling
rates. The system incorporates estimations of time taken to move
between sites, and to convert from RC to diamond on
multi-purpose rigs. Hole planning can be also be adjusted to
factor extraordinary circumstances such as access to benches
affected by production requirements.
Drill control
Drill control is a database that stores information relating to all
chargeable and non-chargeable items affecting drilling rigs.
Geologists transfer the drilling performance information supplied
on daily drilling shift record sheets (plods) on a daily basis. All
information related to metres drilled, consumables used, working
time chargeable and non-chargeable are recorded. This provides
an up-to-date, precision cost control system. Summaries of costs
System Map
Planning
Index
Drill
Control
Invoice
Reconciliation
Forecasting/
Standard Costs
SURPAC /
DATAMINE
DHDB
Barcoding
Rehabilitation
Geological
Logging
Staging
QAQC
Newcrest Mining Limited
FIG 1 - System map.
22
SYSTEMS AND PROTOCOLS OF GEOLOGICAL INFORMATION MANAGEMENT
to-date can be calculated, and the invoices from drilling

companies reconciled against the shift record information. The
system has sophisticated reporting functions that enable graphing
of information including drilling rates, consumable usage and
breakdown time.
Drill hole
The drill hole database (DHDB) is the key process that links all
the other systems (Figure 1). DHDB records all information
relating to geological logging, sampling, sample tracking,
surveying, and assaying for every hole drilled at Telfer. The drill
hole process commences with generation of a drill hole name
formulated in the planning index. Information such as accurate
collar survey, geological logging, assaying intervals, sample
dispatch, returned assay data, and downhole survey data are
added incrementally as the data become available.
Geological logging
The geological logging codes and tables record all data from
logging of drill hole chips and core. This was revised to its
present form in 1998. These logging codes have been derived
using parameters defined over 20 years of logging drill holes at
Telfer. The task required the re-coding of over 1600 different
historic codes into the new codes. The geological logging is split
into five primary tables:
Alteration mineral and intensity;
Lithology stratigraphy, rock type, oxidation, colour,
grainsize, fabric, comments;
Mineralisation mineral, intensity, and occurrence;
Veining type, texture, attitude, percentage, width; and
Structure width, type, attitude.
The geological logging database is used on laptop computers
for entry of data in the field. The system is used to enter all
geological, sampling and preliminary downhole survey data
(Eastman single shot). The data entry of information is validated
as entered via look-up tables. Further validation is completed as
the data are transferred to the primary database.
beginning and end of each hole, both before and after the tags are
attached to the calico sample bags. The system also has an area
for entry of field-split duplicates and standards. This ensures
correct numbering of sample bags, with no missing bags, and
correct sequencing of the duplicates and standards. The samples
are delivered to sample preparation and the barcoder is returned
to the office for data downloading and validation against the drill
hole database.
Diamond drilling
The sampling of diamond drill core is handled slightly
differently, as the intervals associated with sampling of core are
irregular, and defined by the geologist. In this case the geologist
logs the core, and defines the sampling intervals in the geological
logging software. When an appropriate point is reached the
sampling information is downloaded to the server, into a
temporary storage area. The sampling information is then written
to the barcode read ready file which is uploaded to the barcoder,
as well as printed out as a cutting sheet. Core is cut and sampled
to these intervals, packaged into calico bags, and barcoded for
submission to the laboratory.
These processes ensure that a chain of custody is established
for all samples from the point of extraction to the point of
delivery to the preparation laboratory.
Quality Assurance and Quality Control (QAQC)

Quality assurance of analytical results is monitored via a
database linked to the assay, standards, and splits tables within
the primary database. Functionality includes plotting of time line
graphs by date range, standard, laboratory, and assay element.
These plots and associated summary printouts are utilised to
audit assay batches for accuracy and precision of element
analysis. Scatterplots are generated to check duplicate analyses.
Statistics summarising sample and standard submission are
generated in a reporting function. Reporting on a monthly basis
provides up-to-date monitoring of laboratory performance for
management.
Environmental rehabilitation
Sampling
All samples are prepared at the in-house Telfer Laboratory.
Samples are either assayed at Telfer or pulps are transported to a
commercial laboratory in Perth. Details of all laboratory job
number and technique details are stored in the DHDB assay
registry. All assay records are merged from digital files.
Laboratory data are recorded in assay, standards or splits
(duplicates) data tables. Once all records have been received
from the individual sources, the hole data are printed, and
validated by the senior geologist. These paper records are then
stored in a comprehensive filing system.
Assay barcoding
Invoice reconciliation
RC drilling
A system utilising barcoding technology has been developed to
track all drill samples from field collection to the assay lab. The
drilling contractors are responsible for the drilling and splitting
of the hole, whist the collection and submission of RC samples in
the field is the responsibility of the field assistant workgroup.
Once a RC hole has been completed, and all samples are laid out,
in order, on the top of the bulk residue, a field assistant enters
information into a handheld barcoder that records the name of
the hole, the total depth, and drill hole coordinate information if
required. The barcoder software then calculates the number of
samples and prompts users to scan printed barcode tags at the
The geological rehabilitation database (GRD) is a stand-alone

database predominantly utilised on field computers. It records
auditing progress of rehabilitation on all drilling and area
disturbances, including status of tracks. Linking to and updating
from DHDB, drill collar positions are initially populated from
the database records. Field checking is used to provide the audit
details about requirement for further rehabilitation earthworks.
Data such as area of disturbance and nature of soil types are used
to provide a quantitative estimate on cost per item to allow for
calculation of total environmental liability. MapInfo is used as a
graphical database to record all information related to spatial
information such as tracks and borrows pits.
The invoice reconciliation database is primarily utilised to record

contract personnel daily or weekly time sheets. Details such as
nature of work and cost codes are utilised to reconcile invoices
against appropriate budget cost centres. The database is used to
provide accurate guides for calculation of end of month accruals,
which are passed to the MIMS system.
Occupational health and safety

An interim system was developed to record details of safety
incidents and injuries. This has recently been migrated to the
newly introduced corporate safety system based on the Site Safe
package and MIMS HR system.
23
R D CARLSON, G R HOWARD and G BACK
Case 2: Lab sample handling error
QAQC LOSS CONTROL EXAMPLES

Laboratory analytical accuracy and precision case histories are
well documented in a number of recent proceedings and papers
(AusIMM-AIG, 1999; Francois-Bongarcon et al, 1996; Shaw,
1997). The QAQC system now established at Telfer has a
well-defined role in attempting to lower the risks associated with
ore resource and reserve estimation by ensuring that analytical
data are within specifications. The examples that follow were
discovered following routine checks implemented as part of the
monthly QAQC reporting procedures.
Case 1: Sulphur bias

A sulphur bias was recognised in a number of batches from a
commercial laboratory when checks were run following a
disparity in background levels between close spaced drill holes.
The check results of the sulphur (by mixed acid digest-ICPOES)
were plotted against the original results and found to have a 19
per cent negative bias (Figure 2). The laboratory was asked to
explain the reasons for the disparity in results, some of which
were from original assaying occurring 18 months earlier. The
laboratory finally found that the batches were biased by the use
of a commercial de-flocculant used in settling of sediments prior
to ICP analysis. The de-flocculant was originally checked for all
metals such as gold and base metals prior to use in the
laboratory, however, sulphur was not included in the analysis.
Thus depending on the concentration of the de-flocculant, levels
of 200 - 2000 ppm S were introduced to the analyte. The
laboratory was asked to re-calculate sulphur levels for all batches
effected by the use of that particular de-flocculant, which
involved data files over a period of approximately 12 months.
These data were supplied and re-inserted into the database. This
highlights the need to be vigilant on all elements utilised in
resource estimation, not just the primary drivers such as gold or
copper.
A batch of results from a diamond drill hole was returned with

incorrect multi-element values for the inserted standards. The
gold values were all within acceptable limits but the
multi-element ICP suite (ie Cu, As, S, Ni, Pb, Zn) all lacked any
correlation to the expected results. The laboratory investigated
the causes and found that an operator had not followed correct
pulp handling procedures. A procedure has been established with
the laboratory, that 1:50 pulps were to have wet sizing analysis
completed to check grinding adequacy. The laboratory had a
policy that if the 50th sample was a low weight sample, the
following sample was chosen instead. This procedure was
apparently incorrectly applied to the weighing out of the wet
chemistry pulp. The operator came across low weight samples
(in this case always a standard), and took the following sample
and placed the weighed pulp from it into the test tube for the
preceding sample. Thus, every standard had the multi-element
results from the following sample. This procedural error would
not have been detected without checking the multi-element
signature of the standards.
CONCLUSIONS
The systems established at Telfer are examples of loss control
mechanisms. All data generated by the Geology Department (eg
drill company performance, assay accuracy and precision, or
safety incident reporting) have the potential to identify areas for
improvement in costs, safety, and reserve or resource
estimations. The systems developed for Telfer Gold Mine
provide relevant information for management of key
performance indicators.
Re-as s ay of drillhole s ample S-ICP res ults

8000
7000
6000
5000
4000
3000
2000
Bias=~19%
1000
0
0
1000
2000
3000
4000
5000
6000
7000
8000
Original S-ICP (ppm)

Newcrest Mining Limited
FIG 2 - Re-assay of drill hole sample S-ICP results.
24
SYSTEMS AND PROTOCOLS OF GEOLOGICAL INFORMATION MANAGEMENT
ACKNOWLEDGEMENTS
This paper is published with the permission of the CEO of
Newcrest Mining Limited.
REFERENCES
AusIMM - AIG, 1999. Good Project - Wrong Assays: Getting sample
preparation and assaying right! Australasian Institute of Mining and
Metallurgy and Australian Institute of Geoscientists (AIG
Publication No 27).
DNV, 1999. 2 Day Leadership Training Manual, Det Norske Veritas
Consultancy Services. Unpublished DNV Training Manual.
Howard, G R, Hansen, T, Moore, C, Moffitt, P, Inglis, R, Carlson, R D,

Kirchner, I, Coupland, D, Leary, S and Tomset, A, 2000. Current
geological understanding of the Telfer Gold Mine, WA, in
Proceedings 4th International Mining Geology Conference, pp
135-142 (The Australasian Institute of Mining and Metallurgy:
Melbourne).
Francois-Bongarcon, D M, Long, S D and Parker, H M, 1996. Assay
quality assurance - quality control program. Mineral Resources
Development Inc.
Shaw, W J, 1997. Validation of sampling and assaying quality for
bankable feasibility studies, in Proceedings The Resource Database
Towards 2000 (The Australasian Institute of Mining and Metallurgy,
Illawarra Branch: Wollongong).
25

A W Burgess1, J R Foley2, R J Henham3 and A G Shellshear4
ABSTRACT
The Porgera Gold Mine, west of Mt Hagen in PNG is a large open pit
and underground operation currently producing gold at the rate of almost
one million ounces per year. This deposit has produced some 8.8 million
ounces to-date and has current reserves of 103.4 million tonnes @3.2 g/t
Au. The Porgera Joint Venture has invested over $A1 billion in
infrastructure in this project.
Technical management of this operation involves the interrogation of
some 700 000 metres of diamond drilling, some 50 000 blastholes per
year, and very large volumes of laboratory, geotechnical, environmental
and production reporting data.
Over the past seven years this data has been consolidated into a single
centralised, relational database which now acts as the single reference
point for all the technical application systems used for surveying, grade
control, production reporting, deposit evaluation and modelling. Data
transfer to application programs and management systems is controlled
by the centralised database, this process at times involving a two-way,
real-time, binary data transfer.
The benefits of a specialised database application include security and
integrity of the data asset, ease of access, and performance in a multi-user
environment. This replaced fragmented, inconsistent data sets stored
inefficiently as individual, unindexed, variably formatted files associated
with the different application programs.
Most importantly, the database has allowed mine management at all
levels to make timely, confident investment, cash flow and optimisation
decisions as it provides the primary link between the deposit and
operational data sets on one hand, and the application programs and
management and performance measurement systems on the other.
This paper discusses the economic value and importance of technical
data in terms of the resource asset and the means by which this data is
being managed and used at Porgera.
out by sampling blast holes, and by determining the ore

categories prior to mining. Production is budgeted to mine
approximately 210 000 tonnes per day.
Published reserves at the end of December 1999 comprise
103.4 million tonnes @ 3.2 g/t. The reserve is contained within a
measured and indicated resource of 166.4 million tonnes @
2.7 g/t (at a 1.0 g/t Au cut-off). Additional inferred resources of
14.8 million tonnes @ 3.2 g/t Au are also identified on the
property.
The acquisition cost for obtaining geological, engineering,
geotechnical and environmental datasets required to manage the
various phases of project development are considerable. A
conserveritive estimate of drilling, analysis and testing costs for
the 70 km of drilling to-date is $A60 million.
This resulting technical database is the primary link between
the deposit itself, and the decisions being made about that
deposit, in an effort to provide a cash flow and to produce a
profit.
The speed, effectiveness and integrity of those decisions will
therefore be affected directly by the accessibility and accuracy of
that data. The inability to access this data with confidence, will
lead to delays, errors and re-work that can affect this cash flow
and profit in very real (and large) terms.
The value of project technical data needs to be recognised and
consideration given to the cost-effective management of this
data. Proper data storage systems empowers project management
to maximise the return on project investment by making timely
and confident operating and investment decisions.
INTRODUCTION
THE ORIGINAL ENVIRONMENT AT PORGERA
The Porgera gold mine is located in the highlands of Papua New

Guinea, about 139 km west northwest of Mount Hagen and is
owned and operated by the Porgera Joint Venture.
Situated at an elevation of 2500 metres, the deposit is
associated with oxidised Late Miocene alkalic intrusives of the
Porgera Intrusive Complex (PIC), emplaced at shallow levels
within Late Cretaceous and Early Tertiary sediments of the
Papuan Platform. Mineralisation has occurred in two stages with
mesothermal and epithermal events.
Mine production has been from both underground and open pit
mining and since the decision to mine in 1988, over 8.8 million
ounces of gold has been produced. The majority of this gold is
refractory and associated with pyrite (iron sulphide). Ore
processed in the mill is oxidised using autoclaves, and mixed
with cyanide for extraction, before neutralisation and discharge.
The open pit fleet consists of four O&K RH200 shovels and 31
Caterpillar 789 (175 tonne) trucks and associated ancillary
equipment. The current mining procedure at Porgera involves
drilling and blasting ten metre benches. Grade control is carried
As depicted in Figure 1, the original environment within the

Porgera Technical Services Department consisted of some 14
application programs which each had specific areas of
functionality. Each application program had its own specific
format requirements and individual data storage using mostly
sequential, unindexed files. This resulted in duplication of data
and data concurrency problems. Performance using the
unindexed files was poor, with waits of 30 minutes or longer for
the retrieval of assays from a file containing 500 000 records
(1400 underground holes).
The datasets used were often incomplete and simplified from
the data collected, for example, all lithological mineralogical and
structural descriptions were reduced to 13 two number codes.
Sample numbers were not recorded against assay records making
auditing of results difficult.
The datasets were also often fragmented with for example,
separate files for underground and surface holes.
Often interpreted data rather than the raw data were stored, (eg
downhole surveys and averaged assays) removing the ability to
check the validity of this data. With a formatted dataset file
required by each application program, numerous people had
access to the datasets, which meant the data integrity, and data
security was always in question. Back-up and restoration of the
datasets was limited to the daily system back-ups and no audit
trial or the editing of the numerous copies of data sets and subset
could be maintained.
The problems with this original environment commonly
caused delays and errors during project evaluation and
optimisation.
1.
MAusIMM, Senior Resource Geologist Porgera Joint Venture,

Porgera, Enga Province, PNG.
2.
MAusIMM, Senior Mine Geologist Porgera Joint Venture, PO Box

7050, Cairns Qld 4870.
3.
MAusIMM, Senior Exploration Geologist Porgera Joint Venture
4.
MAusIMM, Principal Software Design Consultants PL, 41 Abbey

Street, Forestdale Qld 4118.
27
A W BURGESS, J R FOLEY, R J HENHAM and A G SHELLSHEAR
Chaos
FIG 1 - Data duplication, transfer and reformatting between independent applications.
INITIAL VISION
The initial vision for the Porgera Technical Data Management
system was to develop a single repository for all technical
datasets; these were to be held within a relational database. This
database was to serve data to all application programs in the
required format and provide a multi-user access across the
Technical Services network.
Data accuracy was to be achieved by validation during data
entry while the integrity of the data was to be maintained by
various database tools. Electronic data transfer into the database
and between application programs was to be used wherever
possible.
The system was to provide for high levels of data security,
performance, availability, stability, and possess good audit and
recovery functions.
Design and implementation of the system was to be modular to
enable a staged development; and to be documented at each of
these stages.
Management commitment was required to provide adequate
resources in terms of staff and or consultants, money, time,
hardware, software and training.
The Porgera Joint Ventures management commitment to the
project was strong; hardware was upgraded, and two Ingres
database developments initially approved, being a production
reporting system and diamond drill hole database. Funds were
provided for the project to be developed and implemented by
consultants to the specifications of site personnel.
28
TECHNICAL APPROACH TO DEVELOPMENT

AND IMPLEMENTATION
Following completion of a vision for the project and the
definition of the broad specification, an evaluation was
performed firstly as to the conceptual model around which to
build the system, and then an evaluation of the tools available
with which to achieve this.
On consideration of the major duplication and reformatting
tasks inherent in the existing system, consideration was first
given to a fully integrated system. By this is meant the adoption
of a major proprietary system such as Datamine, Vulcan, Med
Systems or Mincom system, within which all data storage and
manipulation would be performed. This was subsequently
christened the Swiss army knife approach, alluding to the
attempt to find a single system that could be made to perform all
tasks required of the applications. Figure 2 schematically
represents this approach.
It was quickly discovered however, that there were many
application packages already in use on-site, several of which had
merit in some area or other. Like a Swiss army knife, it was
discovered that single integrated systems did many tasks in a
passable manner, but did few individual tasks to the same level of
capability as the specialist packages the mine was currently
using.
The model that was finally adopted was the centralised
database model, as depicted in Figure 3. Here all tasks related to
management of the data are removed from the individual
TECHNICAL DATA MANAGEMENT AT PORGERA
Acme
FIG 2 - Fully integrated one size fits all application.
Ingres
FIG 3 - Applications accessing a centralised data repository.
29
application packages, and vested in a centralised custom-built

database management system. This central system then serves
data in the required format to the application packages on
demand.
In some cases, the application packages then pass new or
derived data back to the database for storage.
In order to achieve this, a competent relational database
management system was seen as a crucial central tool for the
development. Computer Associates Ingres was the database
system chosen for the task, primarily because it had already
proved to be a very efficient, robust and reliable system at other
Placer sites.
Database design
Design of the database has been evolutionary, with modifications
and additions made as the project progressed. Management of the
database design has mostly been through the use of Visio
Professional, a systems design and documentation tool. Rigid
adherence to the majority of practices common to relational
database design has been important.
Development environment
The system was implemented primarily in a Unix based
environment, with Ingres installed on a Unix server. While all of
the databases can be accessed from the PC environment, all
database management is entrusted to the Server. Two
environments were established, one for development and a
second for the production databases, allowing development to
proceed without any chance of interfering with the real data.
Occasionally a third acceptance environment is used to allow
users to test new systems independently of the other two
environments.
The database applications have been developed using the
Ingres ABF application development environment. While this
has now been superceded by the OpenRoad graphical
environment, Porgera have felt that the very stable and efficient
character based environment offered by ABF is worth staying
with.
The majority of the systems are written in the 4GL language of
the Ingres database management system, and while this is an
older character based language, it is fully mouse-driven. In
certain situations dictated by size or complexity, C or C++ is
used for performance reasons; this is however completely
encased in the Ingres environment and the application providing
a single integrated interface for the users.
Development of one of the newer modules has been done in
PowerBuilder, a new database independent GUI development
environment. This application still sits over the Ingres database.
and the Placer OP system, again by transfer to a temporary

transaction file.
This data transfer is effected in a variety of formats, both
ASCII and binary, dependent upon the source system.
Validation and database integrity

Management of the data once in the database is effected by a
variety of tools including:
manual data editing screens, related to the original data entry
screens, again with real time validation;
batch or on request validation tools designed to check the
various database tables for things such as code validity, range
checks, referential integrity and redundant codes;
tools to prevent inadvertent or intentional corruption of the
data through the implementation of database rules,
procedures, events and triggers;
internal database security through restricted access and
privileges assigned on a user, group, object or location basis;
protection of the database from system crashes through the
use of rollback recovery, journalling and checkpointing, and
database dumps, in addition to the normal systems backups
performed by the on site systems administrators; and
automation and tracking of the backup functions for the
server systems.
Data manipulation functions

A number of functions are provided for the manipulation of the
data within the database, to provide additional access capabilities
to the data, and derived values. Two examples of this type of
manipulation are the calculation of de-surveyed coordinates,
and the addition of attribute tags to geological sample data.
De-surveying involves the calculation of co-ordinate data for
drill hole samples from the hole collar, down-hole surveys, and
sample depth. This trigonometrical exercise was previously
performed in an external package, which was a very slow task.
Tagging involves using the de-surveyed coordinates in
conjunction with the wireframe models in Datamine to determine
the inclusion or otherwise of samples within various structural,
geological, geotechnical or mining domains. This then allows the
data to be segregated for statistical or other purposes. This is
achieved by passing a binary file to Datamine, in the Datamine
native format, for tagging, with the resultant tags returned to
the database for storage against the individual samples.
These two capabilities are proving very powerful tools for data
management and evaluation.
Reporting functions
Development strategy
A modular development strategy has been used throughout the
project. Following definition of the initial vision, discrete
modules, for example the Blasthole Grade Control Module, were
specified and developed and commissioned. This removed the
requirement for a large up front commitment on the part of
management, allowing the success of each module to become
justification for the continued development effort.
Data entry
Data entry for the system is achieved through three mechanisms:
conventional screen based manual data entry, with real-time
validation as appropriate;
electronic transfer from on-line systems such as the
laboratory, with new data transferred to a temporary
transaction table for validation prior to posting to the
database; and
30
transfer from other application systems such as Datamine
A large number of standard selective reports are provided by the

system for retrieval of hard copy data for a variety of purposes.
Additionally SQL can be used for ad hoc reports for special
purposes.
Export/Import functions
In order to make use of the data in other applications, three
classes of export mechanisms have been constructed.
Conventional ASCII data export is provided through CSV or
Tab delimited files. This is accomplished through a series of
custom functions available through the system menus, and
through the use of ad hoc SQL queries.
Special routines have been provided to write binary data files
in the individual applications native format directly from the
database. An example is the Datamine binary de-surveyed
format, which can be used directly and immediately by
Datamine, circumventing the time consuming data import, merge
and de-survey functions normally required by Datamine. This

has reduced a four-hour process to around ten minutes.
Live links are also provided to data in the database, which
effectively removes the need to export the data in the first place.
A simple example is the ODBC access to the Ingres database
from Excel for graphical display. A more serious example is the
linking of the blast hole grade control system in Datamine to the
blast holes stored in the Ingres database.
Development management
The following factors and activities were considered important
for the development exercise.
Formal development management. The project was
managed as formal software development exercise through
detailed specification and design of the overall framework
and conventional functional modules.
New technology prototyping. A prototyping approach was
used in many areas to investigate use of new technologies.
Broad involvement. Involvement of many staff members in
the design, testing and implementation.
Standards and procedures. Development and testing
procedures were developed and followed. Formal
development standards were used to ensure consistency,
maintainability and robustness of the software developed.
Documentation standards. Documented standards ensured
all documentation (hard copy and on-line) would follow a
consistent, professional format.
SYSTEM DEVELOPMENT TO-DATE

The Technical Services Departments involvement in relational
databases commenced in 1994 with the development of a
Production Reporting System for both the surface and
underground operations.
This system, along with the others discussed below, now
constitute the data management environment in daily use for
production planning, control, reporting and data management. In
addition, live links between the database systems and the
graphical display tools such as Datamine allow immediate
visualisation of the new data coming into the systems.
Since 1994 the following database functions have been added
in a staged modular fashion.
Grade control database (MARVIN)

Management of blasthole and related data for use by the
Datamine Grade Control System is the purpose of this module. It
is used constantly as the primary tool for monitoring pit activities
pertaining to grade control, and planning daily mining tasks.
The database currently contains some 500 000 blastholes
which is increasing by around 50 000 holes per year. The
Database maintains a live link to the Datamine grade control
program and includes a blasthole sample quality control system.
A holding file is used for temporary storage of assay files from
the site laboratory so that matching to the blast hole location
using the sample number occurs prior to updating of the primary
assay table. The same approach to the verification of assays is
used in DDHS database.
All database storage and validation is done in Ingres and all
manipulation and interactive design is done in Datamine/Guide.
The menus controlling flow through the system are implemented
in a consistent working environment aiding training and provide
intuitive usage.
Benches are selectively exported directly to Datamine via
binary files for direct import into the mining software. This has
resulted in a large performance increase compared to working
with the entire dataset in a non-relational database.
DDHS database (WANTOK)

Used by all of the exploration, reserve and mine planning staff,
the drill hole database is the cornerstone of the overall system.
Links between this and the other modules, and other evaluation
tools forms the basis for all of the text based and graphical
reporting and appraisal activities.
This is the largest module, and comprises a repository for all
drill hole data for exploration, reserves and production control
functions. Currently contains some 700 000 metres of largely
diamond drill core.
Real time validation of codes and depths of manually entered
lithology logs occurs at data entry. Menu functions are run on a
monthly basis to enable table referential integrity checking, ie
depth matching of collar, assay and lithology tables.
Both measured and interpreted values for collar survey and
downhole survey locations are stored to provide a audit trail of
modifications.
Features include internal de-surveying functions and sample
tagging against Datamine model domain and structure
wireframes.
Wireframe tagging provides a powerful means of relating
geological, geotechnical and metallurgical interpretations to the
spatial distribution of blasthole and drill hole samples. Sample
information together with their tagged attributes can be exported
for use in geostatsical, resource, and other application program
requirements. As interpretations change the new set of
wireframes generated are used in the tagging module to update
all relevant databases.
Geotechnical database (ITAMBU)

A more recent module for management of all on-site
geotechnical data, it includes the storage of orientated core and
traverse mapping information. Database interaction with Excel to
produce graphs, scatter plots, etc. Export of data for stereonet
and Dips application programs is catered for.
Acid rock drainage database (DIDIMAN)

Acid rock drainage is a significant environmental management
issue at the minesite. This module provides storage functions for
all geochemical and acid based accounting data, for
environmental management use. Using the spatial tagging
described above, the ARD data can be extracted by domain, to
help model the optimum waste dump destinations for the
material being mined. This data included multi-elements and
acid based accounting analyses such as net acid neutralisation
and pH ranges.
Sample numbers are cross-indexed with drill hole information
to allow spatial location (necessary for modelling). Selective
reports of individual holes or data identified above key values
can be exported from the database for analysis in visual
programs such as Guide or imported into statistical software for
analysis.
Drilling management system (DRILLRODS)

In order to provide an example of a development that provides
management information, this module will be discussed in more
detail. The Drilling Management System consists of the
following functional components:
new drill program and hole planning;
tracking of the drill hole preparation, including drill pad
set-out, allocation of holes to drill rigs, drill rig movement
and setup;
tracking of the drilling progress for current drill holes, rig
plod sheets, consumables, time-sheets and performance
monitoring;
31
management and tracking of drill hole logging and sampling

activities, including lithological logging, summary logging,
geotechnical logging, sampling, sample remainder storage,
core photography and downhole geophysical logging; and
collar and down hole survey management and adjustment;
sample management and tracking of laboratory assays;
monitoring of data entry progress for the geological
database;
reporting functions for program management, including
periodical progress reporting, backlog reporting for all
aspects of the management role, performance analysis, and
identification of problems such as mismatches in sample
numbers and lost samples;
custom functions for such things as calculation of significant
intersections; and
utility functions for maintenance of the system, including
security and integrity of the system, access control, and
maintenance of reference tables for validation.
The monitoring of drill rig performance has improved the
metres drilled against budget by 140 per cent. This system has
enabled a team of six geologists and 12 support staff to manage
some 90 000 m/annum.
Production reporting (MARK2)

Reporting of open pit and underground production. The system
uses a database within the Modular Mining system and replaces
the original Ingres production reporting system that had been in
service for the past eight years.
Reserves database (DOUG)

Prototype built to provide database for storage of Reserves Block
Models and related data to allow selective interrogation and
retrieval by a variety of systems and users.
Reconciliation (GRAVEYARD)
Reconciliation of grade control, mill and production data,
currently under development. A prototype of this system
currently in use has proved very successful in quickly identifying
some problems such as material allocation and stockpile
management.
We intend to investigate Object Oriented Multi-Media

Database technologies for use in storage of a broader range of
data such as; core photography, geophysical logs, technical
drawings and PIMA traces, etc.
In addition, the application of new hardware technologies,
scanning, fibre channel architecture disk arrays will be reviewed.
BENEFITS
The single most significant benefit has been the centralisation of
the technical data to a single, secure, robust and reliable
repository with a relational database.
Following are a list of some of the specific benefits deriving
from the project to-date:
Real time functions for linking databases with other systems
(eg Datamine Grade Control), along with export functions
for passing data to other systems (eg Excel Graphing, OP
system) are broadly used, and saving significant times in
reporting tasks.
Integration of departmental data sources has seen a dramatic
reduction in data entry and maintenance requirements, which
now involves two to three people rather than eight to ten.
High security of the data has prevented many of the nasty
little accidents that used to happen.
Very high availability and stability of the databases takes the
worry out of reporting.
Good access to data for other projects.
Restored confidence in data and in the evaluations derived
from that data.
Dramatically improved performance is saving significant
time. An example is the Datamine drill hole de-survey
process of loading drill hole collars, surveys, assays and
lithology records into Datamine and then generating a
de-surveyed XYZ binary file for Datamine display, a process
which used to take about eight hours. The new database
system now writes this binary file directly in ten minutes.
Allows technical staff to become more powerful data
analysers rather than data gatherers and managers.
LESSONS LEARNED
Understand the value of your technical data, and recognise it
is the primary link to your single major asset (the resource).
Utilities functions (ERIC)

A collection of functions to aid different management areas; for
example management of core storage, lab coarse reject and pulp
sample storage, and database integrity checking. The spatial
tagging of wire-frames against samples allows the rapid
identification of samples and core for retrieval and further test
work.
Ensure management support for the project.

Establish an initial overall vision, but approach the project in
FUTURE DEVELOPMENT PLANS

PJV has recently upgraded its accounting and supply systems
management to the JD Edwards package, and the intention is to
interface with this package and similar non-technical databases.
Rover GPS systems have been proposed to download survey
and hole information in the field. This system is envisaged to use
a bar-code reader mounted to the GPS system being carried by a
geologist or geological technician. Scanning a tablet of options
can accurately enter hole sample number, lithology and sample
quality control data. Upon return to the office download of this
data would eliminate transcription errors and delays. Sample
dispatch forms can be automated from the sample numbers for
transfer to the assay laboratory system.
32
a modular manner, using the successes and lessons of each

stage to improve the system development.
Make sure the project remains focussed on the practical
requirements of the project, on developing systems that are
of real use and value to the production and management staff.
Continue to investigate new technologies, but make an
objective evaluation. Only embrace those that have a
demonstrable practical application to your operation; dont
fall into the trap of using technology for technologies sake.
Dont reinvent the wheel by trying to build into the database
functionality that is better (and less expensively) handled by
third party application packages. Instead, build interfaces to
allow these packages to access the data they need directly
from your database.
Chose a serious database package, and serious hardware to
allow performance, reliability, security and room to grow,
because it will! In reality, the price of the DBMS software
and the hardware will be only a small fraction of the total
cost of the development, implementation, data transfer and
training, so the choice of good tools is not going to blow the
project budget.
Invest in training for staff to allow them to use these
resources to their true capacity.

Ensure that the projects are managed in the way any other
serious project would be managed; make sure you use people
with the relevant training and experience for the projects. Too
many similar projects fail because they are simply thrown at
a junior mining engineer or geologist without access to the
real skill required.
CONCLUSIONS
The PJV Technical Database System has proved invaluable in
increasing the ability to utilise valuable technical data sets.
Located data that has been validated and cross-referenced with
other variables and can be extracted automatically or constrained
by user queries for various data types or areas to be modelled.
It allowed a degree of integration between the central database
and mining software, and a level of functionality that could not
be achieved in any other fashion.
The investment in the development of technical database

management at Porgera has seen real productivity gains within
the operation, firstly directly, in allowing technicians to perform
technical and management tasks better, and secondly resulting in
an optimisation in the way in which the resource is managed and
mined. This second benefit is the one that pays the big dividends,
far overshadowing the smaller savings in mine engineers or
geologists time.
The results have exceeded our expectations and allowed PJV
to get on with the job of finding and producing more ounces at
less cost from what must be considered one of the worlds top
gold deposits.
REFERENCES
Rock, N M S and Finlayson, E J, 1990. Petrological affinities of intrusive
rocks associated with the giant mesothermal gold deposit at Porgera,
Papua New Guinea, Journal of South Asian Earth Sciences,
4(3):247-257.
Fleming, A W, Handley, G A, Williams, K L, Hills A L and Corbett, G J,
1986. The Porgera Deposit, Papua New Guinea, Economic Geology,
81:660-680.
33
Sampling Practice at Vera Nancy Gold Mine

D Sims1
ABSTRACT
The efficient and profit-maximising daily and long-term management of
the Vera Nancy resource depends on decisions based upon assay data.
This paper examines the range of sample sources, the methods of sample
collection, assay techniques and Quality Assurance used by the mine.
The orebody has a relatively low nugget effect for a gold mine (around
25 per cent) and this greatly assists in reducing sampling error and
producing reliable assay data for production decision-making.
Reconciliation between mine and back-calculated mill grades generally
sees the mine under-calling mill grades by around four per cent.
INTRODUCTION
The Pajingo Joint Venture (50 per cent Battle Mountain Australia
Ltd and 50 per cent Normandy Mining Ltd) owns the Vera Nancy
underground gold mine located approximately 53 km SSE of
Charters Towers in North Queensland. Normandy Mining Ltd
manage the operation on behalf of the partnership. Vera Nancy is
one set of deposits in a precious metal epithermal field which
covers 150 square kilometres of sporadically outcropping
auriferous quartz veins hosted by Drummond Basin volcanics
(Porter, 1990).
The field was discovered by Battle Mountain Australia in 1983
with two deposits (Scott Lode which produced approximately
360 000 oz Au/1 000 000 oz Ag and Cindy which produced
approximately 46 000 oz Au/25 000 oz Ag) mined between 1987
- 1994. The Vera Nancy mineralised trend was recognised in
October 1994 and deep drilling in mid 1995 intersected
significant ore grade material. (Evans and Jones, 1997;
Richards et al, 1998).
By mid-1996 surface drilling on a 40 by 40 metre spacing had
delineated an Inferred Resource of 1.7 Mt at 14 g/t Au to a depth
of 400 metres over a 1.5 km strike length. This was considered
sufficient for the joint venture partners to proceed with
self-funded mine development and infill underground drilling in
advance of production at a scheduled rate of 220 000 t/yr to
produce 90 000 oz Au and 100 000 oz of Ag annually. A twin
decline access was developed along a 1.4 km strike length and
definition of the orebodies commenced with 20 by 20 metre
underground infill drilling and ore driving to establish bench
stopes. Full-scale production was reached in August 1997 at a
capital expenditure of $49.3 M (McKinstry, 1998). A review of
mine production rates during 1998-99 resulted in a mill
expansion completed in October 1999 which increased mill
capacity to 500 000 t/yr. At this higher rate the project will
produce in excess of 200 000 oz Au per annum. The mill is
located 4 km from the mine portal.
This paper refers to the on-going resource definition and grade
control sampling work undertaken on the Nancy, Vera and Nancy
North orebodies from March 1997. The methods and processes
used have been developed to suit the nature of the deposits, the
mining methods used and the objective use of the data collected.
DEPOSIT GEOLOGY
The auriferous quartz veins are hosted by andesitic volcanics
which are part of the Devonian - Carboniferous Drummond
Basin. The field contains deposits of the low-sulphidation
1.
MAusIMM, Senior Mine Geologist, Pajingo Joint Venture, PO Box

1271, Charters Towers Qld 4820.
epithermal type which have essentially been emplaced along

fault zones consisting of multiple fault planes. Economic
mineralisation occurs exclusively within the quartz veining
(Figure 1). Gold grade can drop from ounce-plus values in
quartz to 0.01 g/t Au in andesite across a knife-sharp vein
contact.
Being fault zones the structures which host the veins exhibit
bends, kinks and splays both along strike and down dip (Figure
2). Four deposits have been outlined along the NW - SE strike of
the structure ranging in the north from Nancy North, then Nancy,
then Vera and finally Vera South (Figure 3). Oxidisation extends
to around 70 m from surface and does not have a significant
effect on the orebodies although some zones of deeper
weathering occur within the upper parts of each deposit.
Mineralisation occurs dominantly as amalgams of gold and
silver (electrum approximately 1:1 ratio) occurring as fine
disseminated grains within quartz/carbonate veins. Precious
metal grain size averages around 40 microns with a range from
5 to 150 microns and is generally not visible to the naked eye.
The mineralised veins commonly contain traces of pyrite (<2 per
cent) and/or other rare sulphides as accessory minerals. Adularia
is absent while the main carbonates are ferroan in composition.
Alteration mineralogy in and around the quartz veins includes
silica, pyrite, chlorite, carbonate, sericite and haematite.
Movement on the faults during mineralisation allowed dilation
leading to depressurisation, mixing and boiling of gold/silver
bearing fluids within the structures. The veins contain abundant
epithermal textures such as colloform/crustiform banding,
brecciation and moss texture. Dark narrow bands within the veins
contain elevated fine grained pyrite content and generally
correspond to localised higher gold/silver grades. Some parts of
the veins are zones of brecciated mineralised vein material set in
a low-grade or barren silicified matrix.
Vein width ranges between 0.5 and 12 m and veins extend
around 300 m down dip and along a 2100 m strike-length. Where
multiple veins occur generally one vein contains the dominant
proportion of metal. Veins have moderate to steep dips (65 90) while widths can vary rapidly along strike with down dip
width continuity being generally more consistent. Second order
veins 10 to 20 cm wide commonly splay from the main
structures and can extend one to two metres into the footwall or
hangingwall (Figure 1a).
Faults that offset the deposits have been encountered in mine
development but are difficult to model from drilling alone.
Strike-slip faults with subhorizontal offsets have been intersected
on some levels and are steeply dipping structures subparallel to
the veins. These faults can cross cut the veins creating barren
pillars or overlapping the vein over short strike lengths. On other
development levels mining has exposed several subparallel
shears on or near the vein contacts which do not seem to directly
affect geometry but influence dilution during stope extraction,
shallowly dipping faults which strike perpendicularly to the
orebodies and steep reverse faults which post-date ore and create
vein overlaps in the eastern end of the Vera deposit. The ore
structures have been re-activated during later tectonic movement
complicating orebody geometry in parts of the system.
Where dip reversal occurs shallower dipping narrow ladder
veins can extend away from the Main Vein for a number of
metres. Drill intersections may overstate vein width in these
areas if the drill path follows along the ladder veins - it is
important to note core parallel veins in logging. Vein margins are
35
D SIMS
1A
1B
1C
1D
FIG 1 - Ore grade quartz vein appearance at Vera Nancy. (a) Nancy 971W drive cut 42. Left vein 1.5 m wide at 23.5 g/t Au, right veins 0.4 m wide at
21.8 g/t Au, (b) Vera 862W drive cut 3. Vein 1.9 m wide at 7.7 g/t Au, (c) Vera 862E drive cut 5. Vein 2.2 m wide at 36.1 g/t Au,
(d) Vera 862W drive cut 24. Vein 7.5 m wide at 12.7 g/t Au.
generally well-defined by infill drilling and in general ground

conditions are good.
Metallurgical performance in the CIP plant does not fluctuate
significantly between orebodies. Grinding to an 80 per cent
passing 40 micron particle size overcomes many liberation
problems associated with fine grained gold and recoveries
average around 96 per cent. Higher recoveries occurred when
oxidised ore was fed to the mill during the operation of the Vera
36
Pit mined on a small 5000 oz resource while establishing a portal

for the Vera Decline. Oxidised material is not present in the
existing resources. Silver recovery is generally poorer than gold
averaging around 80 per cent. Silver is less readily leached than
gold and so forms a residual coating or core on fine electrum
particles which are lost to tail. Also it may be possible that a
separate silver mineral species occurs and is not as well liberated
as electrum during grinding.
SAMPLING PRACTICE AT VERA NANCY GOLD MINE
FIG 2 - Mine isometric view showing the arrangement of mineralised fault planes with kinks and splays. Nancy North orebody at rear left,
Vera South orebody at front right.
PAJINGO JOINT VENTURE

INS ITU REMAINING RES OURCE
End o f June 1999
FIG 3 - Vera Nancy resource long section.
37
D SIMS
The Resources/Reserves for the deposit as at end June 1999

are as follows (note that Resources include Reserves):
Resources:
Measured
0.69 Mt
8.9 g/t Au
Indicated
1.36 Mt
18.7 g/t Au
Inferred
2.15 Mt
13.8 g/t Au
Total
4.20 Mt
14.6 g/t Au
Proved
0.53 Mt
12.5 g/t Au
Probable
1.59 Mt
13.9 g/t Au
Reserves:
Total
2.12 Mt
13.6 g/t Au
Classification of resources is based on the following criteria:
Measured
20 by 20 m drilling plus ore drives above and below

the stope block. Sublevel interval is 15 m floor to
floor.
Indicated
20 by 20 m drilling - underground drilling.
Inferred
40 by 40 m drilling - surface drilling.
The resource base is growing as exploration along strike
proves up additional resources most notably at Vera South (1.96
Mt at 14.1 g/t Au).
Critical issues relating to the nature of the orebody and mining
method which must be accommodated in sampling and data
collection methodology include:
the requirement that whole mineralised vein width is to be
mined in a way that maximises the grade produced - the
strong geological control on gold distribution confined within
the quartz vein necessitates the need for sharp boundary
definition in the sampling data, a task assisted by strong
visual control;
assay sample precision and accuracy must be maintained
throughout the project particularly around the ore cut-off
limit (approximately 5 g/t Au);
high spatial accuracy requirements due to the high-grade yet
narrow vein widths means that all locational data must be as
accurate as possible; and
dynamic updates of data and interpretation/modelling are
required as the underground drilling and development
programs progress and mine planning proceeds - data
collection and transfer needs to be as streamlined and as
focussed as possible.
TYPES OF SAMPLING DATA

Sampling for resource modelling/grade control occurs in
dominantly four ways:
1.
on surface diamond drill holes ( RC holes) during Inferred

Resource stage drilling and on underground diamond
drilling when infilling to Indicated and Measured Resource
status;
2.
on mining faces during ore drive development (chip

samples);
3.
on underground percussion samples and small diameter

core used to tightly infill splay vein positions adjacent to
ore drives (sludge or LM22 samples); and
4.
on truck loads when the ore is transported from

underground drives and stopes and tipped at mine stockpile
sites (grab samples).
Each of these sample types in a production environment is a
blend of theoretical best practice and operational practicality.
Other sampling undertaken in the mill for metallurgical control
purposes will not be discussed in detail. The plant uses an
38
automated Ramsey sampler which samples the leach feed stream

from the main thickener at five minute increments to determine a
feed grade over a 12-hour period. The sampler cuts the entire
stream and produces around 4 kg of sample per shift which is
pulverised and fire assayed using the same commercial
laboratory as the mine.
Drill hole sampling

The drilling dataset includes surface RC drill holes (255 holes for
28.5 km), surface diamond drill holes with RC pre-collars
(247 holes for 85.8 km) and underground diamond drill holes
(825 holes for 105.2 km). A total of 80 km of infill underground
core is planned to be drilled on the Vera South deposit.
Early RC holes used a cross-over sub while later holes used
face sampling hammers. Samples on one metre intervals were
riffle split at the rig for a 3 - 5 kg sample. For parts of holes
which were below the water table the entire sample was collected
and dried then crushed and riffle split for assay. RC chip sample
trays are retained by the exploration office.
Smearing of grades in RC holes from the upper part of the
Vera orebody above 160 m depth has become evident from
mining drives in these areas in turn leading to a significant drop
in contained ounces in this part of the orebody. In these areas
infill underground diamond drilling has been undertaken and this
data, along with drive information, overrides RC data in
modelling. Twinning of RC holes with diamond holes or
submission of the B split of RC samples was not undertaken in
the program.
Core sizes for the surface drill holes are dominantly HQ but
include NQ2 in recent drilling where it has shown to be
statistically as reliable as the larger size in testing these
orebodies. The underground core is drilled in conventional
LTK60 size which delivers approximately NQ sized core from a
BQ sized hole. Core recovery is generally very good with only
minor loss in strong shear zones.
Drilling orientation is on sections orthogonal to the
Vera-Nancy mine grid which is oriented parallel to the strike of
the veins (127 magnetic). Local vein flexures can see the vein
trend/drill hole intersection angle vary occasionally by 10 - 20.
Some fans of underground holes are drilled skew to the vein
trend due to drill access constraints around the lateral margins
and lower fringe of the deposit.
All underground holes are Eastman single shot surveyed every
30 m after an initial shot at 12 m depth. Surface holes were
commonly not initially surveyed until around 60 m depth in the
pre-collar stage and then every 60 m thereafter. Up until 1999,
when non-magnetic RC rods were introduced, only a dip reading
was possible in the RC portion of the hole. Although the ground
exhibits no discernible magnetic effects the position of surface
holes intersected in underground development indicates that
surface hole location can be out by up to 3 m at the orebody
intersection depth. This is thought to be related to deviation in
the largely unsurveyed pre-collar and accuracy related to the
60 m survey spacing interval.
Recent surface drilling is surveyed at 30 m intervals with
bearing and dip measured in the pre-collar stage. Surface collar
locations are surveyed in while underground collars have been
measured relative to set-up pins at the drill sites which were
located by the mine surveyor. Geological modelling uses the
grade data from surface holes but not the locational information.
All underground diamond drillcore is jig-sawed (reconstituted)
into continuous lengths 5 m either side of and through ore zones
prior to logging and sample preparation. A continuous reference
line is drawn on the core parallel to the core axis once the core is
jig-sawed to provide a uniform cutting line for sawing and
systematic sampling once logging is complete and the sampling
intervals are determined.
Core is logged and sampled on geological boundaries which

reflect the sharp grade distribution changes. Core is sawn into
two halves after being digitally photographed with the right hand
half assayed and the other retained in core storage. Digital photos
are saved as JPG files and stored in a UNIX directory on SGI
workstations.
Maximum sample length is limited to 1.5 m to ensure sample
sizes will allow complete pulverisation of the assay sample
material at the laboratory. Split intervals less than 0.3 m are
generally whole core assayed.
All remaining half core from ore zones, whole core from
hangingwall and footwall zones and representative intersections
of host andesite are retained in storage racks at the mine core
farm. Geological logging of underground drillcore uses an
Australian barcode-based logging system named DATCOL
which delivers a validated ASCII file which is directly loadable
via scripts into the Vulcan drill hole database. Digital core photos
are viewable in Vulcan during interpretation by clicking on the
displayed hole trace at the desired location which brings up the
relevant photo in a window on the screen.
Percussion sludge sampling

Sludge samples drilled with underground percussion rigs
(jumbos or blasthole rigs) are collected to assist in vein boundary
definition and to further infill location and grade information
particularly for flanking veins beside existing drives. Hole sizes
range from 45 mm to 68 mm and samples are collected via a
catcher which attaches to the rig delivering cuttings to a bucket.
Samples consist of 0.5 - 3 mm cuttings are collected over a 1 m
interval and the entire 3 - 5 kg sample is submitted for assay.
Chips are logged for lithology which is used to check assay data
qualitatively. Sludge holes are also entered into Vulcan.
LM22 core drilling
Development chip samples

Every mining face in the underground ore drives is geologically
mapped and sampled with the boundary and direction of the next
cut marked out. The sampling forms the primary grade control
for ore heading development. Drives are mined on approximately
15 m levels and are 4.5 m high leaving a 10 - 11 m high stope
bench between drives. Bench stopes are drilled and fired with
parallel blastholes once the ore drive sampling and mapping has
defined the ore geometry for final blast design. Overall around
60 per cent of production comes from bench stopes with the
remainder coming from the drive mining to establish the bench
stopes.
A minimum mining width of 3.5 m applies in ore drives due to
equipment requirements although veins down to 0.5 m wide can
be economically recovered using the resue mining method. This
is where a 2.5 m wide waste cut is mined to the footwall of the
vein and the ore is stripped into the cut once the waste is
removed. Given that the Nancy orebody has an average
horizontal width of around 2.6 m, control of drive development is
important to minimise hangingwall over-break in stopes due to
undercutting. Drive hangingwalls are optimally kept to a shanty
profile around 20 cm off the vein hangingwall.
The Vera orebody has higher widths (up to 16 m) where
multiple veins can be mined together. In these instances the
footwall of the vein system is driven with a 5 - 6 m wide drive,
the hangingwall is then sludged or drilled with a truck mounted
LM22 rig on 10 m centres to establish the exact hangingwall
position which is then stripped to full vein width. Drive widths
over 8 m are cablebolted.
As with drill holes the drive faces are subdivided into sampling
intervals based on geology. Chip samples are taken across the
interval at around waste height using a geology hammer and a
triangular sampling hoop to hold the sample bag and catch the
sample. Around 2 - 3 kg of material is collected per sample with
intervals around 0.4 - 1.5 m long.
Face location is measured relative to survey grade control pegs
and face geology is sketched into a pro-forma mapping sheet
which shows vein and fault/shear geometries and sample
locations. Mapping sheets are later sketched onto mine
cross-sections to aid interpretation and hand digitised into Vulcan
for stope geology wireframing. Occasional faces are digitally
photographed to assist with mapping. Sampling data is plotted at
1:125 scale and converted into drill hole data for display and
compositing in Vulcan with hole collar and survey data used to
depict the sample location in 3D space.
Once the ore drive is completed the backs are washed and the
vein contacts marked in paint for 3D pick-up as string data by
mine surveyors. This method was developed at MIMs McArthur
River mine by Geoff Phillips. The strings are loaded into
VULCAN to assist in wireframe construction and vein and
structure back mapping is undertaken on a hardcopy of the
pick-up outlines.
Work undertaken in 1998 by mine geologist Allan Bell assessed

the validity of sludge samples to properly represent veins which
were later driven or stripped. He found sludging poorly
represented veins commonly overstating both width and grade.
As a result discussions were undertaken with local drilling
contractor Rowe Enterprises (now Boart Longyear Rowe
Enterprises) to source a Toyota mounted light core rig which
could replace sludge holes with core. A reconditioned LM22 rig
was mounted on a trayback vehicle and commissioned drilling
LTK48 core. Holes range up to 30 m in length and the drill hole
collars are surveyed to allow hole entry into the Vulcan drill hole
database. Holes are logged and photographed as per usual but
whole core is assayed rather than half core. Increases in
interpretation reliability has occurred with the shift from
sludging to LM22 drilling. Drilling is commonly undertaken in
campaigns of 500 - 1000 m.
Truck grab samples

Grab samples are collected on surface stockpiles to calculate ore
classifications and mill feed grades for both development and
stope material. Material is not moved from the stockpile bays
until it is sampled and the results returned. Truck dumps are
either 30 or 21 tonnes (Bell B40/B30 trucks) and the days
production from each source is trucked to a separate stockpile
bay at the two mine portals for sampling. The grab assay results
are averaged per bay to determine the mean grade of the material
while the tonnage is factored by the number and size of the
dumps. Daily production ranges between 1400 - 2200 t of ore
and low-grade material.
A garden trowel is used to collect around 3.5 kg of material
from around and over each dump taking at least five increments.
Any particle which sits on the trowel is sampled. It is realised
that this method is significantly prone to sampling error and in
particular biases fine material. It is expected a brittle and
well-fragmenting quartz vein should lead to overestimating of
grab grades where host andesite may constitute coarse fragments
although in general grab samples underestimate the mill grade
(see below).
SAMPLE PREPARATION AND ASSAY

TECHNIQUES
Preparation and analysis is identical for all samples types. A
commercial laboratory located in Charters Towers is used for the
work.
39
D SIMS
PRECISION, ACCURACY AND THE NUGGET

EFFECT
Ag repeats
Quality Assurance focuses on assessing assay precision using

repeatability and accuracy using standards.
Repeatability is a function of preparation and assay process
quality and how evenly the gold is distributed in the ore (nugget
effect). Vera Nancy ore contains very fine grained gold and
pulverisation homogenises the gold throughout the sample
reasonably well. Plots of repeat gold analysis data for core, chip
and grab samples and repeat silver analysis data for core samples
are shown in Figure 4. Statistics for this data are contained in
Table 1.
Precision for gold assays is within industry standards and
certainly acceptable although repeatability for grades <0.04 g/t
Au is understandably poor. The high correlation coefficients and
tight plots indicate that Vera Nancy is a relatively low nugget
gold orebody. Variography undertaken for grade estimation
indicates a nugget in the order of 25 per cent. The low nugget
factor greatly assists in reducing sampling error and producing
reliable assay data for production decision-making.
Au repeat
Au repeat
Au repeat
Preparation involves jaw crushing to reduce to -25 mm then

pulverisation of the entire sample to -75 microns by LM5 mill. If
any samples are greater than around 3.5 kg then the sample is
jaw crushed and riffle split before pulverisation. A 50 g charge of
material is assayed for gold by fire assay methods (ALS Chemex
method PM209) and for drillcore samples a 200 g pulp is
retained and returned to site. One in ten gold assays are repeated
by taking a second 50 g charge. Silver is assayed in core samples
only by an AAS method (ALS Chemex method G001) with
repeats in higher grade samples (>50 ppm) undertaken at a
different dilution (ALS Chemex method A101).
Samples are delivered at the end of dayshift/nightshift and
results are returned within 15 hours for chip and grab samples
while core and sludge samples return in two to four days.
In-house standards are submitted with every drillcore batch
with a series of 12 standards prepared from Vera Nancy ores over
a range of grades. The values of the standards were determined
by analysis by eight separate laboratories. Standards are given a
coded identifier at submission for disguise.
FIG 4 - Assay precision repeat plots.
40
TABLE 1
Assay precision statistics.
Sample type
Chips
FA1
Element
Grabs
FA2
FA1
Au
Number of duplicates
Core
FA2
FA1
Au
210
Core
FA2
FA1
Au
262
FA2
Ag
2140
202
Minimum
0.01
0.01
0.03
0.03
0.01
0.01
Maximum
219.0
205.0
69.9
69.1
450.0
476
400
430
Mean
12.82
12.70
7.50
7.47
9.08
9.17
39.27
44.87
Median
0.51
0.49
3.40
3.35
1.56
1.56
2
74.33
Std deviation
29.92
29.18
11.38
11.35
22.08
22.52
65.99
Variance
895.21
851.47
129.50
128.82
487.53
507.15
4354
0.999
0.999
0.998
5524
0.997
% difference, between assay and

standard
Correlation coeff
FIG 5 - Assay accuracy drift.
Precision on silver assays are less impressive largely because

of the fact that silver repeats are generally only undertaken on
high-grade samples (>50.0 g/t Ag) using a different
analytical/digestion method which gives higher values. Although
silver is a valuable by-product it is not as critical as gold to the
profitability of the operation hence poorer precision with the data
is acceptable.
Assessment of accuracy is undertaken by the submission of
two standards with each drill core sample dispatch (generally a
group of 20 - 30 samples). The results from each standard is
inspected by the relevant geologist when hard copy preliminary
assay results are faxed from the laboratory. The values returned
need to be less than 10 per cent of the standard value for the
batch to be accepted. If values indicate that multiple batches are
returning with accuracy poorer than 5 per cent of the standard
value then the geological superintendent is notified who
discusses the problem with the laboratory manager. In general
accuracy is within 5 per cent with the laboratory most
commonly under-quoting the grade. A plot through 1999
indicates an accuracy drift whereby the value of returned assays
has been below real values in the early part of the year drifting
to a more accurate figure at the end of the year (Figure 5).

Discussions with the laboratory can be undertaken based on firm
data.
MINE TO MILL RECONCILLIATIONS

Four product types are produced from the Vera Nancy mine for
either mill feed or stockpile:
high-grade ore - greater than 20.0 g/t Au (milled);
ore - 5.0 to 20.0 g/t Au (milled);
low-grade - 1.0 to 5.0 g/t Au (stockpile at mill - to be milled
after mine closure); and
marginal - 0.3 to 1.0 g/t Au (stockpile at mine).
Reconciliations for mine ore production to the end of the
December 1999 indicated that the Reserve overstated the tonnage
by nine per cent and understated the grade by eight per cent
compared to production. Around 99 per cent of Reserve gold
ounces were produced. Production of low-grade material for the
same period saw the Reserve tonnage exceeded by 19 per cent
and the grade figures exceeded by 15 per cent with production
containing 137 per cent of Reserve contained ounces.
41
D SIMS
The mined grade for this period was 13.4 g/t Au compared to a
milled grade of 14.0 g/t Au. Although general assay accuracy
may be a contributing factor to the undercall on grade it is
believed that grade estimation methodology also has an
influence.
The Resource estimate interpolation method used to produce
the Reserve figures for last year utilised Inverse Distance
Weighting methods. Calibrating the global figure for the Nancy
orebody against Indicator Kriging (IK) methods proved
satisfactory although techniques which will produce a better
local estimate (such as IK) in production situations are being
investigated.
CONCLUSION
Vera Nancy is a high-grade mine which requires accurate mining
to maintain managements objective of maximising the head
grade. Quality sampling information is a significant contributor
to achieving this aim in mine management and is strongly
assisted by good visual control on grade distribution and a low
nugget factor in the ore.
42
Mining at Vera Nancy is to undergo an expansion program

which will see production raised to around 500 000 t/yr to
produce in excess of 200 000 oz per year. Continued use of
sound and practical sampling processes will be required to
manage the operation into the future.
REFERENCES
Evans, R C and Jones, B H, 1997. The Discovery and evaluation of the
Vera-Nancy deposit, North Queensland, in Proceedings World Gold
97 Conference, pp 233 - 237 (The Australasian Institute of Mining
and Metallurgy: Melbourne).
McKinstry, J, 1998. The methodology behind the development of the
Vera Nancy mine, in Proceedings Seventh Underground Operators
Conference, pp 217 - 219 (The Australasian Institute of Mining and
Metallurgy: Melbourne).
Porter, R G, 1990. Pajingo Gold Deposits, in Geology of the Mineral
Deposits of Australia and Papua New Guinea (Ed: F E Hughes), pp
1483 - 1487 (The Australasian Institute of Mining and Metallurgy:
Melbourne).
Richards, D R, Elliott, G J and Jones B H, 1998. Vera North and Nancy
gold deposits, Pajingo, in Geology of Australian and Papua New
Guinean Mineral Deposits (Eds: D A Berkman and D H Mackenzie),
pp 685-690 (The Australasian Institute of Mining and Metallurgy:
Melbourne).
Leachwell* Versus Fire Assay: Comments from Assay Quality

Assurance at Tarmoola Gold Mine
D J Elder1
ABSTRACT
A fundamental component of grade control at Tarmoola Mine has been
the monitoring of assaying through quality assurance programs. The
programs have concentrated on issues related to accuracy and precision
of leachwell and fire assay techniques, sample preparation and grade
variability within a sample. An outcome of monitoring has been
identification of shortcomings in the application of assaying techniques
to Tarmoola ore types that contain coarse particulate gold.
Leachwell has shown satisfactory precision since introduction as the
primary assay technique in October 1998 but leachwell accuracy and
recovery through comparison with fire assay techniques has identified
bias. A bias toward higher fire assays in grades above 2 g/t invoked an
investigation into the cause of the discrepancy between leachwell and fire
assay.
Testwork evaluated Leachwell assay recovery by varying leach time
duration, which showed that by increasing the leach time improved
recovery significantly. Monitoring of leachwell against fire assay had
shown that a leachwell (200 g) with a one hour leach has under-called
grade when compared with fire assay by up to 18 per cent, on average,
for Tarmoola ore types. Alternatively, optimisation of the leachwell
assaying technique using a 200 g pulp and a two hour leach has returned,
on average, higher assays than fire assay by up to eight per cent.
Testwork on screen fire assays, duplicate assays and fire assay to
extinction has demonstrated assay variability associated with coarse,
particulate gold. The variability and distribution of gold in Tarmoola ore
types has emphasised the importance of high quality sample preparation
for leachwell assaying to ensure liberation of gold to leaching solutions.
Assay quality assurance protocols are vital to assay data integrity and
enable confidence in delineation of ore-waste boundaries and provide
early recognition of problematic assaying periods and isolation of grade
control practices, which depart from performance standards critical to the
operation.
INTRODUCTION
Tarmoola Gold Mine is situated 280 km north of Kalgoorlie in
Western Australia within the Norseman-Wiluna Greenstone Belt
of the Yilgarn Craton. Gold mineralisation at Tarmoola is hosted
in quartz vein arrays and associated with intense
quartz-carbonate alteration within structurally complex shear
zones on a granitoid-greenstone contact. The mine produced 3.2
million tonnes of ore for 194 000 ounces in the 1998/1999
financial year.
The objective of grade control at Tarmoola is to delineate
optimised ore outlines and accurately predict tonnes and grade of
ROM ore to the mill. The collection of quality assay data is of
primary importance in achieving the objective and can only be
verified by assay monitoring through quality assurance
programs. By evaluating assay quality a level of confidence in
data collected can be ascertained.
Changes in grade control practices have occurred at Tarmoola
to optimise ore zone delineation, with a number of changes
resulting directly from assay quality assurance programs. In
October 1998, the aqua regia (30 g) assaying technique was
TM
Leachwell in this paper refers to Leachwell

accelerated cyanide extraction (ACE).
1.
Mine Geologist, PacMin Tarmoola Operations Pty Ltd, PO Box 67,

Leonora WA 6438.
60x catalyst, an
replaced by leachwell (200 g). Testwork undertaken during 1998

showed leachwell provided accurate and precise results, and
reduced sample variability was an expected benefit of assaying a
larger sample.
In response to grade control estimates under-calling actual mill
head grades the leachwell assaying technique (200 g, one hour
leach) was replaced by fire assaying (40 g) technique in
September 1999. The change from leachwell to fire assay was
supported by quality assurance testwork that indicated leachwell
was under-calling grade with respect to fire assay.
Assay quality assurance for the period October 1998 to
January 2000 has analysed leachwell and fire assay precision,
leachwell accuracy through comparison with fire assay, gold
particle size and distribution in Tarmoola ore types, leach time
duration and sample preparation.
Assay quality assurance programs at Tarmoola have been
fundamental to further understanding assaying issues critical to
grade estimation, reconciliation and ore zone delineation. It has
highlighted the importance of completing assay quality assurance
by identifying key issues related to assaying and has resulted in
optimisation of leachwell assaying techniques for Tarmoola ore
types.
ASSAY PRECISION
Leachwell
Leachwell has been the primary assay technique used between
October 1998 and September 1999. The leachwell assaying
technique involves weighing of 200 g of pulverised sample (d75
microns) into a one litre, plastic screw capped container with the
addition of 400 ml of water (tap temperature) and one leachwell
assay tablet. Containers are capped and placed in a tumbler
revolving at approximately 15 rpm for one hour. The containers
are removed from the tumbler and allowed to settle. A 10 ml
aliquot of clear, cyanide liquor is removed from each and placed
into a 20 ml culture tube. The gold concentration is then
determined using atomic absorption spectrometry (AAS) in
conjunction with appropriate standards. Testwork in October
1999 has resulted in the leach time being extended to a two hour
leach duration with a starting water temperature of 30 degrees
Celsius.
Repeat analyses are performed routinely by the principal assay
contractor on ten per cent of assays submitted. Leachwell repeat
data for the period October 1998 and September 1999 showed a
moderate to tight scatter over 0 - 10 g/t grade range with no bias
observed (Figures 1 and 2). During this period a precision of
0.19 has indicated repeatable assays from the same pulp through
all grade ranges.
Fire assay
Fire assay replaced leachwell in September 1999 as the primary
assay technique. Analysis of assay repeat data for the period
September 1999 to January 2000 showed a moderate scatter with
precision of 0.24 with no bias observed (Figures 3 and 4). Fire
assay has shown a wider scatter than leachwell indicating
leachwell to be a more precise method. Both assaying techniques
have produced unbiased repeat results but comparison of fire
43
LW2
LW2
D J ELDER
FIG 2 - QQ plot of LW1 versus LW2 for the period October 1998 to
September 1999 (n=3827).
FA2
FA2
FIG 1 - Scatter plot of LW1 versus LW2 for the period October 1998 to
September 1999 (n=3827).
FIG 3 - Scatter plot of FA1 versus FA2 for the period September 1999 to
January 2000 (n=1092).
FIG 4 - QQ plot of FA1 versus FA2 for the period September 1999 to
January 2000 (n=1092).
assay against leachwell has shown a bias toward higher fire

assays indicating accuracy or recovery is a limiting factor in the
leachwell assay technique.
duplicate fire assay results have identified higher gold grades

than leachwell. Scatter and quantile-quantile (qq) plots for the
period indicated that fire assay determined, through the 0 - 10 g/t
grade range five per cent more gold than leachwell where a bias
is observed in grades above 2 g/t (Figures 5 and 6). Analysis of
the 2 - 10 g/t grade range showed fire assays determined ten per
cent more gold than leachwell in grades above 2 g/t. These
results have shown that with increasing gold concentration there
is an increasing bias toward higher fire assays over leachwell.
ASSAY ACCURACY: COMPARISON OF

LEACHWELL WITH FIRE ASSAY
October 1998 to August 1999
With the introduction of leachwell (200 g, one hour leach) in
October 1998, assay quality assurance programs were introduced
to test leachwell accuracy by repeat assaying pulverised sample
using the fire assay technique. Pulps were recalled from the
principal assay contractor and submitted to independent
laboratories for duplicate fire assaying and screen testing to
analyse grinding efficiencies. Duplicate fire assays on pulps were
employed to provide averaged fire assay results to minimise
scatter commonly associated with coarse, particulate gold.
Results for the period October 1998 to August 1999 indicated
44
August 1999 to October 1999

In August 1999, the assay quality assurance program was revised
and the principal assay contractor was instructed to complete fire
assay repeats on ten per cent of leachwell assays grading above
0.5 g/t and to complete a fire assay on the corresponding
leachwell sample residue. The change from recalling batches of
pulps for independent analysis to in-house quality assurance was
initiated to provide results allowing fortnightly analysis of
leachwell accuracy.
FA1
FA1
LEACHWELL VERSUS FIRE ASSAY
FIG 6 - QQ plot of LW1 versus FA1 for the period October 1998 to
August 1999 (n=758).
FA1
FA1
FIG 5 - Scatter plot of LW1 versus FA1 for the period October 1998 to
August 1999 (n= 758).
FIG 7 - Scatter plot of LW1 versus FA1 for the period August 1999 to
mid-October 1999 (n=208).
FIG 8 - QQ plot of LW1 versus FA1 for the period August 1999 to
mid-October 1999 (n=208).
Quality assurance monitoring for the period August 1999 to

mid-October 1999 compared leachwell (one hour leach) and fire
assay. A bias toward higher fire assay results was observed above
3 g/t. The average leachwell assay is 1.88 g/t and the average fire
assay is 2.07 g/t and fire assays were on average, nine per cent
higher (Figures 7 and 8).
The continued bias toward higher fire assays resulted in
leachwell being replaced by fire assay in September 1999 as the
primary assay technique. The quality assurance program
continued to monitor leachwell accuracy, with the assay
contractor completing leachwell repeats on ten per cent of fire
assays grading above 0.5 g/t and a fire assay on the
corresponding leachwell tail.
where the average fire assay is 2.62 g/t and the average leachwell
is 2.77 g/t. Leachwell assays are six per cent higher than fire
assays on average. The change from one hour to two hour leach
times had a significant effect on gold recovery and results have
indicated that leachwell has the ability to identify more gold than
fire assay.
October 1999 to January 2000

Analysis of leachwell and fire assay data for the period
mid-October 1999 to January 2000 compared leachwell (two
hour leach) and fire assay (Figures 10 and 11). The results
indicated a bias toward higher leachwell assays above 2 g/t,
Leaching duration testwork

Leachwell testwork in early-October 1999 had investigated one,
two and four hour leach times and starting water temperatures of
7C and 32C (Table 1). The results are combined with previous
screen fire assay results and original leachwell assays (performed
using a one hour leach and tap water temperature). The results
indicated that increases in leaching duration significantly
improved gold recovery. The effect water temperature had on
recovery is less clear but suggested better recoveries at a water
temperature of 32C. This testwork resulted in the duration of
leaching being extended to a two hour leach and installation of a
constant temperature water supply (30C) at the principal assay
contractors laboratory in mid-October 1999.
45
FA1
LW2h
D J ELDER
FIG 9 - QQ plot of LW1h versus LW2h for testwork completed in

October 1999 on leaching duration time (n=87).
FIG 10 - Scatter plot of LW1 versus FA1 for the period mid-October
1999 to January 2000 (n=118).
TABLE 1
Summary of leachwell leaching duration testwork evaluating variations in bottle roll time and effects of water temperature when added
to leaching solution.
7 degress Celcius
32 degress Celcius
Sample no
Screen FA
Orig LW1
1 Hr
2 Hr
4 Hr
Tail
1 Hr
2 Hr
4 Hr
Tail
252.48
133.00
79.10
129.00
187.00
33.30
137.00
138.00
196
38.2
79.59
76.20
40.60
63.70
78.80
0.77
40.70
57.40
70.2
1.72
28.17
27.50
24.80
27.70
28.90
0.11
28.70
30.70
31.1
0.18
10.85
7.69
7.38
14.10
15.10
0.35
9.77
10.40
10.4
0.19
18.71
17.20
16.30
19.60
21.20
0.20
14.10
15.50
16.2
0.21
9.06
9.19
9.93
11.50
12.30
0.14
9.41
9.12
9.46
0.08
6.67
5.80
5.04
7.41
7.98
0.07
5.72
7.05
7.15
0.07
13.86
11.70
7.80
11.80
14.30
0.96
10.40
11.10
11.9
0.83
8.46
7.00
4.34
7.01
7.90
0.20
6.97
7.56
7.78
0.18
10
20.19
15.80
8.24
9.07
9.17
0.16
7.72
7.79
8.02
0.15
Further evaluation of gold recovery and leaching duration

involved leachwell assaying of 100 pulps using a one and a two
hour leach times. The one hour leach produced a population
averaging 1.10 g/t and the two hour leach produced a sample
population averaging 1.25 g/t. The results showed the two hour
leach produced leachwell assays 12 per cent higher than one hour
leaches from the same pulverised sample (Figure 9). The results
of this testwork have shown that assay results are highly
susceptible to error in recovery or accuracy even though
precision for each variation is still good. Fire assays of the same
data set produced a sample population averaging 1.27 g/t,
slightly higher than the leachwell two hour average of 1.25 g/t.
FA1
Fire assay of leachwell tails

The leachwell assaying technique is a partial digest technique
therefore evaluation of gold not recovered in the leachwell
assaying process is an important consideration. Fire assay of
leachwell tails data (FAT) has become a routine part of assay
quality assurance. Comparison of fire assay of leachwell tails
with corresponding original leachwell assays from periodic
testwork between October 1998 and August 1999 indicated
approximately two to four per cent gold was not recovered from
46
FIG 11 - QQ plot of LW1 versus FA1 for the period mid-October 1999 to
January 2000 (n=118).
FAT1
leachwell assays. Testwork for the period between August 1999

to January 2000 indicated that on average three per cent of gold
is not recovered by the leachwell assaying technique (Figure 12).
No significant change was observed in the data where leaching
duration was extended to a two hour leach time.
Screen Fire Assays

Screen fire assay testwork has been used to identify the
distribution of gold in size fractions of ten, high-grade samples
from granitoid-greenstone contact mineralisation (Table 3). The
results have identified significant percentages of gold are
distributed in size fractions greater than 75 microns. At these
particle sizes it is believed that occlusion of gold from leaching
solutions has a considerable impact on gold recovery. This
emphasises the importance of high grind efficiencies in sample
preparation to liberate all gold particles prior to partial digest
assaying. These resistant particles containing proportionally
more gold are most likely to be from zones of high-grade
mineralisation with intense silicification as observed in Tarmoola
ore types.
If the ten screen fire assays (presented in Table 3) are
considered to be one combined screen fire assay result then
averaged percentages of sample in each size fraction can be
plotted against averaged gold content for each size fraction
(Figure 13). The relationship between gold content and size
fraction has indicated that coarser size fractions (>75 microns)
represented, on average 21 per cent of the sample and contained,
on average 51 per cent of the gold.
FIG 12 - Scatter plot of LW1 versus FAT1 for the period August 1999 to
January 2000 (n=332).
SIZE AND DISTRIBUTION OF GOLD IN

TARMOOLA ORE TYPES
Gold particle size
To evaluate gold particle size a coarse gold sample was collected
from the gravity circuit of the Tarmoola Mill in August 1999.
The gravity gold was passed through a series of screens and
results gave an indication of gold particle size in Tarmoola ore
types (Table 2). While gold particles collected for this testwork
would have experienced deformation and flattening through the
milling circuit the results are thought to be indicative of gold
particle size. Gravity gold collected in this part of the milling
circuit was very coarse with 85 per cent gold being greater than
75 microns. Up to 45 per cent of gold in the milling process can
be recovered in the gravity circuit indicating the coarse nature of
gold particles in ore types at Tarmoola. The abundance of coarse
gold is consistent with difficulties associated with assaying
particulate gold mineralisation at Tarmoola.
TABLE 2
Summary of 200 g sample of gravity gold taken from gravity
circuit at Tarmoola Mill to investigate gold particle size.
Size (microns)
Weight (g)
Weight (%)
1000 - 2000
500 - 1000
300 - 500
16
150 - 300
67
31
75 - 150
91
42
<75
33
15
Totals
216
100
FIG 13 - Line graph showing the relationship between averaged

percentage of sample in each size fraction (diamonds) and the averaged
percentage of gold from each size fraction (squares).
The results indicating distribution of gold in high-grade

samples are significant when using a partial digest technique as
failure to liberate all gold particles to leaching solution will result
in an under-call of true grade of the assay sample. The occlusion
of gold from leaching solutions is considered to be a significant
issue in assaying of Tarmoola ore types arising from high
degrees of silicification in mineralised zones. Considerable
attention is paid to sample preparation protocols, to ensure
liberation of all gold particles to leaching solutions.
Duplicates
An extensive duplicate program was undertaken on
granitoid-greenstone contact mineralisation in May 1999. A total
of 403 duplicate field split samples were collected from a Jones
Riffle splitter mounted on a grade control RC drilling rig. The
assaying technique for this testwork was leachwell. A wide
scatter was observed in the scatter plot (Figure 14) with a
precision of 0.57 and a bias toward higher duplicate assays was
noted. The results indicated a poor correlation between the
leachwell original (submitted for normal grade control) and the
duplicate leachwell assay and has demonstrated the grade
variability in a sample. The grade variability has been attributed
to the coarse particulate nature of gold mineralisation at
Tarmoola.
47
D J ELDER
TABLE 3
Summary of screen fire assay testwork on ten samples from high-grade granitoid-greenstone contact mineralisation.
Sample no and sizing
(microns)
Fire assay
ppm
% Retained from
sizing
Calculated Head
Grade
>150
3520.00
2.08
73.33
29.05
106-150
1162.50
3.64
42.27
16.74
75-106
423.00
7.84
33.17
13.14
53-75
297.50
5.87
17.46
6.92
38-53
286.00
5.18
14.81
5.87
<38
94.75
75.39
71.43
28.29
100.00
252.48
Total Sample 1
133.00
% Gold in size
fraction
100.00
>150
2524.00
1.05
26.49
33.29
106-150
751.00
1.09
8.15
10.24
75-106
243.00
4.62
11.22
14.09
53-75
111.50
7.00
7.81
9.81
38-53
93.90
5.10
4.79
6.01
<38
26.05
81.15
21.14
100.00
79.59
Total Sample 2
26.56
76.20
100.00
>150
159.00
5.65
8.99
31.91
106-150
89.30
6.07
5.42
19.25
75-106
66.45
8.94
5.94
21.10
53-75
41.80
7.63
3.19
11.32
38-53
29.40
5.35
1.57
5.58
<38
4.60
66.35
3.05
10.83
100.00
28.17
24.00
3.29
Total Sample 3
27.50
100.00
>150
13.70
106-150
12.90
8.33
1.07
9.90
75-106
16.05
12.00
1.93
17.76
53-75
12.05
6.18
0.74
6.86
38-53
13.85
5.87
0.81
7.50
<38
6.88
43.62
3.00
27.67
Total Sample 4
30.31
100.00
10.85
>150
43.75
11.51
5.04
26.91
106-150
33.05
9.35
3.09
16.52
75-106
32.45
13.27
4.31
23.02
53-75
30.20
6.94
2.09
11.20
38-53
16.85
5.93
1.00
5.34
<38
6.01
53.00
3.18
17.01
Total Sample 5
7.69
17.20
100.00
100.00
18.71
>150
42.80
0.34
0.15
1.62
106-150
48.70
1.79
0.87
9.64
100.00
75-106
26.05
9.14
2.38
26.27
53-75
11.60
8.20
0.95
10.50
38-53
13.60
8.38
1.14
12.57
<38
4.95
72.14
3.57
100.00
9.06
2.59
0.60
8.99
Total Sample 6
39.40
9.19
100.00
>150
23.15
106-150
21.55
4.49
0.97
14.49
75-106
8.88
10.07
0.89
13.41
53-75
8.08
8.87
0.72
10.74
38-53
8.00
8.69
0.69
10.41
<38
4.29
65.29
2.80
41.97
100.00
6.67
Total Sample 7
48
Original LW1
5.80
100.00
TABLE 3 (continued)
Summary of screen fire assay testwork on ten samples from high-grade granitoid-greenstone contact mineralisation.
Sample no and sizing
(microns)
Fire assay
ppm
% Retained from
sizing
Calculated Head
Grade
Original LW1
% Gold in size
fraction
>150
18.05
8.87
1.60
11.55
106-150
15.10
8.03
1.21
8.75
75-106
13.65
10.74
1.47
10.58
53-75
12.10
9.43
1.14
8.23
38-53
12.65
6.57
0.83
5.99
<38
13.50
56.36
7.61
54.89
100.00
13.86
Total Sample 8
11.70
100.00
>150
31.80
1.22
0.39
106-150
9.96
3.50
0.35
4.11
75-106
11.95
9.92
1.19
14.01
53-75
12.30
8.28
1.02
12.03
38-53
9.73
7.03
0.68
8.08
<38
6.91
Total Sample 9
70.06
4.84
100.00
8.46
4.60
57.17
7.00
100.00
>150
95.60
8.03
7.68
38.03
106-150
27.75
9.19
2.55
12.63
75-106
26.10
12.88
3.36
16.65
53-75
15.05
8.17
1.23
6.09
38-53
14.85
6.12
0.91
4.50
<38
8.02
55.61
4.46
22.09
100.00
20.19
Total Sample 10
15.80
100.00
LWdup
the sample sets (Figure 15). For example, sample 7 has a range
of 1.90 to 3.44 g/t from 37 assays. The average for this sample
was 2.59 g/t which is 22 per cent higher than the original fire
assay of 2.02 g/t, initially reported and used in routine grade
control. These results typify some of the inherent problems faced
in providing accurate grade control predictions for reconciliation
against actual, mill head grades.
DISCUSSION
FIG 14 - Scatter plot of LW versus LW duplicate field split, May 1999

(n=377).
Fire assaying to extinction

Fire assaying testwork designed to evaluate variability of gold
grade in pulverised sample was completed in January 2000. The
testwork evaluated seven samples of between 1.5 and 2 kg of
pulverised material. Each sample was repeat fire assayed to
extinction (Table 4) and provided between 36 to 47 assay results
for each sample. Results show high variability in grade ranges in
The introduction of leachwell as the primary assaying technique

in October 1998 was seen as a progressive step in providing
quality assay data for grade control. The improvements in assay
precision achieved by assaying a larger proportion of sample
were considered beneficial for assaying Tarmoola ore types by
reducing sampling error.
Periodic testwork during the period October 1998 to May 1999
was completed to ensure the assay quality of leachwell
technique. Testwork indicated a bias toward higher fire assays
over leachwell in grades above 2.0 g/t, which was initially
attributed to gold not recovered from the leachwell sample
during leaching. The bias indicated leachwell assays were five
per cent lower than fire assay on average. Fire assay testwork on
leachwell tails indicated on average, two to four per cent gold
was not recovered by leachwell and was accepted as a
consequence of the partial digest technique.
Grade reconciliations between grade control predicted
estimates and mill actual head grades were acceptable between
October 1998 to May 1999, with variance of 10 per cent. Assay
quality assurance indicated acceptable assay accuracy and
precision through critical cut-off grade ranges (0 - 1.0 g/t), so
leachwell was considered to be a satisfactory assaying technique
during this period.
49
D J ELDER
TABLE 4
Summary of seven samples fire assayed to extinction to evaluate heterogeneity in Tarmoola ore types.
50
Assay no
1.12
0.10
0.72
0.13
0.16
1.00
1.90
1.13
0.12
0.72
0.14
0.23
1.03
1.90
1.51
0.22
0.82
0.14
0.24
1.04
2.01
1.53
0.28
0.85
0.16
0.27
1.04
2.05
1.56
0.29
0.85
0.16
0.29
1.04
2.11
1.68
0.29
0.86
0.16
0.37
1.04
2.12
1.69
0.30
0.88
0.17
0.44
1.05
2.15
1.69
0.31
0.89
0.17
0.52
1.05
2.15
1.78
0.32
0.89
0.17
0.53
1.06
2.23
10
1.78
0.34
0.90
0.18
0.53
1.06
2.26
11
1.78
0.34
0.92
0.18
0.53
1.06
2.34
12
1.82
0.35
0.93
0.18
0.53
1.09
2.35
13
1.85
0.35
0.95
0.19
0.53
1.10
2.36
14
1.90
0.35
0.96
0.19
0.54
1.10
2.41
15
1.90
0.35
0.98
0.19
0.54
1.11
2.44
16
1.91
0.36
0.98
0.21
0.54
1.13
2.47
17
1.93
0.36
1.00
0.21
0.54
1.13
2.47
18
1.93
0.37
1.01
0.22
0.54
1.16
2.48
19
1.98
0.37
1.01
0.22
0.55
1.16
2.59
20
2.02
0.37
1.03
0.22
0.55
1.17
2.59
21
2.02
0.37
1.03
0.22
0.55
1.17
2.63
22
2.03
0.37
1.05
0.23
0.56
1.18
2.66
23
2.04
0.38
1.05
0.23
0.56
1.19
2.67
24
2.07
0.38
1.07
0.24
0.56
1.19
2.79
25
2.12
0.38
1.09
0.24
0.56
1.21
2.94
26
2.13
0.38
1.11
0.25
0.57
1.21
2.96
27
2.14
0.39
1.16
0.25
0.57
1.26
2.97
28
2.14
0.39
1.20
0.26
0.58
1.27
3.03
29
2.16
0.40
1.20
0.26
0.59
1.28
3.03
30
2.17
0.40
1.26
0.26
0.59
1.29
3.04
31
2.18
0.40
1.27
0.27
0.60
1.32
3.06
32
2.24
0.41
1.28
0.28
0.60
1.36
3.11
33
2.28
0.41
1.32
0.28
0.60
1.45
3.14
34
2.30
0.41
1.76
0.28
0.61
1.48
3.16
35
2.30
0.43
2.35
0.29
0.61
1.49
3.25
36
2.44
0.45
3.51
0.30
0.62
1.54
3.44
37
2.47
0.45
0.30
0.62
1.56
38
2.53
0.45
0.57
0.62
1.75
39
2.55
0.51
0.66
1.84
40
2.60
0.52
0.67
41
2.68
0.52
0.68
42
2.89
0.52
0.76
43
3.39
0.61
44
3.45
0.67
45
3.53
0.97
46
3.96
1.10
47
4.43
1.16
FA average
2.21
0.42
1.14
0.22
0.53
1.22
2.59
FA original
2.15
1.40
1.24
0.56
0.63
1.52
2.02
4.50
4.00
3.50
3.00
2.50
2.00
1.50
1.00
0.50
0.00
1 2 3 4 5 6 7 8 9 101112131415161718192021222324252627282930313233343536373839404142434445464748
As s ay numbe r
FIG 15 - Line graph showing assays from seven pulverised samples each fire assayed to extinction.
During July 1999, assay quality testwork results from grade

control drilling in May 1999 were received from an independent
laboratory. Independent fire assay testwork indicated a strong
bias toward higher fire assays over original leachwell assays
above 1.2 g/t. Fire assay identified 18 per cent more gold in this
testwork on average and shows increasing bias with increasing
grades (Figure 16). Independent sample screening indicated poor
sample preparation by the primary assay contractor where only
55 per cent of screens were 90 per cent passing 75 microns. The
results of the testwork anticipated a grade control under-call in
June 1999 when ore defined by substandard assaying in May
1999 was being milled. The testwork emphasises the relationship
between substandard sample preparation and lowered leachwell
recoveries. Grade control under-call was a reflection of
difficulties determining accurate grades of ore sourced from
zones of high-grade mineralisation using leachwell. The bias
between leachwell and fire assay has been observed in grades
above cut-off grades (ROM cut-off >0.5 g/t) so it was believed
that there was no misclassification of ore during the period of
mining.
A review of quality assurance protocols in August 1999
resulted in a change from periodic, independent testwork to
implementation of a routine quality assurance procedure. The
principal assay contractor was required to perform fire assay
FIG 16 - QQ plot of LW1 versus FA1 for accuracy testwork completed in

May 1999 (n=87).
51
D J ELDER
repeats on ten per cent of leachwell assays above 0.5 g/t. This
provided regular feedback of leachwell assay performance and
enabled a review of accuracy through all grade ranges, most
importantly evaluating higher grade ranges, which were
relatively un-tested in previous testwork.
As leachwell continued to under-call grade in comparison with
fire assay in quality assurance testwork, a change of assay
technique was prompted and fire assay replaced leachwell as the
primary assay technique in September 1999. Monitoring of
leachwell accuracy continued where ten per cent of fire assays
above 0.5 g/t were repeat assayed using leachwell. The bias
between leachwell and fire assay initiated a program of testwork
to resolve assaying discrepancies and provide further
understanding of the leachwell assaying technique.
Investigation of leaching time duration in October 1999
showed an increase in gold recovery as leach time was increased.
The consequences of leaching duration testwork resulted in a two
hour leach being introduced at Tarmoola, replacing a one hour
leach and results immediately showed leachwell identifying
higher grades than fire assay through the period mid-October
1999 to January 2000. The optimisation of leaching duration
time coupled with emphasis on quality sample preparation has
indicated the leachwell assaying technique, although a partial
digest has the potential to provide precise assays and identify
more gold than fire assay.
The increase of leaching duration to two hours showed no
significant change in the leachwell tail gold content. Leachwell
recovery had increased by up to eight per cent on average in
comparison with fire assay during the period October 1999 to
January 2000, which suggests there is a discrepancy with the fire
assay tail results. By definition, increased gold recovery should
equate to decreased recovery of gold content in fire assays of
leachwell tails. The reason for the lack of significant change in
fire assay tails cannot be explained at the present and has
identified an area for continued evaluation. For the purpose of
this research however, fire assay techniques through the principal
assay contractor have remained unchanged and fire assay results
are assumed to be a benchmark upon which leachwell assays
have been measured against.
Fire assay to sample extinction testwork has revealed the
heterogeneity of Tarmoola ore types and has showed wide ranges
in grades from a pulverised sample. The wide grade ranges are a
direct result of subsampling pulverised material containing
52
coarse particulate gold. Duplicate field split analyses have

indicated poor precision and further demonstrated the variability
of gold in Tarmoola ore types. The distribution of gold in a
pulverised sample through screen fire assay testwork illustrated
the need to ensure high standards of sample preparation are
enforced to ensure liberation of gold particles in a homogenised
pulp.
Mineralised samples are frequently highly siliceous and
substandard sample preparation can result in partial to full
occlusion of gold in resistant particles emphasising the
importance of high quality sample preparation. Ensuring all gold
particles are liberated to the leaching solution is crucial to
maximise gold recovery in the leachwell assaying process.
Pulverisation of sample should be completed to levels where
ideally the pulverised sample is 100 per cent-passing 75 microns.
CONCLUSION
Leachwell assaying is a precise assaying technique achieving
good repeatability from the same pulverised sample in Tarmoola
ore types. Fire assay is less precise in comparison and shows
wider scatter and both techniques produce unbiased repeat
assays. Leachwell versus fire assay comparisons have shown that
fire assay and leachwell have difficulties determining grade from
particulate gold mineralisation but leachwell has an advantage
over fire assay by using a larger pulp sample for assaying. The
larger sample reduces variability associated with coarse gold and
results in increased precision. However, leachwell accuracy or
recovery is highly susceptible to sample preparation and leaching
duration time. The quality of sample preparation has a significant
effect on liberation of gold particles. The failure to liberate gold
to leaching solutions will result in decreased recoveries in
leachwell assays and will under-call grades when compared with
fire assay. Leachwell provides higher levels of precision than fire
assay and results have demonstrated that optimised leachwell
assaying has the ability to out perform fire assay on average.
ACKNOWLEDGEMENTS
The author would like to thank Pacmin-Tarmoola Operations Pty
Ltd for their permission to publish this paper and for
managements enthusiastic support of assay quality assurance in
grade control at Tarmoola. Also acknowledged is Amdel
Laboratories Ltd of Kalgoorlie for continued research into
leachwell assaying.
Mine Geology
Controls on High-Grade Gold Distribution at
Vera Nancy Mine
D Sims
55
Geological Modelling and Grade Control in a

Narrow Vein, High-Grade Gold Mine
D Sims
65
Mine Geology Practices at the Sunrise Open

Pit
E Haren and P Williams
77
J R Vearncombe and
S Vearncombe
87
The Bunyip Lateritic Nickel-Cobalt Deposit,

Cawse Nickel-Cobalt Operations, Kalgoorlie,
Western Australia
A Bywater and S M Denn
95
Cadia Hill Gold Mine One Year Down the

Track
C F Moorhead and
Cadia Geology Team
105
The Geita and Kukuluma Mineralised Trends,

Lake Victoria Goldfield, Tanzania Orebody
Characteristics and Project Planning
D Bansah, R Chase,
A Davidson, H Michael,
M Skead and H Stuart
115
S Wilcock
129
Current Geological Understanding of Telfer

Gold Mine
G R Howard, T Hansen,
C Moore, P J Moffitt,
R J Inglis, R D Carlson,
I Kirchner, D Coupland,
S Leary and A Tomsett
135
Geology and Structure of the Morning Star

Mine, Mt Magnet, WA
R Mason, N Archibald,
D Holden, T Blyth,
S Huffadine, R Bradey,
A Jones and P Androvich
143
Controls on High-Grade Gold Distribution at Vera Nancy Mine

D Sims1
ABSTRACT
Determining the controls of high-grade mineralisation at Vera Nancy
requires careful data collection and assessment on a global and specific
deposit basis. The techniques developed over 60 years ago to analyse
structure, grade and width data are just as applicable today and can be
enhanced using modern geological modelling software.
High ore grades at Vera Nancy are controlled by both the primary
geometry of the mineralised structure and the intersections of the main
mineralised veins with subordinate splay veins. The main mineralised
structure dilated as a consequence of extensional tectonics while the
splays acted as conduits bringing oxidising surficial fluids into mixing
zones precipitating high gold grades.
Near-mine exploration is focussed on identifying zones of vein flexure
and splay interaction to locate the relatively small high-grade shoots.
Data modelling in a 3D environment is an essential process for
understanding the geometry of the system being explored.
INTRODUCTION
Determining the distribution of high-grade ore is a process which
involves the collection, compilation and analysis of detailed
spatial data to indicate the relationships which may (or may not)
exist between metal distribution, deposit geometry and
lithological or geochemical aspects of the deposit. Exploration
geologists develop genetic models which itemise and arrange
critical components to develop ore generation models for
targeting while it is the mine geologist who has the best
opportunity to understand why the orebody is really developed
where it is. Since profitability of a mine is enhanced by the
presence and distribution of high-grade ore it is important to
achieve an understanding of what controls the highest metal
grades in any gold deposit.
This paper discusses progress towards the understanding of the
Vera Nancy low-sulphidation epithermal deposits located 55 km
southeast of Charters Towers in North Queensland (Figure 1).
The objective is to establish the interrelationship between
geometry and gold distribution for the deposits and to develop
explanations that account for the localisation of high-grade gold
ore. The techniques used to analyse the data are drawn largely
from the text Mining Geology (McKinstry, 1948) and
augmented using computerised 3D geological modelling tools.
LOW SULPHIDATION EPITHERMAL DEPOSITS

Low-sulphidation quartz vein hosted epithermal mineralisation
(White and Hedenquist, 1995; Corbett and Leach, 1998)
produces deposits which are closely constrained within and
adjacent to the controlling structures which host them. The
structures act as conduits for mineralising fluids which undergo
pressure/temperature changes in the near-surface environment
leading to boiling, compositional change and fluid mixing in the
critical zone. Consequent mineral deposition results in the
structure being infilled with predominantly quartz-carbonate,
sulphides, adularia and precious metals. Relatively narrow
alteration envelopes form adjacent to and above the structure.
Alteration, mineralisation and vein textures have been modelled
into an idealised zonation pattern based on many southwest USA
1.

FIG 1 - Location of Pajingo Joint venture.
deposits which can be used predictively in exploration and

resource definition (Buchanan, 1981; Morrison et al , 1990). Ore
shoots in these deposits typically have plunging elongate shapes
with anisotropic grade distributions. Epithermal deposits form at
shallow levels with precious metals generally occurring over a
vertical extent of a few hundred metres.
Mineralised structures represent fault zones which dilated
allowing ore fluid movement, depressurisation and mineral
precipitation. Kinematics on such structures may be either
normal, reverse or transcurrent while ore fluid transport may be
upwards or along the structure. While it is clear that mineral
textures present in low sulphidation epithermal deposits such as
colloform-crustiform banding are a product of fluid boiling,
more contentious is the mechanism which causes precious metal
precipitation. Boiling alone is promoted by some as the sole
mechanism for gold and silver precipitation (Hedenquist et al ,
1996) while others promote fluid mixing (interaction with
surficial oxidised fluids) as efficient mechanisms for precious
metal deposition in certain circumstances (Corbett and Leach,
1998).
THE VERA NANCY DEPOSITS

The quartz-carbonate Au/Ag bearing veins at Vera Nancy have
been described elsewhere in this volume (Sims, 2000a; Sims,
2000b) and have been reviewed in a number of papers (Evans
and Jones, 1997; Richards et al , 1998; Butler et al , 1999).
Four main deposits, named Nancy, Vera, Nancy North and
Vera South, occur over a 2.2 km strike length to a maximum
depth of 600 m. Figure 2 shows the resource long section for the
Vera Nancy deposits. The contours on the long section show Au
gram-metre (gm) values with the 15 gm contour being the
55
D SIMS
FIG 2 - Long section showing deposits along Vera Nancy trend. Contours in gram-metres.
current limit of ore for the mine. The Nancy North, Nancy and
Vera deposits have two or more equant to steeply plunging
shoots separated by low-grade or barren zones. The deposits are
hosted in an andesitic volcanic sequence on the margins of the
Drummond Basin.
Each deposit has a main vein which carries the bulk of the
metal content while subordinate splay veins diverge from the
main structure and either die out in the host rock or rejoin the
main vein. Splay veins can carry economic grades but generally
only in close proximity to the main structure.
In all deposits the veins are moderate to steeply dipping but
contain bends and kinks in strike and rolls in dip. All gold and
silver is contained within quartz veining. Vein widths range from
a few centimetres up to 16 m but average widths in economic
zones are commonly 2.0 - 3.0 m. Nancy has narrower widths
than Vera but higher grades. Nancy North has mostly vein
networks defining the structures rather than the solid continuous
veining seen in Nancy and Vera.
Resources total 4.2 Mt at 14.6 g/t Au. Mine production began
in 1997 and is increasing to a rate of over 500 000 t per annum to
produce in excess of 220 000 oz Au per annum. All mining is by
underground methods utilising bench stoping on 15 m sublevels.
Total cash costs per oz Au for the three months to December
1999 was AUS $190 (43 607 oz Au produced).
The data compiled for this paper spans the Nancy North,
Nancy and Vera orebodies and includes over 105 km of 20 m
20 m spaced underground LTK60 infill core drilling as well as
ore drive development for bench stoping. Ore drive development
is around 90 per cent completed in Nancy, 50 per cent completed
in Vera and 30 per cent completed in Nancy North. As
56
underground drilling and ore driving has yet to commence in the

Vera South deposit, modelling to-date has relied on 40 m 40 m
spaced surface drilling. Only brief discussion is made of Vera
South and confined mainly to the gross geometry of the vein
arrangements as interpreted from core.
In general the underground infill drilling programs provide
enough information to reliably model the orebody geometry and
grade distribution for mine planning purposes to yield Probable
Reserves however geological understanding of the veins are
greatly enhanced during driving. During ore driving the face of
each cut (3 m long) is geologically mapped and sampled
followed later by backs mapping and 3D stope geology
modelling prior to stope extraction. Veins are generally narrower,
higher grade and more sinuous in driving than modelled from
drilling data alone. Data collection and interpretation focused
toward the resolution of detailed vein geometry is a key function
of mining geology at Vera Nancy (Sims, 2000b).
CONTROLS ON HIGH-GRADE ORE AT

VERA NANCY
Issues considered in assessing ore controls at Vera Nancy
include:
the arrangements of the mineralised structures on a global
scale;
the specific geometry and arrangement of the structures
within each deposit;
grade and width variation in relation to vein geometry; and
genetic evidence within the veins indicated by texture and
grade distribution.
CONTROLS ON HIGH-GRADE GOLD DISTRIBUTION AT VERA NANCY MINE
Arrangement of the structures on a global scale

Globally the Nancy North, Nancy, Vera and Vera South deposits
occur along the same fault zone and generally dip steeply south.
There is a through-going structure hosting the main mineralised
vein in each deposit with the main vein containing around 90 per
cent or more of the metal content in each deposit.
The fault zone has a sigmoidal Z shape in plan with the outer
deposits (Vera South, Nancy North) curving away by about 20
in strike from the reasonably linear trend of the central zone
occupied by Vera and Nancy. Figure 3 shows a plan of vein
geometries at the 1000 m RL (about 250 m below surface) while
Figure 4 shows the long section of Conolly contours for the main
vein hangingwall surface over the same strike length. Conolly
contours display the 3D geometry of a vein and are constructed
by contouring points of equal perpendicular distance on the
veins surface from an arbitrary dipping plane approximately
parallel to the strike and dip of the vein. They show vein
geometry in the same way as topographic contours indicate land
surface geometry. The structure contours in Figure 4 are
constructed from a plane oriented parallel to the central zone
containing Nancy and Vera. They demonstrate the changes in
strike between the central zone and the flanking zones of the
structure - strike changes occur where steep gradients in the
contours are evident.
Each deposit along the trend has in detail kinks and bends in
its strike and rolls in its dip. Splay veins diverge from the main
vein at inflection points and can either terminate in the host
sequence or rejoin the main vein along strike. Other subordinate
splays form linking structures between other splays. Figure 3
shows hangingwall splays in Vera and footwall splays in Nancy
located at inflection points in the main structure. Vera South has
a central cymoid loop geometry (McKinstry, 1948) with a fault
horse occurring between veins along a larger scale inflection.
Figure 5 is an isometric view of stacked plan slices at a 75 m

vertical slice interval showing the arrangement of the ore hosting
structures. The 1000 RL slice shown in Figure 3 is the fourth
slice from the top. The strike changes discussed above are
evident in the isometric view as are splay veins at inflection
points. The horse at Vera South has a longer down plunge extent
than along strike.
Post-ore faulting has disturbed the deposit geometry in
localised areas but not to the extent of removing or masking the
primary geometry of the host structures on a global or deposit
scale. The changes in strike between the central and end zones
are a function of primary fault zone geometry.
It is assumed that the kinematic history of all the Vera-Nancy
deposits along the structure are sympathetic the movement
history on one vein is related to the movement history of the
other veins during ore formation. The classic sigmoidal shape of
the fault trend in plan may at first glance imply that vein
development in the central zone is due to a dilational jog
resulting from dextral (right lateral) shear along the structure.
Given that appreciable dilation has occurred also in the flanking
fault trends (Nancy North, Vera South) such a model is
contradicted by other dilational indicators (see below).
Specific deposit geometry

Examination of the deposits/structures in more detail assists to
resolve kinematic issues along the fault zone. This discussion
will exclude the Vera South deposit as detailed modelling is yet
to be undertaken.
Figure 6 shows an isometric view of stacked cross-sections for
the Nancy North, Nancy and Vera structures. As with Figure 5
and Figure 3, the structures do not necessarily contain ore grade
material over the full extent depicted, but rather show the shape
of the ore hosting structure. Sections are spaced at 40 m easting
intervals along the mine grid.
FIG 3 - Plan of Vera Nancy trend showing vein geometries at 1000 m RL. Grid spacing is 250 m with mine grid north at top of page.
FIG 4 - Long section showing Conolly contours for the main vein hangingwall surface. Grid spacing is 250 m and contours extend to surface.
57
D SIMS
FIG 5 - Isometric view of stacked plan slices along 2.2 km strike length. Slices at 75 m vertical intervals.
FIG 6 - Isometric view of stacked sections showing structure of Nancy North, Nancy and Vera deposits along a 1.5 km strike length. Sections at 40 m
easting intervals.
58
The Nancy North portion of the structure has three splay veins
diverging from the main vein, two of which have steep plunges
and rejoin the main vein along strike and down dip. In
generalised section Nancy North has an upper zone with flatter
dips, a steeply dipping central zone then a lower zone again with
flatter dips. Sections shown in Figure 6 indicate this dipping
sigmoidal profile which lessens to a uniformly dipping vein
towards the deposits western limit.
The Nancy deposit consists of the main vein and four
subordinate veins. In the western portion of Nancy the main vein
dips steeply south in its upper portion to steeply north in its
lower portion and is located on the hangingwall of the deposit.
Subordinate veins splay from the main vein into the footwall
with two well-developed splays joining the main structure in the
western portion of the deposit. Above the junction point the main
amd splay veins are all dipping steeply south but where they
merge the main veins dip reverses to become steeply north
dipping (refer also Sims, 2000b Figure 3). This sigmoidal
profiled roll-over area has some of Nancys best developed vein
widths and grades while economic gold grades extend out within
one of the splays for approximately 50 - 100 m.
The Nancy footwall splays discussed above intersect the main
vein in a horizontal zone along the lower western portion of
Nancy but change trend to join the main vein in a more vertical
orientation in the eastern lobe of the Nancy deposit. When the
vein intersections become steep to vertical the roll-over shape is
much less pronounced or is not present.
Vera contains eight splays plus the main vein which, as with
the other deposits, contains the bulk of the metal. The Vera main
vein is continuous with the Nancy main vein but located to the
footwall of the vein array rather than to the hangingwall as with
Nancy.
Note that the upper Vera vein known as Vera North Upper
(VNU - see Figure 2) is the continuation of the main Vera vein
but displaced by post-ore movement on a steep reverse fault
which slices between the two deposits. The VNU vein has been
defined by drilling to continue some distance down the footwall
of the fault below the upper limit of the main Vera vein in
effect the structure has been duplicated by horizontal shortening
at the eastern end of Vera.
The Vera veins do not exhibit Nancys north-dipping
roll-over zones but have a moderately consistent steep southerly
dip. Splays to the hangingwall either emerge from and parallel
the main vein, form northward dipping link structures between
parallel veins or diverge from the main vein to die out in the
adjacent host rocks. In the eastern portion of Vera, which
contains the widest veins in the mine, south dipping splays
emerge from the main vein in areas where the main vein steepens
(Figure 6, Sims, 2000b, Figure 3).
All of the above geometrical relationships of the veins indicate
dilation was a function of extensional tectonics. Vein dilation in
Nancy North occurs in the steep ramp portion of the sigmoidal
profile during extension while the thickening in Nancys
roll-over zone is related to normal fault movement coupled with
jostling on footwall splays to accommodate opening. Veras
north-dipping linking veins and south-dipping splays also
support extensional movement.
Meso-scale veining supports normal movement on the
mineralised structures. To the footwall of the Nancy main vein
thin third order veins dip steeply to the north where the main
vein is vertical or dips steeply south (see Sims, 2000a, Figure
1a). Similar relationships occur to the hangingwall of the main
Vera vein where third order thin veins dip steeply to the north
while the main vein dips south.
Some component of strike slip movement is thought to be
evident in this dominantly extensional environment as indicated
by the positioning of splay veins branching from the main
structure at slight inflection points. Vera has hangingwall splays
branching at a northerly kink while Nancy has footwall splays
branching near a southerly kink in the main structure (Figure 3).

Resolving the magnitude and direction of shear is not readily
apparent and requires further work.
The outer economic margins of each deposit can change from
a 2 - 3 m wide ore vein to a thin poorly-veined chloritic fault or a
weakly mineralised silicified breccia zone over a 10 m strike
length without any obvious strike or dip change. Sometimes
subtle orientation changes are discernible at the point of thinning.
The rapid diminishing of the mineralised structure along strike is
interpreted to represent the tipping out of the extensional fault
plane which allows dilation of the vein. As can occur with other
normal faulting terrains it is expected that the fault surface is
approximately elliptical in shape. Maximum dilation/fault offset
occurs in the centre of the ellipse with displacement reducing to
insignificant amounts along the fault tip line.
Grade and width variation in relation to vein geometry

This section will cover Nancy North separately to Vera and
Nancy due to the difference in strike between these deposits. The
relationship between geometry, grade and vein width will be
discussed using Conolly diagrams which plot parameter
distribution on contours of vein geometry.
Figure 7 shows the Conolly contours for the main vein for both
Nancy and Vera. The boundary between the two deposits is
around 4850 m E. Intersections between the main vein and splay
veins are plotted as heavy lines on the long section. Note the
ridges evident in the contours are structural highs in the
topography of the vein representing changes in vein strike and
dip. Main ridges occur in Nancy around 4500 - 4600 m E and
4800 - 4850 m E and in Vera around 5300 m E. Nancy also
exhibits a rapid vertical gradient change around 950 m RL
between 4450 - 4650 m E that represents the roll-over zone.
The splay vein intersection lines in Nancy are quite continuous
with the two footwall splay intersections running horizontally
along the roll-over zone then changing strike to become
vertically oriented. The coincidence of the roll-over zone and the
vein intersections is a consequence of the splays role as dilation
structures to allow movement in this zone. The circular and
U-shaped intersection lines represent two other splays which
emerge from and rejoin the main structure.
In Vera the pattern of splay intersections is less continuous
intersecting the main vein as vertically, inclined or arcuate
shapes. There is a preponderance of intersections around the
ridge at 5300 m E.
Figure 8 shows a Conolly diagram with main vein Au grade
contours superimposed over the structure contours and splay
intersection lines. The contour lines are in 10 g/t Au intervals
with the shaded areas being 40+ g/t Au. Figure 9 shows vein
horizontal width contours with contour lines in 1 m width
intervals starting at 2 m. The shaded areas are 5 m+ in width.
In Nancy the distribution of both width and grade contours
show a common pattern with the higher values on or close beside
the ridges described by the structure contours. In the west lobe of
Nancy the best widths are developed in the roll-over zone while
high-grades extend up along the flank of the western ridge spur
around 1000 m RL and 4600 m E. The east lobe of Nancy
follows the ridge line closely and is paralleled and enveloped by
the splay vein intersections. There is a strong positive correlation
in Nancy between both grade and width with the location of
splay vein intersections and structural cusps in the vein trend.
Note however that the best grades do not always coincide with
the best widths. In the lower fringe of Nancy vein widths of 2 3 m occur without significant grades.
In Vera the best grades and widths are developed around the
5300 m E ridge with grades being better developed around the
upper eastern splay vein intersections. Large vein thicknesses
around 900 m RL result in high metal accumulations at lower
grades whilst widths and grades in the 5000 5250 m E section
59
D SIMS
have some correlation with vein intersections. A moderate

easterly plunging intersection between a flat northerly dipping
splay vein to the hangingwall of the main vein in the 5100 5250 m E zone appears to have some influence on grade and
width distribution. As with Nancy considerable vein widths

occur in the lower fringe of Vera with relatively low gold
contents and the best grades do not always occur on the widest
zones.
FIG 7 - Nancy-Vera structure contours with splay vein intersections. Grid spacing is 250 m and contour s extend to surface.
FIG 8 - Conolly diagram showing Au grade contours for Nancy-Vera. Shaded areas +40 g/t Au.
60
FIG 9 - Conolly diagram showing vein horizontal width contours for Nancy-Vera. Shaded areas +5 m width.
Figures 10, 11 and 12 show the main vein structure contours,

vein intersections, grades and widths for the Nancy North
orebody. Contouring intervals and shading are the same as with
Figures 7, 8 and 9.
The structure contours have a central plateau of flat contours
which is the steep portion of the sigmoidal profile. The structural
gradients above and below the plateau represent the flattening of
the structure up and down dip. Splay vein intersections are
concentrated in this plateau zone and are either vertical or
arcuate in trace.
High gold grades are more restricted in Nancy North
compared to Nancy and Vera but widths are generally high. Both
widths and grades are concentrated around the plateau zone but
grades are more tightly constrained in the central part of the
plateau in proximity to the splay vein intersections. Some degree
of structural ridge coincidence is evident for the width
distributions within the plateau zone whilst higher grades are
more coincident with splay intersections. Again, wide zones of
veining occur in the lower portion and fringe of the deposit
without appreciable grade development.
Considering the three deposits gold grade and vein width is
generally related to steeply plunging cusps or steeply dipping
flexures in the main vein structure combined with splay vein
intersections. High gold grades in Nancy and Nancy North
appear to be coincident with the intersection of splay veins and
the main vein whereas widths are related to the primary
geometry of the fault plane as a consequence of extensional
dilation.
movement and vein formation. Although more work is to be

undertaken on gold grade persistence in relation to quartz vein
texture, there is no obvious or clear relationship between grade
and texture within a quartz vein. All banded, moss and brecciated
quartz vein textures can host high-grade mineralisation. No
trends in base metal sulphide distributions are evident in any of
the Vera Nancy deposits and base metal sulphides are extremely
rare in the veins.
Around the lower fringes of the Vera, Nancy and Nancy North
orebodies well developed and continuous quartz veining carries
strong ore-bearing textures in appreciable widths but with only
low or minor gold contents (1 - 2 g/t Au). Strong evidence for
boiling in these areas implies that this mechanism alone is not
responsible for the localisation of high gold grades.
The above observation coupled with the closer correlation
between high gold grades and splay vein intersections implies
that fluid mixing, where oxidising surficial waters interact with
ascending gold bearing solutions, are a significant factor in
high-grade gold localisation. Fluid mixing is proposed by
Corbett and Leach (1998) as a significant mechanism for gold
precipitation in epithermal environments based on geochemical
modelling and field studies.
It is believed that footwall or hangingwall subordinate
structures which intersect the main vein/structure contribute to
ore localisation by not only allowing block movement and
dilation in extensional settings but have served as conduits for
descending surficial waters.
Genetic evidence from textures and grade distributions
CONCLUSIONS
The veins throughout the Nancy North, Nancy and Vera contain
abundant evidence that ore formation was a periodic and
repetitive event. Multiple movement stages on the faults
deposited veins through boiling (banded texture) and
quenching/recrystalisation (moss texture - Morrison et al , 1990).
Veins once formed were often brecciated and overprinted by later
Both the primary geometry of the mineralised structure, which

dilated under a dominantly extensional regime, and the
intersections of the main mineralised vein with subordinate splay
veins control high ore grades at Vera Nancy. The splays appear to
have acted as conduits bringing oxidising fluids into mixing
zones with ore fluids thus precipitating high gold grades.
61
D SIMS
FIG 10 - Nancy North structure contours with splay vein intersections. Grid spacing is 100 m and contours extend to surface.
FIG 11 - Conolly diagram showing Au grade contours for Nancy North. Shaded areas +40 g/t Au.
62
FIG 12 - Conolly diagram showing vein horizontal width contours for Nancy North. Shaded areas +5 m width.
Near mine exploration focusses on determining zones of vein

flexure and splay interaction to locate the relatively small
high-grade shoots. Data modelling in a 3D environment is a
critical process for understanding the sometimes subtle changes
in the geometry of the system being explored.
ACKNOWLEDGEMENTS
The relationships between grade distribution and splay vein
intersections were first recognised by Matthew Readford of
Snowden Associates when he and the author undertook resource
estimation work on the Nancy deposit in 1998. Structural
geologist Roric Smith was instrumental in recognising the
overall extensional structural setting of the deposits in particular
the mesoscale evidence for extension and the normal fault tip line
effects on the deposit margins. Helpful comments and
suggestions were made by an anonymous AusIMM reviewer.
REFERENCES
Buchanan, L J, 1981. Precious metal deposits associated with Volcanic
environments in the southwest, Arizona Geol Soc Digest ,
14:237-261.
Butler, I, Murphy, T and Parks, J, 1999. Vera south: discovery history, in
Exploration under cover, AIG Symposium, pp 25 31.
Corbett, G J and Leach, T M, 1998. Southwest pacific rim gold-copper
systems: structure, alteration and mineralisation, Society of
Economic Geologists, Special publication number 6.

Vera-Nancy deposit, North Queensland, in Proceedings World Gold
97, pp 233 - 237 (The Australasian Institute of Mining and
Hedenquist, J W, Izawa, E, Arribas, A Jr and White, N C. 1996.
Epithermal gold deposits: styles, characteristics and exploration, The
Society of Resource Geology (Japan): special publication number 1.
McKinstry, H E, 1948. Mining Geology , (Prentice-Hall).
Morrison, G, Guoyi, D and Jaireth, S, 1990. Textural zoning in
epithermal quartz veins, Klondike Exploration Services, Townsville.
Richards, D R, Elliott, G J and Jones, B H, 1998. Vera North and Nancy
Melbourne).
Sims, D A, 2000a. Sampling practice at Vera Nancy gold mine, in
Proceedings 4th International Mining Geology Conference, pp 35-42
(The Australasian Institute of Mining and Metallurgy: Melbourne).
Sims, D A, 2000b. Geological modelling and grade control in a narrow
vein, high-grade gold mine, in Proceedings 4th International Mining
Geology Conference, pp 65-76 (The Australasian Institute of Mining
White, N C and Hedenquist, J W, 1995. Epithermal gold deposits: styles,
characteristics and exploration. Society of Economic Geologists,
SEG Newsletter , 23:1, 9-13.
63
Geological Modelling and Grade Control in a Narrow Vein,

High-Grade Gold Mine
D Sims1
ABSTRACT
The Vera Nancy orebodies, although simple in overall geometry, are in
detail generally narrow, multi-veined and fluctuating in width, strike and
dip. Tight geological control is required to optimise gold and silver
production from this bench stoping operation. The processes used to
model the resource and to control ore drive development have evolved
since mining commenced in 1997. Grade control is a combination of
orebody sampling, geological data collection, understanding orebody
geometry and the optimisation of ore driving and stoping activities.
Global resource modelling utilises mainly drill hole data which is
largely superceded by drive development data as stope development
progresses. The orebody has a relatively low nugget effect that assists in
grade estimation. Rather than rely on sophisticated grade interpolation
methods, resource estimation to-date has focussed on developing the
geological understanding of the orebody to ensure the framework for
estimation is correct. Drill data alone is not adequate to define deposit
geometry for stoping. Drive data is used as the dominant data set for
stope geology modelling once development for a stope is complete.
INTRODUCTION
The Joint Venture (PJV - 50 per cent Battle Mountain Australia
Ltd and 50 per cent Normandy Mining Ltd) owns the Vera Nancy
underground gold mine located approximately 53 km SSE of
Charters Towers in North Queensland (Figure 1). Normandy
Mining act as operation managers and utilise a mining contractor
for extraction. Mining began on the Pajingo goldfield in 1987
when the Scott Lode open cut commenced extraction of
quartz-hosted epithermal gold/silver deposits found in the area.
With past production and present resources exceeding 2.5 M oz
Au and 4.0 M oz Ag the field is a significant gold district and
still highly prospective.
Current production comes solely from underground mining
along a 1.6 km strike length incorporating four distinct deposits;
Nancy North, Nancy, Vera and Vera South (Figure 2). The Vera,
Nancy and Nancy North orebodies were discovered in 1995
(Evans and Jones, 1997; Richards et al, 1998) and are in
production while Vera South was located 400 m along strike in
1998 (Butler et al,1999). Access development in Vera South is
underway for underground infill drilling commencing in May
2000 with production following in 2001. An upgraded mill
capacity of 600 000 t per year will see the operation producing
over 220 000 oz Au at peak production in 2001. On current
resources the mine has a ten year mine life.
This paper reviews current grade control and modelling
practice at Vera Nancy which has evolved since the mine
commenced development in 1997.
DEPOSIT GEOLOGY
Precious metal mineralisation at Vera Nancy occurs as shoots
within moderate to steeply-dipping epithermal quartz veins
hosted in an andesitic volcanic sequence. The sequence forms
part of the Drummond Basin with mineralisation dated at around
342 Mya (Porter, 1990; Butler et al, 1999). The Vera Nancy trend
is but one shear system in the field and extends to a vertical
depth of 400 m over a 2 km strike length.
1.

FIG 1 - Location of Pajingo Joint Venture.
Mineralised structures occur as quartz/carbonate infill veins

along faults accompanied by limited alteration halos (silica,
pyrite, sericite, carbonate, chlorite, haematite) flanking the veins.
Although in places sulphides in the form of pyrite can account
for up to two per cent of the rock, the system is a low
sulphidation epithermal type with economic mineralisation
exclusively occurring within the constraints of the quartz
veining. Gold occurs as electrum with silver (ratio approximately
1:1), is generally fine grained (<150 mm) and not visible to the
naked eye. Adularia is absent while the main carbonates are
ferroan in composition. Oxidisation extends to around 70 m from
surface and does not have a significant effect on the resource
which is dominantly below 120 m depth.
Movement on the faults during mineralisation allowed dilation
leading to depressurisation, mixing and boiling of gold/silver
bearing fluids within the structures. Ore occurs in steeply
plunging shoots within the dilatant areas. Orebodies are
generally characterised by strong banding and infill textures in
the quartz although in some areas, notably on the deposits lateral
and down dip margins, strongly banded veins are not
significantly mineralised. Precious metals are contained in quartz
veins and in breccias along the fault plane containing quartz vein
fragments. Silicification and late stage veining lack boiling
textures and are not mineralised. Gold grades can drop from
ounce-plus values in quartz vein to 0.01 g/t Au in andesite or
siliceous andesite breccia across a knife-sharp vein contact.
Each deposit sits on a major through-going fault
plane/mineralised structure which contains the majority of the
metal and has the most continuously developed veining. Dips of
the veins range from subvertical to 50 but average at around 70
to the SW. In detail the veins are not planar but are rather sinuous
65
D SIMS
FIG 2 - Long section of Vera Nancy deposit.
and undulating exhibiting rapid width, strike and dip variations.

Veins are commonly 2 to 3 m in horizontal width but can thicken
up to 16 m in places. They develop significant splays at
inflections where subordinate structures diverge from the main
vein. Splays can carry economic mineralisation and are stoped
over limited strike and dip extents either adjacent to or within
close proximity to the main vein. Splays can rejoin the main vein
or diverge to extend into and die out in the surrounding host
rock. Third order mineralised veinlets 10 to 20 cm wide
commonly diverge from the major structures and can extend one
to five metres into the footwall or hangingwall host rocks at a
low angle or subparallel to the major structure.
The Vera deposit contains the most metal and has the widest
vein widths - up to 16 m. It covers 500 m of strike and 200 m of
vertical extent. A main vein containing the bulk of the gold is
flanked by eight minor veins or splays which contain localised
economic mineralisation and are positioned mainly in the
hangingwall of the main vein. Figure 3 shows a representative
cross-section through the Vera deposit. Two smaller deposits, one
named VNU (Vera North Upper) the other Venue (Vera North
Upper Extended), occur above the Vera Deposit. VNU has been
mostly mined out extracted in the first 18 months of production,
while Venue is currently uneconomic. Note the shallowly
plunging trend described by the four major deposits is possibly a
feature related the to regional tilting of a palaeosurface (Figure
2).
The Nancy deposit exhibits narrower widths but higher grades
in comparison to Vera and consists of a dominant main vein with
four locally mineralised flanking splays. At Nancy, the splays are
positioned mainly in the footwall of the main vein. Nancy
extends 300 m of strike and 300 m of vertical extent with the two
main shoots separated by a barren zone. Figure 4 shows a
representative cross-section through the western shoot of the
66
Nancy deposit. Both Vera and Nancy share a similar NW strike

which differ from Nancy North and Vera South which strike
WNW.
Nancy North has a dominant vein ranging in width from 1 to
5 m with minor flanking veins while Vera South is interpreted to
have three major veins and three minor veins containing
economic mineralisation. Nancy North covers 250 m of strike
and 250 m of vertical extent while Vera South covers 500 m of
strike and 550 m of vertical extent.
Nancy North contains a relatively high proportion of
brecciated vein textures, stockwork veining and short scale vein
disruption which is believed to be a function of the deposits
position at a bend in the main structure. As the understanding of
vein morphology and continuity is a function of drill density and
ore driving data, a detailed understanding of Vera South is yet to
be developed. Exploration drilling indicates a highly variable
quartz vein texture and mineralogy (Butler et al, 1999).
The Resources and Reserves for the Nancy, Vera, Nancy North
and Vera South deposits as at end June 1999 are as follows (note
that Resources include Reserves):
Resources:
Measured
0.69 Mt
8.9 g/t Au
Indicated
1.36 Mt
18.7 g/t Au
Inferred
2.15 Mt
13.8 g/t Au
Total
4.20 Mt
14.6 g/t Au
Proved
0.53 Mt
12.5 g/t Au
Probable
1.59 Mt
13.9 g/t Au
Total
2.12 Mt
13.6 g/t Au
Reserves:
GEOLOGICAL MODELLING AND GRADE CONTROL
FIG 3 - Vera cross-section at 5360 mE showing development profiles, diamond drill hole traces and down-hole vein intercepts. Main vein in dark grey.
67
D SIMS
FIG 4 - Nancy cross-section at 4560 mE showing development profiles, diamond drill hole traces and down-hole vein intercepts. Main vein in dark grey.
68
Ground conditions and mining

Ground conditions in the mine are generally good with two to
three widely spaced (1 - 2 m) joint sets common in the host
andesite. Rock strength is high with andesite averaging around
70 Mpa and quartz vein around 120 Mpa. Poor ground
conditions are associated with faults and shears which occur
either subparallel to the veins or moderately dipping to the NE or
SW. Being emplaced along a fault system, the orebodies have
outer vein surfaces which can be polished or slickensided
reducing cohesion. Additionally the fault network has been
re-activated post-ore to produce orebody subparallel faults which
either cause vein duplication or create waste zones in fault
windows. Orebody parallel faults located 1 - 4 m from the edge
of the orebody are common in Nancy and Nancy North but less
prevalent in Vera. Stope overbreak is controlled by these discrete
structures.
Mining is undertaken using bench stoping using a 15 m floor
to floor sublevel interval. Ore drives are mined over the full vein
width out from a central cross-cut. Stopes are drilled with
parallel blastholes, initially fired into a raise-bored slot then the
brow is retreated to the central cross-cut. All ore driving is under
geological control using two boom jumbos and Elphinstone 1500
- 1700 loaders. Minimum drive mining width is 3.5 m (3.7 m
where dip is less than 65) although veins down to 0.5 m wide
are driven using resue mining methods and stoped as benches.
Stoping progresses in an overhand bottom-up manner with
initial decline access reaching the lower limit of ore in each
orebody prior to stope extraction no floor or crown pillars
remain and all voids are backfilled with mine waste. No selective
mining is undertaken in within the ore boundary.
Critical aspects for successful mining geology at Vera Nancy
include:
Reliably modelling the vein system using the infill
underground drilling data as a guide to assist ore drive
development. Using the best vein boundary locational data
available is crucial.
Maintaining tight geological control to ensure drives follow
veins as closely as possible whilst minimising the
undercutting of ore contacts. With generally narrow vein
widths of 2 - 3 m, undercutting is an unavoidable adjunct to
driving. Geological control can in particular minimise the
impact of undercutting to the stope hangingwall.
Use of all available data to solve daily problems in drive
development control this entails using drill data and
existing drive data below or above the level.
Capturing spatial location and grade data from drive mining
to assist in accurate stope geology modelling.
In a narrow vein, relatively high production rate mine, good
geological control during driving and stoping is a key element in
maximising profitability by reducing ore losses and dilution.
Reliable data collection, interpretation, modelling and data usage
all contribute to quality grade control.
DATA COLLECTION
Data collected at Vera Nancy includes surface and underground
drillcore logging and sampling, face mapping and sampling, back
mapping and geotechnical profile mapping. Detailed discussion
of sampling methods used at Vera Nancy is given elsewhere in
this volume (Sims, 2000).
Resue methods involve the face being fired in two parts one half of
the face is fired first in the waste zone and bogged as clean as
possible before the ore portion of the face is fired. This reduces the
amount of waste taken with the ore thus maximising recovered
grade.
The orebodies are delineated with dominantly HQ/NQ core

surface drilling on a nominal 40 m 40 m spacing to yield an
Inferred Resource. This is followed by 20 m 20 m infill drilling
from underground drill sites using LTK60 core to upgrade the
resource to Indicated status. Average hole length for LTK60
drilling is around 125 m with Eastman surveys initially at 12 m
then every 30 m. Underground drilling is mostly undertaken from
purpose-mined hangingwall drill drives.
Drilling orientation is on sections orthogonal to the Vera
Nancy mine grid which is oriented approximately parallel to the
strike of the deposits. Local vein flexures can see the vein
trend/drill hole intersection angle vary by 10 - 20 from
perpendicular. Once the ore drive in the main vein is established
on each level 10 m spaced infill drilling (sludging) is
undertaken from the drive to investigate splay veins observed in
the drive or indicated by the 20 m spaced drilling. Early sludging
was undertaken using a percussion blasthole rig (H104) but
unreliable width and grade data led to using a truck mounted
LM22 rig drilling LTK48 core. LM22 holes are typically 12 25 m long.
Core is logged and sampled on geological boundaries that
reflect the sharp grade distribution changes within the deposit,
digitally photographed and then half-sawn for fire assay. LTK48
core is whole core assayed. Logging is undertaken using a
barcode-based logging system (DATCOL) which delivers a
directly loadable ASCII file which is uploaded to a Vulcan drill
hole database via UNIX scripts.
Each ore drive cut is geologically mapped and chip sampled
with sample data uploaded into Vulcan as short drill holes
positioned across the face. Once the drive is completed the entire
length is washed down and the boundaries of the quartz
veins/breccia zones and significant fault planes are painted on the
backs and walls for survey pick-up. The pick-up strings are
loaded into Vulcan to provide hard locational data for boundary
control. Although the host rocks are non-magnetic survey
inaccuracy in diamond drill holes routinely sees surface holes 3 5 m out in location while underground drill holes can be 0.5 2 m out in cross-strike location.
GLOBAL RESOURCE MODELLING

Modelling of each deposit (global modelling) is undertaken when
underground 20 m 20 m infill drilling is completed. Two types
of data is used for this stage of modelling:
Spatial data is used to create the 3D orebody interpretation
wireframes and consists of orebody boundary location
information from surface and underground drilling. These
wireframes are subsequently used to construct the vein block
models. Vein boundary information from underground
development is not used in this modelling and in areas where
underground drilling has duplicated or repeated surface
drilling only the underground drilling is used. This aims to
lessen the effect of locational inaccuracy of surface drilling
on the model wireframe geometry.
Sample assay information, which is composited into vein
width intersections and used to interpolate assay grades into
the block model, is restricted to drill hole data alone. A
second wireframe which uses all surface and underground
drill data is constructed solely for sample composite
generation.
This approach produces a block model with vein geometry and
location based on the more spatially accurate underground
drilling data in the bulk of the ore zone and a sample composite
database which includes all drilling data for grade estimation.
Considerable effort is expended in ensuring the geology
interpretation accounts for all data and is sensible in light of
development experience - understanding the deposit geology is
the key to a good estimate and model at Vera Nancy.
69
D SIMS
Wireframing
6.
Wireframe construction uses the following process steps:

1.
2.
Cross-section and plan paper interpretation of structure

trends and orebody arrangement is undertaken on A0 paper
plots. Sections every 20 m are generated at 1:500 scale with
plans plotted at 50 m RL increments. Any available ore
drive vein profiles or plan mapping is drawn on the sections
and plans to assist in interpretation. Vein grades and
mineralisation lithologies are annotated and coloured onto
drill hole traces and interpreted into series of veins and
mapped to plan for 3D validation. The interpretation uses
both surface and underground drill data to develop the
structural arrangement of the veins.
The boundary of the orebody in each intersection is
positioned by locating the change from a mineralisation
lithology (quartz vein, brecciated quartz vein, strongly
silicified brecciated andesite) to host rock lithology
(unsilicified andesite, brecciated andesite, silicified
andesite). Assay data can be used to override lithology data
if assays >4 g/t Au occurred outside of mineralisation
lithology but are on trend of a recognised structure. If
internal grade distribution appears to be indiscernible on
the basis of vein geology then the entire structure is flagged
as the mineralisation intersection even if only part of the
structure carries grade.
Although such an approach may dilute some ore
intersections if there is no obvious visual control for
mining, it is assumed that the whole vein will be mined as a
single entity at the ore drive and bench stope stage.
Selective mining of the vein can not be undertaken if visual
control is absent.
Once the interpretation has been resolved 20 m sections are
then sliced through the drilling dataset in Vulcan on-screen
and separate contact string layers for each vein
(hangingwall and footwall) are constructed per section
using the paper interpretation as a template. The contact
strings consist of points snapped to drill holes at the
determined vein contact locations. For each vein
hangingwall and footwall surfaces are modelled separately.
Upper and lower limits of the interpretation are placed
above the extent of data or beyond the lower limit of drill
investigation. No interpreted points (ie points not on drill
holes) are added at this stage. The boundary string
interpretation is then checked by a second geologist.
3.
A boundary polygon is built linking points around the outer

limit of the interpretation for each vein surface. Each
contact layer is triangulated separately with all
interpretation string points disconnected and using the
boundary polygon as a limiting element. This surface
triangulation is constructed by projecting the points onto a
plane dipping parallel to the orebody. Trending is not used
as it tends to increase the number of crossing hangingwall
and footwall facets.
4.
Pairs of vein boundary surfaces are compared to ensure the

process has produced satisfactory surfaces. If surfaces cross
in zones of narrow widths then control points are inserted
into the boundary layers to ensure the crossing areas are
eliminated.
5.
70
Hangingwall and footwall surfaces are melded together to

create solid triangulations, which are then trimmed via the
Vulcan boolean option against adjacent veins, topographic
surfaces, etc. This method of solid triangulation generation
is used rather than the more commonly used method of
solid triangulation generation by sectional polygons as
surface triangulations are easier to update with infill
information and can be better manipulated to eliminate
touching or crossing facets.
Triangulations are validated to ensure the solids are closed,

not self-intersecting and topologically correct. These
booleaned triangulations are used for composite sample
generation and for block model construction.
Compositing and block modelling

Grade estimation to yield an Indicated Resource model involves
the following stages:
1.
The vein solid triangulations are used to generate

vein-width assay composites from diamond drill hole
assay data. This approach produces a single composite with
values for Au and Ag, which extend from one side of the
vein model to the other. No internal grade distribution detail
is retained in compositing. Each vein has its own set of
composite files and the composite is tagged with a vein
code during construction. Length-weighted averaging of the
uncut assay data is undertaken to produce the composite
grade and a Vulcan software enhancement developed for
PJV also resolves true and horizontal width of each
intersection given a generalised vein dip and strike.
Composites are checked to ensure correct coding. Statistics
are run on the composites to determine cut values which
approximate the 97.5th percentile. There is a degree of
smoothing of individual uncut high assay values during the
compositing stage by averaging the grade across the vein.
2.
The wireframes are used to generate a full-width or

seam type block models where single blocks extend from
one side of the vein to the other. Block width edge
resolution is 0.25 m with block size in the other two
dimensions being 5.0 m 5.0 m. The model is constructed
as dip aligned being parallel to the average vein dip. The
selective mining unit block size of 5.0 m long 5.0 m
high wireframe vein width evolved from running models
with larger block sizes and finding the block size is too
coarse for mine design purposes. Given a 20 m 20 m infill
drilling pattern this is the finest block size used with this
data. Blocks are given a vein identification code at
construction using priorities and inversions of solid
triangulations.
3.
Grade interpolation using Inverse Distance methods then

estimates Au and Ag grades into each vein model using the
veins composites as samples. This in effect creates hard
boundaries on assay data as only composites coded with the
correct vein identification are used for estimation thus
reflecting the sharp grade change evident on vein
boundaries. Since the composites used for estimation
include surface drill hole intersections which are not used
in the block model vein wireframes, a cross dip search of
sufficient width is required to ensure all relevant samples
are used during block estimation. Inverse Distance
weighting is applied using isotropic weighting within the
ellipse. Trials running Indicator Kriging models on global
resources for Nancy determined that Inverse Distance
interpolation delivers acceptable estimates for this stage of
modelling. Variography by Snowden Associates indicates
that a low nugget effect (0.25) is present in these deposits.
4.
The model is then validated both visually to inspect grade

smearing (blocks loaded on screen compared to
composites) and statistically (the average grade of samples
is compared to the average grade of the blocks). A number
of interpolation runs are made to optimise the estimation
results. When the model is deemed acceptable, block model
scripts are run to calculate vein width (block volume/block
profile area) and accumulation values (width gold grade).
Waste blocks are also scripted with a default grade of 0.1
g/t Au to allow reserve reporting of dilution. Resources are
then reported using an accumulation value cut-off of 15 gm

with a 4.0 g/t Au grade minimum for a vein by vein
resource breakdown.
No domaining beyond separating individual veins and deposits
is undertaken as mineralisation styles are similar within each
deposit and no significant oxidation effects occur in the
orebodies.
Early estimations used a block cut-off of 4.0 g/t Au for
resource calculations but it is considered that assessing grade
alone without examining block width is not adequately
representing mineability resources. A 15 g/m (with a 4.0 g/t Au
per block minimum) cut-off equates to a minimum vein
width/grade requirement more aligned to mine design
experience.
Reserves
Once the resource model is complete the reserve estimation work
undertaken by the senior mining engineer commences. All
reserve estimation is undertaken using a bench stoping method
with 15 m sublevel intervals. Reserve generation includes the
following stages:
1.
Model inspection by section and plan slices to determine

grade and width continuity, vein spacing and dip variations.
The orebody can be divided into areas where the vein(s)
horizontal width is either generally less (narrow) or more
(wide) than the drive minimum mining width (3.5 3.7 m).
2.
Drive and stope outline polygons are constructed on 5 m

section intervals for each vein to include expected dilution.
A solid triangulation is built using the polygons and a
reserve is generated against the block model for the drive or
stope. The design limit of the drive or stope is determined
by inspecting the vein grade compared to the required vein
grade to cover the cost of mining ore drives at the periphery
of the deposit (4.5 g/t Au). In areas where veins are too
close to extract separately by leaving a pillar, or in areas
where veins merge, design decisions are determined by
evaluating and reserving a variety of options through solid
triangulation construction. Major fault structures are
modelled as individual elements while ground conditions
and expected dilution are assessed qualitatively based on
ore drive geotechnical profile mapping (see below).
GRADE CONTROL
Grade control during ore driving is a critical aspect of mining
geology at Vera Nancy and has evolved to current practices over
the past two years of mining.
Sampling
During ore driving each round or cut of material is sampled
twice; once as a chip sample across the mining face and a second
time when the ore from the cut is trucked to the surface and
dumped in surface stockpile bays. Sims (2000) discusses
sampling of these materials. The grab samples taken on the
stockpiles are used to classify the ore into a series of grade
ranges used for mill feed blending while the development face
chip samples are used for ore drive control and as sample data
for stope scale block modelling.
Ore feed classifications to the mill are high-grade (+20 g/t
Au), ore (5 - 20 g/t Au), low-grade (1 - 5 g/t Au) and marginal
(0.3 - 1 g/t Au). Only high-grade and ore is fed to the mill with
low-grade being retained for later milling following mine
closure. Marginal material is stockpiled at the mine as it is
currently not economic to treat. The low pyrite content and
non-refractory nature of the ore ensures long-term storage of
low-grade ore does not adversely effect milling performance.
Geological control of development

Although sampling faces for grade estimation and sampling truck
dumps for mill feed blending are important activities the best
opportunities in maximising mill head grade is in optimising
geological control during ore driving and stoping. Having a
bottom-up mining method precludes selective scheduling to
maximise head grade and sequences are relatively inflexible.
Ensuring grade maximisation is a key objective.
Each development face is mapped on 1:50 scale mapping sheet
(Figure 5) noting vein position and dip in the face, vein structure,
texture and geometry, faulting and jointing, alteration and the
presence of thin veining in the host rocks or drive walls. The face
is chip sampled based on geological domains and an estimate of
the grade-range of the face is made. The next cut boundaries are
painted on the face, the desired width and height written on the
face and one or two arrows are painted on the backs to indicate
the direction in which the next cut is to be taken. Arrows may be
in different directions if the vein is opening out in width. The
geologist has the mapping sheet from the previous face with
sample grades to assist in decisions regarding the correct position
of the drive hangingwall or footwall. Plans are also carried
showing the block model outlines for the level and the trace of
the development level immediately below the drive.
A significant issue in ore drive control is the rapid change in
strike direction and width which occurs in some parts of each
deposit. Fluctuations which can cause ore drives to run off or not
fully recover the vein along the level generally occur on a
spacing less than the infill drilling. The resource model may not
predict short-scale changes and so driving involves close
monitoring of information from drives above or below the
current heading.
When driving runs off or fails to completely recover the
vein(s) stripping is undertaken to extract the portion left behind.
Where large amounts of veining appear to be left behind in the
walls LM22 drilling or sludging may be undertaken to determine
the thickness and grade of the veining left behind in the wall. An
increasing level of caution regarding sizeable wall stripping has
developed in the mine as poor control on stripping can often lead
to increased undercutting of ore boundaries and hence higher
dilution levels during stoping. Often small strips in the walls are
taken with the current face to clean up minor amounts of quartz
veining left behind from the previous one or two cuts.
When a heading needs to be turned sharply to follow a change
in vein direction the ability to change direction is limited by the
position of the vein in the face relative to the shape of the drive
where the jumbo is to operate. Since the jumbo booms are
around 5.0 m long sharp turns are not possible as the end of the
boom often hits the wall before reaching the desired drilling
angle on vein boundary holes. In such cases a strip cut is taken to
modify face geometry before the next cut is taken. In all
circumstances where a face requires a significant amount of
turning, the mine shiftboss is consulted to ensure what is desired
by the geologist is possible. This advice often dictates the
requirement for pre-stripping a face or not for the next cut.
The best ore drive geological control is totally ineffectual if the
mining contractor does not ensure markups are rigorously
followed and optimum face boring practices are employed. The
contract currently in place at Vera Nancy has a financial penalty
for the contractor if face mark-ups are not followed. A maximum
overbreak tolerance of 0.2 m on each boundary compared to the
mark-up dimensions is allowed before a penalty applies. For
example if a face is marked up at 4.5 m wide by 4.5 m high a
maximum allowable dimensions at the end following the
extraction of that cut is 4.9 m by 4.9 m.
Oversized faces are inspected by the shiftboss and the Pajingo
Joint Venture mining supervisor to determine if the overbreak is
a mining quality issue or an unavoidable ground conditions issue.
This is generally based on locating the position of the butts from
the last cut and inspecting ground quality. If the overbreak is a
71
D SIMS
FIG 5 - Face mapping sheet.
72
function of poor practices the jumbo operator concerned is

counselled and the mining contractor financially penalised with
greater penalties for larger overbreaks. Generally mining
contractor performance in following mark-ups and minimising
overbreak is very high.
It is possible to have some quartz veins or quartz breccia zones
in the face carrying subeconomic grades where an adjacent vein
may be better mineralised. This is more common near the deposit
limits along strike or down-dip. Not all quartz carries ore and the
previous face sampling data is used to guide the next markup.
Ideally the hangingwall of the vein is positioned 20 cm
inboard of the drive hangingwall. This is to minimise
undercutting of the vein hangingwall, which is important in
reducing stope dilution and to ensure optimal positioning of the
longhole rig to allow drilling along the ore contact above or
below the drive. The minimum mining width of 3.5 - 3.7 m
(determined by loader size) ensures undercutting is a common
occurrence but geological control aims to position it to the
footwall rather than hangingwall.
Veins up to 8 m wide are mined full width in a single pass
while wider veins are mined along the footwall at 6 - 8 m wide,
sludge drilled to fix the contact in detail and cablebolted with
6 m bulbed single strand bolts. The veins are then stripped to full
width in one or two passes with additional cable bolting during
or after stripping. Veins in Vera out to 16 m wide have been
driven in this way.
Veins less than 1 m wide or low-grade veins around 2 m wide
are mined using resue methods which increases recovered grade.
The waste from a resue cut is trucked as marginal ore to a
marginal pad which, following stockpile sampling, may become
low-grade ore or be retained on marginal pad at the mine.
To-date, modelling using surface and underground drilling
alone tends to overstate vein width but understate vein grade.
When driving is completed the vein model is updated for a stope
model between levels (see below). Additionally where LM22
drilling is undertaken on adjacent veins 1:250 scale sections on
10 m centres are plotted and detailed interpretation using face
mapping/sampling data and infill drilling data are undertaken in
section and plan.
The rating for each section is then transferred onto a

longsection showing the drives in each orebody labelled with
either low, medium or high dilution risk ratings along their
lengths on 10 m spacings. The longsections can be used to locate
areas requiring additional hangingwall or footwall ground
support via cablebolting or to determine areas where drill and
blast parameters may be changed. Introduced in mid-1999 the
dilution risk rating is proving successful in locating areas of high
dilution risk.
STOPE GEOLOGY MODELLING

Once ore driving is completed on the upper and lower levels of a
bench stope a remodel of the ore boundaries is undertaken for the
stope between the levels using the most current information. The
resource model based on drilling data is not accurate enough in
its spatial position for blasthole ring design and becomes
superceded by a stope geology model.
Geological boundary back string mark-up, pick-up and
geological mapping is undertaken once a drive is completed. The
vein boundary and fault position back strings are imported into
Vulcan and used to fix geological features in 3D space.
The process for stope geology modelling is as follows:
1.
Drive as-constructed wireframes, face sample drill holes

and ore boundary back strings are loaded on screen and the
geological mapping and chip sampling hard copy plans and
the folder of drive face mapping/sampling sheets are
gathered for reference.
2.
A section is sliced with a 1 m width across the drive at the

location of the first chip sample taken for the drive. The
back strings show the position on that section of where vein
boundary sits in the backs or walls while the face mapping
sketch shows the appearance of the face at the time of
mining. Although the short drill holes constructed from the
sample data taken across the face may not coincide
accurately to the actual position of the face due to plotting
errors, the back strings provide hard locational data for
modelling. Using the face mapping sketch two strings are
digitised on screen representing the orebody hangingwall
and footwall. Any fault planes are also digitised on the
screen.
3.
The section slice is advanced along the drive until the next
face sample is encountered usual 3 - 4 m along strike. The
section plane is reoriented if required to ensure a true
section for that part of the drive is obtained parallel to the
sampled face. The process of matching back string to
geology mapping face sketch is repeated. Special attention
is given to grade distribution in each face which may
indicate ore outside the back string limit. For example, thin
peripheral veining can carry extraordinarily high gold
grades through part of the Nancy system in what appears to
be andesite. The process continues until the end of the drive
is reached. This may involve up to 100 face maps. Once the
boundary strings are digitised they are validated by visual
inspection checking vein dips and locations against face
mapping sheets and by doing fly-through views along the
drives checking digitised strings against back pick-up string
locations (Figure 6).
4.
The strings digitised as face ore boundaries along the length

of the drive are used to construct solid triangulation
surfaces (termed ribbons) representing either the footwall
or hangingwall surface of the orebody as mined in the drive
(Figures 7 and 8). Faults are modelled as triangulated
surfaces in a similar way. Checks on face chip grades
establishes the economic limits of the orebody along strike
as defined during driving and the ore boundary surfaces are
Geotechnical input to stope design

An estimation of predicted ground behaviour during stoping is
required to locate areas of high dilution risk for mine design
purposes. The majority of stope dilution issues encountered
to-date are related to the effects of discrete fault surfaces
paralleling the orebody boundaries coupled with low angle joint
sets particularly in areas where poor drive control has undercut
the ore boundary. In extreme cases up to 4 m of hangingwall
dilution can occur requiring re-slotting of the stope to leave an
ore pillar.
A geotechnical profile mapping method has been devised in
conjunction with Golder Associates which rates the risk of
hangingwall and footwall dilution along each ore drive on
sampling points every 10 m down the drive. The system has
evolved from a scan line mapping methodology through a
window mapping system to the current process which looks at a
section profile of the drive.
A profile sketch is made of the drive showing development
outline, ore boundary position, significant joint sets and fault
planes. Discontinuity orientation and characteristic data is
collected and an assessment matrix is completed on the mapping
sheet that determines the risk of stoping dilution for both the
hangingwall and footwall surfaces. The rating is determined by
summing a set of weights that relate to specific geotechnical
features (such as fault frequency and distance from ore boundary,
presence of low dipping joints, etc) and classifying the resultant
number into a risk category.
73
D SIMS
FIG 6 - View along a drive ascon triangulation with chip sample drill holes, ore contact back strings and ore boundary strings visible.
FIG 7 - Same view along drive triangulation with ore boundary triangulation.
74
FIG 8 - View onto drive with ore boundary strings, ore boundary triangulations and face samples drill holes visible.
only produced within these limits. The ore boundary

surfaces are then linked between levels to produce
hangingwall and footwall boundary surface triangulations
for the stope that encloses only ore (Figure 9).
5.
Final validation of ore boundary wireframes uses LTK60

drilling data to ensure no peripheral veining or wider ore
zones have been overlooked between levels. Once final
validation is complete the stope ore boundary surfaces are
reviewed with the ring design mining engineer and blast
hole design can commence.
6.
A grade control block model is produced which uses the

drive and stope ore boundary surfaces converted to solids
and used to define block edges using the full width type
block modelling approach described above. Block size is
reduced to 2.5 m along strike 2.5 m up dip to reflect the
closer spaced sampling data from face samples. Blocks are
only constructed within the stope model wireframes hence
only occupy strike extents identified during driving as ore.
Sample composites are built using the chip sample drill
holes and coded in the same way as the global resource
model composites. Estimations are made using similar
methods to those described above and the models are
reserved against diluted ring design polygons to yield
expected stope tonnes and grade.
Reconciliation of models with production to-date (Sims, 2000)

indicates model reserves overstate tonnage by around nine per
cent while grade is understated by around eight per cent. Around
99 per cent of reserve ounces are produced. Although Inverse
Distance interpolation methods perform reasonably in stope
estimation a better local estimator of grade is required and
further kriging work is in progress. An accumulation estimation
approach, whereby width and accumulation (width by grade) are
modelled in an attempt to reduce adverse support effects from
different sized samples, is considered an essential improvement
and is being investigated.
CONCLUSIONS
The Vera Nancy orebodies are intensive in the amount of
geological control required to successfully model and extract
their ore. Data collection, interpretation and grade control
focuses on developing an understanding of each deposits
geometry on a detailed scale for both drive control and stope
geology. To-date the focus has been on data collection and
interpretation to understand and predict the shape of the deposits.
Sophisticated grade interpolation methods have not been widely
employed. Inverse distance grade interpolation aided by a low
nugget factor performs adequately on a global scale but needs
improvement as a local grade estimator.
75
D SIMS
FIG 9 - Stope geology models for Nancy 986W and 986FWW stopes.
ACKNOWLEDGEMENTS
The following geologists have contributed to developing the
geological methods at Vera Nancy since mining commenced;
Dave Hall, Allan Bell, Sharon Turner, Michael Jackson, Jacki
Stott and Chris Chambers.
REFERENCES
Butler, I, Murphy, T and Parks, J, 1999. Vera south: discovery history, in
Exploration under cover, AIG Symposium, pp 25-31.
Vera Nancy deposit, North Queensland, in Proceedings World Gold
97, pp 233-237 (The Australasian Institute of Mining and
76
Porter, R G, 1990. Pajingo Gold Deposits, in Geology of the Mineral

Deposits of Australia and Papua New Guinea (Ed: F E Hughes), pp
1483 - 1487 (The Australasian Institute of Mining and Metallurgy:
Melbourne).
Richards, D R, Elliott, G J and Jones B H, 1998. Vera North and Nancy
Melbourne).
Sims, D A, 2000. Sampling practice at Vera Nancy gold mine, in
Proceedings 4th International Mining Geology Conference pp35 - 42
Mine Geology Practices at the Sunrise Open Pit

E Haren1 and P Williams2
ABSTRACT
The Sunrise deposit is one of Australias largest and most profitable gold
deposits. It has consistently and spectacularly outperformed every
resource model created. Currently over the life of the Sunrise pit there
has been a 50 per cent under estimation of ounces mined. Without a solid
basis for predicting the tonnes and grade or the spatial location of
mineralisation, mine planning becomes problematic. Recently, through a
better understanding of the mineralising processes the reconciliation of
the resource model has improved to be in the order of ten per cent under
on ounces mined. This paper will describe how the mine geologists at
Placer Granny Smith have implemented the changes required to improve
the resource model. Work began with the development of a
three-dimensional lithological and structural model based on historical
drilling and pit mapping. Continual updating of the geological model
with grade control logging, assays and in pit face mapping, now allows
for the on-going improvement of the resource model and the planning of
mine development drilling.
INTRODUCTION
The Sunrise Pit is located 32 kilometres south of the Granny
Smith mill, which is 950 kilometres northeast of Perth in Western
Australia. The Sunrise deposit was discovered in 1992 and an
initial resource of approximately 400 000 ounces was estimated
based on predominantly oxide drilling. Mining development
commenced in 1994 and ore production the following year with
mining to-date (January 2000) producing approximately 1.9
million ounces. Current indicated and inferred resource are
6896 kt at 3.37 g/t for 747 000 ounces with reserves of 4019 kt at
3.48 g/t for 450 000 ounces. Mining is expected to continue until
February 2002. Placer Granny Smith, a joint venture between
Placer Dome (60 per cent) and Delta Gold (40 per cent), operates
the mine. The Sunrise Pit forms part of a much larger gold
deposit, with known resources at depth currently being drill
tested.
An important aspect of the development of the Sunrise deposit
has been the consistent underestimation of the total contained
metal. This has led to a number of problems, including
conservative pit optimisation, and uncertainty in definition of ore
blocks, resulting in overly selective oreblocks with potential for
loss of ore to low-grade or waste stockpiles. Solution to the
reconciliation problem therefore has significant cost benefits to
the mining operation at a number of critical points in the mining
process.
To ensure that the mine was operating at peak efficiency, the
geology team at Sunrise approached the problem at its most basic
level at the time of transition between mining the oxide resource
and the more critical fresh rock resource. SRK Consulting were
invited to initiate the geological evaluation while utilising their
structural and geological experience. The fundamental problem
was to develop a geological understanding of the deposit, that
would allow better resource definition, improved grade control
practice and allow better in and near mine targeting A conceptual
framework for the process has been developed, which shows the
inter-relationships between the various mining activities,
identifies the key geological inputs required, and allows an
estimation of the cost benefits of the geological inputs.
1.
MAusIMM, Geologist, Placer (Granny Smith) Pty Ltd, PO Box 33,

Laverton WA 6440.
2.
Managing Director, SRK Consulting, 25 Richardson Street, West

Perth WA 6872.
This paper outlines the processes involved in developing and

managing the geological model. It is then demonstrated how
converting the conceptual framework and geological model into
operational tools used on a regular basis has reaped benefits. The
most important of these are improved reconciliation, improved
resource drilling economics and improved geological management in the daily operation of the pit.
Conceptual framework
The processes that led to accurate definition of ore prior to
delivery to the mill are complex. In many operations, grade
control is the major process that determines the quality and
quantity of mill feed, and the resource model determines the
expected mill feed from pre-mining assay data. The
reconciliation between actual and expected delivery of ore, and
the recovery of metal, is an important measure, because most
planning and development activities at the operation rely on the
resource model. Large differences between the resource
prediction and the actual delivery could mean large losses in
potential reserves due to a number of planning issues:
extremely conservative pit optimisation;
conservative sampling limits may not sample areas of ore;
scheduling for mining is not optimal;
waste dump designs are too large resulting in unneeded
expenditure for release of ground, environmental bonds, etc;
ore pad designs are too small; and
scheduling for milling is not optimal.
The most important factors that lead to improved
reconciliation and planning practices are outlined in the linkages
chart shown in Figure 1.
FIG 1 - Conceptual framework of linkages between factors for the

on-going development and mining of Sunrise.
77
E HAREN and P WILLIAMS
Figure 2 summarises the percentage differences between grade

control and resource models up until the time of implementation
of results from the new geological model to geological practices
in the pit.
Full-scale mining commenced at Sunrise in September 1996.
Small volumes mined have amplified the percentage differences
shown in the graph before this time. By January 1997 most of the
transported mineralisation had been mined. This zone of
mineralisation proved to be more widespread than initially
estimated. The grade however was usually only slightly higher
than predicted. From January 1997 to December 1997 the ore
mined was predominantly oxide. The end of oxide mining
occurred during August 1998, oxide mining recommenced from
June 1999 to the present. At the time of transition between oxide
mining and fresh rock mining, the positive reconciliation was
increasing (Figure 2) and it was recognised that the distribution
and controls on ore were not well enough understood to give
confidence that the operation would extract the maximum benefit
from the orebody. Consequently, the decision to develop a new
resource model, based on better geological modelling, was taken.
The resource model sits at the top of the chart, as the primary
input to the mining process, and the mill process sits at the
bottom of the chart, as the primary output from the mining
process. Directional arrows show the factors and
inter-relationship between the factors. Each of the factors
represents a dynamic process in its own right, and in many
operations these processes are either ignored (for example pit
mapping) or not integrated into the overall thinking of the mining
team. In either of these cases, important data that leads to
improved efficiency or cost-savings for the overall operation are
not available to the operations or planning departments. The
linkages chart highlights the importance of developing a quality
geological model for the deposit. Removal of the sectional
interpretation from the inner circle, for example, results in an
inability to develop a 3D geological model, and also severely
restricts ability to predict geology and grade on the next bench to
be mined. Lack of a geologically realistic expectation of grade
distribution means a total reliance on grade control numbers a
blind acceptance of the results of drilling, assaying and numeric
interpolation without any reality check against an expected
model. Consequently, any needs for improvements in grade
control or ore mark-out processes are unlikely to be recognised,
let alone implemented. In this contribution, we discuss the
various aspects of the chart, and use the results to discuss the
economic advantages of implementing an integrated geological
model in many aspects of the mining process.
DEVELOPMENT OF GEOLOGICAL MODEL

Following the discovery of the transported and oxide resource,
the exploration department discovered the deeper fresh rock
resources through a program of mine-based exploration. The drill
spacing and depth for the initial geological model meant that the
overall deposit style and core scale lithologies and structures
were reasonably well-defined. However, the nature of
mineralisation, controlling structures and distribution of
lithologies at the scale necessary for development of the resource
at a mine scale was not adequately defined.
Mining geologists needed to consider features on the scale of
metres to tens of metres in an active often fast moving mining
environment. Pit personnel such as shotfirers and production
HISTORICAL RECONCILIATION RESULTS
Jun-98
Apr-98
Feb-98
Dec-97
Oct-97
Aug-97
Jun-97
Apr-97
Feb-97
Dec-96
Oct-96
Aug-96
Jun-96
Apr-96
Feb-96
% Difference
Positive reconciliation between grade control and resource

models has characterised the history of mining at Sunrise. Over
the life of the Sunrise pit the reconciliation between grade
control and the various resource models has meant an extra
610 000 ounces of gold has been mined. This is made up from an
extra 16 per cent in tonnes and 27 per cent in grade to make an
overall gain of 47 per cent in ounces.
Month
FIG 2 - Per cent difference between grade control and resource model predictions up to June 1998.
78
MINE GEOLOGY PRACTICES AT THE SUNRISE OPEN PIT
engineers required specific accurate and simple information

about the material they were dealing with. Good predictive
information on structures which would be encountered,
structures known to cause problems, and a working
understanding of how different conditions (wet/dry, shearing,
silicification) interact with mineralisation and affect rock
mechanics was also required.
Development geologists also have a number of concerns that
generally require understanding of the geology of the deposit at a
larger scale than the exploration scale. At Sunrise the
development drilling scope requires information that can define
ore pods or ore shoots at a scale of 15 - 50 m. This requirement
demands a high level of understanding of the relationship
between detailed structural geometry and ore distribution. This
detail of understanding was not available at the completion of
exploration drilling.
Geology of the Sunrise deposit

Improvements in resource modelling, as well as within-mine and
near-mine targeting are fundamentally dependent on an improved
understanding of the geology of the Sunrise deposit. The
geometry of the Sunrise deposit had not been systematically
documented prior to mining of the oxide resource. Work by
Standing (1993), Ojala (1994, 1995), Newton (1997) and Newton
et al (1998) had established the generalised geometry, and
inferred that the localisation of mineralisation is the result of
heterogeneities caused by inflections in major shears. Because of
the generalised nature of this previous work, a program was
established to develop a detailed geological model that would
address mine geology and development geology needs.
This work generated a robust three-dimensional structural and
lithological model of the deposit based on a series of 40
cross-sections and three level plans produced from resource
drilling data and pit mapping data. The rocks are an interlayered
sequence of fine grained sediments, banded iron formation,
volcanic and volcaniclastic rocks of intermediate composition
and minor porphyritic felsic intrusive rocks, forming four major
correlated units with two additional units separated by major
shears. These units were used as part of the 3D geological model.
A mineralogical and geochemical study of the mineralisation
and alteration system was also undertaken in order to develop a
fluid flow model. This model linked the structural framework to
mineralisation envelopes, and allowed mineralisation envelopes
to be better constrained in the resource model.
The key findings from the geological studies are as follows:
Mineralisation within the Sunrise deposit is within three major
sites of deposition (three major styles):
1.
shear-hosted mineralisation - characterised by wide

alteration assemblages and dispersed low-grade
mineralisation haloes;
2.
pod mineralisation - characterised by high-grade, discrete

quartz-carbonate veins in narrow alteration selvedges.
These pods contain the largest number of ounces in the
deposit; and
3.
BIF-hosted mineralisation in anticlinal fold closures

adjacent to thrust faults.
The structural geometry of the deposit is controlled by a series
of west-northwest dipping thrust faults.
The faults in the Sunrise pit connect at depth to a large
reverse shear (the Sunrise Shear).
The thrust faults had a movement direction towards the
east-southeast on a vector plunging approximately 30 to
290.
The BIF units form a series of open anticlines and synclines
adjacent to the faults.
All three mineralisation styles are controlled by the thrust fault

framework.
Mineralisation was at the same time as reverse movement
along the faults.
The thrust faults acted as the major fluid pathways during
deformation and mineralisation and are all intensely altered.
Alteration mineral assemblages associated with gold are
broadly the same in all three mineralisation styles.
The pod mineralisation is related to brittle deformation in
competent volcanic units between shears. Vein orientations
within the pods are predominantly parallel to the shear
foliation, but high-grade veins also dip to the east and the
west at 50 - 60.
Pods are located above flatter portions of the top of the
Sunrise Shear surface.
The strike extent of pods appears to be controlled by lateral
ramps in the thrust system.
BIF-style mineralisation is related to fractures developed in
the BIF during folding, providing fluid access for
de-sulphidation reactions.
Mineralised veins caught up in progressive deformation and
alteration may dominate shear-hosted mineralisation.
Mineralisation is thicker in flats on the shears.
Interpretation methodology
The Sunrise resource drilling was based on a nominal grid of
25 m 25 m of combined percussion and diamond drilling. Most
holes were drilled to the east at a declination of 60.
Geology interpretation
The cross-section geometry was interpreted using a combination
of examining the core photographs and examination of critical
sections of core. The main factors to examine in the core were
bedding orientations and verification of the presence of shear
zones. Bedding orientation was the critical factor in checking
correlation of sediment and BIF units. Where possible,
bedding-cleavage relationships were also determined. The initial
structural and lithological intercepts on section were plotted from
pit mapping data. These preliminary interpretations were not
constrained by the 3D connectivity, or by the latest phase of pit
mapping.
Following the completion of the preliminary interpretations,
concurrent pit mapping was used to provide additional
constraints on the interpretation, and all sections were examined
as a set to ensure there was 3D continuity of both structures and
geological units. A significant problem in the Sunrise deposit is
the low dip of units, which places a considerable sensitivity on
the section interpretation. This is because small variations in RL
of units on the section will produce significant changes in
horizontal location in bench plans produced from the sections.
The mining method employed at Sunrise does not leave clean,
undisturbed faces during mining that can be easily mapped.
Because of this, there is not a very good control on the geology
of the pit floor on any geological maps of the pit. An example of
a section of the geology is shown in Figure 3.
Correlation process
Digitised strings were completed for each section, and these were
used to generate horizontal slices for checking the continuity and
correlation of structures, and to check that structures did not
generate unrealistic intersections in level plans. At the end of this
stage, the model was sliced at 5 m intervals to ensure the faults
and shears formed continuous surfaces or merged realistically
with other structures. These alterations to the interpretation were
crosschecked against the original drilling data to ensure any
changes did not violate the primary data set.
79
FIG 3 - Polygonised geological section at 70 000 N, Sunrise, generated using the processes outlined in the text.
Once the shears and stratigraphic units were correlated through

the model, polygonisation was completed prior to commencing
the wireframing of the major components of the model.
Wireframes were completed for all the shears and faults, and the
BIF units. The final correlation of structures is shown in
Figure 4.
Relationship of mineralisation to the geometric model

The relationship between mineralisation and the geometric
model is determined by:
correlating the 3D structural features and grade or resource
intercepts;
linking that correlation to the results from the study of the
local movement history or kinematics; and
defining the internal structural geometry of ore zones.
This approach leads to the identification of low strain zones,
and means the geometry of the orebody can be used to predict
additional sites of likely gold accumulation.
Grade control data provides a valuable source to test the
structural model against ore distribution. The data down to
295 RL was gridded to produce level plans and sections through
the grade control data. A contoured version of this information is
shown in Figure 5.
The grade control sections demonstrate a very strong
correlation between the shear zones and faults mapped on
sections and the distribution of ore. The main zones of
mineralisation are both within and adjacent to the shear zones.
Large accumulations of ore occur between adjacent shears,
particularly where shears flattened and subsidiary faults
developed at the down-dip point of curvature on the shear (see
Figure 5).
There is also a correlation of gold grade with BIF beds in
particular structural locations. The coincidence of BIF and a
shear zones is an ideal site for mineralisation; (the BIF provides
a good chemical trap), whilst the shear zone provides the fluid
pathway as well as high strain conditions that fracture the BIF.
80
Several sections show how the resource drilling had straddled

some of the larger ore zones, which in general are about 25 m
wide, the same spacing as the original drilling pattern. These
cases highlight why the original resource model and the grade
control data do not reconcile.
The dip of shear zones plays a critical role in controlling ore
distribution in the deposit. There is a strong correlation between
higher grade and lower dip in the mineralised shears, indicating
that mineralisation was during reverse movement on the shears (a
zone of dilatancy and therefore increased fluid flow forms along
flatter parts of shears).
GRADE CONTROL AND MINING

A pie tray collects samples from blast holes by a sampler. The
blast holes are usually 5 m in depth with some subdrill in fresh
material. Two samples are taken from each hole, one at the 2.5 m
mark and one at the 5 m mark. No subdrill is sampled. Each blast
hole collar is individually surveyed and geologically logged.
There are currently well over 100 logging codes which identify
combinations of the rock types identifiable including BIF,
porphyry volcanics, sediments and shears; the intensity of
alteration (1-weak to 3-strong) and presence of pyrite.
Previously the lack of detailed working models of the
structural geometry of the shears, BIF sequences and the general
fluid flow mechanics meant that grade control was carried out
by the numbers. The geologists lacked the tools to be able to
fully interpret the grade control geology logging information and
how this was related to the assays being returned from the
samples. This was particularly relevant in oxide where the shears
were difficult to locate within the mineralisation envelope due to
masking by the supergene dispersion. Figure 6a represents an
early oreblock plan produced without a geological model
influencing interpretation of mineralisation. This has resulted in
discontinuous oreblocks where over-selectivity has occurred.
FIG 4 - Pattern of shears in the Sunrise Deposit, based on the sectional interpretation after correlation to level plans.
The challenge mine geologists faced was to integrate the

improved geological and mineralogical models into the
understanding of the small-scale (4 m 5 m drilling grid)
structural and lithological trends. By interrogating the new
resource models, mineralisation could be related specifically to a
named shear zone or a pod, or observed to be part of a specific
named BIF. Figure 6b shows an oreblock plan where the
orientation of mineralisation of the shears has been taken into
account and more continuity between high grades has been

allowed. The resource model provides confidence when dealing
with isolated gold assays in grade control data. Sporadic but high
grades found in only one or two samples away from main fluid
pathways are probably extremely narrow veins. The probability
of economic extraction of these veins without excessive dilution
is very low.
81
FIG 5 - Contoured grade control at 0.5 g/t for section 70 000 N, Sunrise.
FIG 6 - Comparison of oreblock plans, note the more continuous oreblocks in B.
82
Mapping
NEW BENCH
As familiarity with the named lithological and structural features

that are relevant to mineralisation increased so did the quality of
pit wall and floor mapping. The objective of mapping expanded
from a predominantly geotechnical focus to include the mapping
of features directly related to geological and mineralisation
features presented in the resource model. The geologists were
then able to visualise themselves in the 3D model while they
were pit mapping. When pit mapping, geologists compare the
resource model to both structures observed and marked
oreblocks. This has been crucial in increasing enthusiasm for
collection and processing of mapping data. The mapped
information is transferred in the office to base maps where
variations in expected orientations of structures, and new features
are investigated to understand their significance. If unexpected
ore occurs during grade control, or if ore which is expected does
not appear, the mapping is directed to the area of concern. The
information is then transferred to hard copies of the sectional
interpretation to ascertain if the interpretation is correct. If
updating is required, the digitised sectional linework is modified
ready for the next round of resource modelling.
The most important aspect of the mapping is to determine
boundaries as accurately as possible. To achieve this requires
face mapping as well as flitch mapping. Additional structural
data from the face mapping is also important, and interpretation
of the presence of folded BIF and overturned BIF will rely on
face mapping. Using a digital camera has been extremely
important to check interpretations (Williams, 1999). Digital
photographs also preserve the geology and are later
manipulated to produce panoramic images with geology marked
over the top.
The relationship between the model and bench mapping is
sufficiently reliable to use the model predictively between
benches, as the model is continuously updated by routine bench
and face mapping. This means that a routine of mapping plus
model upgrading can potentially replace detailed blast hole
logging to control modelling blast hole assay data and
determining mine block specific gravity. Currently blast hole
logging is still being carried out.
This methodology has significant potential for savings,
because the geological model remains current for future resource
modelling, the time involved in routine mapping/logging is
reduced, and the geological understanding of the deposit remains
current at all times. The methodology is summarised in Figure 7.
Previous resource estimates at Sunrise have included manual

polygonal estimates performed by Delta Gold in August 1992
and June 1993, and inverse distance squared estimates by PGS in
December 1993, July 1994 and June 1995. Indicator Kriging has
been used for the most recent models in June 1996, March 1998,
September 1998, November 1998 and June 1999. Over time the
resource models have been completed by both external Placer
personnel and site based personnel. The development of a
detailed geology model and an increase in site based skills has
meant the resource modelling processes and responsibility is now
site based with the mining geology department.
To create the geological block model for resource estimation
the sectional line work for both the geology and fluid flow
interpretations are digitised into separate layers. These layers are
then converted to polygons with the intention of connecting each
of the separate BIF and shears into discrete triangulation. (Gotley
et al, 1998).
Achievements
Shear triangulations
The mapping procedure represents both an improvement in

efficiency of mining geology personnel, and has increased the
ability of the mine geologists to rapidly exploit any geological
variability in the deposit. In the longer term, when the deposit is
mined, a comprehensive geological archive for Sunrise, will be
available for future research. This archive will be fully digital
(Williams, 1999).
Mine geologists are now constantly thinking critically about
the deposit in terms of the resource model interpretation and how
it is related to what they see in drill core, chips, pit walls, pit
floors, grade control assays and drilling assays. There is a
heightened sense of ownership towards the resource model and
ensuring it is as accurate as possible.
The communication between development geologists and mine
geologists has also benefited as features which can be seen at
depth in drill core or chips, and perhaps difficult to understand in
a 3D sense, can be related to actual pit exposures. All geologists
are now speaking in the same terms according the model. This
will mean the interpretation of new drilling will be intimately
linked to the current geological understanding gained through
mining.
The Sunrise Shear and nine hanging wall shears are relatively
straightforward to model in 3D. The hanging wall shears vary
from having some thickness (usually less than 10 m) to being
represented as fault planes. The Sunrise Shear is a major
structure up to 30 m across in places, with an average thickness
of approximately 20 m. The shear zones triangulations are
updated regularly as new drilling is completed.
Before bench mined

PLOT BENCH PLAN FROM
SECTIONAL/3D MODEL
UPDATE BIF AND SHEARS FROM
BLASTED GROUND
MODIFY
3D
GEOLOGY
MODEL
Daily
UPDATE BENCH PLANS AS

NECESSARY
Daily
MODIFY SECTIONS AS
NECESSARY
After bench
mined
COMPLETE BENCH PLANS
Archive
COMPLETE FACE MAPPING

MODIFY SECTIONAL
INTERPRETATION
FIG 7 - Flowchart of mapping methodology for the Sunrise deposit.
RESOURCE MODELLING
BIF triangulations
The geometry of the BIF units is extremely complicated due to
the shearing and thrusting which has occurred and the large
amount of inferred movement along two of the major shears.
Owing to this complexity, individual triangulation of each of the
19 BIF units has not been completed. Initially, the polygons were
extruded halfway to the next section to create 3D solids. The
stratigraphy is however not orthogonal to the sections so there
was a stepped effect. This was not ideal therefore the stratigraphy
was separated into six BIF dominated or volcanic dominated
units that were more suitable for resource estimation. In this case
the boundary surfaces were either the shears with large
83
displacement, or the top or bottom of a BIF unit which was

suitable for 3D modelling. This latest method has been continued
with all subsequent models until the present (January 2000).
Benefits from the new model

The new resource model has the benefits of:
closer estimation to actual spatial location of ore, which has

assisted in daily planning of drilling patterns;
Fluid flow model

The alteration study carried out in conjunction with the
geometric model development showed that the alteration halo
around individual veins was very narrow. Alteration haloes
around mineralised shears are wider, but still directly related to
the thickness of mineralisation. A 0.1 g/t grade cut-off was found
to be very close to the visible alteration halo (Ferguson et al,
1998). Because of these relationships, the grade data can be used
reliably to map the effects of fluid-rock interaction, and it is this
envelope that has been used to define the effective pathways of
the mineralising fluid.
The fluid flow model was interpreted onto 29 sections that
were digitised then triangulated into a 3D solid. This solid was
used to constrain the high-grade mineralisation. A typical
example of the grade envelopes developed is shown in Figure 8.
The flow model assumes that the fluids moved from deeper in
the deposit upwards, and that the major feeder system is located
west of the Sunrise lease. This latter assumption needs to be
tested by deeper drilling at Sunrise, as there may be multiple
deep feeder zones to the Sunrise system. The models generated
(for example Figure 8) define the local flow directions, and point
to areas where additional pod or similar linkage structures may
occur.
Modelling
The geological block model was developed from the
interpretation. To obtain the necessary definition from the narrow
shears, this model was created with a small cell size of 5 m East
5 m North 1 m Elevation, then reblocked to a larger block
size of 5 m 5 m 5 m for use in the resource estimate (Gotley
et al, 1998).
separate domaining of discontinuous, high-grade, pod style
mineralisation and continuous lower grade mineralisation

within the shear zones. Controlling the grade estimation is
made easier as the variography and the physically observed
ranges of these two styles are very different even though
mineralisation is considered to be contemporaneous,
occurring during and before the end of shearing (Ferguson et
al, 1998);
3D geological model that can be easily interrogated to
compare modelled geology with actual geology and predict
shearing intersections on pit floors and walls;
having the shears as 3D solids has meant a better
understanding of geotechnical issues that have occurred
previously, including a small oxide wall failure;
prediction of high-risk areas for wall failures and proactive
engineering of these areas in pit wall cutback designs to
minimise risk.
Reconciliation
The new resource models created using the geological
interpretation had an immediate effect on reconciliation. The
period from the start of mining to December 1996 (mainly
transported material), shows a 30 per cent underestimation in
ounces. This increased to 70 per cent underestimation from
January 1997 to December 1997 (mainly oxide), then up to 91
per cent for January 1998 to September 1998 (mainly fresh). The
results for the geological modelling were introduced into the
resource modelling from October 1998 and for this period up to
December 1999 the underestimation of ounces has dropped to 28
per cent. Figure 9 shows the progressive reconciliation by these
groups and graphically details the effect that a concerted effort in
improving the geological understanding has made.
FIG 8 - Typical section of the fluid flow model outlines at 0.1 g/t (fluid flow is hatched), Sunrise deposit.
84
Dec-99
Oct-99
Aug-99
Jun-99
Apr-99
Feb-99
Dec-98
Oct-98
Aug-98
Jun-98
Apr-98
Feb-98
Dec-97
Oct-97
Aug-97
Jun-97
Apr-97
Feb-97
Dec-96
Oct-96
Aug-96
Jun-96
Apr-96
Feb-96
% Difference
Month
FIG 9 Per cent difference between grade control and resource models, January 1996 to December 1999, Sunrise deposit.
When the geological modelling was initiated in March 1998,

the previous three months had returned gains in ounces of over
100 per cent (Dec 1997, Jan 1998 and Feb 1998 totalled over 30
000 ounces mined over the resource model prediction).
TARGET GENERATION
The geological and fluid flow models have provided reliable
tools to improve target generation for development drilling. The
two major styles of mineralisation (ore pod and shear-hosted
mineralisation styles) are located at specific structural sites. The
best shear-hosted mineralisation is in flats in the shears, and
these can be determined from the model. Where drilling is
sparse, the presence of flats is only inferred from deviations from
average dip. The detail in the model then provides a mechanism
for determining drill spacing to locate the actual position of the
flats. Pod-style mineralisation is preferentially located close to
areas where shears diverge, and commonly link lower shears to
the upper shear through a series of steep mineralised brittle
fractures. Subtle lateral features of the thrust system also control
the location of these pods, and zones of these lateral structures
are also predictable from the geological model. As a result of
application of the model, additional pod mineralisation was
discovered at Sunrise, earlier than would otherwise have been the
case.
The relationship between structural features and mineralisation
has allowed direct interrogation of the model for areas that have
been under-explored, and this is particularly applicable to areas
where access to collar locations is difficult. The risk factors
associated with ignoring certain difficult areas can be assessed
against the structural setting of the area. The model therefore
provides a method for prioritisation of development drilling, and
links the mine planning process to the requirement to provide
access to certain critical development drilling areas.
PIT OPTIMISATION
The engineering department, in collaboration with mine geology,
uses the resource model during design stages. On the resource
model, engineering run a variety of pit optimisations, with
varying parameters including mining costs, gold price, haulage
rates, etc. Comparison of the optimisation runs provides
excellent information about areas of high risk and high potential.
Geologists then combine with engineers to use the geological
and resource model to:
produce targets for drilling to increase reserves;
confirm any critical low confidence mineralisation which the
pit may drive down on; and
produce dummy mineralisation to assess whether targeted
mineralisation will significantly affect pit optimisation.
This produces a best case pit design where both the
confidence of interpretation and potential for upside are
considered. The geologist is also given a firm understanding of
whether targets selected are critical for pit designs or whether
they can be considered to be near mine exploration. This last
point can be very important in scheduling as pit development
may mean the loss of access to collar positions for drilling if not
timed correctly.
COST EFFECTIVENESS
There is a cost associated with the development of a
comprehensive geological model for a resource, and it is
important to ensure that the model returns a significant benefit to
the operation. The benefits from the various processes described
in Figure 1 are outlined in Table 1. In this table an estimation of
an increase in revenue or a reduced cost is presented. For
additional ounces recovered, the operation will carry an
associated mining cost, so the figures represent the potential
85
TABLE 1
Cost benefits resulting from application of geological modelling.
Process
Cost ($US)
Resource modelling
Activity
Increased revenue ($US)
Increased grade by ~0.20 g/t of remaining resource which equals

44 300 oz
Geological modelling
300 000
Section analysis, maintenance
6 202 000*
0
Grade control and mining
Improved layout, better recovery high-grade ore in remaining

reserve. Improved ounces by two per cent of 450 000 ounces =
9000 oz
1 260 000
Target generation
Increased drill spacing, reduction of 60 drillholes @ 100 m, $60/m
360 000**
Pit optimisation
Increased grade by ~0.20 g/t of remaining resource which equals

26 000 oz
3 612 000
Efficiency gains
Mapping, geotechnical, block-out procedures, one hour per day

for two geologists
18 000
$300 000
$5 250 000
* Increase in resource is realised in pit optimisation

** Calculated from June 1998 to December 1999
actual saving. At an average production cost of $US150/oz and a

metal price of $US290/oz, the revenue is estimated as
$US140/oz.
Changes to the drilling scope to target specific areas and
prioritisation of drilling in relation to pit optimisation and design
requirements does provide a direct saving to the operation, as do
the efficiency gains provided through improved task definition
and scheduling. In the table, all the savings and revenue gains are
deliberately conservative, because there are always a number of
competing factors in claiming such savings. For example,
reduction in the requirement for cut-backs may carry a higher
drilling cost early in the operation, for savings realised later, and
finance costs have not been factored into this analysis.
CONCLUSIONS
A framework has therefore been created for keeping relevant
information at hand, using the resource model predictively for
grade control and keeping the geological and mineralisation
interpretations up do date with both the pit and floor mapping
and development drilling information. This also has implications
in terms of geotechnical considerations. The ultimate goal is to
have the resource model reflect accurately the expected grade,
tonnes and, possibly most importantly, the spatial location of ore.
Pit designs, waste dump designs, scheduling for mining and
milling and grade control will all be improved with a solid basis
for forecasting ore production.
86
REFERENCES
Ferguson, L, Williams, P R, Lally, J and McQuaig, T C, 1998. Sunrise
Geological Modelling. Placer (Granny Smith) report by SRK
Consulting.
Gotley, S, Haren, E, Ridge, K, Lai, J, Lewis, R W, Titley, M and
Pridmore, C, 1998. Sunrise Resource Report. Placer (Granny Smith)
internal report.
Lally, J, 1998. Fluid Flow Modelling and Target Generation, Sunrise
Lease. Placer (Granny Smith) report by SRK Consulting.
Newton, P G, Gibbs, D, Groves A, Jones, C M and Ryall, A W, 1998.
Sunrise-Cleo Gold Deposit, in Geology of Australian and Papua
New Guinean Mineral Deposits (Eds: D A Berkman and D H
Mackenzie) pp 179-186 (The Australasian Institute of Mining and
Newton, P G, 1997. Notes on the steuctural controls on gold
mineralisation at Sunrise, Laverton, Western Australia. Western
Australia. Placer (Granny Smith) internal report.
Ojala, J V, 1994. Structural and geochemical study of gold mineralisation
at the Sunrise Dam gold deposit, Laverton, Western Australia. Placer
(Granny Smith) internal report.
Ojala, J V, 1995. Structural and Depositional Controls on Gold
Mineralisation at the Granny Smith Mine, Laverton, Western
Australia, PhD thesis, University of Western Australia.
Standing, J, 1993. The structural geology, structural controls of
mineralisation and paragenisis of the Sunrise Dam gold deposit,
Laverton, Western Australia. Placer (Granny Smith) internal report
no. WA 18/93.
Williams, P R, 1999. Recommendations for Pit Mapping. Placer (Granny
Smith) report by SRK Consulting.

J R Vearncombe1 and S Vearncombe1
ABSTRACT
SpaDiS is a revolutionary approach to the spatial analysis of point data
using geometry to analyse geometry! SpaDiS uses each and every
spatial relationship and is not dependent on mathematical models. For n
2
points there are n n spatial relationships and, because of the square
function, the method yields interpretable results for both small and large
data sets. SpaDiS is an alternative to variography for directional
studies. At the regional scale, the analysis can assess distribution patterns
of mineralisation and potential controlling structures. At the deposit
scale, the characteristics of zones of mineralisation such as direction,
spacing, high-grade ore direction and grade distribution can all be
deduced.
INTRODUCTION
Point data on the spatial distribution of mineralisation can be
analysed by a plot on which distance and direction from each
data point to each other data point are recorded by a point at that
distance and direction from the origin. Termed an
all-object-separations plot, these are commonly known as Fry
plots and were developed for the analysis of strain and strain
partitioning in rocks (Hanna and Fry, 1979; Fry, 1979). Similar
methods have been applied to the analysis of inter-atomic
distances in crystals and the spatial auto-correlation of ore bodies
(Patterson, 1934, 1935; Perutz, 1942; Leymarie, 1968; de Paor
and Simpson, 1987; Lagarde et al, 1990; Allison et al, 1997;
Vearncombe and Vearncombe, 1999). We have further developed
the analytical methods of Fry (1979) and apply these to the
spatial distribution of grade as recognised in drill assay data. The
method is an alternative to directional variography.
Spatial analysis can operate manually by placing a sheet of
tracing paper on which a series of parallel reference lines
(typically north pointing on a map) have been drawn, and the
location of each data point is recorded (Fry, 1979). On a second
sheet of tracing paper with a centre point (or origin), a set of
marked parallel lines are kept parallel to those on the first sheet.
The origin of the second sheet is placed on one of the data points
on the first sheet and the second sheet marked with all the
positions of points on the first. Then the origin of the second
sheet is placed on a different data point on the first, and the
positions again recorded on the second sheet. This is continued,
maintaining the same orientation, until all points on the first
sheet have been used as the origin on the second. For n data
points there are n2-n translations. The resulting plot may be
further analysed by construction of a rose diagram recording join
frequency versus directional sector. Although the manual
technique provides totally satisfactory answers, the method is
tedious and cumbersome even for modest databases. For fast and
effective analysis the examples shown here have been analysed
using SpaDiS software.
Computerisation provides additional capabilities not readily
available to the manual user. These include the ability to handle
large data sets (which may result in hundreds of millions of
translations) and to analyse data within specified distance ranges.
Advanced analysis is achieved for data sets recording location
and a value (such as deposit endowment in regional studies or
grade in drill holes for deposit studies). The user defines success
within a selected numeric range, and only those points are
analysed. For prospect- to mine-scale studies, the spatial
1.
MAusIMM, Vearncombe and Associates Pty Ltd, 14A Barnett

Street, Fremantle WA 6160.
distribution of a sampling (usually drilling) pattern can be

analysed, and successful sampling compared per radial sector
with all sampling on a percentage translations basis, helping
deduce the directions of ore shoots.
DRILLING DATA
The spatial distribution of mineralisation as recorded by the
location of successful drill holes is constrained by the
distribution of drilling. We illustrate this with prospect-scale RC
drilling results from Canyon. Based on a real example, these data
have been re-oriented and massaged to prevent identification at
the request of the owner. For this study, data are projected to the
surface (map plane) with the drill hole intersections and
mineralisation recorded as gold grade thickness (g m/t Au).
The purpose of the analysis here is to determine directions of ore
shoots, controlling structures, the repetition and spacing of these
ore shoots.
The map projection of drill intercepts shows the location of
drilling at Canyon (Figure 1). Translations of all drilling and the
rose diagram show the decisions that have influenced drill hole
location including a knowledge of the trend of the regional
stratigraphy, a desire to drill on a systematic grid and problems
of access due to topography. Note the general east bias of the
drilling grid and the overall northwest bias in the drilling pattern
(Figure 1).
For Canyon, drilling results are analysed for relationships
within 500 m of each other (distance range 0 to 500) and for
value ranges >1 g m/t, >3 g m/t, >6 g m/t, >10 g m/t, >25 g m/t,
and >50 g m/t (Figures 2 to 5). Note that the analysis is robust
for economic grades of mineralisation. The results are initially
presented as a translations plot. This plot shows all the spatial
relationships between drill holes fulfilling the value range. The
plot is thus a mix of influences between the (human) choice of
drill hole locations and the underlying geology. This spatial
distribution plot is summarised in the absolute rose diagram of
translations. To overcome bias in the sampling we use the
relative rose.
The relative rose diagrams show the percentage of drill holes
within the defined distance range which achieve the chosen value
range relative to the total number of spatial relationships per
(10) radial sector. Thus we see through bias in the original
sampling pattern to record percentage success. The relative roses
are drawn here with an outer circle at 100 per cent and internal
circles at 20 per cent intervals. There are two types of relative
rose diagrams. The translations relative rose shows spatial
relationships as counted in the translations diagram. The
sampling rose calculates from the translations the non-linear
relationship of the number of drill holes within the value range
relative to the total number of drill holes per radial sector.
The results of the analysis at Canyon show that the data
develop a clear geometric anisotropy at >3 g m/t. This geometric
anisotropy is evident from the first showing of the northeast
directional trend in the relative rose (Figure 3). The geometric
anisotropy may be considered as similar to, although
mathematically distinct from, the range in variography. At
>25 g m/t the translations plot is no longer dominated by the
original drilling pattern and clearly shows the underlying
geology, which at >50 g m/t clearly shows ore shoots oriented to
the northeast with widths of 25 m and a spacing of about 100 m
(Figures 4 and 5). Note that for >50 g m/t spatial relationships in
the northeast sector achieve 26 per cent success and this
translates to 55 per cent success in drilling along this alignment.
87
J R VEARNCOMBE and S VEARNCOMBE
FIG 1 - Map of the drilling at Canyon showing the grade thickness results for gold in drill intercepts projected to the surface. The translations plot shows
all the spatial relationships between the drill holes and these are summarised in the rose diagram. Note that the translations have a rotational symmetry
and occupy an area four times that of the original map, but preserve internal geometry and distance relationships.
DISCUSSION
Mapping at scales relevant to a geological problem, with an
emphasis on structural geology, is the best and most effective
method of understanding the directional controls on
mineralisation. Other methods, such as SpaDiS, are of most
use in conjunction and as a complement to mapping.
There are many data manipulation methods, including
contouring of assay results (or numerical derivatives such as
grade thickness), and directional variography, by which data
can be evaluated. Contouring will remain a major tool in data
presentation, and provides a map as the final product, a feature
not available to directional variography or SpaDiS analysis.
However, contouring of small data sets is of little value, and
contouring data collected on a grid can be strongly biased by the
sampling pattern. Effective contouring of large data sets uses a
search window, usually an ellipse, based on a prior knowledge of
how the values are distributed and the geology, or is derived from
directional variography. The long axis of the search ellipse is
88
oriented along the assumed direction of maximum continuity.

SpaDiS analysis is a new way to determine the direction(s) of
maximum continuity with distinct benefits in many
circumstances, including contouring.
SpaDiS analysis and variography both utilise pairs of data in
computation, but diverge due to the statistical approach and
algebraic models in variography versus a purely geometric
approach in SpaDiS analysis. In variography, a semivariogram
(commonly referred to as a variogram) is defined as half the
mean-squared difference of pairs of values at some defined
distance (Journel, 1986; Thomas and Snowden, 1990). The
variogram is of interest because it incorporates several geologic
features such as the continuity of mineralisation, zone of
influence (known as the range), and an ability to assess the
anisotropy of mineralisation with directional variation. The
mathematical representation of variograms calculated for
different directions are determined by fitting appropriate
theoretical models (Journel, 1980), or by indicator methods
(Journel and Isaaks, 1984; Dowd, 1992).
THE SPATIAL DISTRIBUTION OF GRADE
FIG 2 - Results of spatial analysis of drilling for all holes within 500 m of each other for all value ranges (top) and for >1 g m/t (bottom). Original data are
shown in Figure 1. The translations plot shows the distribution of drill holes achieving the value range criteria (all data top and > 1 g m/t bottom). The
translations plot comprises a mix of information related to the original sampling distribution and the underlying geology. This is summarised in the
absolute rose diagram. The three numbers quoted after samples to the bottom right of each rose diagram comprise firstly the number of drill holes
fulfilling in the numeric criteria (eg >1 g m/t). Second, the number of spatial relationships within the distance range quoted in terms of the number of drill
holes. Third, the total number of drill holes in the data set. The relative rose diagrams show the percentage of drill holes within the defined distance range
per 10 radial sector which achieve the chosen value range (all data top and >1 g m/t bottom) relative to the total number of spatial relationships. Thus we
see through bias in the original sampling pattern to record percentage success. The relative roses are drawn here with an outer circle at 100 per cent and
internal circles at 20 per cent intervals. One hundred per cent of the spatial relationships have numeric values of zero or greater. There are two types of
relative rose diagram. The translations relative rose diagram shows spatial relationships as counted in the translations plot for all data compared to the
data achieving the numeric range. The sampling rose calculates from the translations the non-linear relationship of the number of drill holes within the
value range relative to the total number of drill holes per radial sector.
89
FIG 3 - Similar to Figure 2, results of spatial analysis of drilling for drill holes within 500 m of each other for all value ranges (top) >3 g m/t and for
>6 g m/t (bottom).
90
FIG 4 - Similar to Figure 2, results of spatial analysis of drilling for drill holes within 500 m of each other for all value ranges (top) >10 g m/t and for
>25 g m/t (bottom).
91
FIG 5 - Similar to Figure 2, results of spatial analysis of drilling for drill holes within 500 m of each other for all value ranges >50 g m/t.
Directional variography (Rendu, 1984) is commonly used to

determine the continuity of directions of mineralisation in mines
and advanced prospects. A published example is the Kundana
mine (west of Kalgoorlie, Western Australia) where Hadlow et al
(1993) compare critically results of directional variography with
deposit structure. Variograms were calculated in search windows
of 10 in all sectors in the plane of the orebody using several
different mathematical models, but only a limited number of
resulting variograms were robust. The robust variogram
directions were found to correspond closely to some of the
structures deduced from geological mapping and fabric studies.
However, multi-dimensional variography requires pairs of data at
multiple spacings in all sectors. It is effective only on large to
very large data sets, and even then gaps in the data can result in
erroneous results. In contrast to multi-dimensional variography,
SpaDiS uses each and every spatial relationship and is not
dependent on mathematical models. SpaDiS can produce
interpretable results with modest data sets (14 or more samples),
but typically the larger a sample of quality data the more reliable
the end result. With very large data sets (thousands to tens of
thousands of samples) we have found that the millions to
hundreds of millions of translations restrict the sensitivity, and it
is best to subdomain the data into sample sizes in the hundreds.
CONCLUSIONS
Regional applications of SpaDiS analysis include assessing
deposits and old workings to deduce structural directions
controlling mineralisation and to verify aeromagnetic
interpretations. The assessment can be according to deposit size,
to deduce variations in structural directions controlling
92
mineralisation according to deposit endowment. Mine- and

prospect-scale applications are used to determine the distribution
of successful drill holes relative to all drilling. This reduces the
spatial bias inherent in the distribution of drilling to determine
ore shoot directions. Grade control data can be examined to
deduce directions of continuity of mineralisation within open
pits.
ACKNOWLEDGEMENTS
One of us (JV) was a research student at the University of Wales
when Norman Fry developed some of the practical methods
described here. Applications at that time were to rock
deformation. Fifteen years later it became clear that the method
was applicable to directional studies and an effective alternative
to directional variography. Spatial analysis is performed using
SpaDiS for Windows and Power Macintosh, which are
developed under contract to Vearncombe and Associates Pty Ltd.
SpaDiS is an internationally registered trademark.
REFERENCES
Allison, I, de Paor, D G, Haszeldine, S, Bowman, A and Maguire, M,
1997. Cryptic structural trends in basement revealed by Patterson
diagrams: examples from the Scottish and Irish Caledonian orogen,
in: Evolution of Geological Structures in Micro- to Macro-scales
(Ed: S Sengupta), pp 487-495 (Chapman and Hall: London)
De Paor, D and Simpson, C, 1987. Rf/Fry version 2.0.5 (Software)
Instruction Manual. EarthnWare Inc, Boston, not paginated.
Dowd, P A, 1992. A review of recent developments in geostatistics,
Computers and Geosciences, 17:1481-1500.
Fry, N, 1979. Random point distributions and strain measurement in

rocks, Tectonophysics, 60:89-105.
Hadlow, H R, Khosrowshahi, S and Vearncombe, J R, 1993. The use of
directional variography and structural geology to determine controls
on gold mineralisation in the South Lode, Kundana Mines,
Coolgardie Goldfield, Australian Institute of Geoscientists Bulletin,
14:31-38.
Hanna, S S and Fry, N, 1979. A comparison of methods of strain
determination in rocks from southwest Dyfed (Pembrokeshire) and
adjacent areas, Journal of Structural Geology, 1:155-162.
Journel, A G, 1980. The log normal approach to predicting local
distributions of selective mining unit grades, Mathematical Geology,
12:285-303.
Journel, A G, 1986. Geostatistics; models and tools for the earth
sciences, Mathematical Geology, 18:119-140.
Journel, A G and Isaaks, E H, 1984. Conditional indicator simulation:
application to a Saskatchewan Uranium deposit, Mathematical
Geology, 16:685-719.
Lagarde, J L, Omar, S A and Roddaz, B, 1990. Structural characteristics
of granitic plutons emplaced during weak regional deformation:
examples from late Cretaceous plutons, Morocco, Journal of
Structural Geology, 12:805-821.
Leymarie, P, 1968. Une mthode permettant de mettre en vidence le

caractere ordonn de la distibution des gtes mineraux, Mineralium
Deposita, 3:334-343.
Patterson, A L, 1934. A Fourier series method for the determination of
the component of intra-atomic distances in crystals, Physics Reviews,
46:372-376.
Patterson, A L, 1935. A direct method for the determination of the
components of inter-atomic distances in crystals, Ziet fur
Kristallographie, 90:517-554.
Perutz, M F, 1942. X-ray analysis of Haemoglobin, Nature, 149:491-494.
Rendu, J M, 1984. Interactive graphics for semivariogram modeling,
Mining Engineering, 36:1332-1340.
Thomas, M and Snowden, V, 1990. Improving reconciliation and grade
control by statistical and geostatistical analysis, Australian Institute
of Geoscientists Bulletin, 10:49-59.
Vearncombe, J and Vearncombe, S, 1999. The spatial distributon of
mineralization: applications of Fry analysis, Economic Geology,
94:475-486.
93
The Bunyip Lateritic Nickel-Cobalt Deposit, Cawse Nickel-Cobalt

Operations, Kalgoorlie, Western Australia
A Bywater1 and S M Denn2
INTRODUCTION
The Cawse nickel-cobalt operation is situated 50 km NW of
Kalgoorlie in Western Australia, at latitude 30 2230 south and
longitude 121 0830 east on the Kalgoorlie 1:250 000
geological map sheet (SH 51-9) (see Figure 1). The project is
owned and operated by Centaur Nickel Pty Ltd (CTR) and was
the first in Australia to commercially produce nickel metal and
cobalt sulphide through pressure acid leach (PAL) technology.
Cawse has a total resource base (all categories of resources +
reserves) of 213 million tonnes Mt at 0.7 per cent Ni and
0.04 per cent Co, comprising 127 Mt at Cawse Central which is
100 per cent owned by CTR and 86 Mt at Cawse Extended
where CTR has a 80 per cent equity. Cawse Extended is located
immediately to the northwest of the Cawse Central leases. All
current reserves (proved and probable) of 30 Mt at 1.0 per cent
nickel are contained within Cawse Central. In addition the
Siberia project area 20 kilometres to the NW has a resource of
63 Mt at 0.7 per cent Ni.
Cawse is a dry laterite deposit, with nickel and cobalt hosted
by iron and manganese oxides. Cawse Central mineralisation
extends over a ten kilometre strike length, with three pits
currently being mined. The Bunyip pit has the premier nickel and
cobalt grades within Cawse Central.
There are seven dominant lithology types forming mappable
units within current pit exposures. Four of these units are host to
nickel and cobalt mineralisation. Ore types are defined as
Upgrade, Grind and Non-upgrade, depending on their physical
characteristics with respect to the treatment route.
Mining commenced in February 1998. To the end of calender
1999 the operation has produced 2100 tonnes nickel plate and
470 tonnes of cobalt as a cobalt sulphide product. Future
production will target 9000 tonnes per annum Ni metal and up to
2000 tonnes per annum Co, based on processing 500 000 tonnes
of ore per annum.
EXPLORATION HISTORY
In 1895, a prospector named Cawse discovered gold in quartz
veins from granite outcrop in an area now called Bunyip. In the
late-1970s, Western Mining Corporation (now WMC Resources)
outlined a laterite resource of approximately 30 Mt grading 1.3
per cent Ni and 0.08 per cent Co at Siberia, approximately 20 km
NW of Cawse and hosted within the same geological unit as the
Cawse deposit (Loftus-Hills, 1975). Small pits were developed in
this area from 1978 - 1980, supplying feed and silica flux to the
Kalgoorlie nickel smelter (Marston, 1984). Newcrest Mining
mined gold from a small open pit adjacent to historic gold
workings in the late-1980s. In 1992, Centaur Mining and
Exploration Ltd applied for a group of prospecting licences to
explore for gold in palaeochannels, as mined at its Lady
Bountiful Extended Operations since 1988. Drilling commenced
at Cawse in March 1993 with assays returning significant Ni
values. Analysis for nickel, cobalt and copper returned a best
1.
Senior Geologist, Cawse Nickel Operations, Centaur Mining and

Exploration, PMB 32, Kalgoorlie WA 6433.
2.
MAusIMM, Chief Geologist, Cawse Nickel Operations, Centaur

Mining and Exploration, PMB 32, Kalgoorlie WA 6433.
intercept of 8 m at 2.2 per cent Ni from an area now called Orc

pit some 2.5 km north of the current Bunyip pit. The recognition
of laterite potential led to further drilling in the area during
1994 - 1995, resulting in definition of a resource of 50 Mt at 1.0
per cent Ni. A fast-tracked project incorporated infill drilling and
metallurgical testwork, culminating in a successful bankable
feasibility study in 1997.
REGIONAL GEOLOGY
Cawse is situated within the Ora Banda Mt Pleasant area of the
Eastern Goldfields district in the Yilgarn Block. The geology is
dominated by an Archaean greenstone sequence folded around
intrusive granitoids. Nickel mineralisation is exclusively
developed on ultramafic and differentiated mafic sequences
within the greenstone sequence. The Cawse deposit is formed
above an olivine adcumulate-textured dunite on the eastern edge
of the Walter Williams Formation (WWF), part of the Linger and
Die Group within the Ora Banda domain (Brand et al, 1996,
Figure 1). Regional greenstone stratigraphy is similar to that of
the Kambalda and Kalgoorlie areas as shown in Table 1, the
stratigraphic column for the Siberia-Ora Banda area. The
greenstone belt defines the Goongarrie-Mt Pleasant anticlinal
structure around the Goongarrie granitoids, which intrude the
base of the greenstone sequence. At Cawse the basal basaltic
sequence, the Pole Group, is absent, attributed to the intrusion of
the Goongarrie Granitoid.
Nickel-cobalt mineralisation is confined principally to the
ultramafic rocks of the WWF, bounded by the Goongarrie
Granite to the east, and the Siberia Komatiite to the west.
Localised nickel ore grades have also been reported from the
Siberia Komatiite and peridotites from the differentiated mafic
Ora Banda Sill.
The WWF is up to 300 m thick, comprising coarse-grained
olivine adcumulate and orthocumulates. Thin orthocumulates
occur at the base and top of the formation. The ultramafic has
undergone serpentinisation in the Cawse area. The Siberia
Komatiite overlies the WWF and is separated from the upper
orthocumulate by a zone of pyroxene, gabbro and Mg-rich
leucogabbro (Witt and Harrison, 1989). It is characterised by thin
spinifex-textured komatiite flow units. Overlying the Siberia
Komatiite are high-Mg basalts intercalated with tholeiitic basalts.
The sequence is completed by a succession of epiclastic
sedimentary rocks (Witt and Swager, 1989). Intrusive layered
mafic-ultramafic sills, termed the Ora Banda and Mt Pleasant
Sills occur above the Siberia Komatiite.
The earliest deformation recognised in the Cawse area is the
regional D2 structure called the Goongarrie - Mt Pleasant
anticline. The granite greenstone contact in the Cawse region is
offset by north-trending strike faults and is displaced by
north-east-trending regional (D3) strike-slip fault with a small
dextral component. Drilling has indicated that zones of nickel
enrichment in the regolith at Cawse are, in part, controlled by
primary structures trending ENE-WSW; these D3 structures are
subperpendicular to the regional strike. The regional drainage
pattern is also subparallel to these structures.
The sequence has undergone low to middle grade greenschist
metamorphism, with local upper greenschist to lower
amphibolite grade facies associated with a high strain
deformation aureole close to the granite margins (Witt, 1993).
95
A BYWATER and S DENN
FIG 1 - Cawse nickel project regional geology and location plan.
96
THE BUNYIP LATERITIC NICKEL-COBALT DEPOSIT
TABLE 1
Stratagraphic column for the Siberia-Ora Banda area.
Group
Formation
Thickness
Granitoids
Black Flag Group
Grants Patch Group
Linger and Die Group
Pole Group
Description
Fine to coarse graine , granodiorites and monzogranites.
Kurrawang Formation
> 2000 m
Quartz rich silts, sandstones and conglomerates.
Black Flag Group (undifferentiated)
2000 m
Epiclastic felsic to intermediate volcaniclastics.
Orinda Sill
250 m
Gabbro.
Pipeline Andesite Member
200 m
Fragmental intermediate volcaniclastic rock.
Ora Banda Sill*
2000 m
Layered mafic-ultramafic sill, peridotite to granophyre.
Victorious Basalt
1000 m
Massive to pillowed coarse grained plagioclase-phyric basalt.
Bent Tree Basalt
2000 m
Massive to pillowed, fine grained, basalt.
Mt Pleasant Sill
600 m
Differentiated mafic sill includes an iron rich granophyric unit.
Mt Ellis Sill
600 m
Differentiated mafic sill.
Big Dick Basalt
500 m
Variolitic textured massive to pillowed high Mg basal.
Siberia Komatiite*
2600 m
Olivine spinifex textured flows.
Walter Williams Formation*
200 m
Coarse cumulate textured dunite to peridotite.
Missouri Basalt
1000 m
Tholeiitic basalt.
Wongi Basalt
> 2000 m
High Mg basalt.
Adapted from Witt 1994

* Contains ore grade nickel ore intersections
LOCAL GEOLOGY
Nickel-rich laterite has formed as a residual product from
chemical weathering of olivine-rich cumulate rocks. The Cawse
deposit is an oxide type laterite nickel deposit. Nickel is hosted
by iron and manganese oxides. Silicate minerals of the lower
saprolite ore type host small portions of the orebody near the
base of weathering. The interaction of structural and regolith
features form significant control on the definition of high-grade
nickel and cobalt ore deposits. Mineralisation is hosted within
the residual portion of the laterite profile, although small zones
of enrichment occur at the base of transported material (Brand et
al, 1996).
Serpentinised olivine cumulates of the WWF form the
protolith to the Cawse deposit. Most of the saprolite and bedrock
are derived from an adcumulate dunite. The unit strikes
north-west to the south of Cawse Central, then north-north-west
at Cawse Central and north-north-west at Cawse Extended. A
olivine orthocumulate has been identified from limited bedrock
data at both lower and upper margins of the adcumulate. All
ultramafic lithologies have undergone a degree of
serpentinisation, with primary fabric still preserved except in
fault zones.
The irregular eastern margin of the WWF abuts the granite at a
faulted contact with associated shearing and talc carbonate
alteration. The ultramafic is composed dominantly of forsterite
that has been serpentinised to antigorite and lizardite with minor
magnetite, chromite, magnesite, dolomite, talc, chlorite and
silica.
The nickel abundance of the serpentinite in the WWF varies
from around 0.19 per cent Ni for orthocumulate to 0.24 per cent
Ni for adcumulate (Hill et al, 1987), with local values up to 0.35
per cent Ni.
The protolith is enriched in olivine-compatible elements and
depleted in olivine-incompatible elements as detailed in Table 2.
No economic concentrations of nickel sulphide mineralisation
have been identified to-date from the WWF in three decades of
nickel exploration in the region with at best trace quantities of
primary nickel sulphides in the form of pentlandite.
TABLE 2
Olivine compatible and incompatible elements in the Walter
Williams Formation.
Mean olivine-compatible
elements
Mean olivine-incompatible
elements
42.35 wt % MgO
0.23 wt % Al2O3
36.60 wt % SiO2
0.02 wt % TiO2
8.33 wt % Fe2O3
8 ppm Cu
100 ppm Co
< 5 ppm Zr
820 ppm Mn
1 ppb Pt
9 ppb Ir
1 ppb Pd
12 ppb Os
3 ppb Rh
7 ppb Ru
MINE GEOLOGY
The mine geology can be divided into seven individual units on
the basis of lithology, mineralisation and position in the regolith
profile (Figures 2, 3, 4 and 5). Boundaries represent stratigraphic
horizons and the accumulation of minerals due to a specific
redox environment. These units are described below in detail.
Lateritic duricrust
A veneer of lateritic duricrust averaging 2 m in thickness is the
upper most horizon of the region. It is dark brown-red and
composed of iron-rich pisoliths and nodules (commonly
magnetic in the first metre) set in a loam matrix with occasional
fragments of silica and rounded quartz grains. These siliceous
fragments commonly show dissolution fabrics and may be
rimmed with secondary iron oxides. Pedogenic carbonates are
present in the top metre, decreasing with depth. In drill core, soil
97
FIG 2 - Regolith and mineralisation profile, Bunyip Pit.
98
10 600mE
10 800mE
10 600mE
10 800mE
18 400mN
18 400mN
18 200mN
18 200mN
90
85
LEGEND
Alluvial Zone
75
18 000mN
Manganese Blanket
Massive Silica
Pit Outline 30m depth
Talc Zone
Upper Saprolite Zone
0.6% Ni Contour
40
Lower Saprock Zone

Saprock
1.0% Ni Contour
150
0
metres
Fault
40
18 000mN
LEGEND
17 800mN
Strike and Dip
150
1.4% Ni Contour
metres
0.1% Co Contour
17 800mN
0.3% Co Contour
Pit Outline 30m depth
FIG 3 - Bunyip geological plan, 30 m below surface.
FIG 4 - Bunyip plan nickel and cobalt grade contours, 30 m below

surface.
10 600mE
10 700mE
20m
20m
40m
40m
60m
60m
Upper Alluvial Zone
Massive Silica
Lower Saprock Zone
Lower Alluvial Zone
Talc Zone
Saprock
Manganese Blanket
Upper Saprolite Zone
Fault
50
metres
FIG 5 - Bunyip geological cross-section 18 000 mN.
99
displays a gradational boundary to weakly mottled clays.

Magnesium and calcium values are often elevated (~0.6 per cent
and one per cent) in the duricrust because of calcrete formation
near surface.
Alluvial
Alluvial cover is widespread throughout the Cawse region as
represented by palaeochannels, and it is divided into an Upper
and Lower zone. One channel defines the axis of Bunyip pit,
varying in width from 20 m in the south to 200 m in the north.
The channel trends 005 with localised meandering. The
palaeochannel is situated in part adjacent to major structural
lineaments which have focussed weathering and fluid movement
in the Bunyip Pit area.
The Upper Alluvial Zone has undergone strong mottle
development that differentiates it from the Lower Alluvial Zone.
Mottles are composed of bleached, green and red-brown
kaolinitic clays and round iron rich pisoliths. This zone extends
from a few metres below surface to a depth of 20 m, and deeper
adjacent to structures. Pisoliths range in size from 0.1 - 3 cm
diameter, with maghemite also common in transported material.
The Lower Alluvial Zone is more variable in composition and
can be difficult to distinguish where no coarse fraction is present.
The base of transported sediments may comprise conglomerate,
sands and silts, or undifferentiated clays. Bedding has been
observed in some of the basal units, in which flat lying sandstone
and conglomerate units measure up to 1 m in thickness. Channel
fill is poorly sorted subrounded to angular pebbles and cobbles.
Large ovoid silcrete pods are also present measuring 10 to 40 m
in length and up to 20 m in width. The silcrete is a very
fine-grained matrix of silica containing irregular shaped clasts of
coarser silica and rock fragments measuring 1 mm - 200 mm.
This zone is highly porous.
Manganese Zone
A subhorizontal blanket of cobalt-nickel-manganese enrichment
is situated 10 - 20 m beneath the surface. The manganese zone is
hosted within the lower alluvial profile and the underlying Upper
Saprolite horizon. The blanket dips parallel to the boundaries of
the palaeochannels developed over the north trending fault zone.
Manganese minerals adhere preferentially to siliceous material as
coatings and infilling in small cavities. Chalcedonic quartz rich
zones contain numerous voids 1 mm - 2 m in size.
The Manganese Zone occurs at a redox front associated with
high fluid porosity at the alluvial- upper saprolite contact. The
manganese is present as manganic oxides and hydrates including
pyrolusite, baumite, nirnessite, jacobsite, groutite, todorokite,
chalcophanite and cryptomelane, identified by XRD analysis.
These minerals are associated with cobalt and nickel grades from
0.3 per cent to 5.0 per cent. Nickel concentrations are generally
two times higher than cobalt. Nickel and cobalt appear to be
absorbed by the manganese oxides rather than undergoing ionic
substitution.
Upper Saprolite Zone (Limonite Zone)

This goethite rich unit is located between the manganese blanket
and Lower Saprolite, measuring up to 40 m in thickness. The unit
is variable in composition in terms of iron, kaolin and silica. It is
characterised by a strong brown-mustard colour, low density and
high-porosity. Secondary iron oxides (goethite and haematite)
result in red, brown and yellow colour development. Textures are
vuggy due to fabric collapse caused by mineral breakdown and
remobilisation. The resulting enhanced fluid permeability has
facilitated the deepening of the redox boundary. The boundary
between the Lower and Upper Saprolite is marked by a sharp
colour change from light brown to yellow brown. Goethite, clay,
100
secondary quartz, minor magnetite and iron magnesium

silicates are the principal minerals. The clay mineralogy of this
zone includes kaolinite, montmorillonite, nontronite, magnesite
and talc. Structurally controlled talc alteration is found within
and adjacent to limonite zones. These units host variable nickel
mineralisation. Chalcedonic quartz is distributed throughout the
Upper Saprolite Zone, appearing as iron stained masses
measuring 1 mm - 5 m in size. The amount of quartz is highly
variable, measuring between ten to 90 per cent of the rock mass.
Where secondary chalcedonic quartz development appears
greater than 50 per cent of the rock mass, the local term massive
silica is used to describe the rock.
Massive Silica Lithotype

Massive silica is distributed throughout the upper saprolite
profile at Bunyip pit, reflecting intense weathering of the primary
ultramafic. This secondary alteration is iron stained with Ni
values of ~0.4 per cent. The massive silica zone in south Bunyip
measures 300 m in strike, 40 - 100 m in width, and occurs
variably from 15 m - 40 m below surface. The silica contains
voids 0.5 mm to 2 m in size, and may include small clasts of
laterised ultramafic material. It often retains primary adcumulate
textures, as represented by interstitial iron staining around
individual olivine crystals. This silica is a product of the
pervasive weathering of the primary ultramafic host. Silica
flooding occurs throughout regolith units, including the
Manganese blanket and Upper to Lower Saprolite.
Lower Saprolite Zone

The Lower Saprolite Zone is situated at the base of the
weathering profile at the transition between the Upper Saprolite
and Saprock Zone. The unit is defined as partially weathered
ultramafic with altered serpentine and visible relict igneous
textures. The unit comprises limonite stained nontronitic clays
and is host to minor nickel mineralisation. The unit is commonly
sheared and jointed, resulting in a blocky fabric. It is typically
developed below 40 m depth, with an average thickness of 10 m.
The preserved adcumulate fabric becomes progressively
destroyed upwards through the unit. Carbonate veinlets, rimmed
with magnetite and maghemite, are still present in the Lower
Saprolite where as magnetite alters to goethite in the upper
saprolite. Silicification of saprolite with adcumulate fabrics are
preserved throughout the Saprolite zone and associated with
minor coatings of manganese oxide. The upper boundary of the
Lower Saprolite is marked by a thin accumulation of soft green,
bleached iron-magnesium silicates.
Saprock Zone
The Saprock Zone is defined as primary ultramafic with less than
than 20 per cent of its volume affected by weathering. The
material has undergone serpentinisation but retains primary
adcumulate textures. The lower boundary of the Saprock zone is
marked by oxidation of joint filling thick subhorizontal
magnesite veins 0.5 - 4 cm in thickness. Mineralogy comprises
antigorite, magnesite, magnetite and minor chromite. The
serpentine exhibits a mesh texture and locally a crackle breccia
texture with internal, minor finely disseminated magnetite grains
(Brand et al, 1996). Antigorite has pervasively replaced olivine
crystals, with iron staining occurring along the edges of original
grain boundaries. Dolomite and magnesite veinlets occur locally
in the adcumulate forming veinlets, with some dolomite in the
serpentinite mesh cores. Parts of the Saprock contain tremolite,
chlorite and antigorite as the dominant minerals, suggesting a
source rock composition of peridotite. Gabbro units comprise
plagioclase and amphibole, with minor clinopyroxene and
sphene. Primary sulphides identified (trace only) are fine-grained
chalcopyrite, pentlandite and pyrite.
10 600mE
10 700mE
20m
20m
40m
40m
60m
60m
Pit Outline
1.4% Ni Contour
0.6% Ni Contour
0.2% Co Contour
1.0% Ni Contour
Geology Contacts
0
Fault
50
metres
FIG 6 - Bunyip cross-section nickel and cobalt grade contours, 18 000 mN.
MINERALISATION AND CLASSIFICATION OF

ORE TYPES
Ore with grades greater than 0.6 per cent Ni and 0.1 per cent Co
is situated above or adjacent to the Upper and Lower Saprolite
Zones. This reflects the significance of weathering acting as the
dominant control for all mineralisation. No ore has been
identified within the Saprock or Upper alluvial zone.
Three ore types are defined in the Bunyip deposit: Grind,
Upgrade and Non-Upgrade (Figure 2). Grind ore is classified
as material that will not undergo grade beneficiation through an
upgrade circuit and hence requires further grinding to allow
further processing. This ore type includes most high-grade cobalt
mineralisation associated with the manganese zone. Upgrade ore
comprises limonite ore derived from the upper saprolite that is
processed through the upgrade circuit. This material may also
include a small proportion of lower saprolite (weakly smectitic
clays). Upgrading is possible via screening out coarse low-grade
material, usually comprising silica. Non-upgrade material is
defined as nickel bearing material not amenable to processing
through the upgrade circuit. This is due to the nontronite content,
and results in high-grade nickel ore balling and being passed as
reject material.
Grind ore: silica-cobalt mineralisation

Grind ore is defined as material requiring fine grinding to extract
nickel and cobalt, and is locally termed siliceous cobalt
mineralisation (SICO). This material cannot be processed
through the beneficiation circuit due to the association of
massive chalcedonic silica and manganese, nickel and cobalt
mineralisation. This ore type is hosted within the manganese
blanket. Siliceous cobalt mineralisation represents ten per cent of
the measured resource and is the primary source of high-grade
cobalt feed for the Cawse plant. This ore is enriched in nickel,
cobalt and manganese, commonly reaching grades between 0.3
to four per cent Ni, 0.1 to 1.5 per cent Co, and two to ten per cent
Mn (Figures 4 and 5). This ore is characterised by manganese
staining of silica and clays, and is readily identifiable in the field,
ranging from dark black to steel grey-blue in colour. The zone is
typically situated 15 m below surface, is 20 - 120 m in width,
averages 6 m in thickness and trends north-south intermittently
above the limonite ore zones. The eastern side of the SICO
appears to be defined by a zone of barren chalcedonic quartz and

a talc shear, which appears to have acted as a barrier below and
east of the main lode. The cobalt blanket dips westward and
pinches out with depth, into limonite ore. The SICO zone
appears displaced downwards in some areas by ~5 m, which may
be due to slumping of the underlying residual material.
Upgrade ore: goethitic nickel mineralisation

Limonite mineralisation comprises approximately 80 per cent of
the global resource. Limonitic clays with variable proportions of
vuggy, geothitic silica typically contain grades of 0.4 per cent to
1.6 per cent Ni. Limonite ore comprises predominantly goethite
with minor hematite within a matrix of silica. The ore appears
bright orange in colour, and is friable in hand specimen. This
zone is situated immediately below the Grind ore in the Upper
Saprolite Zone, and transgresses into the top of the Lower
Saprolite. The dominant ore zone extends over a 1.2 km strike
length, measuring 30 - 120 m in width, up to 40 m in thickness,
with typical ore grade intercepts of 30 m @ 1.3 per cent Ni and
0.07 per cent Co. The limonite ore comprises clays with variable
silica content and minor magnesite and nontronite. Thin
subparallel talc shears transect the unit and are variably
mineralised.
The main ore zone is bounded towards the east by a massive
silica pod, which is related to a low angle footwall shear dipping
towards the west. Mineralisation is bounded to the west by
Lower Saprolite material and bedrock. The ore zone pinches out
with depth into lower saprolite and bedrock material.
Goethite-hematite is the dominant nickel host. Numerous
nickel silicate minerals have been identified, including
brindleyite, dwornikite, nepoutite, nimite and tetrataenite. These
are usually of lesser importance as nickel host minerals than
geothite.
Nickel is strongly enriched in upgrade ore, with concentrations
reaching four to five times those in the protore, and is associated
with cobalt in the upper part of the horizon. Nickel grade
decreases as the proportion of silica increases. Cobalt content is
between 0.01 to 0.1 per cent but can reach 0.3 per cent where
manganese staining occurs. Talc mineralisation within limonite
ore comprises approximately five per cent of the resource and is
located toward the base of the weathering profile, proximal to
penetrating shear structures. The talc zones in the residual
101
regolith have elevated nickel grades (>2 per cent) along upper
and lower margins. Minor hangingwall and footwall talc shears
are similarly enriched, but are thinner and show a more restricted
dispersion of nickel.
Variable amounts of silica are distributed throughout the
limonite ore as clasts ranging from 1 mm - 2 m in diameter. This
material contains Ni values between 0.4 to 0.8 per cent. The
presence of silica distributed throughout the matrix facilitates
upgrading of the ore.
Non-upgrade ore
This ore type commonly occurs adjacent to weakly weathered
saprock on the western side of the deposit. The ore strikes NNW
in a zone 10 - 20 m in width, with typical grades of 0.8 to 1.4 per
cent Ni and 0.04 per cent Co. Mineralogy comprises iron and
nickel-rich smectite clays, termed locally as nontronites. It is
characterised by a blocky nature and green colour, due to the
abundance of these smectite clays. Various degrees of weathering
in this unit influence the degree of upgradability which is
dependant on silica and magnesium content. This ore marks the
transition zone between less weathered Lower Saprolite with
strongly weathered Upper Saprolite Zone. It is characterised by
elevated Mg values >4 per cent. This ore type is not a significant
part of the Bunyip resource, however it is similar to that found at
Bulong (Elias et al, 1981). Just as Ni2+ substitutes for Fe2+ in the
limonite zone, it has a similar capacity for cation exchange with
Mg 2+ in the saprolite.
ORE GENESIS
An interpretation of the geomorphology at the time of orebody
formation is essential in clarifying ore genesis. Central to this
interpretation is a palaeochannel, which has been recognised at
all deposits within the Cawse Central area. The presence of
sedimentary sands and conglomerates suggests that the
paleographic setting of the orebody was a valley around a
high-energy alluvial system. This alluvial system developed over
a locally significant north-south trending fault. Smaller creeks
are interpreted to have fed the larger system from the east over
the Massive Silica unit. To the north of Bunyip Dam, a wide, 20
metre thick massive mottled clay deposit signifies the opening of
the system northwards into a larger low energy river or lake. The
saprolite situated west of the alluvial system and main fault
represents the remainder of what was probably a hill or scarp that
has subsequently been eroded down to the current surface level.
The surface of the massive silica, eroded by the smaller creeks
and displaying much deeper weathering, is interpreted as
representing the pre-weathering surface.
Laterite formation occurs in warm climates associated with
large influxes of water and is associated with intense weathering.
The unweathered ultramafic unit with background levels of
0.25 per cent Ni and 0.02 per cent Co, is the primary metal
source for Cawse mineralisation. During weathering,
magnesium, iron and nickel are dissolved from the primary rock.
Minerals with high cation exchange capacities, such as
secondary serpentine, smectite clays and iron-manganese oxides
(Lawrance, 1996) preferentially absorb nickel released from
weathering of primary minerals.
Cobalt is commonly found in the clay minerals and associated
with trace amounts of manganese. Cobalt combines selectively
with Mn3+ and Mn4+ (Hotz, 1964) in manganese minerals due to
adsorption. Serpentine breaks-down during weathering to form
iron oxides (in saprolite) and silica. Silicic acid formed at the
redox front as a decomposition product, reacting with
magnesium rich rocks to form magnesium ions and silica.
However if aluminium is present nontronitic clays will tend to
form (Lawrance, 1999).
102
All economic mineralisation at Bunyip occurs above or

adjacent to the Lower Saprolite-Saprock contact. The major
north-south trending shear has acted as a secondary control on
mineralisation. This shear defines the boundary between massive
silica to the east and iron-rich Lower Saprolite to the west. It is
overlain by a major palaeochannel following along the pit axis.
This has influenced downward permeation of fluids and ore
accumulation at redox boundaries, resulting in nickel and cobalt
mineralisation. Interfaces between lithological units, geological
structures and redox boundaries control ore deposition and
grades. Leaching, re-distribution and accumulation may still be
occurring in the regolith profile today because of current aridity.
Structural controls on mineralisation

Numerous structural features have been identified at Bunyip
which influence ore deposition. Variable weathering of different
rock types has resulted in a complex alteration pattern. The
dominant mineralisation control in Bunyip is a north-south
striking major fault dipping 40 towards the west (Figure 4). This
shear marks the eastern-most limit of high-grade nickel
mineralisation, at the boundary between a mineralised
hangingwall and a barren silicified footwall. This zone contains
the highest grade nickel and deepest known ore extent in the
Cawse region.
A zone on the eastern pit edge has been identified striking at
120 with a dip of 80 towards the NNE, over a 25 m wide area
(Figure 3). This area comprises numerous parallel faults that
disrupt and control the SICO ore zone in some areas. Some
blocks of SICO appear incorporated into the fault zone. The
underlying silica zone is also faulted, with clay infill often
defining fault planes.
A major talc shear exists in the eastern wall, striking 120 with
a subvertical dip. The zone measures 60 m in width with a 200 m
strike extent and is variably mineralised. A fault occurs on the
western limit of the mineralised Upper Saprolite Zone. The fault
strikes at approximately 320 - 340 with a dip of 85 east.
Nontronite is commonly situated adjacent to this fault.
MINING
Mineral resource and ore reserve estimates
The Cawse Central mineral resource estimate is based on 1545
reverse circulation and 34 diamond core holes for a combined
84 830 m of drilling. In addition a total of 14 large diameter
holes for 360 m advance were drilled for metallurgical sampling
(Denn, 1998). Most holes were vertical and sampled over 2 m
intervals. Elements were analysed utilising multi-acid digest with
grade determinations by atomic absorption spectrometry (AAS).
Density measurements were compiled from PQ diamond core
and downhole wire-line gamma-gamma density logging. Crosssection interpretations were wire-framed to produce a threedimensional model of geology and ore grades. Ore grades were
domained at 0.5 per cent Ni and 0.3 per cent Co. Detailed
variography has been undertaken on sample grades within the
0.5 per cent nickel wireframe, and formed the basis for multiple
indicator kriging estimates. A block model was generated
incorporating ore type, multi-element grades and density values.
Block size was 10 m in the local grid easting direction, 20 m in
the northing direction and 2 m in height. Grade estimation was
generated utilising multiple indicator kriging methods.
The total resource including measured, indicated and inferred
for Cawse and Cawse Extended is 213 million tonnes (Mt) at
0.7 per cent Ni, and 0.04 per cent Co using a 0.5 per cent Ni
cutoff (Table 3). This includes 30 Mt at 1.0 per cent Ni and 0.06
per cent Co in the proved and probable reserve category. Bunyip
pit contains proven reserves of 5.1 Mt at a grade of 1.1 per cent
Ni and 0.13 per cent Co at a waste to ore strip ratio of 2.5:1.
TABLE 3
Cawse and Cawse extended mineral resource estimates.
Project
CTR equity
Siberia
75 %
Sibera tank
Cawse central
Cawse extended
80 %
100 %
80 %
Subtotal
Heron strategic
alliance
Indicated
Inferred
Total
46.5
46.5
% Ni
0.7
0.7
% Co
0.03
0.03
mt
16
16
% Ni
0.7
0.7
% Co
0.04
0.04
mt
47.2
11.9
68.3
127.4
% Ni
0.7
0.7
0.6
0.6
% Co
0.04
0.08
0.03
0.04
85.6
mt
85.6
% Ni
0.7
0.7
% Co
0.03
0.03
mt
47.2
11.9
216.4
275.5
% Ni
0.7
0.7
0.7
0.7
% Co
0.04
0.08
mt
Total accessible
resources
Total metal
available
Measured
mt
0.03
0.04
344.3
344.3
% Ni
0.8
0.8
% Co
0.06
0.06
560.7
619.8
mt
47.2
11.9
% Ni
0.7
0.7
0.7
0.7
% Co
0.04
0.08
0.04
0.04
Tonnes
Ni metal
330 400
83 300
3 924 900
4 338 600
Tonnes
Co metal
18 880
9520
224 280
247 920
Grade control
Grade control drilling is undertaken on a staggered 10 m 5 m
drill pattern utilising face sampling reverse circulation hammer
drilling to 20 m depth. This pattern is varied according to ore
continuity and strike. Drill samples are composited for assaying
over 2 m intervals, with each 1 m sample logged for geology
contacts, lithology and ore type. Samples are analysed for nickel,
cobalt, manganese, magnesium, aluminium, iron, chromium,
copper and zinc, using a multi-acid digest and determination by
inductively-coupled plasma spectroscopy. Each sample is split at
the laboratory, with one split submitted for head grade analysis,
and the other submitted for upgrade testwork, involving a
simulated ore benefication process in the laboratory. Material is
processed through a 0.5 mm screen, which rejects coarse
low-grade material. Upgradibility of the ore is back-calculated
from material which does not pass through the screening process.
These figures are utilised to predict leach feed tonnes and ore
grade post-beneficiation. Procedures are currently being
developed to utilise leach feed grades (post-benefication) to
outline ore zones.
Ore zones are initially defined by lithology and classed as
Grind, Upgrade or Non-Upgrade feed. These ore types are then
divided on the basis of nickel and cobalt grades into mill feed
(>1.0 per cent Ni, >0.7 per cent Co), medium-grade stocks (0.8
to 1.0 per cent Ni, 0.3 to 0.7 per cent Co), low-grade stocks (0.6
to 0.8 per cent Ni, 0.1 to 0.3 per cent Co) and mineralised waste
(0.4 to 0.6 per cent Ni, 0.05 to 0.1 per cent Co), resulting in 12
stockpile destinations.
Ore spotting is effective in ~30 per cent of ore contacts, as
related to clear visual distinctions between grind and upgrade
ore, and high-grade to low-grade contacts. Spotters are in
constant communication with the excavator operator during

removal of ore. Dilution of ore is estimated for each ore block,
based on area; perimeter ratios and blasting parameters. Blast
vector indicators are placed in each area requiring blasting, on a
20 m 20 m grid to record pre- and post-blast position of ore.
Ore block mark up of these blocks is then adjusted accordingly.
The direction of mining aims to initially remove hanging wall
mineralisation (Bunyip south) with working faces running
north-south to minimise dilution.
Mining methods
Conventional open pit contract mining methods for ore
extraction. The fleet currently includes one Hitachi EX2500 250
tonne excavator and 4 Caterpillar 785B 150 tonne capacity
dump trucks. Bench heights are 2 m, with six metre benches for
blasting as required. The possibility exists to increase bench
heights below the Manganese Zone to 3 m. Current planned pits
are designed to a depth of 40 to 56 m. Approximately 70 per cent
of the area involves free-dig material with the remainder
requiring drill and blast.
Mining is currently focussed on the highest-grade portions of
the deposit, however the 20 year plan will incorporate the four
existing pits into one pit up to five kilometres in length and 850
m in width. Numerous ore zones are available and active at any
one time to allow in pit blending scenarios of ore type and
grades. Pre-stripping involves removal of 5 - 15 m of transported
overburden that is usually free dig. Open pit design parameters
include a 65 face angle bench from surface to 20 metres depth,
with 5 m berms, reducing to a 55 face angle bench from 40 m
depth. The overall wall angles are 53 with 20 m wide haul roads
with a 1:9 grade incline.
103
Ore is sourced from four open pits over a 6 km strike length

and trucked to a central ROM pad. Ore is stockpiled on the basis
of ore type and then grade range. Non-upgrade ore is currently
stockpiled separately because of elevated magnesium values,
which results in higher acid consumption. High-grade upgrade
ore is blended on individual stockpiles up to 60 000 t in size,
utilising chevron stacking and re-claiming.
As of February 2000, 4.1 Mt of ore at 1.0 per cent Ni and 0.2
per cent Co has been mined from three open pits. This includes
1.2 Mt of high-grade ore at 1.4 per cent Ni and 0.4 per cent Co.
ACKNOWLEDGEMENTS
The authors would like to acknowledge the assistance of Centaur
Nickel Operations in allowing the publication of this paper.
Many professional contributions have been made towards
understanding the Bunyip deposit, particularly Calvin Ferguson,
Toby Morris and David Hope. We acknowledge discussions held
with other personnel who have greatly added to the
understanding of the geology of the Cawse nickel deposit. The
efforts of Ken Hellsten and Neil Phillips in reviewing this paper
has greatly improved its quality and is much appreciated. The
authors also acknowledge the editorial assistance in finalising
some aspects of this paper.
REFERENCES
Brand, N W, Butt, C R M and Hellsten, K J, 1996. Structural and
lithological controls in the formation of the Cawse nickel laterite
deposits, Western Australia - Implications for Supergene ore
formation and exploration in deeply weathered terrains, in
Proceedings Nickel 96, pp 185-190 (The Australasian Institute of
Denn, S M, 1998. Large diameter drilling at Cawse Nickel Project - An
innovative solution to metallurgical sampling, AIG Bulletin,
22:61-66.
104
Elias, M, Donaldson, M J and Giorgetta, N, 1981. Geology mineralogy

and chemistry of laterite nickel-cobalt deposits near Kalgoorlie,
Western Australia, Economic Geology, 76:1775-1783.
Hellsten, K J, Lewis, C R and Denn, S, 1998. Cawse nickel-cobalt
deposit, in Geology of Australian and Papua New Guinean Mineral
Deposits, pp 335-338 (The Australasian Institute of Mining and
Hill, R E T, Barnes, S J, Gole, M J and Dowling, S E, 1990. Physical
Volcanology of Komatities, Excursion Guide Book No 1.
Hotz, P E, 1964. Nickeliferous laterites in southwestern Oregon and
northwest California, Economic Geology, 59(3).
Lawrance, L M, 1996. Geochemical responses within lateritic profiles
over barren and mineralised ultramafic rocks: implications for nickel
exploration in the Yilgarn block, Western Australia, in Proceedings
Nickel 96, (Eds: E J Grimsey and I Neuss) pp 167-171 (The
Australasian Institute of Mining and Metallurgy: Melbourne).
Lawrence, L M, 1999. Geochemistry of nickel in the regolith. In course
notes Nickel laterites Centre for Teaching and Research in Strategic
Mineral Deposits. The University of WA (Unpublished).
Loftus-Hills, G D, 1975. Ora Banda lateritic nickel deposits, WA, in
Economic Geology of Australia and Papua New Guinea, (Ed: C L
Knight) 1. Metals, pp 1010-1011 (The Australasian Institute of
Marston, R J, 1984. Nickel mineralisation in Western Australia.
Geological Survey of Western Australia, Mineral Resources Bulletin,
14, 272 p.
Witt,W and Harrison, N, 1989. Volcanic rocks and bounding shear zones
of the Ora Banda greenstone sequence, in The 1989 Kalgoorlie
Workshops, (Ed: M Glacken) pp 2-7 (The Australasian Institute of
Mining and Metallurgy, and Eastern Goldfields Discussion Group:
Kalgoorlie).
Witt, W K and Swager, C P, 1989. Structural setting and geochemistry of
Archean I-Type granites in the Baardoc-Coolgardie area of the
Norseman-Wiluna belt, Western Australia, Precambrian Research,
44:323-351.
Witt, W K, 1993a. Gold Deposits of the Mount Pleasant Ora Banda
areas, Western Australia-Part 2 of a systematic study of the gold
mines of Menzies-Kambalda region, Western Australia Geological
Survey, Record 1992114.
Cadia Hill Gold Mine One Year Down the Track

C F Moorhead1 and Cadia Geology Team2
ABSTRACT
The Cadia Hill Gold Mine, located near Orange in Central Western New
South Wales, is owned and operated by Cadia Holdings Pty Ltd, a wholly
owned subsidiary of Newcrest Mining Limited. Construction including
pre-strip mining occurred from 1996 to 1998. The mine was officially
opened in October 1998 following commissioning of the concentrator
and mine fleet in June of that year.
The Cadia Hill orebody is a porphyry related gold and copper system.
The final pit has a strip ratio of approximately 1.3:1.0. Mining is
undertaken on 15 metre benches using large capacity diesel hydraulic
face shovels at an annualised rate of up to 70 Mtpa. The concentrator is
currently rated at 17Mtpa and includes the worlds largest single line
grinding circuit. The feasibility Mineral Resource was estimated at
352 Mt at 0.63 g/t Au and 0.17 per cent Cu. Initial reserves totalled
200 Mt at 0.74 g/t Au and 0.17 per cent Cu yielding a mine life of
approximately 12 years.
During the pre-production period mine geology systems were
developed to support the operation. By design these systems were
intended to add value to the business through minimisation of material
miss-classification, provision of accurate and timely production
information and mapping and reconciliation as a continuous model
improvement tool.
1.
MAusIMM, Geology Manager, Cadia Hill Gold Mine, Newcrest

Mining Limited Cadia Hill Gold Mine, via Post Office South
Orange NSW 2800.
2.
Cadia Geology Team, Members of the Mine Geology Team past

and present, Newcrest Mining Limited Cadia Hill Gold Mine, via
Post Office South Orange NSW 2800.
Variances that would prove to be material to the life of mine plan were
recognised during the first year of production. These included
identification of additional ore tonnes, higher copper grades and higher
gold recoveries relative to the model. Gold grades and copper recoveries
also reconciled favourably. Testwork was undertaken to determine the
underlying causes of each variance. Re-modelling was then undertaken to
capitalise on those variances.
The Cadia Hill Gold Mine life of mine plan was re-evaluated with
updated geology, grade, geotechnical, metallurgical and environmental
inputs. Operating strategies were also reviewed and revised. Despite
application of reduced metal prices in the evaluation, the outcome was
one of extended mine life with more metal produced at lower cost.
Development of robust mine geology systems, including a pro-active
reconciliation process contributed to tangible improvement of the Cadia
Hill Gold Mine operation. On-going system development and
improvement initiatives are planned to add further value for the future.
INTRODUCTION
Location and ownership
The Cadia Hill Gold Mine is located in the Central Tablelands of
New South Wales, Australia, approximately 20 km south of
Orange at Latitude 3328S and longitude 1490E. Cadia Hill
Gold Mine is owned and operated by Cadia Holdings Pty Ltd
which is a wholly owned subsidiary of the Australian gold
mining company Newcrest Mining Limited (Newcrest)
(Figure 1).
FIG 1 - Location diagram.
105
C F MOORHEAD and CADIA GEOLOGY TEAM
Geology and operations

The Cadia deposits (Cadia Hill, Ridgeway, Cadia East, Cadia Far
East, Cadia Quarry, Big Cadia and Little Cadia) all occur within
a 6 km long northwest trending corridor within the Molong
Volcanic Belt of the Lachlan Fold Belt of eastern Australia. The
Late Ordovician porphyry-related mineralisation is spatially
related to a small composite stock (Cadia Intrusive Complex) of
predominantly monzonite composition. The stock intruded
Ordovician Forest Reefs Volcanics and Weemalla Formation
sediments.
At Cadia Hill, mineralisation consists of a southwesterly
dipping tabular zone of sheeted veins with chalcopyrite and
bornite predominantly hosted in the orthoclase porphyritic phase
of the Cadia Monzonite. The veins cross the lithological contact
into the roof pendant Forest Reef Volcanics to the east without a
significant change in character.
Post-Silurian reverse faulting has significantly dissected the
ore zone with major thrusts on the western pit limit associated
with the Cadiangullong Fault which represents the northern limit
of the regional Wongalong Fault; and the Gibb Fault juxtaposing
Cadia Hill mineralisation against the Cadia East deposit in the
east.
Detailed descriptions of the geology of Cadia Hill and the
adjacent deposits have been published previously by various
workers (Wood and Holliday, 1995; Newcrest Mining Staff,
1995; Holliday et al, 1998; and Newcrest Mining Staff, 1998)
(Figure 2).
The feasibility life of mine plan at Cadia Hill was based on
starting reserves of 200 Mt at 0.74 g/t Au and 0.17 per cent Cu.
With an average waste to ore stripping ratio in the order of
1.3:1.0, open pit mining was planned to be at rates up to 60 Mtpa
to feed the worlds largest single line grinding circuit at 17 Mtpa.
Flotation copper concentrate would be pumped to a filter plant at

nearby Blayney prior to transport to customer smelters in Japan
via rail to port. Annual production would be approximately
300 000 ounces of gold and 20 000 tonnes of copper over a mine
life of ~12 years (Figure 3).
Open pit mining occurs on a continuous shift basis. Fifteen
metre benches are blasted and mined in a single pass using large
diesel hydraulic face shovels loading 230 t dump trucks. The
majority of the ore is direct dumped into a large gyratory crusher
on a just in time basis. This crusher feeds a coarse ore stockpile
(COS). This stockpile is ~200 kt in size with a live capacity of
~50 kt (one day). The SAG mill draws from three feeders located
below the COS at a rate in excess of 2000 dry tonnes per hour.
In its first full year of production Cadia Hill produced 253 670
ounces of gold at a total cost of production (NAGIS) of $409/oz.
The total cash cost was $257 per ounce (Newcrest Mining
Limited Annual Report, 1999). In addition a significant
underground resource has been discovered and identified nearby
at Ridgeway. Definition of potential additions to the surface and
underground resource inventory is ongoing. Results to-date have
given strong encouragement that reserves will increase.
Feasibility resource model

The Cadia Hill Gold Mine feasibility study was based on a
resource estimate completed by the Cadia Drillout Geology
Team in 1996 (Cadia Drillout Staff, 1996). The development of
the geology model and resource estimation process was further
documented in 1998 for presentation at the Towards 2000
Seminar held at Cobar (Moorhead et al, 1998). The quoted
Mineral Resource at the commencement of the project was
352 Mt at 0.63 g/t Au and 0.17 per cent Cu (Newcrest Mining
Annual Report, 1997).
NW-SE
PROSPECTIVE
HIGHLY
CORRIDOR
FIG 2 - Lease area geology plan.
106
CADIA HILL GOLD MINE ONE YEAR DOWN THE TRACK
FIG 3 - Cadia Hill feasibility life of mine metal production chart.
This model consisted of a three-dimensional block model

(Datamine), based on digitised gold equivalent grade zones and
lithological/structural domains interpreted on well supported
50 m cross-sections and plans. Data consisted of 1 m half HQ3
diamond core samples composited to 15 m on a bench basis.
Grade was interpolated using the inverse distance squared
approach with search ellipses and top-cuts determined
statistically for each domain. Variography showed gold and
copper mineralisation at Cadia Hill to be non-problematic with
respect to grade interpolation. Experimental variograms were
well structured with low relative nugget values and an absence of
short scale structures. Parent block sizes were set at 25 by 12.5
by 15 m with some subcelling allowed at domain boundaries.
Model fields included Au, Cu, CuCN, S, Lithology, and AuEq.
This model was further regularised to a 25 by 25 by 15 m SMU
size for optimisation purposes.
In each regularised cell a calculated profit field was
generated. This spatially variable field was a function of revenue
and processing cost, each of which was able to be derived from
the primary model fields. Profit was subsequently used as the
basis for material classification in mine planning and operations.
A waste model was created for the non-ore volume within the
model prototype. Within this model fields were generated for
MPA (maximum potential acidity) and ANC (acid neutralising
capacity). The MPA to ANC ratio was then used to classify the
waste in terms of potential to generate acid mine drainage.
Geology
MINE GEOLOGY SYSTEMS

Ore control
Data acquisition
The primary means of grade control sampling at Cadia Hill is a
multiple spear blast-hole method. This approach was selected on
the basis of sample quality and efficiency following
heterogeneity studies and sampling testwork. The testwork
included a trial of several blast-hole sampling methods including
pie segment, spear, multiple spear and a simple scoop. The
remainder of the cuttings for each hole in the trial (~1.6 t), was
carefully picked up using a small loader and passed through a

rotary splitter and submitted under controlled conditions to the
laboratory. Results were used to benchmark each of the methods.
All testwork indicated that sampling at Cadia Hill does not
represent a significant problem. This result, although atypical for
gold mineralisation was not unexpected given the disseminated
style of mineralisation.
Labour required to perform blast hole sampling is provided on
a seven days per week basis by the contract assay laboratory.
This innovative approach has proven to be very efficient and
effective. Samples dispatched by midday are generally available
by mid-morning the following day. Samples are analysed for
gold (fire assay 50 gms), copper (multi-acid digest AAS),
cyanide soluble copper (bottle roll AAS) and sulphur (ICP for
less than one per cent S otherwise LECO). Results are up-loaded
into the database electronically on an as required basis.
Each batch of samples submitted to the laboratory is
accompanied by selected standard reference material. Each
sampling and subsampling step also includes a notional one in 20
repeat. This data is up-loaded into the database. Accuracy and
precision data for all elements is monitored daily and reported
formally on a weekly and monthly basis. The quality control
system also includes routine laboratory inspections, submission
to a third party reference lab and review of laboratory internal
QA/QC reports. Sample grind size determinations are also
carried out on a regular basis.
Mine geologists log all blast holes using rugged field computers
(Hammerhead). Data collected includes lithology, sulphide
species and abundance, relative abundance of quartz veining and
alteration type and intensity. This data is down-loaded directly
into a relational database. Geologists also map available pit
exposures which provides critical structural data including
geotechnical information. Blast holes in waste zones are also
monitored for sulphide content as an indication of acid forming
potential. Specific environmental testwork is also undertaken as
an on-going means of AMD (Acid Mine Drainage)
characterisation. Expert consultants are engaged on an as needs
basis for peer review (Figure 4).
107
Gold
CadiaMine
Hill
Resource
Feasibility
& Grade
Geology
Model
Scale
Legend
Silurian sediments
Monzonite
Monzodiorite
Volcanics
Diorite
> 0.35g/t < 0.50g/t AuEq
> 0.50g/t < 1.00g/t AuEq
> 1.00g/t AuEq
Fault
FIG 4 - Feasibility model cross-section 13 920 mE.
Ore control modelling
Production and reconciliation
Data is retrieved from the database for processing using Surpac

Mining Software. Geology and grade information is plotted for
validation and interpretation. Interpreted geology and grade
domains are digitised on screen. Block models are created with
interpolation by ordinary kriging of the four assay fields.
Domain-specific variography is reviewed and updated regularly.
Models are stamped with lithology and a profit value calculated
for each block for material classification purposes.
Mining operations are controlled with the assistance of a real

time GPS based truck dispatch system (Modular Mining Dispatch). This state-of-the-art system provides detailed material
movement and productivity information including source and
destination reports. A Microsoft ACCESS based system was
developed to manage large volumes of production and
reconciliation data and generate reports. This system interfaces
Surpac block model report files with truck movement
information from Dispatch. The system also interfaces with a
similar system developed to manage metallurgical balance and
other mill information including weightometer data (Figure 7).
Material classification
Material classification at Cadia Hill is based on the calculated
profit field. The profit algorithm ($PA) incorporates a
calculation of revenue and processing cost. Revenue is a function
of grade, metal price and recovery. The latter is based on
regression relationships derived experimentally during the
feasibility process for each of six ore-types. The ore-type is a
function of Cu:S ratio and proportion of total copper that is
cyanide soluble (an indication of relative abundance of bornite).
Processing cost is a function of lithology (affects grinding cost)
and concentrate grade (TCRC). Copper grade also effects cost as
copper is treated as a bi-product credit. Profit therefore varies
spatially, the key sensitivity however is gold grade. Inputs to the
profit algorithm are updated as required (Figure 5).
Block delineation
For each blast the block model is simplified to yield practical
mining outlines. These are put out as dig plans for mining crews
and other stakeholders. Blocks are numerically identified by
bench, blast, and ore category for input into the truck dispatch
system (Modular Mining). Dig blocks are set out in the field by
mine surveyors using high precision GPS equipment. Each block
is then delineated using large colour-coded core flute boards
(Figure 6).
108
Production reporting
Production Key Performance Indicators (KPIs) are monitored on
a daily basis using source and destination information
downloaded from the Dispatch system. On a weekly basis
however active pit and stockpile faces are picked up by survey.
Digital terrain models are updated and used to deplete pit and
stockpile block models and stamp them with a date mined field.
Dispatch information is then used to pro-rata depleted volumes to
reconciled destinations incorporating any ore losses and dilution.
Stockpile management
The Crushed Ore Stockpile (COS) and Run of Mine (ROM)
stockpile are treated as live maintaining a daily balance of
delivered and removed material using production and
weightometer information. All other stockpiles are surveyed at
the end of each month and compared to their estimated status.
Block models are constructed in Surpac from DTM surfaces and
the estimated delivered grades assigned. The stockpiles are
closely monitored to ensure their metal inventory is realistic and
recoverable. SAG Mill feed estimates of grade and theoretical
recovery are calculated daily through depletion of the coarse ore
stockpile (COS).
CADIA HILL GOLD MINE - MATERIAL CLASSIFICATION CHART

As At 28 J ANUARY 2000
MATERIAL e g 790023301
F rida y, 5 J une 1998
Defined by Bench eg. 790; Blast No. eg. 023; Material Type eg. 3 (Green);
Block Number eg. 01 (first block of green)
NO
Es timate d Value
> $0 pe r to nne
YES
WASTE
> Trace Vis ible

Pyrite ?
Mate rial
Clas s ificatio n
Dig Blo c k No.

De s tinatio n
ORE
We athe re d?
Oxide ?
Estimated Value
> $-0.97 per tonne
$0.00 to $2.91
> $2.91
Mine ralis e d
Was te
Marg inal
Ore
Ore
NO
YES
Was te
AMD
Blue
Pink
Ora nge
Gre e n
Ye llow
Re d
501,502 e tc
Wa s te Dump
e g Blue 5
401,402 e tc
AMD Dump
e g P ink 2
601,602 e tc
S/PILE
e g Ora nge 2
301,302 e tc
S tockpile
e g Gre e n 3
201,202 e tc
S tockpile e g
YELLOW 4
101,102 e tc
Crus he r or
ROM
FIG 5 - Material classification chart.
FIG 6 - Dig plan example.
109
Mineralisation
Pit mapping has confirmed the majority of the mineralisation
zones based on earlier diamond drilling. However detailed
mapping has also enabled the recognition of an additional
discrete zone of copper mineralisation in the northwestern
portion of the pit not identified in the drillout phase. The copper
sulphides present in this zone are hosted in fractures and breccia
matrices clearly not typical of the main ore zone mineralisation
style. This small zone of very high copper grades may in part
have contributed to the copper grade variation described below.
Production
To the end of the June 1999, 14.5 Mt of ore was processed
through the concentrator. Reconciliation of this ore against the
reserve predictions has shown a 14 per cent increase in tonnes,
with a nine per cent increase in copper grade and a one per cent
increase in grade for gold (Newcrest Mining Annual Report,
1999). Metal recoveries were also significantly underestimated in
the model. The combined upgrade of both in situ metal and
recoveries resulted in 23 per cent additional gold and 26 per cent
additional copper metal production compared to reserve
expectation for the four quarters ending June 1999 (Newcrest
Mining Limited Quarterly Report to Shareholders, 30 June
1999).
FIG 7 - PRS flowchart.
Variance characterisation
Reconciliation
Net production is calculated as a function of mill adjusted
production and net stockpile movements. This is reconciled
against survey based reserve model depletions as a measure of
block model performance as a predictive model for the business.
Results are formally reported on a monthly, cumulative and
rolling basis (three months). In addition detailed reconciliation of
each process step is undertaken to characterise any variances in
the net reconciliation. This includes comparison of the reserve
model to the ore control model verse (information effect), ore
control to dig plans (selectivity), dig plans to as mined
(oreloss/dilution) and as mined to mill adjusted (mill
reconciliation). Results are analysed numerically, as
grade/tonnage curves and spatially.
VARIANCES AND ANALYSIS
Examination of incremental ore showed that additional tonnage

above cut-off was being identified by ore control on the orebody
margins. Testwork targeted at reasons for this reconciliation
variance examined the effects of incremental increases in sample
data density. Using the available high-density blasthole data, a
trial was run by selective inclusion of increasing number of
samples to establish if a reliable relationship between data
density and ore tonnes could be defined. Results from this work
showed a small but consistent increase in tonnes defined above
production cut-offs with increasing data density. Interestingly
however this information effect did not appear to result in the
expected reduction in the average grade of ore mined.
Unidentified high-grade zones
During the first year of production at Cadia Hill mapping and

reconciliation revealed significant variances to results predicted
by the model. Most of these variances were positive and expected
due to intentional and unintentional conservatism built into the
original model. Importantly, during that period ore control based
production figures reconciled within two per cent of net smelter
returns validating the use of this data as the reconciliation
benchmark.
Geology
Structure
Mapping of mine exposures has refined the structural framework
and ore zone configuration developed for the feasibility study
drill hole-based interpretation. While no fundamental changes to
the geological interpretation were required, structures subparallel
to the principal drilling direction were accurately defined and
located. Similarly major thrust faults on the western side of the
pit which were previously poorly constrained have been shown to
be significantly more complex. Resolution of these issues proved
to be vital in improving the model for short and mid term mine
planning as they had material influence on the local distribution
of grade and problematic zones of intense alteration that had a
negative impact on the flotation process from time to time.
110
Information effect
Further investigation revealed that the effect of additional lower

grade tonnage at the orebody margins was more or less offset by
under-sampled higher gold grade pods seen in the ore control
data to a greater extent than the wider spaced diamond drilling
data. Similarly the high copper zone noted above was intersected
by few diamond holes and was under-sampled in the feasibility
study model. The combined effect of under-sampling due to wide
spaced diamond data density was further compounded by
orientation of that data with the copper breccia zone striking
subparallel to the regular drill pattern. Furthermore the original
model served to smooth the data through gold equivalent (AuEq)
domain interpretation and inverse distance interpolation.
Sample biases
Investigation of possible bias between diamond core and blast
hole sample assays was also undertaken. This included a nearest
neighbour study where bench core composites within each
domain were paired with blast-holes within 2 m separation.
Results showed a clear understatement of gold and copper in the
drillout data compared to the production blast-holes in the main
ore zone. A program of re-analysis of drill-out pulps and
re-certification of standard reference materials from that period
was commenced to resolve this issue. (Results however were not
yet available at the time of writing.)
Gold
CadiaMine
Hill
Updated Model
Resource
& GoldGeology
Grade
Scale
Legend
Silurian sediments
Monzonite
Monzodiorite
Volcanics
Diorite
> 0.3g/t < 0.5g/t Au
> 0.5g/t < 1.0g/t Au
> 1.0g/t < 2.0g/t Au
> 2.0g/t Au
Fault
EOM Feb. 2000 Pit
FIG 8 - Updated model cross-section 13 920 mE.
Other variances
The metallurgical recovery models established during the Cadia
Hill feasibility were shown during the first year of production to
clearly under-estimate achievable recoveries. This variance
significantly contributed to the overall metal reconciliation result.
The causes of this are not within the scope of this paper.
REMODELLING
Interpolation
Geology
Interpretation
All lithological, structural and grade domains were re-interpreted
from basic principles using all available diamond, ore control and
mapping data. With respect to the latter, Au and Cu zones were
interpreted separately in this instance highlighting a subtle spatial
separation of the maxima of the two economic metals within the
overall mineralised zones. All domains and subdomains were
digitised and wireframes created that were subsequently used to
flag the data. This did not result in a material change to the
fundamental interpretation however the location of critical
structures was improved.
Variography
Data was composted to 5 m downhole for modelling purposes.
Domainal variography was performed for all domains and
interpolated fields including Au, Cu, Ag, CuCN, and S. Typical
variograms for high-grade gold domains (mean 1.64 g/t) were
interpreted as four structure curves with 50 per cent nugget, total
variance 0.6, range 320 metres, with the plane of best continuity
dipping 50 degrees to 210. Typical high-grade copper domain
(mean 0.39 per cent Cu) variograms were four structure curves
with 20 per cent nugget, total variance 0.025, range 300 with
plane of best continuity similar to Au at 50 degrees to 210.
Lower grade domains have considerably lower variance values
(Figure 9).
Grade interpolation used ordinary kriging with input parameters

for each domain defined by variography described above. A
regular block size of 12.5 by 12.5 by 15 m bench was used with
subcelling allowed on oxidation and topography surfaces only.
The volume model comprised a combination of blocks from
individually filled wireframes for each grade and structural
domain with blocks flagged appropriately for the interpolation
process. Assay grades interpolated included Au, Cu, Ag, CuCN
(cyanide soluble copper) and S (total sulphur). Grades for each
element and domain were interpolated using only those samples
within the same domain plus a 15 m halo around the domain to
soften the hard boundary effect.
The interpolated grade model was subsequently updated with
resource category, density (based on lithology) and an
appropriate volume of waste with modeled AMD field.
Waste model
Sulphide distribution models

The abundance of sulphide species including pyrite, chalcopyrite
and bornite was logged electronically during the Cadia Hill
111
Gold
CadiaMine
Hill
Updated Model
Resource
& CopperGeology
Grade
Scale
Legend
Silurian sediments
Monzonite
Monzodiorite
Volcanics
Diorite
> 0.10% < 0.25% Cu
> 0.25% < 0.40% Cu
> 0.4% Cu
Fault
EOM Feb. 2000 Pit
FIG 9 - Example of fitted experimental variogram.
drillout on a trace, moderate and abundant basis. This

information was taken advantage of to create a series of sulphide
distribution models using indicator kriging (IK). This was
undertaken in an attempt to create improved predictive models of
spatial sulphide distribution for metallurgical and waste AMD
characterization modelling. The logged sulphide database
comprised 157 km of core logged at 1 m intervals which were
composited to 5 m intervals for interpolation. Variography run
on the data showed broad long-range anisotropy of each sulphide
species approximately parallel to Au and Cu mineralisation
trends. The pyrite distribution model derived from this process
was subsequently applied to development of an improved AMD
Model.
AMD model
The pyrite distribution model using the Trace indicator was
calibrated against actual distribution of AMD observed over the
first year of production. It was found that at a cut-off of 0.57 a
good numeric and spatial reconciliation was achieved. That is, all
blocks in the model with >57 per cent probability of containing
greater than trace (0.25 per cent) pyrite corresponded closely
with material classified in production as potentially acid forming.
In lieu of results from on-going AMD Characterisation testwork,
this model was adopted as the predictive model for AMD
distribution in the life of mine plan.
At the time of writing this model continued to perform well as a

predicitve model for AMD distribution at Cadia. Characterisation
work may result in a lifting of the cut-off necessitating a further
change to this model.
FIG 10
112
FIG 11
113
Other
As part of the re-evaluation process the geotechnical and
metallurgical models were also reviewed and updated. The detail
of this re-modelling is not within the scope of this paper.
application of a lower gold price as the basis for revenue

calculation. Ongoing mine geology initiatives aimed at
continuous improvement of the Cadia Hill and adjacent models
will result in further value added to the Cadia Hill Gold Mine
business.
Metallurgical recovery
ACKNOWLEDGEMENTS
Real production data was extracted from the geology and

metallurgy databases. This data allowed metallurgical recovery
models to be simplified and modified to yield a significantly
improved set of predictive models for each ore-type. The detail
of this re-modelling work is not within the scope of this paper.
Geotechnical
Detailed mapping of open pit exposures allowed the geotechnical
pit-slope model to be updated and refined. This process was
audited by peer review and resulted in modifications but no
material change to overall design pit-slopes.
RESULTS
The updated Cadia Hill model was input into a pit optimisation
and mine design process. This was evaluated under prevailing
(reduced relative to feasibility) metal prices. The resulting life of
mine plan was shown to yield more metal at a lower cost of
production than the previous plan. Reconciliation of the new
model over recent months shows recovered metal to be within
five per cent of expectation. This also represents a significant
improvement in the model as a planning tool.
CONCLUSION
Implementation of an effective reconciliation process led to early
identification of variances that were material to the business.
Testwork was conducted to verify underlying causes of, and
quantify these variances. The geological model was able to be
updated using comprehensive mapping information and the
resource re-estimated for input into the pit optimisation process.
The resulting mine plan yields more metal at lower cost despite
114
The authors would like to thank Newcrest Mining Limited for

permission to publish this paper. Thanks also to current and past
members of the Cadia Geology Team, Garnet Halliday General
Manager Cadia Hill Gold Mine and Foy Leckie Chief Geologist
Mining and Development Newcrest Mining Limited for their
on-going strong support of Mining Geology within the Group.
REFERENCES
Cadia Drillout Staff, 1996. Resource Model Report - Cadia Hill Copper
Gold Mineralisation - April 1996, unpublished Internal Report,
Newcrest Mining Ltd.
Holliday, J, Wood, D, McMillan, C and Tedder, I, 1998. Discovery of the
Cadia Au-Cu Deposits, Lachlan Fold Belt, Australia. Pathways 98
Extended Abstracts Volume, BC and Y Chamber of Mines/SEG.
Moorhead, C F, Dunham, P B, Eastwood, G J and Leckie, J F, 1998.
Cadia Hill From Discovery to a Mine A Case Study, in
Resource/Reserve Estimation Practice in the Central West New
South Wales Mining Industry Cobar NSW, pp 67-77 (The
Newcrest Mining Staff, 1995, The Cadia Wallrock-Porphyry Style
Gold-Copper Deposit, NSW. Porphyry Related Gold and Copper
Deposits of the Asia Pacific Region, pp 16.1-16.10.
Newcrest Mining Staff 1998 Cadia Gold-Copper Deposits: Geological
Update, AIG Bulletin, No 23 p 83.
Newcrest Mining Staff, 1998. Cadia gold-copper deposit, in Geology of
Australian and Papua New Guinean Mineral Deposits (Eds: D A
Berkman and D H MacKenzie), pp 641-646 (The Australasian
Institute of Mining and Metallurgy: Melbourne).
Wood, D G and Holliday, J R, 1995. Discovery of the Cadia Gold/Copper
Deposit in New South Wales By Refocusing on the Results of
Previous Work. New Generation Gold Mines: Case Histories of
Discovery, pp 11.1-11.10 (Australian Mineral Foundation:
Adelaide).
The Geita and Kukuluma Mineralised Trends, Lake Victoria

Goldfield, Tanzania Orebody Characteristics and Project
Planning
D Bansah1, R Chase2, A Davidson2, H Michael3, M Skead2 and H Stuart2
ABSTRACT
INTRODUCTION
Ashanti Goldfields Companys (Ashantis) most recent development is

the Geita Project in the Lake Victoria Goldfield of NW Tanzania. It
comprises six distinct gold deposits within two mineralised trends:
Nyankanga, Lone Cone, Geita Hill (Geita trend) and Matandani,
Kukuluma and Area 3 West (Kukuluma trend).
Between 1936 and 1966, the Geita area produced an estimated 5.5 M
tonnes of ore from five underground deposits at an average recovered
grade of 5.3 g/t. The discovery of Nyankanga was made by Ashanti in
1996 from RAB drilling of a +100 ppb Au anomaly in laterite soil above
ferricrete cover. The Kukuluma trend was discovered by SAMAX
Resources (taken over by Ashanti in 1998) between 1996 and 1997 from
RC drilling of a +80 ppb Au laterite soil anomaly.
Oxide facies banded iron formation (BIF) of the Upper Nyanzian
System forms prominent topographic ridges and is the dominant host
lithology for gold mineralisation. The ridges are flanked by felsic tuff
units which lie stratigraphically above. Syn-genetic lamprophyre
(vogesite) and post-genetic felsic dykes are also seen in the Geita trend.
At Nyankanga a granodiorite-diorite suite has intruded as dykes and
sills into BIF from a plug centred on a WNW fault and has produced
wallrock alteration assemblages similar to those seen in Cu-porphyry
systems (potassic, propylitic, etc) as well as sulphide and gold
deposition. However, the main ore-grade mineralisation (as in the rest of
the Geita trend) is controlled by low-to-medium angle structures dipping
north-west and north with numerous splays in the hanging wall.
High-grade shoots at Nyankanga plunge to the west and to the
north-west, being lithologically and fault-controlled respectively. At
Geita Hill the higher grade shoots are controlled by fold noses in BIF
dipping to the north. Free gold occurs within fractures of sulphides.
The deposits of Kukuluma and Matandani occur on an ancient
(possibly Cretaceous) erosional plateau and are related to two WNW to
NW trending shear zones, discordant to the local strike of the BIF
sequence. Complete oxidation occurs up to 120 m depth and there is a
strong supergene component to the gold distribution. Primary gold
mineralisation is divisible into two distinct assemblages:
arsenopyritepyrrhotitemagnetitechlorite associated with brecciated
cherts and pyrite associated with shaley banded ironstones. Area 3W is
most likely the fault-displaced extension of the same mineralised trend
and comprises thin, intermittent zones of high-grade gold mineralisation
occurring adjacent to a ridge in banded cherts and minor BIF.
Current resources (measured, indicated and inferred) for the project
are estimated at 89.24 Mt at an average grade of 4.17 g/t (11.97 Moz of
contained gold) at a 1.0 g/t cut-off grade. Of this 9.83 Moz is within the
Geita trend and 2.14 Moz is within the Kukuluma trend. A resource of
2.74 Moz is from an inferred resource at depth at Nyankanga where the
viability of an underground operation is being studied. Proven and
probable reserves are 49.77 Mt @ 3.46 g/t (5.54 Moz) based on a
$325/oz gold price providing a mine life of ten years at an average
production rate of 500 000 oz per year. Reserve estimation was carried
out using mineable polygons to select resource blocks based on a series
of mining and costs parameters. In addition key cost parameters were
tested to provide sensitivities for pit optimisations and pathfinders for
negotiations with suppliers and contractors. Construction began in March
1999 and first gold is expected to be poured in the second-quarter 2000.
Ashanti Goldfields Company is an African gold mining and

exploration company which produced over 1.5 million ounces of
gold in 1998 and which has exploration interests in over 12
African countries. Ashantis most recent development is the
Geita Project located within the Lake Victoria granite-greenstone
terrane of NW Tanzania (Figure 1). It comprises six distinct gold
deposits within two mineralised trends: Nyankanga, Lone Cone,
Geita Hill (Geita trend), Matandani, Kukuluma and Area 3 West
(Kukuluma trend). The total combined resource for the six
deposits is currently 11.97 million ounces. Ashanti began
construction of a 4 Mtpa process plant in January 1999 and
production is scheduled to commence in second-quarter 2000.
1.
Ashanti Exploration Ltd, Box 2665, Accra, Ghana.
2.
Ashanti Goldfields Tanzania Ltd, Box 532, Geita Tanzania.
3.
MAusIMM, Ashanti Goldfields Tanzania Ltd, Box 532, Geita

Tanzania.
PRODUCTION AND EXPLORATION HISTORY

Gold mineralisation in the Geita area was first discovered in the
early-1930s as part of a survey carried out by a small Kenyan
syndicate. A processing plant was constructed in 1934 and
between 1936 and 1966, the Geita Mine constituted the largest
gold operation in East Africa, producing an estimated 5.5 Mt at
an average recovered grade of 5.3 g/t from five deposits.
By 1940 smaller underground mines were also in operation at
Lone Cone and Ridge 8 with open pit material being mined by
glory hole from Prospect 30. Ore from Ridge 8 and Prospect 30
was transported via aerial ropeway to the Geita mill. A small
quantity of high-grade ore was trucked from the Mawe Meru
mine 20 km to the south in the Rwamagaza greenstone belt.
Mining methods and cut-offs varied with deposit. Mill recovery
by cyanidation was approximately 85 per cent.
The Geita mine closed in 1966 due to a combination of
inadequate financing and the fixed gold price although according
to historical records, good grades were being encountered and
there was no shortage of ore at the time of closure. Total
production is shown by deposit in Table 1.
TABLE 1
Geita district historic production.
Deposit
Tonnes
milled
Grade
recovered g/t
Ounces
produced
Geita Mine
3 827 587
5.07
623 912
Lone Cone
163 415
8.22
43 187
Prospect 30
392 970
5.72
72 768
Ridge 8
833 092
7.64
204 633
Mawe Meru
(Rwamagaza)
82 928
21.65
57 723
5 292 992
5.88
1 002 223
Total
Modern exploration in the Geita belt was initiated in 1994 by

Cluff Resources Plc in the Geita East and Geita West licences
(190 km2) (Figure 2). Initial work focussed on the previous
production centres of Lone Cone and Prospect 30. At Lone Cone
115
D BANSAH, R CHASE, A DAVIDSON, H MICHAEL, M SKEAD and H STUART
Lake Victoria
MWANZA
GEITA
Buck Reef
Mawe Meru
SUKUMALAND GREENSTONE BELT
BULYANHULU
GOLDEN RIDGE
Shinyanga
BUZWAGI
Kahama
N
GOLDEN PRIDE
ADVANCED GOLD PROJECT
50km
ARCHEAN GREENSTONE BELT

FIG 1 - The Lake Victoria goldfields.
Granite terrane
BIF
BUGULULA
Quartz Gabbro Dyke

Deposit
Late Intrusive
Mining Licence
KUKULUMA
Exploration Licence
Major Structure
Fault
Fold Axis
GEITA EAST
GEITA WEST
Geita Trend
Bukolwa Hill
Kukuluma Trend
GEITA
HILL
Ridge 8
N
3 km
FIG 2 - Ashanti Goldfields Tanzania Ltd Geita licences with regional geology.
116
THE GEITA AND KUKULUMA MINERALISED TRENDS
trenching in late-1994 defined two mineralised zones which were

drilled on an 80 m 40 m grid in early-1995. Following the
award of the Geita Hill licence in late-1995 and the acquisition
of Cluff by Ashanti in early-1996, work commenced on 9500 m
of underground channel sampling which led to a first phase of
drilling on an 80 m 40 m grid. Resource drilling on a 40 m
20 m grid to a maximum depth of 175 m below surface was
completed by December 1997.
Mineralisation at Nyankanga was discovered in May 1996
from rotary air blast drilling (RAB) of a 200 ppb soil anomaly.
Initial reverse circulation (RC) drilling commenced in mid-1996
and was accelerated in late-1997 following the discovery of the
1120 shoot and high-grade BIF hosted mineralisation at depth
in the eastern part of the deposit.
Work by SAMAX Gold Inc. on the Kukuluma (100 km2) and
Bugalula (120 km2) Prospecting Licences began in 1995 (Figure
2). Reconnaissance soil sampling (Figure 3) outlined four main
anomalies on which trenching was carried out. Anomalies from
the BIF ridge to the south-east (Area 3) were more encouraging
and were drilled between late-1995 and mid-1997. In June 1997,
eight holes were targeted on Area 1 based on an erratic soil
anomaly (+100 ppb) around two artisanal pits. Five of these
holes returned significant results in what is now the Kukuluma
deposit. In early-1998 RAB drilling of extensions discovered the
Matandani deposit 700 m to the north. In late-1998 the
Kukuluma and Matandani deposits were included in the Geita
project following the acquisition of SAMAX by Ashanti.
DISTRICT GEOLOGY
The Geita Greenstone belt forms the east-west trending northern
arm of the regional Sukumaland Greenstone belt and is
approximately 60 km long by 15 km wide (Figure 1). The belt is
dominated by isoclinally folded oxide facies BIF ridges

averaging 500 m in thickness and flanked by younger felsic
pyroclastics. The BIF units have been later deformed into
west-plunging open folds with WNW axial trends. These have
subsequently undergone major displacement along NW trending
shears. Regional Proterozoic quartz-gabbro dykes intrude along
reactivated NE-SW faults (Figure 2).
THE GEITA TREND

The Nyankanga, Lone Cone and Geita Hill deposits form a
semi-continuous NE mineralised trend near the hinge of the
principal west-plunging synform. All deposits dip N-to-NW,
subparallel to stratigraphy. They are located on structures
subsidiary to the regional NW trending shears, close to which
local compression causes breaks in continuity. (Figure 3).
Nyankanga
The Nyankanga deposit forms the SW limit of current known
resources along the Geita Trend and subcrops in low ground
below 5 - 15 m of barren, transported laterite cover (Figure 4).
NNW-to-NW trending regional shears, Iyoda and Veda, bound
the deposit along strike, towards which a gradual decrease in
gold mineralisation is observed.
A microdiorite intrusive plug, whose emplacement is thought
to have been focussed along the Iyoda Shear (Figure 5) is
situated in the south and west of the deposit. Sills penetrate
north-eastwards from this plug into a 30 NW-dipping
sedimentary package of Banded Iron Formation with lesser
mudstone and siltstone (hereafter grouped as BIF), giving rise to
a repetitive stratigraphic sequence which forms the main host to
mineralisation. BIF increases in abundance away from the
intrusive centre along strike to the NE and down dip.
FIG 3 - The Geita trend orebody distribution
117
FIG 4 - Nyankanga subcrop geology with orebody and with final pit outlines.
Mineralisation is principally related to a sigmoidal NE

trending structure formed between the Iyoda and Veda shears
during dextral displacement. Stratigraphy has also been drag
folded into a similar orientation (Figure 3).
The main orebody is up to 50 m thick in the central part of the
deposit and dips subparallel to stratigraphy at 20 - 30 to the NW
and N. Numerous steeper mineralised structures up to 10 m thick
occur as imbricate splays in the hangingwall. Discontinuous
low-grade zones occur in the footwall and outside the main
deposit and are thought to be intrusion related.
Areas of high-grade seen on the grade thickness plan
projection (Figure 5) generally represent uniformly mineralised
BIF units grading in the order of 20 g/t over 10 m thickness and
ranging up to 45 g/t over 18 m. Within the microdiorite,
mineralisation comprises a wide anomalous zone up to 60 m
thick typically containing multiple intersections of 1 - 2 g/t over
5 - 8 m. Rheological contrast with the microdiorite has focussed
shearing within the BIF and mineralisation has followed these
units wherever possible, resulting in higher grades. Where BIF is
not present (as in the main microdiorite plug in the SW of the
deposit) or where the ore zone passes through a significant
microdiorite sill a lack of stress focus has dispersed ore fluids
away from the main shear along stockwork fabrics and
pre-existing structures such as higher angle joint sets. In these
areas the orebody is wider but has a more erratic gold
distribution and a lower average grade. Thus, grade distribution
within the orebody is primarily controlled by lithology (Figure
6).
118
There is also a structurally controlled high-grade zone (the

1120 shoot) in the SW of the deposit (Figure 4). This is situated
along the NW plunging intersection lineation of the main shear
with a subvertical NW-trending structure. This has provided
focus for ore fluids in the microdiorite and has resulted in
extremely high-grade-thickness values for this host.
Two phases of syn-to-post-mineralisation dykes occur
throughout the deposit and are thought to represent late felsic
evolution of the intrusive system at Nyankanga. Initial
emplacement of en-echelon felsic porphyry dykes 50 - 100 m
apart occurred dominantly along pre-existing joint sets dipping
40 - 50 NW and N with some dilation parallel to the ore zone
where they cross-cut. Late-stage deposition and/or
re-mobilisation of gold may have occurred during emplacement
since very high grades (50 - 200 g/t) are often observed in the
orebody adjacent to contacts and low-to-moderate grades can be
seen within the margins of the dykes themselves. A single,
barren, final stage quartz porphyry dyke, which has emplaced
along joint-related structures dipping 40 - 50 to NW and N
cross-cuts and displaces all lithologies and mineralisation. With
respect to the Iyoda and Veda shears all dykes follow the same
sigmoidal, drag-folded pattern as the bedding and orebody.
Gold mineralisation is intimately associated with fine-grained
pyrite and silicification. Free gold occurs as inclusions and
within fractures in pyrite (Schandl, 1999) and as discrete grains
within secondary quartz. In BIF, ductile shearing and
mineralisation have occurred preferentially within magnetite
2000mN
1750mE
1500mE
1250mE
1000mE
750mE
2000mN
>250gm
150 - 250gm
1750mN
1750mN
100 - 150gm
1500mN
1500mN
50 - 100gm
25 - 50gm
1750mE
1500mE
1000mN
1250mE
1000mN
1000mE
1250mN
750mE
1250mN
FIG 5 - Nyankanga grade thickness distribution, plan projection.
bands and along magnetite-chert bedding contacts. Areas of

higher strain have resulted in brecciation and the resultant higher
fluid flow has caused silica flooding and pervasive pyrite
mineralisation, eventually leading to total replacement of the
protolith in areas of very high-grade (approximately >50 g/t). In
microdiorite, pyrite occurs as fracture fill in biotite/calcite veinlet
stockworks and in quartz veins. In areas of higher grade
(approximately >3 g/t) pyrite is also finely disseminated
throughout the groundmass and silica replacement is pervasive.
The Nyankanga orebody exhibits a complex distribution of
overprinting alteration assemblages. Multiple intrusion-related
alteration systems are seen in the microdiorite and less frequently
in BIF and although work to-date on this aspect is limited it is
evident that similarities exist with porphyry deposits. Potassic
(K-feldspar, biotite, magnetite) and sodic (albite) alteration is
most common in the west of the deposit within the main intrusive
and close to the footwall of the orebody. Outside these high
temperature zones haloes of epidote and sericite are further
surrounded by widespread propylitic (calcite, chlorite) alteration.

Subeconomic gold and anomalous copper (chalcopyrite), lead
and zinc are thought to be related to intrusive mineralisation.
When coincident with the main shear, this early alteration and
mineralisation has frequently been brecciated and overprinted by
silicification and ore-grade gold deposition related to the main
mineralising event. In these cases early subrounded discrete gold
grains can be seen rimmed by later phase platy over-growths of
gold (Schandl, 1999).
Geita Hill
The North East Extension mine is connected along strike to the
Geita Mine by historic underground workings. Open stoping up
to 20 m width at a cut-off of approximately 4.5 g/t took place on
nine levels each 45 - 50 m apart. Ore was preferentially extracted
from higher grade hangingwall and footwall zones which
converge in the central part of the Geita Mine.
119
1400mRL
2200mE
2000mE
1800mE
1600mE
1400mE
1200mE
1000mE
1400mRL
LATERITE
NW
SE
1200mRL
1200mRL
BIF
1000mRL
OREBODY
1000mRL
MICRODIORITE
(UNSHADED)
800mRL
800mRL
FELSIC/QUARTZ
PORPHYRY DYKES
100 metres
2200mE
2000mE
600mRL
1800mE
1600mE
1400mE
1200mE
1000mE
600mRL
FIG 6 - Schematic cross-section, Nyankanga line 1080 mE looking southwest.
The lithological succession is dominated by BIF with interbeds

of mudstone, siltstone and felsic pyroclastics. Feldspar porphyry
(FP) sills seen in the Geita Mine (Schandl, 1999) are thought to
be genetically related to microdiorite in the SW of Geita Hill and
other parts of the Geita Trend. The volume of sill intrusions
decreases NE-wards. Pre-to-syn genetic biotite-rich lamprophyre
(vogesite) intruded along pre-existing E-W structures and other
planes of weakness such as bedding, lithological contacts and the
principal mineralising shear.
Mineralisation is controlled by a major north-east trending
shear dipping between 50 and 55 NW with imbricate splays in
the hangingwall. Shearing is subparallel to the regional strike of
the BIF sequence but cross-cuts local folding and lithological
contacts. The outer contacts of the orebody are sharp with a rapid
drop in grade over 1 to 2 m to background levels of less than
0.02 g/t gold. Ore grade mineralisation is continuous over a
strike length of 2.3 km, within which four lithologically
controlled high-grade shoots plunge to the north, with grades up
to 8 g/t over 40 m. These are the main Geita and North East
Extension mines (BIF dominated) and two newly discovered
shoots; The Gap (interbedded BIF and felsic tuff) and Geita West
(interbedded BIF and microdiorite) (Figure 7). Widespread
pre-mineralisation faulting is responsible for abrupt changes in
120
host lithology between these shoots bringing less-favourable

lithological packages into the path of the mineralising structure.
Extensive microdiorite occurs between Geita West and Geita
Mine and a highly disjointed interbedded BIF and felsic tuff
sequence is found between Geita Mine and The Gap. The
orebody is thinner (5 - 15 m) and lower grade (1 - 2 g/t) in these
areas.
Gold mineralisation is intimately associated with fine-grained
sulphide mineralisation and silicification. Pyrite is dominant with
localised pyrrhotite-rich zones. Sulphide content is typically ten
to 20 per cent of total rock volume in high-grade areas with
occasional zones of massive replacement. The style of sulphide
mineralisation within BIF is similar to that at Nyankanga
whereas in other lithologies sulphides are disseminated. Most
gold occurrences are free as fine-grained inclusions and
fracture-fill in pyrite (Schandl, 1999). Several generations of
quartz veins are present. Veins may attain widths up to 2 m.
Their margins are generally heavily deformed and may contain
high gold grades, most notably in North East Extension, where
they have been targeted by stoping. Base metal occurrences are
minor.
Pervasive sericitisation common in the felsic porphyry is
thought to be early in the genetic sequence. Thin (up to 1 cm),
irregular veinlets of calcite with subsidiary ankerite and dolomite
NE
SW
Level 4
The
Gap
Planned
Open Pit Outline
Geita West
North East Extension
Stope
High Grade Shoot
Geita Mine
Mine Level
500m
FIG 7 - Geita Hill deposit long section looking southeast showing location of high-grade shoots.
post-date and cross-cut both mineralisation and silicification/

quartz veining. Late secondary biotite rims calcite and quartz
veins and also cements some breccias (Schandl, 1999).
Lone Cone
Two orebodies are defined at Lone Cone: the North Zone, which
was mined between 1940 and 1953 on three levels accessed by a
vertical shaft, and the South Zone, recently discovered by
artisanal miners. To-date only initial drilling has been completed.
The North Zone is hosted by a distinct BIF unit within a
regular sequence of interbedded BIF and microdiorite.
Mineralisation in the South Zone is hosted by BIF with minor
interbedded felsic tuffs. A late felsic dyke intrudes along the
mineralised structure in the South Zone.
Both orebodies strike ENE and dip 50 to the north. The North
Zone has been delineated over a strike length of 480 m. Ore
occurs as two high-grade lodes (6 - 7 g/t over 5 - 8 m) at the
hanging wall and foot wall margins with lower grade material
between contributing to a total thickness up to 25 m. As at Geita
Mine, ore was preferentially mined from hanging wall and
footwall zones. The South Zone outcrops over a strike of 240 m
and differs in character from the North Zone in being made up of
various thin and sometimes patchy and inconsistent mineralised
sections forming a total thickness of up to 40 m. Gold
mineralisation and wallrock alteration are similar to that at Geita
Hill.
THE KUKULUMA TREND

The Kukuluma Trend comprises five NW-trending areas of
significant gold mineralisation within a 3 km ESE trending gold
in soil anomaly (+80 ppb) that cuts obliquely across a
NW-trending horseshoe ridge of BIF called Bukolwa Hill

(Figure 8). The five areas comprise three on the eastern limb of
the horseshoe in Area 3, and two close to the apex on the western
limb now referred to as the Kukuluma and Matandani deposits.
Mining in the area is limited to artisanal pits located within the
Kukuluma deposit and sporadic workings at several other
locations along the ridge.
Bukolwa Hill comprises prominent ridges of BIF and chert
interbedded with mudstone, siltstone and wacke. The BIF
sequence is underlain by a distinctive unit of carbonaceous
pyritic mudstone and overlain by felsic tuffs that flank the ridges.
Tight anticlinal folding, with the BIF and pyritic mudstone in the
core, has been refolded along a NW axis to produce the
horseshoe ridge. NW shears trend parallel or subparallel to
bedding. A major E-W shear cuts across the fold and the
Kukuluma Trend. The hill is also cross-cut by N-S and N-E
trending structures.
The Kukuluma and Matandani deposits are located in
topographic bowls incised into the ancient (Cretaceous) hill-top
ferricrete plateau at the head of seasonal streams. Mineralisation
is fault controlled, stratabound and localised by the intersection
of two shears (the Kasata and Juma shears) trending NW and
dipping steeply to the NE with iron-rich lithologies of a similar
strike but shallower dip (Figure 9).
Kukuluma
The Kukuluma deposit comprises two distinct zones of
mineralisation, the Fold Nose and the Fold Limb. Artisanal
miners have worked sporadically at two locations. Both zones are
focussed where the general 300 strike of the western BIF ridge
is deflected to 280 by the influence of the E-W trending shear.
High strain deformation is evident in rocks from both zones
121
FIG 8 - The Kukuluma Trend - geochemistry and deposit locations.
(Figure 10). As a result of the deep weathering the upper 20 m of

the profile is leached hence the poor soil and trench results
obtained from this area. An altered, unmineralised intrusion of
diorite underlies the Kukuluma deposit.
Mineralisation on the Fold Limb occurs along a 680 m section
of the Juma shear, in a sequence comprising highly brecciated
chert beds (5 to 10 cm thick) and grunerite-chlorite-magnetite
mudstone. The orebody dies out to the SE where it approaches
the E-W shear and is truncated to the NW by a N-S fault. The
Fold Nose lies SW of the Fold Limb and comprises a core of
mineralisation 150 m by 50 m trending 290 and confined to the
well bedded, oxide facies BIF in the nose of the fold (axial plane
strike 100, plunging 70 to 90 SE) where it is cut by the Kasata
shear.
Gold in fresh rock is very fine grained (<20 ) and occurs
solely in the iron-rich mudstone, included in magnetite and
arsenopyrite within fibrous grunerite aggregates and spatially
associated with pyrhotite and pyrite. The margins to the
mineralisation are extremely sharp, with the intercalated
sediments being rarely mineralised in fresh rock.
Matandani
The Matandani deposit comprises two parallel zones, trending
315 referred to as East and West. The mineralisation is covered
by a blanket of soil and laterite and was identified from an erratic
122
gold-in-soil anomaly consistent with the northward projection of

the Kukuluma deposit. The host lithologies tend to be more
clastic than those at Kukuluma and include shaley BIF and
wackes.
Mineralisation is localised along NW trending shears, with the
stronger mineralisation of the eastern zone being associated with
the dislocated, northern extension of the Juma shear. Although
the bulk of the gold is contained in discrete BIF units, occasional
high-grade, quartz vein mineralisation is encountered in the
volcano-sedimentary lithologies.
Both the Matandani and the Kukuluma deposits are weathered
to over 100 m in some areas and this appears to have had a strong
influence on the current gold distribution. Initial trenching
through the soil overburden gave few indications of the ore
bodies below, yielding a maximum of 0.85 g/t over 2 m.
However, free leaf gold is found along fractures in BIF taken
from the Kasata artisanal workings (10 m maximum depth).
Exploration drilling has revealed a general increase in mean
grade of 2 to 3 g/t between 60 and 105 m depth at Kukuluma,
and a planar zone of gold mineralisation between the base of
complete oxidation and the top of fresh rock, identified by
panning the drill cuttings and quantified by analytical results. In
modelling these deposits it is therefore recognised that narrow
stringer in fresh rock coalesce to form wider zones of
mineralisation in the zone of weathering.
FIG 9 - Kukuluma/Matandani geology plan.
FIG 10 - Kukuluma deposit schematic section looking northwest.
123
Area 3
Area 3 comprises three mineralised zones located along the
inside edge of the 1.5 km long eastern limb of Bukolwa Hill. The
BIF sequence comprises more chert than at Kukuluma although
the same carbonaceous pyritic mudstone is seen at the core of the
fold and iron-rich sediments are found interbedded with the BIF
at the southern end of the ridge. The limb is linear, with a vertical
dip and NW strike. At the southern end of the limb, an ESE
plunging fold nose is intruded by a granodiorite plug. Garnet and
coarse grained actinolite are ubiquitous in the iron-rich
sediments. The laterite cap and underlying leached profile
observed at Kukuluma has been eroded from this area which
accounts for the stronger, more coherent geochemical anomaly
delineated along this part of the trend and the superior results
achieved by the trenching that led to the first phases of drilling.
Mineralisation at two locations along the limb occurs at the
intersection of crosscutting NE faults with the principal NW
shear. Gold is associated with arsenopyrite in magnetite rich
bands or occurs with slicification and quartz veins at the sheared
contact between the BIF and the felsic tuff. Grades are therefore
either high over relatively small widths (eg 18.85 g/t over 6 m,
13.9 g/t over 8 m) or wider and lower grade (eg 2.2 g/t over 26
m, 2.5 g/t over 11 m). At the southern end of the limb the fold
nose is cut by the same northwest shear, but mineralisaton is of
low-grade and appears to be relatively unfocused (eg 30 m at
1.65 g/t, 15 m at 1.02 g/t). A low-grade (0.5 to 1 g/t) halo is
associated with the granitoid intrusive.
RESOURCE ESTIMATION
Resource estimation was carried out for Nyankanga, Geita Hill,
Lone Cone, Kukuluma and Matandani. All drill data was
recorded on field sheets and entered into Field Marshall.
Micromine was then used for data validation and generation of
sections on which geological and orebody interpretation are
carried out before data is transferred to Datamine for modelling.
A 3D wireframe geological model was created and used as a
guide for interpreting the orebody for which outlines were
generated using a cut-off minimum of 3 m grading 0.5 g/t. Each
zone of mineralisation was considered independently. Additional
surfaces were modelled to allow classification into ferricrete,
oxide, transition and sulphide zones and the assignment of
density factors.
The block model was created by filling the respective
wireframes with rectangular blocks to produce a volume model.
Strike:across strike:elevation dimensions of 10 m:5 m:5 m were
used with subblocking along boundaries to a minimum size of
2.25 m:0.5 m:0.5 m.
The distribution of sample lengths in the two sampling
methods employed to evaluate the deposits show that 95.3 per
cent of the samples have a length of 1 m. Consequently, irregular
core samples lengths were composited to one metre to yield
common sample lengths. Univariate statistics were calculated on
all samples within the wireframe boundaries selected, and then
on subsets representing individual mineralised zones. Top cuts
were applied to each zone based on distribution in cumulative
frequency plots. These ranged from 20 to 76 g/t.
Variographic analysis was undertaken primarily to determine
the suitability of the drill spacing used, to determine the optimum
dimensions of the search ellipse and to study the grade
distribution in each deposit. With the exception of down-hole
semi-variograms, all directional variograms were noisy with
minimal apparent structure and did not give sufficient confidence
to derive appropriate cross-validation models to enable the use of
kriging as an interpolation method. As a result inverse distance
squared (ID2) was used to estimate block grades. Directional
weightings of 1:1:0.17 were applied to limit the averaging of
grades in the across-strike direction. To control the orientation of
124
the search during grade interpolation, each block was first

assigned an azimuth and dip value derived from digitised plan
and cross-section strings that trace the trend of each mineralised
zone. The grade interpolation was conducted as a series of
multiple loops. For a given run, all blocks with local dip and
azimuth values that locate within a 10 azimuth/dip window
were interpolated. Each zone was estimated separately such that
only samples relating to a given zone/lode was used to
interpolate grades into it.
Rock densities were calculated based on the weight and
displacement of whole NQ and HQ core samples and are
summarised in Table 2. Owing to the large difference between
BIF and microdiorite density throughout the area and the strong
lithological control on the distribution of high grades, especially
at Nyankanga, ore blocks were assigned densities based on the
geological model.
TABLE 2
Rock densities.
Rock type
Oxide
Transition
Sulphide
BIF
2.60
2.90
3.05
Microdiorite
1.76
2.27
2.70
At Nyankanga, the barren felsic dykes were modelled

separately and the grade within those outlines excluded from the
statistics and grade interpolation. Where they intersect the
orebody, the volume of the dykes was deducted from the orebody
model.
At Geita Hill and Lone Cone the underground workings were
wireframed based on the mine closure plans, actual intersections
with resource holes and on recent survey data. This volume was
also deducted from the orebody model.
Resource Figures are shown in Table 3. The resource has been
classified based on the September 99 JORC code guidelines for
reporting mineral resources and ore reserves. The resource
classification is based on the density of sampling and varies from
deposit to deposit. At Geita Hill the deposit has been drilled at a
spacing of 40 m (strike) by 20 m (dip) and as such a large portion
of the resource is classified as measured. As less information is
available in the area surrounding the underground workings these
blocks have been included as indicated. At Kukuluma,
Matandani and Lone Cone, drilling is on a 40 m by 40 m spacing
and as such the bulk of the resource is classified as indicated. At
Nyankanga the near-surface portion of the orebody has been
drilled at 40 m by 20 m and is classified as measured. Below this
to the 1000 m RL the drilling is based on a 40 m by 40 m grid
and is classified as indicated.
Drilling in 1999 was aimed at delineating the full potential of
the Nyankanga open pit and the potential for an underground
operation. The drilling was widely spaced and extended the
orebody to the 750 m RL (500 m below surface). This part of the
resource has been classified as inferred. Results from this drilling
have shown significant high-grade extensions at depth which will
be mineable from an open pit. It has also defined an underground
resource of 12.03 Mt grading 8.09 g/t which is included in the
data below based on a cut-off between open pit and underground
resources at the 930 m level.
RESOURCE OPTIMISATION AND PIT DESIGN

The resource block models for the six deposits were interrogated
visually in plan at 5 m intervals. The continuity of ore blocks
both along and across strike was used to estimate the potential
mine recovery of each ore bench compared to the in situ resource
estimated.
TABLE 3
Geita project resources.
Million tonnes
Gold g/t
Million ounces
Million tonnes
Gold g/t
Million ounces
Nyankanga
Oxide
1.89
3.36
0.20
Measured
14.52
5.05
2.36
Transition
2.19
3.29
0.23
Indicated
10.63
4.98
1.70
Sulphide
24.07
5.46
4.23
Inferred
3.00
6.24
0.60
Subtotal
28.15
5.15
4.66
Subtotal
28.15
5.15
4.66
10.14
8.42
2.74
Indicated
0.20
6.49
0.04
Inferred
9.94
8.46
2.70
Subtotal
10.14
8.42
2.84
Total
38.29
6.01
7.40
Underground
Sulphide
Total
38.29
6.01
7.40
Geita Hill
Oxide
4.70
2.42
0.37
Measured
16.26
2.50
1.31
Transition
6.63
2.51
0.54
Indicated
4.63
2.98
0.44
Sulphide
11.14
2.81
1.01
Inferred
1.58
3.08
0.16
Total
22.47
2.64
1.91
Total
22.47
2.64
1.91
Kukuluma
Oxide
5.47
3.32
0.58
Measured
5.08
3.22
0.53
Transition
0.42
3.84
0.05
Indicated
2.46
3.47
0.27
Sulphide
2.73
3.02
0.27
Inferred
1.08
2.90
0.10
Total
8.62
3.25
0.90
Total
8.62
3.25
0.90
Matandani
Oxide
5.20
2.70
0.45
Measured
4.98
2.84
0.45
Transition
0.53
2.52
0.04
Indicated
3.85
3.34
0.41
Sulphide
5.29
2.97
0.51
Inferred
2.19
2.13
0.15
Total
11.02
2.84
1.01
Total
11.02
2.84
1.01
Lone Cone
Oxide
1.89
2.52
0.15
Measured
0.00
0.00
0.00
Transition
1.97
2.86
0.18
Indicated
5.05
2.66
0.42
Sulphide
2.28
2.63
0.19
Inferred
1.09
2.73
0.10
Total
6.14
2.67
0.52
Total
6.14
2.67
0.52
Area 3
Oxide
0.70
2.30
0.05
Measured
0.00
0.00
0.00
Transition
0.00
0.00
0.00
Indicated
0.00
0.00
0.00
Sulphide
2.00
2.80
0.18
Inferred
2.70
2.70
0.23
Total
2.70
2.70
0.23
Total
2.70
2.70
0.23
Total
Oxide
19.58
2.82
1.80
Measured
40.84
3.54
4.65
Transition
8.74
2.81
0.79
Indicated
26.82
3.80
3.28
Sulphide
60.65
4.81
9.37
Inferred
21.58
5.82
4.04
Total
89.24
4.17
11.97
Total
89.24
4.17
11.97
Selection criteria were established based on the size and

selectivity of the mining equipment (100 t trucks with 6.5 m3
excavators), the minimum mining unit (3 m 3 m at Nyankanga,
3 m 2 m at Lone Cone and Geita and 2 m 2 m at Kukuluma
and Matandani), the dip of the orebody and the envisaged density
of grade control drilling that would be carried out to define the
ore. These are summarised as follows:
included waste blocks at 33 per cent along strike and 50 per
cent across strike;
extractable waste at three continuous blocks along strike and

three continuous blocks across strike;
minimum polygon segments of 20 m along strike and 5 m

across strike; and
maximum angle change between polygon segments 45.

On the basis of these criteria, mineable polygons containing
the recoverable resource were selected. Some portions of the
resource were excluded from the optimisation (eg hanging wall
splays at Nyankanga were excluded). The areas defined within
125
the mineable polygons were then re-estimated for grade to the

boundary limits of the polygon outline rather than including the
resource blocks that were >1.0 g/t within the outline. The
following dilution skins of specified widths were then added to
the polygons: Lone Cone 0.25 m, Nyankanga and Kukuluma 0.5
m, Geita Hill 1.0 m and Matandani 1.2 m resulting in a diluted
reserve model that was used for pit optimisation and definition
and estimation of proved and probable reserves.
A geotechnical evaluation was carried on each deposit using
probabilistic limit equilibrium analysis and kinematic stability
assessments. In the zone of weathering the lithological units at
Geita Hill, Lone Cone and Nyankanga are of poor to fair rock
mass quality, while at Kukuluma and Matandani these have very
poor to poor rock mass quality. On a multi-bench scale,
rotational failure is expected to be the main mode of instability in
the weathered zones.
In fresh rock, the rock mass is of fair to good quality and
characterised by high intact rock strength with numerous well
developed subvertical and subhorizontal joint sets striking
normal and parallel to the trend of mineralisation. Bedding is
well developed in BIF but is generally tight or healed. Footwall
bedding dips are generally parallel to or steeper than the dip of
the orebody footwall at Nyankanga, Lone Cone and Geita Hill,
though in some areas of Geita Hill the orebody dip is steeper
than the bedding dip. Instability in the unweathered zone is
therefore expected to be structurally controlled. Bench scale
planar failure could occur on footwall slopes where
mineralisation dips more steeply than bedding. Single and
multi-bench instability on the hangingwall slopes could be due to
the failure of tetrahedral wedges. In addition, at Kukuluma and
Matandani, geometric conditions exist for potential toppling
instability in the unweathered rock mass.
Pit optimisation and design were carried out using Datamine
LG3 based on a gold price of $US325/oz. Table 4 shows a
summary of reserves as of January 2000.
Initial mine scheduling has been carried out and encompasses
the following aims:
the development of the Nyankanga Stage 1 pit to provide
sufficient waste material suitable for construction purposes
(tailings dam, ROM pad, road construction).;
to expose high-grade ore in preparation for full processing by
April 2000.;
to defer development of other pits to focus early supervision
and training efforts and concentrate site security.;
accelerate material movement from 2001 to 2004 to gain
early access to the deeper, higher grade fresh mineralisation
of Nyankanga.; and
provide continuous feed from the satellite pits of Matandani
and Kukuluma (16 km to NE) to justify the introduction of
the higher unit operating cost road trains into the operation.
Mining begins and ends in Nyankanga over the life of the
project. The Nyankanga staged pits commence during the
pre-production period with the second pit, Lone Cone coming on
line in 2001 primarily due to early low strip ratio followed by

Kukuluma and Geita Hill in 2002. This is followed by Matandani
the following year to maintain a steady ore production rate from
the satellite pits. The final cutback on Nyankanga which
represents the access into the deepest, high-grade section
commences in 2004.
Ore feed blends to the process plant vary throughout the
project life with higher than 80 per cent sulphide feed occurring
in 2001 and 2009 - 12, primarily from the deeper Nyankanga ore.
No high oxide feed years occur except for 2002 (76 per cent)
with the introduction of the oxides from Kukuluma into the
circuit.
DISCUSSION AND CONCLUSIONS

Mineralisation in the Geita greenstone belt is primarily
controlled by regional, NW trending shearing developed in a
brittle-ductile environment. At Kukuluma the NW shears form
the principal mineralised structure, whereas in the Geita trend the
deposits are located along NE and ENE subsidiary structures. In
both cases mineralised shears trend subparallel to stratigraphy.
Late-stage displacement on NW shears is probably minor since
orebodies do not appear to be significantly displaced.
BIF is the dominant host for mineralisation. Within the
structurally-controlled regime competency contrast between
lithologies is the most important factor affecting grade
distribution. This is seen to maximum effect at Nyankanga where
the ratio of BIF to microdiorite is optimal and has given rise to
widespread high-grade mineralisation.
Intermediate intrusives in the Geita trend are thought to be
derived from a common source, possibly centred on a
diorite-granodiorite body occurring immediately to the south of
Nyankanga. Intrusives of similar composition are also observed
at Kukuluma and Ridge 8 (Figure 2). Although porphyry style
alteration and mineralisation is seen at Nyankanga, the authors
feel that this is distinct from, and has been overprinted by, the
main shear controlled mineralisation.
The use of mineable polygons in estimating the recovery of
resource blocks is suited to the Geita and Kukuluma trends
where parts of the resource occur in thin (<3 m) hanging wall
splays, which may not be amenable to the mining equipment
optimal for the bulk of the orebody. By using mineable polygons
a more practical assessment of minable reserves has been
possible.
In the last five years more than 14 Moz of contained gold has
been discovered in the Geita Greenstone belt, making it one of
the most prospective gold targets in the world. The reason for the
large number of gold occurrences is thought to be the thickening
of the belt as a result of folding, coincident with widespread NW
shearing. Work to-date in Geita represents first phase exploration
in an area where 60 per cent of the belt lies below transported
cover and the potential exists for further discoveries that will
continue to expand the Geita Project.
TABLE 4
Geita project reserves (diluted polygon method).
Orebody
Ore tonnes (M)
Grade g/t
Ounces (Moz)
Waste tonnes (M)
Strip ratio
Nyankanga
19.55
4.77
3.00
190.45
9.74
Lone Cone
3.54
2.57
0.29
13.35
3.8
Geita Hill
17.37
2.31
1.29
61.49
3.5
Kukuluma
6.28
3.28
0.66
24.05
3.8
Matandani
2.98
2.96
0.28
6.65
2.2
Total
49.77
3.46
5.54
295.98
5.9
126
REFERENCES
Ashanti, 1999. Geology of the Geita Project, Ashanti Goldfields
Tanzania, (unpublished internal report).
Barth, H, 1990. Explanatory notes on the 1:500 000 provisional
geological map of the Lake Victoria Goldfields, Tanzania.
Bundesanstalt fur Geowissenschaften und Rohstoffe und
Geologische Landesamter in der Bundesrepublik Deutschland
(BGR), Geologisches Jahrbuch Reihe B, Heft 72.
Borg, G, 1994. The Geita gold deposit in NW Tanzania - Geology, ore

petrology, geochemistry and timing of events, Geol Jb, D100, pp
545-595.
Schandl, E, 1999. Petrography of the Nyankanga and Geita Deposits,
Internal Consultants Report.
127

S Wilcock1
ABSTRACT
The Kunwarara magnesite deposit is the largest known magnesite
(MgCO3) deposit in Australia. It is located under black soil plains to the
north of Rockhampton in Queensland. The deposit consists of magnesite
nodules set in a fine-grained sandstone, siltstone and mudstone matrix.
The magnesite is derived from weathering of serpentinite in adjacent hills
and transport of magnesium in solution to a Tertiary/Quaternary river
system. The magnesium has come out of solution in the active river
system, and combined with atmospheric and biogenic CO2 to form
magnesite nodules. The magnesite can be very pure with certain varieties
up to 98 per cent MgO (LOI free). The deposit has been mined since
1991 and processed into magnesia (MgO) which is sold into refractory,
agricultural and other markets. The geologist is an integral part of the
small mine technical team. The role of the geologist includes
characterisation of the ore through drilling and bulk sampling,
supervision of geological aspects of mining and assistance in the
optimisation of ore processing.
INTRODUCTION
The Kunwarara magnesite deposit is located adjacent to the
Bruce Highway, 70 km north-west of Rockhampton in
Queensland. The topography of the area consists of broad
black-soil plains about 20 m above sea level flanked by hills up
to 400 m elevation. The location of the deposit is shown on
Figure 1.
Magnesite occurrences have been known in the Kunwarara

district since the early-1900s. All of the early discoveries were of
vein magnesite in the ultramafic rocks in the hilly areas.
Small-scale mining occurred at a number of locations between
Kunwarara and Marlborough, 30 km north-west of Kunwarara.
These mines operated intermittently from the 1920s through to
the 1980s.
In 1985, Queensland Metals Corporation (QMC) discovered a
large deposit of magnesite hosted in fluvial sedimentary rocks
under the Kunwarara black-soil plains. The deposit can be traced
for about 60 km from near Yaamba to Broad Sound. QMC holds
exploration and mining titles over the central section of the
deposit. A subsidiary company of QMC, Queensland Magnesia
(Operations) Pty Ltd (QMAG), commenced mining in 1991 on
the KG1 lease. To the end of 1999, 20.5 Mt of ore had been
mined.
Magnesite from the Kunwarara mine has been used by QMAG
for the production of deadburned and electrofused magnesia for
the making of refractory bricks. In addition calcined magnesia is
produced for use in agricultural, environmental and other
industrial applications. These products are sold to national and
international markets.
Future developments are planned based on the Kunwarara
magnesite orebody. The most advanced is that by the Australian
Magnesium Corporation Pty Ltd (QMC 50 per cent, Normandy
50 per cent). They are proposing to build Australias first
magnesium metal smelter. This project is currently being
assessed with a pilot plant operating in Gladstone. The plant uses
a new process, developed by QMC and CSIRO over the past
decade to produce magnesium metal at lower cost than
traditional techniques.
REGIONAL GEOLOGICAL SETTING

The magnesite deposits are found in Late Tertiary (Post Eocene)
to Quaternary sediments deposited in a fluvial environment and
are associated with Paleozoic ultramafic complexes. The
ultramafic complexes are the source of the magnesium required
for deposit formation. The ultramafics are part of the
Marlborough terrane, a component of the northern New England
Fold Belt. The Marlborough terrane is an ophiolitic assemblage
that separates the Carboniferous Yarrol terrane from the
Devonian-Carboniferous Wandilla terrane. The exact mode of
formation of these basement rocks is still a matter of debate
(Leitch et al, 1994). Serpentinites are a common component of
the ultramafic assemblage and these are the source of magnesium
for magnesite formation.
During the late Tertiary and Quaternary, a sinuous fluvial
system developed adjacent to the ultramafic complex and the
resulting sedimentary sequence has formed the host for the
magnesite deposits. Three semi-continuous deposits have been
recognised stretching from Yaamba (near the present day Fitzroy
River) through Kunwarara to Herbert Creek (adjacent to Broad
Sound), a distance of approximately 60 km.
DEPOSIT GEOLOGY
FIG 1 - Location of Kunwarara magnesite deposit.
1.
Senior Mine Geologist, Queensland Magnesia (Operations) Pty Ltd,

PO Box 5798, Central Queensland Mail Centre Qld 4702.
Sediment hosted magnesite deposits can be very large. The

Kunwarara deposit has an Inferred Resource of 1.2 billion tonnes
with greater than five per cent (mass) magnesite content
(Queensland Metals Corporation, 1999).
129
S WILCOCK
The magnesite has been derived from the weathering and

erosion of the adjacent ultramafic complex during the
Tertiary/Quaternary. The ultramafic complex is, in some places,
immediately adjacent to, or underlying, the magnesite deposit.
All of the magnesite deposits are within 10 km of the
ultramafics. The ultramafics are characterised by hilly landforms
rising 40 to 400 m above the surrounding plains. The ultramafic
rocks have provided the source of magnesium rich fluids, which
have been transported into the paleo-fluvial system where the
deposits are now found.
The deposits are located in topographically low areas, and
there is no outcrop.
The Tertiary/Quaternary fluvial sequence is up to 40 m thick.
It generally fines upwards from gravel and coarse unconsolidated
sand at the base through finer grained weakly cemented
sandstones to siltstone and mudstone. The magnesite orebodies
are found in the upper half of the sedimentary sequence, within
the weakly cemented sandstone and siltstone. The sediments are
overlain by 1 to 4 m of black clay, which has been deposited by
periodic sheet flood events, which have continued to the present
day.
No fossils have been found in the sediments, so the only age
indication comes from the fact that they overlie Eocene oil shales
near Yaamba and may be partly overlapped by Holocene alluvial
sand deposits at Kunwarara.
Earlier depositional models for the host sediments indicated a
lacustrine depositional environment (Schmid, 1987; Queensland
Metals Corporation, 1992). Exposures in pit walls during mining
at Kunwarara and data from extensive exploration drilling have
shown higher energy sedimentary features than would be
expected in a purely lacustrine environment. These include
gravel bars, erosionally based sandy channel deposits, and also
clay filled abandoned channels. In addition, the sinuous nature of
the deposits indicates that a more likely depositional environment
is a low to moderate energy fluvial system.
The magnesite orebody is found in the upper parts of the
sedimentary sequence and ranges in thickness from 1 to 26 m.
Ore thickness at the Kunwarara mine has averaged 12 m over the
8 years of mining operations. A diagram of the typical orebody
intersection is shown on Figure 2. Magnesite content in the
deposit varies from five to 90 per cent by mass (Queensland
Metals Corporation, 1992). Testwork during mining operations at
Kunwarara has shown an average magnesite content of around 35
per cent, by mass, in the areas which are suitable for mining.
The orebody is characterised by discontinuous zones of high
quality magnesite, separated by areas where the magnesite is less
well-developed.
FIG 2 - Typical deposit cross-section.
Magnesite types
The magnesite itself is cryptocrystalline and pure white in colour.
Scanning electron microscope (SEM) studies show crystal sizes
ranging from 1 - 10 microns. Inclusions include amorphous
silica, clays and iron and manganese oxides. The magnesite
occurs as distinctive concretionary nodules and as a stockwork of
veins and sheets. The nodules range in size from a few
millimetres up to 60 cm, with cemented aggregates of nodules up
to 1 m across. A typical nodule is shown on Figure 3.
Magnesite texture ranges from hard, pure, porcellanous bone
type through to softer, less dense, porous and chalky types.
Bone magnesite has a conchoidal fracture and is
characterised by well-developed rounded nodules. The porous
and chalky types are more common as veins and sheets. All
nodules have a skin of amorphous silica, which forms a rough
crusty surface. In bone magnesite, the interior of the nodules can
be > 98.5 per cent MgO and < 0.5 per cent SiO2 (all analyses are
loss on ignition (LOI) free basis). In the more porous types,
amorphous silica penetrates deeply into the nodules along cracks
130
FIG 3 - magnesite nodule in sandstone matrix.
KUNWARARA MAGNESITE DEPOSIT
and around pores, leading to lower MgO and higher SiO2

contents. In the upper parts of the orebody, additional
silicification has occurred and the nodules may show deep
desiccation cracks and can be as high as three to four per cent
SiO2.
Lime occurs in solid solution in magnesite. It ranges from one
per cent CaO to four per cent CaO in different parts of the
deposit. In a bulk sense the main source of lime in final products
is from the mineral dolomite (CaMg(CO3)2) which ranges in
abundance from nil to 100 per cent of the carbonate assemblage.
Dolomite often occurs as separate nodules although intergrowths
with magnesite are locally common. Dolomite nodules are
generally smaller and more irregularly shaped than magnesite
nodules. They also differ in that they are characterised by
inclusions of quartz grains, and have higher amounts of iron and
manganese oxides. As a result, dolomite has higher silica values
than magnesite (av ten per cent SiO2) and higher A12O3, Fe2O3
and MnO.
Dolomite is generally more abundant in the lower parts of the
ore zone and may form a pure dolomite layer (within host rocks)
at the base of the deposit.
Typical analyses of magnesite and dolomites are shown in
Table 1.
TABLE 1
Typical magnesite and dolomite analyses, Kunwarara deposit.
MgO %
SiO2 %
CaO %
Al2O3 %
97.5
0.4
1.93
0.04
0.09
95.7
1.46
2.5
0.12
0.12
0.09
94
2.45
3.03
0.22
0.24
0.09
49.2
15.3
33.6
1.35
0.47
0.13
1.
2.
3.
4.
Fe2O3 % Mn3O4 %
0.08
Bone magnesite
Bone and porous magnesite
Porous magnesite
Dolomite
Variations in the host rock are related to variations in quality of

magnesite. At Kunwarara, bone magnesite is more abundant in
red-brown fine-grained sandstone, whereas more porous types of
magnesite are more common in the grey siltstone facies. It would
appear that this distribution is related to higher permeability in
sandstone allowing greater circulation of magnesium rich fluids.
DEPOSIT FORMATION
The close spatial association of the magnesite deposits with a
large ultramafic complex gives an obvious pointer to the source
of the magnesium. No studies of magnesium dissolution from the
Marlborough terrane serpentinite have been undertaken, but
studies elsewhere give valuable pointers as to the likely
mechanisms operating for this class of deposit (Zachmann and
Johannes, 1989; Fallick, Ilich and Russell, 1991). Weathering of
the ultramafic complex during the late Tertiary-Quaternary has
mobilised magnesium in ground waters.
Cryptocrystalline magnesite formation has been linked with
selective dissolution of magnesium from serpentinite under the
influence of waters rich in atmospheric and biogenic CO2.
Hydrated magnesium carbonates precipitate in suitable
environments given an appropriate trigger mechanism, such as
mixing with high pH waters or by concentration through cyclic
evaporation. The hydrated magnesium carbonates are
transformed to magnesite by diagenetic processes (Zachmann
and Johannes, 1989).
Limited studies have been undertaken on the carbon and

oxygen isotopes of Kunwarara magnesite. Typical 13CPDB
values are -7.5 per mil. Typical 18OSMOW values are 31.5 per
mil. These values suggest the magnesite is derived from meteoric
water and a mixture of atmospheric and plant derived CO2
(Russell, M, 1999 Personal communication. July). Kunwarara
isotope values are comparable to those found in the Nevade,
Shilopaj and similar sediment hosted magnesite deposits in
Yugoslavia (Fallick, Ilich and Russell, 1991). In the Yugoslav
deposits similar isotope values have been interpreted to reflect
the formation of magnesite from dissolution of magnesium from
serpentinite by gravity-driven circulating carbonated meteoric
waters, which precipitate as magnesite on pressure drop as CO2
escapes. Atmospheric and biogenic CO2 is also implicated in the
formation of these deposits.
A modern day analog for the Kunwarara type of deposit, albeit
in a lacustrine environment, has been documented at Salda Lake
in Turkey (Schmid, 1987). The lake is flanked by serpentinite
hills, which shed magnesium rich waters and particulate matter
into the lake. Magnesite is currently being deposited around the
lake shore and occurs in rubbly dunes up to 10 m high.
Cryptocrystalline nodules and lumps of magnesium carbonate
and hydroxide are forming at the mud-water interface under the
influence of seasonally varying water levels. Field evidence
shows that magnesite crystallisation can occur very rapidly given
an adequate source of magnesium (Schmid, 1987). Chemically
precipitated nodules from Salda Lake are very similar to
Kunwarara nodules.
In the Kunwarara deposit, the magnesite precipitated in situ
very soon after the deposition of the host sediments. Evidence
for this is shown in mine pit exposures where channels have
eroded nodule aggregates and formed pebbly channel-floor
deposits (Wilcock, 1998). In situ precipitation continued above
such minor hiatuses. It would appear that nodules and veins
continued to develop for some time after sediment deposition.
The original sedimentary structures are still visible in pit
exposures, but they are heavily disrupted and distorted by the
growth of nodules and the penetration by veins.
The nodules of magnesite do not incorporate any of the host
sediments. Field observations show that they displace the host
sediments (Figure 3). Similar disruption of host sediments is
noted in the Yugoslav deposits.
Conversely, dolomite appears to be at least partly replacive.
Field observations show gradations from fine sandstone into
dolomite, and SEM studies show abundant quartz grains in
dolomite. This indicates that dolomite has replaced the clay
cement in sandstone.
A diagram of the accepted model of the formation of the
Kunwarara deposit is shown on Figure 4.
Magnesite nodule precipitation (in situ) is very common in the
vicinity of ultramafic complexes. In the Kunwarara region
magnesite nodules have been observed in numerous locations in
the weathering profiles above diverse rock types within a few
kilometres of serpentinite bodies. This indicates that magnesite
will readily precipitate out of solution given a suitable trigger
mechanism.
Within the serpentinites, veins of magnesite are locally very
common. This type of magnesite is related to weathering of
serpentinite and deposition of magnesite through interaction of
weathering products with descending meteoric waters loaded
with atmospheric and biogenic CO2 (Zachmann and Johannes,
1989; Fallick, Ilich and Russell, 1991).
The Thuddungra magnesite deposit near Young in New South
Wales has many features in common with the Kunwarara deposit
(Diemar, 1998).
131
S WILCOCK
FIG 4 - Model of formation of Kunwarara magnesite deposit.
MINE GEOLOGICAL METHODS

Mining and processing techniques
The mine is open cut, with a 100 tonne backhoe excavator and 50
tonne capacity trucks used for ore mining and haulage. Scrapers
are used to remove overburden clays and soil. The host sediments
are only weakly cemented but do form a competent rock mass,
which causes few geotechnical problems during mining. No
blasting is required. However the magnesite nodules themselves
are extremely hard. All metal surfaces in the mining equipment
and the process plant have to be wear protected.
Current production of ore is around 3 Mtpa, for the production
of about 400 000 t of magnesite. Primary beneficiation of the
magnesite takes place at the mine site. This involves crushing,
scrubbing, screening, heavy media separation and optical sorting
to separate the magnesite from the sedimentary host rocks and
dolomite. Conversion of the magnesite to various grades of
calcined, deadburned and electrofused magnesia occurs at a large
processing plant on the outskirts of Rockhampton.
Role of the geologist

The geologist at the Qmag mine is required to provide geological
interpretations to allow efficient day-to-day operation as well as
resource estimates decades into the future. Interpretation of
geological data must be integrated into the mine planning and
mineral processing areas of the operation.
The geology of the deposit is complex, with rapid vertical and
lateral variations in magnesite quality and abundance. These
variations are usually related to different niches in the primary
sedimentological environment. Geological techniques used to
assess the orebody have been developed to quantify this
variability. This involves rotary air blast (RAB) drilling on a
reducing grid to differentiate mineable areas. In early stages of
exploration the drill holes are collared on 200 or 400 m centres.
At this level of detail, Inferred Resources can be estimated. The
grid is gradually closed in areas that have mining potential to a
minimum spacing of 25 m. At this level, Measured Resources
and Proved Reserves can be estimated.
132
Large numbers of drill holes are required to assess the ore

zones. Five thousand RAB holes have been drilled in the current
operating mining lease, which covers an area of 550 ha, to fully
assess its potential. The drilling technique used is rapid and
relatively inexpensive.
The RAB drilling technique produces rock chips up to 25 mm
across. Magnesite content is determined using image analysis of
photographs of raw drill samples. This technique was developed
by Qmag using advances in digital camera technology and image
analysis software. It is used to efficiently quantify the amount of
magnesite relative to waste materials in the drill samples.
Magnesite quality is determined by chemical analysis of
magnesite separated from drill samples. Specific chemical
analyses are for MgO, SiO2, CaO, Fe2O3, Al2O3 and MnO. The
magnesite content and chemical analysis data from drill holes
give a qualitative guide to the likely performance characteristics
of different parts of the orebody. Judgement is required on the
part of the geologist to determine which areas should be included
in reserves and subsequently mined.
Qmag magnesia products have tight specifications for all the
oxides listed above, but especially MgO, SiO2 and CaO. One of
the main challenges is that not all magnesite in the mined ore is
of suitable quality to be considered final product. The overall
magnesite content of the orebody is about 35 per cent (by
volume), but the yield of final magnesite product from the mine
is about 15 per cent. The magnesite that is mined, but not
accepted as Qmag product, usually fails because of high levels of
SiO2 and/or CaO.
The processing plant at the mine upgrades the run of mine ore
using crushing, scrubbing, screening, heavy media separation
and optical ore sorting to extract magnesite of the required grade
from the ore. The behaviour of the ore through this complex
plant cannot be predicted from RAB drill chip samples, so
selective bulk sampling of the deposit has been undertaken.
These bulk samples have ranged from five to ten tonnes from one
metre diameter Calweld drill holes, to 40 000 t pits. Ore from
smaller bulk samples is used in simulations of the processing
plant, and ore from larger pit samples is run through the plant as
full-scale production trials. The bulk samples provide a much
lower density of sample points than RAB drilling, but if carefully
sited using information from the RAB drilling campaigns, they
KUNWARARA MAGNESITE DEPOSIT
provide the information required to better predict final product

quality distributions from the processing plant and determine
which areas should be included in reserves.
Future planned projects, based on the magnesite resource, such
as the magnesium metal smelter will use some of the lower grade
magnesite not currently suitable for Qmag requirements. This
will lead to higher yields of magnesite from the deposit.
The mining engineer in consultation with the geologist plans
the mine pits. Pits normally contain about 0.5 Mt of ore. Grade
control in the pit is visually based. The white magnesite contrasts
strongly with the grey and red-brown host sediments. This allows
the digger operator to readily identify important features such as
waste zones within the ore, and the bottom of the ore zone. The
geologist provides guidance to the mining crews, based on
information from drilling.
Various geophysical techniques have been trialled, but none
have proven cost-effective relative to drilling. The technique with
the greatest potential is a high-resolution resistivity system. This
system could successfully differentiate significant magnesite
development from lower quality areas by focussing on the
resistivity contrast between magnesite and the fluvial sediments.
Other techniques which have been trialed, but were not
successful included ground-penetrating radar and down-hole
geophysical methods including gamma, neutron, resistivity and
sonic.
CONCLUSION
Sediment hosted cryptocrystalline magnesite deposits have only
gained world scale importance as a source of magnesite since the
discovery of the Kunwarara deposit in 1985.
The mode of formation of these deposits is broadly
understood, although the reasons for some localised, but
practically important features such as dolomite distribution
within the orebodies require further study.
The magnesite deposits will only form where an ultramafic
complex with common serpentinite is present to provide a source
of magnesium. Tertiary-Quaternary weathering of the ultramafics
produces magnesium rich ground waters that may precipitate
economically important deposits in permeable TertiaryQuaternary terrestrial sedimentary sequences.
Mine geological methods focus on locating high magnesite
content zones within the deposit. Once located these zones are
mined in a conventional open cut mine. The greatest challenge
for the geologist is predicting performance where not all
magnesite will be recovered to final product.
ACKNOWLEDGEMENTS
The author wishes to thank the management of Queensland
Magnesia (Operations) Pty Ltd and Queensland Metals
Corporation Limited for permission to publish this paper
REFERENCES
Anon, 1993. Major expansion at Thuddungra magnesite mine, Minfo,
(Department of Mineral Resources: New South Wales) 38:14-16.
Burban, B, 1990. Kunwarara magnesite deposit, in Geology of the
Mineral Deposits of Australia and Papua New Guinea (Ed: F E
Hughes), pp 1675-1677 (The Australasian Institute of Mining and
Diemar, V A, 1998. Thuddungra magnesite deposits, in Geology of
Australian and Papua New Guinean Mineral Deposits (Eds: D A
Berkman and D H Mackenzie), pp 655-660 (The Australasian
Fallick, A E, Ilich M and Russell, M J, 1991. A stable isotope study of
the magnesite deposits associated with the alpine-type ultramafic
rocks of Yugoslavia, Economic Geology, 86:847-861.
Leitch, E C, Fergusson C L, Henderson, R A and Morand, V J, 1994.
Ophiolitic and metamorphic rocks in the Percy Isles and Shoalwater
Bay region, New England Fold Belt, Central Queensland, Australian
Journal of Earth Sciences, 41:571-579.
Milburn, D and Wilcock, S, 1994. The Kunwarara magnesite deposit,
central Queensland, in Proceedings Field Conference 94, Capricorn
Region, pp 99-107 (Geological Society of Australia: Brisbane).
Milburn, D and Wilcock, S, 1998. Kunwarara magnesite deposit, in
Geology of Australian and Papua New Guinean Mineral Deposits
(Eds: D A Berkman and D H Mackenzie) pp 815-818 (The
Queensland Metals Corporation Limited. 1992, Annual Report.
Queensland Metals Corporation Limited. 1999, Annual Report.
Jones, M R, 1995. Magnesite in review, Queensland Government Mining
Journal, April 1995, pp 11-20 (Department of Minerals and Energy:
Queensland).
Schmid, I H, 1987. Turkeys Salda Lake: A genetic model for Australias
newly discovered magnesite deposits, Industrial Minerals,
239:19-31.
Wilcock, S, 1998. Sediment Hosted Magnesite Deposits, AGSO Journal
of Australian Geology and Geophysics, 17(4):247-251.
Zachmann, D W and Johannes, W, 1989. Cryptocrystalline magnesite,
Monograph Series on Mineral Deposits: Magnesite, 28:15-28.
(Gabruder Borntraeger: Germany).
133
Current Geological Understanding of Telfer Gold Mine

G R Howard1, T Hansen1, C Moore1, P J Moffitt1, R J Inglis2, R D Carlson3, I Kirchner1,
D Coupland4, S Leary1 and A Tomsett3
Location
The Telfer Gold Mine is located in one of the most remote areas
of Western Australia, some 485 km by road south-east of Port
Hedland or 680 km north-east of Newman, within the Great
Sandy Desert (Figure 1). The mine falls within the boundaries of
the East Pilbara Shire, an area covering 386 000 km2 that
contains less than one percent of Western Australias population.
The climate of the region is characterised by hot summers
(January average daily temperatures exceed 40C) and warm
winters (July average daily temperatures exceed 10C). Rainfall
is strongly seasonal, falling between December and March and is
usually associated with remnant cyclones and thunderstorm
activity. Annual evaporation is 4160 mm.
18o
118o
119o
120o
122o
121o
123o
BROOME
18o
lf
hw
ay
TELFER
Nor
t h
100
19o
er
50
Kilometres
he
t
th
s
We
H ig
19o
or
Telfer Gold Mine combines a world-class gold and copper resource and
longevity of operation as a major gold and copper producer.
Telfer Gold Mine has undergone a series of transformations since
mining commenced as a high-grade narrow-vein oxide open pit operation
in 1977. In 1986, the open pit operation converted from a narrow-vein,
high-grade operation, to a low-grade high-volume oxide mining
operation utilising dump leach processing technology. Mechanised
underground mining of the high-grade narrow-veins commenced in 1990.
Total recovered gold production between 1977 to 1999 has totalled 5.7
million ounces.
The Telfer Extension Project consists of two pre-feasibility studies, the
Surface Extension Project (SEP) and the I-Series Project (ISP). The
combined studies represent one of the more significant resource
evaluations undertaken in the gold industry in recent times.
Geological activities for the SEP pre-feasibility study have included
the drill definition of a large low-grade resource within an area of
25 km2. The surface drilling program has completed 15 000 metres of
HQ3 diamond drilling and over 30 000 metres of RC between last quarter
of 1998 and late-1999.
The ISP pre-feasibility study is currently evaluating the resource
potential of both a high-grade low-volume and a lower grade bulk mining
scenario, 800 m to 1200 m below the surface of Main Dome. 15 000
metres of underground diamond drilling have been completed between
last quarter of 1998 and August 1999 as part of this study.
Systems and cultures have been developed to maximise safety, monitor
the quality assurance/quality control (QA/QC) on all aspects of the
operation, and minimise impact on the environment. All data are stored
in database systems, specifically designed for Telfer operations.
3D geostatistical resource estimation using ordinary kriging has been
carried out for over 220 geological domains within the 25 km3 of
mineralised system. Non-linear geostatistical techniques are being
investigated as part of the pre-feasibility study.
In seeking to sustain the operation well into the 21st century a

dedicated team of geologists, engineers and metallurgists has
identified additional gold and copper mineralisation from the
open pit and underground. The project to evaluate this
mineralisation is termed the Telfer Extension (TEX)
pre-feasibility study. This paper summarises the geological
component of the TEX pre-feasibility study.
ABSTRACT
Gre at
20o
RA
er
Riv
ne
gi
lla
Fortescue R
iv
NG
119o
er
120o
121o
22o
Lake Dora
r
R a n ge
Tom Price
Rang
ell
er
118o
TELFER
nbush
iso
e
Po Rang
Oa ko ve r R iv er
LE
21o
ss
RS
st
Lake Waukarlycarly
Nu
ro
ME
ck
Th
HA
AD
Tr a
ge
he
HILLS RO
an
ic
22o
RIPON
Marble Bar
WESTERN
AUSTRALIA
MAusIMM, MMICA, Geologik Solutions, PO Box 73, Mosman

Park WA 6012.
er
la
4.
R iv
WOO
DIE
W
ROADOODIE
21o
20o
on
Shay Gap
Gre y
r b le
el
Newcrest Mining Ltd, Telfer Gold Mine, PO Box 6380, East Perth
WA 6004.
De
Ma
Ro ad
3.
ay
GAusIMM, Newcrest Mining Ltd, Telfer Gold Mine, PO Box 6380,

East Perth WA 6004.
hw
Ba
2.
ig
No rth ern
MAusIMM, Newcrest Mining Ltd, Telfer Gold Mine, PO Box 6380,

East Perth WA 6004.
lH
Great
1.
r th
ta
ab
No
The Telfer Gold Mine, a wholly owned and operated subsidiary

of Newcrest Mining Limited, is a world-class gold and copper
resource that has been mined continuously since 1977. It has
undergone a series of transformations from a high-grade
narrow-vein oxide open pit operation in 1977 to the integrated
open pit, dump leach and underground operation that it is today.
Total gold production from the narrow-vein underground and
oxide open pit operations between 1977 and 1999 is more than
5.7 million ounces.
The Telfer Gold Mine is undergoing a renaissance with respect
to understanding the geological controls on mineralisation as
well as the potential application of alternative metallurgical
process routes.
West C
s
oa
Is
Summary
K id s
PORT HEDLAND
ve
INTRODUCTION
122o
d
Ru
a ll
Ri
Lake
Blanche
123o
FIG 1 - Regional locality map.
Discovery
The Bureau of Mineral Resources (Australian Geological Survey
Organisation) first geologically mapped the Telfer district in
1959, however during the mapping no gold or copper
mineralisation was identified.
Prospectors and exploration companies targeted the Telfer
district in the late-1960s and early-1970s as a copper province.
In 1971, Day Dawn Minerals NL undertook a regional
sampling program in the district under the direction of R
Thompson. Anomalous copper and gold values were returned
from gossanous outcrops that were sampled in Main Dome,
(Tyrwhitt, 1995).
135
G R HOWARD et al
Summary of Telfer Gold Mines development history

An intensive exploration and resource drilling program was
undertaken by Newmont Pty Ltd during the period 1972 to 1975.
This program defined an open pit reserve of 3.8 Mt @ 9.6 g/t Au
containing in excess of 1 Moz (Turner, 1980). Following a
feasibility study, Newmont brought BHP into the project and
mining commenced during 1975 at Main Dome and reached full
production of 0.5 Mtpa in 1977. Resources and reserves were
maintained at 2 and 1 Moz respectively in the early part of

Telfers mine life (Chamberlain, 1990).
By the early-1980s, reserves were nearing exhaustion and an
accelerated exploration program commenced in Main Dome and
West Dome (Figure 2). This program increased the reserves to
2.5 Moz and resources to 5 Moz by the end of the 1980s. The
increase in ounces was at the expense of gold grade which
dropped from 10 g/t to 2 g/t (Chamberlain, 1990).
FIG 2 - Telfer Main and West Dome plan.
136
CURRENT GEOLOGICAL UNDERSTANDING OF TELFER GOLD MINE
Extensive testwork was conducted during the early-1980s on

heap and dump leach technology. These processes were weighed
against the alternative of expanding the milling operation. The
latter option was chosen and in 1985 Telfer Gold Mine made the
transition from a low-volume, high-grade operation to a
high-volume, low-grade operation by expanding the mill to 1.5
Mtpa. Resource definition drilling during the late 1980s defined
additional resources of submill-grade material (1.2 g/t Au). This
increase led to further extensive metallurgical testwork on the
feasibility of establishing a dump leach operation. Commercial
dump leach treatment commenced in 1988, with an initial
processing rate of 4 Mtpa. This was expanded to 17 Mtpa during
1996.
As the pits deepened, the ore gradually changed from oxide to
supergene-sulphide material. This change was accompanied by
an increase in the content of cyanide soluble copper minerals.
Consequently the current open pit resources are process
constrained.
In 1989 a sulphide flotation circuit was established to process
Middle Vale Reef (MVR) supergene-sulphide ore, initially from
open pit sources and in 1990, from underground. This treatment
circuit was modified to its current configuration in 1996 to treat
mainly sulphide ore from the underground M10 and M12 reefs.
Exploration programs during the 1990s successfully delineated
additional reefs on the eastern flank of Main Dome. These reefs,
which include the M10, M12 and M50 reefs are currently being
mined using narrow-vein underground mining methods. Ongoing
resource assessment is being undertaken on the M20, M30, M35
and M40 reefs. In addition, decline access is currently being
established to the I30 reef system. This high-grade mineralised
system lying 1000 m below the surface was discovered in 1992.
PREVIOUS GEOLOGICAL MODELS

The Paterson Province regional geology and lithology is
described in Wells (1954), Blockley (1979), Turner (1982) and
Hill (1989).
The tectonic setting of the Paterson Province is described in
Williams and Myers (1990) and by Etheridge et al (1987). An
interpretation of the structural geology of Telfer is compiled in
Hewson (1996).
Early genetic models for Telfer mineralisation invoked
syngenetic-exhalative processes (Tyrwhitt 1979, 1985; Turner,
1982). Royle (1985) proposed a variation of this with the MVR
being formed as an evaporative horizon in a sedimentary basin.
The evaporite horizon was subsequently replaced by
quartz-sulphide assemblages.
Exposures of sulphide mineralisation in deep exploration
drilling and open pit/underground headings resulted in the
development of the epigenetic replacement models presented in
Levet (1986) and Dimo (1990). Goellnicht (1987, 1992)
proposed that mineralisation was derived dominantly from
magmatic fluid sources. This is in contrast to the interpretation of
Hall and Berry (1989) and Rowins (1994) which suggests that
granites acted as heat sources to circulate hydrothermal fluids
through the sedimentary sequence. Detailed mapping of the M10
- M12 reef system has changed the focus of the genetic models
from the replacement of fine grained siltstone units to
mineralisation being structurally controlled (Inglis, 1995).
REGIONAL GEOLOGY
The Telfer Gold Mine occurs within weak to moderately
deformed, low-grade metasedimentary rocks that form part of the
Neoproterozoic Yeneena Group in the Paterson Province (Chin
and Hickman, 1977). The Yeneena Group unconformably
overlays the intensely deformed Mesoproterozoic Rudall
Complex (Williams and Myers, 1990). A veneer of Neopaleozoic
to Holocene sediments forms the cover rocks in the Telfer
district.
Granitoid intrusives make up approximately 20 per cent of the

Telfer district (Goellnicht, 1992), with the main plutons being the
Mt Crofton Granite Complex, the Wilkie Granite, the Deserts
Revenge Granite, the Minyari Granite Complex and the
OCallaghans Granite.
MINE GEOLOGY
The stratigraphy and lithology of the Telfer Formation is
described in Dimo (1990). Since 1990 there has been
considerable progress in the understanding of the lithology and
controls on mineralisation in the Malu Quartzite Member and the
Isdell Formation. These units form the local mine stratigraphy
below the Telfer Formation (Figure 3), and focus discussion in
this chapter.
Malu Quartzite Member stratigraphy

The Malu Quartzite Member (MQM) is conformably overlain by
the Telfer Formation. The MQM is subdivided locally into three
main stratigraphic units: the Upper Malu Member (UMM), the
Upper Limey Unit (ULU) and the Lower Malu Member (LMM)
(Figure 3).
Upper Malu Member

The UMM is conformably overlain by the Telfer Formation
(TFM). The UMM sediments are similar in lithology to the
LMM. The UMM hosts the M10, M12, M20, M28, M30, M35
and M38 Reefs. The Tiger Beds, a 40 cm package of sheared
and laminated sandstone and siltstone between the M10 and
M12, provide a useful stratigraphic marker for distinguishing
between the M10 and M12 Reefs (Inglis, 1995).
Upper Limey Unit

The ULU is a light grey, fine to coarse-grained, calcareous,
altered sandstone unit. Sand grains are dominantly angular to
rounded clastic quartz grains, accompanied by muscovite flakes
with interstitial micritic dolomite and calcite. Internal fabric
displays a penetrative schistosity, with shear fabrics up to
mylonitic levels. Both bedding and veins can be folded and
boudinaged. The average thickness of the ULU is 0.4 m. The
ULU hosts the M40 Reef.
Lower Malu Member

The LMM is conformably overlain by the ULU and overlies the
Isdell Formation. The LMM is a sequence of very fine to coarse
grained interbedded quartz sandstone and argillaceous siltstone.
Quartz sandstone units are commonly greater than ten metres
thick with siltstone units typically thinly banded or laminated
and less than a few metres thick. Individual sandstone beds are
commonly massive but can display sedimentary features such as
cross bedding, graded bedding and flame structures. The LMM
hosts the M45, M50, M55, M60, M65, M67 and M70 Reefs.
Isdell Formation stratigraphy

The Isdell Formation (IDF) is the lowest unit of the Upper
Yeneena Group that occurs within the mine sequence.
The IDF is comprised of a sequence of very fine to
coarse-grained interbedded quartz sandstone and siltstone beds
with varying amounts of dolomite and carbonaceous material.
The ratio of sandstone to siltstone is 3:1.
The IDF is subdivided into three main stratigraphic units: the
Upper Isdell Member (UIM), the Lower Limey Unit (LLU) and
the Lower Isdell Member (LIM) (Figure 3).
137
G R HOWARD et al
FIG 3 - Mine area stratigraphic column.
138
FIG 4 - Main Dome I30 Project cross-section 11 325 N.
Upper Isdell Member

The Upper Isdell Member (UIM) is conformably overlain by the
Malu Quartzite Member and overlies the Lower Limey Unit. The
average thickness of the UIM is 280 m and it hosts the
I30 Quartz Reef and eight associated hangingwall reef structures
(A10, A15, A20, A22, A25, A30, A35, A40) (Figure 4). The
hangingwall reefs vary in thickness from 5 cm to 3 m and occur
within 50 m of the I30 Quartz Reef.
The hangingwall Black Silt (BZ) is the lowest bed within the
UIM. It is a 2 m to 5 m thick dark grey or black laminated and
carbonate spotted, carbonaceous siltstone which directly overlies
the LLU and hosts the I30 Quartz Reef.
I30 Quartz Reef

The I30 Quartz Reef is a massive stratabound
quartz-carbonate-sulphide vein that occurs approximately 900 m
to 1000 m beneath the surface at the contact of the UIM and the
LLU. Drilling has delineated the I30 Quartz Reef over an area of
875 m (north-south) by 160 m (east-west) in the south-east
corner of Main Dome.
The geometry of the I30 Quartz Reef is controlled by the
intersection of the I30 monocline fold structure, a north-south
trending reverse fault and a near vertical north-east trending fault
corridor. The quartz reef has a maximum true thickness of 10 m
in the hinge of the I30 Monocline, and reduces to a true thickness
of approximately 0.5 m on the flanks of the fold structure
(Figure 4).
The I30 Quartz Reef mineralisation is characterised by a
gangue consisting of quartz, grey, white and pink dolomite;
calcite and rare siderite. The dominant sulphide minerals are
pyrite and chalcopyrite. The I30 Quartz Reef also contains trace
amounts of bornite, covellite, galena, stibnite, scheelite,
chalcocite and rare telluride minerals. Visible gold is commonly
logged in the diamond core.
Lower Limey Unit

The Lower Limey Unit (LLU) conformably overlies the LIM.
The LLU is a light grey, fine to coarse-grained carbonate altered
sandstone unit. Sand grains are dominantly angular to rounded
clastic quartz grains, accompanied by muscovite flakes with
interstitial micritic dolomite and calcite. The unit varies from
massive to bedded with occasional thin dark silty laminations.
Internal fabric displays a penetrative schistosity, with a shear
fabric reminiscent of a mylonite. Both bedding and veins can
display folding and boudinage. The average true thickness of the
LLU is 9 m but it can vary between 1 m and 22 m in thickness.
In places the LLU contains intense veining concordant and
discordant to bedding.
Lower Isdell Member

The Lower Isdell Member (LIM) is the deepest stratigraphic unit
intersected by mine based drilling programs.
Interpretation of diamond drill holes has identified four
concordant quartz-carbonate-sulphide veins (reefs) in the LIM.
These are named B10, B15, B20 and B30. The B10 reef has an
average true thickness of 0.2 m and is hosted by a 2 m to 3 m
thick carbonaceous silt bed. This reef occurs 20 m to 25 m below
the LLU and has been traced 400 m along strike along the
eastern limb of the fold.
139
G R HOWARD et al
Footwall reefs are not developed in the hinge of the fold. The
deepest reef intersected to-date is the B30. This occurs 65 m
stratigraphically below the LLU Lower Limey Unit.
TELFER EXTENSION PROJECT

PRE-FEASIBILITY STUDY
In 1998, Newcrest Mining Limited Board approval was given to
undertake the evaluation of significant gold and copper
mineralisation below the current open pits at Telfer. These
collectively form the basis of the Telfer Extension (TEX) Project.
The TEX Project consists of the Surface Extension Projects
(SEP) and the I Series Project (ISP). The SEP is evaluating the
potential for deeper open pit gold and copper mining, whilst the
ISP is evaluating the economic potential of the I30 Quartz Reef
and associated narrow-vein hanging wall reef structures,
approximately 1000 m below the surface in the south-east corner
of Main Dome (Figure 3).
Surface Extension Project

The primary geological objectives for the SEP project include:
complete drilling programs in Main Dome and West Dome to
confirm the potential for a bulk open pit minable gold and
copper resource;
estimate gold, copper transitional (cyanide soluble copper)
and sulphide resources for Main Dome and West Dome; and
co-development of geotechnical, hydrological, acid mine
drainage (AMD) and metallurgical models for the project.
The SEP pre-feasibility study was estimated to take 15 months

to complete and commenced in late-1998.
The justification for the Pre-feasibility Study was the
favourable SEP Conceptual Study (Thompson et al, 1998) which
focussed on the south-east corner of Main Dome. The conceptual
study indicated that there was potential for a resource of at least
130 Mt, based on the first generation Main Dome 1 (MD1) and
West Dome 1 (WD1) resource estimates, completed in December
1996.
MD1 and WD1 were the first resource estimates undertaken at
Telfer that incorporated all of the drilling to 300 m below the
surface in two seamless models. Prior to MD1 and WD1 there
were 22 separate resource models and these were not optimal for
evaluating the economic potential of the Telfer mineralised
system.
Drilling
A total of 30 000 m of reverse circulation drilling and 15 000 m
of HQ3 or HQ3.68-3 diamond drilling was completed by
November 1999. The drilling programs focussed on testing the
south-east corner of Main Dome (Figure 2 and 5) with
reconnaissance drilling programs in Pit 9 and 10 in West Dome.
Drill hole spacing has averaged 50 m 50 m.
SEP Resource estimation

Five model updates were completed as part of the SEP
Pre-feasibility Study. These included three estimates covering
Main Dome (MD1.2, MD1.5, MD2) and two in West Dome
FIG 5 - Main Dome SEP Project cross-section 11 050 N.
140
(WD1.5 and WD2). WD1.5 and MD1.2 were interim estimates

using Inverse Distance (ID) interpolation techniques to estimate
the distributions of gold, copper and cyanide soluble copper.
MD2 and WD2 are ordinary kriged estimates and as such form
the basis for the Telfer June 1999 Mineral Resource statement.
The MD2 resource estimate encompasses an area of 3.9 km by
1.9 km and includes all material from surface to the M35 reef
structure, approximately 0.5 km below the surface. The WD2
resource estimate covers an area of 3.3 km by 1.5 km and
includes all material from surface to 0.4 km below the surface.
The primary objectives for developing MD2 and WD2 were to:
utilise linear geostatistic techniques such as ordinary kriging
to compare with ID techniques and to minimise conditional
bias and other effects;
update the models with additional drilling data and a revised
geological interpretation;
model the gold, copper, cyanide soluble copper and sulphur
distributions;
develop AMD models for the mine area; and
undertake a study of the optimal drill hole composite size and
drill hole spacing, for future Telfer resource estimations.
The MD2 estimate comprises 28 primary geological domains,
these include:
12 reef structures (E reefs, MVR, M10/M12, M20, M30 and
M35 reefs),
one M10/M12 stockwork vein domain,
one M10/M12 link structure, and
one surface regolith domain.
The MD2 model is constrained by geological interpretations
based on cross-sections 25 m or 50 m apart. All of the
lithological units are wireframed to provide geological control on
the distribution of mineralisation.
Copper and cyanide soluble copper distributions are modelled
into three domains reflecting the copper depletion, supergene
enrichment and the primary hypogene distribution.
The model parent cell size (X=12.5 m, Y=25 m and Z=4 m),
uses approximately half the drill hole spacing in both the X and
Y directions. The Z direction reflects the 4 m bench height used
in the current open pit operation. Variography including kriging
neighbourhood analysis was used to determine kriging
parametres for each domain.
Resource classification was conducted individually for each
element with the resources being classified based upon
geological confidence, interpretation, QA/QC, number of
samples used and distance criteria.
Prior to release of the MD2 estimate a series of qualitative and
quantitative validation processes were applied to the final grade
model. Table 1 outlines the resources delineated in the study.
I Series Project
The geological objectives of the ISP pre-feasibility study
include:
to confirm the geometry, distribution and tenor of the gold
and copper mineralisation within the I30 Quartz Reef and
proximal reefs; and
to identify the overall geological potential of the deep
mineralised system.
The ISP pre-feasibility study was estimated to take 15 months
for completion and commenced in late-1998.
Historic drilling targeted the I30 Quartz Reef from the surface.
These drill holes are steeply dipping and generally intersected
the I30 Quartz Reef on the western flank of the monocline fold
structure, with occasional drill holes and associated wedge holes
intersecting the eastern flank of the I30 Quartz Reef.
A systematic geological evaluation was undertaken from
November 1997 to June 1998, with the aim of re-interpreting the
available geological data for the I30 Quartz Reef and associated
narrow-vein reef structures. In addition, several targeted diamond
holes were completed from the decline to confirm the apparent
thickening of the mineralisation associated with the monocline
fold and the eastern flank of the I30 Quartz Reef. Drilling
identified significant potential to expand the 1996 I30 Quartz
Reef resource estimate in the hinge area. It was this study that
provided the justification to proceed with the ISP pre-feasibility
study.
Drilling
A total of 15 000 m of HQ 3.68-3 and HQ3 diamond drilling was
completed over 12 months with the average intercept spacing
being 50 m. The drill holes targeted the I30 Quartz Reef hinge
zone and eastern flank from underground drill cuddies at the base
of the decline approximately 500 m above the target. Sample
preparation, assay and QA/QC procedures are the same as for the
SEP Pre-feasibility project (Carlson et al, these proceedings).
ISP Resource estimation

Resource estimation of the I-Series reefs has been carried out
using Ordinary Kriging of a metal accumulation variable (the
product of grade, true width and density). Block grades are
back-calculated by dividing the kriged accumulation by kriged
density and true width. It is important to use an accumulation
variable in the case of I-Series reefs so as to create an additive
variable suitable for kriging from unequal length samples of
varying density.
An unfolding approach has been applied to modelling which
essentially unravels the complex overturned geometry onto a
plane. This method results in greatly improved variography
without compromising the spatial integrity of the model.
TABLE 1
SEP (MD2 and WD2) 1999 Resource estimate (press release 27 September 1999).
Measured
Dry tonnes
(millions)
Gold grade
(g/t Au)
Indicated
Copper grade
(% Cu)
Dry tonnes
(millions)
Gold grade
(g/t Au)
36
1.5
1.7
9.9
Inferred
Copper grade
(% Cu)
Dry tonnes
(millions)
Gold grade
(g/t Au)
110
0.9
Copper grade
(% Cu)
Copper and Gold resources

Total Gold
2.2
Total Copper
0.64
7.8
0.23
80
0.16
Note: Figures are not additive due to rounding. Figures are based on a lower cut-off using a gold price of $A450/oz and copper $A1.10/lb. The resource
was constrained spatially by $A650 notional gold pit shells.
141
G R HOWARD et al
TABLE 2
I Series 1999 Resource estimate (press release 27 September 1999).
Measured
Dry tonnes
(millions)
Gold grade
(g/t Au)
Indicated
Copper grade
(% Cu)
Dry tonnes
(millions)
Gold grade
(g/t Au)
Total Gold
19
Total Copper
Inferred
Copper grade
(% Cu)
Dry tonnes
(millions)
Gold grade
(g/t Au)
1.7
6.8
Copper grade
(% Cu)
Copper and Gold resources

3.3
1.7
0.88
Note: Figures are not additive due to rounding. Figures are based on a lower cut-off using a gold price of $A450/oz and copper $A1.10/lb. The resource
was constrained spatially by a $A650 notional gold shell.
Results of the ISP pre-feasibility study resource estimate were

released in September 1999 and are summarised in Table 2.
CONCLUSIONS
The TEX projects have the potential to extend the mine life of
the Telfer operation. Dedicated management of geological issues,
understanding of critical path elements in resource evaluation,
and continued support from the Newcrest Board have contributed
to identifying this world class gold and copper resource. The
development of geological/structural interpretations and
improved resource estimation techniques have led to a 50 per
cent increase in gold and 70 per cent increase in copper resources
for the ISP project since 1996. The Telfer Gold Mine has played
an important role in the development of the gold industry in
Western Australia, and it is hoped that it will continue to
contribute for a long time to come.
ACKNOWLEDGEMENTS
This paper was compiled from the dedicated efforts of many
geologists from the Telfer Geology Department, and is published
with the permission of the CEO of Newcrest Mining Limited.
REFERENCES
Blockley, J G and De la Hunty, L E, 1975. Paterson Province, in Geology
of Western Australia, Geological Survey of Western Australia,
Memoir 2, pp 109-127.
Carlson, R D, Howard, G R and Back, G, 2000. Systems and Protocols of
Geological Information Management at Telfer Gold Mine, Western
Australia, in Proceedings 4th International Mining Geology
Conference (The Australasian Institute of Mining and Metallurgy:
Melbourne).
Chamberlain, C C, 1990. Telfer - A Case Study of Transition at
Macquarie Bank Gold Conference, Sydney.
Chin, R and Hickman, A H, 1977. Proterozoic geology of the Paterson
Range, WA. 1:250 000 Sheet. Geological Survey of Western
Australia, Record 1977/11 (unpublished).
Dimo, G, 1990. Telfer gold deposits, in Geology of the Mineral Deposits
of Australia and Papua New Guinea (Ed: F E Hughes) pp 643-651
Etheridge, M A, Rutland, R W R and Wyborn, L A I, 1987. Orogenesis
and tectonic processes in the Early to Middle Proterozoic of northern
Australia, Amer Geophys Union Geodynam Ser, 17:131-147.
142
Goellnicht, N M, 1987. Constraints on the Timing and Source of Gold

Mineralisation at Main Dome, Telfer, Western Australia, BSc (Hons)
thesis (unpublished), The University of Western Australia.
Goellnicht, N M, 1992. Late Proterozoic Fractionated Granites and their
Role in the Genesis of Gold and Base-Metal Mineralisation in the
Telfer District, Western Australia. PhD Thesis(unpublished) The
University of Western Australia, 132 p.
Hall, D H and Berry, M, 1989. A discussion of the Telfer exploration
model. Internal Newmont Australia Pty Ltd, Report (unpublished) 7
p.
Hewson, S A J, 1996. A Structural Examination of the Telfer
Gold-Copper Deposit and Surrounding Region, northwest Western
Australia: The Role of Polyphase Orogenic Deformation in
Ore-deposit Development and Implications for Exploration.
Hill, A P, 1989. Structure of West Dome, Telfer, Western Australia and
its significance to mineralisation and regional tectonics. BSc Hons
Thesis, University of Western Australia (unpublished) 63 p.
Inglis, R J, 1995. The paragenesis and structural evolution of the M10
Reef, Telfer, Western Australia. BSc (Hons) thesis, University of
Tasmania (unpublished).
Levet, B K, 1986. Genesis of Telfer ore bodies some alternatives.
Newmont Australia Ltd Rep (unpublished), 8 p.
Newmont Perth and Melbourne Staff, 1972. Report on Exploration of
Paterson Range Gold Prospect Western Australia. Telfer internal
report.
Royle, D Z, 1985. Gold mineralisation at Telfer and its environment
within the Proterozoic Paterson Range Province. Newmont Australia
Ltd Rep (unpublished) 8 p.
Thompson, P, 1998. Telfer Gold Mine Sulphide Extension Project
Conceptual Study, Internal Newcrest Mining Limited Report.
Turner, C C, 1982. The Telfer Gold Deposits, Western Australia,
stratigraphy, sedimentology and gold mineralisation of the
Proterozoic Yeneena Group, PhD thesis (unpublished) University of
New England, 296 p.
Tyrwhitt, D S, 1979. Stratabound gold deposits of Proterozoic age at
Telfer, Western Australia. 85th Annual Convention, Northwest
Mining Association, Spokane, Washington (American Institute
Mining Engineering).
Tyrwhitt, D S, 1985. Exploration, development, and geology of the Telfer
Gold Mine, Great Sandy Desert, Western Australia, in Prospecting in
Arid Areas, First International Conference, Rabat, Morocco, pp
11-19 (The Institution of Mining and Metallurgy: London).
Tyrwhitt, D S, 1995. Desert Gold: The discovery and development of
Telfer (Louthean Publishing Pty Ltd: WA)
Williams, I R and Myers, J S, 1990. Paterson Orogen, in Geology and
Mineral Resources of Western Australia, Western Australian
Geological Survey, Memoir 3, pp 274-275.
Geology and Structure of the Morning Star Mine, Mt Magnet, WA

R Mason1, N Archibald2, D Holden1, T Blyth3, S Huffadine3, R Bradey3, A Jones3 and
P Androvich3
ABSTRACT
The Morning Star mine is owned by Hill 50 Gold NL. Gold is hosted in
quartz-carbonate stibnite molybdenite veins that occur in a series of
deformed, altered and metamorphosed mafic volcanics with subordinate
sediments and felsic intrusive rocks. The irregular gold distribution and
its relationship to structure and alteration intensity makes the definition
of stoping limits difficult. Understanding the structural evolution of the
orebodies and the contemporaneous and later structures that affect the
orebodies has been very useful in this context. This paper reports on a
detailed mapping exercise (as part of a more extensive and on-going
investigation) and the integration of this mapping with other geological
and grade control data in a 3D modelling environment to provide an
expanded basis for stope design.
Three types of mineralised zones can be discriminated based on the
morphology of the veins within. They are:
those associated with strongly folded, thick veins (approximately
1 m thick eg 05 orebody);
those comprising planar, boudinaged and some folded veins which
consolidate to form tabular, steeply dipping bodies (eg 010 orebody);
and
zones containing the above vein types and thin, highly deformed
veins within zones of intensely foliated country rock that result in
localised, irregularly distributed high-grade gold zones (eg
02 orebody).
The mineralised zones associated with thick folded veins are typically
predictable over many tens of vertical metres. The second and third types
of ore zone is generally less predictable. Detailed mapping and structural
kinematic studies indicate all ore zone types are affected by progressive
deformation involving strong strain partitioning. It is interpreted that high
strain zones, recognisable by intense foliation development and highly
deformed veins, are the latest stage of a progressive deformation which
approximates bulk flattening.
Structural controls on mineralisation can be defined at two different
scales.
1.
Regional scale folding has deformed a sequence of mafic volcanics

with interlayered sediments and felsic intrusives. Initial geometry
of felsic bodies relative to orientation of the primary lithological
layering and the bulk shortening direction is interpreted to be
critical for initiation and development of strongly dilatant zones
within this package of rocks, and
2.
Local scale dilation is evidenced by development of several vein

sets whose range in orientations and opposed vergence
relationships suggest a pre-folding origin comprising both
extensional and hybrid shear-extensional types. Continued
deformation resulted in folding of these veins with strong strain
partitioning. Major vein packages are closely spatially associated
with the felsic/mafic contact and an interpreted early structure
which appears to have partitioned early dilation. This structure is
now obscured by strong deformation but is currently represented
by a zone of planar veins.
INTRODUCTION
The Morning Star mine is owned by Hill 50 Gold NL and
includes a number of historically mined pipe-like ore shoots. It is
located approximately 3 km NW of the township of Mt Magnet,
Western Australia (see Figure 1). Mineralisation is usually hosted
1.
Fractal Graphics, 39 Fairway, Nedlands WA 6009.
2.
FAusIMM, Fractal Graphics, 39 Fairway, Nedlands WA 6009.
3.
Hill 50 Gold NL, PO Box 1547, West Perth WA 6872.
FIG 1 - Map showing interpretive geology on pit maps for Star and
Nathan Pits. Interpretation of Evening Star chert between Nathan and
Star Pits represents stratigraphic form surface. The felsic intrusion
defines an opposing fold to that in stratigraphy with a similar axial
surface. Some of the Morning Star orebodies referred to in the text
are shown approximately.
in a strongly quartz-carbonate veined sequence of mafic rocks

(interpreted as basalt lavas) near their contact with a felsic
intrusive body. The basalts belong to the Viqueries Formation
(Thompson et al, 1990) which also includes the Nathan-Easter
chert and a polymict, generally felsic, breccia unit that occurs to
the east of the Morning Star orebodies. To the north and west of
the Morning Star orebodies is the Latecomer fault zone which
comprises a series of discrete strike slip (dextral) faults.
Much of the work described in this paper is a continuation of
work carried out by previous geologists who worked at, and
mapped the lodes of the Morning Star mine. The backs mapping
carried out by these geologists has provided invaluable records of
the geology of the Morning Star lodes. Much of this geology has
been captured in a digital format and has been used extensively
in this study. Geological investigations such as the honours thesis
by Henderson (1981), the report by Ward and Evans (1985) and
the paper by Thompson et al (1990) have detailed various
interpretations of the geology considered during this study.
Regional accounts of the geology by Archibald (1980, 1982)
have provided useful constraints on the regional structure of the
Mt Magnet area.
143
R MASON et al
In this study, detailed geological mapping and 3D modelling

was undertaken in order to characterise the geology, and the
structure in particular, of the Morning Star orebodies. Geological
attributes have been recorded consistently and graphically
represented so that each of these could be manipulated visually
in a computer environment, thus allowing 3D geological models
to be built. The purpose of the mapping and modelling was to
provide a factual record of geological features which could be
used to study and interpret their relationship to mineralisation.
After the 3D geological mapping was completed, analysis and
modelling was undertaken to develop a predictive geological
framework for understanding the distribution of mineralisation
based on the nature of structures, veining and alteration. A
detailed sampling program together with a microstructural study
has been initiated with the aim of finding out exactly where the
gold occurs in relation to the systematically mapped geological
features. To accomplish this, faces with detailed sampling are
being mapped at 1:25 scale. The detailed sampling and
microstructural studies are incomplete at this time. The
microstructural study should enable the relationships between
gold precipitation and vein formation, as well as foliation
development and alteration, to be established.
This paper describes the geology of the Morning Star lodes
with emphasis on a detailed underground mapping and 3D
computer modelling program as part of a more exhaustive but as
yet incomplete study of the geology and mineralisation. This
study is aimed at a better understanding of the mineralisation and
will continue to result in better sampling parameters, stope
definition as well as better risk assessment and management.
initial buckling geometry of the stratigraphy. That is, the initial

shape of the felsic body must have approximated a low
amplitude synform plunging south such that E-W compression
resulted in amplification of this shape. Simultaneously, the
stratigraphy (as outlined by the Evening Star chert) must have
contained an amplifying fold that formed from an initial
low-amplitude buckle whose shape was a south plunging
antiform. The result of this geometric arrangement of
mechanical stratigraphy is a significant space problem between
the two amplifying folds such that accumulating shortening
results in concomitant dilation between the folds (Figure 2).
GEOLOGICAL SETTING
Geometry
Mapping in the Morning Star area has been carried out at 1:500
scale for open pit exposures in both the Star and Nathan pits
(Figure 1) and at 1:100 scale for underground exposures
accessible at the time of mapping campaigns. Several distinct
marker lithologies and important structures have been
recognised. The rocks in the area consist of a series of
metamorphosed and deformed mafic and felsic rocks with minor
sedimentary units. Mapped contacts and the gross geometric
shapes of the various felsic bodies suggest that most are
intrusive. Overall, the main Morning Star felsic body comprises a
synform plunging to the south (Figure 1).
Sediments in the area comprise chert and ferruginous chert,
known locally as both the Evening Star and Easter cherts
(Thompson et al, 1990). The cherts form discontinuous, irregular
bodies which are confined to narrow linear belts and large folded
pods. This distribution along linear belts is interpreted to reflect
the pre-deformation stratigraphy which has been dismembered
mainly by boudinage and extension on fold limbs. The hinge
areas, by contrast, form large bodies with dimensions in tens of
metres. These bodies are typically occupied by folded and
faulted cherts and typically have trailing, narrow, linear belts of
chert (ie the limbs of the fold). Two hinge areas have been
recognised. These are in the Nathan Pit and the Evening Star part
of the Star Pit. These hinge zones are joined by discontinuous,
narrow belts of chert and together form the only well defined
stratigraphic marker in the area. The south plunging, antiformal
hinge zone exposed in the Nathan Pit is immediately south of the
Morning Star system of orebodies and the Morning Star felsic
synform. The two folds approximately share a common axial
surface parallel to the main foliation in the area.
Interpretation
This arrangement of folded lithologies with two folds essentially
opposing each other is interpreted to have formed because of the
initial geometry of the felsic body and its relationship to the
144
FIG 2 - Sequence illustrating buckling of felsic intrusion and stratigraphy

forming opposed folds sharing the same axial surface and generating
significant bulk dilation in the Morning Star area.
This mechanism is proposed to account for the large amount of

dilation, evidenced by the quantity of veining in the Morning
Star area. A question that arises from this interpretation is: Is
dilation concentrated near the felsic/mafic contact or is it
distributed in some systematic way between the two folds? It is
clear that significant dilation is taken up near the felsic/mafic
contact (ie in the vicinity of the Morning Star orebodies) and this
may be due to a large competency contrast between the felsic
rocks and the mafic rocks which focuses dilation on the contact.
GEOLOGY AND STRUCTURE OF THE MORNING STAR MINE, MT MAGNET, WA
The veining in the vicinity of the felsic/mafic contact forms

the majority of the lode deposits mined in the Morning Star area.
Historically these orebodies are numbered from 01 through to the
011 and are generally disposed about the felsic/mafic contact.
The orebodies generally comprise large packages of subparallel,
often folded, veins however it is not yet clear that an entire vein
package is mineralised. Because of this, we use a terminology
that discriminates vein packages from ore (viz the 03 vein
package comprises a series of closely spaced veins, part or all of
which may be mineralised forming the 03 orebody).
veins mapped in the current development with a high degree of

confidence (Figure 4). Reconstruction of the 3D shape of the
felsic contact in some detail has been possible from old and
current mapping and drill core (Figure 5).
PREVIOUS MAPPING
Many of the Morning Star lodes have been mined from a shaft
with a series of levels and sublevels that were developed from the
shaft. Mapping of these exposures, by previous mine geologists,
has provided an invaluable record of the geology of this deposit
from surface down to approximately 100 mRL (2100 level,
approximately 300 m below the surface). Current development is
from a decline and overlaps, in part, with the shaft development.
Visualisation of all mapped veins in 3D demonstrates consistent
and repeatable geometries over significant vertical distances,
albeit with some anomalous zones (Figure 3). Many veins
mapped in the shaft-developed levels can be correlated with
FIG 4 - Correlation of 01W vein package from 2050 level through the
2075 and 2100 levels (good correlation) up into the shaft-developed
level mapping (reasonable correlation).
A planar (NNW oriented) vein zone is readily interpreted

through the shaft-developed level mapping. Here the zone
appears as both highly strained veins and as planar, moderately
deformed veins that terminate other vein sets. In addition, this
planar vein zone appears to have had a very marked effect on the
terminations and shapes of old stopes.
The correlation of vein packages from level to level is based
on vein thickness and fold amplitude and wavelength. Many
veins are remarkably similar over great vertical distances and this
gives much confidence in predicting down plunge continuity.
Importantly, some of these vein packages can be confidently
interpreted throughout the mine although some apparently
discontinuous zones occur.
DETAILED MAPPING AND GEOLOGICAL

OBSERVATIONS
Data collection
FIG 3 - Veins mapped in shaft-developed levels showing consistent vein

geometries over significant vertical distances.
The backs and walls of the nominal 5m 5m development of all

available current development levels were mapped at 1:100 scale.
All mapping has been digitised and accurately registered in 3D
space.
145
R MASON et al
FIG 5 - Detailed model of felsic contact showing irregular nature of the contact suggesting an intrusive (but later folded) origin.
Note the cuspate lobate fold shapes. Felsic contact reconstructed from drilling and mapping. Based on 1 - 5 m spaced interpreted
horizontal sections. Looking approximately North.
The mapped geological features fall into five groups:

1.
lithology,
2.
veins,
3.
alteration,
4.
5.
TABLE 1
Vein types.
Vein
ductile structure (foliations and other fabric elements), and
brittle structure (faults and fractures).

Lithologies comprise two main groups, felsic and mafic rocks
with an intermediate dyke cutting both groups. Many of the
lithological variants are possibly derived from the same, or
similar, protolith but have been discriminated because of
mappable differences due to variations in deformation and
alteration. The felsic rocks are likely to have been derived from a
single, felsic rock of generally uniform composition. Most of the
relatively undeformed and weakly altered fine-grained mafic
rocks are identifiable as pillowed lavas or their unpillowed
equivalents. These are interpreted as a series of basaltic lavas
with intermittently developed pillows. Some lithological types
are interpreted as highly deformed and strongly altered variants
of pillowed and massive basaltic lavas and are so coded to reflect
this deformation and alteration.
Veins have been subdivided on the basis of composition.
Schemes for discriminating vein types based on orientation
become ambiguous when final orientation depends largely on
initial orientation and fold shape. Veins were mapped as
accurately as possible at 1:100 scale and deformation features
such as boudinage, folding and fold asymmetry are represented.
However, many small veins exist which cannot be represented at
the scale of mapping. The vein types discriminated in the
mapping are described in Table 1. Four main alteration types
have been discriminated and are shown in Table 2.
Recorded ductile structures include foliations and their
intensity, and shear zones. Two foliations have been recognised
in some areas, and are common in less altered felsic rocks. The
first formed foliation (based on overprinting relationships) has
been recorded as SI and the second as SII. The second of these
two foliations is by far the most penetrative and forms the
dominant fabric in mafic rocks. Variation in foliation intensity is
assumed to be due to differential shortening, which is partly
dependent on rock type. Foliation intensity has been categorised
146
Description
Vu
Vein undivided.
Vy
Quartz carbonate molybdenite stibnite chlorite

sulphides.
Vm
Quartz carbonate dirty veins.
Vb
Quartz carbonate clean veins with brown carbonate

on vein margins associated with reverse faults.
Vc
Vein, quartz carbonate. Occur in subhorizontal sets.
Vqc
Vein, clean quartz carbonate undivided.
TABLE 2
Alteration types.
Alteration
Description
O-type
Molybdenite and/or stibnite in Vy veins and silver

dusting (? Fine molybdenite) on foliation surfaces.
B-type
Development of khaki-brown-yellow phyllosilicate

mineral in mafic rocks (sericite). Intense and
strong-moderate variations have been discriminated.
S-type
Strong sericitic alteration of felsic rocks; obliterates

primary fabrics.
F-type
Fuchsite (bright green).
as weak, moderate and strong with a moderate foliation intensity

the most common and weak foliation generally developed in
competent lithologies. Strong foliation is recognised by the
development of very continuous foliae in fine-grained mafic
rocks and, if very strong, foliation boudinage (Platt and Vissers,
1980) is developed.
Brittle structures are defined here as those which constitute a

discontinuity in the rock mass (ie veins are treated separately).
These structures comprise:
1.
thrust faults, which often develop within strongly foliated

zones;
2.
strike-slip faults which are steeply dipping and strike NE;

and
3.
barren fractures (eg joints or other fractures with no

infilling).
Several other features are recorded during the mapping as
point measurements. These include orientation measurements of
foliations, lineations and other structural elements. These data
are stored as databases in the modelling environment where they
can be viewed in various formats.
Distribution of lithologies
The contact between the felsic rocks and mafic rocks within the
main mineralised zone has an irregular, but generally arcuate
shape that has the form of a steep south plunging synform. Two
foliations developed, and preserved in the felsic rocks, attest to a
complex deformation history. The irregular shape of the felsic
body is suggestive of an intrusion rather than a conformable part
of the stratigraphy; ie volcanics (Figure 5). The current shape of
the contact has been determined partly by its initial geometry and
partly by deformation events subsequent to emplacement of the
felsic intrusion. Folds in the felsic/mafic contact typically have
cuspate-lobate forms characteristic of folding a competent/
incompetent interface. Surprisingly, the form of these folds
suggests that the felsic was less competent at the time of folding
than the mafic, possibly reflecting the strongly sericitised nature
of the felsic.
A broad zone containing intermittently developed pillowed
basalts trends NNE in the eastern parts of the mine. This may be
subparallel to the local trend of stratigraphy. A medium grained
massive basalt unit locally forms a marker horizon, however its
origin is uncertain. It may be a contiguous part of the volcanic
stratigraphy or it may be an intrusive dyke.
An intermediate dyke develops consistently along an
approximate N-S zone although its detailed shape is often
non-continuous and complex (Figure 6). The dyke is deformed
and two foliations have been observed in it although it is oriented
subparallel to the main foliation (SII).
FIG 6 - Detailed model of intermediate dyke based on current level

detailed mapping and diamond drilling. The dyke is often
non-continuous and complex.
Distribution of strain
The amount of strain accommodated by a particular volume of
rock is assumed to be reflected by the intensity of the foliation.
This assumption allows rapid mapping of strain variations to be
carried out and is based on readily observable characteristics of
the foliation. Weakly foliated rocks show faint or discontinuous
cleavage surfaces whilst moderately foliated rocks have
continuous well-developed cleavage defined by subparallel
alignment of phyllosilicate minerals. Strongly foliated,
fine-grained mafic rocks have very continuous foliation, strong
parallel alignment of phyllosilicates such that the rock has a very
platy look and sometimes develops foliation boudinage. Strongly
foliated zones (high strain zones) are intermittently developed
and appear to form anastomosing arrays. They are generally not
discrete structures and so have poorly defined boundaries and
terminations that can be difficult to discern from moderately
foliated zones.
High-strain zones are generally parallel to the foliation in
strike but can have a shallower dip. The foliation within
high-strain zones is generally parallel to foliation in moderately
strained zones (Figure 7). High-strain zones intersecting the
Morning Star felsic commonly contain fingers of felsic
material. It is unclear what controls the location of areas of
greater strain accommodation, especially since it is not solely
FIG 7 - Equal area projection of poles to foliations:

a) in high strain zones; b) outside high strain zones.
147
R MASON et al
due to intrinsic rock properties. However, high-strain zones

appear to coalesce to the south, away from the felsic contact
(Figure 8). In general, low strain domains occur adjacent to
relatively straight, NW-SE oriented segments of the felsic
contact suggesting that parts of the felsic body acted as buttresses
to the deformation.
Critical relationships
Pre- or syn-SII veins
FIG 8 - Coalescing high strain zones away from (to the south of) the
felsic contact. Mapping of strongly foliated (dark shading) zones and
veins from 2075 level. There are relatively few high strain zones near the
felsic contact but 100 m to the south, exposed rocks are dominated by
high strain zones.
The majority of veins observed in the area of the Morning Star

orebodies are deformed. Commonly, veins are folded or
boudinaged with the main SII foliation either axial planar (folded
veins) or parallel (boudinaged veins) to the veins. Deformed
veins are usually compositional types Vy or Vm.
Folded and boudinaged veins are interpreted to have formed
pre- or syn-SII formation. Folded veins would have formed
initially with an orientation at a low angle to the SII compression
direction while boudinaged veins would have formed with an
initial orientation at a high angle to the SII compression direction.
These two types may have been part of an orthogonal array of
extension veins. Veins with an initial low angle to the
compression direction may have originated as conjugate sets of
hybrid shear-extension veins. Figure 9 shows the geometries of
early-formed Vm and Vy vein types relative to the SII foliation.
Early veins commonly occur in packages which comprise
multiple interconnected and overprinting, composite veins and
vein sets generally with similar orientations. Such packages
indicate significant dilation over, at least, several fracturing
episodes. Moreover, many of these large vein packages occur in
strongly mineralised zones (eg 05, 03, 01W) and there appears to
be a spatial association between large vein packages and
mineralisation. This relationship is obvious from stope shapes
and the veins mapped from parts of the shaft-developed
workings.
FIG 9 - Maps of Vy and Vm veins showing their relationship to the SII foliation. a) Folded Vy veins of the 01W vein package showing
axial planar SII foliation both strongly (closely spaced continuous lines) and moderately (dashed lines) developed. b) Folded and weakly
boudinaged veins of the 010 vein package showing geometric relationship to SII foliation.
148
Some large vein packages (such as the 03 and the 05) are
developed near the felsic/mafic contact. In places these vein
packages occur entirely within the felsic (as evident in the
shaft-developed level mapping) whilst in other areas they form
on the contact zone. Both the veins and the contact are folded.
These observations are interpreted as follows:
the Vc veining event clearly separates (in time) reverse shear

zone and high strain zone formation from strike-slip faulting.
Although they may have a common maximum principal stress
(1) direction the timing relationship described above clearly
separates their development.
1.
Zone of planar veins
2.
a series of generally planar veins formed on the contact, but

where perturbations in the contact occurred the veins
continued straight through, such that in some areas the
entire vein package is within the felsic rocks; then
the veins and contact were folded together during DII
deformation.
Shear zone related veins

Vb veins are commonly associated with high strain zones and
occur as either west dipping reverse shear arrays or as foliation
parallel arrays within, and deflected by, a reverse shear. Unlike
the reverse shear array Vb veins, foliation parallel Vb veins
related to reverse shear zones in high strain zones are difficult to
interpret. This is because they are at a high angle to the
compression direction yet clearly spatially (and probably
genetically) related to the reverse shear zones (Figure 10).
FIG 10 - Sketch showing relationships of Vb veins to reverse shear zones

in high strain zones. The veins form reverse shear arrays and foliation
parallel arrays.
A zone of approximately NNW striking, planar, foliation-parallel

veins, together with folded veins, occurs in the area between the
eastern end of the 05 vein package and the 01W area (Figure 11).
This area is notable and may be distinguished from adjacent
zones by the common occurrence of generally continuous,
planar, foliation parallel veins. Planar, foliation parallel veins
could have been formed in several structural settings, these are:
1.
formed pre-SII and have been strongly flattened;
2.
formed late in SII and have been weakly flattened;
3.
formed post-SII parallel to the foliation; or
formed pre-SII at a high angle to SII, but have been so

strongly deformed during SII (high strain zones) that they
are now transposed and parallel with SII.
These veins are generally slightly boudinaged to planar and so
are probably formed late in SII at a high angle to 1 (ie type 2
above). Some may also be strongly deformed in high strain zones
but are probably not transposed from initially high angles to SII.
4.
FIG 11 - Planar vein zone defined by long relatively continuous planar

veins amongst foliated zones (current level mapping 2075 level
showing planar veins).
Post-SII veins
Vc veins commonly fill a series of subhorizontal fractures, both
in high strain zones and near the felsic contact. Vc veins are
essentially undeformed and appear to post-date all Vb veins.
Quartz-carbonate veins commonly occur in strike-slip faults.
In one critical exposure a strike-slip fault offsets, and therefore
clearly post-dates, a Vc vein. This timing is important because
02 vein package and the felsic/mafic contact

An important geometric relationship is observed in the shape of
the felsic contact and the 02 vein package. Where the 02 vein
package is well developed it coincides with a similarly oriented,
NNW or NW segment of the felsic contact. However, below this
149
R MASON et al
level that segment of the felsic contact becomes more aligned

with the felsic contact immediately to the east (at the 05 vein
package) and the 02 vein package becomes less well-developed
(Figure 12). The interpretation of this relationship is that the
far-field stress tensor is perturbed by the 02 aligned segment of
the felsic/mafic contact, and locally extension is accommodated
by the NW oriented 02 vein package. As this felsic/mafic contact
segment becomes less planar and more east west at depth,
extension is accommodated by the enhanced development of the
05 vein package.
Alteration
Coarse molybdenite in and around veins (O-type alteration) is
strongly associated with gold mineralisation. However, O-type
alteration also occurs in zones outside veins, often near reverse
shear zones in high-strain zones, as finely disseminated
molybdenite which appears as a silver-grey dusting on foliation
planes. This alteration together with high-strain zones is also
strongly spatially associated with Vb veins. B-type alteration is
widespread in mafic rocks particularly around the 02 and 05 to
01W orebodies. S-type alteration occurs in felsic rocks and is
mineralogically different to B-type alteration but is also referred
to as a sericite alteration. Timing relationships between foliation
development and alteration are difficult to constrain in
hand-specimen but it appears that O-type and S-type alteration
may post-date or form in late stages of SII development. B-type
alteration may have pre-dated SII development.
STRUCTURAL SYNTHESIS
General
The felsic/mafic contact appears to have been the locus of
significant dilation evidenced by the intermittent development of
large vein packages. The contact geometry is an important

control on orebodies that are developed in vein packages on the
felsic/mafic contact. These include the 03 and 05 which develop
along the contact and also the 02 which is developed from the
western end of the 05 and to the south (see discussion below).
The 04 orebody also appears to be developed in the vicinity of
the contact whilst the 010 vein package is a series of planar veins
which probably cut the contact but appear to be mineralised
south of the contact.
The 01 and 01W (and the 06) orebodies are somewhat
different in that they are not proximal to the felsic/mafic contact.
However, each of the vein packages associated with these
orebodies terminates at the zone of planar veins which itself
intersects the felsic/mafic contact just east of the 05 vein
package. Moreover, the planar vein zone and the felsic/mafic
contact together connect all of the above mentioned Morning
Star orebodies (01, 02, 03, 04, 010, 05, 01W, 06). Such a
connection suggests that this may be form a significant part of
the plumbing system responsible for fluid flow leading to
mineralisation (Figure 13).
The origin of the planar vein zone remains enigmatic. On the
2100, 2075 and 2050 levels this zone is represented by planar
veins that trend NNE amongst folded veins. It is partly defined
by veins in high-strain zones. To the west of this planar vein
zone, other planar veins have been mapped although their
continuity as zones is less persistent. In the previous mapping
from shaft-developed levels, vein geometries in some areas
suggest the planar vein zone is part of a high strain zone. The 01,
01W and 06 vein packages appear to terminate at this zone and,
in many cases, this is appears not to be a simple case of offset but
rather the vein packages do not exist on both sides of the planar
vein zone. These relationships suggest the possibility that the
planar vein zone may represent a pre-DII structure that
influenced early vein development by partitioning dilation, and
subsequently partly concentrated DII deformation.
FIG 12 - A series of horizontal sections depicting the geometric relationship between a NW oriented segment of the felsic contact and the 02 vein
package. a) 150 mRL. Well-developed NW contact segment and well-developed 02 vein package. (10 m thick horizontal slice) (Veins mapped by
previous mine geologists). b) 130 mRL. oblique view of well-developed NW contact segment and well-developed 02 vein package. (20 m thick horizontal
slice) (Veins mapped by previous mine geologists). c) approximately 50 mRL. NW oriented segment of felsic contact becoming less
planar as part of a transition to a more WNW orientation and concomitant poorly developed 02 vein package. (20 m thick horizontal slice).
150
13B
13A
FIG 13 - Planar vein zone defined by long relatively continuous planar veins amongst foliated zones. a) current level mapping 2075 and 2100 levels
showing planar veins; b) interpretation of planar vein zone through shaft-developed level mapping showing relationship of planar vein zone to old stopes.
Structural evolution
The earliest recognisable deformation in the Morning Star area is
represented by an early foliation (SI) which is commonly
observed in felsic rocks but less so in mafic rocks. Most of the
early foliation has been strongly overprinted by the main
foliation (SII). It is uncertain what macro-scale structural features
may have resulted from this early deformation event but it may
have influenced, in part, the shape of the felsic/mafic contact
prior to DII deformation. Early folds have not been recognised in
the Morning Star area, however their recognition probably relies
on interpreting the larger scale structural geometry. Because little
is known of this early deformation and, by contrast, much is
known about the main foliation forming structural event (SII/DII)
the following discussion on the structural evolution is focussed
mainly on the DII event.
Folds are defined by the felsic/mafic contact, basalt pillow
margins and by veins. No other suitable markers exist in the
current development of the Morning Star underground mine area.
Penetrative deformation resulted in the development of an axial
planar foliation. It is likely that the amplitudes and wavelengths
of folded veins are related to vein thickness as well as to the
amount of shortening. Apart from different areas accommodating
different amounts of shortening, as evidenced by intermittent

development of high-strain zones, folded veins may exhibit
different amounts of shortening if they formed at different times
during a progressive shortening event (eg DII). That is, veins
formed early in DII will be shortened more than veins that have a
comparable orientation but that formed late in DII.
The DII deformation is a progressive, approximately E-W
shortening event can be considered as occurring in a series of
incremental steps. In the first increments, initial buckling of the
felsic/mafic contact occurred and early veins formed in response
to dilation between the Morning Star and the Nathan fold. Thus,
early veins will probably comprise three spatially distinct groups:
1.
a group of approximately E-W steeply dipping extension

veins;
2.
two groups of hybrid shear-extension veins, forming

conjugate shear arrays trending approximately NE and
approximately SE. The acute angle between these sets is
dependent on the differential stress and intrinsic rock
properties.
Early veins may have cut across an incipient SII foliation. The
next increments of shortening resulted in further buckling of the
felsic/mafic contact, further foliation development, initial
151
R MASON et al
buckling of early formed veins and the formation of new veins

(given that every increment of shortening produces some amount
of dilation between the two macro-scale folds). The final
increments of shortening appear to be complex, however
observed relationships (high strain zones and Vb vein types)
indicate strong flattening with (at least) transient periods of high
fluid pressure. Strong flattening is indicated by the presence of
high strain zones. These zones progressively accommodate more
strain than can be accommodated by foliation development
alone. Consequently foliation boudinage develops where small
amounts of slip on curved foliation surfaces helps to
accommodate additional increments of strain. Further
deformation results in development of west dipping, reverse
shear zones whereby strain accommodation is achieved by
movement across the shear zones and finally if the strain rate
becomes too high, brittle failure occurs and a discrete reverse
fault develops. Figure 14 shows the progressive deformation in
increments from incipient SII development through to reverse
shear zones and finally discrete reverse faults. Reverse shear
zones develop en-echelon shear arrays and foliation parallel
veins within the shear zones (discussed previously). This may
also coincide with O-type alteration of wall rocks as this
alteration is frequently associated with high-strain zones. The
development of discrete reverse faults is essentially the last
increment of deformation in the DII event.
Several important processes appear to have operated
contemporaneously during the DII deformation, they are:
folding of the felsic/mafic contact;
formation of the SII foliation;
dilation between the Morning Star synform and the Nathan
antiform;
veins formed progressively and were folded accordingly; and
some zones accommodated relatively greater bulk shortening
and formed high-strain zones.
These processes are depicted in Figure 15.
Following the DII deformation, the next structures to form are
flat lying fractures often filled with vein material. These fractures
are commonly developed in zones of strong foliation (high strain
zones) and this is interpreted to be due to the increased
anisotropy in these zones. Such a strong anisotropy would tend to
concentrate subvertical extensional strain in the strongly foliated
zones and hence develop subhorizontal fractures. At some time
during the development of these fractures the nature of the pore
fluid changed such that some fractures are filled and some are
not.
Strike-slip faults are interpreted to post-date development of
the subhorizontal fractures because of one clear overprinting
relationship.
RELATIONSHIP BETWEEN GEOLOGY AND

MINERALISATION
Introduction
There is a clear association between old stopes and large vein
packages. This suggests that mineralisation is contained in, and
perhaps near, zones of strong dilation. This association may be
interpreted in several ways. Gold may have been precipitated
during vein formation, in which case there should be an equal
chance of finding gold anywhere in that particular vein or
contemporary veins. Alternatively, the large vein packages may
have acted as a competent host lithology and gold may have been
introduced during a later deformation (or increment of
deformation) than that which formed the vein. In this case the
distribution of gold within the vein package may be more
complex and will be influenced by structures that formed during
gold deposition.
152
FIG 14 - Sketch showing interpreted progressive development of high

strain zones during increased shortening: a) incipient SII foliation;
b) well developed SII foliation; c) shortening is no longer accommodated
by foliation development and foliation boudinage forms and reverse
shear zones form; d) discrete reverse faults accommodate further
shortening and vertical extension.
FIG 15 - Cartoon showing three incremental stages of the interpreted evolution of major structures and veins in the Morning Star mine (map view).
a) initial shortening, very low strain, forms various sets of veins including orthogonally arranged extension veins and conjugate sets of hybrid
shear-extension veins. Large extensional veins form along felsic/mafic contact and abut interpreted early structure (planar vein zone). b) increased
shortening, low to moderate strain, early-formed veins and felsic/mafic contact have initiated significant buckle folds, new veins form and SII foliation
is incipient to well-developed. c) maximum shortening now results in attenuated folds in felsic/mafic contact and veins with foliation well developed
and partitioned into high strain zones with lower strain zones buttressed by felsic/mafic contact.
153
R MASON et al
Model for gold mineralisation

The observations described above and those pointed out in earlier
sections have been developed into a model for gold
mineralisation that, at this stage of the study, must be considered
preliminary. The description of this model follows.
Clearly, many of the veins in the large vein packages are
deformed by the DII deformation. The association between
O-type alteration and high strain zones (ie late DII deformation)
has been highlighted earlier. If all O-type (molybdenite)
alteration post-dated Vy vein formation then molybdenite must
have been introduced during late DII. Because of the association
between O-type alteration (molybdenite) and gold mineralisation
it may be inferred that gold was also introduced during late DII
deformation and post-dated many of the veins in the large vein
packages. This would also imply that the molybdenite in Vy
veins was introduced after vein formation and that Vy veins may
have formed originally as just quartz-carbonate veins. This
model is consistent with late-DII deformation being strongly
ductile (eg high strain zones) and yet dominantly brittle in vein
packages. Early-DII deformation appears to have been
dominantly ductile and may have coincided with higher pressure
and temperature conditions and/or a lower strain rate. Because
some high strain zones develop discrete faults (ie brittle
deformation) it is likely that late-DII deformation was near the
brittle-ductile transition given the prevailing pressure,
temperature and strain rate conditions. Therefore, while
deformation in the wall rocks was largely ductile (albeit with
some brittle behaviour as reverse faults), deformation of the
considerably more competent vein material would have been
brittle enabling enhanced fluid flow via fracture induced
permeability. This mechanism has the potential to develop
significant fluid conduits within continuous, large vein packages.
If the pore fluids were transporting gold at this time it would
have been preferentially deposited in the large vein packages
perhaps due to pore fluid pressure drops associated with vein
package fracturing episodes.
Fluid flow is interpreted to have been primarily controlled by
the felsic/mafic contact and the planar vein zone because of the
spatial link these features provide between all orebodies in the
Morning Star deposit.
Implications of the model

If the model for gold mineralisation described above holds true
then several geological features are important for outlining
mineralised zones. Molybdenite bearing (Vy) veins together with
O-type alteration are the key indicators of mineralisation. The
geometries of large vein packages are critical for outlining the
shapes of mineralised zones. These geometries may be predicted
from fold vergence and the plunges of nearby folds but will also
be influenced by high strain zones and late faults. Further work
(detailed sampling and microstructural studies) should establish
what parts of large vein packages are best mineralised and
whether this is systematic and can be predicted.
154
Whilst the aim of this study is to better understand

mineralisation, Morning Star is an operating mine and it is
important that information gathered has been developed into
operating knowledge. The study has combined all historical data
with current data and interpretations into one coherent database,
resulting in efficient use of the input of many geologists over a
vast time span and a common environment for all geological
data. The knowledge accumulated has manifest itself as real
economic benefit, and continues to be a profit augmenting
mining tool.
By better comprehending the nature of the geology and
mineralisation:
sampling procedure has been fine tuned;
the significance of drill intersections is better understood;
mine personnel are now quickly able to discern what
structures and alteration types are important.
3D modelling has allowed:
better down plunge predictions;
confirmation of stope design, ensuring common structures
are connected by design between levels;
decreased risk in all stope and development designs;
the development of a tool for the interpretation and
exploration of undiscovered ore zones; and
better management (and use of) collected geological data.
ACKNOWLEDGEMENTS
The authors acknowledge Hill 50 Gold NL for permission to
publish this paper. Review by Bill Power significantly improved
the paper.
REFERENCES
Archibald, N J, 1980. Geology and structural controls to Gold
Mineralisation, Mt Magnet Area, Carpenteria Exploration Company,
Unpublished report.
Archibald, N J, 1982. Structure, lithological association units and gold
mineralisation, Mt Magnet area, in Archaean Geology of the
Southern Murchison (Compiler: J L Baxter) pp 15-27 (Geological
Excursion Guide 1982 GSA (WA Division)).
Henderson, C A, 1981. The nature and genesis of quartz-gold vein
systems in the Morning Star deposit, Mt Magnet, Western Australia,
BSc (Hons) thesis (unpublished), University of Western Australia,
Perth.
Thompson, M J, Watchorn, R B, Bonwick, C M, Frewin, M O,
Goodgame, V R, Pyle, M J and MacGeehan, P J, 1990. Gold
deposits of Hill 50 Gold Mine NL at Mt Magnet, in Geology of the
Mineral Deposits of Australia and New Guinea, (Ed: F E Hughes),
Melbourne).
Platt, J P and Vissers, R L M, 1980. Extensional structures in anisotropic
rocks, J Struct Geol, 2(4):397-410.
Ward, M and Evans, G, 1985. Untitled, Unpublished company report.
Resource and Reserves

The 1999 JORC Code What Does it Mean
for Todays Mining Geologist?
P R Stephenson
157
Resource Evaluation of Nuggety Slate-Hosted

Gold-Quartz Reefs
S C Dominy, B W Cuffley,
G F Johansen, A E Annels
and I M Platten
169
Multiple Indicator Kriging Is it Suited to My

Deposit?
J Vann, D Guibal and

M Harley
187
Mining Bench Height Evaluation for the Wallaby

Resource A Conditional Simulation Case Study
I M Glacken, M Nopp and

M Titley
195
Ore Definition at the Henty Gold Mine

Tasmania
N Schofield
207
Computer-Based Resource Estimation in

Accordance with the 1999 JORC Code
J Duke and P Hanna
215
The 1999 JORC Code

What Does it Mean for Todays Mining Geologist?
P R Stephenson1
ABSTRACT
The 1999 edition of the Australasian Code for Reporting of
Identified Mineral Resources and Ore Reserves (the JORC
Code) became effective in September 1999 and is the latest
revision of a Code which was originally issued in 1989, but
which has an ancestry dating back to the first JORC report
released in 1972. The modifications introduced in 1999, which
are briefly discussed in this paper, substantially improved the
document, but did not materially alter the fundamentals of the
Code, these being the Mineral Resource/Ore Reserve
classification framework and terminology, and the requirements
and guidelines for public reporting of Resources and Reserves.
Codes or guides similar to the JORC Code exist or are under
development in other countries, including the United States of
America, Canada, the United Kingdom and South Africa. It is
pleasing to note that the 1999 JORC Code has been used as a
model by these countries in developing their latest standards, and
has been a major influence in the formulating of international
reporting standards.
The JORC Code is primarily a document covering the
reporting of the results of mining companies activities to the
public. However, mining geologists are interested in practical
assistance and guidance in carrying out their duties. The JORC
Code provides this in four ways. It establishes a framework
within which those preparing and publicly reporting Mineral
Resource and Ore Reserve estimates must operate, by setting
standards for the classification and reporting of Resources and
Reserves and for qualifications of Competent Persons. It
establishes the division of responsibility between the companys
Board of Directors and Competent Persons with respect to the
preparation and reporting of Resources and Reserves, and
empowers Competent Persons to materially influence how their
Resource/Reserve estimates are reported to the public. It
provides an extensive, detailed and practical check list of criteria
to be taken into account in the preparation of Mineral Resource
and Ore Reserve estimates. Finally, it provides a strong measure
of due diligence protection in the event of litigation if it is
followed with diligence and honesty.
INTRODUCTION
The JORC Code has become well accepted within Australasia as
a standard for public reporting, and has, in recent years, become
increasingly used as a template for other countries reporting
standards, and as a de-facto world standard. In spite of the
ten-year history of its operation in Australasia, some
stakeholders, and particularly mining professionals, are
sometimes not fully aware of the importance of the Code to their
activities and responsibilities, or that it includes many features
which are of positive benefit to their careers. This paper
examines the JORC Code from the perspective of one of those
stakeholder groups, the mining geologist, the professional who is
often at the cutting edge of Mineral Resource and Ore Reserve
estimation. It has been prepared for the 4th International Mining
Geology Conference, Coolum, Queensland, May 2000.
1.
FAusIMM, CPGeo, MMICA, Principal, P R Stephenson Pty Ltd,

Consulting Geologists, PO Box 805, Bendigo Vic 3552.
The author is Chairman of the Joint Ore Reserves Committee,

but the views expressed in this paper are his own and should not
be taken as official guidance from JORC. The paper includes a
number of quotes from the 1999 JORC Code and extracts from
the 1999 Australian Stock Exchange (ASX) listing rules. While
care has been taken to ensure that the quotes and extracts are
accurate, readers should refer to the original documents for
authoritative information.
BRIEF HISTORY OF JORC CODE

The following description is a brief summary of the history of the
JORC Code. For more details, readers are referred to Stephenson
and Miskelly, 1999, Stephenson and Miskelly, 1997 and
Stephenson and Glasson, 1992.
The first edition of the JORC Code was released in 1989, but
the committee itself originated in 1971, following public,
industry and regulatory concern with unacceptable reporting
practices associated with the nickel boom and bust in Western
Australia in the late-1960s. Essentially, JORC was formed
because regulators hinted that unless the mining industry
developed appropriate reporting standards, they (the regulators)
would do so instead. Recognising the warning signs, the
Australian Mining Industry Council (AMIC), now the Minerals
Council of Australia (MCA), established a committee to
examine the issue. It was promptly joined by The Australasian
Institute of Mining and Metallurgy (The AusIMM), resulting in
the formation of the Australasian Joint Ore Reserves Committee
(JORC). The Australian Institute of Geoscientists (AIG)
became JORCs third parent body in 1992.
Between 1972 and 1985, a number of reports were issued by
JORC which made recommendations on public reporting and
Ore Reserve classification and which gradually developed the
principles now incorporated in the JORC Code. The
recommendations had the status of guidelines only, but were,
over time, gradually adopted by most Australasian mining and
exploration companies. It is worth noting that the core concept
on which the Code is built, the Competent Person, was
introduced in JORCs first publication in 1972.
During this period, two key non-JORC documents were
published which had a fundamental effect on the development of
the JORC Code. In 1980, the US Geological Survey released a
document entitled Principles of a Resource/Reserve
Classification for Minerals (US Bureau of Mines and the US
Geological Survey, 1980, commonly known as Circular 831),
in which, for the first time, a clear division between Resources,
representing in situ material, and Reserves, representing
economically extractable material, was presented. Two years
later, Conzinc Riotinto Australia Ltd (CRA) released a seminal
document, entitled A Guide to the Understanding of Ore
Reserve Estimation (King, McMahon and Butjor, 1982), which
set out many of the concepts which now underpin the Code.
Following two years of review, JORC issued, in February
1989, the first version of the JORC Code (Joint Committee of
The Australasian Institute of Mining and Metallurgy and
Australian Mining Industry Council, 1989). Apart from updating
and improving previous documents, this publication differed
from those preceding in two critical ways. It was immediately
incorporated into ASX listing rules, thereby becoming binding
on companies listed on the ASX. It was also immediately
157
P R STEPHENSON
adopted by The AusIMM as an Institute Code, and therefore

became binding on members of The AusIMM. Through these
processes, it became mandatory for both individuals and
companies to conform with the Code, and this has been the
dominant factor underpinning its success. It was adopted as an
AIG Code in 1992 and, in the same year, was incorporated into
New Zealand Stock Exchange (NZSE) listing rules.
Guidelines to the Code were published in 1990 and the two
documents were revised and released in a combined form in
1992 (Joint Committee of The Australasian Institute of Mining
and Metallurgy, Australian Institute of Geoscientists and
Minerals Council of Australia, 1992). In 1993, an Appendix
covering diamond reporting was issued and in 1996, the JORC
Code was again slightly revised, incorporating the Diamond
Appendix and changing the term Pre-Resource Mineralisation
to exploration results, with restrictions on its use (Joint
Committee of The Australasian Institute of Mining and
Metallurgy, Australian Institute of Geoscientists and Minerals
Council of Australia, 1996). A major revision was completed in
early-1999, and the 1999 edition of the JORC Code became
effective in September 1999 (Joint Committee of The
Australasian Institute of Mining and Metallurgy, Australian
Institute of Geoscientists and Minerals Council of Australia,
1999).
MAIN DIFFERENCES BETWEEN 1996 AND 1999

CODES
The 1999 revision of the JORC Code, which had over ten
person-years of committee and industry input, was major, but
none of the modifications materially altered the fundamentals of
the Code. The main changes were:
introduction of a clause (Clause 4) covering the underlying
principles of the JORC Code, those being Transparency,
Materiality and Competence;
adjustments to the definitions of Mineral Resource,
Measured Mineral Resource, Indicated Mineral Resource,
Inferred Mineral Resource, Ore Reserve, Proved Ore
Reserve and Probable Ore Reserve to reflect agreements
reached between organisations participating in an initiative
by the Council for Mining and Metallurgical Institutions
(CMMI) to develop international standard Resource/
Reserve definitions (Miskelly, 1997);
recognition that Measured Mineral Resources may, in certain
circumstances, be convertible to Probable Ore Reserves;
improvements to the definition of a Competent Person to
make it more clearly applicable to those estimating Ore
Reserves as well as those estimating Mineral Resources;
adjustment to the provision for public reports to fairly reflect
documentation prepared by Competent Persons, in order to
maintain compatibility with the 1999 ASX listing rules
which were modified to remove a previous inconsistency
(this is dealt with more fully later in the paper);
expansion of Table 1 to include a check list for the reporting
of exploration results and to provide a more comprehensive
check list for the reporting of Mineral Resources and Ore
Reserves;
clarification of the types of reports covered by the JORC
Code;
simplification of the section on coal reporting by the
insertion of several coal-specific clauses rather than the
inclusion of a separate Coal Code which formed an appendix
in previous editions of the Code;
simplification of diamond reporting requirements by
incorporation of diamond-specific clauses into the main body
of the Code, rather than listing them as a separate section,
thus removing duplicated and superfluous text;
158
strengthening of the provision to explain whether Mineral

Resources have been reported inclusive of, or additional to,
Ore Reserves; now a mandatory requirement;
merging of the Code and Guidelines into a single document
in order to make it more concise and user-friendly, with
guideline notes appearing indented and in a different type
face to Code clauses immediately after the clauses to which
they refer.
In addition to these changes, the ASX introduced in its 1999
listing rules the concept of the recognised mining professional
(refer to Appendix 1). In brief, this allows a company reporting
to the ASX on Resources or Reserves for an overseas deposit, to
base the report on documentation prepared by a person who does
not qualify as a Competent Person because he or she is not a
Member or Fellow of The AusIMM or AIG (but who would
qualify in terms of relevant experience), as long as that person is
a member of a recognised overseas professional body that has
agreed to sanction the person if he or she does not comply with
the JORC Code.
THE 1999 JORC CODE AND THE MINING

GEOLOGIST
So what does the JORC Code mean to todays mining
geologists? What is its relevance to their functions and
responsibilities? In brief, for those mining geologists involved in
Resource/Reserve estimation, its role and importance is
fundamental to their activities. It is a key regulatory standard,
through its incorporation into stock exchange listing rules, and it
therefore has legal ramifications. The Corporations Law requires
a listed company to comply with the ASX listing rules, thereby
indirectly giving these rules and the JORC Code the force of law.
The JORC Code is also the mining industrys accepted standard
in this field, having been adopted as a Code by The AusIMM and
AIG and accepted as a contribution to best practice by other key
organisations. It is therefore of vital importance to mining
geologists professional interests and they should be thoroughly
familiar with its content and implications.
The main aspects of the 1999 JORC Code relevant to mining
professionals are summarised below, and are expanded upon in
sections following. However, as an introductory comment, the
author urges all mining geologists and Competent Persons to
read the 1999 Code, carefully, from beginning to end. Many
users of the Code may have read an earlier edition, and have
assumed that the 1999 Code contains nothing new of relevance to
them. This would be an incorrect assumption, as this paper will
show. Others have probably filed it without any intention of
examining it, a very unwise move. It takes less than an hour to
read the Code from front to back, a worthwhile investment of
time given its importance to the industry and to professionals
practicing in the industry.
Briefly, the JORC Code, in relation to mining professionals:
establishes minimum standards for the public reporting of
Mineral Resources and Ore Reserves which are binding on
members of The AusIMM and AIG and on companies listed
on the ASX and NZSE;
establishes a framework for the classification of Resources
and Reserves;
requires that public reports are based on documentation
prepared by a Competent Person (often, in a mine situation, a
mine geologist);
empowers the Competent Person to influence the content of
public reports based on their documentation;
if followed with diligence and honesty, provides a measure of
due diligence protection to the Competent Person and
company in the event of litigation arising from such public
reports (a point often not appreciated);
THE 1999 JORC CODE
provides a comprehensive check list and guidance notes to
public reports of Mineral Resources and Ore Reserves must
assist the Competent Person in preparing and reporting

exploration results and Resource/Reserve estimates.
In short, the JORC Code is a mixture of minimum standards,
mandatory provisions, empowerment, protection and guidance.
And all in 16 pages!
subdivide the estimates into the appropriate categories, and

these categories must not be reported in a combined form
unless details for the individual categories are also provided;
estimates must not be reported in terms of contained metal or
mineral content unless corresponding tonnages and grades
are also presented; Mineral Resources must not be
aggregated with Ore Reserves (Clauses 26 and 34);
any matters relating to criteria listed in Table 1 which might
materially affect a readers appreciation of public reports on
exploration results, Mineral Resources or Ore Reserves must
be included in public reports (Clauses 18, 27 and 36);
if Resource or Reserve estimates are reported after allowing
for adjustments such as cutting of high grades, or factoring
based on reconciliation with mill or other data; this should be
clearly stated in a public report and the nature of the
adjustment or modification described (Clause 27);
in situations where figures for both Mineral Resources and
Ore Reserves are reported, a clarifying statement must be
included in the report which clearly indicates whether the
Mineral Resources are inclusive of, or additional to the Ore
Reserves (Clause 35);
if a Competent Person prepares documentation for internal
company purposes that does not comply with the JORC
Code, the documentation should include a statement that it
does not comply with the Code, in order to minimise the
likelihood of non-complying documentation being used as a
basis for public reports (guidelines to Clause 5).
In addition, ASX listing rules require the Competent Person(s),
on whose work the public report of Mineral Resources or Ore
Reserves is based, to be named in the report. The report or
attached statement must say that the person consents to the
inclusion in the report of the matters based on their information
in the form and context in which it appears, and must include the
name of the persons firm or employer if that person is not a
full-time employee of the reporting entity (ASX listing rules 5.12
and 5.13).
One way of illustrating how the JORC Code should, and
should not be used in public reporting is by hypothetical
examples of inappropriate reporting. Three such examples, taken
from Stephenson and Miskelly, 1999, are presented in Appendix
2. They are simplified and exaggerated in order to help get the
message across, but all are based largely on real public reports
released in Australia in the last ten years.
Standards: minimum standards for public reporting

The whole of the Code is the minimum standard for public
reporting of exploration results, Mineral Resources and Ore
Reserves in Australasia. To the extent that a mining geologists
activities result in, or affect, a public report on these matters, it is
therefore essential that he or she is thoroughly familiar with the
Code and with the relevant ASX or NZSE listing rules. An
extract from the current (September 1999) ASX listing rules
which relate specifically to mining and exploration activities is
attached as Appendix 1.
Company Directors, not Competent Persons, take
responsibility for public releases of information. However, it may
be necessary under ASX listing rules and Clause 8 of the JORC
Code for listed companies to obtain the approval of Competent
Persons in respect of public releases based on their
Resource/Reserve estimates (discussed later). Competent Persons
should, therefore, be aware of certain specific requirements of
the Code covering such public releases. The more important of
these are:
public releases on Resources and Reserves may use only the

terms specified in the Code (Clause 12);
public releases include, but are not limited to, company
Annual Reports, quarterly reports and other reports to the

ASX or NZSE or required by law; the Code is also a
recommended standard for other reports such as
environmental statements; Information Memoranda; Expert
Reports and technical papers in respect of reporting on
exploration results, Mineral Resources or Ore Reserves, if
they have been prepared for the purpose of informing
investors or potential investors: (guidelines to Clause 5);
the Code encourages the concept of team responsibility in
both Mineral Resource and Ore Reserve estimation, and
recommends that, where there is a clear division of
responsibilities within a team, separate Competent Persons
should accept responsibility for their particular contributions
(guidelines to Clause 10);
if a Competent Person signs off a Mineral Resource or Ore
Reserve estimate prepared by a person who does not qualify
as a Competent Person under the Code, he or she should
appreciate that they are accepting full responsibility for the
estimate and supporting documentation under ASX or NZSE
listing rules and should not treat the procedure merely as a
rubber-stamping exercise (guidelines to Clause 10);
there are specific requirements when publicly reporting on
diamonds (Clauses 19, 27, 36) and coal (Clauses 37 to 40)
estimates of tonnage and average grade must not be assigned
when reporting exploration results which are not part of a
Resource or Reserve estimate (Clause 17);
a Mineral Resource, by definition, must have reasonable
prospects for eventual economic extraction; this requires
preliminary judgements by the Competent Person(s) in
relation to such factors as a lower cut-off grade and minimum
mining parameters (guidelines to Clause 20);
Mineral Resource and Ore Reserve classification is the
responsibility of the Competent Person (Clauses 24 and 32);
public reports of tonnage and grade figures should be
rounded to reflect the order of accuracy of the estimates
(Clauses 25 and 33);
Classification: framework for classification of Mineral

Resources and Ore Reserves
Figure 1 of the Code depicts graphically the terms and categories
used in the Code and the inter-relationship between these
categories. The concept of Mineral Resources as the pre-cursor
to Ore Reserves is now well accepted in Australasia and in most
western countries, although it is a relatively recent development
in the history of the world mining industry, having first been
introduced in Circular 831 of the USA Geological Survey/USA
Bureau of Mines in 1980 (although JORC was the first to apply
the concept to individual companies and deposits).
As previously mentioned, it is important to appreciate that a
Mineral Resource is not an inventory of all mineralisation
drilled or sampled, regardless of cut-off grade, likely mining
dimensions, location or continuity, to quote from the guideline
to Clause 20. It is a realistic inventory of mineralisation which,
under assumed and justifiable technical and economic
conditions, might, in whole or in part, become economically
extractable. The reason for applying preliminary economic
criteria at the Resource stage is that the JORC Code is a
primarily a document which specifies standards for reporting to
the public. The public, when presented with an estimate of
159
P R STEPHENSON
tonnes and grade, must have confidence that the figures represent
material which has either been shown to be economically
extractable (Ore Reserves) or which may reasonably be expected
to become economically extractable (Mineral Resources). While
a company may, for its own internal purposes, prepare estimates
based on cut-off grades or minimum mining dimensions well
below any likely to apply in practice, the public release of such
estimates could be extremely misleading in terms of potential
economic viability.
Prior to a company acquiring sufficient data on a deposit, or
having sufficient confidence in geological and/or grade
continuity to classify any tonnage/grade estimates as at least
Inferred Resources, information may only be reported under
exploration results (Clauses 17 to 19). At this stage, reporting
of tonnage/grade estimates are prohibited, although this does not
restrict a company from providing indications of its exploration
targets or exploration potential. Any tonnage/grade figures
mentioned in this context must be clearly order-of-magnitude
and conceptual in nature and expressed so as not to misrepresent
them as an estimate of Mineral Resources or Ore Reserves
(Clause 17).
The conversion from Mineral Resources to Ore Reserves is
achieved by considering and applying a number of relevant
technical and economic factors, listed in the Code as mining,
metallurgical, economic, marketing, legal, environmental, social
and governmental. This would usually be undertaken as part of a
feasibility study. However, the Code has been deliberately
non-specific in specifying the level of study required to convert
Resources to Reserves, since this may vary according to the
company, commodity, and technical/economic circumstances (in
addition, there currently appears to be no accepted definitions in
the industry for the varying levels of feasibility studies). As long
as the study is sufficiently rigorous to demonstrate at the time of
reporting that extraction could reasonably be justified (Clause
29, definition of Ore Reserve), the conversion would be in
accordance with the Code.
Deciding between classifications

Classification requires the Competent Persons to consider many
factors, and it is easy to lose sight of reality or common sense in
such a situation. A technique which mining geologists may find
useful in focusing the mind when making classification decisions
is to try to imagine the effect which additional sampling data
(usually infill drilling) might have on the tonnage, grade, shape
and location of the mineralised bodies (Stephenson, 1995;
Stephenson and Stoker, 1999).
For example, if it is felt that closer drilling or sampling would
not greatly affect the geological interpretation and/or confidence
in grade distribution, or even if it could affect the interpretation
or grade distribution, would not result in a significantly different
estimate of tonnage, grade, shape and location of the mineralised
bodies, then (assuming that the quality of the data on which the
estimate is based is acceptable) the particular section of the
deposit under question can probably be classified as Measured.
A similar approach can be taken in deciding between Indicated
Resources and Inferred Resources. If it is thought that additional
drilling or sampling could significantly affect the shape and/or
distribution of the mineralised zones, but not substantially affect
the tonnage-grade estimate, then the portion of the deposit in
question can probably be classified as Indicated. An even more
basic approach to this decision can also be used. If the
Competent Person has sufficient confidence in the Resource
estimate for it to be used for decisive mine planning and for
investment decisions, then it probably meets the requirements for
Indicated Resources, and the Competent Person has probably, in
effect, intuitively classified it accordingly. If he or she does not
have sufficient confidence for the estimate to be used for these
purposes, it should probably be classified as Inferred.
The most important criterion to be considered in separating
mineralisation at the exploration results stage from Inferred
Resources relates to assumptions regarding continuity.
Continuity has two components - continuity of geology and
continuity of metal values (Sinclair and Valle, 1994). At the
FIG 1 - 1999 JORC Code

General Relationship between exploration results, Mineral Resources and Ore Reserves.
160
THE 1999 JORC CODE
exploration results stage, the author suggests that there would

usually be doubts as to assumptions that could be made with
respect to both of these components due to the sparsity and/or
quality of data. At the Inferred Resource stage, there would
usually be some confidence in assumptions of geological
continuity but possibly some doubts regarding assumptions of
continuity of metal values, and possibly other concerns of a
technical nature. In both situations there must be sufficient
sampling data available on which to base the judgement of
continuity.
Inferred Mineral Resources may also reasonably be estimated
on the basis of little or no sampling data where the mineralisation
being considered covers limited extensions beyond identified
Indicated and/or Measured Mineral Resources. Knowledge of the
adjoining Resources would usually be sufficient to support such
estimates of Inferred Resources. It is important to keep any such
extensions to reasonable extrapolations which can be clearly
supported by the adjoining data.
An important point to bear in mind when classifying Mineral
Resource and Ore Reserve estimates is that there is nothing in
the JORC Code which requires classification to be a complicated
process, or to be based on a particular procedure such as, for
example, the use of statistical data available from a block model
estimation method. Increasingly practitioners in the industry
appear to becoming fixed on the latter process as if it was
somehow mandatory or the only acceptable approach. In
addition, translating such detailed computational data directly
into Mineral Resource or Ore Reserve categories without the
application of over-riding judgements by the Competent Person
can result, and to the authors knowledge has resulted, in
erroneous or even nonsensical classifications with all the serious
consequences which can flow from misunderstanding the
confidence with which the Resource and Reserve estimates are
known.
As long as classification is carried out by Competent Persons
and as long as they take into account relative confidence in
tonnage/grade computations, confidence in continuity of geology
and metal values, and the quality, quantity and distribution of the
data on which the Mineral Resource or Ore Reserve estimate is
based, the resulting classification, however it is done, should be
acceptable and in accordance with the JORC Code.
Competent Person: requirement that public reports

are based on documentation prepared by a Competent
Person
As previously mentioned, the concept of the Competent Person
has been a core principle of JORC documents since 1972. The
requirement for professional experience and competence in
Resource/Reserve estimation, combined with accountability and
a clear separation of responsibilities between professionals and
company directors, gives the Code the flexibility to be applicable
to a wide range of commodities and situations without the need
to become unreasonably prescriptive.
The reasoning behind the Competent Person concept is that the
public must have confidence that tonnage/grade estimates
presented to them have been professionally compiled, are
soundly based and represent mineralisation that either has been
shown, or has a reasonable prospect of being shown, to be
economically exploitable. Directors take responsibility for any
such public statements, but, as most mining geologists would
know, Resource/Reserve estimation requires a combination of
knowledge, experience, science and art which would rarely be
found on the Boards of most mining and exploration companies.
Hence the requirement for the involvement of Competent
Persons and for the Competent Person to have a minimum of five
years experience which is relevant to the style of mineralisation
and type of deposit under consideration and to the activity which
that person is undertaking (Clause 10 of the 1999 Code).
One of the improvements made in the 1999 revision of the

Code was to amend the definition of a Competent Person (Clause
10) to refer to Ore Reserves as well as to Mineral Resources. In
addition, the guidelines to Clause 10 state that Ore Reserve
estimation is almost always a team effort, with mining engineers
usually occupying the pivotal role. This is an important
clarification for mining geologists, as it has been common
practice in the past (and perhaps still is) for geologists to be
expected or required to sign off Ore Reserve estimates as well as
Mineral Resource estimates, even though the conversion from
Resources to Reserves involves mainly mining-related decisions.
Given the serious responsibilities attached to acting as
Competent Persons, geologists should be careful to ensure that
they have the required qualifications and experience to accept
liability for Ore Reserves estimates. If in doubt, they should
decline to sign and should encourage those with the requisite
experience, usually mining engineers, to accept responsibility.
The Corporations Law requires a listed company to comply
with the ASX listing rules (which incorporate the JORC Code),
and failure to comply can have legal consequences for the
company. Any internal company reports which support public
reports may become discoverable in the event of litigation
arising from such public reports; that is, they may become part of
court evidence. It therefore behoves mining geologists preparing
internal Resource/Reserve reports to ensure that such
documentation is prepared to the highest standards. As a
consultant who has been involved in expert witness activities in
litigation, the author can assure readers that it can be an
unnerving experience for a professional to have to defend his or
her actions in the witness box, and this experience can be
extremely demoralising if the work they are having to defend is
sub-standard or sloppy. Regardless of the outcome of the
litigation, any such public questioning of a Competent Persons
competence or professionalism can have damaging consequences
for that persons reputation and employability.
In any event, since a mining companys Mineral Resources and
Ore Reserves are its fundamental asset, and are often the basis on
which debt finance is raised, it should simply be standard
practice that internal documentation is prepared to high standards
with a full audit trail. Nothing is more frustrating to a
Resource/Reserve auditor than an inability to follow the practices
and procedures employed in the estimation process. If this
frustration leads to costly delays, a requirement for additional
(otherwise unnecessary) work or to the bank declining finance,
the mining geologists employer is not likely to be amused.
Competent Person accountability

The Code gives Competent Persons freedom to use their
experience to decide appropriate estimation and classification
approaches, and provides extensive guidance. This system is
likely to be effective only if the Competent Persons can be made
to account for their actions. In Australasia, obligatory
membership of either The AusIMM or AIG provides the
mechanism by which Competent Persons can be brought to
account, since both organisations are national professional bodies
which have effective and enforced codes of ethics. There is no
doubt that this requirement has been instrumental in achieving
high standards of reporting in Australasia. There have been
instances in which the Ethics Committees of The AusIMM or
AIG have investigated complaints made in respect of reporting
by Competent Persons and action has been taken when deemed
justified. Such action would not have been possible if the
person(s) concerned had not been members of The AusIMM or
AIG.
Overseas, other national Ore Reserve committees have
introduced, or are introducing, similar accountability provisions
161
P R STEPHENSON
for qualified professionals. Although the means of enforcing this

accountability may vary from country to country due to different
regulatory and professional regimes, the principle is accepted as
fundamental to effective reporting codes. International
negotiations in the challenging area of reciprocal recognition of
Competent Persons across international boundaries have been
underway for some time, and encouraging progress is being
made. The introduction of the recognised mining professional
by the ASX in its 1999 listing rules will assist some companies
with overseas deposits to report more easily in Australia, while
maintaining the strict qualification requirements which apply to
Competent Persons.
Competent Person empowerment: the rights and

responsibilities of Competent Persons to influence the
content of public reports based on their documentation
In the 1996 ASX listing rules, there was a difference between the
treatment of the Competent Person as a consultant and the
Competent Person as a company employee. If a Competent
Person was a consultant, the company had to obtain his or her
written approval to the release of a public report which included
information based on the Competent Persons work. However, if
the Competent Person was an employee of the reporting
company, no such approval was necessary. It was only required
that the public report fairly reflect the Competent Persons work.
JORC identified this inconsistency during the revision of the
1996 Code and brought it to the attention of the ASX. The ASX
rectified the situation and the relevant 1999 listing rule (5.13)
now states:
The person referred to in rules 5.10 and 5.11
who compiles the information must consent in
writing to the inclusion in the report of the
matters based on the information in the form and
context in which it appears. The report or
attached statement must state that the person
consents, contain the name of the person and, if
the person is not a full-time employee of the
entity, the name of the persons firm or
company.
Clause 8 of the 1999 JORC Code reflects this listing rule
change.
This change was a significant strengthening of Competent
Persons rights. They are now empowered to materially
influence public reports based on documentation prepared or
supervised by them, and to ensure that such public reports do not
misrepresent their work. However, with rights come
responsibilities. Mining geologists who act as Competent
Persons should ensure that their companies are aware of this
listing rule provision and should exercise their right to review
public reports based on their work. In the event of litigation
arising from such public reports, it would be in Competent
Persons interests to be able to demonstrate that they took
reasonable steps to exercise this right.
Competent Person protection: the role of the JORC

Code in providing due diligence protection
It is often not appreciated that the JORC Code is not just a
document setting out minimum standards for reporting and
providing extensive guidelines for estimating, classifying and
reporting. It is also an invaluable tool for a due diligence
defence, as is the VALMIN Code with respect to assessment and
valuation of mining and exploration properties (Lawrence,
Hancock and Dewar, 1995). The JORC Code represents industry
best practice in Australasia in the field of public reporting of
exploration results, Mineral Resources and Ore Reserves. If
162
professionals/Competent Persons, such as mining geologists,

involved in these fields diligently and honestly comply with the
JORC Code and ASX (or NZSE) listing rules to the best of their
ability when preparing documentation which will form the basis
of a public report, they will have provided themselves with a
strong due diligence defence in the unfortunate event that
litigation arises or is threatened as a result of the public report.
Competent Person guidance: the JORC Code as a

guide to good practice
As has been previously mentioned, the JORC Code does not
constrain Competent Persons in terms of the technical procedures
and practices involved in the preparation of Mineral Resource
and Ore Reserve estimates. In respect of such matters, it leaves
Competent Persons free to exercise their professional judgement
as to the best techniques and methods to apply in their particular
circumstances. However, the Code is not silent on this issue.
Throughout the Code, but particularly in Table 1, extensive
guidance is provided on matters which should be taken into
account in the reporting of exploration results, Mineral
Resources and Ore Reserves, and in the estimation of Resources
and Reserves.
Table 1 in the 1999 Code is a major expansion of the same
table in previous versions of the Code, and was revised with
considerable industry input. It is subdivided into five sections:
Sampling techniques and data
Reporting of exploration results
Estimation and reporting of Mineral Resources
Estimation and reporting of Ore Reserves
Estimation and reporting of diamond mineralisation.
It has the following introductory comment: The order and
grouping of criteria in Table 1 reflects the normal systematic
approach to exploration and evaluation. Criteria in the first
group Sampling Techniques and Data apply to all succeeding
groups. In the remainder of the table, criteria listed in preceding
groups would often apply to succeeding groups and should be
considered when estimating and reporting.
The Table is described as a Check List of Assessment and
Reporting Criteria, and it provides an invaluable list of criteria
for the mining geologist and others involved in the relevant fields
to use when preparing and reporting exploration results, Mineral
Resources and Ore Reserves. The author himself uses it as a
check list when he undertakes audits and due diligence reviews.
All definitions and key clauses in the Code have
accompanying guidance notes to assist in their interpretation.
These are important and should be read by all those involved in
Resource/Reserve estimation. Several are worthy of special
attention by mining geologists acting as Competent Persons
(some of these have been mentioned previously in this paper, but
bear repeating):
As a general guide, persons being called upon to act as
Competent Persons should be clearly satisfied in their own
minds that they could face their peers and demonstrate
competence in the commodity, type of deposit and situation
under consideration. If doubt exists, the person should either
seek opinions from other colleagues or should decline to act
as a Competent Person (guideline to Clause 10, definition of
a Competent Person);
The Competent Person or Persons undertaking this activity
(signing off a Resource/Reserve report prepared by others)
should appreciate that they are accepting full responsibility
for the estimate and supporting documentation under ASX or
NZSE listing rules and should not treat the procedure merely
as a rubber-stamping exercise (guideline to Clause 10,
definition of a Competent Person);
THE 1999 JORC CODE
The term reasonable prospects for eventual economic
extraction implies a judgement (albeit preliminary) by the

Competent Person in respect of the technical and economic
factors likely to influence the prospect of economic
extraction, including the approximate mining parameters. In
other words, a Mineral Resource is not an inventory of all
mineralisation drilled or sampled, regardless of cut-off
grade, likely mining dimensions, location or continuity. It is a
realistic inventory of mineralisation which, under assumed
and justifiable technical and economic conditions, might, in
whole or in part, become economically extractable
(guideline to Clause 20, definition of a Mineral Resource);
Interpretation of the word eventual in this context may vary
depending on the commodity or mineral involved. For
example, for many coal, iron ore, bauxite and other bulk
minerals or commodities, it may be reasonable to envisage
eventual economic extraction as covering time periods in
excess of 50 years. However for the majority of gold
deposits, application of the concept would normally be
restricted to perhaps 20 to 30 years, and frequently to much
shorter periods of time (guideline to Clause 20, definition of
a Mineral Resource);
Mineral Resource or Ore Reserve estimates are sometimes
reported after adjustment by cutting of high grades, or after
the application of modifying factors arising from
reconciliation with mill data. If any of the data are materially
adjusted or modified for the purpose of making the estimate,
or if the estimate is subsequently adjusted, this should be
clearly stated in a Public Report of Mineral Resources or
Ore Reserves and the nature of the adjustment or
modification described (guideline to Clause 27);
The term economic implies that extraction of the Ore
Reserve has been established or analytically demonstrated to
be viable and justifiable under reasonable investment
assumptions. The term Ore Reserve need not necessarily
signify that extraction facilities are in place or operative or
that all governmental approvals have been received. It does
signify that there are reasonable expectations of such
approvals (guideline to Clause 29, definition of an Ore
Reserve).
CONCLUSION
For all mining professionals, but particularly for geologists
involved in Mineral Resource and/or Ore Reserve estimation, the
1999 JORC Code is a very important document.
It is a minimum standard for public reporting, with which the
ASX and NZSE require compliance.
It is a Code which has been adopted by both The AusIMM
and the AIG, and which is therefore binding on the majority
of professionals in Australasia.
It is a source of extensive guidance on matters to be taken
into account in Resource/Reserve estimation and reporting of
exploration results, Resources and Reserves.
It provides professionals with a strong due diligence defence
at law if followed with care and honesty.
Close familiarity with the JORC Code should therefore be
mandatory for all mining geologists. This familiarity should be in
respect of the entire document, not just selected clauses.
Australia can be proud that the JORC Code is being used as a
model for national reporting standards in many countries and for
developing international standards. This is testament to the
robustness of the original concept established over 25 years ago,
to the commitment of JORCs parent bodies and the ASX, and to
the many hours of voluntary effort by members of JORC and
contributors to revisions of the Code.
The JORC Code is a dynamic document which will continue
to evolve to meet the needs of investors, companies,
professionals and regulators in the industry. This evolution is
greatly facilitated by the active participation of interested parties

such as the delegates to this mine geologists conference. The
Joint Committee welcomes all contributions, comments and
constructive criticisms, and will maintain its commitment to
keeping Australasian reporting standards at the forefront of
world best practice.
ACKNOWLEDGEMENTS
The author expresses his appreciation to Mr N Miskelly, Ms R
Phillips, Mr D Nicholls and Mr A Vigar, who kindly reviewed a
draft of the paper.
REFERENCES
Joint Committee of The Australasian Institute of Mining and Metallurgy
and Australian Mining Industry Council, 1989. Australasian Code
for Reporting of Identified Mineral Resources and Ore Reserves.
Joint Committee of The Australasian Institute of Mining and Metallurgy,
Australian Institute of Geoscientists and Minerals Council of
Australia, 1992. Australasian Code for Reporting of Identified
Mineral Resources and Ore Reserves.
Australia, 1996. Australasian Code for Reporting of Identified
Mineral Resources and Ore Reserves.
Australia, 1999. Australasian Code for Reporting of Mineral
Resources and Ore Reserves.
King, H F, McMahon, D W and Butjor, G J, 1982. A Guide to the
Understanding of Ore Reserve Estimation. Supplement to The
AusIMM Proceedings, No 281, March 1982.
Lawrence, M J, Hancock, R G and Dewar, G J A, 1995. The Due
Diligence Defence against Personal Liability for Consultants, in
Proceedings Ethics, Liability and the Technical Expert, a joint
seminar sponsored by The Australasian Institute of Mining and
Metallurgy, Mineral Industry Consultants Association, New South
Wales Section of the Australian Mining and Petroleum Law
Association, Sydney, December 1995.
Miskelly, N, 1997. International Standard Definitions for Reporting of
Mineral Resources and Reserves, in Proceedings The Australasian
Gold Conference, Kalgoorlie, 4-6 March 1997.
Sinclair, A J, and Vallee, M, 1994. Reviewing Continuity: An Essential
Element of Quality Control for Deposit and Reserve Estimation, in
Exploration Mining Geology, 3(2):95-108. (Canadian Institute of
Mining, Metallurgy and Petroleum).
Stephenson, P R and Glasson, K R, 1992. The History of Ore Reserve
Classification and Reporting in Australia, in Proceedings AusIMM
1992 Annual Conference, pp 121-125 (The Australasian Institute of
Stephenson, P R, 1995. Reporting using the Australasian Code for
Reporting of Identified Mineral Resources and Ore Reserves, in The
AusIMM Bulletin, No 2, March 1995, (The Australasian Institute of
Stephenson, P R and Miskelly, N, 1998. The JORC Code, 1987-1997, in
(Eds: D A Berkman and D H Mackenzie) pp 45-51 (The
Stephenson, P R and Miskelly, N, 1999. Reporting Standards and the
JORC Code, in Proceedings JORC Code Seminar, held as part of
Mining Week Victoria, November 1999, Melbourne (Victorian
Chamber of Mines and The Australasian Institute of Mining and
Stephenson, P R and Stoker, P T, 1999. Classification of Mineral
Resources and Ore Reserves, in Proceedings APCOM 99, Computer
Applications in the Mineral Industries, October 1999, Colorado
School of Mines, Denver, USA.
Stephenson, P R and Vann, J, 1999. Common Sense and Good
Communication in Mineral Resource and Ore Reserve Estimation, in
US Bureau of Mines and the US Geological Survey, 1980. Principles of a
Resource/Reserve Classification for Minerals. Geological Survey
Circular 831.
163
P R STEPHENSON
APPENDIX 1
EXTRACT FROM AUSTRALIAN STOCK
EXCHANGE LISTING RULES AS PUBLISHED
1 SEPTEMBER 1999
CHAPTER 5
Additional reporting on mining and exploration
activities
Mining exploration entities and others

5.2 A +mining exploration entity, and an entity which has or
whose +child entity has +acquired an interest in a +mining
tenement, must complete a report (consolidated if
applicable) concerning each quarter of its financial year and
give it to ASX. It must do so no later than 1 month after the
end of the quarter. The report must include each of the
following.
Introduced 1/7/96. Origin: Listing Rule 3B(5) and 3B(6)(d).
5.2.1
TABLE OF CONTENTS
The main headings in this chapter
Rules
When to report
5.1 - 5.3
Requirements for reports
5.6 - 5.17
Terms of a mining tenement joint venture
Appendix 5A
Appendix 5B
Mining exploration entity quarterly report
5.2.2
5.18
Australasian Code for Reporting of Mineral

Resources and Ore Reserves (the JORC Code)
Details of the exploration activities of the entity or

group (including geophysical surveys), and a
summary of the expenditure incurred on those
activities. If there has been no exploration activity,
that fact must be stated.
Details of the mining production and development
activities of the entity or group relating to mining,
mining exploration and related operations, and a
activities. If there has been no production or
development activity, that fact must be stated.
5.2.3
EXPLANATORY NOTE
This chapter sets out disclosure requirements additional to those
in chapter 3 (continuous disclosure) and chapter 4 (periodic
disclosure) that +mining entities and others must satisfy. A
+mining entity includes a +mining producing entity and a
+mining exploration entity.
Where indicated, other entities must comply with requirements
in this chapter. Usually the disclosure is required from an entity
which has, or whose +child entity has, +acquired an interest in a
+mining tenement.
Information for release to the market must be given to ASXs
+company announcements office.
WHEN TO REPORT
5.1 A +mining producing entity, and any other entity that ASX
asks, must complete a report (consolidated if applicable)
concerning each quarter of its financial year and give it to
ASX. It must do so no later than one month after the end of
the quarter. The report must include each of the following.
Details of the mining production and development

activities of the entity or group relating to mining
and related operations, and a summary of the
expenditure incurred on those activities. If there
has been no production or development activity,
5.1.2
A summary of the exploration activities (including

geophysical surveys) of the entity or group, and a
activities. If there has been no exploration activity,
164
Introduced 1/7/96. Origin: Listing Rule 3B(6).

Note: The obligation in this rule will apply to all quarterly reports
from the time when ASX asks until ASX no longer requires it.
Mining exploration entity to complete Appendix 5B

(Quarterly report)
Mining producing entities and others
5.1.1
If ASX asks, the +mining exploration entity, or

entity which has or whose subsidiary has
+acquired an interest in a +mining tenement, must
include each of the following items in each
quarterly report.
The location of +mining tenements held.
The location of +mining tenements +disposed of
during the quarter.
Beneficial percentage interests in farm-in or
farm-out agreements +acquired or +disposed of
during the quarter.
Notes by P R Stephenson. Appendix 5A is the 1999 JORC Code.

Listing rules relating specifically to oil and gas have not been
included in these extracts, nor have all accompanying notes.
Interested readers are urged to obtain a full copy of the 1999 ASX
listing rules and should not rely on the accuracy of these extracts.
Note that the superscript + denotes a term defined in Chapter 19 of
the listing rules.
5.3 A +mining exploration entity must also complete Appendix

5B and give it to ASX. It must do so immediately the
information is available, and in any event within 1 month
after the end of each quarter of its financial year.
Introduced 1/7/96. Origin: Listing Rule 3B(10).
5.4
5.5
Introduced 1/7/96. Origin: Listing Rule 3A(24). Deleted 1/7/98.

Introduced 1/7/96. Origin: Listing Rule 3B(9B). Deleted 1/7/98.
REQUIREMENTS FOR REPORTS

Reports to comply with Appendix 5A
5.6 A report prepared by a +mining entity, or an entity which
has or whose +child entity has an interest in a +mining
tenement, must be prepared in accordance with Appendix
5A if the report includes a statement relating to any of the
following.
Exploration results.
+Mineral resources or +ore reserves.
Introduced 1/7/96. Origin: Listing Rule 3C(5), 3M(13), (14). Amended
1/9/99.
Note: This rule also applies to statements in the annual report. Paragraph 14
of the JORC Code says: A company must promptly report any material
changes in its Mineral Resources or Ore Reserves. Paragraph 15 of the
JORC Code says Companies must review and publicly report on their
Mineral Resources and Ore Reserves annually. Rule 5.6 does not require
entities to comply with these paragraphs - it requires an entity to comply
THE 1999 JORC CODE
with requirements of the JORC Code when it does report, for example, under
a listing rule requirement. As well as the rules in chapter 5, other listing rules
that may requiring reporting include rule 3.1.
5.6.1
However, an entity need not comply with

Appendix 5A to the extent that if rule 5.10 allows a
report to be based on information compiled by a
+recognised mining professional, the report need
not be prepared by or under the direction of and
signed by a +competent person. The requirements
of Appendix 5A applying to a +competent person
apply to the +recognised mining professional.
Introduced 1/9/99.
Note: Paragraph 9 of the JORC Code says: Documentation
detailing Mineral Resources and Ore Reserves estimates from
which a Public Report on Mineral Resources and Ore Reserves is
prepared, must be prepared by or under the direction of, and
signed by, a Competent Person or Persons. The guidelines under
paragraph 10 of the JORC Code say that a listed entity with
overseas interests must nominate a competent person to take
responsibility for mineral resource and ore reserve estimates and
the guidelines in Appendix 1 to the JORC Code include sample
statements relating to the person being a competent person.
However, in the case of mineral resources and ore reserves not
located in Australia, listing rule 5.10 allows a recognised mining
professional (as defined in the listing rules) instead of a
competent person (as defined in the JORC Code) to compile
information.
Example: Paragraph 32 of the JORC Code requires a competent
person to approve categories of ore reserves. If rule 5.10 allows a
report to be based on information compiled by a recognised
mining professional, the recognised mining professional must
approve the categories of ore reserve.
The flow rate.

The choke size used during testing.
Any other relevant basic data.
Introduced 1/7/96. Origin: Listing Rule 3M(10)(b).
Note: This rule also applies to an entity which has, or whose child entity has,
acquired an interest in a well.
Competent person or recognised mining professional to

compile information about minerals
5.10 A report relating to an entitys +mineral resources or +ore
reserves, must be based on information compiled by a
+competent person. However, if the resource or reserve is
not located in Australia, the report may be based on
information compiled by a +recognised mining
professional.
Introduced 1/7/96. Origin: Listing Rule 3M(7). Amended 1/9/99.
Note: This rule applies to an entity which has, or whose child entity has,
acquired an interest in a mining tenement.
5.10.1 The report must either state that it is based on the

information, or be accompanied by a statement to
that effect signed in the same manner as the report.
Introduced 1/7/96. Origin: Listing Rule 3M(7).
5.10.2 If the report is based on information compiled by a

+recognised mining professional, it must include
each of the following statements.
(a)
A statement by the +recognised mining

professional that the report complies with
Appendix 5A (except paragraph 9).
(b)
A statement by the entity that the person is a

+recognised mining professional and the
basis on which each of the requirements for a
+recognised mining professional are met.
Content of reports
5.7 During the exploration stage, a report in the field of mineral
exploration must include the following information.
The type and method of sampling.
The distribution, dimensions, assay results and relative
location of all relevant samples.
Person compiling information about hydrocarbons
Cross reference: paragraph 15 of Appendix 5A.
Any other relevant basic data.
Introduced 1/7/96. Origin: Listing Rule 3M(9)(b).

5.7.1
If true dimensions (particularly width of

mineralisation) are not stated in the report, an
appropriate qualification must be included.
Introduced 1/7/96. Origin: Listing Rule 3M(9)(b). Amended
1/7/98.
5.8 Assay results must be reported using one of the following

methods. The method used must be the most suitable
according to the entitys geologist or mining engineer and
must be stated.
Method 1 All assay results, with sample widths or size in
the case of bulk samples.
Method 2 The weighted average grade of the mineralised
zone, indicating clearly how the grade was
calculated. When high values are recorded, they
must be given in context, with full supporting
data.
Introduced 1/7/96. Origin: Listing Rule 3M(9)(d).
5.9 During the +pre-hydrocarbon reserve stage, a report,

statement or assessment on +hydrocarbon exploration must
include the following information.
The depth of the zone tested.
The age and, if appropriate, the rock type and formation
name of the zone tested.
Any liquids recovered.
Introduced 1/9/99.
5.11 A report relating to an entitys +hydrocarbon reserves must

be based on information compiled by a person who has a
degree (or equivalent) in geology, geophysics, petroleum
engineering or a related discipline; is practising or teaching
geology, geophysics or petroleum engineering; and has
practised or taught one of them for at least five years.
Introduced 1/7/96. Origin: Listing Rule 3M(8). Amended 1/9/99.
5.11.1 The report must either state that it is based on the

information, or be accompanied by a statement to
that effect signed in the same manner as the report.
Person compiling information to be identified

5.12 If the person referred to in rules 5.10 and 5.11 who
compiles the information is a full-time employee of the
entity, the report or attached statement must say so and
name the person.
Introduced 1/7/96. Origin: Listing Rule 3M(7) and 3M(8). Amended 1/9/99.
5.13 The person referred to in rules 5.10 and 5.11 who compiles
the information must consent in writing to the inclusion in
the report of the matters based on the information in the
form and context in which it appears. The report or attached
statement must state that the person consents, contain the
name of the person and, if the person is not a full-time
employee of the entity, the name of the persons firm or
company.
Introduced 1/7/96. Origin: Listing Rule 3M(7) and 3M(8). Amended 1/9/99.
165
P R STEPHENSON
Progress report on geophysical survey

5.14 A report on the progress of any geophysical survey must
include the name, nature and status of the survey, and the
permit under which the survey is being conducted.
APPENDIX 2
HYPOTHETICAL EXAMPLES OF
INAPPROPRIATE REPORTING UNDER THE JORC
CODE
Hypothetical Example 1
HYPOTHETICAL EXAMPLE 1
Deposit
Hydrocarbon reports
5.15 +Probable hydrocarbon reserves must only be reported in
conjunction with +proved hydrocarbon reserves. +Possible
hydrocarbon reserves must only be reported in conjunction
with +proved hydrocarbon reserves and +probable
hydrocarbon reserves.
Mineral
Resources
Ore Reserves
Total
Tonnes
Gold
Tonnes
Gold
Tonnes
Gold
(000s)
(g/t Au)
(000s)
(g/t Au)
(000s)
(g/t Au)
Promising
450.1
2.6
327.8
2.9
777.9
2.7
Auspicious
386.5
2.1
351.0
2.3
737.5
2.2
Introduced 1/7/96. Origin: Definition of hydrocarbon reserves.
Encouraging
56.4
4.2
40.0
4.5
96.4
4.3
Total
893.0
2.5
718.8
2.7
1 611.8
2.6
5.16 A report relating to the +pre-hydrocarbon reserve stage

must not use the word reserves in isolation.
The Mineral Resources and Ore Reserves for the
Propitious epithermal gold deposit are
presented in the above table. The Resources and
Reserves have been calculated in accordance
with the JORC Code by Mr V Sanguine, who has
had extensive experience in the coal and bauxite
mining industries, and who is a member of a
recognised professional association. Although
the Mineral Resources include a substantial
amount of Inferred Resources, the company has
great confidence that further drilling will prove
up these Resources and has therefore included
them in Ore Reserve calculations.
Introduced 1/7/96. Origin: Listing Rule 3M(10)(a).

5.17 A report relating to the results of exploratory investigations

which have reached the stage where a +hydrocarbon
reserve can be estimated must use the expressions for
categories of +hydrocarbon reserves in the listing rules.
Introduced 1/7/96. Origin : Listing Rule 3M(12).
TERMS OF A MINING TENEMENT JOINT

VENTURE
5.18 An entity must not enter a joint venture agreement to
investigate or explore a +mining tenement, unless the
agreement provides that if the entity requires it the operator
will give the entity all the information the entity requires to
comply with the Listing Rules; and that the information
may be given to ASX for release to the market if necessary
for the entity to comply with the listing rules.
Analysis of non-compliance with the JORC Code

1.
There is no breakdown of the Mineral Resources and Ore

Reserves into component categories. This is one of the
most basic reporting requirements of the JORC Code, and
is particularly important in this case, because the report
states that a substantial proportion of the Resources are in
the Inferred category which, by definition, is of low
confidence.
2.
Mineral Resources have been added to Ore Reserves.

Mineral Resources and Ore Reserves are apples and
pears and should never be added together.
3.
The estimator would not qualify as a Competent Person

under the JORC Code. Mr Sanguines experience has been
in the coal and bauxite industries, and he would not,
therefore, meet the relevant experience criteria of the
JORC Code. In addition, being a member of a recognised
professional association is not sufficient to qualify as a
Competent Person in terms of the JORC Code, and may not
be sufficient to satisfy the ASX provision for a recognised
mining professional. The relevant ASX listing rule
requires that the professional body has agreed to sanction
the person if the person does not comply with JORC Code.
4.
The conversion of Inferred Resources to Ore Reserves is

not permitted under the Code. There is no Ore Reserve
equivalent of Inferred Mineral Resources in Australasia.
5.
The statement that the Resources and Reserves have been

calculated in accordance with the JORC Code is not
applicable. Resources and Reserves cannot be calculated
in accordance with the JORC Code, since it does not

Cross reference: rule 3.1.
EXTRACT FROM CHAPTER 19 OF

1999 ASX LISTING RULES
INTERPRETATION AND DEFINITIONS
Definition of a recognised mining professional
Recognised mining professional
A person who has each of the following:
a degree or an overseas equivalent in geology, mining
engineering or a related discipline relevant to the estimation
of the type of mineral resource or ore reserve referred to in
the report.
at least five years experience in the estimation, assessment
and evaluation of the type of mineral resource or ore reserve
referred to in the report.
membership of a recognised overseas professional body that
has agreed to sanction the person if the person does not
comply with Appendix 5A.
Introduced 1/9/99.
166
THE 1999 JORC CODE
regulate estimation techniques or procedures. It establishes

a system of Resource/Reserve classification and sets
minimum standards for public reporting, so any such
statement should refer to classified and reported in
accordance with the Code. In addition, use of the word
calculated in this context is inappropriate; estimated
would be preferable.
context of a description of exploration targets or

exploration potential, in which case any figures mentioned
must be clearly order-of-magnitude and conceptual in
nature and expressed so as not to misrepresent them as an
estimate of Mineral Resources or Ore Reserves.
3.
No information has been provided on the consulting

geologists concerns with regard to the quality of the
drilling data. The Code requires that a public report should
include information on any matters that might materially
affect a readers understanding or interpretation of the
estimates being reported, and in particular on any matters
which might significantly affect confidence in an estimate
of Mineral Resources or Ore Reserves.
Deposit
Mineral Resources
Measured
Indicated
Inferred
Total
4.
Gold
(ounces)
Gold
(ounces)
Gold
(ounces)
Gold
(ounces)
The company has used the term Geological Resources

which has no meaning under the JORC Code.
5.
There is no statement that the public report of Mineral

Resources is based on information compiled by a
Competent Person, nor is the Competent Person named.
These are specific requirements of Australian Stock
Exchange Listing Rules.
Impeccable
495 000
656 000
542 000
1 693 000
Exquisite
210 000
398 000
987 000
1 595 000
Total
705 000
1 054 000
1 529 000
3 288 000
See Table Hypothetical Example 3.
The Geological Resources for the Flawless

gold deposit are tabulated above. In addition to
these Resources, a further 4.6 million ounces of
pre-resource mineralisation has been delineated.
The Companys consulting geologist has
expressed reservations about the quality of the
drilling data on which these estimates are based,
however the Company is confident that these
problems will be readily resolved.
The Mineral Resources and Ore Reserves for the

Congenial copper mine in Outer Harmonia are
12 857 215 tonnes @ 1.382% copper, and 1 605
527 tonnes @ 2.866% copper respectively (see
table below). The figures have been calculated by
Mr M Convivius, a member of the Harmonium
Institute of Engineers, Geologists and Tractor
Mechanics. The mine has a history of positive
reconciliation between Reserve estimates and
mill production, and so the Reserves have been
factored upwards.

1.
2.
The tabulation presents only contained metal figures. The

1999 JORC Code does not permit reporting of contained
metal without corresponding tonnes (or volume) and grade
(in the 1996 Code, such reporting was strongly
discouraged). It is critically important that a reader is made
aware whether a quoted quantity of contained metal is the
product of a large tonnage of low grade material or of a
small tonnage of high grade material.
An estimate has been made of pre-resource
mineralisation. The term pre-resource mineralisation was
deleted from the JORC Code in 1996 and replaced with
exploration results, with a prohibition on the reporting of
tonnage/grade/contained metal figures. The only situation
in which reporting of tonnage/grade figures not classified as
a Mineral Resource or Ore Reserve is permitted is in the

1.
The tabulation includes Possible Ore Reserves, a category

not recognised or permitted under the JORC Code.
2.
There is no statement as to whether the Mineral Resources

are reported inclusive of, or additional to, the Ore Reserves.
The 1999 Code requires such a statement (the 1996 Code
strongly encouraged such practice).
3.
There is a very substantial difference between the tonnages

of Mineral Resources and the tonnages of Ore Reserves.
Where such large differences occur, the public report
should provide a brief explanation to assist the reader in
assessing the likelihood of the remaining Mineral
Resources eventually converting to Ore Reserves.
Mineral Resources
Measured
Indicated
Inferred
Total
Tonnes
Cu (%)
Tonnes
Cu (%)
Tonnes
Cu (%)
Tonnes
Cu (%)
4 367 323
1.322
4 566 778
1.154
3 923 114
1.713
12 857 215
1.382
Ore Reserves
Proved
Probable
Possible
Total
Tonnes
Cu (%)
Tonnes
Cu (%)
Tonnes
Cu (%)
Tonnes
Cu (%)
359 927
2.633
421 156
2.967
824 444
2.916
1 605 527
2.866
167
P R STEPHENSON
4.
168
Membership of the Harmonium Institute of Engineers,

Geologists and Tractor Mechanics would not qualify Mr
Convivius to be a Competent Person under the JORC Code.
In such a situation, and assuming that that Institute was not
able or willing to sanction Mr Convivius for
non-compliance with the JORC Code (a qualifying
requirement under the ASXs recognised mining
professional provision), the company must nominate a
Competent Person to review and sign off the Resource and
Reserve estimates.
5.
The tonnages and grades are given as very precise figures,

which is a poor and potentially misleading practice. If
tonnages and grades are not rounded so as to reflect the
uncertainty surrounding their estimation, then an unrealistic
degree of accuracy is implied and unreasonable
expectations may be generated in the mind of the reader.
6.
No information is given on how and by how much the Ore

Reserves have been adjusted to match mill performance.
Resource Evaluation of Nuggety Slate-Hosted Gold-Quartz Reefs

S C Dominy1, B W Cuffley2, G F Johansen3, A E Annels4 and I M Platten5
ABSTRACT
Nuggety gold-quartz reefs are characterised by localised, high-grades and
in many cases coarse free-gold. Geometrical and grade complexities are
often observed. The high-grade regions are generally erratic and have a
low continuity; though may make the deposit economic. Viable grades
are generally contained within discrete oreshoots, which are surrounded
by barren/low-grade material. Drilling is an effective measure of
geological continuity, however grade distribution can only be reliably
obtained from underground development and bulk sampling. In general
terms, only an inferred resource is likely to be estimated from surface
drill data alone. Underground development, further drilling and probably
bulk sampling will be required to delineate inferred and indicated
resources. Closely spaced development with bulk sampling is likely to be
the best way to determine a probable reserve. The resource evaluation
process must driven by a clear geological model that should attempt to
understand the controls to reef geometry and grade distribution.
Geological interpretation has traditionally been undertaken in
two-dimensions, though computer software is now being used to produce
three-dimensional models. Grade interpolation has generally been
undertaken using classical methods, typically weighted averages, zone of
influence weighting and polygonal/sectional blocks. Tonnages can
generally be calculated from diamond drill and development information
with a reasonable degree of confidence. Grade is much more difficult to
define with confidence because of its highly erratic and discontinuous
nature. To account for variability it is recommended that a best estimate
be reported within a grade range, as this more accurately reflects the
uncertainties involved. Reporting should be undertaken within the
framework of the JORC code, but should involve special consideration of
grade reliability.
INTRODUCTION
Around the world, gold production from reef systems hosted in
metamorphic terrains has dominantly come from greenstone
sequences such as the Yilgarn Block in Western Australia.
Significant production has also come from reefs hosted in low
metamorphic-grade Palaeozoic and younger turbidite sequences
(slate-belt and black-shale type) such as the world-class Central
Victorian Goldfield (eg Bendigo, Ballarat and Tarnagulla) of
Australia and the small Dolgellau Gold-Belt of Wales, United
Kingdom.
Nuggety slate-hosted gold-quartz reefs are characterised by the
presence of highly-localised, erratic occurrences of coarse-free
native gold (Johansen, 1998; Dominy et al, 1997; 1999). Large
variations in grade ranging from <0.1 up to 8000 g/t are
common. Mineable values are either related to erratically
distributed highs throughout the orebody or as erratically
distributed highs throughout discrete oreshoots. Head grades are
often in excess of 8 g/t, though can reach 30 g/t with resource
1.
MAusIMM, Lecturer in Mining Geology - Department of Earth

Sciences, Cardiff University, PO Box 914, Cardiff CF10 3YE, Wales,
UK.
Email: dominys@cardiff.ac.uk
2.
MAusIMM, Exploration Manager - Reef Mining NL, Commercial

Road, Tarnagulla Vic 3551.
3.
MAusIMM, Exploration Manager - Bendigo Mining NL, PO Box

2113, Bendigo Vic 3554.
4.
Principal Mining Geologist - SRK Consulting, Summit House, 9 - 10

Windsor Place, Cardiff CF10 3SR, Wales, UK.
5.
Visiting Fellow - School of Earth and Environmental Sciences,

University of Greenwich, Chatham Maritime, Kent ME4 4TB,
England, UK.
tonnages of a single reef ranging from small (<0.1 Mt) to large

(>1 Mt) (Table 2). Surface diamond drilling is generally only
able to define an inferred resource, with underground
development, in-fill drilling and bulk sampling required to define
indicated/measured resources and reserves.
In this paper, the nature of nuggety slate-hosted gold-quartz
reefs and problems of resource evaluation, grade control and
resource reporting are discussed through a series of case studies.
The critical role of geological understanding is emphasised.
REEF GEOLOGY AND DEVELOPMENT

Quartz-dominated slate-hosted reefs show variations in width
from narrow (<3 m) up to 20 m or more, and display grade and
sometimes-geometrical complexities (Table 1; Figure 1a/b).
Within a particular deposit, the geological development of a reef
is likely to have involved a unique sequence of events that
localise rich-gold domains. Mines in the same district may share
a number of common factors as their deposits formed in the same
general geological environment. The factors controlling
TABLE 1
Characteristics of nuggety slate-hosted gold reefs.
Characteristic
General features
Geological
characteristics
Dip between 0 - 90
Highly variable width (<0.1 - 20 m)
Structural variability: splitting and
branching
Coarse free gold
Slate/shale hosted
Data characteristics
Surface drill holes on typically >40 m

centres
Difficult to get representative samples
Low average concentration and high
variance of Au
Highly skewed grade distribution
Large grade range
Evaluation data
sources
Spatial continuity
Low to high geological continuity

Low-grade continuity
Domaining
Often sharp definition of overall vein

structure although internally there may be
complex intermingling of ore and waste
Vein outlines, vein splitting and faulting
Presence of oreshoots
Interpretation focus
Characterising grade distribution

Identifying extreme high-grades
Study of grade distribution along folds,
faults, etc
Understanding geology
Grade estimation
requirements
Handling extreme high-grades

Nuggety distribution
Estimation along folds and across faults
Grade estimation
techniques
Classical methods
Inverse distance weighting
Geostatistics (rarely used)
Geological mapping
Diamond core drilling
Underground development
Underground bulk sampling/trial mining
169
S C DOMINY, B W CUFFLEY, G F JOHANSEN, A E ANNELS and I M PLATTEN
FIG 1A - Chidlaw Lode (40 level, Link Zone), Gwynfynydd mine,

Gwynedd, UK. Vein (under hammer head), about 0.2 m wide, comprises
quartz intergrown with massive auriferous sphalerite and coarse gold
grading about 1500 g/t. This footwall gold vein (FGV) lay about 2.5 m
off footwall of a larger, 4.5 m wide mother reef. Footwall of FGV in
black Clogau shale and hangingwall in lighter greenstone.
(Source: Welsh Gold PLC).
localisation are at a scale intermediate between that available on

old workings and that provided by a drill core. This means that
interpretation of the core data will be significantly improved if
the gold-rich structures and the controls on their distribution can
be characterised by study of the old working and production data
for the site. Interpretation of the Nick OTime shoot at
Tarnagulla (Cuffley et al, 1998) has been successful because of
detailed data from the mine openings, but at Gwynfynydd mine
(Dominy and Phelps, 1996; Dominy et al, 1996) interpretation
during recent workings was severely restricted by the lack of
detailed geological data on previous mined rich-gold pockets.
There are three levels of geological complexity involved in
localisation of the nuggety gold in many of these deposits:
The internal distribution in the immediate host structure may
be controlled by reef thickness and/or textural development
of the major reef filling components to give an irregular
distribution of gold.
The primary shape and dimensions of the immediate host
reef, which are mostly controlled by slip directions and
initial fracture shape. Fracture pattern and deformation style
may be controlled in part by the behaviour of the rest of the
reef and the reef host rock.
Relationships of the carrying structure to the whole reef,
particularly the effect of later deformation and veining on its
position relative to the gross reef envelope.
These geological features generate the erratic and short-range
characteristics observed during statistical analysis of gold grade
in this style of mineralisation. Of major importance is the
investigation of both geological and grade continuity during the
resource evaluation process (Sinclair and Valee, 1994). Geology,
its interpretation and understanding, forms a key part of resource
evaluation (Dominy et al, 1997; 1999).
170
FIG 1B - Nick OTime shoot (Block 2E/1/mRL990-994), Poverty Reef,

Tarnagulla Project, Victoria, Australia. High-grade laminated
hangingwall vein with spurs.
RESOURCE EVALUATION
Sampling
Diamond drilling of nuggety gold reefs generally gives a good
indication of reef location, etc but is generally not an effective
measure of grade. Assay of drill core gives a feel for grades, but
not an accurate assessment. The importance of drilling lies in the
determination of local geological structure, etc. Drilling
generally enables estimation of a resource tonnage.
Underground development permits detailed mapping and
sampling, the results of which enable grade and geological
continuity to be assessed. Development permits evaluation of the
orebody by bulk sampling (5 - 150 tonne samples) and trial
mining (up to 10 000 tonnes).
Comparison between surface diamond drilling, underground
linear/panel and bulk sampling; shows that bulk sampling is the
best method for grade evaluation in nuggety gold reefs (Dominy
et al, 2000). In summary, the general observations are (Johansen,
1997, 1998; Dominy and Cuffley, unpublished data):
assays using small assay charge sizes (25 - 100 g)
consistently understate assays using large assay charge sizes
(1.5 - 3 kg);
wide-spaced drilling results consistently understate
close-spaced drilling results;
close-spaced drilling results consistently understate bulk
sample results; and
linear and panel sample results understate bulk sample
results.
RESOURCE EVALUATION OF NUGGETY SLATE-HOSTED GOLD-QUARTZ REEFS
TABLE 2
Tonnage and grade of selected nuggety vein gold deposits.
Mine name
County/State
Country
Ballarat East
Victoria
Australia
Various
3 300 000
9.5
Ballarat East
Victoria
Australia
Various
*4 100 000
9.1
Cassidy Creek
Colorado
USA
Cassidy Vein
*50 000
38.5
Cefn Coch
Gwynedd
UK
Cefn Coch Lode
*2 500
17.0
Victoria
Australia
Outer, Inner & Rowes

Reefs
*465 000
14.0
16.7
Central Deborah
Structure
Tonnage(tonne)
Grade(g/t)
Clogau
Gwynedd
UK
Main Reef
*160 000
Cohens
Victoria
Australia
Cohens Reef
*1 400 000
33.3
Eaglehawk
Victoria
Australia
Eaglehawk Reef
*338 000
28.8
Gwynfynydd
Gwynedd
UK
Chidlaw Lode
*113 000
11.9
Gwynfynydd
Gwynedd
UK
180 000
15.0
Gwynfynydd
Gwynedd
UK
Maxwells
Victoria
Australia
New Bendigo
Victoria
Australia
New Bendigo
North Star
Victoria
Australia
California
USA
Chidlaw Lode
3
Collett Reef
10 000
Maxwells Lode
51 000
8.3
800 000
6.6
5
6
Deborah
Williams 40
120 000
7.9
North Star Vein
*1 500 000
16.5
German-Beehive Reef
*116 000
36.7
20 000
36.0
*122 000
92.0
South German
Victoria
Australia
South Gordon
Victoria
Australia
Tarnagulla
Victoria
Australia
Poverty Reef
*56 000
28.6
Poverty Reef
*8 000
7.0
S Poverty reef
433 000
5.0
*1 230 000
10.5
South Gordon Reef

Poverty Reef
Tarnagulla
Victoria
Australia
10
Tarnagulla
Victoria
Australia
11
Tarnagulla
Victoria
Australia
Wattle Gully
Victoria
Australia
12
Phillips Reef
Where possible an individual named orebody is given. An orebody can include one or more oreshoots.
*Historical/recent milled grades/tonnages. All other grades/tonnages based on resource estimates.
1. Drill inferred resource includes a number of reef structures;
2. From West oreshoot;
3. Inferred resource;
4. Recent inferred resource estimate - larger historical tonnage;
5, 6. Inferred resource;
7. Bulked tonnage from a number of oreshoots;
8. Inferred resource;
9. Bonanza oreshoot,
10. Nick OTime oreshoot global production estimate and
11. Poverty Shaft oreshoot; and
12. Inferred resource.
Resource evaluation
Most reefs are relatively narrow tabular structures amenable to
2D representation and evaluation on plans and vertical
longitudinal projections. Computer-based 3D geological and
resource modelling is being applied to this style of mineralisation
more, and is particularly useful when attempting to interpret
historical stoping and geological information.
Conventional evaluation methods that have been applied to
nuggety reefs in the past include weighted averages and
triangular and polygonal techniques (eg Annels, 1991; Vallee,
1992). More recently IDW methods have been found to be most
suitable/acceptable for reefs as they are easy to apply. Typically
for very nuggety structures with least grade continuity, inverse
distance cubed weighting is applied, and an elliptical search area
is used whose long axis is aligned parallel to the oreshoots and
whose axial ratios may be defined by preliminary variographic
analysis. Though 2D-block kriging has been applied, often
combined with indicator kriging where high-grade
subpopulations exist, the nature of these reefs makes variography
difficult and the estimation of parameters unreliable. Frequently
the drilling density is inadequate to define the structure of
variograms at low lags. This is also a problem where composite

or hole-effect variograms are anticipated but where these cannot
be defined because of the inadequate sample coverage at the
exploration stage. Kriging may, however, become more
applicable during the operational phase of a mine when more
detailed sampling has been undertaken.
Reporting resources and reserves

The classification and reporting of reserves and resources in a
mineralised system dominated by erratically distributed gold is
not straightforward. In most cases the following characteristics
are observed:
the tonnage of mineralisation can be defined with reasonable
accuracy;
gold grade shows a high nugget effect and short range;
it is only possible to define a global grade for each zone of
mineralisation; and
drilling and other small volume sampling techniques
significantly understate bulk sampling grades and most likely
the true grade of the mineralisation.
171
Resource evaluation aims to produce an estimate of the deposit

grade and tonnage within the framework of a classification such
as the JORC (1999) code. By their nature, it is extremely difficult
to estimate measured resources (and proven reserves) in nuggety
gold reefs especially at the surface exploration stage when only
inferred and perhaps indicated resources are possible. Indicated
and measured resources require much higher levels of geological
data/exposure which are only likely to be achievable during
underground development. In many cases it is still impossible to
estimate the grade of a block even when it is fully developed and
sampled (Dominy and Phelps, 1996). In this case grades from
historical production data will probably form the basis of the
estimate.
It is recommended that a possible grade range be presented in
a bankable document as well as a grade estimate (Johansen,
1997; Johansen and Stephenson, 1999; Dominy et al, 1999).
Both grade and tonnage figures will be rounded to reflect the
levels of precision of the estimations. For instance grade may be
rounded to one decimal place within a grade range rounded to
the nearest g/t. Tonnage may be rounded to the nearest 10 000
tonnes. It is felt that this approach better reflects the current
situation than reporting a single grade, and as a consequence
better informs investors and the stockmarket.
CASE STUDIES
The following section presents case studies of nuggety
slate-hosted gold reef operations. The examples are drawn from
the authors practical experience and report how the difficulties
of resource evaluation and grade control were approached.
Chidlaw Lode (Link Zone), Gwynfynydd Mine,

Dolgellau Gold-Belt, North Wales, UK
The Gwynfynydd mine is located 8 km to the north of the town
of Dolgellau in North Wales. Since production was first reported
in 1863 over 1.5 tonnes of gold have been produced. Recent
activity began in 1981 under the auspices of Gwynfynydd Mine
Ltd, eventually closing in 1989. The last period of working
commenced during 1992 and terminated in March 1999 due to
lack of accessible resources and environmental constraints. The
operator, Welsh Gold PLC, raised 1.5 million under Rule 4.2 of
the London Stock Exchange in January 1995. Gold produced was
turned into Welsh Gold jewellery, which yields a substantial
premium on the London Bullion Price. Approximately 8000
tonnes were milled annually from underground and surface (old
tips) sources. Underground production was based in an isolated
fault-bounded block of the Chidlaw Lode known as the Chidlaw
Link Zone (Dominy et al, 1996; Dominy and Phelps, 1996;
Figure 2). The company currently (May 2000) operates a tip
reprocessing plant on the site producing gold for the premium
jewellery market.
Geology and mineralisation

The
east-north-east-trending
Chidlaw
Lode
is
a
quartz-sulphide-bearing reef, which is hosted by Cambrian
sediments. Metadolerite sills known locally as greenstones
intrude the sediments. The reef occupies a normal fault and
possesses variable strike, dip and width according to the host
rock type. Multi-stage book and ribbon, and pinch and swell
structures combine to give the reef a complex geometry.
Late-stage barren quartz veins intersect and displace the
gold-bearing structure by metre-scale amounts (Platten and
Dominy, 1999). Gold mineralisation is most likely to occur
where the reef is hosted in the Clogau Formation (carbonaceous
black-shales). Previous studies demonstrate that complex
interaction of auriferous fluids with the carbonaceous rocks led
to gold precipitation. Gold is generally coarse (up to 3 mm),
172
though highly variable, and sometimes intimately associated with

sphalerite. Within the high-grade veins over 65 per cent of the
gold reports to the +300 m fraction and over 80 per cent to the
+150 m fraction. Less than five per cent generally reports to the
53 m fraction.
Within the Chidlaw Lode gold occurs in rich pockets or
subshoots found within an overall easterly plunging oreshoot
(Dominy et al, 1996; Dominy and Phelps, 1996). The oreshoot is
dominantly low-grade, essentially describing a fertile-zone in
which gold-bearing pockets are likely to be encountered
(Dominy et al, 1996). The oreshoot position is partly controlled
by the bedding-reef intersection, with location of pockets related
to variations in reef geometry and local geology. Localised zones
of reef flattening are often gold-bearing and related to dilational
zones proximal to a rigid greenstone body. In some of the older
(1860-1900) Chidlaw workings, subshoots are reported to be
related to reef branching/intersections. Historical and recent
production in the Chidlaw Lode show that the pockets are
generally found in a narrow (<0.4 m wide) footwall gold-vein
(FGV - Figure 1a). This is hosted in the footwall of the larger 4 6 m wide mother structure (Figures 2, 3 and 4). The footwall
zone is considered to be the region up to 5 - 6 m off the true
mother reef footwall. An FGV can lie immediately on the mother
footwall or diverge from it, and display changes in dip from 80
to subhorizontal (Figure 3). The FGV are laterally and vertically
impersistent (in most cases no more than 10 - 40 m along strike
and up dip). A key part of the grade control process is the
effective following and local exploration of the footwall zone.
The importance of the footwall zone was realised early-on
during recent activities. Initial development of the Chidlaw Lode
(Link Zone, 6 level; Figure 2) in 1981 pursued a sharp contact
between the mother reef footwall and Clogau shale. The
driveage, approximately 50 m long, followed sphalerite-galenarich mineralisation along the footwall, but failed to find any
visible gold or notable assays. Later 1.5 - 3.5 m slashing along
the footwall zone revealed a 0.6 - 1.2 m wide zone of sheared
shale/greenstone containing a discontinuous gold-bearing quartz
vein. This zone was traced for 25 m along strike, 15 m down-dip
(below 6 level) and 10 m up-dip (above 6 level); and represents
the FGV. Gold was found throughout the structure, with a
particularly rich patch 12 m below 6 level (Figure 1a); overall the
FGV yielded 31 kg gold from about 2000 tonnes (mill head
grade about 17 g/t).
Grade distribution
Gold grade distribution within the Chidlaw Lode is complex.
Studies of historical and recent production areas have shown that
grades are located within a 40 - 50 m wide (true width), 40 - 50
east-plunging oreshoot. Economic accumulations within the
oreshoot are related to highly localised pockets of gold hosted
within a FGV. Local grades within an FGV may reach 6000 g/t.
Historical records show that the oreshoot contained less than five
per cent high-grade pockets. The largest recorded pocket (1887)
was in the form of a westerly plunging subshoot which produced
approximately 373 kg of gold from 8400 tonnes of rock. The
shoot produced at metallurgical grade of 44 g/t, though the mill
head grade was likely to have been nearer 60 g/t. Smaller
high-grade pockets can contain from 6 kg upwards. Small
background shows will typically yield between 30 - 60 grams of
gold from less than one tonne of rock. Elsewhere within the
oreshoot isolated specks of gold are sometimes observed. Overall
three gold domains are present:
barren: mother reef outside the oreshoot <0.05 g/t;
very-low/medium-grade: oreshoot background grades in
mother reef >0.3 - 30 g/t*; and
high-grade: oreshoot gold-pockets associated with FGV 30
g/t.
FIG 2 - Chidlaw Link Zone, Gwynfynydd mine, Gwynedd, UK. Composite level plan showing location of major block bounding A-B and C faults,
mother reef, FGV (marked quartz leader vein), stopes and major gold pockets between the upper 6 level and lower 110 level.
Levels in the Chidlaw Link Zone are measured in feet below 6 level. (Source: Welsh Gold PLC).
FIG 3 - Chidlaw Link Zone, Gwynfynydd mine, Gwynedd, UK. Schematic section showing reef shape as observed in the CLZ 110 level-west stope,
section through 990E (see Figure 2). FGV (marked quartz leader vein) is seen to flatted and then rise towards the 40 level above.
At the base of the raise to the 40 level, a small show of gold was encountered in the FGV yielding about 30 g gold. Levels in the
Chidlaw Link Zone are measured in feet below 6 level. (Source: Welsh Gold PLC).
173
FIG 4 - Chidlaw Link Zone, Gwynfynydd mine, Gwynedd, UK. Geological plan of the CLZ 110 level-west stope sublevel (8 m above 110 level).
The FGV (marked quartz leader vein) is well exposed in the first 10 m of the stope, where it lies on the base of a Clogau Shale horse on the
mother reef footwall. Beyond 10 m, it passes further into the footwall greenstone body. Later stope lifts corrected for this position.
Channel sampling across the stope back yielded low grades (less than 1 g/t over stope width), though within the FGV grades were locally
17 g/t over 0.35 m. A small show of gold was located approximately 9.5 m along the stope in the FGV yielding 30 g gold. Levels in the
Chidlaw Link Zone are measured in feet below 6 level. (Source: Welsh Gold PLC).
The bulk mineable grade range generally lies between

0.6 - 3.5 g/t which would be achievable by milling all stoped
very-low/medium-grade rock within the oreshoot. In situ grades
are generally within the >0.2 - 6.0 g/t range, though values up to
30 g/t are occasionally found. The presence of high-grade
pockets would raise the overall grade to nearer the historically
mined average grade of 15 g/t.
Resource evaluation
Geological and tonnage evaluation. Prior to full-scale
re-opening in 1992, an underground diamond-drilling program
was undertaken to guide the proposed development in the
Chidlaw Link Zone. It was well understood that drilling would
not enable the estimation of grade, though would provide an
estimate of potentially mineable tonnage. The program was
aimed at determining the:
width and geometry of the Chidlaw Lode below 6 level;
structure and mineralogy of the reef and nature of the
footwall zone;
orientation and position of the A-B and C faults bounding the
Chidlaw Link Zone;
location of the stratigraphic boundary between the Gamlan
and Clogau Formations; and
geomechanical nature of orebody and wallrocks.
Location of the A-B/C faults and Gamlan/Clogau Formation
boundary was of major significance. Strike extent of reef was
controlled by the faults and the lower limit of likely gold
mineralisation by the boundary.
A total of 810 m of BX-core was drilled from 6 level
Little-Lode drive. Twelve holes were drilled on approximately
174
15 m centres to intersect the reef between 30 - 80 m vertically

below 6 level. All holes were advanced by at least 10 m beyond
the intersected reef footwall to investigate footwall geology and
were surveyed every 6 m using a Reflex Fotobor system. Reef
intersections were cut into 1 m lengths and the core sent for gold
and silver assay after detailed logging. In most cases grades were
less than 0.2 g/t, though a few grades of 1 - 3 g/t over 1 m were
found.
The drilling program was successful, showing that:
reef true-width varied between 1 - 4 m, on average 3.5 m;
and the reef showed only a small variation in dip and strike;
reef structure (book and ribbon-type) and mineralogy (quartz
dominated with sphalerite-galena, etc often on the footwall
zone) was comparable to that seen elsewhere in the Chidlaw
Lode;
the position of the A-B and C-faults which diverged with
depth;
Gamlan/Clogau Formation boundary was 70 m vertically
below 6 level;
reef wallrocks were not likely to provide any serious
geomechanical problems; and
the Chidlaw Link Zone contained 24 000 tonnes of potential
ore below 6 level.
In 1992 the classification of reserve (proven and probable) was
applied to the mineralisation defined in the Chidlaw Link Zone.
However, as discussed later, this terminology was subsequently
considered inappropriate considering the extremely low
confidence in the grade estimate. A global specific gravity of
2.65 t/m3 was used for all tonnage calculations.
Grade evaluation. As previously noted, gold grades at

Gwynfynydd are particularly complex and difficult to predict.
Resource grade estimates have been based on bulk sampling/trial
mining and historical production. In 1985 a trial stope centred on
the projected oreshoot yielded 550 tonnes from the Chidlaw Link
Zone above 6 level. Channel samples in the 1.2 - 1.5 m wide
stope were taken on a 1 - 1.5 m grid revealing a wide range of
values (<0.01 - 649 g/t). Weighted averaging of the uncut data
yielded a geological in situ grade of 10.7 g/t. A mill grade of
16.3 g/t was calculated from the tonnage throughput, recovered
gold and tailings assay.
Between 1863-1916 stoping on the Chidlaw Lode 2 level
(above 6 level and the west of A-B fault) yielded 105 000 tonnes
of ore with an average grade of 11.8 g/t. During 1984 - 1990 and
1992 - 93 stoping on the Chidlaw Link Zone (6 level and below)
yielded 7000 tonnes at an average grade of 15 g/t, which was
taken as the adopted historical grade (Unpublished consultants
report, 1994). Tonnages were determined in the usual way, based
on drill hole intersections and geological interpretation. Prior to
mining, a block was defined by the development of a lode drive,
sublevel and raise(s) allowing vein continuity and geometry to be
tested.
The original evaluation used to raise capital in early-1995
reported a grade based on historical production and bulk
sampling and quoted proved and probable reserves (Unpublished
consultants report, 1994). This usage was flawed for this deposit,
because the degree of confidence implied for a reserve grade was
not there. At best only an inferred resource was present.
Subsequent estimates by the technical staff for company reports,
etc used the terms:
potential resource (undeveloped) generally undeveloped
with drill hole intersection(s). Some geological evidence
(probably historical) to suggest that the oreshoot is present.
Preliminary estimate of tonnage based on inferred geological
continuity from drill holes;
inferred resource (partly developed) partly developed
(one level), evidence that an oreshoot is present (possibly
historical data or data extrapolated down-dip). Some
development mullock batch-milled as bulk samples and
backs/faces sampled; and
inferred resource (fully developed) developed on at least
three sides (preferably four), mapped and sampled.
Development mullock batch milled as bulk samples and
backs/faces sampled - ready to mine (highest degree of
confidence, but still low confidence with respect to grade).
Whichever class used, the caveat was always that there was no
guarantee of a high-grade gold pocket(s) being present and that
the mined grade may only lie within the low-grade background
range of the oreshoot. Indicated and measured resources and
reserves were never defined.
Grade control
As no formal resource grade could be estimated, reconciliation
was irrelevant. Grade control activities were critical in expanding
the geological database and developing a better understanding of
the mineralisation. Grade control studies were central to the
mining of a stope panel. As previously discussed, recent
production and historical data showed the presence of a
discontinuous FGV within the mother reef footwall. It was thus
important to understand, as best as possible, the nature of the
footwall zone prior to mining. Historical records and recent
production experience showed that it was easy to develop along
an incorrect footwall structure or loose it during mining (Figure
4). For any block, footwall evaluation was undertaken by the
development of upper and lower lode drives and raises. Only
small stoping blocks were used approximately 15 - 20 m down
dip and along strike. The target horizontal stope width was
1.5 m. Short cuddies, slashes and/or sludge holes were driven
periodically into the stope footwall and hangingwall to check

reef position and possible splitting, etc. Geological mapping of
development was used to delineate reef geometry, the effects of
faulting and identify areas of poor ground conditions (Dominy
and Phelps, 1997). Sections were produced on which the final
stope design was based. Geological data was digitised into
DATAMINE, enabling plans and cross-sections to be produced.
In this style of mineralisation, linear/panel sampling was of
limited value for resource grade evaluation. It was however used
to monitor local grades and to confirm the presence of the
oreshoot which carried a low-medium grade background value.
Development drive backs were chip sampled (about 15 kg) every
3 m and submitted for triplicate standard fire assay externally
(sometimes SFA). The decision on which fire assay method to
employ was based on local geological knowledge and expected
grade. Where assays indicated an anomalous (with respect to
other nearby samples) grade (generally 2 - 8 g/t), a box cut was
fired into the backs to check for visible gold.
Daily chip-panel face samples (drives and stopes; about 50 75 kg) were taken and processed in a surface laboratory. These
were jaw-crushed (-6 mm), ball-milled (-1 mm) and a
concentrate produced using a half-size Wilfley table and spiral
panner. A bench-scale 3 Knelson Concentrator was trailed; this
proved highly effective in processing about 45 kg/hour of
pulverised sample (-1 mm). Sludge holes were drilled into both
hanging- and foot-walls (sampled every 0.5 m); panned and
sometimes sent for fire assay. All development muck within the
oreshoot zone was milled and a mining grade back-calculated.
An individual round(s) of ore could be batch milled as a bulk
sample if required (minimum practical sample size five tonnes).
Grab samples were not routinely taken, though were occasionally
used to monitor muck-pile or truck grades. In most cases a 15 kg
sample was taken, crushed, milled and either split and sent off
for analysis, or a panned concentrate produced.
Underground mill - bulk sampling facility

The underground gravity-based mill was able to process up to 3.5
tonnes/hour. Ore was passed through jaw (-70 mm) and gyratory
(-6 mm) crushers prior to grinding in rod mill (-1 mm) and then
into a 12 Knelson Concentrator. A proportion of the 12
Knelson outflow (about 40 per cent) was passed through a
smaller 7.5 Knelson. The original mill design used the smaller
Knelson, with expansion of the operation in 1995 the 12
Knelson was purchased, replacing the 7.5 as the primary
concentrator. Gold traps were located immediately after the
gyratory crusher and rod mill, the latter collecting up to 40 per
cent of the gold. Products from both Knelson Concentrators was
further cleaned-up using a half-size Wilfley Table prior to on-site
smelting and refining. The final mill concentrate (including
material from both gold traps) was composed of gold and galena
with minor amounts of pyrite, grading up to 6500 g/t.
There was always the likelihood of gold particle lock-up in the
rod mill liners. As a result the following was undertaken:
after stripping and cleaning of the mill circuit flush through
with waste to fill gaps in mill liners (eight tonnes); and
flush the mill through with waste between samples
(minimum 3.5 tonnes).
As mill head grade was highly variable (eg no reserve
estimate), mill recovery was monitored through the smelter gold
return and controlled tailings assay; in general the mill recovered
between 94 - 97 per cent.
Main Reef, Clogau St Davids Mine, Dolgellau

Gold-Belt, North Wales, UK
The Clogau St Davids mine is located 4 km to the northwest of
Dolgellau in North Wales. Since production was first reported in
1854, the mine has produced in excess of 2.6 tonnes of gold.
175
Clogau Gold Mines PLC undertook the first comprehensive,

modern evaluation of the mine during 1984 - 1987. The company
was floated on Londons Unlisted Securities Market in 1984
raising 2.5 million to finance an exploration program. The
program was terminated in 1987 when it was concluded that the
highly erratic, pockety-nature of the deposit would not support a
viable bullion mining operation. More recently the mine has been
operated on a small scale during 1989 - 1998 by Clogau St
Davids Gold Mines Ltd. In early-2000, Cambrian Goldfields Ltd
acquired the mine and are currently evaluating its potential for a
premium Welsh gold jewellery business.

The general geological characteristics are comparable to those
previously reported in Gwynfynydd mine, though the reef is
notably narrower (0.1 - 3.5 m wide). Gold mineralisation is
hosted in the Main Reef and its associated branches (John
Hughes, Paraffin, etc). Within the mine, gold occurs in rich
pockets (or subshoots) found within a series of southwest
plunging oreshoots. It must be stressed that like Gwynfynydd
oreshoots, the Clogau shoots are not dominantly economic,
describing fertile-zones in which gold-bearing pockets are most
likely to be encountered. The oreshoots are related to zones of
vein branching or refraction by metadolerite (greenstone) sills.
Gold is generally coarse (up to 5 mm) and is variably associated
with tellurides (principally tellurobismuth and tetradymite) and
galena. Within high-grade zones (grades >30 g/t) over 75 per
cent of the gold reports to the +300 m fraction and over 85 per
cent to the +150 m fraction. Less than two per cent generally
reports to the -53 m fraction.
Post-mineralisation faulting resulted in normal movements
which occurred on reef margins and fracture planes within
individual reef elements. This caused portions of vein to be
downthrown by up to 10 m, effectively diluting grades within
gold-rich areas of an oreshoot. Minor controls that influence
oreshoot geometry and gold localisation are related to the dip of
the reef and irregularities in its walls. Gold can be preferentially
deposited on the footwall, particularly in shallower dipping
sections and associated with footwall irregularities. Late
post-gold reef veins (elephant quartz; Platten and Dominy, 1999)
occur as thick developments of white quartz. Where this occurs
within rich sections of an oreshoot, they have a deleterious effect
on structure and grade, locally truncating, dislocating and
sometimes folding the gold reef.
Grade distribution
Gold pockets are located within a series of 3 - 15 m long
(along-strike), >50 m along plunge, 20 - 80 southwest-plunging
oreshoots. Historical records show that the most productive
oreshoots contained gold-rich pockets in less than five per cent of
the total shoot volume. Overall, three gold domains are generally
recognisable:
barren:
reef outside the oreshoot <0.05 g/t
very low/
medium-grade:
oreshoot background grades >0.1-15 g/t
high-grade:
oreshoot gold-pocket grades 15 g/t
Activities during the 1980s followed the Llechfraith-No 1

shoot down dip. The shoot plunges 60 southwest, is
approximately 10 m along strike and can be traced 40 m
down-dip from surface to the No 4 level below. The shoot is
related to a split in the reef with high-grades within the
immediate split zone. Gold is extremely erratic with high-grades
(>3000 g/t) within 10s mm of very low-grades (<0.1 g/t). The
overall grade of ore mined was approximately 9 g/t.
176
Resource evaluation and grade control

During its recent periods of operation (post-1984) the mine has
been run as an exploration project with little resource evaluation
sensuo-stricto undertaken. Because of the extremely nuggety
nature of grade, geologists have been pre-occupied in
understanding the geology and controls on oreshoot/gold pocket
localisation rather than trying to define resources. As with the
mineralisation at Gwynfynydd, only an inferred resource could
possibly be defined based on the position and continuity of an
oreshoot. Clogau has the added advantage that the reef is narrow
(ie all reef exposed in development) and the gold is hosted within
it (there are no footwall leaders).
Diamond drilling has rarely been used at Clogau, because of
the impracticability of obtaining representative samples. During
1984 - 1987 some 2000 m of surface and underground drilling
was undertaken for structural/location purposes. As most of the
known oreshoots have already been worked at surface, activities
have been centred on underground exploration. This approach
enabled access to the entire strike length of the workings
(1.5 km) allowing geological mapping and sampling in these
areas to identify potential oreshoots. Since 1984 over 800 m of
new development has been completed. Some 12 000 tonnes of
material has been mined and processed in a gravity mill on-site.
Activities during 1984 - 1986

During this period a systematic comparison was made between
1 m spaced-channel samples (in ore shoot area) and five tonne
bulk samples (box cuts). The latter showed poor grade
correlation, with the channel samples understating the bulk
samples. Ten-tonne bulk samples were used to overcome the
erratic grade distribution and thereby establish reasonably
realistic grades. A small on-site gravity mill was used. All mill
concentrate fractions were assayed to enable the back calculation
of the geological grade for each bulk sample. A split of
underground samples were crushed and panned by hand to give
an indication of gold content, these results generally correlated
well with assays.
Activities post-1989
During 1989 - 1998 mining was on a small-scale (<1500 tonnes
per year) and driven by geological control. Mine plans were
digitised into DATAMINE in an attempt to model local geological
conditions around oreshoots. Mining was guided by geological
mapping which aimed to locate the presence of, and relationship
between, reef splits and greenstone bodies. Mining was highly
selective, employing air-leg development of raises, winzes and/or
sublevels into the geologically defined target zones. When gold
was encountered a small over- or under-hand stope was
developed. Bulk mining of reef proximal to oreshoots was not
routinely undertaken (unless free gold was present), however one
stope on the Upper Level yielded a continuous low-grade source
(2 - 6 g/t). Grade control channel and bulk sampling was
undertaken as part of the development process.
A small 0.75 tonne/hour gravity mill allowed batch milling of
bulk samples backed up by chip-channel samples where
appropriate. The small surface mill comprised jaw (-50 mm) and
gyratory (-10 mm) crushers, a ball mill (-1 mm) and two Micron
Mill Wave Tables (Mark II) for final concentration. Gold traps
were located on the outflow of both crushers, generally
collecting up to 40 per cent of the liberated gold. Grade control
samples were either sent externally for assay or treated on-site.
On-site samples were prepared with a lab-scale jaw crusher
(-5 mm) and ball mill (-1 mm). The sample was then passed over
one of the Micron Mill Wave Tables to produce a concentrate for
visual assay.
Current activities (May 2000) involve historical research and

refinement of the geological model based on detailed mapping
prior to a program of trial mining.
Nick OTime Shoot,

Goldfield, Victoria
Poverty
Reef,
Tarnagulla
The Tarnagulla Goldfield is located 45 km west of Bendigo in

the Central Victorian Goldfield. Gold-bearing quartz reefs
(Poverty Reef and Specimen Reef) were discovered in 1853,
shortly after the initial alluvial gold rush. The Poverty Reef
dominated the field for high-grade and gold yield. Total primary
gold production from the Tarnagulla Goldfield to-date is
estimated to be 14.4 tonnes, of which 11.2 tonnes were mined in
1853 - 1866 from the high-grade Bonanza Shoot within the
Poverty Reef. The Bonanza Shoot was worked to a depth of 120
m from a small outcrop, for a production of 122 000 tonnes at a
head grade of 92 g/t gold (Figure 4 - 1a). From 1995 until
October 1999, Reef Mining NL milled 49 000 tonnes at a head
grade of 28 g/t from the Nick OTime Shoot. It is estimated that
a further 7000 tonnes remains at a grade of 33 g/t which will be
extracted by March 2000. This gives a global resource of 56 000
tonnes at 28.6 g/t (1.6 tonnes contained gold).

The Poverty Reef is one of a number of ellipsoid, en-echelon
quartz segments, which make up the Poverty line of Reef
(Cuffley et al, 1998; Figure 5 - 1b). The steep west dipping
Poverty Reef segment is 650 m in strike length and up to 300 m
in down-dip extent. The segment plunges at 20 south, with the
Bonanza Shoot at the north end and Nick OTime Shoot at the
South end (Figure 5 - 2). The Poverty Reef is a typical fault vein
within tightly folded Ordovician quartz turbidites (Figure 6). The
Poverty Line of Reef has developed as an en-echelon fracture
system located between en-echelon anticlinorial domes. This is
consistent with a torsional stress field setting, the effects of
which can be observed down to mine scale. Detailed geological
interpretation and mapping has resulted in a clearer picture of the
structural development of both reef and oreshoot (Cuffley et al,
1998).
High-grade, narrow, laminated reefs have developed at each
end of a massive quartz segment, where torsional, oblique
strike-slip movement has occurred, creating zones of dilation
(Figure 5 - 3). The competency contrast between the pre-existing
massive quartz and the enclosing metasediments has created an
environment in which mineralised crack-seal quartz reefs have
developed. The gross oreshoot has been broken up into blocks by
post-mineralisation reverse faults (Figure 6). Overall, the Nick
OTime Shoot is a vertical pipe like body extending from 70 m
below the surface to 370 m. The upper 180 m of the shoot is
high-grade and below that a rapid transition to lower grade
occurs. Only the high-grade section has been mined, to a depth
of 250 m below the surface. The laminated high-grade structures
range up to 1.5 m in width and occur as oblique hangingwall and
footwall reefs in an en-echelon pattern around a central core of
low-grade massive quartz, which ranges up to 7 m in width
(Figures 5 - 3 and 7). The combined strike length of the structure
varies from 20 m at the top of the shoot to 70 m at the bottom.
The gold generally occurs in laminated reefs associated with
pyrite, arsenopyrite, galena, tetrahedrite, sphalerite and
chalcopyrite. Isolated gold specks can also occur alone in quartz.
Gold, pyrite and arsenopyrite also occurs in stylolitic fractures
developed around the massive quartz core of the shoot. Several
phases of gold mineralisation are apparent. The primary phase is
associated with the laminated reefs of the shoot and two other
phases are associated with cross-cutting quartz and quartz-albite
veining in faults and fractures.
With the high-grade laminated reefs, gold specks are

disseminated and generally up to 7 mm in size. Larger
gold-quartz specimens have been found in the supergene zone.
Approximately 46 per cent of the gold reports to the +425 m
fraction, 57 per cent to the +300 m fraction and 79 per cent to
the +100 m fraction. Less than ten per cent generally reports to
the 53 m fraction.
Grade
The highest grades within the Nick OTime Shoot occur in the
narrow laminated, oblique reefs splaying off to the north
(footwall reef - FW) and south (hangingwall reef - HW) of the
central quartz core (Figure 7). Vein grades can be in 1000s g/t
(up to 6000 g/t) and commonly in 100s g/t. Grades are erratically
distributed throughout the laminated reefs (Figures 1b and 7) and
appear to be in two distinct domains:
Low-grade:
central massive core 0.1 - 5 g/t
High-grade:
HW and FW laminated reefs >20 g/t
Geologically these statistical units correlate with early phase

low-grade massive quartz reef and late phase shoots
characterised by the laminated reefs formed along reactivated
structures.
Resource evaluation
During 1989, Western Mining Corporation carried out an
evaluation of the Poverty Reef. The study was based on a small
8000 tonne shoot located 150 m below surface adjacent to the
Poverty Shaft. A 970 tonne bulk sample yielded a head grade of
7 g/t. Diamond drill core assays from both surface (NQ) and
underground (BQ) drilling indicated a grade in the 1 - 3 g/t
range. Underground grade control methods included face
chip-channel sampling of each cut (15 - 20 tonnes) and truck
grab sampling. The face samples provided two to three 1.5 kg
samples that were sent for fire assay. Both the drill core and face
samples generally understated the mill grade.
During 1994, Reef Mining discovered the Nick OTime Shoot
by surface diamond drilling on a 50 m grid pattern, along an
interpreted south plunging zone south of the old Poverty Reef
workings. Three of the first 25 drill holes were direct hits down
the vertical axis of the shoot. The nature of the mineralisation in
the discovery holes (disseminated specks in laminated quartz
rather than isolated specimen gold) was very encouraging. The
drilling also demonstrated the continuity of the Poverty Reef to
the south of the old workings.
In 1995 a preliminary estimate based on weighted averages
and polygonal blocks projected to a midpoint between drill holes
was calculated. A top-cut of 30 g/t was selectively used to
remove the influence of abnormal high-grades in the areas of
lower grade mineralisation. The estimate was based on limited
drilling plotted on a vertical longitudinal projection. A global
specific gravity of 2.65 t/m3 was used. Some 30 000 tonnes at
30 g/t was defined in the Nick OTime Shoot (inferred resource)
and 497 000 tonnes of low-medium grade elsewhere in the
Poverty Reef (inferred resource). The low- to medium-grade
resource was purposely not assigned a definite overall grade, as
the density of drilling did not justify this. Rather, the grade was
quoted as ranging up to 10 g/t (ie further exploratory work could
define orebodies within the inferred resource up to 10 g/t in
grade). Classification in the inferred resource category was
regarded as prudent, since uncertainty existed regarding the
overall grade.
Since mining of the Nick OTime Shoot commenced in 1996,
the resource estimate for the high-grade shoot has been
progressively modified and elevated to measured resource
category on the basis of 15 - 20 m spaced underground drilling,
development and bulk sampling. An allowance for mining
177
5924 500N
5925 000N
5925 500N
ec
l
in
e
5926 000N
5926 500N
5927 000N
5927 500N
5928 000N
5928 500N
5929 000N
5930 000N
5929 500N
SOUTH
C
ry
st
al
Hi
ll
NORTH
RMD 105
RMD 85
RMD 83
RMD 112
RMD 111
RMD104
RMD 107
RMD 107A
RMD 104A
1a. Poverty Line of Reef - Long Section
NICK O'TIME
SHOOT
BONANZA SHOOT
RMD 116
RMD 114,118
RMD 84
RMD 106
RL(metres)
0
W1 RMD 79W2
RMD 81
RMD 87C
RMD 87 RMD 67
RMD 74
RMD 79
RMD 87C
RMD 72
500
Unexplored to 1000 metres

1000
North
Yorkshire
Shaft
Poverty Shaft
See 2. Poverty Reef

Quartz Segment Plan
MIN 4756 / EL 3640

0
1b. Poverty Line of Reef - Plan

0
Scale
100 m
North
Nick O'Time High Grade

Shoot
Scale
Bonanza High grade

Shoot
1000 m
POVERTY
SHAFT
See 3. Nick O'Time Shoot Plan
Low - Medium Grade massive Quartz Reef
PRINCE OF WALES SHAFT
2. Poverty Reef Quartz Segment - Inclined Plan View (down 20o S plunge)
0
25 m
Scale
North
Low grade Massive Quartz Core

High Grade Footwall vein
High Grade Hangingwall Vein
Footwall Structure
3. Nick O'Time Shoot. Block 2E/1 (mRL990m) Level Plan
REEF MINING N.L.
Author : B. Cuffley
Date : August 1999
MIN 4756 TARNAGULLA GOLD PROJECT
POVERTY REEF QUARTZ SEGMENT & SHOOTS

Mine Scale to Regional Setting - Schematic Plans
FIG 5 - Tarnagulla Project, Victoria, Australia. Poverty Reef quartz segment and oreshoots; regional to mine scale structural data. 1a Long section of the
Poverty line of reef showing location of Bonanza and Nick OTime shoots; 1b Poverty line of reef plan showing individual reef segments; 2 Poverty
Reef quartz segment showing Bonanza and Nick OTime shoots and low-medium grade massive reef; 3 plan of Nick OTime shoot on 990 m RL in
mining block 2E/1. (Source: Reef Mining NL).
178
POVERTY
SYNCLINE
753050E
752900E
752800E
BIRTHDAY
ANTICLINE
mRL 1200
OCK
RE
HAV
EL
POV
ERT
Y
ZO
EEF
TY R
'A' Fault
NE
mRL 1100
West
Poverty
'Black
Wall'
Structure
EF
REE
F
VER
T PO
WES
Block
4
Block
3
'B' Fault
Concrete Sill
pillar mRL1070
NICK
Block
2W
'D' Fault
O'TIME
BIRTHDAY REEF
Crystal Hill
Decline Loops
SHOOT
Block 2E/1
mRL 1000
REE
F
Cemented sandfill
Sill pillar mRL990
mRL 965
HAV
ELO
CK
Block 2E/2
Block 2E/3
mRL940
Footwall Vein
Hangingwall
Vein
mRL 900
Footwall Fault
Structure
?
0
50m
Scale
REEF MINING N.L.

NICK O'TIME SHOOT - POVERTY & PARALLEL REEFS
Author : B. Cuffley
Date : August 1999
CROSS SECTION 5926220N

FIG 6 - Tarnagulla Project, Victoria, Australia. Cross-section 5926220N through the Poverty Reef (Nick OTime shoot) and parallel reefs. This figure
exemplifies the possible structural (brittle and ductile deformation effects) complexities confronting operators. The high-grade laminated reefs are seen on
the margins of the low-grade core reef. Late-stage reverse faults displace the Poverty Reef (Nick OTime shoot) by 8 - 16 m into a series of isolated
blocks. (Source: Reef Mining NL).
179
SCALE 1 : 500
N
55O-65O
Calculated
Milled
576 t / 0.4g/t
1.2g/t
405 t / 1.4g/t
3.8g/t
981 tonnes
2.3g/t
'Oblique Vein'
(visible gold)
FOOTWALL VEIN
D69.2
NICK
O' TIME
MASSIVE
QUARTZ
CORE
SHOOT
HANGINGWALL VEIN
GRADE OF NICK O'TIME SHOOT

60m STRIKE LENGTH
Preliminary Estimate 20g/t Bulk Grade
(Treatment in progress)
REEF MINING N.L.

CRYSTAL HILL DECLINE, HIGH GRADE NICK O'TIME SHOOT
LEVEL PLAN mRL 990 , BLOCK 2E

GEOLOGY PLAN
FIG 7 - Nick OTime ore shoot, Tarnagulla Project, Victoria, Australia. Plan of the 990-m level showing gold-bearing quartz reef and grades within the
high-grade ore shoot. The maximum channel sample grade is 759 g/t, with metal accumulation values ranging from 0.18 to 773 m.g/t. The system
comprises two en-echelon reefs (footwall and hangingwall veins) around a central, low-grade massive quartz core. (Source: Reef Mining NL).
180
dilution is included in the measured resource. Underground

development has increased the resource, in this case double the
initial estimate, as mining and exploratory development is
concurrent. Resource and geological modelling is undertaken
using Gemcom software.
Reconciliation
The Nick OTime Shoot is accessed via a 2 km decline that

allows exploratory development of the whole of the Poverty
Reef. Ore has been mined using a combination of mechanised
methods (3.5 - 7.0 m widths) and conventional hand-held (airleg)
methods (1.2 - 2.0 m widths). Hand-held methods have been
used where geological parameters have allowed selective mining
of high-grade laminated reefs, leaving the massive quartz core as
a pillar. In both cases flat-back cut and fill mining has been the
predominant method, using either mullock or tailings as backfill.
Ore is trucked to surface and then transported to an off-site
mill run by Reef Mining on a dry-hire basis. A 98 per cent
recovery is achieved; 80 per cent of gold is recovered in the
gravity circuit and the remainder in a CIP circuit.
The face grade is determined by weighting according to the

proportion of the face occupied by a particular geological/grade
domain. The grade of a block, representing one cut (3.5 m wide,
4.0 m high, 2.7 m long), is simply determined by averaging the
face grades either end of the block. A weighted average of
tonnes/grade for each block along the 3.5 m stoping lift is then
made.
This simplistic classical mining block-modelling method,
using uncut assays, usually overstates the grade by about 15 per
cent in relation to mill grade. Experience has shown that by
applying a 100 g/t top cut to the calculated block grades (not the
individual unit grade assays) and applying the appropriate
dilution factor for the particular stoping level, the correlation
with mill grade is good. Other grade interpolation methods have
been considered, with work in progress.
Overall the history of resource evaluation compared with
actual mining suggests a general tendency for an increase in
actual tonnes relative to the estimate, and a general consistency
in grade.
Grade control
New Bendigo Project, Bendigo Goldfield, Victoria
Mining and milling
Grade control is essentially geological. The majority of the gold

inventory is contained within the generally clearly defined
laminated reefs. No gold occurs within the altered metasediment
wall rocks. However, the grade boundaries can be difficult to
pick around the massive quartz core or where high-grade
laminated reefs are associated with earlier massive quartz spurs.
Chip-channel samples are taken across the face every cut. A cut
is 2.7 m advance in the mechanised operations and 1.8 m in the
airleg stopes. Mechanised mining faces are 3 - 4 m wide and
usually four 5 - 10 kg samples are taken. The narrower airleg
stope faces are 1.2 - 1.7 m wide and three 7-kg samples are
taken. Each sample is crushed, split to 3 kg, pulverised to -75 m
then subjected to a BLEG assay (solvent extraction, AAS finish).
Alternatively, if data on size distribution of gold is required, a
1 kg SFA is performed.
Grade control is very much visual and experienced based,
supported by a database of retrospective face assays of the
various vein types and styles of mineralisation.
Studies undertaken in 1989 involved grab samples from rail
trucks. Despite the apparent random nature of this method they
often provided a grade nearest to reality (Table 3). Two duplicate
1 kg samples were taken out of each truck at the shaft station. All
samples were bulked together to make a larger 10 - 15 kg
sample, which was pulverised and split down to 1 kg.
TABLE 3
Comparison between face, truck (grab) and mill grades for the
Poverty Shaft Shoot, Tarnagulla Gold Project, Victoria, Australia
(Source: Reef Mining NL).
Lower Sublevel (49 m drive)
Face
Average Grade (g/t)

4.3
Truck
5.8
Mill (350 tonnes)
7.1
Upper Sublevel (18 m drive)

Face
3.4
Truck
7.6
Mill (250 tonnes)
7.0
The Bendigo Goldfield is centred on the urban area of Bendigo

in Central Victoria, some 130 km northwest of Melbourne. Gold
was discovered in the area in 1851 and production continued
until 1954. The Goldfield is the second largest in terms of total
gold produced (~529 tonnes) in Australia after Kalgoorlies
Golden Mile. Recent exploration started in 1978 until 1993 when
Western Mining Corporation Ltd held title over the Goldfield
(Sharpe, 1988). They undertook extensive historical research
together with RC and diamond drilling, but did not define any
resources (Johansen, 1998). In 1993 Western Mining
relinquished the title which was taken over by Bendigo Mining
NL. Since then Bendigo Mining have undertaken an extensive
re-evaluation of the area including detailed 3D geological
modelling (using Gemcom), RC and diamond drilling,
dewatering and accessing historic workings and bulk sampling.
The company estimate that in excess of 120 million tonnes of
quartz existed in the historically mined areas, of which
approximately 40 million tonnes was mined. In November 1997
Bendigo Mining raised $A35 million to fund the sinking of a
decline to investigate targets on the Deborah and Sheepshead
Lines of Reef. The project will also involve 1000 m of
exploration driving, 5000 m of underground fan drilling and
3500 tonnes of bulk sampling (Syddell, 1999).

The Bendigo Goldfield is a classic slate-belt style of
mineralisation with sediment hosted auriferous quartz reefs
closely associated with anticline axes and reverse faulting. A
geological model has been developed to explain the development
and distribution of reefs within the Goldfield (Figure 8).
Structurally the region is comprised of compression-ruptured
chevron folds. The new model envisages that, as compression
progressed, stress was released through reverse faulting of the
anticlines at regular vertical intervals. The structures so formed
became the focus for siliceous mineralisation, resulting in the
development of various types of quartz reefs on, or in close
proximity to the anticlinal axes at vertical intervals of 150 250 m. These structural zones, which have continuous
along-strike extents of many kilometres, are termed ribbons and
the regular repetition of the ribbons at depth was, historically, a
well-known feature of the Goldfield (Figure 8). With later
re-fracturing of the ribbons and an influx of gold-bearing fluids,
some of the reefs and other quartz-filled structures became hosts
for subsequent gold mineralisation. The famous saddle reef
structures of Bendigo actually only accounted for ten per cent of
historic production.
181
FIG 8 - New Bendigo Project, Victoria, Australia. Schematic 3D block diagram showing potential mineralised quartz-reef ribbons at depth on the
Sheepshead, Deborah and Garden Gulley Lines of Reef. (Source: Bendigo Mining NL).
Grade
The gold within the quartz reefs is typically both coarse-grained
and erratically distributed. All the Bendigo reefs contain
low-grades from 0.3 to 1.5 g/t with higher grade areas above
8 g/t; high-grade ore shoots can contain as much as 95 g/t
(Laidlaw, 1993). Historic production grades for the entire field
were 10 - 15 g/t. Current estimates are that historic operators
extracted approximately one-third of the reefs accessed, with the
remaining two-thirds deemed to be too low-grade (their cut-off
was around 6 - 8 g/t). Investigation of historical stoping patterns
and bulk sampling data indicate that the grade, though erratic in a
local sense, tends to be more gradational on a larger scale.
High-grade oreshoots may relate to late, small-scale
displacement, slightly oblique structures and as a result stoping
patterns may occur as 20 - 100 m wide zones pitching gently
across reef strike. Where sufficient information is available, a
high-grade core to the stopes can be discernible. Moving away
from the stoped regions, historic bulk sampling values drop off in
a fairly regular fashion. As a result of the nugget effect, small
samples such as drill core will not detect this broader pattern.
Resource evaluation
There are currently four separate areas of deep remnant
mineralisation with identified resources; Deborah, Sheepshead
(S2), Williams 40 Foot Spurs and New Chum (NC6). The
methodology for estimating resources varies slightly depending
on the data available.
Deborah tonnage estimation. The Deborah workings were
dewatered to approximately 400 metres below surface allowing
detailed geological mapping of the accessible portions of the
historic workings. This mapping, in conjunction with
reconstruction of historic workings and geology from historical
data and three diamond drill holes, enabled the construction of
182
detailed, computerised 3D models of the quartz reefs, associated

structures and historic workings. Tonnages for individual reefs
were determined by applying an average specific gravity
(calculated from samples collected underground) to their
volumes from the models, after removing the volume of historic
stoping and development. Tonnage estimates are considered to
be reasonably accurate.
Deborah grade estimation. A total of two-hundred and
twenty-four 25 kg chip samples were collected from the
accessible reefs. Each sample was crushed to -10 mm, riffle split
into two equal subsamples and each subsample pulverised to -75
m. Three kilograms were split from each subsample and
assayed for gold by BLEG (3 kg). Drill core from two diamond
drill holes were assayed by 3 kg BLEG on the whole core, with
the remaining diamond drill hole assayed by fire assay on half
core. In addition three historic bulk samples each of
approximately 100 tonnes were investigated.
Comparison studies between the subsamples of the rock chips
and also the bulk sample results were completed. This confirmed
the high nugget effect seen elsewhere at Bendigo. It is possible to
generate a range of grade estimates for the reef depending on the
statistical treatment of the data and the inclusion, exclusion, or
weighting of the various sample types. The accessed portion of
the reef was divided into 30 m blocks and assay data within these
blocks length weighted. This effectively de-clustered the data.
The bulk sample data was included in the calculations as
pseudo-chip samples over the bulk sample width. This provided
the best estimate of the grade of the reef and fell within a range
of possible grades depending upon the estimation methodology
employed.
Sheepshead (S2) tonnage estimate. The historic workings and
geology of the southern Sheepshead Line of Reef (including the
S2 ribbon located 500 metres below surface) were reconstructed
from the available maps, plans and managers reports. The
computerised 3D model included all recorded mine development,

stoping, faults and reefs. An estimate of the remnant tonnage
remaining in the Big Blue Neck Reef was determined by
applying the SG of quartz to the volume of the reef remaining
within the model after removing areas of historic development
and stoping. Two Western Mining diamond drill holes provided
confirmation of reef dimensions. Tonnage estimates are
considered to be reasonably accurate.
Sheepshead (S2) grade estimate. Research located a total of 17
historic bulk samples within the Big Blue Reef. A range of grade
estimates could be calculated from these data by varying the
statistical treatment. For instance, simple averaging or weighted
averaging of the data and by dividing the reef into geological
domains and generating grade and tonnages for each domain.
After careful consideration the reef was divided into three
geological domains and a grade determined for each domain.
This provided a best estimate grade within a range of possible
grades depending on the statistical treatment chosen. Assay
results from the Western Mining drilling were not included in the
calculations due to their small sample size and poor repeatability
on re-assay, though they did confirm the tenor of mineralisation.
The new geological model has been used to support the
resource potential estimate of ten million ounces for the New
Bendigo Goldfield. This estimate, which covers only part of the
Goldfield, does not constitute reserves or resources as defined by

the JORC Code, and therefore does not form part of Bendigo
Minings published resource base. It is, however, considered to
represent a soundly based and technically supported,
conservative exploration target.
Proposed resource evaluation methodology

Based on the experience gained from exploring and defining
resources and reserves within the Bendigo Goldfield, an
exploration strategy has been developed to overcome the
expected sampling difficulties in the New Bendigo allowing the
estimation of resources and reserves.
Tonnage estimation. Initial diamond drilling utilising orientated
core on 120 m spaced sections will be undertaken from the
current decline, located well above the target reefs (Figure 9). As
reefs display good continuity along strike, this will be sufficient
drilling for an initial tonnage estimate. Once the decline reaches
the target area, fan drilling on 60 m sections will be completed to
refine the tonnage estimates. This level of drilling is expected to
produce reasonably accurate tonnage estimates suitable for
resource definition. As underground development continues, fan
drilling will be closed down to 30 m sections for mine planning
enabling further refinement of the tonnage estimations.
FIG 9 - New Bendigo Project, Victoria, Australia. Schematic 3D block diagram showing the potentially mineralised quartz-reef D3
(Deborah Line of Reef). Orientated diamond fan diamond drill holes on 120 m spacings will be collared on the Swan Decline to provide an
initial tonnage estimate. (Source: Bendigo Mining NL).
183
Grade estimation. It is anticipated that a reasonable estimate of

grade and grade distribution within the reefs will be possible by
using a combination of close-spaced drilling and bulk sampling
information.
All mineralised drill core will be halved and assayed for gold
and a suite of trace elements. It is not expected that reef grade
can be determined directly from this drilling, however it is
anticipated that the drilling will provide:
an indication of the grade of mineralisation;
an indication of the grade distribution;
a correlation between gold grade and indicators such as
gangue mineralogy and textures; and
a correlation between gold grade and a suite of trace
elements
A systematic program of bulk sampling will be undertaken to
provide grade data. The location of samples will be based on the
interpretation of drilling results, mindful of both structural
setting and assay data. It is envisaged that every second or third
development round in ore will be taken as a bulk sample.
Samples will be an entire individual round of between 60 - 100
tonnes depending upon development size. They will be crushed
to -8 mm and passed through a splitter where 1 to 20 per cent (up
to 100 per cent possible if required) will be split off to a small
pulveriser (hammer mill) and gravity circuit (Knelson
Concentrator). The bulk sampling will provide:
a reasonably accurate estimate of grade; and
a correlation between bulk sample grade and drill grade.
Statistical treatment of grade data sets derived from drilling
and bulk sampling will be undertaken. Bulk sampling is
considered crucial to grade estimation within the New Bendigo.
The current bulk sampling plant is being modified to enable the
minimum sample size to be determined for both crushed and
broken ore for a variety of grade ranges and ore types. Eventually
it is anticipated the sampling plant will be relocated underground
to facilitate sampling and grade estimation.
The classification and reporting of reserves and resources in a

nuggety reef is not straightforward. By their nature, it is
extremely difficult to estimate measured resources (and proven
reserves), which has clear negative implications for project
financing (Spark, 1990; Battersby, 1993). Inferred and indicated
resources will require differing levels of diamond drill
intersection, underground development and bulk sampling.
Tonnage can be determined to a reasonable level of confidence
but not grade, it is therefore difficult to produce bankable
documents. When presenting resource statements, it is
recommended that a best estimate grade be defined within a
grade range. Both grade and tonnage figures will be rounded to
reflect the level of precision of the estimations. For instance
grade may be rounded to one decimal place within a grade range
rounded to the nearest g/t. Tonnage may be rounded to the
nearest 10 000 tonnes. This approach better reflects the current
situation than reporting a single grade, and as a consequence
better informs investors and the stockmarket.
ACKNOWLEDGEMENTS
This on-going work benefits from the support of Cardiff
University, University of Greenwich, SRK Consulting, Reef
Mining NL, Bendigo Mining NL, Welsh Gold PLC and Reef
Mining Services. Discussions with numerous company personnel
are noted, in particular from SMC Resources Ltd, Clogau Gold
Mines Ltd, Elenith Mining Co Ltd, Central Norseman Gold
Corporation Ltd and OCX. Resources. Discussions with Roland
Phelps (Welsh Gold PLC), Dr Steve Henley (Resources
Computing International Ltd), Dr Bon Royle (Consultant), David
Pelham (Consultant) and Ian Gilmore (AMMTEC Ltd) are
acknowledged. Simon Camm (Camborne School of Mines),
Roland Phelps (Welsh Gold PLC) and AusIMM referees are
thanked for their critical review of this contribution. Presentation
of this paper at The AusIMM Fourth International Mining
Geology Conference was supported by a Royal Society Travel
Grant and the Cardiff University Earth Sciences Staff
Development Fund.
Proposed resource/reserve classification and reporting

practice
It is anticipated that resources will be restricted to the inferred
and indicated categories due to the difficulty in estimating grade
(Johansen, 1997; Johansen and Stephenson, 1999). Initially bulk
sampling will be required to define resources. It is hoped that
following the processing of many bulk samples, a correlation will
be established between bulk sampling, trace element
geochemistry, gangue mineralogy, reef texture and drilling
results to enable inferred resources to be defined through drilling.
Indicated resources will most likely always require bulk
sampling. Bulk sampling will always be required to estimate
reserves. It is likely that only probable reserves will be reported,
reflecting the degree of uncertainty with determining grade.
Reporting of probable reserves will be as a best estimate within a
grade range with the proviso that all of the nominated grade
range falls within the economic parameters.
CONCLUSIONS
Nuggety coarse-gold reefs are arguably one of the most
challenging types of mineralisation to evaluate and exploit. This
is generally because of their inherent grade complexities. The
grade characteristics of the deposits make sampling difficult.
Drilling is an excellent indicator of position and structure, but
grade must be evaluated from underground development through
face sampling, etc and bulk sampling. Geological understanding
is the key to improved confidence in the operation. It is vital to
get the geology right before doing the sums this cannot be
over-emphasised. Detailed geological interpretation supported by
3D computer modelling is critical.
184
REFERENCES
Annels, A E, 1991. Mineral Deposit Evaluation, A Practical Approach,
436 p (Chapman and Hall: London).
Battersby, J, 1993. The Inglewood Goldfield, in Proceedings Narrow
Vein Mining Seminar, pp 107-108 (The Australasian Institute of
Cuffley, B W, Krokowski de Vickerod, J, Evans, T and Fraser, R. 1998. A
new structural model for fault-hosted gold mineralization: an
example from the Nick OTime oreshoot, Poverty Reef, Tarnagulla,
in VICMIN 98: The Second GPIC Conference on Developments in
Victorian Geology and Mineralisation, pp 53-63 (Australian Institute
of Geoscientists: Perth).
Dominy, S C and Phelps, R F G, 1996. Gwynfynydds gold veins:
geology plays an important role in the mining of complex narrow
veins, Mining Mag, 175(6):325-333.
Dominy, S C and Phelps, R F G, 1997. Small-scale vein-gold exploitation
in Gwynfynydd mine, Wales, UK, in Mine Planning and Equipment
Selection 1997, pp 133-139 (A A Balkema: Rotterdam).
Dominy, S C, Phelps, R F G and Camm, G S, 1996. Geological controls
on gold grade distribution in the Chidlaw Link Zone, Gwynfynydd
mine, Dolgellau, North Wales, UK, Trans Inst Min Metall (Section
B), 105:151-158.
Dominy, S C, Annels, A E, Camm, G S, Wheeler, P and Barr, S P, 1997.
Geology in the resource and reserve estimation of narrow vein
deposits, Explor Min Geol, 6(4):317-333.
Dominy, S C, Annels, A E, Camm, G S, Cuffley, B W and Hodkinson, I
P, 1999. Resource evaluation of narrow gold-bearing veins: problems
and methods of grade estimation, Trans Inst Min Metall (Section A),
108:52-70.
Dominy, S C and Platten, I M, 1999. Criteria for the recognition of a
pre-cleavage age for gold-bearing quartz veins in weakly cleaved
terrains, Trans Inst Min Metall (Section B), 108:73-76.
Dominy, S C, Cuffley, B W and Johansen, G F, 2000. Planning nuggety

vein-gold operations: the role of underground exploration, bulk
sampling and trial mining, in Mine Planning and Equipment
Selection 2000, in press (A A Balkema: Rotterdam).
Johansen, G F, 1997. Sampling, assaying and reporting in a coarse gold
environment, paper presented to the Assaying and Reporting
Standards Conference, Singapore, 10-11 August.
Johansen, G F, 1998. The New Bendigo Goldfield, in VICMIN 98: The
Second GPIC Conference on Developments in Victorian Geology
and Mineralisation, pp 47-51 (Australian Institute of Geoscientists:
Perth).
Johansen, G F and Stephenson, P R, 1999. Estimating and reporting
reserves and resources in the Bendigo Goldfield, paper presented to
the 7th Victorian Resource Conference, Melbourne, November.
JORC, 1999. Australasian Code for Reporting of Identified Mineral
Resources and Ore Reserves. Report of the Joint Committee of The
Australasian Institute of Mining and Metallurgy, Australasian
Institute of Geoscientists and Minerals Council of Australia (JORC).
16 p.
Laidlaw, N, 1993. Some considerations in the mining of saddles, spurs

and backs on the Bendigo Goldfield, in Proceedings Narrow Vein
Mining Seminar, pp 25-34 (The Australasian Institute of Mining and
Platten, I M and Dominy, S C, 1999. Re-evaluation of quartz vein history
in the Dolgellau Gold-Belt, North Wales, UK, Geol Journ,
34(4):369-391.
Sharpe, E, 1988. Bendigo gold project, in Bicentennial Gold 1988
Excursion Handbook: Central Victorian Gold Deposits, pp 32-38
(University of Western Australia: Perth).
Sinclair, A J and Valle, M, 1994. Reviewing continuity: an essential
element of quality control for deposit and reserve estimation, Explor
Min Geol, 2(2):95-108.
Spark, R F, 1990. The Ballarat East project, in Ore Reserve Estimates:
The Impact on Miners and Financiers, pp 57-62 (The Australasian
Syddell, M, 1999. Bendigo - poised for its moment of truth, Aus Mining,
May 1999:8-12.
Valle, M, 1992. Guide to the Evaluation of Gold Deposits. Special
Volume 45, 300 p (Canadian Institute of Mining and Metallurgy).
185
Multiple Indicator Kriging Is it Suited to My Deposit?

J Vann1, D Guibal2 and M Harley3
ABSTRACT
1.
Many sample value distributions are highly skewed and present

variograms with high relative nugget effects and/or short-range
structures. In consequence, significant deskewing of the histogram and
variance reduction occurs when transforming between sample and block
support, where blocks are of significantly larger volume than samples.
When making estimates in the mining industry we usually wish to
estimate on the basis of block support rather than sample support. The
selective mining unit (SMU) refers to the minimum support upon which
ore/waste allocation decisions can realistically and practically be made.
The SMU is usually significantly smaller than the sampling grid
dimensions, in particular at exploration/feasibility stages. In a non-linear
estimate, for each large block (by convention called a panel) we
estimate the proportion of SMU-sized parcels above a cut-off grade
threshold. A series of proportions above cut-off defines the SMU
distribution. Use of such non-linear estimates reduces distortion of
grade-tonnage curves and provides a better basis for mine planning and
economic decision-making. At present, Multiple Indicator Kriging (MIK)
is the most widely used technique for performing estimation of
recoverable resources. MIK is not the only approach available and it has
some potentially serious limitations. The MIK technique is described and
a number of theoretical and practical implementation issues are
examined. In particular: MIKs assumptions about spatial distribution of
grade; the behaviour of indicator variograms; application of change of
support models; order relationship problems; and modelling the tail of
the distribution. Some simple tests are described which allow us to
determine the suitability of different estimators for a given deposit and
criteria to identify a preferred estimator are also examined. Uniform
Conditioning (UC), an alternative estimator which may be used instead
of MIK, is described.
2.
a specified data/block configuration (in other words, the

geometry of the problem).
Criteria 2 above involves knowing the block dimensions and
geometry, the location and support of the informing samples, and
the search (or kriging neighbourhood) employed. Minimum
estimation variance simply means that the spread of estimation
errors is minimised by OK. Given an appropriate variogram
model, OK will outperform IDW because the estimate will be
smoothed in a manner conditioned by the spatial variability of
the data (known from the variogram).
Non-linear geostatistical estimators contrast with linear
estimators in that they allocate weights to samples that are
functions of the grades themselves, ie not solely dependent on
the location of data.
Linear estimators have a number of inherent limitations, which
are discussed below.
Limitations of linear estimators

Linear estimates have well known limitations:
1.
INTRODUCTION
Since the early-1980s, Multiple Indicator Kriging (MIK) has
become the most frequently applied non-linear geostatistical
estimation method. It is especially widely applied in Australia by
both consulting companies and in-house resource estimation
teams. This paper presents the underlying motivations and
concepts of non-linear estimation and outlines some of the
potential problems with MIK.
CONCEPTS
Linear versus non-linear
A linear interpolator has the following important property: the
weights assigned to each of the N sample locations inside the
estimation neighbourhood are independent of the specific data
values at these locations. This is true for linear geostatistical
estimators, eg Ordinary Kriging (OK) and Inverse Distance
Weighting (IDW).
OK is a more sophisticated linear interpolator than IDW and
was proposed by Matheron (1962, 1963a, 1963b). Note that OK
is, in essence, simply a regression approach to obtaining the
best weights for a linear estimator. OKs advantage over IDW
as a linear estimator is that it ensures minimum estimation
variance given:
1.
MAusIMM, Technical Director, SRK Consulting, PO Box 943, West

Perth WA 6872.
2.
FAusIMM, MMICA, Technical Director, SRK Consulting, PO Box

943, West Perth WA 6872.
3.
Principal Geostatistician, SRK Consulting, PO Box 55291,

Northlands 2116, South Africa.
a specified model for spatial variability (eg variogram); and
We may want to estimate the distribution of grades within

a block rather than simply estimate the average (or
expected) grade. Linear estimators cannot do this. An
example is the estimation of recoverable resources, ie the
proportion of smaller selective mining unit (or SMU)
blocks within a larger block (often called a panel) that
exceed some cut-off grade. More generally, linear
estimation of non-linear functions of the grades does not
work well, if at all.
2.
When dealing with a strongly skewed sample distribution,

eg many gold, tin and uranium deposits, estimating the
mean by a linear estimator (for example by OK) is risky. In
effect, as the weights do not depend on the sample grades,
the presence of extreme values making any linear estimate
very unstable. In this situation, knowledge of the
distribution of grade within a block may improve our
estimate of the mean grade of that block. This usually
involves making assumptions about the distribution (for
example, what is the shape of the tail of the distribution?)
even in situations where we are ostensibly distribution
free (for example using MIK).
Note: In many cases where we require an estimate of
recoverable resources (ie need to account for future selectivity at
mining stage) we are also dealing with highly skewed sample
distributions.
From a mathematical viewpoint, the best possible estimate of a
function of the grade is the conditional expectation, which is a
complex, non-linear function of the known grades, in general.
Only in the very particular case where the distribution of the
grades is multi-gaussian is the conditional expectation linear
(simple kriging). Geostatistical non-linear estimation is an
attempt to estimate the conditional expectation, and further the
conditional distribution of grade within a panel by using a
broader class of estimators than the traditional linear estimates.
In such a case we wish to estimate the mean grade (expectation)
at some location under the condition that we know certain nearby
sample values (conditional expectation). Interested readers can
get a good overview of the mathematical background to
non-linear geostatistics from Rivoirard (1994).
187
J VANN, D GUIBAL and M HARLEY
In summary, we define non-linear geostatistical estimators as

those estimators that use non-linear functions of the data to
obtain (or approximate) the conditional expectation. Once we
know (or approximate) this distribution, we can predict grade
tonnage relationships (eg how much of this block is above a
cut-off Zc? At what grade?). In a non-linear estimation we thus
estimate, for each large block (by convention usually called a
panel) the proportion of SMU-sized parcels above a cut-off
grade. A series of proportions above cut-off defines the SMU
distribution.
Available methods
There are many methods now available to make local (panel by
panel) estimates of such distributions, most of which are
summarised and discussed by Vann and Guibal (in press). Those
in current use in the mining industry form a more restricted list,
and include:
1.
Disjunctive Kriging DK (Matheron, 1976; Armstrong

and Matheron, 1986a, 1986b);
2.
Indicator Kriging IK (Journel, 1982, 1988) and variants

(Multiple Indicator Kriging; Median Indicator Kriging,
etc);
3.
Probability Kriging PK (Verly and Sullivan, 1985);
4.
Lognormal Kriging LK (Dowd, 1982); and its

generalisation
to
non-lognormal
distributions,
Multigaussian Kriging MK (Verly, 1983);
5.
Uniform Conditioning UC (Rivoirard, 1994); and
6.
Residual Indicator Kriging RIK (Rivoirard, 1989).
Even estimating very small blocks directly by a non-linear

estimator may be incorrect and risky. When using non-linear
estimation for recoverable resources estimation in a mine, the
panels should generally have dimensions approximately equal to
the drill spacing, and only in rare circumstances (ie strong
continuity) can significantly smaller panels be specified.
Non-linear estimation provides a solution to the small block
problem. We cannot precisely estimate small (SMU-sized)
blocks by direct linear estimation. However, we can estimate the
proportion of SMU-sized blocks above a specified cut-off, within
a panel. Thus, the concept of change of support is critical in most
practical applications of non-linear estimation.
Definition of support effect

Support is a term used in geostatistics to denote the volume
upon which average values may be computed or measured.
Grades of mineralisation measured on drill hole samples can
be much richer or poorer than grades measured on larger
supports, say SMU blocks. The grades on smaller supports are
said to be more dispersed (variable) than grades on larger
supports.
Note that the overall (or global) mean grades on different
supports (applying no cut-off) are the same. However, the
variance of smaller supports is larger. Support effect is this
influence of the support on the distribution of grades. Figure 1
illustrates the impact of support upon the histogram. Note on
Figure 1 that, if we apply a cut-off above the mean grade, the
recovered portion predicted from samples (v) over estimates the
true recovered proportion on blocks (V). The proportion above
cut-off is shown on the figure by shading.
The small block problem

Why not simply estimate small block (SMU) grades directly by a
linear method? This is the small block question, familiar to
every geostatistician. By small blocks we mean blocks that have
small dimensions relative to the sample grid. For example, 5 m
5 m 5 m blocks in a deposit drilled on a 30 m 30 m grid.
Perhaps the question most frequently asked of professional
geostatisticians is: what is a safe block size? This is a question
that simply cannot be answered without knowledge of the spatial
grade variability (eg the variogram). However, we make a few of
generalisations, below.
In (unusual) situations where the nugget effect is very low, the
variogram has no pronounced short-range structure, and a long
range that is significantly longer that the drilling mesh, small
blocks may possibly be reasonably estimated using linear
methods.
In general, estimating blocks that are considerably smaller
than the average drilling grid (say appreciably less than half the
size) is potentially very risky. In very high nugget situations
(epithermal and shear hosted gold, for example), even blocks
with dimensions approximating the drill spacing may still be
highly risky.
There is a plethora of warnings in the geostatististical literature
against estimation of small blocks by linear methods (Armstrong
and Champigny, 1989; David, 1972; David, 1988; Journel, 1980,
1983, 1985; Journel and Huijbregts, 1978; Krige, 1994, 1996a,
1996b, 1997; Matheron, 1976, 1984; Ravenscroft and
Armstrong, 1990; Rivoirard, 1994; Royle, 1979). The authors
strongly reiterate this warning.
The commonplace practice of estimating blocks that are far
too small is symptomatic of misunderstanding of basic
geostatistics. In the authors experience, many professionals
estimating resources, have not read, or still dont understand, the
earliest exposition of this problem by Krige (1951). It is
interesting that Krige has felt the need to re-issue his warnings in
recent years (Krige, 1994, 1996a, 1996b).
188
FIG 1 - Impact of support on the histogram.
The necessity for change of support

Before committing the capital required to develop a selective
mining operation, an economic decision is made based on an
estimate in which (usually) the only grade data are from
exploration drill holes. Because mining does not proceed with a
selection unit of comparable size to the samples, the difference in
support between the samples and the proposed SMU must be
accounted for in any estimate to obtain achievable results. When
there is a large nugget effect, or (equivalently) an important
short-range structure, then the impact of change of support will
be pronounced.
MULTIPLE INDICATOR KRIGING IS IT SUITED TO MY DEPOSIT?
In practice, we see that the histogram of samples usually has a

much longer tail than the histogram of mining blocks. Note that
simplistic variance corrections, for example affine corrections, do
not reflect the fact that, in addition to variance reduction, change
of support also involves symmetrisation of the histogram. This is
especially important in cases where the histogram of samples is
highly skewed (eg most precious metal deposits, uranium
deposits, impurity variables for iron, bauxite and manganese
deposits, etc).
Recoverable resources
Recoverable resources are the portion of in situ resources that
are selected as ore during mining. The concept of recoverable
resources involves both technical considerations, such as cut-off
grade, SMU definition, machinery selection, etc, and also
economic/financial considerations such as site operating costs,
commodity prices outlook, etc.
The end result of an estimation of recoverable resources is, for
each panel, a tonnage and grade of material above cut-off within
that panel, assuming a specific SMU. The tonnage is simply the
proportion of the panel above cut-off multiplied by the panel SG.
The product of recovered grade and recovered tonnage yields
recovered metal.
The end result of a non-linear estimate tells us what percentage
of the panel is likely to be mined above cut-off, but does not tell
us which part of the panel will be above/below cut-off. In other
words, grade control drilling during production will usually be
required to define dig lines. The recoverable resource estimate
will, however, be a better basis upon which to estimate the cash
value of each panel when performing economic optimisation or
mine design (for example pit optimisation).
probabilities (the probability that the grade is above the

specified indicator); or
proportions (the proportion of the block above the specified

cut-off on data support).
Multiple indicator kriging

Multiple indicator kriging (MIK) involves kriging of indicators at
more than one cut-off (see various publications by Journel,
Hohn, 1988 and Cressie, 1993). MIK is an approach to
recoverable resources estimation which is robust to extreme
values and is practical to implement. It is fair to say that it is
currently the method most widely used in the mining industry for
estimating recoverable resources.
Theoretically, MIK gives a worse approximation of the
conditional expectation compared to disjunctive kriging (DK),
which can be shown to approximate a full cokriging of the
indicators at all cut-offs. However, MIK does not have the strict
stationarity restriction of DK (see Vann and Guibal, in press).
The major practical difficulties with MIK are summarised
below. Be aware that these are not merely theoretical points their impact on the reliability of results (in each case) can be
significant.
Adequate representation of the histogram

In MIK we discretise the histogram of sample values into n
classes each bounded by an upper- and lower-cut-off boundary
value. The number of classes and the values of their bounding
values is decided by the practitioner. Particular problems occur in
three regards:
1.
The values are often selected to be evenly distributed with

respect to the percentiles of the data distribution. For
example, boundaries are set at the deciles of the
distribution. The problem here is that we may end up with a
discretisation that does not adequately resolve the more
critical parts of the histogram. Alternative approaches
include additional discretisation of these critical parts (the
tail in particular) or discretisation such that equal metal is
contained in each defined class. This is certainly preferred.
2.
The tail of the distribution generally has very few samples

(for example in gold deposits). This means that a model
must be defined to give the shape of the histogram in the
tail. Choices between different models may materially
affect the estimates at higher cut-offs, but there is little
objectivity in choosing between models.
3.
For the grade (as opposed to the tonnage) estimation, it is

necessary to have estimates of the average grades of the
classes. The common practice is to choose as mean grade of
a class the average value of the samples belonging to that
class. A possible exception is the last class where often a
different estimate is chosen: median or trimmed mean for
instance. Now the consequences of this choice which is
often arbitrary can be very significant and strongly impact
the estimation of the richest zones of the orebody.
INDICATOR GEOSTATISTICS
Indicators
The use of indicators is a strategy for performing variogram
analysis in order to characterise the spatial distribution of grades
at different cut-offs. The transformed variable has a binary
distribution (everything is now a 1 or a 0), and so by definition
does not contain extreme values. Furthermore, the indicator
variogram for a specified cut-off zc is physically interpretable as
characterising the spatial continuity of samples with grades
exceeding zc. Indicator transformations may thus be conceptually
viewed as a digital contouring of the data. They give very
valuable information on the geometry of the mineralisation.
A good survey of the indicator approach can be found in the
papers of Journel (eg 1983, 1987, 1989). A simple, easy to
follow case study is provided by Hohn (1988).
An indicator variable I(x, zc) is defined, at a location x, for the
cut-off zc as the binary function that assumes the value 0 or 1
under the following conditions:
I(x, zc) = 0
if Z(x) zc
I(x, zc) = 1
if Z(x) > zc
After transforming the data, indicator variograms can be

calculated easily by any program written to calculate an
experimental variogram. An indicator variogram is simply the
variogram of the indicator.
Variography
The variography of indicators presents several difficulties:
1.
Destructuration of high-grades - the variograms for higher

cut-offs often tend towards pure nugget effect. With such
variograms, at the estimation stage, large search
neighbourhoods should be used, in order to avoid
conditional biases. The impact of this is to distribute metal
in the upper class(es) very smoothly.
2.
The number of variograms required may be large. In

practice this leads to three possible solutions:
Indicator kriging
Indicator kriging (or IK) is kriging (usually OK) of indicator
transformed values using the appropriate indicator variogram as
an input. An IK estimate (ie kriging of a single indicator) must
always lie in the interval [0,1].
The resulting estimate can be interpreted as either:
189
the job takes a lot longer (disliked intensely by

management!);
b. the job is rushed and insufficient time is taken with
variogram modelling; or
c.
automated variogram modelling routines are
employed.
The last two options (b) and (c) can lead to compromised
estimation of resources, because the kriging results may be
sensitive to the variography. In particular (as we will see when
we discuss order relation problems) we must ensure that the
variograms at various cut-offs are theoretically compatible, or
results can be fatally compromised.
dangerous and this is the reason why some authors

(Deutsch and Lewis, 1992) propose to use the same
elementary structures for all cut-offs, with partial sills
slowly changing.
a.
Order relation problems
3.
When increasing cut-offs, at least one data in the search

neighbourhood will change from 1 to 0. Clearly, this is not
possible if the cut-offs are defined once and for all. Thus,
the idea is to define cut-offs at a local level (for each search
neighbourhood), depending on the local data. The same
number and types of structures are used for all cut-offs, but
the partial sills are linearly interpolated between cut-offs
(Deutsch and Lewis, 1992). Note that this procedure is
fairly complex and tends to become purely theoretical, the
link with the experimental variograms getting very tenuous!
Because indicator variogram models may be inconsistent from

one cut-off to another we may estimate more recovered metal
above a cut-off zc2 than for a lower cut-off zc1, where zc1 < zc2,
which is clearly impossible in reality! If results are inconsistent
from one cut-off to another, in this manner, we say we have an
order relation problem.
While there has been emphasis in published literature on the
triviality of order relation problems and the ease of their
correction, the authors have observed quite severe difficulties in
this regard with MIK, especially when anisotropies show
significant changes and search neighbourhoods are modified
from one cut-off to another. The theoretical solution is to account
for the cross-correlation of indicators at different cut-offs in the
estimation by cokriging of indicators (Vann and Guibal, in
press), but this is completely impractical from a computational
and time point of view.
Most commercial and public domain MIK programs correct
order relation problems by smoothing the grade-tonnage vector
of a panel if they violate order relations. Too often, the correction
algorithms are not documented and/or not reported. The authors
find this disturbing and observe that such corrections are entirely
arbitrary. From an auditing viewpoint, it is very important that
the number and magnitude of the corrections are reported: if they
are numerous and significant, they may have a material impact
on the stated Resources.
It is important to note that order relation violations are
problems of methodology, and can be reduced or eliminated by
modifying the methodology. However, mechanical fixes (such as
smoothing misbehaving estimates!) may simply disguise the
inherent distortion of the grade-tonnage relationship estimated by
MIK.
Order relationship problems can be reduced by ensuring that:
Change of support in MIK
1.
Aside: median indicator kriging
2.
190
The variogram models at various cut-offs are theoretically

consistent. The most obvious case is when all the indicator
variograms are proportional, allowing the use of a single
indicator variogram model (this is Median Indicator
Kriging or Median IK - see below). Unfortunately, it is a
very rare occurrence that the indicator variogram at all
cut-offs can be reasonably approximated by a single
(proportional) model. In the general case, because the
indicator variables at various cut-offs are correlated, there
are also relationships between the corresponding
variograms, which need to be respected.
The search volumes defined during kriging at various
cut-offs are the same (identical shape/volume, ie the same
informing data are used across all cut-offs). Unfortunately,
this is only a good idea when the variograms for all
indicators have roughly the same shape (or are
proportional) which is not the usual case for mineral
deposits. Major changes in anisotropies are in particular
Again this is the assumption made in Median Indicator Kriging.
Change of support is not inherent in the MIK method. We apply

change of support in MIK as a post-processing step. The
estimation and change of support are thus not integrated.
In the authors experience, many practical applications of MIK
involve using the affine correction, which assumes that the shape
of the distribution of SMUs is identical to that of samples, the
sole change in the distribution being variance reduction as
predicted by Kriges Relationship (David, 1988).
Clear warnings can easily be found in the literature by Isaaks
and Srivastava (1989), Vann and Sans (1995), among others,
about the inherent deskewing of the distribution when going
from samples to blocks. The affine correction assumes no
deskewing, and thus is not suited to situations where there is a
large decrease in variance (ie where the nugget is high and/or
there is a pronounced short-scale structure in the variogram of
grades).
Other approaches can be utilised with MIK, eg direct or
indirect lognormal corrections (very distribution dependent), or
conditional simulation (often perceived as too complex and
costly in time). The authors have also seen assumptions of
normal (or Gaussian) block distribution. In this case, the
underlying assumption is that block grades are totally deskewed
(symmetrical). Our experience suggests that this assumption may
be true only for very high nugget situations.
Whatever the method used, there is no guarantee that the
corrections applied at a local level are consistent with the same
type of correction applied at a global level, and this is not very
satisfactory, to say the least.
Median indicator kriging is an approximation of MIK which

assumes that the spatial continuity of indicators at various
cut-offs can be approximated by a single structural function, that
~
for zc = m , where ~
m is the median of the grade distribution. This
has obvious advantages in speed (only one variogram model to
deal with) and in reduction of order relationship problems (see
above).
In Median IK the indicator variogram at (or close to) the
median is considered to be representative of the indicator
variograms at other cut-offs. This may or may not be true, and
needs to be checked. In the authors experience, most precious
metal deposits do not even approximate this assumption of
Median IK.
The most obvious advantage of median indicator kriging over
MIK is one of time (both variogram modelling and estimation).
The critical risk is in the adequacy of the implied approximation.
If there are noticeable differences in the shape of indicator
variograms at various cut-offs, one should be cautious about
using median indicator kriging (Isaaks and Srivastava, 1989,
p 444).
DISCUSSION: UNDERLYING ASSUMPTIONS OF

METHODS
MIK, like all other non-linear geostatistical methods, assumes a
model for spatial distribution of grades (contrary to the assertions
of many practitioners that MIK is free of assumptions).
The estimation of recoverable resources, in the simplest case,
corresponds to a two-fold problem: estimation of geometry and
estimation of grade.
A mineralised envelope is usually inferred from geological
interpretation. Within most orebodies there are good reasons to
suspect that the mineralisation will stand-up within a broad
envelope at a low grade or geological cut-off. More concern
arises when a cut-off is considered that is significantly higher
than this geological cut-off. A traditional practice for mining
geologists consists of drawing nested envelopes (or contours),
with the highest grade in the centre and grade decreasing
systematically towards the border in any direction. Lets call the
resulting model Situation 1.
Contouring grades may be helpful when interpreting possible
geological trends in the mineralisation, but experienced mine
This situation is the underlying assumption for methods like

Uniform Conditioning, discussed below, and it commonly suits many
metalliferous deposits, in the authors experience.
geologists know that in some cases it departs completely from

reality. The mineralisation can in reality exhibit moreor
lessrapid change in grade at different scales. At the scale of an
SMU, high-grade can be adjacent to sterile or medium-grade. We
will call such a situation Situation 2.
Geostatisticians say that in case of Situation 1 there are
border effects or edge effects (see Figure 2a), ie going from
grade 1 to grade 3 one will encounter grade 2. In the case of
Situation 2 (see Figure 2b), we say that there are no border
effects, ie that going from grade 1 we can meet grade 2 or grade
3. Several other cases can occur, with more or less strict
transitions, including the case of total independence between
zones at different grades.
Since the early-1970s several techniques have been used for
estimating
recoverable
resources,
based
on
either
multi-lognormal or multi-gaussian distributions, cokriging of
indicators (DK) or kriging of indicators (MIK).
Recall the definition of indicators introduced previously:
I(x, zc) = 0
if Z(x) zc
I(x, zc) = 1
if Z(x) > zc
The indicators are generally not independent: at a point x, with

two cut-offs Zc1 > Zc2, if I(x, zc1) > 1 then, I(x, zc2) > 1 and there
is dependence between indicator values at these two cut-offs.
MIK presumes that this dependence does not exist, and this is
a fundamental source of order relation problems.
FIG 2 - Spatial models for grade distribution - see text for discussion.
191
When estimating by any non-linear technique, do we

implicitly choose a particular border model, or can we identify
simply in which situation we are: border effect or no border
effect, to determine which estimating technique is more suitable
for our deposit? The answers are given in some detail in
Rivoirard (1993), but in short, the basic solution consists of:
1.
Computing the variograms and cross variograms at

different cutoffs z0, z1, z2, zi. This can be automatically
performed with modern software (eg Isatis, from
Geovariances).
2.
Comparing the cross variograms of indicators z0z1(h),

z0z2(h), ... z0zi(h) to the variograms of the lower cutoff
indicator z0(h).
Then the following situations can occur:
1.
If the ratio z0zi(h)/z0(h) increases with (h) there are

border effects, and a diffusion-type model is applicable
(eg discrete gaussian model, lognormal model).
2.
If the ratio z0zi(h)/z0(h) is constant for a given cut-off zi, ie

z0zi(h) and z0(h) are proportional, then the different
cut-offs are nesting without edge effects one within the
other, we are in a no border effect situation. In such case a
residual indicator kriging (RIK) method is applicable
(Rivoirard, 1989).
3.
If all variograms and cross-variograms are proportional to

the same model (h), we have a particular model called
mosaic model (see Figure 2c), which corresponds to a
situation where the separate kriging of indicators can
theoretically be applied (Journel, 1982). This is the implicit
assumption made when using Median Indicator Kriging.
This type of calculation has been performed in a detailed
variographic study of a Western Australian gold deposit by two
of the authors (see Figure 3). The results show that a complex
situation can occur with a border effect model occurring in two
main directions of variogram anisotropy, and a no border effect
occurring in the third direction of anisotropy. In this case the use
of both multi-gaussian and indicator residuals would have to
be tested and the results carefully compared.
Aside: Uniform Conditioning

Uniform conditioning (UC) is a variation of Gaussian disjunctive
kriging DK more adapted to situations where the stationarity is
not very good.
In order to ensure that the estimation is locally well
constrained, a preliminary ordinary kriging of relatively large
panels is made, and the proportions of ore per panel are
conditional to that kriging value.
All three authors of this paper have concluded through
experience that UC is a robust technique. However, it does
depend heavily upon the quality of the kriging of the panels. As
for DK, the discrete Gaussian model ensures consistent change
of support.
The authors believe that this method (which has seen increased
in usage in Australia in recent years) has much to recommend it.
It is more theoretically consistent with the observed geological
distribution of grade in many deposits, it has a robust change of
support and it suffers much fewer order relationship problems
than MIK. For more details on UC refer to Rivoirard (1994).
RECOMMENDATIONS
This papers title asks a question is MIK suited to my deposit?
As with most worthwhile questions in geology and geostatistics,
the answer requires that we ask further questions. The key
questions are:
1.
What are the relations between indicator variograms and

cross variograms in my deposit?
2.
Is there significant deskewing of the histogram when

changing support?
If the answer to question one above suggests that we have a
border effect situation then we should be wary of MIK. If the
answer to question two is yes than we should worry about
affine corrections. In both cases we would recommend UC as an
alternative method (which is, coincidentally, faster to
implement).
If the answer to question one above, suggests that we have no
border effect, then we recommend consideration of RIK
(because of the improved change of support compared to MIK).
Both UC and RIK require additional knowledge and skill in
geostatistics, compared to MIK. However, when large
investments are made on the results, this is a small price to pay
for more realistic assessment of mineral deposits.
CONCLUSIONS
FIG 3 - Variogram across veining of indicator at cut-off 0.5 g/t and cross
variogram of indicators at higher cut-offs.
192
1.
Small block OK or IDW estimates should no longer be

acceptable as inputs to important financial decisions.
2.
Understanding
the
underlying
assumptions
and
mathematics of these methods is critical to making
informed choices when selecting a technique.
3.
Testing of assumptions is rarely performed, in our

experience. We therefore recommend that such tests be
implemented (see Rivoirard, 1994).
4.
The issue of change of support is critical in estimation of

recoverable resources. Our major criticism of MIK, the
most widely applied non-linear estimation method at the
end of the 20th century, centres on change of support. The
whole problem of recoverable reserves is the problem of
change of support. We recommend that practitioners
become very familiar with the issue of change of support
and bring a sophisticated appreciation of this problem to
their practice.
ACKNOWLEDGEMENTS
The authors wish to acknowledge Henri Sanguinetti, an old
colleague of two of us (DG and JV) who contributed ideas to the
final discussion part of our paper. All three authors have
benefited from discussions with our professional geostatistical
colleagues at SRK. We also wish to acknowledge the support of
SRK in the preparation of this paper, and the valuable experience
gained by working over many years with clients making real
decisions on real projects.
REFERENCES
Armstrong, M, 1989. Geostatistics, in Proceedings of the 3rd
International Geostatistical Congress, Avignon, France. September,
1988, (2 volumes) (Kluwer Academic Publishers: Dordrecht).
Armstrong, M and Champigny, N, 1989. A study on kriging small
blocks, CIM Bulletin 82(923):128-133.
Armstrong, M and Matheron, G, 1986a. Disjunctive kriging revisited,
Part 1, Mathematical Geology, 18(8):711-728.
Armstrong, M and Matheron, G, 1986b. Disjunctive kriging revisited,
Part 2, Mathematical Geology, 18(8):729-742.
Baafi, E and Schofield, N A, 1997. Geostatistics Wollongong 96.
Proceedings of the 5th International Geostatistical Congress,
Wollongong, NSW, Australia, September, 1996, (2 volumes).
Cressie, N, 1993. Statistics for spatial data (revised edition) 900 p (John
Wiley and Son: New York).
David, M, 1972. Grade tonnage curve: use and misuse in ore reserve
estimation, Trans IMM, Sect A, 81:129-132.
David, M, 1988. Handbook of applied advanced geostatistical ore
reserve estimation. Developments in Geomathematics 6, 216 p
(Elsevier: Amsterdam).
Deutsch, C V and Lewis, R W, 1992. Advances in the practical
implementation of indicator geostatistics, in Proceedings 23rd
APCOM, pp169-179 (Port City Press: Baltimore).
Dowd, P A, 1982. Lognormal kriging the general case. Mathematical
Geology, 14(5):475-489.
Dowd, P A, 1992. A review of recent developments in geostatistics.
Computers and Geosciences, 17(10):1481-1500.
Guarascio, M, Pizzul, C and Bologna, F, 1989. Forecasting of selectivity,
in Geostatistics Proceedings of the 3rd Int Geostatistical Congress at
Avignon (Ed: M Armstrong) pp 901-909 Volume 2 (Kluwer
Academic Publishers: Dordrecht).
Guibal, D, 1987. Recoverable reserves estimation at an Australian gold
project, in Geostatistical case studies, (Eds: G Matheron and M
Armstrong) pp 149-168 (Reidel: Dordrecht).
Hohn, M E, 1988. Geostatistics and petroleum geology, 264 p (Van
Nostrand Reinhold: New York).
Isaaks, E H and Srivastava, R M, 1989. Applied Geostatistics, 561 p
(Oxford University Press: New York).
Journel, A G, 1980. The lognormal approach to predicting local
distributions or selective mining unit grades, Mathematical Geology,
12(4):285-303.
Journel, A G, 1983. Non-parametric estimation of spatial distributions,
Mathematical Geology, 15(3):445-468.
Journel, A G, 1982. The indicator approach to estimation of spatial data,
in Proceedings 17th APCOM, pp 793-806 (Port City Press: New
York).
Journel, A G, 1985. Recoverable reserves the geostatistical approach,
Mining Engineering, June 1985, pp 563-568.
Journel, A G, 1987. Geostatistics for the environmental sciences. United
States Environmental Protection Agency Report (Project CR
811893) 135 p (USEPA: Las Vegas).
Journel, A G, 1988. New distance measures the route toward truly
non-Gaussian geostatistics, Mathematical Geology, 20(4):459-475.
Journel, A G, 1989. Fundamentals of geostatistics in five lessons. Short
Course in Geology: Vol 8, 40 p (American Geophysical Union:
Washington).
Journel, A G and Deutsch, C V, 1998. GSLIB Geostatistical software
library and users guide, Second Edition, (Oxford University Press:
New York).
Journel, A G and Huijbregts, Ch J, 1978. Mining geostatistics.
(Academic Press: London), 600 p.
Krige, D G, 1951. A statistical approach to some basic Mine Valuation

Problems on the Whitwatersrand, Journal of the Chemical,
Metallurgical and Mining Society of South Africa (Dec 1951, March,
May, July and August 1952).
Krige, D G, 1994. An analysis of some essential basic tenets of
geostatistics not always practised in ore valuations, in Proceedings of
the Regional APCOM, Slovenia.
Krige, D G, 1996a. A basic perspective on the roles of classical statistics,
data search routines, conditional biases and information and
smoothing effects in ore block valuations, in Proceedings of the
Regional APCOM, Slovenia.
Krige, D G, 1996b. A practical analysis of the effects of spatial structure
and data available and accessed, on conditional biases in ordinary
kriging, in Geostatistics Wollongong 96, in Proceedings 5th
International Geostatistical Congress, Wollongong, NSW, Australia,
September, 1996, pp 799-810.
Krige, D G, 1997. Block kriging and the fallacy of endeavouring to
reduce or eliminate smoothing, in Proceedings of the Regional
APCOM, Moscow.
Lantuejoul, Ch, 1988. On the importance of choosing a change of support
model for global reserves estimation, Mathematical Geology,
20(8):1001-1019.
Marechal, A, 1978. Gaussian anamorphosis models. Fontainebleau
Summer School Notes C-72, 22 p (Centre de Morphologie
Mathematique: Fontainebleau).
Matheron, G, 1962. Traite de geostatistique applique, Tome I. Memoires
du Bureau de Recherches Geologiques et Minieres. No 14 (Editions
Technip: Paris).
Matheron, G, 1963a. Traite de geostatistique applique, Tome II: Le
Krigeage. Memoires du Bureau de Recherches Geologiques et
Minieres. No 24 (Editions Technip: Paris).
Matheron, G, 1963b. Principles of geostatistics, Economic Geology,
58:1246-1266.
Matheron, G, 1973. Le krigage disjonctif. Internal note N-360. 21 p
(Centre de Geostatistique: Fontainebleau).
Matheron, G, 1976. A simple substitute for conditional expectation: The
disjunctive kriging, in Advanced Geostatistics in the Mining
Industry. Proceedings of NATO ASI. (Ed: M Guarascio) pp 221-236
(Reidel: Dordrecht).
Matheron, G, 1984. Selectivity of the distributions and the second
principle of geostatistics, in Geostatistics for natural resources
characterisation (Eds: G Verly et al) pp 421-433 (Reidel Publishing
Co: Dordrecht).
Matheron, G and Armstrong, M, (Eds), 1987. Geostatistical case studies,
247 p (Reidel: Dordrecht).
Ravenscroft, P J and Armstrong, M, 1990. Kriging of block models the
dangers re-emphasised, in Proceedings of APCOM XXII, Berlin, pp
577-587.
Rivoirard, J, 1987. Geostatistics for skew distributions. South African
Short Course Notes, C-131, 31 p (Centre de Morphology
Mathematique: Fontainebleau).
Rivoirard, J, 1989. Models with orthogonal indicator Residuals, in
Proceedings of the 3rd International Geostatistical Congress,
Avignon, France. September, 1988, (Ed: M Armstrong) (2 volumes).
Rivoirard, J, 1993. Relations between the indicators related to a
regionalised variable, in Geostatistics Troia 92, (Ed: A Soares) pp
273-286 (Kluwer Academic Publishers: Dortrecht).
Rivoirard, J, 1994. Introduction to disjunctive kriging and non-linear
geostatistics, 180 p (Clarendon Press: Oxford).
Royle, A G, 1979. Estimating small blocks of ore, how to do it with
confidence, World Mining, April 1979.
Soares, A, (Ed) 1993. Geostatistics Troia 92, in Proceedings 3rd
International Geostatistical Congress, Troia, Portugal, (Kluwer
Academic Publishers: Dordrecht) (2 volumes).
Sullivan, J, 1984. Conditional recovery estimation through probability
kriging theory and practice, in Geostatistics for natural resources
characterisation, Part 1 (Eds: G Verly et al), Reidel (Dordrecht), pp
365-384.
Vann, J and Guibal, D, in press. Beyond ordinary kriging An overview
of non-linear estimation, in Mineral Resource and Ore Reserve
Estimation - The AusIMM Guide to Good Practice, (The
193
Vann, J and Sans, H, 1995. Global Resource Estimation and Change of

Support at the Enterprise Gold Mine, Pine Creek, Northern Territory
Application of the Geostatistical Discrete Gaussian Model, in
Proceedings APCOM XXV, pp 171-179 (The Australasian Institute of
194
Verly, G, 1983. The multigaussian approach and its applications to the

estimation of local reserves, Mathematical Geology, 15(2):259-286.
Verly, G and Sullivan, J, 1985. Multigaussian and probability krigings
an application to the Jerrit Canyon deposit, Mining Engineering,
June, 1985, pp 568-574.
Mining Bench Height Evaluation for the Wallaby Resource

A Conditional Simulation Case Study
I M Glacken1, M Nopp 2 and M Titley3
ABSTRACT
A study on the feasibility of mining the recently discovered Wallaby gold
deposit is due for completion by mid-2000. Open pit mining is the
favoured method. Determination of the impact of mining bench height on
the recovery of ore, including the estimation of dilution and ore loss, is
critical to the economics of the operation.
Conditional simulation was used to derive a number of gold estimation
scenarios within a typical volume of the deposit. These scenarios were
interrogated for a selection of mining bench heights and ore block
dimensions. The results were reported as a range of grade tonnage
relationships and compared to the kriged resource model. This allowed
the resource model to be calibrated to a specific mining bench height and
ore block size. Once calibrated, more realistic dilution and ore loss values
for a number of mining bench heights were calculated from the resource
model. These values were then used to determine the economic
sensitivity of particular mining methods.
The spread of results from the simulations also provided an indication
of the confidence in the kriged grade estimates for the different zones of
mineralisation within the deposit and highlighted areas of any significant
uncertainty.
INTRODUCTION
Wallaby is one of Australias largest gold discoveries in recent
years. Two exploration tenements cover the deposit. The Granny
Smith Joint Venture (Placer Dome Asia Pacific 60 per cent and
Delta Gold Ltd 40 per cent) (GSJV) holds Wallaby, which is
situated within the southern lease and includes the majority of
the resource. Homestake Gold of Australia Ltd holds Just In
Case which is in the northern lease. The Wallaby Mineral
Resource (Indicated and Inferred) as at 31 December 1999 was
52 million tonnes at 2.7 g/t for 4.5 million ounces of gold at a
1.0 g/t cut-off. The GSJV is completing an open pit mining
feasibility study that is due for completion around June 2000.
The nature of gold mineralisation in a deposit determines the
dilution (waste mixed with ore) and ore loss (ore lost to waste)
for different mining block sizes, referred to hereafter as the
selective mining unit (SMU). A feasibility study must optimise
the economic benefits gained from using larger ore mining
equipment, hence larger SMUs, against the potential loss of
revenue due to dilution and ore loss.
Conditional simulation is a tool which will generate a number
of equally-likely images of the interpolated data. Importantly,
simulation honours the statistical distribution of the input data.
This differs to other estimation techniques which usually produce
a single output image with a smoother statistical distribution than
that of the input data, particularly when the spacing of the input
data (drill holes) is much larger than the SMU. This smoothed
model does not adequately represent the actual grade tonnage
relationship of the deposit, and so cannot be used to determine
the impact of different SMU sizes. By deriving a number of gold
estimation scenarios using conditional simulation, a range of
possible grade tonnage relationships for a given SMU can be
1.
FAusIMM, Snowden Mining Industry Consultants, PO Box 77,

West Perth WA 6872.
2.
MAusIMM, Snowden Mining Industry Consultants, PO Box 77,

West Perth WA 6872.
3.
MAusIMM, Placer Granny Smith, PO Box 33, Laverton WA 6440.
calculated. These simulated models can then be compared to the

resource estimate and used to calculate more realistic dilution
and ore loss values to assist in determining the economic
sensitivity of a particular mining method.
Time constraints on the study meant that the entire model area
could not be simulated, so a representative volume of the
resource was selected. To ensure that the results can be used in
the feasibility study, the kriged resource model was calibrated to
a specific SMU, and adjustment factors provided for alternative
SMUs.
The spread of results from the best, median and worst
simulations also provide an indication of the confidence of the
kriged grade estimates for the different zones of mineralisation
within the deposit and highlights areas of significant uncertainty.
THE WALLABY DEPOSIT

Background
Wallaby is located in the North Eastern Goldfields region of
Western Australia, approximately 27 km south-southwest of
Laverton and 11 km southwest of the Granny Smith Mine, at
latitude 28 51S, longitude 122 19E (Figure 1). The deposit lies
on the northeastern shore of Lake Carey, which is a large salt
lake. The discovery history is complex and has been published in
Nielsen and Currie (1999).
FIG 1 - Wallaby deposit location plan.
Recent exploration history

The GSJV commenced fieldwork at Wallaby during November
1997. The program comprised reconnaissance aircore drilling.
Follow-up drilling in June 1998 confirmed continuity of grade
between the initial anomalies. Reverse Circulation drilling began
later that month but had difficulty penetrating the thick
water-saturated lacustrine clays. A combination of aircore
pre-collars with diamond tails proved to be the most cost
195
I M GLACKEN, M NOPP and M TITLEY
effective method of drilling. By December 1998, 65 holes for a

total 22 500 m were completed on 100 m centres. This drilling
broadly outlined the lateral limits of the deposit. An inferred
resource of 30 Mt at 2.4 g/t for 2.3 Moz was reported at a cut-off
of 1.0 g/t.
Drilling continued through 1999, concentrating on; infill to
increase grade confidence, determination of open pit mining
limits, improving geological understanding, collection of core for
metallurgical testwork, de-watering pump tests and geotechnical
evaluation.
Project status
An open pit mining feasibility study is due for completion
around June 2000. The current combined Indicated and Inferred
Mineral Resource is 52 Mt at 2.7 g/t for 4.5 Moz gold at a 1.0 g/t
cut-off. The potential open pit mine covers an area of 1300 mNS
1000 mEW 300 m vertical.
A program of diamond drill holes is being completed to test
extensions to the Wallaby mineralisation at depths down to
1000 m below the surface. Preliminary exploration of possible
southern extensions of the mineralisation system beneath the
Lake Carey sediments is on-going.
Regional geology of Wallaby

The Wallaby deposit is located within the Eastern Goldfields
Province of the Archaean Yilgarn granite-greenstone terrain of
Western Australia. The geology within this zone has been
divided into two associations; association 1, comprising mafics,
ultramafics and banded iron formations, and association 2,
comprising andesitic volcanics and volcaniclastics interbedded
with mafic volcanics and overlain by clastic and chemical
sediments. Association 2 occurs in the western and eastern parts
of the district, while association 1 occurs in a north-south
corridor through the central part of the Laverton area. Wallaby
occurs within association 1 and is located on the eastern flank of
the Mt Margaret Anticline. Figure 2 shows the location of
Wallaby and the regional geology around the Laverton area
FIG 2 - Wallaby deposit regional geology.
Mineralisation occurred after the emplacement of the felsic and

alkali intrusives but prior to the emplacement of the carbonatitic
fractionate. Figure 3 shows the geology in schematic section.
Structure, mineralisation and alteration

Deposit geology
The Wallaby deposit is covered by 1 to 20 m of aeolian dune
sands and between 24 to 120 m of Tertiary lake clays and sands.
This transported cover is shallowest on the eastern margin and
steadily thickens to the west. A deep regional palaeochannel
trending south-southeast runs through the western side of the
Wallaby deposit. No economic gold mineralisation is hosted in
the transported cover.
Oxide saprolitic material occurs mainly in the eastern part of
the deposit. Saprolitic material in the west has been mostly
stripped by the palaeochannel. Approximately ten per cent of the
gold in the Wallaby deposit is in oxide material.
The host rock at Wallaby is an Archaean matrix-supported
polymict conglomerate. The dominant clast type is mafic
volcanics. Other clast types include felsic porphyries, sediments
(mostly banded iron formations and cherts), and carbonate and
quartz clasts. The conglomerate is predominantly massive,
although occasional graded bedding, grit beds and clast
alignment indicate the unit is upright and has a dip of around 45
to the south east. The conglomerate has been metamorphosed to
upper greenschist facies.
Two north-south trending subvertical dyke swarms have
intruded the conglomerate. The dykes represent a fractionated
alkali syenite suite and range in composition from various
monzonites through syenites, to carbonatitic syenites and a
carbonatite. A number of the more alkali-rich fractionates have
been emplaced as sills within the shear structures. These
generally have an east-west orientation and southerly dip.
196
Mineralisation occurs along shears formed during a dominant top

block north to north east thrust regime. The shear zones display
limited shear foliation. This is due to the ability of the host
conglomerate to absorb large amounts of strain with limited
physical deformation The main shears are relatively flat lying,
with a gross gentle dip towards the south east. Lower grade
mineralisation is associated with steeper on echelon linkage
structures, with a dominant dip to the north east. The shear zones
are defined more by alteration than foliation, and range in size
from 1 to 40 m.
The alteration at Wallaby can be classified as an inner,
intermediate, or outer halo.
The inner halo defines the dominant ore-zones and is a
dolomite-albite-pyrite-chalcopyrite and gold assemblage. It is
bleached, lacks a significant magnetic response, and primary
textures are often obliterated. Small amounts of visible gold
occur in small, late-stage quartz/carbonate veins. Gold is closely
related to pyrite content. The highest grades tend to be related to
abundant fine pyrite. Mineralisation behaves differently in the
intrusives than in the conglomerate, with the grades tending to be
lower in the intrusives.
The intermediate halo comprises biotite, pyrite, magnetite and
chalcopyrite. It is dark green to brownish grey and may have a
significant magnetic response. This alteration is derived from the
same fluids as the inner zone but with a lower fluid to rock ratio.
A petrological study of the magnetite shows magnetite and pyrite
are in textural equilibrium throughout the deposit at all scales.
This implies they formed at the same time.
MINING BENCH HEIGHT EVALUATION FOR THE WALLABY RESOURCE
FIG 3 - Wallaby deposit: schematic east-west section regolith and geology.
The outer halo is generally unmineralised conglomerate. It is

split into a chlorite-albite assemblage and an actinolitemagnetite-pyrite alteration. The actinolite assemblage has a
significant magnetic response and is the distal alteration from the
mineralising fluid. The chlorite-albite assemblage is from the
regional upper greenschist facies metamorphism.
Description of mineralisation zones

There are two main flat lying ore-zones, termed 50 and 60,
which are separated by about 100 to 150 m. These zones range in
thickness from 5 to 35 m. Ore-zone 40 is thought to be a flat
lying above ore-zone 50, but is almost entirely in oxide and is not
as well constrained as the other ore-zones.
The north east dipping ore-zones, 240 and 250, lie above and
below ore-zone 50 respectively. Ore-zone 240 is well
constrained, is very strongly mineralised, and ranges in thickness
from 3 to 25 m. Ore-zone 250 comprises a series of stacked
northeast dipping structures which cannot be easily correlated
between holes. There is structural evidence indicating that parts
of ore-zone 250 flatten, and in some cases dip to the south.
Ore-zone 250 is not well constrained and has the lowest grades.
The individual ore-zone 250 structures range in thickness from
2 to 7 m.
Ore-zone 70 lies directly below ore-zone 60 and being the
deepest, is not well defined by drilling. The current interpretation
of ore-zone 70 is a series of gently northeast dipping stacked
planar zones. The stacking suggests it may be similar to ore-zone
250 but with areas of better grade continuity.
The ore-zones cross intrusive/conglomerate boundaries with
little change in geometry. There is a subtle increase in gold
dispersion through the felsic intrusives, especially for the very
low gold values. Figures 4 and 5 show the ore-zones in schematic
north-south and east-west section.
RESOURCE ESTIMATION
Solid 3D triangulations representing the major mineralised
ore-zones were created. Mineralised ore-zones were based on a
structural model and defined by alteration, depleted magnetic
susceptibility and a gold grade cut-off around 0.2 g/t. Ore-zone
250 consists of a number of small, poorly defined mineralisation
zones that could not be adequately linked to form a
homogeneous interpretation. The ore-zone lies between the base
of ore-zone 50 and the top of ore-zone 60 and there is a relatively
high proportion of unmineralised material included within this
profile.
The geometry of the intrusives could not be interpreted in
enough detail to create a solid wireframe, as it is not uncommon
for a drill hole to intersect intrusives having a thickness greater
than 100 m while surrounding holes may not contain any
intrusives greater than 1 m thick. Also, the majority of the
intrusives are unrooted, ie drilling continued through them and
back into conglomerate. The geometry of the intrusives is an
important issue as the intrusives have a different gold grade
distribution. An intrusive lithology code was assigned to samples
from the geology logs. This code was initially used for statistical
analysis and then later, during block model construction, to
estimate the proportion of intrusive material in a model block.
Drill hole samples were assigned weathering, lithology and
ore-zone codes, by using the wireframe models and logging
codes. The three codes were combined into a single unique open
pit ore-zone code (OPDOM) representing the different
combinations of codes. Two metre downhole composites were
created for statistical analysis, variography and grade estimation.
The 2 m composite length was selected as it provides the best
resolution of the ore-zone geometry while minimising the mix of
short and long composites. All composites were terminated on
changes within the OPDOM code to ensure different materials
were not mixed.
197
FIG 4 - Wallaby deposit ore-zone geometry, east-west section.
FIG 5 - Wallaby deposit ore-zone geometry, north-south section.
198
Statistical analysis and variographic analysis of the composites

was completed, with the following significant conclusions:
The grade of felsic intrusive hosted mineralisation needed to
be estimated independently of the conglomerate-hosted
mineralisation, as it has a different grade distribution and
different variogram parameters.
The alkali intrusive samples could be combined with the
conglomerate samples for the following reasons:
1. The number of alkali intrusive samples within some
ore-zones is insignificant.
2. Where significant number of samples are available the
grade distribution is similar to the conglomerate.
3. The alkali intrusive geometry was difficult to model
separately and too few samples were available to use
proportional indicator modelling. Thus, the conglomerate
and alkali intrusive material were combined, and are
referred to as conglomerate in the rest of the paper.
Hard mineralisation boundaries must be used to constrain the
mineralisation within ore-zones. However, ore-zones that
intersect each other (such as 240 and 50) may use samples
from either ore-zone.
Ordinary Kriging with cutting would be suitable for grade
estimation for the following reasons:
1. The variography study showed no spatial grade
anisotropy. That is, high-grades were not oriented
differently to the lower grades.
2. The distributions within each ore-zone and lithology
have a relatively low degree of skewness.
3. Only 0.1 per cent of high-grade composites required
trimming.
4. Lack of geological evidence for the alignment of grades,
either high or low, in veins, shoots or any structures
within the ore-zones.
The ore-zones have considerable local variation, so the average
orientation does not give the best result for variography and
grade estimation. This is especially relevant for ore-zones 50, 60
and 240 as their orientations are not consistent throughout the
deposit. To avoid having to split these ore-zones into numerous
smaller zones, the samples were unfolded for variography and an
in-house anisotropic modelling package was used for the grade
estimation. The Placer Open Pit software allows construction of
an anisotropy model. The anisotropy model is a block model
containing local azimuth, dip and plunges for each block. This
local information is used as directions for the search and
variogram calculations when performing grade estimations.
A geology block model was constructed using a constant
10 mN 10 mE 5 mRL block size. An OPDOM code was
assigned from the relevant wireframe models. As an accurate
geology model of the felsic intrusives could not be created, an
indicator method was used. The proportion of felsic intrusive
within each block was calculated using ordinary kriging of
conditional 0 (absence of felsics in a sample) and 1 (felsic
sample) indicators.
Both the conglomerate and felsic intrusive grades were
estimated for each model block using ordinary kriging. A final
average tonnage-weighted gold grade was calculated for each
model block, based on the proportions of conglomerate and felsic
intrusive within each block.
CONDITIONAL SIMULATION
Objectives of conditional simulation
Conditional simulation is a technique which has developed as an
alternative to estimation techniques such as polygons, inverse
distance weighting, ordinary kriging or indicator kriging.
Simulation generates many equally likely scenarios, in contrast

to estimation which provides one only. As the name implies, it
uses the principles of Monte Carlo simulation, random number
drawing from a specific distribution, while ensuring that the
outcome is conditional to the input data and the geological
model. In particular, each simulation honours the statistical
distribution of input grades (the grade returned at any point
where a data value exists is that data value) including the
variability, as typified by the range, variance, or coefficient of
variation. Each simulation also honours the spatial continuity of
the raw data as represented in the variogram, and any other
features of the geological model which have been built in.
Kriging, which is theoretically the most optimal estimation
technique, also uses the spatial variability of the data to derive
the weighting scheme for each unknown point to be informed,
but the output kriged map does not preserve the variogram. In
contrast to simulation, a single kriged estimate provides the
minimum variance set of values. Each individual simulation,
taken in isolation, is more variable than any one kriging estimate,
but the ensemble of simulations (typically between 20 and 100
are generated), treated together, provide more information than
kriging. Simulation preserves more characteristics of the input
data than other estimation techniques; it also has the benefit of
providing the risk dimension to resource evaluation studies that
kriging cannot readily provide.
Although the principles of simulation in other fields have been
known for decades, conditional simulation in the mining and
petroleum fields has been practised for only 20 to 25 years, and it
is only with the advent of fast, affordable computers and memory
that practical simulation has been available to the mining
industry. Good descriptions of simulation are provided in
Goovaerts (1997), by Srivastava (1994), and in Chils and
Delfiner (1999).
Applications of conditional simulation

The applications of conditional simulation in the mining industry
fall into four broad categories:
Probability and confidence interval analysis

In this field of application, the suite of simulations generated are
used to provide indications of the range of likely outcomes. This
involves ranking each individual simulation in some way
average grade above a cut-off, maximum tonnage, or maximum
metal and then tabulating or processing the extreme values. It is
possible to generate a true probability or confidence interval
around a median, mean, or any value using simulations for
example, the 95 per cent grade and tonnage confidence interval
around the cut-off at 1 g/t in a gold deposit, or the range of
tonnage expected to be delivered to a plant from one bench of a
nickel laterite deposit. Using this approach, the extreme cases
can be further processed to yield best and worst outcomes,
using such techniques as pit optimisation, underground stope
optimisation, schedule analysis or cashflow forecasting. It is also
possible using the suite of simulations to derive the probability of
exceeding any particular key value, such as a cut-off grade or
minimum level of contaminants. A case study of this type of
approach is presented in Coombes et al (1998).
Optimisation
Optimisation of simulations takes the entire range of outcomes
and produces a best result, which is optimal for a given set of
known criteria. These criteria are often presented as a loss
function, or more generally an economic function combining
profit and loss components (Srivastava, 1987; Glacken, 1996).
The typical application of optimisation of conditional simulation
output is in grade control applications, whereby an optimal
199
dig-line is generated according to the balance of economic

criteria applied in designating a block to either ore, waste, or one
of a number of stockpiles. This approach has been explored by
Shaw and Khosrowshahi (1997) and by Schofield (1998), and
has been adopted in a number of available proprietary software
systems.
Reblocking/Resampling
This application of conditional simulation takes one or several of
a suite of simulations and treats that simulation as reality; in
other words, a model of the phenomenon under study. This can
be readily achieved as simulations do not smooth the data as with
kriging, and also by the fact that it is possible to simulate into
very small volumes (essentially nodes or points) without
encountering conditional bias. The fine-scale simulations can be
sampled to represent various grade control or drilling patterns, or
also to represent production over a range of timeframes or
production intervals. Another typical application is the analysis
of a range of mining unit (SMU) sizes, which can easily be
generated by reblocking or averaging of a fine-scale simulation,
providing that the variable under study can be linearly averaged
(which is certainly the case for most mining applications). It is
this application of simulation which was used to investigate the
range of bench heights at Wallaby; by simulating at a fine-scale,
it was possible to aggregate the node values over a range of
bench heights and then to investigate the ore loss and/or dilution
relative to the kriged model. This was particularly important
given the strong ore-zone control on mineralisation at Wallaby
and the potential for more or less dilution at various mining
scales.
Instead of reblocking a single simulation, a best and worst case
scenario (actually the 5th and the 95th percentile of the
simulations when ranked by average grade) were considered in
order to assess the extra dimension of the uncertainty due to the
risk in the overall grade.
Calibration of a resource model

By comparing the selectivity of a resource model with that of a
similar kriged model, it is possible to assess the effective
selectivity of the kriged model. This assumes that the simulation
is a more accurate representation of reality than the kriged
model, and given the accurate reproduction of statistical and
spatial characteristics, coupled with the optimal change of
support for simulation and lack of conditional bias, this should
be the case. By comparing the simulation tonnage-grade curve
with the kriging curve on a similar block size, it is possible to
calibrate the kriged model (effectively increase or decrease the
selectivity) to more accurately represent the likely mining
conditions. This approach was also adopted in the Wallaby study.
DETAILS OF STUDY
Selection of test area
One downside of simulation is that it can be a fairly
time-intensive technique as multiple realisations of very small
blocks are required. Validation of the results can also be a
lengthy exercise. It was therefore decided that the simulation
should be carried out within a representative volume of the
Wallaby orebody which was large enough to allow meaningful
analysis of results, but also small enough to be completed in a
reasonable time frame. The test volume was selected to include
representative areas through the main ore zones in terms of grade
and thickness, and also to allow sufficient vertical extent for the
effects of dilution and ore loss to be modelled over a range of
bench heights. A typical cross-section through the test area is
shown in Figure 6.
200
FILTER
Category
Background
Rcode 50
Rcode 60
Rcode 240
Rcode 250
FIG 6 - East-west cross-section of the test area at 808 200 mN showing
the arrangement of grade ore-zones.
Generation of conditional simulations

The variography study established that the degree of rotation of
anisotropy between the high- and low-grades was small. In other
words, the continuity of the high-grades was not oriented
particularly differently to the low-grades within any or all of the
ore-zones. This information, together with the low degrees of
skewness displayed by the data, led to the choice of sequential
Gaussian as the simulation algorithm. This approach was further
supported by the lack of geological evidence for strong
connectivity of high-grades or low-grades; in other words,
neither the high-grades nor the low-grades appear to be aligned
in veins, shoots, or structures to any significant degree.
The use of sequential Gaussian necessitated the generation of
normal scores variograms for the gold grades for both the
conglomerate samples and the intrusive samples within each of
the five ore-zones represented in the study area. The global data
sets for the entire resource, composited to 2 m, were used for the
variography. Care was taken to ensure that the directions of
maximum continuity coincided both with the known structural
directions and with those directions previously modelled in the
traditional variography used for the kriged resource estimate. The
variograms are characterised by low to moderate nugget
variances (25 to 40 per cent of total variability) and overall
ranges up to 100 m. The flat structures (ore-zones 50 and 60)
show southerly strikes on the mineralisation, but the link
structures (ore-zones 240 and 250) have primary directions of
continuity which dip shallowly to the northeast.
Since the style of mineralisation in the intrusive suite is
different to the conglomerate, these units were treated
independently within each ore-zone. The grades within the
conglomerate and intrusive suites were simulated separately and
twenty simulations were produced for each of the five ore-zones
within each suite, with the grades simulated into a fine grid of
2.5 mN 2.5 mE 2.5 mRL.
A block model of the spatial occurrence of the intrusive and
conglomerate suites was created from the proportion of intrusive
determined using categorical kriging (indicator kriging of the
presence or absence of a single variable). To this end all
composites within the extended area were coded to either one or
zero depending on whether they were felsic intrusive or
conglomerate. A variogram model based upon these indicator
codes was modelled and used to estimate the proportion of

intrusive material for each block using ordinary kriging, with a
steeply orientated sample search.
All of the simulations for each ore type were validated by
comparing the input data histograms with the output simulated
values by ore-zone. Q-Q plots indicated very good correlation
between the input data and the simulated values per ore-zone. In
addition, spot checks on the variography showed good
reproduction of variogram models, which are constrained by the
wide spaced data.
The ordinary kriged proportion model was then used to weight
the relative grades of conglomerate and intrusive when the
simulations were merged, using the following formula:
Simulated grade in block = proportion of intrusive simulated
intrusive grade +(1 proportion of intrusive) simulated
conglomerate grade
The densities are similar between the rock types (three per cent
difference) and, given the variability generated by the
simulations, tonnage weighting of the grades was not considered
to be necessary.
Each of the 20 simulations was merged in this way, and the
average of all simulations after merging was also generated. The
merged density was allocated using the proportion of intrusive in

a block and the density for pure intrusive and for pure
conglomerate.
The simulation models were ranked according to the grade at
zero cut-off, and the second lowest and second highest
simulation models (effectively the 5th and the 95th percentile
models) were selected to represent worst and best case scenarios
respectively. The middle, or median, model was selected as a
middle case scenario. Note that the ranking is based on the
average grade at zero cut-off over the entire test area, and not for
any particular ore-zone, since ranking of individual ore-zones
could cause unrealistically conservative worst case and overly
optimistic upper case models, particularly when benches span
boundaries between ore-zones.
As an additional validation, the mean grades and coefficient of
variation (COV) per ore-zone for the input data and the median
simulation, together with the means from the kriged resource
model, are compared in Table 1. The median simulation data
clearly honours the input data, while some of the kriged model
results display more significant differences. Three 25 m spaced
east-west sections through the test volume for the overall median
simulation are shown in Figure 7. The simulation has been
reblocked to 10 mE 10 mN 5 mRL.
FIG 7 - Diagram of three 25 m spaced east-west cross-sections through the median simulation, reblocked to 10 mN 10 mE 5 mRL.
TABLE 1
Comparison of gold statistics: composites, simulation and kriged model.
Ore-zone
2 m composites
Median simulation
Wallaby kriged model
Mean
COV
Mean
COV
Mean
Ore-zone 0
0.11
2.609
0.11
2.609
0.25
Ore-zone 50
2.11
2.607
2.10
2.514
2.03
Ore-zone 60
1.66
1.325
1.66
1.316
1.53
Ore-zone 240
2.69
1.284
2.69
1.269
2.83
Ore-zone 250
0.45
2.039
0.45
2.039
0.48
201
CALIBRATING THE RESOURCE MODEL

One of the desirable outcomes from the simulation study was the
ability to investigate the selectivity of the kriged resource
estimate. Because the simulations were generated at a fine-scale,
and honoured both the statistical and spatial variability of the
input composite data, when reblocked to the dimensions of the
kriged estimate (ie 10 mN 10mE 5 mRL) they provide a
range of outcomes, each of which is free of conditional bias, and
which can be used to look at the smoothness of the kriging.
The results for the overall study area, including all ore-zones,
are shown in Figure 8. This shows that, overall, the kriged model
sits within the range of the 10 mN 10 mE 5 mRL reblocked
simulations.
However, when individual ore-zones within the simulations are
compared with similar blocks in the kriged model, the results are
different. Figure 9 shows the results for ore-zone 50, and Figure
10 shows the results for ore-zone 250. The ore-zone 50 results
show that the grades for the reblocked simulation
(10 mN 10 mE 5 mRL) and the kriged model at 0 g/t cut-off
grade are very similar, at 2.06 g/t and 2.03 g/t, respectively.
However, the slopes and positions of the curves indicate that the
kriged model for ore-zone 50 is oversmoothed, and probably
represents the selectivity to be expected from blocks of at least
20 m 20 m 15 m. An explanation for the lower grade of the
simulations reblocked to larger block sizes compared to the
kriged model is that the reblocking includes the effect of
boundary dilution with increasing bench height, whereas the
kriged model does not.
The comparison of kriging results and various reblocked
(median) simulations for ore-zone 250 show a different result
(Figure 10). The grades for the reblocked simulation
(10 mN 10 mE 5 mRL) and the kriged model at 0 g/t COG
are similar, at 0.45 g/t and 0.48 g/t, respectively. As expected, an
increase in simulation block size results in a general increase in
tonnes and a decrease in grade. The decrease in grade is most
dramatic for higher cut-offs in ore-zone 250. The implication is
that it may be difficult to achieve high-grades if large selective

mining units are used. A large increase in tonnes with increasing
block size is evident in ore-zone 250 for low cut-offs. This is to
be expected from the volume-variance effect, and contrasts with
a larger change in grade for increasing block size at the higher
cut-offs. The slope and nature of the curves suggest that the
kriged model for ore-zone 250 is insufficiently smoothed and
represents the selectivity to be expected from blocks of less than
5 m 5 m 5 m. This is not generally expected from a kriged
model, and an explanation may be found in the search and
sample selection applied during the kriging.
Implications for kriging

The comparison of the reblocked simulations and the kriged
model indicates that the kriged model performs reasonably well
overall, with the degree of smoothing commensurate with blocks
in the order of 5 mN 5 mE 5 mRL to 10 mN 10 mE
5 mRL. Based on this overall comparison, tonnage and grade
factors were prepared relative to the 10 mN 10 mE 5 mRL
reblocked simulation so as to report the grades and tonnages
expected for a range of block size and bench height scenarios at a
range of cut-off grades.
However, it is clear from the comparison of grade-tonnage
curves for ore-zones 50 and 250 that there are differing degrees
of smoothing per ore-zone in the kriged model, with the
comparison for ore-zone 50 suggesting a moderate degree of
smoothing and ore-zone 250, in contrast, very little smoothing.
These differences are important since they highlight where the
kriged model may be expected to under- or over-perform relative
to the actual selectivity of the modelled block size. While it
would be possible to factor the model to reflect the differing
degrees of smoothing, a better approach would be that the
kriging parameters first be examined to identify whether the
necessary modifications can be effected by changing certain
kriging parameters.
FIG 8 - Tonnage-grade curves for selected simulations reblocked to 10 mN 10 mE 5 mRL together with kriged model (in bold);
1 g/t cut-off has larger symbol.
202
FIG 9 - Ore-zone 50 kriged model (in bold) and median simulations at various block sizes.
FIG 10 - Ore-zone 250 kriged model (in bold) and median simulations at various block sizes.
203
BENCH HEIGHT STUDY
1.40
Tonnage and grade factors
1.30
The reblocked simulations were used to investigate the effects of

using different bench heights for the proposed Wallaby pit. The
results from the various reblocking exercises were presented
relative to the 10 mN 10 mE 5 mRL reblocked simulation
which has been compared to the kriged results above. The other
simulations were presented as tonnage and grade factors for a
range of cut-off grades. The results for ore-zones 50 and 250 are
presented in Figures 11 to 14.
1.20
1.10
1.00
0.90
0.80
0.5
0.6
0.7
0.8
0.9
1.0
1.1
1.2
1.3
1.4
1.5
1.6
Cut-off Grade g/t

5x5x5
5x5x10
10x10x7.5
10x10x10
10x10x12.5
10x10x15
15x15x15
Factor
FIG 14 - Grade factors for ore-zone 250 relative to 10 mN 10 mE

5 mRL blocks.
Cut-off Grade g/t
FIG 11 - Tonnage factors for ore-zone 50 relative to 10 mN 10 mE

5 mRL blocks.
The results show generally that block sizes larger than the base
case generate more tonnage for any given cut-off grade, but at
lower grades than the base case. For ore-zone 50 the tonnage
increases with increasing cut-off grade up to a point, beyond
which the tonnage relative to the base case remains constant. The
tonnage curves for ore-zone 250 show that the larger blocks
show a drop-off in tonnage as the cut-off grade increases,
suggesting that the homogenisation of the higher grade and lower
grade portions of the mixed ore-zone 250 occurs more quickly at
larger bench heights. The grade factors for ore-zone 250 are also
fairly insensitive to changes in cut-off grade, reflecting the
dilution shown in the tonnage factor chart. The results show that
the choice of bench height may be more consequential for
ore-zone 250 than for ore-zone 50, and that the cut-off grade in
this ore-zone has relatively little effect on the tonnage and grade
factors relative to the base case. In contrast, the choice of bench
height for ore-zone 50 appears to be less important than the
cut-off grade.
The tonnage and grade factor curves were each modelled by a
polynomial regression method so that the tonnage and grade
relative to the base case could be determined for any intermediate
cut-off. This enabled the factors to be built directly into the pit
optimisation.
Dilution and ore loss factors
FIG 12 - Grade factors for ore-zone 50 relative to 10 mN 10 mE

5 mRL blocks.
FIG 13 - Tonnage factors for ore-zone 250 relative to 10 mN 10 mE

5 mRL blocks.
204
The results of the reblocking were also presented as percentage

dilution and ore loss. The dilution was defined at a given cut-off
as the difference between the tonnage from larger blocks and the
base case (10 mN 10 mE 5 mRL) blocks, relative to the base
case blocks. This is effectively the relative change in tonnage
above a given cut-off. Ore loss was defined as the change in
grade between the larger blocks and the base case blocks relative
to those base case blocks. This ore loss is directly proportional to
the percentage change in revenue per tonne with changing block
size.
The dilution and ore loss factors were tabulated for a range of
cut-off grades, and the issue of risk was introduced by examining
the differences between the low, median, and high-grade
scenarios. Overall, dilution rates vary between ten and 25 per
cent relative to the base case block size, with larger block sizes
incurring more dilution. For tightly constrained ore-zones such
as 50, the change in dilution and ore loss is relatively small
moving from smaller blocks to larger blocks. However, for
ore-zone 250, larger block sizes incur proportionally more
dilution. The spread of outcomes between the high and low
simulations is much greater for ore-zone 250 than for ore-zone
50, reflecting the much greater uncertainty in the definition of the
250 ore-zone. The dilution and ore loss results show that
recovery from ore-zone 250 would be maximised with smaller
benches.
OUTCOMES OF THE STUDY
ACKNOWLEDGEMENTS
The use of simulation at Wallaby has provided a number of

benefits for the mine planning and feasibility study activities.
Firstly the resource estimate has been validated, and areas of
potential under- and oversmoothing highlighted. Secondly, the
analysis of different bench heights as provided by the reblocked
simulations has led to the generation of a range of tonnage and
grade correction factors which can be used in reserve
optimisation scenarios and over the range of expected cut-off
grades. The calculation of dilution and ore loss percentages for
various ore-zones, cut-off grades, and simulation scenarios has
shown the sensitivity of the ore-zones to bench heights and
cut-off grade combinations. Upon the input of mining costs
based on the predicted equipment and blasting scenarios for the
various bench heights it will be possible to generate actual
forecast revenue figures. Finally, the analysis of high, low, and
median scenario simulations has highlighted those areas which
are very robust (low range of outcomes) and those areas which
are high risk (high range of outcomes). The optimisation of
end-member simulations, ranked on those critical areas (such as
the grade and tonnage of ore-zone 250) will reveal the robustness
of any pit design.
A further application of conditional simulation, not yet used at
Wallaby, would be to sample a chosen simulation or simulations
on various grade control grids. This could be used as a first pass
analysis of the implications and costs of different grade control
patterns and could be used to optimise drilling costs.
Overall, the application of conditional simulation has lent an
extra dimension to the feasibility study through the quantification
of risk, the validation and calibration of the resource, and the
important information for the economic and mining evaluation of
suitable bench heights.
The management of Placer (Granny Smith) and Delta Gold have

given permission to publish these findings. Tim Keleman and
Mathew Matheson at Granny Smith assisted with geological and
grade control aspects of the study, and Elizabeth Haren
completed the variography, geological modelling and data
preparation. Jacqui Coombes, Steve Potter, and Craig
MacDonald of Snowden helped generate and post-process the
simulations. Vivienne Snowden acted as project reviewer.
REFERENCES
Chils, J-P and Delfiner, P, 1999. Geostatistics, Modelling Spatial
Uncertainty, 696 p (John Wiley and Sons).
Coombes, J, Thomas, G S, Gifford, M and Jepsen, L, 1998. Assessing the
risk of incorrect prediction a nickel/cobalt case study, in
Proceedings Mine to Mill 1998, pp 63-68 (The Australasian Institute
of Mining and Metallurgy: Melbourne).
Glacken, I M, 1996. Change of support and use of economic parameters
in block selection, in Geostatistics Wollongong 96 (Eds: E Y Baafi
and N A Schofield) 1997 pp 811-821 (Kluwer: The Netherlands).
Goovaerts, P, 1997. Geostatistics for Natural Resources, 483 p (Oxford
University Press: New York).
Nielsen, K I and Currie, D A, 1999. The discovery of the Just In
Case/Wallaby gold deposit, Laverton District, Western Australia, in
Proceedings New Generation Gold Mines 99 Conference,
November 1999, pp 1-13.
Shaw, W J and Khosrowshahi, S, 1997. Grade control sampling and ore
blocking: Optimisation based on conditional simulation, in
Proceedings Third International Mining Geology Conference, pp
Melbourne).
Schofield, N A, 1998. The myth of mine reconciliation, in Proceedings
Mine to Mill 1998, pp 87-92 (The Australasian Institute of Mining
Srivastava, R M, 1987. Minimum variance or maximum profitability?
CIM Bulletin, 80(901):63-68.
Srivastava, R M, 1994. The visualization of spatial uncertainty, in
Stochastic Modeling and Geostatistics. Principles, Methods, and
Case Studies, Volume 3 of AAPG Computer Applications in Geology,
(Eds: J M Yarus and R L Chambers) pp 339-345.
205
Ore Definition at the Henty Gold Mine Tasmania

N Schofield1
ABSTRACT
The Henty Gold Mine is a good example of a situation where conditional
simulation and optimisation techniques may have useful impact on
modelling of mineralisation distribution and ore definition. Much of the
gold mineralisation at Henty is concentrated within an alteration zone
which is both physically discontinuous and forms only a small fraction of
the overall mineralised volume. The alteration units are mappable from
diamond drill core and underground workings.
Conditional simulation and optimisation techniques for ore definition
have found considerable success in the open pit gold mining
environment. For many reasons including computational efficiency, lack
of consistent geological information, and lack of familiarity with ideas
and methods among mine geologists, implementations of these methods
in open pit mining have been comparatively simplistic. Modelling is
commonly reduced to treating the sample data as a single geologic
population.
This paper discusses the application of the MP method (a particular
implementation of conditional simulation and optimisation) to modelling
the distribution of potential ore blocks in the Henty gold mineralisation.
The approach allows the mine geologist to model both the spatial
geometry of the alteration types and the gold grades associated with each
alteration type in a single process. Options allow the geologist to include
or exclude underground geological mapping and face sampling
depending mainly on the abundance of drill hole sampling and the quality
of the face sampling. An optimisation algorithm based on maximisation
of profit from mining is used to assist in ore definition.
INTRODUCTION
The Henty Gold mine is located approximately 30 km north of
Queenstown on the west coast of Tasmania (De Mark and
Callaghan, 1997). The mine encompasses four major zones of
gold mineralisation trending along strike and down plunge within
a package of highly altered volcanic rocks. The shallowest zone
is termed the Sill Zone, followed by the Intermediate Zone, Zone
96 and Mt Julia. Zone 96 has been the focus of initial production,
the Intermediate and Mt Julia Zones being explored extensively
over the past three years.
Henty lies within the Cambrian Mount Read Volcanic Belt, a
rich mineral zone that hosts the Hellyer, Que River, Rosebery,
Hercules and Mt Lyell mineral deposits. Henty is contemporary
with the high grade Cambrian Pb-Zn-Cu-Au-Ag mineralisation
in these deposits. However, the high-grade gold deposits at Henty
are a new style of mineralisation in the Mt Read Volcanics and
are the only known economic deposits hosted in the Tyndall
Group rocks.
The Henty mine lease covers rocks of the Central Volcanic
Sequences, the Henty Fault Sequences and the Tyndall Group
rocks of the Mount Read Volcanics and the overlying Owen
Conglomerate. Near the mine, the Henty Fault splays into the
North and South Henty Faults, dividing the geology into
segments to the east and west of the faults, and a package
between the splays. Gold mineralisation is hosted in the Tyndall
Group rocks to the east of the Henty Fault.
The Henty orebodies are hosted east of the Henty Fault on the
steeply west dipping overturned western limb of a shallowly
south plunging asymmetric syncline trending into the Henty
Fault. The orebodies plunge at around 45 degrees to the south
between the Sill Zone and Zone 96, and shallow at depth toward
Mt Julia. Zone 96 is up to 300 metres long, 30 metres wide and
1.
MAusIMM, Hellman and Schofield Pty Ltd, PO Box 599, Beecroft

NSW 2119.
200 metres high. The lithologies and regional foliation strike at

340 degrees with a steep westerly dip, and swing parallel to the
Henty Fault as they approach the structure.
The structure of the Henty Gold Mine is dominated by the
steeply west dipping Henty Fault Zone. The 20 metre wide South
Henty Fault forms the upper boundary to mineralisation, and
truncates it to the south. Adjacent to the South Henty Fault,
foliation intensifies and lithologies become brecciated and
mylonitised. The orebodies are disrupted by numerous south
trending, steeply west dipping brittle-ductile faults with
displacements of up to a few metres.
Nearly all of the stratigraphic units of the Tyndall Group
present at the Henty Gold Mine have undergone hydrothermal
alteration. The most intense quartz-sericite-sulphide alteration
and gold mineralisation has affected the Lynchfold Member of
the Comstock Formation and is referred to as the A-Zone type
alteration. A Zone alteration types include MA, MZ, MV, MA,
MP and CB. The main mineralised zone comprises MQ, MV and
MZ and these are the alteration types followed in the
underground drives in Zone 96. Of these, MV alteration
comprising a yellow-green fine grained and highly foliated
quartz-sericite-pyrite rock and MQ comprising a grey, cream or
pink massive to recrystallised brecciated quartz rock carry most
of the gold mineralisation with MQ carrying around 80 per cent.
Figure 1 (after De Mark and Callaghan, 1997) shows a
schematic cross-section through the Henty deposit and
surrounding geology. Figure 2 shows the spatial distribution of
gold grades in a number of drill holes and underground samples
at the 2142RL. The figure illustrates the very narrow
intersections in drill holes and the underground sample grades
provide a reasonable illustration of the grade continuity between
drill hole intersections. The sample symbols show the alterations
types mapped at each sample location. Figure 3 shows the
histogram of sample grades in the MQ and MV alteration types
in Zone 96. Typical of gold, the histogram is strongly positively
skewed with a coefficient of variation in excess of six and a
maximum grade exceeding 7000 grams per tonne.
ORE DEFINITION IN UNDERGROUND MINES

The problem of ore definition in underground mines, particularly
gold mines, can be significantly more difficult when compared to
the problem of ore definition in open pit mines. Ultimately
however, the solutions to these problems are based on the same
spatial modelling methods and economic concepts.
There are several reasons for the greater difficulty. Higher
underground mining costs generally increase the cut-off grade at
which ore is defined. Limited access to the mineralisation to be
mined and the high-cost of underground development are not
always compatible with highly selective mining practices. Ore
definition drilling in underground mines, most commonly
diamond drilling in fans, is usually carried out from sites located
some distance from the mineralisation. Consequently, unit
sampling costs are higher and usually smaller numbers of
samples are available for ore definition compared to open pit ore
definition.
In contrast to the open pit environment however, geological
mapping can often play a more important role in ore definition in
the underground mine. Diamond drill core provides a better basis
for lithological and structural mapping at the larger scale while
mapping and sampling of the backs and faces of drives provide
more detailed information strongly focused on the main areas of
207
N SCHOFIELD
FIG 1 - Schematic cross-section at 54 900 mN through the Henty deposit and host geology showing the Henty fault and the distribution of the main
alteration types (after De Mark and Callaghan, 1997).
208
ORE DEFINITION AT THE HENTY GOLD MINE TASMANIA
FIG 2 - Plan showing the distribution of alteration types and gold grades in drill hole and underground sampling, 2142RL.
209
N SCHOFIELD
FIG 3 - Histogram of gold grades in drill hole and underground samples in the MV and MQ alteration types.
ore bearing mineralisation. The quality of assay information

generated from face and back sampling is usually inferior to that
generated from diamond drill core.
The greater uncertainty in locating the ore that is introduced by
wider drill hole spacing, inherently clustered underground
sampling and higher cut-off grades, can be mitigated to some
extent by better quality sampling from core and the potential for
better geological modelling, when compared to the open pit
environment.
At the Henty Gold mine, ore is defined with the help of
conditional simulation models of the alteration geometry and the
gold grades within each alteration type. The spatial distribution
of the three main gold bearing alteration types, MQ, MV and MZ
are generated using the method of Probability Field simulation
(Froidevaux, 1992). These models use both the geological
logging of alteration types from the drill core as well as the
underground mapping of the drives. The distribution of gold
grades within each alteration type is simulated using the
Sequential Gaussian Simulation method (Gomez-Hernandez and
Journel, 1992). Inclusion of the underground mapping and
assaying information is optional.
An optimisation algorithm similar to that described by
Schofield and Rolley, 1997 (the MP Method) and Glacken, 1996
is applied to a set of multiple simulations to provide the mine
geologists with an optimal definition of ore prior to final
geological interpretation and stope design.
Reproduction of the histogram is important because it provides

some guarantee that a model will be globally and conditionally
unbiased.
Reproduction of the spatial continuity is also important
because it provides some guarantee that the way in which high
and low grades are grouped together in the sample data will be
reproduced in the model. This property is integral to the
definition of ore.
Similar properties apply to the generation of conditional
simulations which represent the distribution of geological
variables such as the distribution of MV, MQ and MZ in the
Henty deposit. Indicator simulations and Pfield simulations
generate MV, MQ and MZ at locations in the model in
proportions similar to those present in the data. The continuity of
each alteration type, represented by the indicator variograms is
also represented in the indicator simulation.
Presently, conditional simulation methods for generating
models of distribution of alteration types and grades are the only
methods which can consistently reproduce these fundamental
properties of the sample data.
The advantage of the additional optimisation procedure in
assisting the definition of ore is that it introduces the local
variable costs and revenues into the ore definition in a better way
than simply applying a previously defined cut-off grade to the
estimate of grade in a block in a deposit.
WHY CONDITIONAL SIMULATION AND

OPTIMISATION?
MODELLING THE SPATIAL GEOMETRY OF THE

ALTERATION TYPES
Conditional simulation methods for building models of grade in

mineral deposits are designed to generate models that reasonably
reproduce certain average properties of the sample grades,
specifically the histogram of the grades and the spatial continuity
of the grades.
The spatial distribution of alteration types such as MQ, MV and

MZ may be simulated at an appropriate scale using a number of
conditional simulation methods such as Sequential Indicator
Simulation (SIS) and Simulated Annealing (Deutsch and Journel,
1992). Both of these methods are presently too slow for practical
210
implementation in a producing mine with current computing

capabilities. Another approach which provides reasonable
indicator simulations in a fraction of the time taken for both of
these methods is Probability Field Simulation (Pfield). This is the
approach currently taken at Henty to generate models of the
distribution of alteration types in the mine sequence.
Figure 4 shows simulations of the distribution of alteration
types in small blocks within Zone 96 of the Henty deposit. The
traces of drill holes in close proximity to the level are also
plotted showing the sample alteration type. The figure also shows
directional indicator variograms describing the continuity of the
MZ to MV, MQ transition and the MZ, MV to MQ transition. As
might be expected, the latter transition is significantly less
continuous than the former.
MODELLING THE GRADE DISTRIBUTIONS IN

EACH ALTERATION TYPE
Conditional simulation of the gold grades in each alteration type
allows the reproduction of the histogram and variogram of the
gold grades in each alteration. In the underground fan drilling at
Henty, the proportion of samples that are MQ averages around
eight per cent and the proportion of samples that are MV
averages around 15 per cent. Consequently, some assumptions
regarding the spatial continuity of gold grades in the MQ must be
made because the amount of information about grade in MQ and
the discontinuous nature of the MQ geometry does not permit
any reasonable calculation of the variogram of gold grade within
the MQ.
The MP modelling at Henty assumes that the spatial continuity

of gold grade in MZ, MV and MQ is similar and that the
differences between the gold grade distributions in each
alteration type are entirely reflected in the histogram of sample
gold grades for each type. Each simulation model reasonably
reproduces the histogram of gold grades for each alteration type
as measured from the diamond drill hole sample grades in that
type.
Figure 5 shows a level plan of a conditional simulation of gold
grades with traces of the underground drill holes overlain and the
locations of underground samples shown. Directional variograms
of the normal scores of gold grade used to generate the
conditional simulations are also shown in the figure alongside the
level plan.
The grade control process allows for up to 100 pairs of
conditional simulations of the gold grade distribution and the
alteration type distribution to be generated in the first stage of the
modelling process. Each pair of simulations is designed to
reasonably reproduce the average univariate and spatial
properties of the input sample data and honour the known sample
grades and geological classification.
ORE DEFINITION USING CONDITIONAL

SIMULATIONS OF GRADE
In the traditional practice of resource estimation and grade
control, the economic cut-off grade is applied to an estimate of
the grade of a mineable volume to decide if the volume should be
accepted as ore or rejected as waste. In this approach, the single
FIG 4 - Indicator variograms of the alteration transitions and a level plan of a conditional simulation of alterations types in Zone 96 (easting scale is
exaggerated).
211
N SCHOFIELD
FIG 5 - Variogram of gold grade and a level plan showing a conditional simulation of gold grade.
model of block grades produced by a particular estimation

method (such as kriging or polygonal estimation) is assumed to
provide the best representation of the block grade for the purpose
of deciding ore and waste.
With optimisation approaches like MP, the set of conditional
simulations is used to build an estimator of ore and waste that is
implicitly customised to each individual deposit. There is no
prior assumption that an estimator such as the polygonal
estimator of grade is the best for grade modelling in a particular
deposit. The optimisation method uses the information contained
in all the simulations to decide the economic advantage of
accepting a block of mineralisation as ore or rejecting it. In doing
so, the process takes into account the probability that the block
will have a grade exceeding the cut-off grade. This is part of the
information provided by the conditional simulations.
Figure 6 shows the result of applying the optimisation process
to the set of conditional simulations to help define ore on two
levels in Zone 96.
RECENT RECONCILIATION INFORMATION

The MP grade control system was introduced at Henty in
December 1998 and became operational in early-1999. CIBC
World Markets magazine (July 1999) reported that The
operating performance at Henty showed significant improvement
over the course of the past year, with production ranging from
13239 oz at $434 per oz in the June 1998 quarter to 26199 oz at
$217 per oz in the March 1999 quarter. The improvement was
largely due to the modified mining method being employed
which has allowed mining of the higher grade central portions in
212
Zone 96 whilst minimising dilution and lead to an increased

mining rate.
In the current year to the end of February, some 84 900 tonnes
of reserve have been mined at a predicted grade of 19.6 grams
per tonne for 53 500 oz of contained gold. Allowing for
overbreak (dilution), some 100 300 tonnes were actually mined
at a grade of 17.1 grams per tonne for 55 130 oz of contained
gold. Ounces produced are within three percent of ounces
predicted.
CONCLUSIONS
The Henty Gold mine presents some difficult challenges to ore
definition which are related to the high-cost of mining and the
concentration of most of the gold within a very small proportion
of the mineralised rocks. The application of the conditional
simulation and optimisation method known as MP at Henty
provides tools to allow the mine geologist the ability to model
both the distribution of the main mineralised alteration types and
the gold grade associated with those types. The mine geologist
can choose to include or exclude underground mine sampling
and mapping in the modelling. The method is comparatively fast
and allows re-modelling of large areas of the deposit within
several hours.
The approach has potential for application to other
underground gold mines where the alteration characteristics
associated with mineralisation can be readily mapped and where
underground mine sampling and mapping can play an important
part in ore definition.
FIG 6 - Plan and section plots showing ore defined by simulation and optimisation.
ACKNOWLEDGEMENTS
The author acknowledge the contributions of colleagues at
Hellman and Schofield and the Henty Gold Mine in preparing
and reviewing this paper. He also thanks the conference
reviewers for their thoughtful comments in improving the
presentation of this paper.
REFERENCES
De Mark, P and Callaghan, T, 1997. Geology and Mineralisation of the
Henty Gold Mine, Tasmania, in Proceedings of the Third
International Mining Geology Conference, pp 25-29 (The
Goldfields Limited, 1999. CIBC World Markets Australian Gold Book .
Deutsch, C V and Journel, A G, 1992. GSLIB Geostatistical Software

Library and Users Guide , 340 p (Oxford University Press: New
York).
Froidevaux R 1992. Probability Field Simulation, in Geostatistics Troia
92, Volume 1 (Ed: A Soares) pp 73-84 (Kluwer Academic
Publishers: London).
Glacken, I M, 1996. Change of Support and Use of Economic Parameters
for Block Selection, in Geostatistics Wollongong 96, pp 811-821.
Gomez-Hernandez, J and Journel, A G, 1992. Joint Sequential
Simulation of Multigaussian Fields, in Geostatistics Troia 92,
Volume 1 (Ed: A Soares) pp 85-94, (Kluwer Academic Publishers:
London).
Schofield, N and Rolley, P, 1997. Optimisation of Ore Selection in
Mining: Method and Case Studies, in Proceedings Third
International Mining Geology Conference , pp 93-97 (The
213
Computer-based Resource Estimation in Accordance with the

1999 JORC Code
J H Duke1 and P J Hanna2
ABSTRACT
Following the Busang fiasco, more detailed guidelines were appended to
the latest revision of the JORC Code released in September 1999 (JORC,
1999). Although the definitions and terminology essentially remained
unchanged, the additional information in the guidelines has highlighted a
number of issues for consideration by geologists undertaking
computer-based resource estimations.
The three main issues are:
Data Integrity details such as drilling and logging techniques,
sampling procedures, assay tests, site visits, data recording and
aggregation methods need to be considered.
Geological Interpretation is the Competent Person applying the
appropriate software interpretation to the deposit in accordance with
the regional setting. Are the correct computer-based tools being
applied and are they producing the desired result (without
compromise) to the satisfaction of the Competent Person?
Modelling Technique the importance of the relevant experience of
the Competent Person not only in regard to the geology, but also
with respect to the modelling algorithms and resource calculation
methods of the computer software system involved.
At the end of the day, it is essential that the Competent Person
undertake a reality check on the resource estimation itself. They should
also check that the borehole spacing is in accordance with the level of
continuity for each critical parameter (such as structure, grade,
impurities, etc) for which the resources are categorised.
INTRODUCTION
It has long been accepted practice amongst mine geologists to
provide conservative resource estimates. It is easier, the
geologist argues, to accept that some ore was missed in the
estimate than to admit to mine management and the mining
engineers that the estimate over stated the resource. The mine
geologist does not usually have to wait very long before mining
reveals what is actually in the ground. Additionally, the
accumulated knowledge of the existing mineralisation provides a
solid basis for the predictive resource model on which the
estimate is made.
On an exploration project however, the exploration geologist is
often prone to provide a more optimistic resource estimate.
There is usually a long lead-time to development and hence to
any reconciliation of the resource to mine production.
Consequently the knowledge base for the resource model is very
limited and grows only by further exploration.
Whether the resource estimate is conservative or optimistic,
either way it is not necessarily the best estimate based on the
available data and geological understanding. According to the
JORC Code, what investors and mine management need are
resource estimates which are as accurate as possible so they can
realistically assess the geological risk of the project.
For many years now the resource geologist, whether attached
to an existing mine or assigned to an exploration project, will
carry out resource estimation using specialised resource
modelling software. This implies that the resource geologist not
1.
MAusIMM, MMICA, Principal Geologist Metals,

International Pty Ltd, PO Box 160, Bowral NSW 2576.
2.
FAusIMM, Senior Geologist Coal, ECS International Pty Ltd, PO

Box 160, Bowral NSW 2576.
ECS
only works to understand the deposit but can use the appropriate
software methods to model it accurately. Applying the
appropriate methods involves both having access to those
methods in the available software and being able to use them to
generate a realistic and accurate model.
DATA INTEGRITY
The 1999 JORC Code provides us with guidelines (refer to Table
1 of the JORC Code) which are recognised as industry standard
procedures for resource estimators. The first nine points are
concerned with the integrity of the geological data, such as,
logging, sampling, assaying and data location.
There are two important criteria to consider whilst developing
a geological database for resource estimation, Materiality, and
Transparency. It is essential that the geologist remains focussed
on these criteria when designing and compiling a geological
database.
Geologists should recognise the fact that the database will
evolve with a project as it develops from exploration to
development through to the mining stage. At all stages it is
important that the investors, project managers and the mine
planners are aware of the impact of geology on the viability of
mining the resource.
For these reasons it is essential that geologists know why data
was collected, which data is relevant to the project and the
reliability of the information.
When compiling the database, geologists should ensure
adequate data validation has been conducted. It is the authors
opinion that geologist should not only sign-off on a Resource
Statement, but that they should be prepared to sign-off on the
database itself and that this should become a standard practice
for the industry.
Todays geologists need to fully understand how data can be
manipulated with the software system they are using to store the
data. A database should be all encompassing, integrating all
types of information including lithology, analytical data,
downhole geophysical logs, geotechnical logs, hydrogeological
data, gas data, etc. Every piece of geological information that has
an impact on the project should be retrievable from the database
to be placed on appropriate plans or cross-sections, even if it is
not related to a borehole (Figure 1).
Comments such as micro-faulting, H2S odour emitted from
borehole, and lost water circulation in borehole (possible
breccia/fractured zone), should be stored in an appropriate form
in the database.
It is also important that the database agrees entirely with the
final geological model. Some systems use an extract summary
file of the database to generate a model. Often, a geologist may
edit the extract file to update the model, but neglect to update the
database with the same edits. This is regarded as poor practice, as
it often complicates an independent audit of the resources. The
best method to check the integrity of the database is to plot the
data onto sections and plans of the final model (Figure 2).
The database should well documented, as it will inevitably be
scrutinised by external auditors. The documentation should note,
among other things, the extent of validation, assumptions, any
shortcomings of the data, verification of laboratory results and
listings of the data, such as the borehole names, stratigraphic
sequence, etc.
215
J H DUKE and P J HANNA
FIG 1 - Comments on a conceptual mine plan highlighting geological hazards to mine planners.
Ultimately, the database will be the basis for which the

geologist will communicate:
the level of confidence an investor can bank on from the
Resource estimate reported; and
relevant information to the mine planners who will determine
a reserve estimate based on the geological conditions.
GEOLOGICAL INTERPRETATION
Geological interpretation is a fundamental part of the resource
estimation process. In computer-based resource estimation the
geologists understanding of the structure and grade continuity of
the resource are translated as defined zones or domains with
their unique geometric shape. In the case of a single coal seam
deposit for example, interpretation might involve defining the
floor and thickness of the seam (domain) and a boundary to limit
its extent either at subcrop or at depth. Software would often
need to be able to handle faulting and seam thickness
complexities such as washouts and seam splits.
Nature of data used
FIG 2 - Validation of borehole data against the geological model is

critical, and needs graphical capability in software systems.
216
Geological interpretation for mineral resource estimation is

always based on both the available data and the geologists
knowledge of the deposit. The nature of the relevant data used
for the interpretation often dictates how the geologist conducts
the interpretation and what assumptions are derived about the
geology. For example, in the single seam deposit shown in Figure
3, coal intercepts might be either visually logged by the geologist
from drill core or, interpreted seam intercepts from downhole
geophysical logs in open holes (see Figure 4). It is assumed that
enough cored holes have been logged and compared to
geophysical logging data such that the geophysical signature of
the seam can be accurately recognised.
COMPUTER-BASED RESOURCE ESTIMATION
In the case of a complex gold vein deposit (see Figure 5), the
data used to carry out the geological interpretation is derived
from a variety of sources surface costean sampling,
underground stope channel sampling, reverse circulation
borehole sampling data and diamond borehole sampling data.
Because overall data density is often sparse, all available data is
used in the interpretation and the grade estimation even though
there are differences in sample size and technique.
Alternative interpretations
It is sometimes useful to compare alternative interpretations,
especially in an effort to arrive at the best resource estimate. In
the gold vein example both a 3D block model vein envelope
and a 2D gridded high-grade vein interpretation were compared.
The 3D approach, whilst easier to define, caused problems of
smearing high-grade and low-grade gold estimations. The
definition and correlation of 2D high-grade veins (see Figure 5)
to generate a vein model resulted in a much more accurate and
useful model for mine planning than the block model derived
from the envelope interpretation.
Grade and geological continuity
FIG 3 - Accommodating seam splitting in a single seam interpretation

by splitting seam A in DH2 into three seam splits with a complex
father-son relationship.
The factors that affect grade and geological continuity need to be

understood and this understanding carried through to both the
geological interpretation and modelling process. In modelling a
sedimentary uranium deposit it was found that interpretation of
lode intercepts based on economic grades only, resulted in
some problems with the grade model. For example, as grades
diminished at the margins of the deposit it was important to
include a lower grade intercept at the edge rather than a zero
thickness intercept. The lower grades at the margin of the lode
prevented the higher grades from being extrapolated out to the
margin as it did with the zero thickness intercept (see Figure 6).
FIG 4 - Downhole geophysical data is often used to define splitting and coalescing coal seams.
217
FIG 7 - Block model of a folded, base metal vein-type deposit.

Transformed blocks are used to control grade estimation within the fold
- allowing for local changes in orientation of the search ellipsoidal.
FIG 5 - Section through a complex gold vein deposit showing the

hangingwall and footwall surfaces derived from vein envelope
intercepts interpreted from a variety of exploration data.
metal lode (see Figure 7) use was made of transformed blocks (a

proprietary method called Z-Grid Control see Cram and Duke,
1992) to control grade estimation within the fold.
ESTIMATION AND MODELLING TECHNIQUES

As discussed in some detail in our earlier paper (Duke and
Hanna, 1997) geological interpretation involves the geologist in
using all relevant data to define geological domains which
differentiate the various types of mineralisation in the deposit
and which stand up to the scrutiny of statistical and variogram
analysis. The resource model is then built by filling these
domains with blocks and estimating block grades in such a way
as to accurately reflect the geologists understanding of spatial
distribution within each domain.
It is important that block models are constrained by the
controlling geological elements of the deposit, such as lode
boundaries, structure and spatial distribution of grades and other
attributes. These controls are incorporated into the computer
model using such tools as wireframes, triangulated and gridded
surfaces and grade estimation procedures adapted to emulate
grade distributions.
Choosing the appropriate estimation technique

Having defined the geological domains during the geological
interpretation stage, choosing the appropriate estimation
techniques involves finding ways of modelling the spatial
distribution of grades and other attributes within the domains.
FIG 6 - Alternative geological interpretations of a sedimentary uranium
deposit affect the grade continuity in the resource model.
Using geology to control estimation

The use of geology in guiding and controlling mineral resource
estimation is particularly relevant to folded and faulted deposits.
For example, in modelling a steeply dipping gently folded base
218
Data analysis within domains

A study of the spatial distribution of assay data that lies within
each domain usually increases the geologists understanding of
the nature of mineralisation and provides the basis for
determining grade estimation strategies. This study involves
examination of data spacing, and trends in values displayed on
sections, plans and in 3D, along with statistical and variogram
analysis.
COMPUTER-BASED RESOURCE ESTIMATION
Treatment of extreme grade values
Reconciliation and mine production records
If extreme grade values exist within a single domain, there is a

danger that these values will cause undue influence on local
grade estimation. Strategies to rectify this include cutting high
grades, multiple estimations using different search parameters for
different grade ranges and use of indicator kriging. These
extreme values are often due to smaller scale structures such as
mineralised stockwork veins, which are discontinuous and
impossible to domain separately.
The real test of a resource models validity is how it reconciles

with what is actually mined. In practice however it is often
difficult to provide accurate reconciliation data, as the first step is
to compare the reserve estimate with production so as to
eliminate the differences between predicted and actual mining
losses and dilution.
Interpolation parameters
In compiling a resource estimate, the JORC Code specifies that

the main governing principles are transparency, materiality and
competence. Computer systems, when used correctly, will
provide an abundance of support material to assist the estimator
in classifying the resources into their appropriate categories.
Histogram and cross-correlation graphs, variograms and
confidence models, and area of influence plots are usually quick
and easy to produce by computer-literate geologists today. As we
know with the interpretative nature of geology, it is easy to
produce enough statistical evidence to prove whatever result we
want.
However, as the JORC Code points out in Table 1, at the end
of the day, the geologist should at least review the results to
determine whether they appropriately reflect the Competent
Person(s) view of the deposit. A reality check will assist the
Competent Person in determining whether they believe their
peers would agree with their classification. This is the ultimate
test of a resource estimate, and has given rise to the international
acceptance of the JORC Code itself.
One change of the 1999 JORC Code was the addition of the
diagonal arrows in Figure 1 linking Measured Resources to
Probable Reserves. This has provided the geologist with the
ability to recommend to the reserve estimator that although a
resource is considered as Measured, there may be other factors
that will influence the extraction of the resource thus reducing
the confidence of the reserve estimate.
One such example could be an underground coal resource,
which has been explored to the extent, whereby further drilling
would not alter the estimate of tonnage or coal quality by more
than say ten per cent. Although this would be considered as a
Measured Resource in the coal industry, other factors, such as
roof or floor conditions, gas regime, etc may require further
drilling data to confirm their effect on the mining viability of the
resource. Therefore the geologist should ensure that their report
reflects these concerns that may lead the reserve estimator to
classify the reserves as Probable rather than Proved.
Grade interpolation methods include inverse distance weighting

and kriging. Kriging is a geostatistical method of allocating
optimal weights for local estimation based on the variogram
model. Grade interpolation parameters include such parameters
as scan distance, sample/sector selection criteria and anisotropic
ratios and rotation. Anisotropic control is particularly useful for
modelling deformed deposits (Figure 7). It is particularly
important to use the appropriate set of search strategies for each
domain.
Maximum scan distance

Choosing the maximum scan distance for grade estimation in any
domain involves both geological and practical considerations.
Practically, the maximum scan distance should provide for
sufficient samples in each search ellipsoid in order to generate a
valid block grade in each block. Geologically, the maximum scan
distance is in the direction of greatest grade continuity as
determined by the geologists knowledge of the data.
Block size vs borehole spacing

As a rule of thumb blocks should be no smaller than a quarter
of the average data spacing in each direction. Some grade
estimation techniques, such as indicator kriging, require larger
block sizes. However if regularised blocks are used in folded
domains, such as the complex gold vein shown in Figure 5, then
the blocks will need to be much smaller than one-quarter the
average data spacing in order honour the folded geometry of the
lode.
Model validation The reality check

To validate the resource model there is no substitute for
comparing between the model and the relevant borehole data
plotted systematically on a full set of plans and sections. It is
particularly useful for this checking to be carried out both by the
Competent Person who will sign off on the resource estimate and
a geologist who has good knowledge and geological
understanding of the deposit (if not the same person).
Check estimates
A check estimate is carried out as a way of validating the model
on which the resource estimate is based. It is important that the
check estimate is carried out using an entirely different method
(often a more manual method such as polygonal or sectional
method) so that it provides a valid basis for comparison. In the
case of the complex gold vein deposit it proved to be very
important that the 3D block model of the vein envelope be
checked against the 2D vein model. In fact, it was the 2D vein
model that proved to be the most accurate and useful model.
RESOURCE REPORTING
CONCLUSIONS
The old adage Give six geologists one data set and you will end
up with seven different interpretations reflects the true nature of
our profession. Todays sophisticated geological and mine
planning software systems are designed with great flexibility to
be able to produce those seven different interpretations.
In this day and age, therefore, it is essential that the Competent
Person should be experienced not only in the geological
interpretation of a deposit, but also in the technology behind the
software system used to produce the resource model. This
technology know-how needs to cover two areas:
1.
Previous estimates
Previous estimates must be used as a means of checking and
comparing any resource estimate. Differences need to be
explained satisfactorily which maybe attributed to such factors as
a revised geological interpretation, additional data or differences
in modelling method.
the methodology used to create an accurate resource model

including such tools as database validation, geological
control techniques and extrapolation algorithms; and
2.
the methodology of estimation of resource tonnage and

associated grade.
Without this knowledge, it is quite likely that the computer
will produce the wrong result (garbage in, garbage out).
219
FIG 8 - General relationship between exploration results, Mineral Resources and Ore Reserves in the 1999 JORC Code.
More importantly, with or without this technology know-how,

it is essential that the Competent Person conduct a reality check
before publicly reporting any resource estimate.
REFERENCES
Cram, A A and Duke, J H, 1992. Geological Control in Computer-Based
Resource Estimation, SMEDG/AIG Resource Evaluation
Symposium, Sydney, 1992.
Duke, J H, 1987. Geological Modelling Techniques in Mine Planning, in
Proceedings Pacific Rim Congress 87, pp 851-854 (The
Duke, J H and Hanna, P J, 1997. Geological Interpretation for Resource
Estimation, in Proceedings The Resource Database Towards 2000,
pp 99-109 (The Australasian Institute of Mining and Metallurgy,
Illawarra Branch: Wollongong).
220
Guibal, D, 1997. Variography, A Tool For The Resource Geologist, in

Proceedings The Resource Database Towards 2000, pp 85-91 (The
Australasian Institute of Mining and Metallurgy, Illawarra Branch:
Wollongong).
Hanna, P J and Cameron, J L, 1997. Computer Database and Geological
Modelling of Hunter Valley Geology, in Proceedings Thirty First
Newcastle Symposium on Advances in the Study of the Sydney Basin,
April 1997, Newcastle, NSW, Australia, pp 53-55.
Isaaks, E H and Srivastava, R M, An Introduction to Applied
Geostatistics (Oxford University Press: New York).
JORC, 1999. Australasian Code for Reporting of Mineral Resources and
Ore Reserves (The JORC Code), The Joint Ore Reserves Committee
of The Australasian Institute of Mining and Metallurgy, Australian
Institute of Geoscientists and Minerals Council of Australia.
Grade Control
Mining Grade Control Past, Present and
Future
W J Shaw
223
V M OBrien and T Cutts
229
The Use of Magnesium Oxide and Iron to

Predict Host Units in Grade Control and
Exploration at Bulong Nickel Operation
W J Bollenhagen
239
Mining Grade Control Past, Present and Future

W J Shaw1
ABSTRACT
The objectives of mining grade control are presented and examples of the
techniques used in various open pit and underground mines are used to
define the attributes of good grade control. Reasons are discussed for the
success of various improved practices. The progressive development of
grade control is presented and the balance between quantitative and
qualitative data is analysed. The skills and resources required for grade
control on mine sites that are using best practice are also presented.
A number of problems in grade control have been solved recently
however there are still issues that need to be addressed. Examination of
the grade control process frequently stops after sample quality has been
examined. There is much more beyond this that can now be optimised
using conditional simulation models. Other future directions of mining
grade control are also suggested.
An extensive bibliography of published papers on grade control is
included that will be maintained and updated on a website.
OBJECTIVES OF MINING GRADE CONTROL

Objectives for grade control and for mining geology have been
presented previously: The objective of grade control is to
maximise the value of ore mined and fed to the mill. The
objectives of mining geology are primarily to enable the mining
company to maximise profits by reducing the risk associated
with geological (and other) uncertainty. (Shaw, 1990a). This
issue of risk management will be addressed later as it is now
more amenable to quantitative analysis than it was ten years ago.
In analysing different approaches to grade control it becomes
apparent that there are similarities and differences between
various mining styles, various commodities and between the old
ways and the new. Some comparisons are presented here to
explore what grade control is, how it has developed and where it
is going.
There are always assumptions and unless these are stated at the
outset they can remain hidden. Let us start by ensuring that we
are in all cases talking about good practices; there is little to be
gained by supposing that grade control is now done by people
who are more dedicated, or better trained than was the case in the
past. We will try to identify concrete improvements in techniques
that have contributed to improving grade control.
SAMPLING
In most parts of the world grade control sampling is now done
mechanically, although in the past, and still in some less
developed countries, cheap labour meant that sampling was done
by hand. Very successful grade control of high-grade
underground gold mines relied in the past on panning of gold
samples during mining. It is remarkable today to see remnant
underground workings that now define the high-grade trends in
an open pit mine. Large open pit gold mines currently exist
where artisan miners using hand panning steal gold overnight,
leaving behind trenches in the richest parts of the ore. The
primitive grade control techniques that defined such high-grade
trends are obviously very successful.
All sampling, whether it is manual or by trenching or drilling,
should conform to the principles of sampling defined by Gy
(1979), to ensure that the sample collected is representative.
Elaboration and improvements on this theory by Pitard (1993)
and Francois-Bongaron (1998) have made it more relevant to
1.
FAusIMM, Principal Consultant, Mining and Resource Technology

Pty Ltd, Level 3, Kirin Centre, 15 Ogilvie Road, Mt Pleasant WA
6153.
gold deposits. Good grade control depends on good data.

Application of sampling theory at many mine sites has made a
significant improvement to the quality of that data.
One of the biggest issues that faces open pit mining operations
is the balance between using blast holes and dedicated reverse
circulation (RC) drilling for sampling. It is not easy to balance
the costs and the benefits of these methods, although some new
strategies are further discussed later. Blast hole sampling is
attractive in hard rock mining because the holes are already
available. The disadvantages of blast hole sampling
(contamination and sample loss, leading to poor precision and
bias) are not always easy to quantify and there are few good
published comparative studies. The trend setting operations often
had to rely on subjective trials to justify a change to RC drilling,
but now many successful mines take the view that such dedicated
drilling provides better quality data in a more timely manner,
enhancing their capacity for selective mining. This was certainly
not the case ten years ago when dedicated RC drilling (eg as
discussed in Shaw, 1990b, at the first Mining Geology
Conference) was novel. The quality of RC drilling has improved
significantly with the introduction of face-sampling hammers
rather than crossover subs. Drilling companies have also
responded to pressure to reduce costs while meeting higher
standards of sampling quality. In many cases now the sampling
quality is stipulated in drilling contracts and mine sites routinely
carry out validation tests on sampling equipment.
Trenching techniques have changed from using rip lines to
cutting sampling trenches using rotary saws or chain saws (both
variants of DitchWitch equipment). The smaller consistent
particle size overcomes many of the disadvantages of trenching
and enables samples that have demonstrably better repeatability
to be collected systematically. There is little to be gained now by
comparative test work between rip lines and DitchWitch trench
samples, although very little has been published. Case studies
such as Pratt and Jankowski (1993) provide an insight into the
importance of geological controls in deciding the correct
orientation for grade control sampling.
Underground face sampling continues to rely mainly on
manual methods, ie collecting rock chips using a hammer and
chisel. There are variations on this approach depending on the
orientation of sampling and whether channels are cut or samples
are collected over an area. Alternatives generally rely on sludge
sampling of blast holes, or sampling muck piles of blasted
material or trucked ore. Sampling broken ore often shows poor
representativity and bias due to the particle size variation.
Dedicated drill sampling is generally adopted in large
underground mines once safety factors are seen as an issue.
Where the geometry of the orebody and the mining method can
support it, drilling well in advance of mining enables design of
the ore blocks to be based on true grade control sampling. Any
other sort of sampling is really too late to have much impact on
the grade of the ore blocks and is best described as reconciliation
sampling. Such sampling can only explain why the grade was
different to expectations; it cannot help to control the grade.
In examining the various sampling methods we have defined
some important considerations. Good grade control sampling
must be:
safe,
unbiased,
representative (ie correct in terms of Gys sampling theory),
and
timely (so that the results can usefully define the ore blocks).
223
W J SHAW
ASSAYING
GEOLOGICAL DATA
A similar approach can be taken to examining assaying or indeed

any other part of the grade control process. Assaying is one of
the most easily verified stages in grade control since the use of
Standard Reference Samples and inter-laboratory checks can
identify and control bias, while the submission of field duplicates
and pulp repeats can quantify precision at various stages of
subsampling and assaying. Assaying techniques vary
significantly with different commodities.
The adoption of standard assaying techniques is usually
unassailable. Some operations however opt for partial extractions
or quasi-metallurgical testwork that may emulate some part of
the mining or processing. For example it is not unusual in
bauxite operations to measure the silica that is digested by
caustic soda at a specific pressure and temperature. Some
mineral sands operations define attributes as rock, sand and
slimes at specific sieve sizes. Ferrous metal mines may use wet
screening to emulate scrubbing or the effects of crushing and
sizing in the mill. Such metallurgical assays can all suffer from
the same problem: they assume that the test mimics some part of
the ore treatment or handling process. The problem with such
tests is that the mill configuration can change over time,
rendering the testing technique less relevant. Even standardised
empirical tests (such as the decrepitation index for iron ore)
cannot account for changing practices over time and so need to
be reinterpreted for different conditions.
Of course grade control must be orientated towards predicting
those aspects of the ore processing that affect the value of the ore
and there will always be a need for such information. Effort has
to be put into developing cost-effective techniques that can be
reinterpreted as the mill configuration changes.
One of the big problems facing bulk shipping ores is the issue
of the Lump/Fines ratio and partitioning of grades and penalty
elements into various size fractions. The Lump/Fines ratio of the
ore changes with increased crushing, transport and handling.
While this is hard to predict it is evident that the response of the
ore to such physical treatment may be controlled by the
mineralogy of the ore.
Normative mineral analysis is being adopted for more and
more deposits (Lipton et al, 1999). By assaying for a suite of
elements and then combining these according to theoretical
mineral compositions, the ratios of various minerals can be
estimated. Tying the physical characteristics of ore to such
normative mineral estimates may lead to a better understanding
of such physical attributes as the Lump/Fines ratio.
Problems in ensuring the useful application of consistent
techniques for longer term reconciliations can also arise with
partial extraction assays. Of course care should be taken to
ensure that the digestion techniques have not been changed over
time. However even when consistently applied the use of partial
extractions such as cyanide bottle roll tests (or even aqua regia)
for gold can be misinterpreted when reconciling with the mill.
Again it is best to ensure that such metallurgical assays are
constantly benchmarked.
The use of modern assaying techniques such as the Leco
furnace for S (which can provide assays for other volatile
elements) and ICP assays (which can provide assays for a broad
spectrum of elements) means that in many operations there is
now more information available than in the past. Assaying for
more than one element for grade control will become more
common. This imposes new demands on data management and
interpretation.
Adding to the previous list, good grade control assaying
should be:
based on measurable attributes that may safely be used to
predict metallurgical performance; and
comprehensive so that all the required data is collected.
A grade control system that is not built on a relational database

to manage the assay data is heading for trouble. While this may
be now well recognised, few operations perform well at
collecting and managing soft data (mapping and drill hole
logging). The incorporation of mapping data into mining grade
control is still an important goal for most mines. This will require
accessible 3D-visualisation software built on a relational
database.
Finding the correct balance between soft geological data and
hard assay data has proved difficult for many operations. Where
there is abundant assay data (for example in those operations that
sample blast holes) there is often a tendency to discount
individual assays. If the sampling repeatability is poor, even the
average grades of the ore blocks may be regarded with some
scepticism during mining and the ore blocks may then be
radically altered, reclassified to a lower grade stockpile or even
set aside for resampling after mining. The ultimate reflection of
poor confidence in sampling and assays (the hard data) is a total
reliance on visual control (soft data) with close supervision of the
contacts by mining geologists who make subjective judgements
about whether material is ore or waste.
At the other end of the spectrum is a total reliance on the hard
assay data with a complete disregard of geological controls. The
impact of such a policy has been long appreciated in resource
estimation with a significant move away from the unconstrained
kriging approaches that were prevalent during the 1980s.
However the use of contouring or even kriging to define ore
block boundaries is still evident at some sites. Sometimes there is
such an emphasis on collecting hard assay data and interpreting it
that the mining geologists do not have time to do mapping, let
alone incorporate the results into the ore block designs. This
approach appears most prevalent in operations that use close
spaced trenching or blast holes to sample every mining flitch. It
is often worth investigating whether the geology (soft data) is
being effectively used, especially if assay turn-around times are
critical to ensuring the ore blocks can be designed before the ore
is dug.
A grade control system that disregards the geological controls
will never be optimal. On the other hand, a system that
maximises the use of good quality information in an efficient
manner can release staff away from the tedium of manually
interpreting ore blocks in a subjective manner so that they can
put effort into the mapping that will control future ore blocks.
The balance between hard and soft data is dynamic. Good
assay data provides new insights into geological controls. Good
mapping can significantly reduce grade control costs when the
operation moves away from sampling everything towards
targeting mineralised zones with high quality sampling. The
geological data also provides the controls that enable the best
balance between quality and quantity of sampling to be achieved.
Adding to the criteria for good grade control:
all available data should be stored in a relational database
with 3D coordinates (for integration with mining software);
an appropriate balance should be achieved between hard and
soft data for defining ore blocks; and
the gathering of expensive hard data (assays) should be
guided by the availability of cheaper soft data (mapping).
224
INTERPRETATION OF ORE BLOCKS

The need to convert data to information (and eventually to
knowledge and then wisdom) must not be overlooked. Many
grade control systems get bogged down in data collection,
validation and subjective manual interpretations. These aspects
of the grade control process can be very time consuming and
frustrating, when the data (hard and soft) is being under-utilised.
MINING GRADE CONTROL PAST, PRESENT AND FUTURE
In the past, for open pits the grade control assay data and
mapping were plotted on cross-sections and bench plans for
manual interpretation. The process of defining ore block
boundaries consisted of interpreting major lode directions and
contouring the cut-off grade boundary based on the assays. These
two separate sets of information were then merged and the final
cut-off grade boundary was defined. At this stage the mining
selectivity needed to be considered and so the mineable ore block
boundary was then superimposed, digitised and finally marked
out on the ground. As described, this process involved a number
of iterations and under the pressure of production it was not
always possible to ensure that an audit trail existed that would
validate the final ore blocks. It was apparent that different
practitioners on the same mine site could interpret the same data
quite differently and that there was no easy way to arbitrate or
define the objective truth. This frequently resulted in the
definition of quite different ore block shapes on consecutive
mining benches in open pits.
For underground mining based on drill sampling well in
advance of ore block development, the interpretation problem
can be addressed using computer systems to ensure
three-dimensional integrity of the interpretations. For operations
relying on face sampling of the next development cut, or
sampling stoped ore, the decision as to what will be mined has
already been made and the sampling is of no value in controlling
the grade.
A risk management approach to mining grade control can be
developed based on conditional simulation (Dowd, 1996;
Khosrowshahi and Shaw, 1997; Shaw and Khosrowshahi, 1997;
Schofield and Rolley, 1997). Conditional simulation involves
generating a set of independent possible outcomes (realisations)
on a close spaced grid, that honours the real data (hence the word
conditional) and models the spatial variability of the real data
set. The construction of a robust conditional simulation model
that encompasses a large number of realisations can provide
extensive information about the expected grade and the degree of
variability at closely spaced points. A number of questions
frequently come up that have not perhaps been adequately
addressed in the literature:
Conditional simulation appears to artificially increase the
level of sampling how can this be possible? The approach taken
is to generate a series of new values between the true samples.
These new values should have a similar mean, dispersion and
histogram as the available true data and should produce a similar
variogram. In addition, all the true values are retained and
honoured. Of course the new values are artificial; they are not
real new sample points and we would be foolish to believe them
immediately since we have generated only one possibility (a first
realisation) from the infinite number of alternatives that meet our
target parameters.
If conditional simulation can give us possible realisations of
the small-scale variability - which one of these is correct? No
single realisation is more valid than another. The approach taken
is to use these to define the expected variability at each simulated
point. Every realisation is conditioned to the real data this
means that the real data influences every point. In areas where
there are a lot of high grades you can have a greater expectation
of similar high grades. In areas where the real grades are mostly
waste the effect is the opposite. This is intuitively acceptable.
That is why it is important to deal with a large number of
realisations. The approach to use is thus to generate many
realisations and to use them to indicate the likely risk of grades
being high or low at any point. If we have 50 realisations (or 100,
or 200), the distribution of the values at each point is used to
represent our expectation at that point. If we have modelled the
grade (and any attribute can be modelled) we could determine
the probability that the grade will be above or below a nominated
cut-off grade. From this we can make a decision based on how
we perceive this distribution will affect us. Since the risk is the
inverse of the confidence we can select our risk position

according to our current circumstances.
A risk analysis approach enables the mining geologist to
produce ore blocks that are less subjective, that honour the
nominated cut-off grades, and that allow the risk profile (less
dilution or less ore loss) to be varied for each ore type. The dense
grid of simulated values enables the final ore boundaries to be
defined to a better resolution than the actual grade control
sampling pattern. Even in manual methods this is often done by
the more experienced practitioners who push the ore blocks out a
bit around the high grades near boundaries to be sure they do not
lose any ore, at the expense of taking a bit more dilution.
At operations that find this approach successful the geologists
have been able to devote more attention to mapping and to
incorporate real geological controls on grade.
Is the technique being widely used for grade control mark
outs? No, it is not yet common practice. MRT currently has six
sites using our Ore Block Optimisation (OBO) system, and there
are a number of other practitioners offering systems. There is
little published yet on the approach used or the results. The most
recent summary of the current state of the art was at the
International Symposium on Geostatistical Simulation in Mining
(28 - 29 October, 1999 in Perth) for which the proceedings are
still in press.
OBSERVED IMPROVEMENTS IN
GRADE CONTROL PRACTICES
Ten years ago reverse circulation drilling was rarely used for
grade control but now it is accepted as best practice. The same
can be said for good quality sample preparation: it is now
generally the practice to split 3 kg from the sample at the drill rig
and to pulverise this split using total prep, ie reducing it all to
better than 90 per cent passing 75 microns before subsampling
for assaying.
Other innovations are also gaining in popularity, including
better stockpile management, more comprehensive ore tracking,
audit trails, quality assurance and comprehensive reconciliation
reporting for management.
The reason for the success of various approaches is not hard to
find. The evolutionary process, ie survival of the most fit, is at
work ensuring that less successful practices (and mining
companies) are gradually eliminated through open transparent
competition. In some places the older ways survive, due to
misallocation of resources or protectionism by their adherents,
however in the long run the most successful practices eventually
get adopted and then persist. Even five years ago there was less
consideration given to validating assay quality than there is today
because there are now more practitioners around that understand
the problems that arise if data is dirty.
The mining industry has had a reasonably successful track
record at adopting these many innovations. Both the measurable
benefits and the intangible advantages of conditional simulation
will ensure that risk based definition of ore blocks will be
adopted at more and more operations.
SKILLS AND RESOURCES

Training of mining geologists that supervise grade control is of
major importance. The author has run a grade control workshop
course (Shaw, 1991) with over 450 participants. Many of the
ideas presented are now implemented routinely in grade control
all over the world. Other specialist courses in sampling and in
geostatistical estimation techniques have also made a significant
contribution. There is little evidence yet of the ultimate objective
being reached, that of a specific mining geology discipline with
specialisation in mining grade control. Without doubt this will
happen as more practitioners find that a systematic approach to
225
W J SHAW
grade control makes it a rewarding profession and more mine

managers see the economic benefits of scientific grade control
practices.
The skills required of a mining geologist involved in good
grade control are:
data management and presentation,
geological mapping,
statistical analysis, and
reporting and communication.
A sound appreciation of the principles in a number of
specialist areas is also required:
geostatistics and resource estimation,
sampling theory,
cut-off grade theory,
discounted cash flows (the time value of money),
mining engineering, and
mineral processing and metallurgy.
Resources that enhance the ability of grade control
practitioners to perform well are now commonplace on most
mine sites, but include:
a data audit trail,
a relational database, and
appropriate mining software for plotting and interpretation of
data.
PROBLEMS IN GRADE CONTROL THAT HAVE

BEEN SOLVED RECENTLY
Examination of the grade control process often starts with
sample quality but sometimes the analysis never progresses.
There is much more beyond sampling that can now be optimised
using conditional simulation models.
There is a clear relationship between the sampling density and
costs of sampling. The relationship is less clear between the costs
of sampling and the benefits in terms of reduced ore loss,
dilution and misclassification. Understanding and balancing the
costs and benefits of various strategies is a major issue for many
operations. This relationship can sometimes be defined for
individual deposits by trial sampling or similar programs, or by
conditional simulation. In the former there are often many
parameters and operational factors that cannot be controlled (for
example the deposit can only be finally mined in one way).
Using conditional simulation, many different strategies can be
trialled. Cases have been previously presented (eg Shaw and
Khosrowshahi, 1997) that indicate the application of conditional
simulation to solving the problems of sampling grid optimisation
and ore block optimisation. While there is still much to do in
refining such techniques a sound general approach has been
established.
Thus good grade control must be demonstrably cost-effective
(ie it must show a positive cost-benefit).
FUTURE DIRECTIONS OF MINING GRADE

CONTROL
Supervision of actual mining practices and control of the ore
through various stockpiles will continue to improve. The actual
monitoring of ore movements will become easier as sites adopt
truck-despatch systems based on Differential GPS survey
control. However these systems need to interface with the data
management systems used by grade control if they are to produce
useful information. Real time accessible survey control in all
parts of the mine will also make geological mapping significantly
more effective.
226
In forecasting the future directions for grade control it is still

apparent that there needs to be improvements in the following
areas:
better sampling on drill rigs;
safer sampling underground;
better representativity in gold mines with coarse gold;
continuous survey control;
better integration of mapping and assay data at various
scales;
more integration of mine and mill tracking systems with
grade control predictions; and
better prediction of ore movement during blasting.
CONCLUSIONS
Mining grade control has come of age in the last 15 years with
recognition on many mine sites that it plays an integral part in
the successful economic exploitation of mineral resources. Mines
cannot do without grade control because while geological
boundaries may be sharp in many cases grades are continuous
rather than discrete. Since grade control is by definition about
controlling the grade, the relationship between the discrete
geological controls and the continuous mineralisation controls
must be clearly established. This can be done by:
understanding all the various inputs to good mining;
keeping a balance between hard and soft data;
ensuring that grade control is optimal by continuously
evaluating the costs and the benefits; and
using risk based strategies to support decision-making.
ACKNOWLEDGEMENTS
Many of the ideas presented by the author on mining grade
control over the last ten years have been developed in
conjunction with mine site personnel and grade control course
participants whose support is greatly appreciated. Many
operations develop and solve specific problems and the industry
benefits when such advances are presented in open forums.
Many of the real innovations leading towards a quantitative
approach to mining grade control have been developed with the
support and assistance of Dr Sia Khosrowshahi and my other
colleagues at Mining and Resource Technology Pty Ltd and their
support is gratefully acknowledged.
BIBLIOGRAPHY
This bibliography includes references sourced from many
publications and previous reviews (Bettenay and Shaw, 1990;
Bell and Reeves, 1979). While it is extensive at this stage it is
strongly influenced by the authors personal views. This
bibliography
will
reside
on
a
web-site
at
www.mrtconsulting.com.au and new entries will be added to
provide a comprehensive on-going reference to published work
on mining grade control.
Australian Standards Committee QR/4, 1990. Quality control: guide to
the use of control chart methods including Cusum techniques.
Standards Australia, Standards Australia, Sydney, 59 p.
Ayris, M, 1990. Determining drill hole deviation. Strategies For Grade
Control, Australian Institute of Geoscientists Bulletin,10:37-41.
Barnes, J F H, 1987. Practical methods of drill hole sampling.
Meaningful Sampling in Gold Exploration, Australian Institute of
Geoscientists Bulletin, 7:145-170.
Bell, G D and Reeves, M, 1979. Kriging and geostatistics: a review of the
literature available in English, Proc Australas Inst Min Metall,
269:17-27.
Bettenay, L F and Shaw, W J, 1990. A check list for effective grade
control. Strategies For Grade Control, Australian Institute of
MINING GRADE CONTROL PAST, PRESENT AND FUTURE
Border, S N, 1991. Optimisation of cut-off grades during design of

underground mines, in Proceedings Mining Industry Optimisation
Conference, pp 21-27 (The Australasian Institute of Mining and
Carras, S N, 1987. Gold Sampling - The importance of getting it right.
Meaningful Sampling in Gold Exploration, Australian Institute of
Coles, A J, Hadlow, H R and Levy, I W, 1993. Reconciliations at the
Kundana Gold Mine, WA (Selective blasting of ore and waste, and
its effect). Open Pit to Underground - Making the Transition, AIG
Bulletin, 14:95-102.
Collings, M, Lenartowicz, D and Alexander, K R, 1989. Dominion
Mining Limited - Cosmo Howley Gold Mine, in Proceedings World
Gold 89: Gold forum on technology and practices, Proceedings of
the First Joint International Meeting between SME and The
AusIMM, Reno, Nevada, November 5-8 (Eds: R B Bhappu and R J
Harden).
Cook, J N, 1971. Grade control procedures at the Iron Prince Quarry,
Whyalla, SA, in Proceedings 1971 Adelaide Regional Conference,
paper No 4 (The Australasian Institute of Mining and Metallurgy:
Melbourne).
Cook, J N and McIntyre, R I, 1973. Use of computers for grade
estimation, iron ore quarries, South Australia, BHP Tech Bulletin,
17(2):28-33.
David, M, 1977. Geostatistical ore reserve estimation. Developments in
Geomathematics, 2, 364 p (Elsevier: Amsterdam).
David, M, Dowd, P and Korobov, S, 1973. Forecasting departure from
planning in open pit design and grade control, in Proceedings 10th
International APCOM Symposium, 1972, Colorado 1974, pp
131-153.
Davis, B M, Trimble, J and McClure, D, 1989. Grade control and ore
selection practices at the Colosseum gold mine, Mining Engineering,
pp 827-830.
Dowd, P A, 1996. The use of conditional simulation in grade control, in
Proceedings Mining Geostatistics Conference, Kruger National Park,
South Africa, 19-22 September 1994, pp 11-25 (Geostatistical
Association of South Africa).
Dudley, R J, 1988. Contribution of grade control costs to production costs
in open pit gold mining. Economic Geology Research Unit,
Contribution 18, Geology Dept, James Cook University, Queensland.
Eyde, T H and Eyde, D T, 1984. Application of geology to problems of
sampling and grade control unique to industrial mineral deposits.
1984 fall meeting of the Society of Mining Engineers of AIME,
Denver, CO (Ed: R Metz), Soc Min Eng Am Inst Min Metall and Pet
Eng, pp 83-89.
Fietz, G C, 1993. Short term mine management systems - an integrated
approach to blast design, grade control, stockpile management and
reconciliation, in Proceedings International Mining Geology
Francois-Bongaron, D, 1998. Gys Formula - Conclusion of a New
Phase of Research, AIG Bulletin, 22:1-10.
Gill, J E, Blais, R A and Haw, V A (Eds), 1968. Ore Reserve Estimation
and Grade Control, in Proceedings of a Symposium held in Quebec,
September 25-27, 1968: CIM Special Volume No 9.
Gillies, A D S, 1992. Application of geostatistics to construct a coal
quality model, Proc Australas Inst Min Metall, pp 1-8.
Guarascio, M and Raspa, G, 1974. Valuation and production optimization
of a metal mine, in Proceedings 12th International APCOM
Symposium, Colorado 1974, Colorado 1974, F50-64.
Guibal, D G, 1987. Procedure and example of grade control in open pit
mining, in Proceedings Pacific Rim Congress 87, pp 855-857 (The
Gy, P, 1979. The sampling of particulate materials: theory and practice.
Developments in Geomathematics, 4, 431 p (Elsevier: Netherlands).
Hazen, S W, 1967. Assigning an area of influence for an assay obtained
in mine sampling. US Bureau Mines, Rep Invest, US Bureau of
Mines, 6955.
Hazen, S W, 1967. Some statistical techniques for analysing mine deposit
sample and assay data, US Bureau of Mines Bulletin, 621.
JORC, 1999. Australasian code for reporting of Mineral Resources and
Ore Reserves (The JORC Code), prepared by the Joint Ore Reserves
Committee of the Australasian Institute of Mining and Metallurgy,
Australia (JORC), effective September, 16 p.
Journel, A G, 1974. Grades fluctuation at various scales of mine output,

in Proceedings 12th International APCOM Symposium, Colorado
1974, Colorado, F78-94.
Journel, A G and Sans, H, 1974. Ore grade control in sub horizontal
deposits, Transactions of Institution of Mining and Metallurgy,
83(812):A79-A84.
Kavanagh, M E, 1987. Comparison of RAB, RC percussion and grade
control drilling Gabanintha gold mine, Western Australia.
Meaningful sampling in gold exploration, Perth, 26 October,
Australian Institute of Geoscientists Bulletin, 7:171-180.
Keough, D C and Hunt, S J, 1990. Statistical determination of a safe
sampling protocol: case studies at Granny Smith and Big Bell,
Western Australia. Strategies for Grade Control, Australian Institute
of Geoscientists Bulletin, 10:31-36.
Khosrowshahi, S and Shaw, W J, 1997. Conditional simulation for
resource characterisation and grade control - principles and practice,
in Proceedings World Gold 97, pp 275-282 (The Australasian
Krige, D G, 1951. A statistical approach to some basic valuations
problems on the Witwatersrand, J Chem Metall Min Soc South
Africa, 52(6):119-139.
Krige, D G, 1959. A study of the relationship between development
values and recovery grades on the South African Goldfields, J S Afr
Inst Min Metall South Africa, 59:317-331.
Krige, D G, 1961. Developments in the valuation of gold mining
properties from bore hole results, in Proceedings 7th Comm Min and
Metall Congress South Africa, pp 537-561.
Krige, D G, 1964. A review of the impact of statistics on mine valuation
in the gold mining industry, J S Afr Inst Min Metall, 64:373-380.
Krige, D G, 1970. The development of statistical models for gold ore
valuations in South Africa, in Proceedings 6th International Mining
Congress, Madrid.
Krige, D G, 1970. The role of mathematical statistics in improved ore
valuation technique in South African gold mines, in Topics in
Mathematical Geology, A Special Research Report (Eds: M A
Romanova and O V Sarmanov), pp 243-261 (Consultants Bureau:
New York).
Krige, D G, 1977. Applications of statistics and geostatistics in ore
valuation in South Africa, 109 p (Australian Mineral Foundation:
Adelaide).
Krige, D G and Munro, A H, 1968. A review of some conceptual and
practical implications of the use of valuation surfaces for gold ore
reserve estimation, in Ore reserve estimation and grade control,
Special Publication, 9, pp 33-40 (The Canadian Institute of Mining
and Metallurgy).
Lacy, W C, 1985. Mining geology; the industrys hope, in Applied mining
geology; problems of sampling and grade control. Battle Mt Gold
Co, United States, (Ed: R Metz) pp 1-6.
Lacy, W C, (Ed), 1983. Mining Geology. Benchmark Papers in Geology,
69, 466 p (Hutchinson Ross: USA).
Lipton, I T, Shaw, W J and Waltho, A E, 1999. Characterisation of ore
types and beneficiation behaviour using normative minerals, in
Nami, M and Verrezen, L, 1986. The use of sedimentology to improve
valuation in gold mines, Gold Mining Technology, Johannesburg,
1:15-23.
Newton, M J, 1974. The application of geostatistics to mine sampling
patterns, in Proceedings 11th International APCOM Symposium,
Colorado 1974, Arizona, D44-58.
OLeary, J O, 1979. Ore reserve estimation methods and grade control at
the Scully Mine,Canada - an integrated geological/geostatistical
approach, Mining Magazine, pp 300-314.
Peele, R, 1944. Mining Engineers Handbook, 3rd Edition in 2 Volumes
(John Wiley and Sons: New York).
Pitard, F F, 1993. Pierre Gys sampling theory and sampling practice,
2nd Edition, 1, 488 p (RC Press: Florida).
Pratt, J D R and Jankowski, P, 1993. The geology and grade control at
Bannockburn gold mine Leonora District, WA, in Proceedings
Robertson, I G and Sehic, O A, 1993. Ore quality control for process
optimisation, in Proceedings International Mining Geology
227
W J SHAW
Royle, A G, 1979. Estimating small blocks of ore: how to do it with

confidence,World Mining, April, pp 55-57.
Royle, A G, 1980. Geostatistics, 168 p (McGraw-Hill, Inc: New York).
Schofield, N and Rolley, P, 1997. Optimisation of Ore Selection in
Mining: Method and Case Studies, in Proceedings Third
Shaw, W J, 1990a. Management of grade control and mining geology an overview, Strategies for Grade Control, Australian Institute of
Shaw, W J, 1990b. Grade control for gold mining at Mt Magnet, Reedy,
Rothsay and Youanmi in the Murchison District, WA, in Proceedings
Mine Geologists Conference, pp 83-92 (The Australasian Institute of
Shaw, W J, 1991. Grade control course notes: mining geology, grade
control and reconciliation. A short workshop course on grade
control. Mining and Resource Technology Pty Ltd, Perth, WA.
Shaw, W J, 1993. Practical solutions to some grade control problems and
why they work: a few case studies, in Proceedings International
Mining Geology Conference, pp 171-176 (The Australasian Institute
Shaw, W J and Khosrowshahi, S, 1992. Optimising grade control
procedures in large and small open pit mines, in Proceedings Third
Large Open Pit Mining Conference, pp 251-254 (The Australasian
Shaw, W J and Khosrowshahi, S, 1997. Grade control sampling and ore
blocking: optimisation based on conditional simulation, in
Melbourne).
228
Shaw, W J, Khosrowshahi, S, Horton, J and Waltho, A E, 1998.

Predicting and monitoring errors in sampling, sample preparation
and assaying, More Meaningful Sampling in the Mining Industry,
(Eds: B Davis and S E Ho) AIG Bulletin, 22:11-19.
Siddall, G B and Baxter, K G, 1990. Does the mill control the grade?
Strategies for Grade Control, Australian Institute of Geoscientists
Bulletin, 10:77-82.
Stephens, A J and Chapman, G J, 1993. Optimisation of sampling
procedures at the Fimiston open pit Kalgoorlie, in Proceedings
Storrar, C D, 1977. South African Mine Valuation, 472 p (Chamber of
Mines of South Africa: Johannesburg)
Taylor, M J, 1993. Grade control review at Boddington gold mine, SW
region WA, in Proceedings International Mining Geology
Truscott, S J, 1947. Mine Economics, Sampling - Valuation Organisation - Administration, second edition, 366 p (Mining
Publications Ltd: London).
Wright, A, Feyerabend, W C and Kastelic, R L, 1992. Integrity of
samples acquired by deep, reverse-circulation drilling below the
water table at the Chimney Creek project, Nevada (and subsequent
discussion in Jan 1993), Mining Engineering, 44(4):343-351.

V M OBrien1 and T Cutts2
ABSTRACT
Since 1989 Kalgoorlie Consolidated Gold Mines has managed the
Kalgoorlie gold operations of joint venture partners Normandy Mining
Ltd and Homestake Gold of Australia Ltd. Current production from the
Fimiston Open Pit is approximately 650 000 ounces per annum. The
mine incorporates a multitude of narrow shear hosted, subvertical lodes,
predominantly within Golden Mile Dolerite.
Traditionally KCGM has used blasthole based grade control in the
Fimiston pits. A single sample is taken from vertical production
blastholes, drilled on ten metre benches. This process was improved with
the addition of infill drilling along blasthole rows to provide additional
samples across strike. However the limitations of adequately defining
subvertical mineralised structures with vertical blastholes remain.
In 1997 the first of two trial areas were drilled with close spaced,
inclined Reverse Circulation (RC) drilling. These tests indicated that RC
based grade control had the potential to provide significant advantages.
These include:
improved delineation of mineralised structures, particularly across
strike;
reliable short-term mine schedules based on the advance data
available from drilling over three benches; and
the angled RC drilling is capable of providing samples of near stope
and stope fill material which often remains unsampled by blasthole
drills due to the safety constraints around unstable underground
openings.
Based on the success of this initial work RC based grade control was
gradually implemented across production areas commencing September
1998. Some of the many challenges faced include:
development of effective procedures and systems for the planning,
logging, interpretation and modelling of the RC data;
obtaining adequate drill coverage due to mine scheduling constraints
and restricted access around underground openings;
requirement for integrated systems to handle either blast hole or RC
data; and
development of effective methods to evaluate the performance of RC
grade control.
RC based grade control is delivering significantly improved lode
definition, grade estimation and subsequent ore classification.
INTRODUCTION
Kalgoorlie Consolidated Gold Mines manages the open pit and
underground Kalgoorlie operations owned by 50/50 joint venture
partners Homestake Gold of Australia Ltd and Normandy
Mining Ltd. Approximately 80 per cent of gold production is
sourced from the Fimiston Open Pit located adjacent to the city
of Kalgoorlie-Boulder (Figure 1).
KCGM has traditionally used blast hole based grade control at
the Fimiston Open Pit. In late-1998 a transition to grade control
based on angled RC drilling was implemented after trials
indicated substantial improvements in ore delineation and grade
estimation were achievable.
This study presents the development of systems and practices
during implementation and the many challenges met.
1.
MAusIMM, Project Geologist, KCGM Pty Ltd, Black Street,

Kalgoorlie WA 6430.
2.
Senior Mine Geologist, KCGM Pty Ltd, Black Street, Kalgoorlie

WA 6430.
FIG 1 - Location of the Fimiston operations.
MINE GEOLOGY
The Kalgoorlie gold field of Western Australia is located within
the NNW trending Archean Norseman-Wiluna Greenstone Belt.
The greenstone belt has been successively deformed and
regionally metamorphosed to grades varying from lower
greenschist to amphibolite rank. (Binns et al, 1976 in Clout et al,
1990).
Local geology consists of a 3.3 to 4.0 km thick lower sequence
of ultramafic to mafic volcanic rocks and mafic sills overlain by
a 1.0 km thick volcano-sedimentary sequence. The mafic
gabbroic sills locally termed dolerites are conformable with other
rock units and as such are documented as part of the stratigraphy.
The main host rock for gold mineralisation is the Golden Mile
Dolerite, the largest of the intrusive sills in the district. This is a
differentiated gabbroic sill 600 - 900 m thick subdivided into ten
subunits. (Clout et al, 1990)
The ore deposits which occupy an area of approximately
10 km 2 km adjacent to Kalgoorlie-Boulder occur in two
distinct forms. The Fimiston style lodes of the Golden Mile and
the younger Mt Charlotte type quartz stockworks. The Fimiston
lode system consists of an extensive range of mineralised shear
zones. The lodes are divided by the Golden Mile Fault into the
Eastern and Western Lode groups.
A textbook lode contains the highest gold grades within a
siliceous/pyritic core with grades decreasing with distance from
the central shear. Highest gold grades are also typically
associated with tellurides and alteration minerals with higher
vanadium content. Individual lodes may be up to 1800 metres
long, 1200 metres in vertical extent, and ten metres wide.
Approximately one thousand lodes have been identified at
Fimiston. These are grouped into lodetypes based on orientation.
229
V OBRIEN and T CUTTS
The dominant lodetypes are the steep westerly dipping Main

Lodes and the moderate westerly dipping Caunter Lodes. The
distribution of lodetypes within the central mining area is shown
in Figure 2.
MINING HISTORY
Historic
The Fimiston operations are centred on the Golden Mile area
which has been mined almost continuously since discovery by
prospectors in 1895. A legacy of this historic mine production is
an extensive network of underground workings that extend down
to 1300m below surface. Stoping methods employed on the
Golden Mile by the various operating companies included shrink
stoping, cut and fill stoping and open stoping. The oxide and
sand fill within the older cut and fill stopes is a gold reserve in its
own right. This material typically averages gold grades in excess
of 3.5 g/t. The last of the underground operations were shut down
in early 1994.
Prior to the extensive open pit mining which commenced with
the formation of KCGM in 1989 the Golden Mile had produced
in excess of 38 million ounces at an average grade of ten grams
per tonne.
were mined by various companies during the 1980s but

amalgamation of these pits was not possible until the leases were
brought under common management with the formation of
Kalgoorlie Consolidated Gold Mines.
Amalgamation of all mining and milling activities, subsequent
upgrade of the Fimiston Mill and the removal of infrastructure
constraints has seen the mining rate increase from 25 Mt per
annum in 1989/1990 to the current 65 Mt. Further rationalisation
of mining operations commenced in September 1999 with the
transfer from contract based mining to owner mining operations.
Drilling, blasting and excavation is carried out on 10 m
benches. The mining fleet consists of three Komatsu PC8000
hydraulic shovels with a bucket capacity of 60 tonnes and a fleet
of 18 Caterpillar 793 trucks.
Fimistons current annual ore production target is
approximately 10 Mt at a head grade of 2.3 g/t. This ore is
refractory in nature and is all processed through the Fimiston
Mill and the Gidji Roaster. Since 1989 the Fimiston Open Pit has
contributed 80 per cent of KCGMs total production of 7.2 Moz.
Based on current reserves the open pit has a mine life of 15
years. The dimensions of the ultimate pit design are
approximately 4.3 km along strike by 1.5 km wide and 610 m
deep at its lowest level
Remnant mining
Modern
Open pit mining in the Fimiston area commenced in 1983 with
North Kalgurli Mines Limiteds Kemlo Pit followed shortly after
with the Croesus and Eclipse Pits. Numerous small-scale pits
Open pit mining is complicated by the extensive historic

underground workings within the pit which present a hazard to
both personnel and equipment. The majority of the workings are
stopes less than 2 m wide, however large open stopes (up to 50 m
FIG 2 - Plan of interpreted lodes at -190 mRL with the current pit outline.
230
EVOLUTION OF GRADE CONTROL AT KCGM
Safety Exclusion Zone

25
0
25
Metres
19 100m E
LEGEND
Millfeed & Subgrade
19 000m E
E-W by 150 m N-S by 120 m RL ) are of particular concern.

These larger stopes were predominantly mined in the later years
of underground mining using long hole stoping methods.
Although most historic workings were well recorded, the
deterioration of openings over time has required the development
of comprehensive systems to ensure safe mining through these
areas. Where underground workings are determined to be a
potential hazard the working bench is clearly flagged with tape to
delineate the area. No personnel or machinery may cross safety
exclusion zones which are suspected void areas deemed to have
insufficient cover.
The extensive network of safety exclusion zones in the pit
restrict access for drill rigs and prevent complete sample
coverage for grade control. Drill rigs are not permitted to sit
within these areas and any drill hole which intersects
underground workings within the first ten metres must be
abandoned. The exclusion zones are prevalent within ore zones
as seen in Figure 3.
50
48 000m N
48 000m N
47 900m N
47 900m N
47 800m N
47 800m N
1997 drilling and sampling of infill holes along blast hole rows
was implemented in known ore zones to aid in lode delineation
and grade estimation.
Interpretation of ore zones utilising a 0.3 g/t cut-off boundary
is undertaken for each blast pattern. Historic information
including underground level plans and mapping, in addition to
bench above mining blocks is used to guide the interpretation.
Grade estimation is performed using the multiple indicator
kriging method. Mining blocks are then designed using this
model with mining selectivity in mind. The tonnage and average
grade of each block is calculated and a destination applied based
on cut-off grade criteria. Current block cut-off grades are:
0 - 0.3 g/t
Waste
0.3 - 0.9 g/t
Subgrade
0.9 - 1.2 g/t
Marginal
1.2 - 2.2 g/t
Low-grade mill feed
2.2 g/t
High-grade mill feed
Prior to blasting mining blocks are delineated on each shot

using flagging tape weighted with rocks, with the blast hole
collars used as reference points. This preblast delineation is
used as a guide to the marking up of mining blocks on the
blasted shot. Shot movement is generally minimal and is
reflected by the position of the preblast tape.
After a 12-hour settling period the blast is cleared and Survey
tape out the exclusion zones. After watering down and visually
inspecting the potential ore zones blocks are marked out using
taped peg lines and coloured flagging to delineate destinations.
Block boundaries are adjusted along lode contacts where
required.
Unless access constraints dictate otherwise ore and waste
contacts are excavated across strike from the hangingwall side.
This is done to utilise the natural rill angle of the blasted rock
which corresponds with the dip angle of most lodes (60 to 80)
(Figure 4). The mining of contacts is supervised by geologists as
required.
Limitations of the existing system
47 600m N
19 100m E
47 600m N
19 000m E
47 700m N
18 900m E
47 700m N
FIG 3 - Plan of the -250 metre bench in the Western Lodes showing the
extent of the safety exclusion zones.
GRADE CONTROL PRACTICE

Blast hole based grade control
The blast hole based grade control system at Fimiston has
evolved over time. A single blast hole sample is taken from each
blast hole drilled on a ten metre bench. The assay information is
supplemented with logging of blast hole cuttings and mapping of
blast faces. The standard blast hole pattern is 5.2 6.0 metres. In
Blast hole based grade control has obvious limitations when

sampling subvertical mineralised structures. There is poor
sample resolution across the mineralised plane, potential for
downhole sample contamination and the inability to sample
within the safety exclusion zones.
The most critical of these is the safety constraints which
prevent collaring of blast holes near underground workings,
precluding the sampling of near stope and stope fill material. In
areas of extensive stoping the subsequent broad safety exclusion
zones often prevent the sampling of any mineralised material.
The delineation of mill feed blocks in such areas is therefore
highly subjective. Even where the mineralised margins of a lode
are sampled the inability to sample potentially high-grade
material adjacent to stope margins has a significant impact on
grade estimation. Areas unable to be sampled by blast holes can
be large and mostly occur within mill feed blocks or near mill
feed/subgrade or mill feed/waste contacts. Geologists have to
determine a suitable destination for the unsampled material and
assign a grade without adequate samples. These decisions are
further hampered by restricted access for geologists to visually
inspect this material on broken shots.
As blast holes cannot adequately sample stope fill material or
define stope locations, modelling of stopes is not possible.
Therefore voids and stope fill are treated as if they didnt exist.
These areas are assigned the same density as the in situ ore and
are not distinguished from in situ material during grade
estimation, resulting in local overestimation of grade and
tonnage.
231
19 800m E
19 750m E
19 700m E
48 250m N
48 250m N
48 200m N
48 200m N
48 150m N
48 150m N
LEGEND
High Grade Millfeed
Low Grade Millfeed
19 700m E
Subgrade Stockpile
19 650m E
Blast hole based grade control at Fimiston suffered from a

system which focussed on individual blasts rather than
geological domains. Blast hole assays became available for each
single blast shortly before being fired. As such the interpretation
of face mapping, blast hole cuttings and assay grades for each
blast could not incorporate assays from any neighbouring blasts
which hadnt been fired. This often produced poor mark-up
correlation from one blast to the next as seen in Figure 5.
Blast holes have historically produced variograms with
extremely high nugget effect values, suggesting poor correlation
between samples flagged within ore interpretations. When
variograms with high nugget effect values are used in the grade
interpolation of areas containing irregular high-grade samples,
the resultant model grades tend to be oversmoothed and
overestimated.
The problem of grade overestimation is evident in
reconciliations of previous grade control models. Reconciliations
between the grade control estimate and bullion calculated grade
for the 12 month period to December 1998 are within one per
cent. On face value this shows a remarkable correlation between
predicted and actual grade, however evaluation of the grade
control models during this period revealed that approximately ten
per cent of all mark-up blocks had a corrected grade assigned.
Corrected grades are visually estimated by production geologists
when the modelled grade doesnt reflect what is seen on the
broken blast. The large majority of corrected grades are lower
than the estimated grade.
19 650m E
FIG 4 - Schematic of mining across strike to minimise dilution.
25
25
50
Metres
RC based grade control
FIG 5 - Plan of the -130 metre bench in the Eastern Lodes highlighting the
poor correlation of mark-up blocks across blast boundaries.
Acknowledging the deficiencies of the existing blast hole based

grade control system, a review of potential improvements was
carried out in late-1997. Grade control utilising inclined RC drill
holes appeared capable of addressing many of the problems

associated with blast holes. These benefits included:
232
coverage within safety exclusion zones, allowing accurate

delineation of mineralisation contacts, improved grade
estimation and confidence in the mark-up blocks.
The ability of RC drilling to accurately define stoped areas
and sample stope fill material would allow modelling of the
stopes. Stope modelling incorporating independent
estimation of stope fill grades and the application of realistic
densities to stoped areas is a large advance on the blast hole
system.
By drilling multiple benches in a single pass RC based grade
control would provide sample data in advance. Ore
interpretations could be conducted on a larger scale and in
3D as opposed to the individual blast focus of blast hole
based grade control.
The advance data obtained would be used for the short-term
scheduling of production. This would give certainty to the
scheduling of production equipment and mill feed with the
associated cost benefits.
The obvious challenges associated with implementing RC
based grade control would be to achieve adequate sample
coverage and develop new systems to manage and interpret the
RC data. It would require the geology department to manage the
drilling programs and integrate these into the mining cycle.
18 900m E
Inclined RC drilling would provide improved sample
18 950m E
47 800m N
47 800m N
47 750m N
47 750m N
LEGEND
47 700m N
Stope
RC Ore
10
20
Metres
19 500m E
20 000m E
FIG 6 - Plan of the RC trial area on the -190 metre bench in the Western
Lodes highlighting the dilution incorporated using blasthole based grade
control.
19 000m E
In late-1997 a RC drilling program was conducted on a small test

area (75 m by 150 m over three benches) targeting 2 Lode and
3 Lode within the Western Lode system. RC drill holes were
drilled at a dip of 60 towards the east on 10 m E-W sections.
Drill hole spacing along each section line was 6 m and samples
were taken at one metre intervals. Mineralisation and structural
models were developed on-site and Snowden Associates Pty Ltd
were retained to analyse the drill hole data geostatistically and
perform grade estimation using a range of methods.
The RC sample data revealed a more complex mineralisation
system than that anticipated by the mine geologists who had
analysed the blast hole sampling on benches located above the
test area. The results of the trial were favorable and led to
another small trial targeting 4 Lode in the Western Lodes area.
Interpretation and modelling of these trial areas indicated that
excessive external and internal dilution was being incorporated
using blast hole sampling as shown in Figure 6.
10
18 950m E
Blasthole Ore
Grade control trials
49 000m N
100
49 000m N
100
200
Metres
STO RES
M A IN P IT
O RO Y A
NO RTH
48 500m N
48 500m N
Implementation of RC based grade control

Following the encouraging results of the two trial programs RC
based grade control was gradually implemented across
production areas commencing in September 1998 (Figure 7). In
order to successfully implement the new system a range of issues
had to be addressed to ensure both budget and technical
requirements were achieved. Furthermore the RC grade control
system had to be fully integrated with the existing blast hole
based system during this implementation phase.
Challenges included the development of effective procedures
and systems for the planning, logging, validation, interpretation
and modelling of the RC data. All systems were reviewed during
implementation, and modified as required. Most refinements
were due to the site specific problems caused by the presence of
underground workings while others were undertaken to achieve
time and cost-savings.
O RO Y A
SO U TH
48 000m N
48 000m N
O FFICE
HO RSESHO E
-190mBD
47 500m N
47 500m N
20 000m E
The standard drill hole design was for holes 38 metres deep,
dipping 60 towards 090 giving coverage over three mining
benches. Coverage of 30 vertical metres was preferred as it
19 500m E
19 000m E
-80mBD
Drill pattern and drill hole planning
FIG 7 - Plan of the various production areas at Fimiston.
233
provides ample advance data and requires that production is

seldom disrupted by RC drilling in any one pit. The collar
co-ordinates, the design azimuth and dip and one single shot
survey at the end of hole are used to trace drill hole location.
Given the large area to be covered it was decided that it was
impractical to maintain the drill pattern used in the trials. During
the initial trial it was reported that sixty to eighty percent of the
gold variability occurred at less than ten to 15 metres in the along
strike direction for indicator grades approximating mining
cut-offs. (Snowden Associates, 1998) The drill pattern was
reviewed prior to implementation and changed to 15 m sections
with drill holes spaced at 8 m along section.
A one metre sample interval was used initially however this
was soon increased to two metres. Implementing two metre
samples carried a financial saving from reduced assay costs,
reduced sample handling and made interpretations easier. During
initial interpretations the geologists attempted to model the
mineralisation to match the intricate detail provided by one metre
assays. Considering the ten metre bench height and the mining
equipment used this level of detail was not required.
The initial area covered by RC drilling was the Western Lodes
which hosted the two RC trials. The next area to be drilled was
the Oroya South cutback which is characterised by thin,
relatively discontinuous lodes in the vicinity of the Paringa
Basalt contact. A review of the initial Oroya South drilling
program prompted a change in the standard drill hole design
from 60 to 55 to improve sample coverage on each bench.
The drill pattern is under review with the aim to reduce the
section spacing to ten metres for geologically complex areas and
to rotate the drill pattern 20 in Oroya South so that drilling
intersects the dominant NNW striking Caunter Lodes at right
angles.
Drill hole planning is conducted on screen using Datamine
software. Owing to the presence of the safety exclusion zones at
Fimiston the planning of a RC drilling program is more complex
than simply fitting a grid of standard holes on the pit floor. The
geologist designs a grid of standard holes which are then
reviewed in conjunction with the safety exclusion zone
perimeters for the bench and the sectional ore and stope
interpretations based on Resource drilling.
Drill holes which cannot be collared due to the presence of
safety exclusion zones are deleted and replaced with drill hole
fans from the margins of the safety exclusion zones. The
resultant drill pattern is often far removed from a standard grid.
This variability in drill hole design is necessary in order to
achieve representative sampling of ore zones.
The area below designed holes is reviewed and areas where
safety exclusion zones are expected to restrict drilling are
identified. The drill holes are then extended to obtain sample
coverage in these areas. As seen in Figure 8 the combination of
holes drilled from multiple bench levels can provide complete
sample coverage across a bench.
Drill hole depths are restricted to less than 70 metres, which
covers up to six benches. Drill hole deviation in drill holes less
than 70 metres deep has traditionally been minor and single shot
surveys were considered adequate. Holes greater than 40 metres
have single shot surveys taken every 25 metres.
The RC drill hole planning process has to be flexible to
account for changes in access for drilling due to changes in
safety exclusion zones or changes in production schedules.
Another obstacle to obtaining complete sample coverage is the
difficulty in successfully penetrating multiple stopes. In a typical
drill program only 70 per cent of holes designed through stopes
successfully reach target depth. To maximise sample coverage
the drill hole planning system involves a continuous process of
review and infill drilling.
234
FIG 8 - Section at 48265 N showing that a range of drill hole orientations

are required to achieve adequate sample coverage.
Drilling programs
The initial drilling contract called for 90 000 metres per annum
with payment based on a flat rate per metre for all drilling
activities. In addition a worktime rate is applicable for recovering
lost equipment and hole conditioning around stopes. The contract
calls for mobilisation of between one rig on a single shift to two
rigs on double shift. This flexibility is required due to mine
scheduling and access constraints within production areas.
To-date drilling requirements have been met using one
dedicated track mounted RC rig using a carousel system for rod
changeouts. Backup is provided by standard wheel mounted rigs
which are utilised as required.
Before drilling can commence each drill hole site must be
approved by the geotechnical department to ensure sufficient
advance probing in an area has been completed and hazardous
areas flagged off. Drilling programs are managed by the mine
geologist assigned to coordinating all grade control work within
a particular cutback.
Drill schedules are updated weekly and tie in with the weekly
production plan. Orderly completion of drill programs requires
continual liaison between all departments including survey, drill
and blast, mining and geotechnical.
A review in April 1999 saw the annual drilling budget revised
up to 160 000 metres per annum. This is required to ensure
complete RC coverage in all cutbacks.
Data processing
Successfully managing and processing the enormous amount of
grade control data generated is a formidable task. A major
undertaking was to devise a simplified logging system which
would allow documentation of the significant geological features
of each logged interval while minimising the time spent
collecting and processing this data.
After a review of available logging hardware and software it

was decided to develop a simplified site based logging system.
A4 logging sheets were developed which recorded lithology,
oxidation state and two alteration codes for each interval. The
sheets are scanned using Remark software with manual entry of
the drill hole name, final depth and sample numbers. This system
allows fast logging of drill cuttings in the pit followed by
minimal data entry work, thereby providing large time savings.
Pederson and Reardon (1999) outline the development of this
system.
This was supplemented by the development of a simple drill
hole validation system which summarised the hole depths,
number of samples, logged intervals and surveys. Final drill hole
validations are carried out on-screen using Datamine software.
Interpretation
RC based grade control required the development of a new
interpretation and modelling process which could be
incorporated into the current system. Interpretation of ore, stopes
and lithology boundaries are carried out on screen using
Datamine with initial interpretations conducted on 15 metre
sections followed by plan interpretations every ten metres. Ore
perimeters are defined by a nominal 0.3 g/t assay boundary with
a five metre minimum interpretation width enforced for plan ore
perimeters.
Every known lode has a unique, numeric Lodename value
which is assigned to the corresponding ore perimeters. The
Lodetype field is used to record the mineralisation style and
geometry of each lode.
The interpretation system links to the KCGM Oracle Database
to provide access to the existing Resource Definition lode and
stope interpretations, structural mapping data, underground level
workings and geotechnical data. Resource Definition drill holes
are used to supplement RC grade control drilling during
interpretation and modelling.
Modelling and grade estimation

Modelling of the grade control interpretations is performed using
a menu driven macro with Datamine software. Perimeter
projection of the plan interpretations is used for block and
sample selection.
Standard ore and stope cell dimensions are two metres E-W,
by five metres N-S and ten metres vertically. All Lodetypes
except Cross Lodes are subcelled to one metre in the E-W
orientation to improve representation of narrow ore zones in the
model. Cross Lodes are subcelled to 2.5 metres in the N-S
orientation.
Any one metre drill hole samples are composited to two metre
intervals prior to estimation. Topcuts applied to sample data are
based on statistical evaluation of each Lodetype within the
different geological domains. Most Lodetypes in the Western
Lodes have 25 g/t topcuts applied whereas the dominant
Lodetypes in the Eastern Lodes have topcuts of about 20 g/t.
Indicator variograms produced from the RC data had the lowest
nugget effect values seen at Fimiston, in the order of 30 per cent
as seen in Figures 9 and 10.
Grade estimation is conducted using either Ordinary Kriging
(OK) or Inverse Distance (ID) methods. Each lode is identified
by its Lodename value and interpolated separately. Estimation of
the ore is performed using OK for all lodes except those that are
poorly sampled. Search ellipse ranges used during the first pass
of estimation of Main and Caunter Lodes are 20 20 6 metres
(dip strike cross dip). The second and third estimation passes
employ double and triple the search ranges respectively. The
third estimation pass is based on Lodetype rather than
Lodename. Any model cells not estimated during the first two
passes may be estimated using samples from nearby lodes with
the same Lodetype. This ensures all ore cells have an estimated
grade.
FIG 9 - Downhole median Indicator variogram for Main Lodes in the Western Lodes Group.
235
FIG 10 - Along strike median Indicator variogram for Main Lodes in the Western Lodes Group.
Stope fill material is estimated independently using the Inverse

Distance Cubed method. A single estimation pass is conducted
using a 40 metre search range, with any cells remaining
unestimated given a conservative default grade. The ore and
stope models are combined and density values applied according
to lithology and any previous stoping.
Modelling can be conducted for entire mining benches or
individual blasts and are then incorporated into the standard
grade control computer system used for blast hole based grade
control. The ore/waste blocking, preblast and markout procedure
is the same as used for the blast hole based grade control.
BENEFITS TO-DATE
The primary benefit of the RC system is improved delineation of
lode contacts and confidence in grade estimation. The mine
geologists can better correlate the ore block design based on the
RC drilling with block inspection during markup. This has
virtually eliminated the need to arbitrarily assign grade estimates
due to poor confidence in the model estimates. This equates to
improved block classification and more importantly reduces
misclassification of material type. Prior to July 1999 there were
long periods of significant overcall in the mine predicted grades.
Since this date the proportion of mill feed delineated using RC
grade control has increased significantly. This has resulted in a
dramatic reduction in the grade variance.
Previous short-term planning models were based on a
combination of projected grade control data from previous
mining benches and the Reserve model which is based on
relatively sparse Resource definition drilling. Owing to the
heterogeneous nature of the lode system this can result in major
differences between tonnages and grades predicted from the
short-term scheduling model and those seen when the latest blast
hole data is modelled just prior to blasting and markout. This
discrepancy leads to continual rescheduling of mining priorities
in order to meet grade and tonnage targets.
236
The new scheduling models which incorporate the three bench

RC grade control models correlate well with actual mine
performance. This leads to an environment of proactive and
efficient mine planning and enables the operation to significantly
reduce the costs associated with suboptimal scheduling.
Another major benefit is the accurate ground condition data
available from the RC drilling. The advance three bench data
leads to large voids being detected sufficiently in advance of
mining to enable modifications to the mine plan if required.
Standard probe drilling conducted by the Geotechnical
Department only provides information one bench in advance.
The RC holes are also used for cavity monitor surveys which
result in vastly improved 3D models of the previous mine
workings.
FUTURE DIRECTIONS
The introduction of RC based grade control has been a process of
continual evaluation of systems and review of potential
improvements. This evaluation is ongoing and several projects
have been flagged including wireframing and the use of
conditional simulation techniques.
Wireframing of the ore and stope interpretations is being
assessed. Wireframing would allow improved representation of
the flat dipping lodes in the model and refined flagging of lode
samples. The application of conditional simulation techniques
has potential and will be trialled in the near future.
The improved lode definition from RC drilling has prompted a
review of the economic benefits of split bench mining for narrow
and shallow dipping lodes. A trial of split bench mining is
planned to be performed in the Oroya North cutback in the near
future.
Comprehensive assay quality control procedures are currently
being established for grade control. These incorporate the use of
duplicate samples, waste standards, repeat assays and pulp
standards.
SUMMARY
ACKNOWLEDGEMENTS
RC grade control was implemented at Fimiston after trials

indicated excessive dilution was being incorporated with existing
grade control methods based on Blast hole sampling. There was
potential to improve ore delineation and grade estimation by
utilising inclined drilling with two metre samples.
The new system has led to more accurate ore interpretations
and reliable grade estimates. This is reflected in the confidence
the mine geologists have shown in the grade control models. This
confidence translates into a consistently close correlation
between these predicted models and the actual ore/waste markup.
Economic benefits of the new system are apparent in improved
mine to mill reconciliations and more robust short-term mine
schedules.
The implementation has been a major undertaking, with full
coverage in all mining areas taking much longer than anticipated.
With the majority of issues now addressed the new system is
expected to be in use across all production areas by January
2000.
From trial programs in late-1997 to full implementation of RC

based grade control across Fimiston in December 1999 there has
been a considerable effort by the Mine Geology group. The
authors would like to thank all those who have contributed to the
development and implementation of RC based grade control.
Thanks to all reviewers of this paper and to KCGM management
for allowing us to present this paper. We also appreciate the
support and guidance of both Homestake Gold of Australia Ltd
and Normandy Mining Ltd throughout the entire process.
REFERENCES
Binns, R A, Gunthorpe, R J and Groves, D I, 1976. Metamorphic patterns
and development of greenstone belts in the eastern Yilgarn Block,
Western Australia, in The Early History of the Earth (Ed: B F
Windley), pp 303-313 (John Wiley: London).
Clout, J M F, Cleghorn, J H and Eaton, P C, 1990. Geology of the
Kalgoorlie gold field, in Geology of the Mineral Deposits of
Australia and Papua New Guinea (Ed: F E Hughes), pp 411-431
Pederson, L and Reardon, C, 1999. Adaptability The key to an
integrated grade control computing system, in Proceedings APCOM
Computer Applications in the Minerals Industries, (Ed:
K Dagdelen), pp 601-609.
Snowden Associates, 1998. Kalgoorlie Consolidated Gold Mines Reverse Circulation Grade Control Conditional Simulation Study.
237
The Use of Magnesium Oxide and Iron to Predict Host Units in

Grade Control and Exploration at Bulong Nickel Operation
W J Bollenhagen1
ABSTRACT
At Bulong, the mineralised (>0.8 per cent nickel) ore exists within
numerous lithologies, which can and do occur coincidently. It is very
important to discriminate between the different host units for estimation
purposes, stockpiling and scheduling. This allocation of host unit has
relied upon visual interpretation by the geologist recording the
information, which can be very subjective by its nature.
By using quantified constraints, such as assayed magnesium oxide and
iron, the subjectivity of describing the host material of nickel laterite ore
is reduced. The constraints permit consistent identification of the various
host units, which improves the confidence in the overall estimation and
scheduling of the ore.
INTRODUCTION
The Bulong Nickel Project (Figure 1) has a nickel resource that
is contained within the laterite profile of highly weathered
ultra-mafic rocks.
In the most simplistic way the laterite is described as being

comprised of a number of different units, termed lithologies.
Each clay lithology of the laterite bears its own unique physical
and chemical characteristics. It is these characteristics that also
determine the degree, if any, to which these lithologies are
enriched with nickel and/or cobalt. The notion that these
lithologies exist as completely flat lying entities is a notion that
should dispelled, they can be highly irregular and strongly
structurally controlled.
With the passing of long periods of time nickel mobilises
from the parent minerals (olivine) of the bedrock and relocates
to other vacant sites in the profile. This occurs because of a
change in the surrounding environment, such as pressure,
temperature and pH. Which results in the mineral lattice being
modified to a different stable physical and chemical form.
This is a simplification of the process known as nickel
laterisation.
At Bulong a number of variables influence the development of
the laterite profile and many types of laterite profile can
develop in a selected area. A number of high-grade nickel
deposits have been identified that have characteristics that are
unique to each other.
It is because of these characteristics that each deposit needs to
be investigated separately when considering the processes of
nickel enrichment and mining scheduling.
This paper will investigate the characterisation of different
lithologies using their direct relationship with iron (Fe) and
magnesium oxide (MgO) values. This method allows for
objective and consistent ore discrimination, which improves
confidence in resource and grade control delineation.
Project description
Preston Resources NL has a large lateritic resource of nickel and
cobalt at Bulong, from which it intends to develop the high-grade
portions during the initial years of the project. The
hydro-metallurgical processing comprises:
acid pressure leaching at 250C,
solvent extraction (SX), and
electro-winning.
Ore containing high-grade nickel and cobalt is processed to
produce London Metal Exchange (LME) grade nickel and cobalt
cathode.
The project is located close to the abandoned Bulong townsite,
approximately 30 km due east of the City of KalgoorlieBoulder
in Western Australia.
The climate of the area is classified as arid, with an average of
257 mm of rainfall per annum. The frequency of rainfall is low
and the variability is high, with infrequent very heavy rains.
Owing to high average temperatures, the losses by evaporation
exceed the rainfall amounts in all months of the year.
FIG 1 - Bulong Nickel Operation location map.
Geology
1.
MAusIMM, Mine Resource/Development Geologist, Bulong Nickel

Operation Pty Ltd, PO Box 10391, Kalgoorlie WA 6430.
The bedrock comprises a thrust sequence (up to 4 km wide), of

steeply dipping Archaean ultramafic volcanic flows - the Bulong
Complex - within a sequence of mafic volcanic and sedimentary
rocks. Olivine-rich ultramafics, particularly peridotites and
239
W BOLLENHAGEN
dunites constitute the significant proportion of the sequence,

these have undergone low-grade (greenschist facies)
metamorphism, (Elias, Donaldson and Giorgetta, 1981; Burger,
1996).
The sequence has been extensively serpentinised, and in places
suffered talc and carbonate alteration. The talc possibly reflects
the zones where shearing was most active within the ultra-mafic
suite, the talc units appear to form very important buffers for
manganese/cobalt mineralisation development.
The Bulong laterite deposits developed in Mesozoic and
Tertiary times under a climate more humid than at present.
Profile characteristics in places suggest multiple episodes of
weathering while elsewhere the laterite has been partially eroded
or covered by younger detrital clays and sands (Elias, Donaldson
and Giorgetta, 1981).
There is potential for nickel and cobalt laterite development to
be a result of biological activity (Burger, pers comm).
LATERITE
Processes of formation
The typical laterite profile is characterised by a number of
distinct lithological subdivisions. These lithological divisions are
based upon:
mineralogical composition, and
chemistry.
For nickel laterite at Bulong, the most important lithologies are
referred to as:
ferricrete,
limonite,
nontronite,
serpentine,
talc, and
chert.
Fresh rock is encountered at some point beneath these
lithologies and the laterite profile stops.
The depth and extent of laterite development is quite varied
and is influenced by a number of coincident factors.
Workers (Parianos and Rivers, 1996) have identified these
factors, which can be grouped under two categories:
1.
physical, and
2.
chemical.
Physical
Is typified by water movement and environments that influence
its rate and direction.
Water flow is influenced by:
rainfall,
rock structure and texture, and
geomorphology.
Chemical
Primary chemical influences are from:
bedrock composition (competing chemical solubilitys);
input from vegetation cover (organic); and
the relative resistance of different existing laterite
mineralogies, to changing water chemistry.
The physical and chemical factors determine the degree of
weathering that a rock undergoes. The degree of weathering
influences what elements and volume of these elements are
released into solution, which in turn, defines the depth, chemistry
and physiography of the laterite profile development.
240
The more soluble minerals (Si, Mg and Ca) typically mobilise

to the base of the profile, whereas the relatively less soluble ones
(Al, Fe, Ni, Co and Cr) are concentrated at the surface.
The rate of weathering is faster in the tropics because of the
warmer and wetter environments. The rate of ionic exchange of
cations from the crystal lattice and hydrogen ions from water
increases in humid environments (Twidale, 1976)
The movement of ions occurs to maintain equilibrium in the
rock profile and requires parent minerals to undergo chemical
reactions that permit the formation of other more stable minerals.
Possibly in the future these physical and chemical factors will
be comprised in a more unifying description biological.
NICKEL LATERITE
This is one class of laterite and it hosts the primary ore
encountered at Bulong, another example of a laterite deposit is
bauxite, which is aluminium (Al) enriched.
Nickel laterite deposits are derived from olivine rich rocks
(ultramafics), these are often dunites or peridotites. These rocks
have magnesium oxide (MgO) and nickel (Ni) contents
substantially higher than other crustal rocks (mafics and
granitoids).
Whilst the rock is in a molten state (magma) below the earths
surface conditions are right for nickel to partition into the lattice
of available olivine minerals, likewise cobalt (Co) partitions itself
into available pyroxene lattice.
These magmas are then extruded onto the earths surface, as
volcanic flows (often submarine) and harden as broad sheets.
Other later flows consequently are extruded and cover these
original ones.
The individual flows have unique chemical signatures, but the
effects of weathering and alteration often mask those signatures,
making them indistinguishable from each other. Over large
periods of time these sequences of rocks are subjected to various
structural and metamorphic changes, resulting in the original
sheet appearance being greatly altered.
At Bulong this geological situation exists, where these sheets
now exist in an upright (near vertical) nature and due to
deformation have been faulted and thrust past one another
resulting in some thickening of the sequence.
After tectonic movement and at relatively shallow crustal
depths (<450C and abundant H2O) the primary olivine and
pyroxene minerals are altered to varieties of the mineral
serpentine. The degree of serpentinisation ranges from minor
alteration around fractures to complete metasomatism of the
rock.
During weathering, acidic (CO2 enriched) soil waters percolate
downwards and react with the ultramafic rocks, this causes SiO2
and MgO to be leached from the mineral lattice into a solution
where it can become mobile. It is this mass removal that
contributes to the preferential enrichment of other elements into
the sites made available by the leaching of the MgO and SiO2.
With a considerable passing of time, a volume of mobile ions
move out of the mineral lattice, which re-organises itself to
accommodate the new environmental conditions (chemical,
physical and biological). This can be described as a chemical
system achieving a metamorphic equilibrium, thus laws of
thermodynamics can be applied to these exchanges.
At Bulong this process has resulted in economical nickel
deposits developing in the near surface part of the laterite profile.
These deposits have been identified by exploration methods,
including Rotary Air Blast (RAB), Reverse Circulation (RC) and
Air-Core (AC) drilling.
THE USE OF MAGNESIUM OXIDE AND IRON TO PREDICT HOST UNITS
The ore is typically partitioned into three types of lithology

within the laterite profile. These are as follows.
Limonite zone
Strongly enriched in Fe and depleted in MgO.
This zone is characterised by goethite and hematite with minor
kaolinite and quartz. The goethite is present as dry, loose clay,
which is overlain (in places) by a ferruginous hard-cap.
The physiography and extent of development of this unit is
quite varied. Limonite development is greater where
ground-water drainage is good, if drainage is poor then limonite
development is significantly restricted. The colour varies from
light brown to dark brown; it has an earthy texture.
The nickel content is typically lower in this lithology than the
smectite lithologies. The Co content is typically higher in this
lithology than the underlying smectite and serpentinite. Towards
the surface the goethite is replaced by hematite, Fe2O3 and the
Mn-oxides are leached away.
The presence of limonite in certain parts of the laterite profile
is a strong indicator of structural complexity.
Smectite zone
This clay zone is not always present, but characterised by
strongly Ni enriched nontronite clays together with
montmorillonite, beidellite and saponite. A wide range of
chemical variation is reported in association with this lithology
The clay zone is characterised by the presence of nontronite
clays ().
Nontronite is the primary smectite encountered at Bulong and
for simplification all clays containing high moistures (+20 per
cent) are referred to as nontronite. The nontronite is pale-green to
deep green, often with black Mn-oxides. The green colour is
attributed to the presence of Fe+++, however where species
enriched with Fe++ exist, these are brown to red-brown in colour.
The physiography and extent of the nontronite development is
constrained by the amount of ground-water retained in the
laterite profile. For nontronite development to occur, drainage
has to be poor, so in a heavily fractured rock it would be
expected that nontronite development would be restricted.
The Bulong deposit is fortunate to have a number of areas that
are dominated by nontronite rich profiles.
Serpentinite zone
The serpentinite zone contains relict bedrock mineralogy
(serpentine) together with some smectite and secondary silica,
MgO is significantly elevated in this lithology. It is the
transitional part of the laterite profile where there is a decrease in
strongly weathered material and an increase in less weathered
rock.
The material can comprises a mixture of clays (nontronite
and/or limonite) and partially weathered fresh rock (antigorite),
often as remnant fragments. However, depending on the degree
of weathering that the rock has been subjected to, a core-stone
may have developed in situ. This core-stone exhibits
mineralogy that is more representative of the unweathered
bedrock. Mineral features and textures are preserved in this
lithology and parent rock identification can be made. At Bulong,
the serpentinite is derived from an olivine rich ultra-mafic, thus
its texture is dominated by the features of the weathered olivine
crystals. The serpentinite sample exhibits a grey-green colour
and small (<0.5 cm) weathered, sub-rounded olivine crystals can
be seen in the rock fragments.
A minor to strong presence of silica with a lesser, variable
magnesite may also be encountered in this zone, in the form of
chert or chalcedony.
Relative to the nontronite clays, considerably lower ore grades

are encountered in this part of the profile, although some
intervals report high Ni grades in serpentinite.
The MgO level becomes elevated at the interface of the
serpentinite and the unit above (smectite, transported clays,
ferricrete or limonite/goethite) and the level continues to rise
down through the profile until the background MgO level
(between 25 and 40 per cent wt.) is reached. The presence of
magnesite can distort the background MgO values for the
serpentinite.
ORE PROFILING
Now that the complexity of the laterite has been introduced, this
section will focus on the other complexities that have to dealt
with by the geologist, for the purpose of providing the required
ore feed blends to the auto-clave.
Because of the high pressure and hot acid treatment of the
lateritic ore, many chemical reactions occur during the ores
digestion in the autoclave. The complexity of these reactions is
a direct result of the mineralogy being treated.
In the initial steps of processing the ore, reactions are sensitive
to the amounts of Fe, MgO, Mn and Cr present in the ore feed.
Subsequent steps in refining the ore are sensitive to the Cu and
Mn levels. The geologist has to have a high degree of confidence
in their ability to differentiate between different ore types, so
that the ore that is being scheduled for processing conforms to
the specifications stipulated by the metallurgists.
Differentiation between the lithological units can become very
subjective. This can lead to misclassification of the ore
lithologies and consequent dilution or miss allocation of feed
grade. Fortunately, a relationship exists between both interpreted
lithology (mineralogy) and bulk sample chemistry, which has its
origin in the lateritic formation process.
The nickel deposits at Bulong have unique location-specific
ore and lithological characteristics as a consequence of the
processes behind their development. Any relationship developed
between the chemistry and the lithological units remain unique to
the deposit. To be practical for grade control and resource
delineation any technique needs to be simple and flexible.
In order to develop an understanding between chemistry and
lithology, the analysis was performed using drill hole and
available grade control data. For selected areas of interest, all
assay data was sorted into its various different lithologies
(provided by geological logging), with averages and tolerances
calculated for each lithology. To simplify the lithological
representation, the relationship developed is restricted to the
elements Fe and MgO.
These two elements are selected because, in addition to silica
they comprise the dominant parts of the clay mineralogical
composition. The mineralogy of a selected part of a laterite
profile will be reflected by the percentage of Fe and MgO
present. In the simplest terms, the weathering process of
ultramafic rocks preferentially enriches Fe in a profile as MgO is
removed with silica from mineral lattices.
Flow of weathering
An ideal representation of the path that olivine in ultramafic
rocks undergoes during weathering is presented below;
Olivine
MgO
Serpentine
MgO + SiO2
241
W BOLLENHAGEN
Nontronite
MgO + SiO2
Limonite
MgO + SiO2
Haematite
Importantly at every stage during weathering, silica and/or
magnesium oxide are released to the environment. With good
drainage conditions these compounds are removed away from the
original site, however with the passing of time and changes in the
tectonic regime they can be retained within the system. This
circumstance has occurred at a number of locations within the
Bulong laterite profile, where significant volumes of magnesite
and chert have been intercepted during drilling.
FIG 2 - Represents the lithologies as a function of their assayed
MgO and Fe.
Method of analyses
The analyses were performed in a series of steps.
1.
extraction of the data from the various databases;
2.
sorting the data into its separate mineralogies and

subordinate mineralogies;
3.
calculating and presenting data to determine elemental

cut-offs;
4.
calculating the lithological code in terms of the intervals

MgO and Fe content; and
5.
graphically presenting the data against field logging of

lithologies.
A total of 13 279 assays were extracted from grade control
data sets representing three different mining areas.
This sample set comprised:
3177 nontronite,
6338 serpentine, and
3764 limonite.
The data was not selected on its economic viability.
The lithological codes created in step four were then read into
a database with the rest of the assay suite, these new lithological
codes are expressed as a numerical value, they can be then
represented graphically. This data along with corresponding
headers and lithologies were then read as drill holes into
Gemcom (mining software). Separately the exploration data was
also manipulated into this format so that it could be presented
along with the grade control data for comparison.
In Gemcom, the newly created lithological codes were brought
on screen graphically and coloured to distinguish them from
one another. Polygons were then created around the separated
lithologies, ie nontronite, ferricrete/limonite and serpentinite.
This was done irrespective of the nickel and cobalt grades.
RESULTS AND APPLICATIONS

Figure 2 is a graph of MgO versus Fe for each of the three
different lithological units, nontronite, limonite and serpentinite
recorded during grade control.
Immediately obvious from the graph is the degree of scatter in
the data points. This can be attributed to the simplification of
lithologies during data recording (logging) and the mineral
complexity of the in situ materials.
Despite the scatter, there is some correlation within each
lithology to a range of Fe and MgO values.
242
The biggest influence on the results is the accuracy of the

original lithological descriptions (geological subjectivity), this
influences destination of the corresponding sample intervals.
Being able to account for penalty minerals (MgO, Fe, and Mn)
is very important at all stages in the resources life.
One penalty mineral is highlighted in the graph. The elevated
MgO (>5 per cent) in the limonite sample lithology reflects the
presence of magnesite (MgCO3). The sample intervals have been
reported as being comprised of a dominant limonite mineralogy,
however they also have appreciable magnesite (MgO) (ie five to
ten per cent).
Also reflected in the graph above is serpentine ore that
contains subordinate nontronite. The presence of nontronite in a
serpentinite dominant sample lowers the MgO value and
relatively increases the Fe value being reported (Figure 3).
FIG 3 - Photomicrograph of ore grade nontronite within barren serpentine

(scale = 100 mm).
From the assay data, mean and standard deviation was

calculated for each of the elements in each of the lithological
units (Table 1). The population means support the idea that there
can be some distinction made between each of these lithological
units in terms of their MgO and Fe contents. From the averaged
data presented in Table 1, the following ranges for either MgO
and/or Fe are used for lithological differentiation.
THE USE OF MAGNESIUM OXIDE AND IRON TO PREDICT HOST UNITS
>10 per cent MgO

>30 per cent Fe
<10 per cent MgO and
Serpentine
Resource application
Limonite
The use of these Fe and MgO constraints ensures that the ore
lithologies are delineated properly. This is reflected in the
variography performed during resource modelling, where lower
population variances and nuggets are estimated on application
of these constraints. It also serves as a check on how well the
different lithologies are modelled, as poorly modelled data will
show higher nugget effects and greater population variances. By
examining the laterites in terms of Fe and MgO a better
understanding of the local geological structure can be gained.
Weathering is more pronounced in zones where ground water is
free to move. Zones that are heavily fractured or sheared act as
very open systems for water movement, whereas more
geologically quiet zones restrict water movement more.
Limonite (Fe rich) and ferricrete (Fe-rich) reflect these zones of
strong water movement and nontronite reflects zones of water
flow restriction.
Interpretation can be made using these chemistries to better
understand and model the lithological units. Structural
complexities can be interpreted and modelled at the resource
stage, assisting in better pit design and scheduling.
Nontronite
<30 per cent Fe
The above cut-offs are used by geologists to assist in the ore
delineation. To provide both accuracy for scheduling purposes
and consistent ore feed to the auto-clave for processing.
TABLE 1
Average MgO and Fe for the data population, with
standard deviations (SD).
Magnesium Oxide (MgO)
Lithology
Mean (%)
Standard deviation (%)
Nontronite
6.29
2.41
Serpentine
15.36
2.95
Limonite
2.64
1.78
Iron (Fe)
Lithology
Mean (%)
Standard deviation (%)
Nontronite
18.81
2.49
Serpentine
14.43
3.00
Limonite
39.41
5.64
It is these two aspects of ore delineation that are critical to the

smooth flow of ore through processing. Accurate delineation of
ore prior to mining is very important for the mining sequence.
The percentages of ore lithologies delineated by the geologist
need to be very precise so that the sequence and scheduling of
mining equipment can be prioritised to ensure maximum profit is
gained. The blend of ore to meet processing requirements is very
specific.
Grade control application

The primary aim of grade control is to deliver ore that meets all
the processing requirements. Confidence in the nickel and cobalt
grades is not the only focus. Other factors such as the ores
temporary destination on the stockpiles, the host lithology and
MgO content are as equally important. The use of Fe and MgO
constraints by the mine geologist improve the confidence and the
required consistency of the ore. It enables the geologist to
discriminate between nontronite units that have high MgO
contents and those of high Fe contents. These have different
mineralogies and material behaviour characteristics in
processing. Figure 4, shows a complex geological zone that can
be delineated by the use of chemistry. The different lithologies
can be delineated and in a clear and consistent fashion, which is
FIG 4 - Grade control trenches coloured by chemically derived lithology, with recorded geological type projected against the trench.
Trench colouring; black Serpentinite, dark grey Limonite Ferricete and light grey Nontronite.
243
W BOLLENHAGEN
essential to the operations performance. It can be seen that both

the recorded (by geologist) lithology (displayed next to the
trench (line) and the calculated geology (colour of trench)
correlate well, for a two metre lateral and vertical sample
composite shown.
There are however some discrepancies at the lithological
contacts where the geologist may not have identified the
probable dominant mineralogy. A complete lithological picture
provided by chemical delineation is shown in Figure 5. This is a
geological interpretation of Figure 4, showing the hard
boundaries.
The orebodies can appear complex and have the potential to
cause some difficulties, but by understanding the many other
facets of the ore lithologies it enables greater ore definition. By
understanding and modelling the complexities the geologist can
aid in the front end materials handling of the ore within the
mill, as well as understanding discrepancies in sample values at
this point of processing.
CONCLUSION
For a cut-off of ten per cent MgO all samples are comprised
dominantly of serpentine. The samples with greater than 30 per
cent Fe can be considered limonite. The remainder of the
population below these two values is assigned to the nontronite
lithology. This discrimination is specific to the three operational
pit areas currently being mined at the Bulong Nickel Operation.
Different MgO and Fe parameters are used in other resource
areas to delineate the ferricrete, clays (transported or in situ) and
talc units from the three primary ore lithologies.
The simplification of lithology is necessary in both the mining
and resource modelling phases. Situations exist where the
individual collecting the geological information may not be the
same person responsible for the modelling of the data.
Understanding the mineralogical complexities away from the
field and being able to make informed decisions about
mineralogical changes needs to be consistent and modestly
accurate. This approach is to ensure the performance of the ore
feed through the processing phases in the plant is maintained at
an optimum.
An unfortunate function of this approach is that sub-ordinate
mineralogies are ignored, even though they contribute to the
sample intervals bulk chemistry. At the mining scale, large
relative differences in elements are important, since they
contribute directly to our understanding of the deposit.
Without accounting for changes in chemistry, potential
problems can arise in scheduling of both the identified resource
and the stockpile destined for processing.
244
FIG 5 - Geology created by Fe and MgO per cents from grade control data
Criterion Pit, 370 RL.
At Bulong because of the variation in the individual deposit

mineralogy, the step towards using chemistry as a constraint
against mineralogy has proven to be an important ore control
tool. It removes some of the variation associated with
subjectivity in geological logging encountered by geologists
during ore delineation and contributes to consistency and
confidence in the identification of important mineralogies.
REFERENCES
Burger, P A, 1996. Origins and characteristics of Lateritic Nickel
Deposits, in Proceedings Nickel 96, pp 179 - 183 (The Australasian
Elias, M, Donaldson, M J, and Giorgetta, N, 1981. Geology, Mineralogy
and Chemistry of Lateritic Nickel-Cobalt Deposits near Kalgoorlie,
Western Australia, Economic Geology, 76:1775-1783.
Parianos, J M and Rivers, C J, 1996. Geology and Mineralogy of the
Brolga Laterite Deposit Central Queensland, in Proceedings Nickel
96, pp 63 - 68 (The Australasian Institute of Mining and
Twidale, C R, 1976. Geomorphology, 136 p (Thomas Nelson: Australia).
Mine to Mill
Ore Characterisation for Mine to Mill
Fragmentation
A Scott, J Segui and

S Kanchibotla
247
Upgrade Ability and Geology of Cawse Nickel

Ore
S M Denn, C G Ferguson
and S L Makin
255
The Interaction Between Geology, Mining and

Metallurgy at Stawell Gold Mines Pty Ltd
D Fredericksen
263
Ore Characterisation for Mine to Mill Fragmentation

A Scott1, J Segui2 and S Kanchibotla3
ABSTRACT
The Mine to Mill concept is not new. In fact, it is the underlying
objective of every mine. Unfortunately, in many operations the
management structure and reporting systems have created independent
businesses within the overall mining enterprise. This has created
incentives to optimise the performance of each component of the
operation rather than the enterprise as a whole. The Mine to Mill
Conference organised by The AusIMM in October 1998 documented a
number of case studies where run of mine ore was not presented to the
mill in the best possible form for subsequent stages of comminution and
processing. A number of examples were quoted where modest
expenditure in the mine had a very significant impact on downstream
process efficiency and hence costs.
Examples included:
the minimisation of fines in the mining and handling of iron ore;
the management of fines in blasting and crushing for a copper heap
leach operation;
the control of blast damage, mining recovery and dilution in open cut
coal operations; and
the optimisation of mine fragmentation to increase mill throughput
in gold and copper operations.
The latter topic has been trialed at a number of mines and is referred to
as Mine to Mill fragmentation.
A number of Mine to Mill type investigations have been reported in
the literature including Pease et al, Simkus and Dance, Nielsen, Scott et
al, Kojovic et al, Kanchibotla et al, Bulow et al and Hawkes from the
1998 Mine to Mill Conference. In addition, a number of studies are in
progress or the results have yet to be published. These studies have
demonstrated the feasibility of obtaining the data required to characterise
the breakage behaviour of ores and to model the impact of changes to
blasting practices on process efficiency.
Characterising the breakage behaviour of ores is a vital step in the
management of fragmentation and comminution in mines. The
characteristics of importance to fragmentation fall outside the parameters
conventionally required to describe the economic geology of a deposit.
Mine to Mill optimisation of fragmentation means that the mining
engineer needs information from the geologist in order to be able to help
the metallurgist. Providing this information requires close interaction
between these groups.
MINE TO MILL OPTIMISATION OF

FRAGMENTATION
In classic metalliferous operations the ore undergoes at least
three stages of breakage:
blasting to prepare the ore for excavation and transport;
crushing to improve its handling characteristics and to
prepare the ore for grinding; and
grinding which is usually undertaken in two stages.
In large open pit mines, rule of thumb figures suggest the
relationship between the energy requirements and costs for these
three stages of breakage are as shown on Table 1.
In the last 20 years grinding circuit design has undergone
considerable change and has seen the rise in popularity of
autogenous (AG) and semi-autogenous (SAG) mills to the point
1.
FAusIMM, Scott Mine Consulting Services Pty Ltd, 45a Edson

Street, Kenmore Qld 4069.
2.
SAusIMM, Research Scholar, JKMRC, Isles Road, Indooroopilly,

Qld 4062.
3.
Research Fellow, JKMRC, Isles Road, Indooroopilly, Qld 4062.
TABLE 1
Relative energy and cost of breakage.
Energy
Factor
Cost
Factor
Blasting
0.2 kwh/t
$0.15/t
Crushing
2 kwh/t
10
$0.75/t
Grinding
20 kwh/t
100
$3.75/t
25
where it is rare to find a comminution circuit where they are not

used. This trend has enabled unit throughput size to increase
enormously, with consequent economies of scale. The downside
of these machines is that they are relatively sensitive to the size
distribution of the feed, particularly AG mills. This is because
the feed provides the media responsible for grinding as well as
providing the feedstock that is ground.
The feed size distribution has to provide a balance between
the amount and size of the media (coarse fraction) and the
feedstock (fine fraction). Of course the rock media as it wears in
the mill also provides a source of ground product in its own right,
but this is a relatively slow process and so coarser rocks tend to
accumulate in the mill. Hence if a feed size distribution is
produced with too much coarse material or very large top size,
the mill will tend to overfill with pebble size rocks (25 mm 75 mm) and limit throughput. If too little coarse material is fed
there will be insufficient grinding media to grind the feedstock to
the required size.
In both scenarios pebble sized material is found to be a
problem and this fraction is often referred to as the critical size.
Hence it is also found that feeding a size distribution which is
rich in pebbles will also limit throughput. It follows from these
experiences that the ideal feed size distribution is one which has
relatively little pebble size material, does not have excessive
coarse material and has as much material in the below-grate size
region (10 mm - 15 mm) as possible.
The Mine to Mill concept involves increasing the amount of
breakage achieved in both blasting and crushing in order to
relieve the mill of as much new breakage as possible. In essence
the breakage is moved back down the production chain where the
energy requirements (and the costs) are lower. Figure 1 shows the
concept.
The first objective of Mine to Mill blasting is to reduce the
top size of the material to improve the ease of excavation and
transport within the mine. Reduced top size also allows the
primary crusher gap to be reduced, generating material which
needs less breakage in the mill. With reduced top size and
continuous material supply, the crusher can be choke fed without
the risk of blockages. Choke feeding allows more inter-particle
breakage and therefore produces more fines in the crusher
product than without choke feeding. The top of the cumulative
fragment size distribution curve shown in Figure 1 is moved to
the left.
The second objective is to increase the proportion of fines
(material less than the grate size in the mill, say 10 mm) because
this should pass freely through the mill and require no further
breakage. This is represented by lifting the lower end of the size
distribution curve. These changes steepen the curve, reducing the
proportion of material in the critical size range of around
25 mm to 75 mm.
247
A SCOTT, J SEGUI and S KANCHIBOTLA
Physical characterisation
FIG 1 - Changes in size distribution sought through blasting.
The impact of these changes can be simply evaluated. If the

additional blasting cost was $0.10/t and mill throughput
increased by ten per cent, then according to the figures on Table
1, the average milling cost would fall by $0.38, or four times the
extra investment made in the blast. Of course this simple
approach ignores many factors including likely benefits to the
mining operation itself from higher excavation productivity,
higher availability and reduced maintenance costs for equipment.
If the increased revenue generated from additional sales made
possible by the higher production rate is factored into the
equation, the rewards can become very significant.
CHARACTERISATION OF ORES
Ores are conventionally characterised in three ways prior to
mining. These are:
economic characterisation,
mineralogical characterisation, and
physical characterisation.
Economic characterisation
Clearly the most fundamental aspect of orebody characterisation
is whether the material is in fact ore. This is an economic
definition. The amount of potentially valuable minerals and their
distribution within the rock mass are the basic driver for the
definition of the orebody. Of course the net value of the ore
(value less costs) also depends on the costs associated with
mining, transport and processing. These costs depend on the
characteristics of the orebody and so a cyclic process of
evaluation is required where initial estimates of mine geometry,
blasting, excavation, haulage, comminution and processing
performance are made to provide a first pass definition of the
orebody. Subsequent planning cycles refine each of these
estimates to converge on the bankable mine plan. Optimising
this plan depends strongly on the adequacy of the mineralogical
and physical characterisation of the ore.
Mineralogical characterisation
The nature of the minerals present and their association with
each other control the ease with which the desired minerals can
be separated from the gangue. Clearly, liberation verses particle
size and the presence and influence of contaminants on the final
product are key factors. Texture is now being quantified to relate
readily observable features of mineral specimens to their
mineralogy and potential processing performance (Napier-Munn
et al, 1996). Great strides have been made in recent years in the
application of tools such as QEM-SEM and the JKMRCs
Mineral Liberation Analyser in the routine quantification of
mineral composition and character (Gu et al, 1998).
248
The physical characteristics of ores are generally given less

attention by mine geologists than the economic or mineralogical
aspects. It is the physical characteristics that are the missing
links in the mine to mill chain. Most mine geology groups will
keep records of significant structures in core and log material
against a nominal hardness or strength scale (Berkman and Ryall,
1976)
The core may be separately logged or re-sampled for
geotechnical purposes where a more careful look at structure and
sampling of material for basic strength testing may be
undertaken. Indices of rock mass competence such as RMR and
Q (Bieniawski, 1973; Barton, Lien and Lunde, 1974) are often
calculated and reported.
Breakage behaviour is commonly estimated in the form of the
Bond Index (Bond, 1952) which is a measure of the amount of
work (expressed in kWhr/t) required to reduce material to 80 per
cent passing 100 microns. The Bond Index is obtained from
controlled breakage tests and is the major parameter used by
most comminution equipment manufacturers.
CHARACTERISTICS THAT AFFECT

FRAGMENTATION
Importance
Treating blasting as the first stage of the comminution process
and managing that process to optimise the overall economics of
an enterprise requires an understanding of:
the blastability of the in situ ore;
the impact of blast fragmentation on the efficiency and cost
of mining and transport;
the crushing and grinding behaviour of the ore; and
the impact of feed size distribution on the performance of
crushing and grinding circuits.
The management of this process requires adequate:
characterisation of the in situ ore;
measurement of run of mine and post crusher fragment size
distributions; and
monitoring of mining, crushing and milling performance and
efficiency.
Of these, the mine geologist can contribute significantly to the
characterisation of the blastability of the ore.
Blastability
Rock is a complex material and its behaviour when blasted is
indescribably complicated. A review of approaches to rock mass
characterisation for blasting is provided in Scott et al (1996).
There is still much to be learned about the most effective
technical approaches to rock mass characterisation for blasting.
Approaches used to-date generally utilise data available from
geotechnical characterisation. This involves parameters such as
unconfined compressive strength, Youngs modulus and a
description of structure either estimated from field exposures or
derived from RQD or other indices derived from logging core.
Blasting involves both the creation of new fractures and the
extension of existing cracks and joints to loosen and liberate in
situ blocks within the rock mass structure. The fine end of the
fragment size distribution curve is controlled by the substance
strength and the breakage energy provided by the explosive. The
coarse end of the size distribution curve is strongly influenced by
the in situ block size and the spatial distribution of explosive
within the rock mass. Muckpile movement is influenced by the
rock density and stiffness, the heave energy and distribution of
the explosive. The minimum description of rock mass
blastability therefore involves:
ORE CHARACTERISATION FOR MINE TO MILL FRAGMENTATION
density,
strength,
stiffness, and
in situ block size.
Armed with a reasonable description of these properties,
alternative blast designs can be modelled to predict
fragmentation and muckpile shape. This allows mining
performance (and hence costs) to be predicted and provides an
estimate of the feed size distribution for downstream crushing
and milling operations.
Density
The measurement of the density of low porosity materials is
relatively straight forward and is generally accomplished by
measuring the weight of a sample in air and while immersed in a
liquid of known density. Porous materials such as coal offer
some complications as described by Preston and Sanders (1992).
Strength
Blasting is a dynamic process and it is reasonable to assume that
some measure of the dynamic strength of the rock is required in
fragmentation modelling. However, to the authors knowledge no
fundamental or mechanistic model of blast fragmentation has
been successfully scaled up to production blast geometries to
provide practical guidance to blasting operations (Scott et al,
1993). The most successful blast fragmentation models are
basically empirical, although the most useful mine to mill models
(Kanchibotla, 1998; Djordjevic, 1999) do utilise some
mechanistic components, particularly in the estimation of the
fines end of the distribution. Empirical models can satisfactorily
utilise static strength parameters which are far easier to measure
and more likely to be available from conventional geotechnical
investigations.
The most commonly reported strength parameter is unconfined
compressive strength which requires measurement in the
laboratory using carefully prepared samples. The test is subject
to an international standard procedure (Brown, 1981) and has the
benefit of providing both stiffness (Youngs modulus) and a
deformation parameter (Poissons ratio) at the same time. Intact
core makes an ideal source of specimens for determining
unconfined compressive strength.
The point load strength (Brook, 1985) provides a very
convenient estimate of compressive strength from either core or
hand samples from the field. A sample is broken between two
conical platens and the failure load recorded. The load at failure
is normalised to represent an equivalent 50 mm diameter
specimen. The unconfined compressive strength can be estimated
from the point load strength provided a statistically significant
number of point load tests have been undertaken.
Tensile strength can be conveniently measured by the Brazilian
Test in which a disc of core is loaded diametrically between two
platens. The specimen actually fails in tension induced normal to
the axis of the applied load so the failure load can be related to
an estimate of the tensile strength through the geometry of the
sample.
Liberation of fines
When rock is broken, fine material may be liberated from the
new surface that has been created through the rock matrix. This
material might be fine clay particles resident within the rock
structure or particles resulting from the abrasion of the new
surface by other rocks. Particles from both sources are found to
be in addition to the daughter particles expected to result from
the crushing or breakage of the rock by blasting, crushing or
milling.
The Dean David Index (Scott et al, 1998) is a parameter used
to describe the liberation of fines when a rock is broken. The
index is a measure of the proportion of the material presenting as
-38 microns when the ore is broken in a laboratory crusher.
While not relevant to conventional indicators of blasting
performance such as the appearance of a muckpile or its
excavation characteristics, the proportion of this very fine
material can be significant to milling or leaching performance.
The index is determined by passing hand samples of the ore
through a laboratory crusher and plotting the proportion of -38
micron material against the new surface area created by crushing.
The new surface area is estimated by sieving the initial samples
and the crushed product and approximating the fragments as
spheres. The resulting index is then used to estimate the
proportion of material liberated as fines during blasting or
comminution.
In situ block size

In any rock mass the rock substance may be disrupted by a
number of types of discontinuities. These may range from faults,
joints, bedding planes and foliation to more subtle healed
structures or micro-fractures. In terms of blasting, these
micro-structures are important because they provide distinct
planes of weakness which assist the generation of blast
fragments.
Structures within a rock mass are generally not continuous but
do tend to occur in geometric patterns or sets which share
common characteristics such as orientation and spacing. While
not always intersecting to form free blocks within the rock mass,
inspection of muckpiles generally reveals a large number of
blocks (usually the larger fragments in the muckpile) that are
bound by joints on several sides. Blasting extends the existing
discontinuities to complete the definition of these blocks which
are then freed by the movement of the blast volume to form the
muckpile (Figure 2). The coarse end of the fragment size
distribution is strongly influenced by the size of these in situ
blocks.
Stiffness
The empirical nature of the fragmentation models allows them to
be tuned to use a static value for stiffness rather than the
notionally more correct dynamic value. The most common
description of stiffness is Youngs modulus which is derived
from a suitably instrumented unconfined compressive strength
test.
Values for the dynamic Youngs modulus can be derived from
measurements of the sonic properties of a suitable rock sample.
FIG 2 - In situ blocks in a mine bench.
249
In the literature (Harries, 1999) there are a number of models

that attempt to simulate the size and shape of the in situ blocks
and their corresponding distributions based on discontinuity data
collected using detailed line or window mapping methods. These
models include JOINTS (Villaescusa, 1991), SIMBLOCK
(Peaker, 1990) and STEREOBLOCK (Hadjigeorgiou et al, 1995)
and those being developed in the JKMRC-Itasca International
Caving Study (Harries and Eadie, 1999). All require data
describing joint spacing and orientation. These data can only be
obtained from detailed mapping and are seldom available
throughout an orebody yet to be mined.
Simpler approaches (Palmstrom, 1995; White, 1977) can be
used effectively if representative fracture frequency data are
available. To be effective the data must satisfy two criteria:
1.
the data must be from cored holes or exposures at

sufficiently close spacing and an appropriate range of
orientations to provide a relatively unbiased sample of the
rock mass structures in the volume of interest; and
all fractures that affect blasting performance must be

included in the logging.
The second point is vital. Geotechnical engineers concerned
with the failure of a slope are inclined to ignore structures that
cannot contribute to the failure of that slope. Engineers
concerned with the stability of a development drive are only
concerned with the structures that affect the stability or
reinforcement requirements of that drive. Blasting is affected by
all structures, large and small. Anything that provides a plane of
weakness through the rock mass might be exploited by the blast
fracture mechanisms to form fragments. However, it is common
for material logged for geotechnical (stability) purposes to
overlook the smaller and more common of these discontinuities.
Figure 3 shows an example of healed structures affecting
blasting results. The core in Figure 3 is relatively broken, but
stick lengths and hence block sizes of greater than 300 mm are
readily seen. The muckpile from this material contained no
fragments greater than 100 mm which was quite an anomalous
result for the mine. Close inspection of the core from this area
(Figure 4) shows that even the intact lengths of core contained
many planes of weakness that had not yet fractured, but would
take little explosive energy to break. These micro-fractures
were the controlling influence in blasting performance in this
material, but were not logged or noted in the conventional
geotechnical appraisal of the core.
FIG 4 - Micro-fractures in intact core.
2.
CHARACTERISTICS THAT AFFECT

CRUSHING AND GRINDING
Standard rock mechanics tests such as uniaxial compressive
strength and fracture toughness are not appropriate to describe
comminution behaviour on their own. The product size
distribution and energy required to generate that size are
important descriptions of the comminution process and are not
available from these tests. Laboratory tests to evaluate
comminution behaviour are designed to establish a relationship
between the energy input and size reduction achieved. Such tests
include Bond tests, batch grinding and single particle tests. Of
these, the JKMRC utilises the results of a particular single
particle test, the drop weight test, in its models of crushing and
grinding.
To characterise ore breakage at different energy levels, the
JKMRC method uses two complimentary techniques:
to characterise breakage at moderate to high energy levels (ie
impact breakage), a drop weight device is used; and
to characterise breakage at low energy inputs (ie the abrasion
component of breakage), a tumbling test is used.
FIG 3 - Core with some significant intact lengths.
250
The drop weight device comprises a steel drop-weight which is

raised by a winch to a known height. A pneumatic switch
releases the drop weight which falls under gravity and impacts
the rock particle which is placed on a steel anvil. The device is
enclosed in perspex and incorporates a variety of features to
ensure operator safety. By varying the height at which the
drop-weight is released and the mass of the weight, a wide
ranges of energies can be applied to break the target rocks. After
several tests at a given particle size and energy level, the rock
fragments are then collected and sieved. Figure 5 shows the
device in use.
Recent work (Napier-Munn et al, 1996) has identified useful

correlations between some geomechanics properties and
comminution behaviour. Examples include relationships between
fracture toughness and crushing performance and point load
strength and grinding hardness. Further work may improve
these correlations allowing indicative data to be gathered from a
greater number of samples. This should reduce errors associated
with representing large volumes of material with just one or two
laboratory tests.
REMOTE CHARACTERISATION OF
FRAGMENTATION PROPERTIES
Existing access to data
The first task in routinely characterising mine to mill
fragmentation properties is to add to the observations currently
made during exploration and routine proving work by mine
geologists. This would involve:
detailed fracture mapping of available drill core and face
exposures;
sampling and testing of ores for strength and breakage
properties; and
defining blasting domains to guide short-term planning and
blast engineering staff.
While this undoubtedly represents additional work, it provides
the data required to engineer blasts to meet the mine to mill
requirements of the operation. However, exploration drill hole
spacing is determined based on the economic delineation of the
orebody, not the physical characteristics. It is possible that
exploration drilling alone will not provide an adequate definition
of the physical characteristics of the ore.
It would be ideal to be able to remotely sense parameters of
importance to the characterisation of breakage. There is
significant potential to utilise information from the drilling of
blast holes and from geophysical surveys to enhance the data
available from physical samples of the target material.
Data from blast holes
FIG 5 - JKMRC Drop Weight Tester.
The size distribution of the resulting fragments is normalised

with respect to the original particle size. The percentage of
material by weight passing one tenth of the original particle size
is taken as an indicative parameter and referred to as t10. The t10
values are related to the energy input during the test, Ecs, using
the relationship:
t 10 = A(1 e bE cs )
The JKMRC also uses a standard abrasion test to describe the
low energy degradation of an ore. The correlation parameters, A
and b, (from the drop weight test) and ta (from the abrasion test)
are then used by the JKSimMet comminution models to describe
the ore specific breakage properties. In using these models data
are required to describe the machine specific aspects of their
operation.
The most detailed possible data regarding the characteristics of

an area to be blasted can be obtained from blast hole drilling.
This has the advantage of being an evenly distributed sample
through the blast volume and the disadvantage of becoming
available too late to affect the design of the blast hole pattern.
Nevertheless, the amount and type of explosive used and the
detailed initiation pattern for the blast can still be changed to suit
the properties of the ore defined to lie within the blast volume.
The drilling rate is an obvious parameter to indicate the
relative hardness of the ground to be blasted. While providing an
indication of relative hardness, the rate of penetration is affected
by many machine and operator dependent factors that render the
results loosely qualitative at best. Again, taking a lead from the
oil industry where the very large investment in each hole makes
it imperative to maximise the value of the information gathered,
considerable work has been done on monitoring and interpreting
information available from drilling data while the hole is being
drilled.
The application of detailed drill monitoring and the
interpretation of material properties from drill performance
advanced considerably during the 1990s (Peck and Pollit, 1995).
Further sophistication has been added by increasing the data
processing power on-board the machine and using sophisticated
pattern recognition systems to provide real-time advice and
information to the drill operator. Explosive charging details can
then be modified to suit the local geology (and its variation)
provided the relationship between blast design, run of mine size
and comminution performance is adequately understood.
251
Research is being conducted at the JKMRC (Segui, 1999) to

identify the intensity and character of rock mass structure, rock
substance strength and likely comminution properties from
drilling data. Figure 6 shows the concept.
It is recognised that vibration is generally developed in a drill
rig when the bit encounters a region of intense fracturing. It
should be possible to relate the nature and intensity of fracturing
to the dynamic response of the drill. Future research will address
the utilisation of information gathered from the attenuation of the
vibrations generated in the rock as drilling proceeds.
Data from geophysics

Remote geophysics has played a major role in minerals
exploration for many years. Gravity, magnetic and seismic
techniques have been responsible for the targeting of otherwise
hidden resources for proving using conventional drilling.
Near-field techniques such as natural gamma, focussed electric
resistivity, magnetic susceptibility, gamma-gamma and sonic
velocity logs have been successfully used to delineate and
characterise strata in the oil, gas and coal industries for many
years and are now finding application in the hard rock minerals
industry for mineral identification and grade estimation.
Work on the identification of minerals and evaluation of grade
falls under the banner economic geology and most development
effort has been applied in this area. Sophisticated tools such as
automatic strata identification through the statistical analysis of
multi-parameter logs are becoming available for routine
industrial use. (Schepers, 1996).
Geophysical techniques such as radar, seismics and radio

tomography are finding increasing application in on-going mine
development work, particularly in the definition of structure.
Wedepohl et al (1998) argue that under the right circumstances
these techniques have the potential to provide structural
information to assist medium- and short-term planning, but that
no single tool can achieve this under all conditions.
Wedepohl et al suggest that electromagnetic techniques are
typically applied over metres to tens of metres and have mapping
accuracies of centimetres to metres. Seismic techniques are
typically applied over tens of metres to hundreds of metres and
have a mapping accuracy in the order of metres. Electromagnetic
techniques depend on the rock mass having suitable electrical
properties, and seismic techniques are not effective in noisy
environments or highly structured ground.
Each technique will only map a feature that is associated with
a change in the relevant physical property. This usually means
that as associated physical property is mapped rather than the
geological feature itself. For instance, radar provides a reflection
image depicting boundaries associated with changes in radio
wave velocity which is controlled by the dielectric constant and
radio imaging depicts changes in electrical resistivity.
Problems of logistics and access and the need for
interpretation currently limit the routine application of these
techniques in mines. A satisfactory solution must be found in
each application for the inevitable trade-off between range and
resolution. It is therefore likely that structural information of
relevance to stability issues will be available before the more
detailed information necessary to describe blasting behaviour can
be routinely obtained using these tools.
Comminution
Index
Blastability
Index
Rock Mass Characterisation
for Blasting Purposes
Rock Characterisation
for Comminution Purposes
MWD + Image Analysis

PLT
PLT +Blasting
+ BlastingChamber
Chamber
HPB
HSB++Microscopy
Microscopy
Mapping
FaceFace
Mapping
Mill Input
parameters
Comminution
Comminution
optimisation
optimisation
Rock Fragmentation
and rock conditioning
Rock strength
+
Major joints
Data treatment
Process
Borehole image
acquisition
Physical action
Data Acquisition
Breakage Process
FIG 6 - Mine to Mill data capture being studied by the JKMRC (Segui, 1999).
252
Seismics appear to offer the greatest potential contribution for

the definition of structure relevant to blasting. Seismics can be
used to depict boundaries associated with changes in acoustic
impedance (density velocity). Seismic reflection surveys have
been successfully used to delineate suitable structure free blocks
for long wall mining of coal (Poole et al, 1998). Shear waves are
polarised by alligned structures and can be used to infer
alignment and fracture intensity (Urosevic and Hatherly, 1998).
Micro-seismics are now making important routine contributions
to monitoring seismic events and release of stress in deep mines.
Changes in rock mass condition through blast damage was
studied by Adamson (1992) using the cross-hole seismic
technique. The technique was successfully applied to quantifying
an increase in fracture frequency over time resulting from
near-by blasting operations. However, the target volume was
small and the experimental set-up was cumbersome rendering the
approach impractical for routine application. Because the
presence of a discontinuity in the rock substance will tend to
attenuate a seismic wave, it seems reasonable that the attenuation
of seismic signals offers the best prospects for quantifying
structures of relevance to blasting. Further research is warranted
in these areas.
CONCLUSIONS
The breakage behaviour of ores controls the cost of mining and
comminution. This is a very significant component of the overall
cost of production from most mines. It is now possible to
quantify the physical characteristics that control breakage
behaviour and use this information to manage fragmentation
from Mine to Mill. This provides the opportunity for each stage
of the production process to be managed to benefit the overall
enterprise, not just to optimise individual cost centres.
REFERENCES
Adamson, W R, 1992. Application of the Cross Hole Sesimic Technique
to the Quantitative Assessment of In-Situ Rock Structure, PhD
Thesis, University of Queensland.
The Australasian Institute of Mining and Metallurgy, 1998. Proceedings
Mine to Mill 1998 (The Australasian Institute of Mining and
The Australasian Institute of Mining and Metallurgy, 1976. Field
Geologists Manual, (Ed: D A Berkman) (The Australasian Institute
Barton, Lien and Lunde, 1974. Engineering Classification of Rock
Masses for the Design of Tunnel Support, Rock Mechanics, Vol 6,
No 4.
Bieniawski, Z T, 1974. Engineering Classification of Jointed Rock
Masses, Trans S African Inst Civil Engineers, 15:335-344
Bond, F C, 1952. The Third Theory of Comminution, Trans SME/AIME,
193:484-494.
Brook, N, 1985. The Equivalent Core Diameter Method of Size and
Shape Correction in Point Load Testing, Int J Rock Mech Min Sci
and Geomech Abstr, 22(2):61-70.
Brown, E T, 1981. Suggested Methods for Rock Characterisation, Testing
and Monitoring (International Society for Rock Mechanics).
Djordjevic, N, 1999. A Two-Component Model of Blast Fragmentation,
The AusIMM Proceedings, 304(2):9-13 (The Australasian Institute of
Gu, et al, 1998. Measuring and Modelling Mineral Liberation with the
JKMRC/Philips MLA, in Proceedings Minerals Processing 98,
Cape Town RSA.
Hadjigeorgiou, et al, 1995. Characterising In-Situ Block Size
Distribution Using a Stereological Model, Canadian Tunnelling
Journal, pp 210-211.
Harries, N, 1999. Estimation of In-Situ Block Size for Mine to Mill
Optimisation , Report to BHP Cannington Mine, December 1999.
Harries, N and Eadie, 1999. A Model to Predict Primary Fragmentation,
Report submitted to the International Caving Study, JKMRC, May
1999.
Kanchibotla, S, 1998. Importance of Mine Fragmentation on
Downstream Operations, in Proceedings Blast 98, IIR Conference.
Napier-Munn, et al, 1996. Mineral Comminution Circuits-Their
Operation and Optimisation, JKMRC Monograph No 2, University
of Queensland.
Palmstrom, 1995. RMi A Rock Mass Characterisation System for Rock
Engineering Purposes, PhD Thesis, University of Oslo.
Peaker, S M, 1990. Development of a Simple Block Size Distribution for
the Classification of Rock Masses, MA Sc Thesis, University of
Toronto.
Peck and Pollit, 1995. Lithological Recognition Based on Rotary Blast
Hole Drill Performance Monitoring, in Proceedings of Advances in
Equipment Performance Monitoring, Montreal, pp 59-71.
Poole, et al, 1998. Development of 2DE and 3D Seismic Methods for
Coal Mine Planning, in Proceedings 1988 Australian Mining
Technology Conference, Fremantle.
Preston and Sanders, 1992. Calculating Reserves A Matter of Some
Gravity, 26th Newcastle Symposium on the Advances in the Study of
the Sydney Basin, Newcastle, April.
Segui, J L B, 1999. Rock Mass Characterisation for Mine to Mill
Optimisation, PhD Progress Report, Dept of Mining, Materials and
Metallurgical Engineering, University of Queensland.
Schepers, R, 1996. Application of Borehole Logging to Geotechnical
Exploration, HANWHA Symposium, Seoul, Korea, October 1996.
Scott, et al, 1993. The Challenge of the Prediction and Control of
Fragmentation in Mining, 4th International Symposium on Rock
Fragmentation by Blasting, Vienna, July 1993.
Scott, et al, 1996. Open Pit Blast Design Analysis and Optimisation,
JKMRC Monograph No 1, JKMRC Australia.
Scott, A, 1996. Blastability and Blast Design, in Proceedings of Fifth
International Symposium on Rock Fragmentation by Blasting,
Montreal, August pp 27-36.
Scott, A, David, D, Alvarez, O and Veloso, L, 1998. Managing Fines
Generation in the Blasting and Crushing Operations at Cerro
Colorado Mine, in Proceedings Mine to Mill 1998, pp 141-148 (The
Urosevic and Hatherly, 1998. Analysis of Fracture Intensity and
Direction from Surface Seismic Surveys, in Proceedings 1988
Australian Mining Technology Conference, Fremantle.
Villaescusa, E C, 1991. A 3-D Model of Rock Jointing, PhD Thesis,
University of Queensland.
Wedepohl, et al, 1998. A Geophysical Toolbox for Deep Level Gold
Mining in South Africa, in Proceedings 1988 Australian Mining
Technology Conference, Fremantle.
White, D, 1977. Predicting Fragmentation Characteristics of a Block
Caving Orebody, MSc Thesis, University of Arizona.
253
Upgrade Ability and Geology of Cawse Nickel Ore

S M Denn1, C G Ferguson2 and S L Makin3
INTRODUCTION
The Cawse nickel project is a new generation pressure acid leach
(PAL), solvent extraction and electrowinning process used to
produce LME grade nickel metal and cobalt sulphide concentrate
containing 40 per cent cobalt by treating nickel and cobalt oxide
ores from a lateritic ultramafic profile. The project is located 50
kilometres NW of Kalgoorlie in Western Australia.
The commercial success of the Cawse nickel operation is
dependent in part upon the ability of the principal ore type to be
beneficiated. A simple process of rejecting lower grade material
on the basis of particle size achieves this. The process is called
upgrading and allows subgrade material to become commercial
ore grade and also allows higher ore grades to be processed in
the early parts of the operation. The success of the upgrade
process is dependent upon the physical and chemical properties
of the ore being processed and the nickel grade of the ore.
The upgrade ore type represents >80 per cent of the Cawse
nickel resource and is hosted mostly within a limonite unit,
represented by goethitic clays in the upper saprolite zone.
In the upgrade processing circuit the ore is crushed and
slurried. It is then passed over a series of screens that remove all
material over 0.5 mm. This ensures that coarse material is
removed from the main ore stream. Upgrade is reliant on
selectively rejecting lower grade or barren material, which can
include silica, magnesite and relict saprock.
To predict leach feedgrade, the grade control estimates the
head grade and upgrade percentage for all ore delivered to the
run of mine ore pad. To do this reverse circulation (RC) drill
holes are logged for ore type, assayed for a range of elements
and undergo simulated upgrade testwork.
The percentage and grade of material rejected through the
upgrade circuit has the most influence on upgrade per cent. This
reject is determined by the physical properties of the rock mass
and drum scrubber parameters. The correlation between nickel
upgrade and reject can be established to predict upgrade from the
known mass reject.
Parcels of upgrade ore milled to date closely follow projected
correlation trends. Upgrade circuit results have been excellent,
with average nickel grades upgrading by 34 per cent, cobalt by
23 per cent, with 55 per cent mass passing.
UPGRADE DEFINITIONS
Upgrade is possible due to the heterogeneity of nickel content
between the coarse and fine fractions of the mined material. The
following mass balance equation applies:
NiMinedX MassMined = NiRejected X MassRejected +
NiAutoclave feed X MassAutoclave feed
The terms used in this paper to define the upgrade are defined
in Table 1. Upgrade is measured as the ratio of autoclave
feedgrade to mined grade.
1.
MAusIMM, Chief Geologist, Cawse Nickel Operations, Centaur

2.
GAusIMM, Mine Geologist, Cawse Nickel Operations, Centaur

3.
Mine Geologist, Cawse Nickel Operations, Centaur Mining and

Exploration, PMB 32, Kalgoorlie WA 6433.
TABLE 1
Definition of upgrade terminology.
Term
Symbol
Definition
Head grade
Starting elemental grade.
Mass passing
% mass of material passing to the leach

circuit.
Feedgrade
Elemental grade of material passing to

the leach feed.
Upgrade
The percent increase or decrease in the

elemental concentration in the mass
passing compared to the head grade.
Reject grade
Elemental grade of the reject stream.
Reject mass
Recovery
Mass of the material in the reject

stream.
R
Amount of element recovered to the

leach feed compared to the original as
a per cent.
Material at Cawse falls into and is mined in three categories:

upgrade, grind and non-upgrade. The upgrade category is ore
with grade heterogeneity between coarse and fine material. The
upgrade ore is processed by progressively removing the less
mineralised coarser fractions. Non-upgrade ore is defined as ore
that does not upgrade under the current process, such as smectite
rich ores. Grind material is ore that requires pulverisation
through a separate crushing circuit prior to the autoclave leaching
process in order to economically recover the nickel and cobalt.
This ore type is essentially a hard, manganese-rich coarse
fraction.
LOCAL GEOLOGY
The Cawse mine geology is summarised in the following
paragraphs with emphasis on the key geological aspects related
to the upgrade characteristics. The Cawse geology is discussed in
detail by Bywater and Denn (2000) and also by Brand, Butt and
Hellsten (1996) and Hellsten, Lewis and Denn (1997).
Four main types of mineralisation have been identified at
Cawse. These are the oxide ores of limonite, talc, and siliceous
cobalt and the smectite-saprolite ore (locally termed nontronite).
The identification of these ore types, their grade ranges and
physical characteristics proved crucial in the subsequent
metallurgical sampling due to their differing metallurgical
properties.
Limonite mineralisation comprises approximately 75 per cent
of the global resource. Limonitic clays with variable proportions
of vuggy, goethitic silica typically contain grades of 0.4 per cent
to 1.5 per cent nickel. Nickel grade decreases as the proportion of
silica increases. Cobalt content is between 0.01 per cent and
0.3 per cent at which point manganese staining of the silica
occurs. The ore type is upgradeable by screening out the coarse
silica fraction. This ore type comprises the bulk of the upgrade
ore.
Siliceous
cobalt
(SICO)
mineralisation
comprises
approximately five per cent of measured resources. It occurs as
one metre to six metre thick subhorizontal layers in the upper
20 metres of the weathering profile. The ore is characterised by
255
S DENN, C FERGUSON and S MAKIN
Upgrade testing began following recognition of the difference in

grades between coarse and fine materials in samples. Initial
testwork was completed using RC drill chips and diamond core
using screen sizes between ranging between 0.038 mm and
12.5 mm. This sieve work showed that grade increased with
decreasing screen size in most samples tested with optimum
recoveries occurring in a 212 mm screen. Nickel was contained
in the minerals goethite, talc, chlorite and manganese oxides and
removal of coarse chalcedonic silica increased the ore grade.
Further tests using a simulated drum scrubber and samples
from large diameter drill holes (Denn, 1998) were utilised to
confirm the initial testwork. Results supported initial
observations that upgrade increases with decreasing nickel grade.
Testwork also concluded:
wet scrubbing using a rock charge would increase the
amount of fines passing through a scrubber and reduce the
amount of clay balling due to nontronitic clays;
increased scrubbing time reduced clay balling;
gravity separation was an effective way of further upgrading
the fine material; and
overcrushing decreased the upgrade ability of the ore by
pulverising low-grade chalcedonic silica.
Ore slurry is prepared in a 2250 mm diameter by 3700 mm

long drum scrubber, to which the ore feed is controlled by a
weightometer. Water addition to the scrubber can be varied or act
as a set ratio to the feedrate. A 25 mm trommel screen on the
scrubber discharge removes the larger oversize rock. This
material passes to the coarse reject stockpile via the coarse reject
belt. Material rejected from the trommel is sampled along this
belt.
Slurry passing through the trommel is pumped to a
double-decked vibrating screen in series. The first has an
aperture of 5 mm, the second has an aperture of 0.5 mm to
1.0 mm. The -25 mm to +5 mm washed fraction is rejected,
sampled and sent to the coarse stockpile via the coarse reject
conveyor.
The -5 mm to +1 mm screen is situated immediately below the
+5 mm screen. The ore is washed and the fine +1 mm fraction is
rejected to the fine reject stockpile. By the screen stage, 95 per
cent of the material to be rejected has been removed.
Desliming of the minus 0.5 mm fraction can be effectively
achieved by cycloning, with a cut at approximately 75 m.
Slurry passing through the 0.5 mm screen is pumped to a set of
250 mm cyclones, which remove the slime fraction (less than
75 m). This fine material is pumped with the ball mill cyclone
overflow through a trash screen to the leach feed thickener.
Testwork showed that gravity separation, with reject of the
heavy split, achieved the best nickel recovery and upgrade from
this fraction. This also provides the added benefit of removing a
large proportion of the chromite from the autoclave feed, thus
reducing the abrasiveness of the slurry.
Slurry underflow from the desliming cyclones is passed
through spirals, which remove material greater than 0.2 mm in
size by gravity separation. This intermediate size fraction, screen
oversize (>0.2 mm <0.5 mm) is dewatered on a low vacuum filter
and conveyed, together with the 0.5 mm screen oversize, to a
temporary waste stockpile. The fine (<0.2 mm) fraction from the
spirals is pumped to the ball mill cyclone circuit for further
classification.
The cyclone overflow slurry is pumped through a vibrating
trash removal screen, which has 0.7 mm apertures, into a 34 m
diameter high-rate leach feed thickener prior to entry into the
leaching circuit.
Each of the temporary reject stockpiles provides three days
storage. The mining contractor relocates the material to larger
waste and low-grade stockpiles, subject to grade.
UPGRADE PROCESS
GRADE CONTROL
The generalised ore process has the ore mined and stockpiled on
the run of mine (ROM) pad area from where it is fed into the
plant via the primary crusher. The ore undergoes crushing, and
washing, to produce a slurry of fine ore particles. The process
then separates the size fractions by screening and cycloning. In
general, the finer the screen size, the higher the upgrade and the
lower the retention of nickel. The optimum split size was
selected as 212 m, being the best compromise between nickel
recovery and upgrade.
Screening has been selected for removal of material above
0.5 mm. This is considered a practical lower limit, as screening
of finer sizes can lead to operational difficulties, particularly
clogging of the screens and balling of clays.
The ore is fed initially from six fingers on the ore pad (each
finger contains up to 40 000 dry tonnes of ore) into the primary
crusher (where it is sampled to report a crushed head grade) and
then to the main ore bin. The use of ore fingers is used to
facilitate reconciliation from the mine to the mill.
The ore is fed from the main ore bin along the drum scrubber
feed conveyor to the drum scrubber. The ore is sampled from this
conveyor to give the scrubber head grade.
The aim of the grade control drilling is to predict the leach

feedgrade of the ore sent to the autoclave. To achieve this an ore
block has head grade, upgrade and mass passing estimated to
predict the feedgrade and mass passing.
RC drill chips are logged as upgrade, non-upgrade or grind.
Talc and limonite ores (goethitic upper saprolite) are logged as
upgrade, and lower saprolite or nontronite/smectite rich material
is logged as non-upgrade. Manganese/cobalt rich material is
logged as grind.
In order to best predict upgrade through the current plant,
procedures have been put in place that mimic the process of the
plant, and define optimal upgrade parameters of the ore.
The data is obtained from RC grade control drilling
undertaken on a 10 m line spacing with 5 m hole spacing. The
drill spacing is reduced to a 5 m line spacing in areas requiring
additional definition. Drill samples are collected every one metre
and composited to a two metre sample. This sample is then sent
for assay and upgrade testwork as shown in Figure 1. Material
requiring upgrade is split before sample preparation. One split
(Sample one) is processed as normal being assayed for nickel,
cobalt, manganese, aluminium, magnesium, iron and chrome.
dark blue to black siliceous rock with abundant cobalt-rich

manganese oxides such as todorokite, chalcophanite and
cryptomelane, which carry cobalt grades from 0.3 per cent to
seven per cent and nickel grades from 0.5 per cent to five per
cent in the rock mass. The ore is not upgradeable due to the
association of nickel and cobalt to manganese, which is then
associated to coarse chalcedonic silica.
Talc mineralisation comprises approximately ten per cent of
the global resource and is located throughout the weathering
profile proximal to penetrating talc-chlorite shear structures. It
typically contains high nickel grades, generally between one per
cent and two per cent and high magnesium levels between eight
per cent and 15 per cent.
Iron and nickel enriched smectite clays and saprolite form a
minor constituent of the Cawse mineralisation. They are
generally higher in aluminium (>3 per cent) and also magnesium
(>5 per cent). A minor ore type is alluvial ore which is also
non-upgradeable due to its high aluminium (> 4 per cent).
UPGRADE TESTWORK
256
UPGRADE ABILITY AND GEOLOGY OF CAWSE NICKEL ORE
FIG 1 - Upgrade equations and flowsheet for grade control testwork.
The result grade is used as the head grade. The other split
(Sample two) is weighed, before being placed in a bottle roll
with water and rolled for five minutes. The slurry produced is
then passed through a 0.5 mm screen where the reject material is
collected, weighed and assayed for cobalt and nickel. This is the
equivalent of the material rejected from the plant circuit.
Barrel rolling is undertaken as it is considered more precise
than wet sieving alone. It agitates the sample, breaking down
indurated clays, clay balls and weakly competent upper saprolite
material. All material submitted for upgrade is placed in the
bottle roll. The sieving of samples through 0.5 mm post bottle

rolling involves around 30 seconds of washing with hose to pass
fine fraction. No material is physically pushed through the
sieve.
Although initial testwork found a 212 m screen as an
optimum screen size for upgrade, grade control utilises a 0.5 mm
screen that provides a robust test and good reconciliation to the
current plant configuration. Wet sieving, compared with dry
sieving, breaks down indurated clay and oxide particles
(increasing the upgrade and amount of mass passing a screen).
257
The amount of water added to each sample is equivalent to the

sample size. Any wet samples that require drying and then
crushing are crushed to a max of 10 mm. The current best
practise for screen upgrading testwork as used for all grade
control samples and on-going exploration work is detailed in
Figure 1.
Some differences between the plant process and grade control
testwork are:
No spiral and cyclones. The reject from these process steps is
of minimal mass, but higher grade.
RC drill chips are crushed to a finer grainsize than mined ore
allowing more material to pass through the screen and
reducing the upgrade in grade control samples. Crushing
size, however, makes little difference to the upgrade results
in RC samples.
Upgrade testwork is normally conducted every fourth line of
RC drilling. This provides a sufficient density of drilling to
enable upgrade calculations on most ore blocks while keeping
analytical costs to a reasonable level.
The predicted autoclave feedgrade can then be back calculated
from the original mass of sample two, the mined grade (from
sample one) and the reject mass and grade after screening.
The calculation method of upgrade is detailed in Figure 1.
Reproduceability of data
Testwork to confirm the ability to reproduce results from the
upgrade test procedure have been completed using original drill
cuttings stored in green plastic bags (1 m per plastic bag). These
were resubmitted as the same 2 m intervals after they were
placed through a splitter (to maintain the representative nature of
the sample).
The results showed that the upgrade testwork is reproducible
for nickel and cobalt head grades, reject grades, passing grades,
mass passing, upgrade and recovered metal.
ANALYSIS OF UPGRADE RELATIONSHIPS

A plot of nickel upgrade versus mass passing is shown in Figure
2. The graph shows a scatter plot of nickel upgrade versus
percentage mass passing of a total of 5659 samples used from
grade control RC samples from Bunyip and Phoenix open pits. A
well defined maximum curve occurs with a scatter of data points
below the curve. Using this graph we can illustrate the basic
mathematical concepts of the upgrade process.
Reject
Upgrading
Ore
Sampling procedures
Feed (to auto clave)
Sampling of the RC drill cuttings is critical for the successful

evaluation of the upgrade characteristics of the ore. To ensure
good sample collection the following procedures are used.
all samples are taken from face sample RC drilling;
no water injection is allowed and holes are drilled dry (there
is no formation water);
sample collection systems on both grade control rigs and
exploration rigs are designed to use dust collectors to collect
as much of the fine material as possible; and
all laboratory work is carried out using a standard procedure.
In simplest terms the screening upgrade process takes in one

ore stream and splits it into two streams.
The governing equation for calculation of upgrade is found to
be:
upgrade =
feed grade
1
mass passing.(feed grade reject grade) + reject grade
FIG 2 - Theoretical nickel upgrade versus mass recovered, and actual grade control data.
258
Each stream contains a certain percentage of an element of

interest (eg nickel). Let us call these h (for head grade), r (for
reject grade) and f (for feedgrade) (per cent w/w). The combined
mass flowrate of the reject and feed streams equals the mass
flowrate of the input stream. We shall call this M (mass/time). If
we use x to represent the fraction of mass passing to the feed
stream a mass balance can be performed:
mass rate of nickel in = mass rate of nickel out.
This factor is a measure of the inhomogeneity of the nickel

distribution. It ranges from U = 1 (minimum rejected grade) to
U = 0 (homogeneous distribution) and U<1 (higher grade in the
rejects).
Thus Equation 3 becomes:
h . M = f . M . x + r . M . (1-x)
It is possible to characterise different ore types by their

U range. The average of all the data points of the grade control
data gives a U value of 0.54 while the moving average trend line
as shown in Figure 3 has a U value of 0.75 which increases to 0.9
at 30 per cent mass recovered. This indicates that the grade
inhomogeneity is not constant for the data set and is composed of
at least two populations.
or
h = fx + r(1-x)
(1)
The percentage upgrade y is the difference in nickel grade

between the feed and ore streams, as a fraction:
y=
1
1
U( x 1) + 1
(4)
f h
h
y=
or
y=
f
1
h
(2)
Combining Equations 1 and 2 gives:

y=
f
1
x( f r ) + r
(3)
As head grade (h) is a function of reject and feedgrades (r, f)

and mass passing (x), upgradeability (y) is a function of f, r and x
(ie y = y(f, r, x) ). These are geologically determined parameters.
Some bounding conditions can be found using Equation 3. The
best possible upgrade will be achieved when no nickel is rejected
(ie r = 0).
with r = 0:
f
y=
1
fx
y=
1
1
x
This defines the upper limit of data from upgrade testwork as

shown in Figure 2. When there is no inhomogeneity between the
reject and feed streams (ie f = r) there will be no upgrade.
ie with f = r:
y=
r
1 = 0
r
If the rejects carry more grade than the feed stream (r > f) a
downgrade will result.
r > f:
f
<1
x( f r ) + r
y<0
To display more detailed behaviour, simplifications must be
made to reduce the function to two variables. Feed and reject
grades can be expressed as a single factor (U) to simplify the
equation.
U =1
r
f
FIG 3 - Nickel upgrade versus aluminium, magnesium and iron,

grade control data trends.
UPGRADE CHARACTERISTICS OF
ORE TYPES-RELATIONSHIP BETWEEN
LITHOLOGY, MINERALOGY AND UPGRADE
Upgrade is dependant on two things, the heterogeneity in grade
between coarse and fines and the relative amounts of coarse and
fines.
Figure 2 shows the relationship between upgrade and mass
passing while Figure 3 shows the relationship between upgrade
and each of the elements iron, aluminium, and magnesium. The
data in Figure 2 falls into two general trends. The first trend
(U>>0), where upgrade increases with decreasing mass passing,
occurs because there is a distinct heterogeneity in grades
between coarse and fine material. The second trend (U<=0),
where upgrade is poor or there is no upgrade regardless of mass
passing, occurs where the coarse and fines have similar grades.
Upgrade is relatively poor for material low in free silica or
high in aluminium (>3.5 per cent aluminium), magnesium
(>4 per cent magnesium unless talc-rich-8 per cent), or iron (25 30 per cent iron) as illustrated in Figure 3. Upgrade is less
dependent on the amount of iron because iron does not occur as a
silicate (nontronite) in large quantities at Cawse. The effect of
removing aluminium and magnesium rich material from the data
set, however, is an increase in average nickel upgrade from
30 per cent to 45 per cent. Upgrade within material with greater
than three per cent aluminium is less than 20 per cent. Average
upgrade in material with greater than four per cent magnesium is
also less than 20 per cent. These materials represent alluvial and
lower saprolite material respectively. However, material with
little or no magnesium or aluminium upgrades greater than
40 per cent. In this case, the nickel upgrade is dependent on the
amount of silica in the rock.
The extremely high upgrades observed where iron is <10 per
cent is due to the high silica content of these rocks as there is an
inverse relationship of silica and iron.
259
The low aluminium content of the protore at Cawse seems to

limit the amount of smectite formed, and the rapid breakdown of
the protore to goethite, kaolinite and silica means that upgrade
ore makes up almost 70 per cent of the profile at Cawse.
The multi-element data above relates well to lithological units.
High aluminium and iron values occur in the mottled alluvial
clays where the mineralogy consists of kaolinite, haematite,
goethite and little free silica. Pisoliths are the main constituent of
the coarse reject, and they make up only a small proportion of the
alluvium.
Kaolinite, which does not contain significant nickel, and
haematite which contains little nickel both pass into the
autoclave feed reducing the grade of the passing fine material.
The result is a small amount of rejected coarse material and very
little difference in grade between coarse rejected material and
fine autoclave feed material.
Ore containing high magnesium is found in talc zones and in
the lower saprolite. Mineralised lower saprolite material consists
of serpentines and smectitic clays. Little iron oxide material is
found except along grain boundaries and in fractures. Nickel is
contained within smectite, iron oxide and to a small degree in the
serpentine. The coarse reject material consists of blocks of
saprolite (serpentinite) and smectite clays, which tend to absorb
water and ball up, creating an artificial oversize. Smectites,
goethite and minor serpentine also make up the fine material
passing through to the autoclave. The rejection of blocky
non-mineralised saprolite means that this material can be
upgraded, because coarse saprolite is generally slightly lower in
grade than the finer material, and this does help the upgrade of
saprolite ore slightly. However, the balling smectite clays and
high viscosity of the material make good upgrade results difficult
to accomplish within the current plant design.
Talc zones, limonite zones and silica zones all contain free
chalcedonic silica, silicified saprolite and nickel bearing minerals
such as talc, chlorite or goethite. Barren silica (nickel less than
0.4 per cent) forms the bulk of the reject, with only minor
goethitic saprolite found. The huge difference in nickel content
between silica and nickel bearing minerals such as goethite and
talc means that the nickel grade will increase almost
proportionally to the amount of silica rejected.
Manganese oxides are associated with massive chalcedonic

silica and form as coatings on the walls of cavities. During the
upgrade process they are, therefore, rejected along with the
coarse material they are associated with. This can even lead to a
downgrade in nickel and especially cobalt grades as these
elements are concentrated with manganese oxides. Values as low
as one per cent manganese generally indicate that a sample will
not upgrade.
Cobalt-rich, manganese oxide ores are processed by crushing
and grinding all fractions of the ore as Grind ore. Saprolite ore,
alluvial ore and ore rich in smectite are all stockpiled as
non-upgrade ore and will not be processed with the current plant.
RECONCILIATION
A concern with grade control upgrade testing was its
effectiveness in reproducing the upgrade process occurring
through the plant. Re-sampling and testing of ore from Bunyip
pit has shown that grade control upgrade testing is repeatable and
representative.
Of the total (upgrade) ore processed, upgrade was predicted to
be 28 per cent for nickel with an autoclave feedgrade of 1.72 per
cent nickel. The plant produced an upgrade of 33 per cent for
nickel with an autoclave feedgrade of 1.70 per cent nickel. The
moderately higher upgrade is due to a slightly lower than
expected ROM grade. Lower grade material generally contains a
higher coarse fraction and upgrades are therefore higher.
In this way, upgrade is capable of compensating for the effects
of dilution. Coarse, lower grade material will not effect autoclave
feedgrades if included in higher-grade blocks. In Finger BD-001,
for example, high dilutions and a correspondingly lower ROM
grade (1.17 per cent nickel) were encountered, yet upgrading
increased the autoclave feedgrade to 1.65 per cent nickel. Further
testwork showed that material with nickel grades as low as 0.3
per cent nickel could theoretically produce an autoclave
feedgrade of 1.5 per cent albeit with a low mass recovered of ten
per cent.
Compared to the predicted upgrade for the stage 1 feasibility
study, the reconciliations show that the plant does not achieve the
expected outcomes. This is shown in Figure 4, where the upgrade
FIG 4 - Nickel upgrade per cent versus Ni head grade from mill production data to the end of December 1999.
260
trend for the feasibility study is shown, along with all the daily
production upgrade results. Three additional trends are shown,
for each pit mined to-date. The results show that with a head
grade of 1.4 per cent nickel the plant averages 34 per cent
upgrade compared with the initial testwork of 40 per cent. At 1.1
per cent nickel head grade the Bunyip pit trend crosses the
predicted trend and indicates (from a limited data set) that that at
0.8 per cent nickel head grade a 100 per cent upgrade is possible
to a 1.6 per cent autoclave feedgrade. The Orc pit trend follows
the Bunyip pit, but is biased lower due to its ore being blended
with ore from Phoenix pit. The Phoenix pit trend shows an ore
type that is non-upgrade due to it being homogenous, containing
no coarse chalcedonic silica component and being high in
aluminium, magnesium and manganese.
CONCLUSIONS
The screening upgrade method provides the Cawse Nickel
Operation with a cost-effective and flexible way to provide an
increased grade of nickel through the process plant. The process
is simple and relies on the separation of coarse lower grade
material from a higher grade fines fraction. The process uses
trommels and screens to achieve this separation. The upgrade
value of the ore is successfully predicted by using a simple
analytical process using standard RC samples. The different
upgrade characteristics of samples are related to different ore
types, and their mineralogical associations. The upgrade data has
been shown to follow a simple mathematical relationship that can
be used to help predict long-term trends in the data for future
studies on plant expansion options.
ACKNOWLEDGEMENTS
The authors would like to acknowledge the assistance of Centaur
Nickel Operations in allowing the publication of this paper and
also the professional contributions made towards understanding
the Bunyip deposit by geologists from Centaur Mining and
Exploration, in particular Ken Hellsten, Toby Morris and Jenna
Shuttleworth. We acknowledge the many other discussions held
with other personnel who have greatly added to the
understanding of the geology of the Cawse nickel deposit.
REFERENCES
Brand, N W, Butt, C R M and Hellsten, K J, 1996. Structural and
Lithological controls in the Formation of the Cawse Nickel Laterite
Deposits, Western Australia-Implications for Supergene Ore
Formation and Exploration in Deeply Weathered Terrains, in
Proceedings Nickel 96, pp 185-190 (The Australasian Institute of
Bywater, A and Denn, S, 2000. The Bunyip Lateritic Nickel-Cobalt
Deposit, Cawse Nickel-Cobalt Operations, Kalgoorlie, Western
Australia, in Proceedings 4th International Mining Geology
Conference, pp95 - 104 (The Australasian Institute of Mining and
Denn, S M, 1998. Large Diameter Drilling at Cawse Nickel Project- An
Innovative Solution to Metallurgical Sampling, AIG Bulletin
22:61-66.
Hellsten, K J, Lewis, C R and Denn, S, 1997. Cawse Nickel Cobalt
Deposit, in Geology of the Australian and Papua New Guinean
Mineral Deposits, (Eds: D Berkman and D Mackenzie) pp 335-338
261
The Interaction Between Geology, Mining and Metallurgy at

Stawell Gold Mines Pty Ltd
D Fredericksen1
ABSTRACT
Stawell Gold Mines Pty Ltd operates two underground decline mines
within the township of Stawell. An active exploration program and
commitment by owners to improving the ore reserves has led to an
improved mine life and reduced operating costs since the purchase of the
operation by Mining Project Investors and Pittston Mineral Ventures in
late-1992. Over the past six years of operation a cumulative 1 150 000
ounces of gold have been added to the mineral inventory. An improved
geological understanding of the orebody has evolved over time increasing
the ore reserves and allowing the development of long-term plans for the
mines future. This has enabled important capital decisions to be made
providing the impetus for improved productivity and thereby reduced
operating costs. Co-operative mine planning and mine geology utilising
best practice geological systems have enabled the effective extraction of a
structurally complex and mineralogically varied deposit within a rural
Victorian township. Detailed metallurgical research has improved the
function of an aged and poorly designed circuit to boost the mill capacity
whilst also improving the metallurgical recovery. The combination of
metallurgical expertise and an understanding of the geological setting
enable an effective control of metallurgical recovery.
INTRODUCTION
Stawell Gold Mines Pty Ltd operates underground decline mines
at the Magdala and Wonga deposits. These are the two largest
orebodies of the Stawell Goldfield and lie within ML1219 which
is along the eastern margin of the city of Stawell (Figure 1),
245 km NW of Melbourne Victoria.
Stawell Gold Mines Pty Ltd (SGM), a wholly owned
subsidiary of Mining Project Investors Pty Ltd (MPI), is the
manager of a joint venture between Mining Project Investors Pty
Ltd and Pittston Mineral Ventures. SGM operates the Stawell
Gold Mine and conducts exploration on surrounding tenements.
Mining Project Investors Pty Ltd is a private exploration and
development company with assets in Australia and the USA. The
purpose of this paper is to demonstrate how an active exploration
program and commitment by owners to improving the reserves,
has lead to an improved mine life and how in an operation such
as this continued co-operation amongst mining professionals is
required.
HISTORY
Alluvial gold was discovered near Stawell in 1853 (Clappison,
1965) and an estimated 0.8 million ounces of gold was won from
the alluvial leads in the period 1853 to 1860. Production from the
leads waned in the early-1860s and ceased in 1912 (Watchorn,
1986).
In 1855 gold hosted within quartz reefs was discovered at Big
Hill. High-grade ore (>15g/t) was mined up until the early-1880s
from the Sloanes Flat, Cross Vertical and Scotchmans
Flats/Vertical reefs. As production from these systems declined,
payable ore was discovered within shear zones to the east and
further to the south in metamorphosed schists, which form the
basis of the current production from the Magdala and Wonga
mines respectively. It is estimated some 1.9 million ounces of
gold was produced from underground before the closure of the
1.
MAusIMM, Chief Geologist, Stawell Gold Mines, Leviathan Road,

Stawell Vic 3380.
FIG 1 - Location of the Stawell Goldfield and Victorian geological zones,

after Cayley and Taylor, (1996).
mines in 1926. The main reason for the decline in production

was poor recovery, as much of the gold within the shear zones
was associated with sulphide minerals.
263
D FREDERICKSEN
Production and resource figures for the goldfield are

summarised in Table 1. Historic production amounts to some 2.7
M oz gold with recent production and current reserves and
resources giving a total endowment for the field of approximately
4.7 M oz. A further 1.0 M oz of gold has been produced from the
nearby Ararat alluvial goldfield with lesser and poorly
documented amounts of gold produced from smaller mines
worked in the Stawell area.
TABLE 1
Stawell Goldfield production resources and reserves.
The reserve and resource discovery since acquisition of the

project by Mining Project Investors is shown in Table 2. In all
mine development and exploration has added 1.2 million ounces
of gold resources (including production and addition to proved
and probable reserves) at a discovery cost of approximately
$11.00 per ounce (all exploration costs included).
During the calender year 1999, Stawell Gold Mines produced
more than 94 000 oz of gold and maintained proved and probable
reserves able to sustain a further four to five years of
underground production. Long-term plans incorporating current
inferred resources from underground and recently delineated
surface resources fulfill production requirements for a further ten
years of mine life.
Production
Ore Source
Gold Produced
(oz)
Historical
- Alluvial
- Underground
Recent (WMC - SGM)
- Underground and Open cut
- Tailings Sands
770 000
1 900 000
(1853 - 1912)
(1855 - 1926)
(1984 - Dec 1999)
(1984 - Dec 1999)
920 000
16 000
Total Production
3 606 000
Ore Reserves And Resources
As at (Dec 1999)
Ore source
Contained gold
(oz)
Magdala
- Proved and probable
- Inferred
Wonga
- Proved and probable
- Inferred
Surface Resources
417 000
402 000
1 097 180
Total Goldfields
4 703 180
The unique structural and geological setting of the Stawell gold

deposit and the intense deformation of the volcanic and
sedimentary package is a direct result of the complex history of
the western margin of the Lachlan Fold Belt (Figure 1). East
west compression of the Lachlan Fold Belt during the Ordovician
against the Pre-Cambrian Adelaide Fold Belt has uplifted a
multiple thrust-stacked sequence of basalt flows and sea floor
volcanogenic sediments from the base of the stratigraphic pile
(Fredericksen and Gane, 1998). The Stawell Corridor between
the Coongee Fault and the Moyston Fault (Figure 2) represents a
structural high within this sequence, associated with over five
million ounces of gold endowment along the exposed 50 km
strike length. The belt is obscured by basalt cover south of Ararat
and by Murray Basin sediments to the north of Stawell.
MINE SCALE GEOLOGY AND EXPLORATION
28 804
9300
226 000
Total Reserves and Resources
REGIONAL GEOLOGY
Period
The current mining operation is based on the Magdala Decline

Mine, which was commenced in 1981 and has now reached a
depth of 750 m below surface. In addition the Wonga mine, was
commenced in 1984, as an open cut on near surface ore
structures within the contact aureole of the Stawell Granite,
approximately 2 km south of Magdala. A decline was
commenced from the bottom of this pit in 1985 and is currently
inactive and will not be discussed further.
TABLE 2
Resource inventory discovered December 1992 to December 1999.
Magdala
Wonga
Resource and Reserve (Dec 1999)
Ounces
423 000
Mine
Production
Tonnes
Grade
Ounces
1993
488 268
4.40
1994
380 703
4.51
1995
449 234
1996
495 804
1997
510 218
1998
1999
Total
Total
Ounces
Ounces
183 466
606 466
Tonnes
Grade
Ounces
69 072
94 507
4.60
13 977
55 202
176 972
4.78
27 197
4.45
64 272
166 293
4.74
25 342
4.86
77 471
186 558
5.22
31 309
4.22
69 224
82 601
4.92
13 066
668 268
4.66
100 122
738 138
4.70
111 540
3 307 347
4.55
546 903
706 931
4.88
110 892
657 795
Resource and Reserve (June 1999)
822 000
53 000
875 000
Underground ounces discovered
954 903
(19 574)
926 329
Surface ounces discovered
226 000
Total ounces discovered
264
1 152 329
THE INTERACTION BETWEEN GEOLOGY, MINING AND METALLURGY
FIG 2 - Geological cross-section of the Stawell Corridor looking N. Modified after Cayley and Taylor, (1996).
Ore from the Magdala mine is produced from a series of

subparallel lodes hosted by faults and shear zones on the western
flank of a large basalt antiform the Magdala Anticline (Figure
3). These lodes, the most important of which are Central lode
and the Basalt Contact lodes, are intimately associated with an
intensely deformed package of volcanogenic sedimentary rocks.
Historical production as previously discussed was predominantly
from high-grade Hangingwall Reefs, hosted by the Stawell
Fault in psammo-pelitic rocks stratigraphicly overlying the
volcanogenics. The Stawell Fault is a large regional scale reverse
fault, which locally forms the contact between the footwall
Mine Schist and the hangingwall Wonga Schist.
Mineralisation within the Magdala mine is related to a regional
tectonic event termed D4 by Watchorn (1986). The major
structural event related to D4 other than the Basalt anticlines is
the development of the Central lodes shear system (striking at
330 - 340 and dipping 50 - 80 SW) which overprints earlier
D1 - D3 events. The Scotchmans Fault and South Fault are both
D5 and effect offsets in the mineralisation. Recent diamond
drilling beneath the South fault has established that the Magdala
mineralised system is offset by the South Fault and that the
Magdala system as we know it represents an upper block which
has been thrust upward and south on the South Fault. Diamond
drilling of this target continues to outline the geometry,
mineralisation and economic potential of the block beneath the
fault.
Ore shoots developed within the Magdala Lodes generally
plunge steeply north, but are constrained within a moderate
northerly plunging corridor bounded by the Scotchmans Fault
and the South Fault (Figure 4). It is within this corridor that all
recent production activity has been focussed and all of the
resource inventory gains achieved and currently contains nearly
all of the resources outlined to date. Exploration during 1997 and
1998 of the near surface potential of this corridor identified
significant resources suitable to open cut mining. Feasibility
studies and environmental effect statements have been prepared
with the aim of being able to bring this into production in the
coming year.
A deep surface diamond drilling program aimed at identifying

the resource potential below -900 m RL has recently been
completed. The targets for these holes were up to 500 m further
north and 200 m vertically below the nearest underground
drilling information and were developed from structure contours
of underground drilling and mapping information. They were,
despite the distances involved, very close predictions of the
position of the major surfaces identified, Scotchmans Fault,
Central Lode and the South Fault. Intersecting the targets to
within 20 - 50 m of plan coupled with strong mineralisation in
several of the intersections shows very good continuity and
predictability of the mine scale geology. This predictability
becomes and issue on the smaller stope scale where orebody
geometrys and grade can change dramatically within metres
along strike or up and down dip.
Other exploration targets around the Magdala Anticline but
outside mine corridor include:
1.
Mariners system above the Scotchmans Fault, including

down plunge extensions of the Big Hill mineralisation.
2.
Mineralisation on the East Flank of the Magdala anticline,

where some significant volcanogenic hosted intersections
have been produced.
3.
Below the South Fault for a possible repetition of the

entire Magdala System, as discussed previously.
MINE DEVELOPMENT
The Stawell operation utilises modern mining methodologies. All
of the underground ore is hauled to the surface in 50 tonne haul
trucks via a conventional 5.5 m by 5.0 m decline with variable
gradients from 1:9 to 1:8. Prior to mid-2000 the Magdala decline
consisted of a north and south branch below -468 m RL (Figure
4). As a result of a shorter strike length the two decline branches
will merge at -786 m RL. A fleet of ELPHINSTONE R1700,
R2800 and R2900 LHDs tram ore from the stopes and load the
haul trucks. During the calender year some 730 000 tonnes of ore
265
D FREDERICKSEN
FIG 3 - Geological cross-section through Magdala mine looking NW.
grading 4.7 g/t were hauled to the surface. In addition to this a

further 350 000 tonnes of mullock and low-grade ore were
handled either underground as stope fill or to low-grade
stockpiles on the surface. Increasing annual underground
production has been progressively achieved despite the cessation
of mining the Wonga orebody and increasing depth in the
Magdala mine (Table 2).
266
Current development strategies involve levels spaced at

nominally 19 m backs to backs intervals in the ore. Very little if
any level development is placed in waste except for creating
ventilation links. A wide variety of orebody geometrys exist
over the 1.0 km strike length of the orebody with widths varying
from <1.5 m to greater than 15.0 m. These range in dip from
45 degrees to vertical and appropriate development strategies are
required to maximise the extraction of ore.
THE INTERACTION BETWEEN GEOLOGY, MINING AND METALLURGY
FIG 4 - Longitudinal projection of the Magdala mineralised system showing major stuctural features, resource and mined ore blocks, decline development
and exploration drill results and targets.
Over the past five years handheld stoping methods have

largely been phased out of the operation and significant emphasis
placed in improving the accuracy and reliability of the longhole
open stoping methods. Much of the ore is fired from
conventional uphole retreat stopes in panels up to 30 m in strike
and ranging in tonnage from 2500 - 50 000 tonnes. Other areas
involve more complex stoping plans in which ore is mined from
stopes extending over 3 to 4 levels. Cemented rock fill has
recently been introduced to the stoping program as a way of
optimising the recovery of the resource. To measure performance
of the stoping process and to accurately capture final void
geometries for filling, future firing and mine design all stope
voids are surveyed with a cavity monitoring system.
Geotechnical ground conditions vary significantly and range
from very good to very poor. The development cycle utilises
resin bolts and mesh with cable bolts introduced for intersections
and as hangingwall support in the stoping cycle.
METALLURGY
Ore is treated in a plant initially built to treat tailings sands and
oxidised open pit ore. Significant upgrades over time have seen a
plant evolve with the capacity to treat 800 000 tonnes of
hardrock feed. This coupled with an improved understanding of
the metallurgical characteristics of the ore feed has boosted gold
production substantially. Campaign treatment of different ore
streams based upon geologically distinct ore types was
introduced in 1993 and significantly enhanced gold recoveries
and enabled cost reduction strategies to be implemented.
All ore types currently pass through a two stage crushing
circuit before being ground in a ball mill to nominally 95 per
cent passing 100 microns. The cyclone underflow passes through
Knelson concentrators to recover any coarser free gold before

being added back to the mill for further grinding. At this stage of
the process differing ore types are managed in different ways
Central Lode ore is treated through a floatation circuit with the
floatation concentrate taken to a stirred ball mill to reduce its size
to <20 microns before being added back to the leach circuit.
Under conventional CIP treatment this ore will exhibit
metallurgical gold recoveries ranging from as low as 50 per cent
up to 88 per cent. Much of the gold is associated with pyrite and
is also affected by moderate to high amounts of graphite, which
robs gold from the pregnant solution. The fine milling coupled
with carbon management strategies maintain gold recoveries for
this ore in excess of 80 per cent.
Basalt Contact ore is taken to the leach circuit direct from the
ball mill discharge. Much of the gold in this ore is free milling
and is associated with arsenopyrite and pyrrohtite with minor
amounts of pyrite. Coarse gold recovery in the Knelson
concentrators ranges from 25 to 50 per cent of the final gold
recovered which is generally 88 to 92 per cent of the feed.
Stockworks ore from irregular zones of mineralisation either in
the hangingwall of the basalt lodes or from the footwall of
central lode exhibit varied metallurgical response and requires
either of the two treatments identified.
Previously the ore mined from underground at Wonga
exhibited recoveries in the high 75 - 80 per cent range. Gold in
this ore was locked within fine grained arsenopyrite crystals
which were concentrated well in the floated circuit. Subsequent
fine grinding to less than ten microns improved overall
recoveries to >80 per cent.
An additional benefit of campaign treatment is the ability to
stockpile and treat stopes separately to gain additional geological
reconciliation information. Where possible this is carried out.
267
D FREDERICKSEN
MINE GEOLOGY
SUMMARY AND CONCLUSIONS
An on-going program of Resource Definition drilling is

completed some two years in advance of mining. As the orebody
envelope plunges 30 degrees to the north and is up to 1.0 km in
strike length, specifically placed drill drives in the hangingwall
and extending well north of the orebody are required. To-date
two of these drives have been completed with the third to
developed during 2000. NQ diamond drill holes are nominally 30
m spaced and enable the estimation of an indicated resource
model in MINESIGHT and MEDSYSTEM. Preliminary mine
designs are completed and decline designs committed based on
this data.
An enhanced ACQUIRE database system developed by
Metech Pty Ltd to handle this drilling data has been established.
This also caters for all production and exploration drilling
information, which represents some 30 000 - 35 000 metres of
underground diamond drilling annually. To aid logging and data
capture, geologists enter data via a barcode logging system. A
geological legend developed over the life of the project
incorporates detailed geotechnical and detailed point data
descriptions. Standard SURPAC data extraction macros enable
all of the geological and geotechnical data to be plotted in any
orientation. It is important to be able to extract the data in a
number of different grid sets as the orebodies locally rotate
significantly and working perpendicular to the strike enhances
the geological interpretations.
Further production diamond drilling to nominally 10 m centres
is required prior to the final commitment to ore development.
This function along with the estimation of a measured resource is
the function of the mine geologists and is an on-going process
until final stope mining is committed.
In the Central Lode orebodies development is guided by
survey control determined from the closer spaced diamond
drilling. In the Basalt Contact lodes some development is under
survey control whilst others are guided by daily geological input.
All development faces are mapped and sampled and followed up
with mapping of the backs of the advancing development. This
mapping also collects sufficient detail to aid in the design of
appropriate ground support for the final stoping designs. A
system of capturing the face samples in a spatial database has
been developed and further enhanced by capturing face and
backs mapping information in true 3D space for extraction to
plans, cross-sections and for use in MINESIGHT for on screen
geological interpretations and modelling.
Over the past five years geostatistical and three-dimensional
resource modeling techniques have progressively been integrated
into the geological functions. It is now at the stage where nearly
all mine design work can be carried out utilising a detailed
geological model in real 3D space.
In conjunction with metallurgy and mining personnel the mine
geologists ensure that all material to be mined in upcoming
production schedules has some form of metallurgical testwork
carried out. Over the past five years the recovery estimation has
primarily been carried out on development face samples and final
stope drilling samples. This reactive process has been replaced
by procedures that identify samples from the various phases of
diamond drilling for detailed metallurgical analysis. An onsite
facility conducts leach tests and characterises the ore relative to
its ability to act as a preg robber in the treatment circuit and its
amenability to floatation. A site team assess the information as it
comes to hand and provide the appropriate feedback to the mine
planning, metallurgical and geology groups.
The Stawell Goldfield has an extensive history and a significant

gold endowment. Recent exploration success has continued to
add to the endowment of the field and given time for the mining
operation to continue to modify its operating practices in the
mining, metallurgical and geology areas. The evolution of
operating practices has occurred in an operating environment in
which underground ore production and mill throughput have
been increasing.
Geologically the Magdala orebody is unique and significantly
different to other Victorian gold deposits. On a gross scale it is
predictable as demonstrated by the deeper drilling recently
completed. As the scale decreases back to stope size blocks it
becomes less predictable and geometries and grade distributions
vary significantly. This in itself necessitates a high level of
geological involvement in the mining process and a need to
collect information well in advance of the mining process. As the
orebody plunges away from the access development significant
capital commitment is required in appropriately placed
development from which drilling can be completed.
Metallurgically the orebody poses some interesting challenges
exhibiting preg robbing and sulphide associated refractory gold.
To-date significant improvements in overall recovery of gold
through the circuit have been achieved and research work
continues aimed at reducing the effect of the preg robing and
recovering more of the sulphide associated gold.
Systems for data collection are required that provide ready
access to geological information and have portability across mine
planning and modeling software packages. For such a large and
diverse deposit three-dimensional modeling has been a challenge
but once instigated has shown benefits to the design and
scheduling process.
The geology functions from exploration through to day to day
grade control operations are integral to the success of the
operation and involve integration with the other mining
professionals. SGM has come some way toward optimising the
output from this deposit but clearly recognise a number of areas
in which continued effort is required.
268
ACKNOWLEDGEMENTS
The author wishes to acknowledge the permission of Stawell
Gold Mines Pty Ltd to publish this information. A large number
of personnel have contributed much to the development of the
Stawell operation and deserve much credit for their inputs.
REFERENCES
Cayley, R A and Taylor, D H, 1996. Geological Evolution and Economic
Potential of the Grampians Area, Victoria, in Recent Developments
in Victorian Geology and Mineralisation, Australian Institute of
Geoscientists Bulletin 20.
Clappison, D J, 1965. A study of the petrology, mineralogy and
geochemistry of the Stawell goldfield, Victoria, MSc thesis,
University of Melbourne, Melbourne.
Fredericksen, D C and Gane, M, 1998. Stawell gold deposits, in Geology
of Australian and Papua New Guinean Mineral Deposits (Eds: D A
Berkman and D H Mackenzie), pp 535-542 (The Australasian
Watchorn, R, 1986. Deformation in the Stawell Goldfield and its
Relationship to Gold Mineralisation, MSc Thesis, University of
Melbourne, Melbourne.
Geotechnical and Structural

The Regional Controls Exerted on Rock
Discontinuities at the Iron Duke Deposit, South
Australia and Extrapolation Throughout the
Rock Mass
C N Winsor
271
Pasminco Century Mine Open Pit Slope Design

A Geotechnical Perspective
A J Dutton
283
Modelling Bulk Density The Importance of

Getting it Right
I T Lipton
291
The Regional Controls Exerted on Rock Discontinuities at the

Iron Duke Deposit, South Australia and Extrapolation
Throughout the Rock Mass
C N Winsor1
ABSTRACT
INTRODUCTION
Analysis of rock discontinuities, collected using a scanline method, has

been undertaken at the Iron Duke open cut mine, on the southern margin
of the Middleback Range in the Gawler Craton, South Australia. The iron
mineralisation, which is a product of supergene enrichment of carbonate
sediments, is located on the W (western) limb of a gently N plunging D2
macrosyncline. Mesofolds plunge shallowly to the NE or steeply to SE,
suggesting two shortening events, of which the second produced
significant macrofolds. Discontinuities include:
The stability of rock excavations is largely controlled by

discontinuities which are rock fractures of low tensile strength.
These include bedding, foliations and joints, which exhibit
characteristics commonly influenced by the rock type and the
deformation history (Schultz, 1996). For the purpose of
geotechnical assessment extrapolation back into an exposed face
is required, although difficult without an appreciation of the
regional geology. For rock engineering assessment, the precise
values of a number of factors as shown in Figure 1 on the left of
the diagram should be determined, with the ultimate aim being to
assign a rock mass index. To achieve this, assumptions are often
made that characteristics are uniform throughout the rock mass,
although not usually valid (Chiles, 1988). Difficulties in
predicting rock mass responses result from the discontinuous and
variable rock nature, ie heterogeneous, anisotropic,
discontinuous, porous, permeable and multiphase (Priest 1993).
Problems also arise from the multiphase nature. Despite the
common understanding that rock excavation can be disrupted by
the occurrence of unpredictable geology arising from
macro-deformation (ie large-scale folding and faulting), few
technique are available to assess the complete 3D nature of the
rock mass prior to or during excavation. Recently attempts have
been made to determine a fractal - scale invariance relationship
for discontinuities as discussed by Barton and La Pointe (1995).
In making a detailed rock assessment, a variety of rock
properties should be systematically determined, such as by a
scanline sampling method (Priest, 1993), as used by rock
engineers, where a tape is extended across a rock mass and
measurements made concerning each discontinuity that intersects
the tape. This approach has been used here, but is supplemented
with regional geology assessment. To accommodate differences
between engineering and geological nomenclatures, structural
planes are denoted in this article by their dip or plunge (two
figures) and dip or plunge direction (three figures). Owing to
regional variations it is not often possible to uniquely determine
and assign a number to a structural orientation, so terms such as
1.
faults and shears which display variable orientations, mainly

striking WNW dipping steeply N;
2.
mesofractures or joints maintaining a well constrained orientation,

dipping steeply S, subvertical striking E-W; and
3.
bedding surfaces which are folded about a gently N plunging axis.
Through this example of the regional control exerted on local

discontinuities, it is clear that discontinuities are not randomly distributed
as often assumed in purely statistical approaches. Commonly for rocks of
any age there is a distinct discontinuity pattern, reflecting the geological
deformation history. If the controls exerted on local rock discontinuities
can be determined and assigned a number related to their significance,
the way forward is available to develop a site specifc knowledge base that
can make realistic predictions for any position in a given rock mass. Such
a procedure would involve assessment of the regional geology, linked to
scanned discontinuities either via manual or automatic means. Recent
advances in automatic scanning rock faces (eg by the I-SiTE Laser
Radar) to determine discontinuities distributions, enable the rapid
calculation of rock characteristics. This is a significant advance in rock
engineering, providing a method of safe and quick collection of data.
However it creates problems in the presence of an increase in the
geotechnical data available, which should be assessed in the context of
the regional geology. Although techniques are underway to analyse
post-blast discontinuities, the next signifant potential advance in this area
could be the integration of a knowledge based system to assess
discontinuities ahead of mining, incorporating regional assessment and
human intuition.
1.
MAusIMM, Mineral Resources, University of South Australia SA

5095. Present address: Applied Structural Geological Analysis,
28 Wildwood Drive, Salisbury Park SA 5109.
FIG 1 - Stages involved in undertaking rock assessment and determining engineering parameters in order to determine the Rock Mass Index.
271
C N WINSOR
north (N) and west (W), etc are commonly used in describing the
macrostructure. Deformation events are denoted by the letter D,
folds by F, penetrative surfaces by S and lineations by L (after
Bell and Duncan, 1978).
THE GEOLOGICAL CONTROLS ON ROCK MASS

DISCONTINUITIES - NON RANDOM!
Geoengineers traditionally apply geostatistics to establish the
relationship between discontinuities (Priest, 1993), assuming that
they exhibit a random distribution. However, distributions are not
generally random, although in highly foliated rocks they may be.
Statistical methods of analysing geotechnical data do not provide
a realistic model to enable extrapolation, as they assume
homogeneity or stationarity. Interpretative procedures have been
developed by geologists to extrapolate in 3D and determine the
relationship between structural elements. These are often based
on inituitive or gut feelings, and are difficult to codify, being
often influenced by the availability of rock mass data and time.
Fuzzy logic (Zimmermann, 1991) and/or neural networks
(Freeman and Skapura, 1991) potentially offer a means of
treating complex geological situations and interpretations in a
mathematical way. Results are considered to be more realistic
to natural situations. The possibility of incorporating fuzzy logic
and/or fuzzy set theory into all aspects of rock mechanics
investigations dealing with a level of uncertainty, is enormous.
Progress in this area has increased with the articles by Kacewice
(1988, 1990) being relevant. The idea of a fuzzy logic or neural
network knowledge based system to determine ground conditions
is not entirely new, as it has been applied to geological and
mining investigations (eg Hammah and Curran, 1998; Nordlund,
1996). Nordlund (1996, p 689) has commented that fuzzy
methods are very useful, using qualitative geological data in
mathematical models. Studies have been initiated by others to
incorporate structural geology into rock engineering practise
(Thomas and Kor, 1992). However a knowledge based system to
access regional and local structural geology and enable rock
mass predictions is not currently available, although the ability to
develop such a system is apparent.
Programming a fuzzy knowledge based system would involve
identifying a number of crisp (well constrained) and fuzzy rules,
based on local conditions - relations and regional associations.
Through a literature review, preliminary crisp, semicrisp and
fuzzy relationships are recognised. Discontinuities are usually
controlled by the rock type, primary - secondary terrains and
modification phases. A wide range of terrains is evident in the
geological record, dependent on the tectonic conditions and
history. Generally for each of the environments, discontinuities
display a consistent pattern, controlled by the structural
overprints and rock types. Where an area is subjected to
recurring tectonic events the discontinuities present reflect this
history as previously discussed (Winsor, 1995; Winsor and
Priest, 1996; Winsor and Fowler, 1997; Winsor, 1998). An
example of an application of fuzzy logic to potentially assess the
discontinuities at the Iron Duke deposit is proposed below.
IRON DUKE GEOLOGICAL CONTROLS

Regional geology
The Iron Duke deposit (Figure 2) is situated on the southern
margin of the Middleback Range, 50 km S of Whyalla (Fietz
1989; Theologous, 1991), a N-S discontinuous set of hills,
forming the NE part of the Gawler Craton (Yeates 1980) in South
Australia. Fietz (1989) undertook detailed adit and face mapping
and drill hole logging across the deposit. The Iron Duke deposit
lies within the early proterozoic Hutchinson Group which rests
unconformably above the older complexly deformed Sleaford
Complex basement. This forms part of the early to middle
272
FIG 2 - Structural plan of the Iron Duke Deposit displaying outcrop of

banded iron formation and ore, bedding trends, cleavage and faults.
Proterozoic orogenic belt in the Gawler Craton, with the

stratigraphy documented by Parker and Lemon (1982) and Fietz
(1989). Mineralisation of the range is a product predominantly of
supergene enrichment of carbonate facies rocks. The structure on
the range as recognised by Ashworth (1973) comprises two
synclines separated by an anticline.
Deformation
The Iron duke deposit has been subjected to three deformations,
of which D2 produced the most significant response. Fietz (1989)
indicated that during D1 and D2 the rocks were subjected to
high-grade metamorphic conditions, but D3 was a retrograde
event. The deposit is located in the hinge and the W- limb of a D2
macrosyncline, plunging gently N-5<>012 (plunge<>plunge
direction), with bedding folded about this orientation (Figure 3a).
THE REGIONAL CONTROLS EXERTED ON ROCK DISCONTINUITIES
FIG 3 - Structural data regionally from the Iron Duke Deposit.
273
C N WINSOR
The contoured maximum of bedding is oriented 170/65E

(strike/dip), clearly indicating that most of the region is on the
W- limb of a syncline. Mine mesofolds plunge, shallowly to the
NE- or steeply to the SE- (Figure 3b), suggesting two shortening
phases. Although no direct overprinting evidence has been found
through this investigation or that by Fietz (1989). A penetrative
S2 slaty cleavage is weakly developed, oriented 004/88E,
subvertical NS trending (Figure 3c). Faults and shears display
considerable variable orientations, many trending to the WNW
dipping steeply N (Figure 3d). Mesoscale fractures or joints
show a near constant orientation, dipping steeply S, subvertical
striking E-W (Figure 3e). These joints are oriented normal to the
fold axis and can be described as ac joints (cf Hancock, 1985).
Igneous dykes trend mainly N-S, with a variable dip (Figure 3f).
The intersection of bedding and ac joints produces a lineation
plunging in the direction of the fold axis (Figure 3g). Dissolution
stylolite surfaces with peaks plunging to the E, W, SW or
vertically, are considered a result of either horizontally directed
tectonic or overburden stress. The asymetry apparent in the
stylolite peak orientations as well as the fault and shear
directions, suggests that as well as shortening about an E-W axis,
there may also have been shortening about a NE-SW axis, so as
to produce the moderately SE plunging mesofolds in Figure 3b.
The following history is proposed after Fietz (1989) and
Parker and Lemon (1982):
D1 Regional high-grade metamorphism, locally layer parallel
transposed S1, not recognised on the range. Layer parallel
intrusions.
D2 Tight to isoclinal macro - similar style N-S folding, slaty
cleavage S2 variably developed, strong, shallow N-plunging
mineral elongation and quartz rods. Main structure in the
range, no axial plane cleavage. High-grade metamorphic
event. Plunge reversals due to noncylindrical folding.
D3 Major non-coaxial deformation, open folding, retrograde
metamorphism, broad open folds. Major mylonite-shear
zones developed, which truncate the macrosyncline. Axial
planar cleavage not common. No significant feature
apparent to distinguish D2 and D3 folds. Conjugate shear
zones, with local S3, L3 and quartz rods, with a minor
effect on F2.
D4 NE/NW trending cross faults, shear zones, fold - kink
bands with E-W axial planes.
The Middleback Range and more specifically the Iron Duke
deposit, has been subjected to a number of episodes of igneous
intrusion, as described by Fietz (1989). These episodes are:
1.
bedding parallel intrusive pre D2;
2.
complex dykes that intruded post D2; and
3.
fresh dolerite that intruded post deformation.
OBSERVED DISCONTINUITIES
Discontinuity data
Detailed discontinuity data was collected at four locations across
the Iron Duke deposit, to determine their characteristics and the
major controls on the distribution. In undertaking a manual
scanline discontinuity survey and determining its position, an
important determining factor is the level of exposure and the
access. Such restrictions are not so important in the use of an
automated system, however, the results of this analysis highlight
the need for good macrogeological control and the influence of
variable lithologies.
Site 1 308 RL Structural data
The following discontinuities, which include penetrative surfaces

(bedding or cleavage) and non-penetrative surfaces (fractures,
faults, veins and stylolites) are identified from an assessment the
Iron Duke deposit:
Bedding - defined by compositional layering. Lower and middle
Middleback Formations are particularly well bedded with
alternating layers of hematite and chert (Fietz, 1989). Locally
graded bedding is identified (Lemon, 1979). Bedding is well
preserved in jaspilite units. Other units are intensely folded
preventing the recognition of bedding.
Joints - very common mesofractures which display little or no
evidence of movement or offset. Often joints display orientations
controlled by the macroscopic folding or faulting as evident for
274
any region and sedimentary or metamorphic rock type that has

been deformed (cf Hancock, 1985). In regions that have been
affected by folding, the dominant orientations of joints are
normal to bedding and normal (ac) or parallel to the fold axis
(bc). This constraint on orientation means that for a plunging
fold, such as that at Iron Duke, the geometry of the bc joints will
vary, depending on the attitude of bedding (cf Winsor, 1985).
Joints at the Iron Duke deposit as at other locations, display their
best development in the competent rock units.
Faults and shear zones
Bedding parallel faults - these are mesodiscontinuities with
evidence of displacement. They may be present at all scales.
Normal faults - meso and macrodiscontinuities, where the
hanging wall has moved down relative to the footwall.
Thrust faults - meso and macrodiscontinuities where the
hanging wall has moved up relative to the footwall.
Cross faults - meso and macroconjugate faults which strike
NW-SE and NE-SW. As noted by Fietz (1989) they are
poorly recognised, but show evidence of minor horizontal,
but not vertical, displacement.
Veins - discontinuites that underwent dilation and have been
filled with a precipitate eg quartz or carbonate which may have
been later removed. Veins can include fractures that are infilled
with carbonates or iron oxides. The geometry of the veins or the
vein filling material is often related to the stress or strain
directions active during deformation.
Penetrative cleavage - A layer parallel foliation is best developed
in schistose units in the fold hinges, but in jaspiites is absent. A
crenulation cleavage is locally developed, usually subparallel to
S1, and is axial plane to small-scale folds. The wavelength of
folds is variable from a few centimetres to tens of metres.
Layer parallel shear zones - bedding parallel discontinuities,
developed as a result of D2, due to differences in competency
between different layers.
Stylolites - planes along which dissolution occurred. Stylolite
peaks are often normal to the dissolution surfaces, with the peaks
defining the orientation of the stress direction 1.
Dykes - dolerite that intruded pre, syn- or post-deformation.
Site 1 was situated in banded iron formation close to the highest

point on the Iron Duke deposit at RL 308. The site as depicted in
Figure 4 encompasses the limbs and hinge of a S-plunging F2
macrosyncline, which has been transected by a NNW striking,
SW dipping fault. Although joints over the syncline display
variable orientations, they can be readily distinguished into a
number of sets (Figure 4a), ie bedding parallel and bedding
normal which are ac joint normal to the fold axis and bc joints
normal to bedding and ac joints. A plot of normals to bedding
and cleavage are shown respectively in Figure 4c and e, revealing
the nearly subvertical bedding, N-S striking. The cleavage has a
subparallel orientation.
FIG 4 - Structural plan and data from site 1.
275
C N WINSOR
Fourteen scanline traverses were undertaken to determine the

complete 3D characteristics of discontinuities across the
syncline. This involved a series of long horizontal scanlines and a
number of short vertical lines. Stereoplot of normals or poles to
discontinuities are displayed in Figure 4b, d and f, which, when
compared to the stereoplot of bedding, joints and cleavage,
makes clear what each set of discontinuties corresponds to.
For scanlines 1 - 4 (Figure 4b), in the hinge of the fold where
bedding max dips 86 towards 180, Cleavage S1 NS, NW-SE
trend subvertical, three discontinuity sets are apparent, see Table
1:
Set 1
- bc joints, subhorizontal, spacing 0.25 m;
Set 2
- ac joints, subvertical EW trend, spacing 0.27 m; and
Set 3
- bedding surfaces spacing 0.1 m.
By far, most of the discontinuities belong to set three.
On the W limb across scanlines 5 - 9 (Figure 4d), three
discontinuity sets were distinguished, see Table 1:
Set 1
- ac joints, subvertical EW trend, spacing 0.27 m;
Set 2
- bedding surfaces, with spacing of 0.11 m. Common

discontinuities; and
TABLE 1
Discontinuity data site 1 (all distributions are negative
exponential).
Site
Rock type
Site 1
hinge
Set no
East limb
scans 10-14
- bedding, 0.17 m spacing;
Set 2
- ac joints, 0.12 m spacing;
Set 3
- subhorizontal, 0.17 m spacing; and
Set 4
- bc joints , 0.12 m spacing
In comparing the discontinuities recorded across at site 1, it is
clear that in the hinge of the fold and for each limb, the main sets
of discontinuities are present, ie bedding, ac and bc joints. It is
only on the west limb, where the orientation of bedding has
swung to a NW trend that there is an apparent additional set,
which may be considered subhorizontal. Generally for the hinge
and the east limb the ideal bc joints plot close to a
subhorizontal normal (pole) position. As might be expected the
spacing of bedding surfaces is considerably less than that of the
joints.
Site 3 284 RL Discontinuity data

Situated to the N of a NW-SE trending strike slip fault, site 3, is
located at 284RL (Figures 1 and 6). Site 3 lies across a minor
276
Mean
spac (m)
bc
51
90
0.25
ac
40
186
0.27
S0
12
405
0.10
870
121
ac
21
187
0.17
S0
11
418
0.11
bc
50
44
0.04
Total
unassigned
568
57
S0/S2
27
127
0.17
ac
20
211
0.12
bc
27
71
0.22
conj-S0/sh
34
102
0.12
TABLE 2
Discontinuity data site 2- 5 (all disitributions are negative
exponential).
Site
Rock type
Site 2

Four discontinuity sets are identified on the W limb of a syncline
at site 2, 300RL (see Table 2 and Figure 5). The position of site 2
is close to site 1, being located on the N-extension of the fold in
Figure 4. As with the discontinuities identified at site 1, similar
sets are present at site 2, ie 1 a set of bc joints, 2 a set of
subhorizontal joints, 3 a set parallel to bedding and 4 a set of ac
joints. The limited number of discontinuities of each set makes it
difficult to make definitive conclusions regarding the set spacing,
although for this site the spacing of the subhorizontal set appears
less than for site 1. Significantly for site 2, a greater number of
bc joints have been recorded than at site 1, this is because at site
1, these joints are close to horizontal and would be difficult to
access through the manual method adopted in this investigation.
The adoption of an automated scanning system should provide a
more representative sample. However this would still be
restricted by the fact that the degree of exposure is much greater
laterally, than vertically. Also the data collected is limited to a
large degree by the geometry of the available surfaces and faces,
so as to provide a complete 3D representation of discontinuities.
No
discon
762
81
Total
unassigned
Set 3
- bc joint or subhorizontal, spacing 0.04 m.
Four discontinuity sets are distinguished on the E limb of
syncline, along scanlines 10 - 14 (Figure 4f), these are:
Set 1
Fisher
constant
Total
unassigned
scans 1 - 4
West limb
scans 5-9
Discont
type
Site 3
Site 4
Set no
Discont
type
Fisher
constant
Total
unassigned
No
discon
Mean
spac (m)
762
164
bc
24
248
0.15
subhoriz
54
61
0.08
S0
22
168
0.17
ac
25
121
0.19
Total
unassigned
480
258
S0/bc
48
44
0.11
S0/bc
24
144
0.04
S0/bc
91
17
0.59
S0/bc
94
17
0.52
Total
unassigned
1
173
85
S0
36
45
0.21
Sh
168
16
0.46
s0/sh
95
27
0.37
south plunging syncline, with a NW dipping axial plane. A post

D2 dolerite dyke is located on the W side of the fold. At site 3
the orientation of bedding and the fold axis is more to the NE,
than apparent regionally, which might be a result of movement
on the NW-SE trending fault, present immediately to the S. Such
faulting can easily complicate the regional correlation of
discontinuities.
Discontinuities
set 4
set 3
set 2
GC 3
set 1
GC 1 GC 4
GC 2
Bench outline
At site 3, discontinuities have been determined across a

macrofold, there is continual variation in the geometry of
bedding and associated discontinuities across the fold. It is for
this reason that there are a large number of discontinuities that
cannot be assigned, ie 54 per cent unassisted in Table 2. All the
sets represented are normal to the fold axis and can be designated
as either parallel (S0) or normal to bedding (ac joints). The
assessment of discontinuities at site 3 indicates the need for
careful consideration of the regional and structural setting, and
the possible requirement to subdivide some survey sections into
homogeneous portions. However, where folds and to a lesser
extent faults, are present at the scale of the survey, this is not
always possible. It is for this reason that a knowledge based
system might provide a valuable adjunct to discontinuity
assessment.

Site 4 is situated on the western limb of a syncline where
bedding dips moderately steeply to the east. Three main
discontinuity sets are recognised, either comprising shear joints
at a low angle to bedding or bedding surfaces (Table 2).
The way forward

Fuzzy logic was first described by Lofti Zadeh in the 1970s as a
means of modelling roughness (Zimmermann, 1996). Zadehs
main contention was that, although probability theory is
appropriate for measuring randomness of information, it is
inappropriate for measuring the meaning of information. He
proposed a possibility theory as a measure of vagueness.
Whereas traditional logics express values as either true or
false, Zadehs theory expresses lack of precision in a
quantitative fashion by introducing a set membership function
with real values between 0 - 1. Fuzzy set theory is concerned
with the rules for computing the combined possibilities over
expressions that each contain fuzzy variables. The theory has had
success in enabling the problem-solving system to find the rule
that most closely matched the input data. Initial investigations of
fuzzy logic dealt mainly with theoretical aspects, but since then
real applications have been developed for a wide range of
situations. Fuzzy controllers have been used in guiding ships
rudders, helping to maintain balance of free-moving loads,
problems of classification, pattern recognition, database
management, and in making motors more efficient.
277
C N WINSOR
Discontinuities
set 1
set 2
set 4
set 3
GC 3
GC 2
GC 4
GC 1
Bench outline
FIG 7 - General fuzzy sytem configuation.
Fuzzy logic provides an inference morphology that enables

approximate human reasoning capabilities to be applied to
intelligent systems, but it should not be considered fuzzy, but
instead, to provide a mathematical strength to capture the
uncertainties associated with human cognitive processes like
thinking and reasoning. While the fuzzy theory provides an
inference mechanism under cognitive uncertainty, the artificial
neural networks offer invaluable advantages such as adaptation,
generalisation, learning and parallelism. Fuzzy logic is a tool for
modelling uncertainties associated with human thinking and
278
perception. Figure 7 illustrates a general fuzzy system, which

involves:
1.
the real input,
2.
its fuzzification,
3.
the estabished rules,
4.
memory capability to enable pattern determination, and the

final stage
5.
the diminishing of the data fuzzyness.
Fuzzy set theory is the theory of graded concepts, which can

be used to examine situations and phenomena which are not
readily analysed using classical, probability approaches, that is
where there is a transition between member and nonmembership
(Pedryce, 1995; Zimmermann, 1991, 1996). As noted by Kruse
et al, (1993) a physical interpretation of a fuzzy set is the
description of a vague object, eg a grey scale picture which must
be analysed by mathematical methods. Fuzzy logic uses the
possibility approach to likely situations, but focuses mainly on
imprecision which is a common feature of human language
rather than a probability method. In geology where information
about the third dimension is often not known with certainty, and
little meaningful data can be gathered in terms of the probability
of a certain situation arising, a theory which is based on
possibility would appear a more realistic approach. A high
degree of possibility does not necessarily imply a high degree of
probability. Possibility is described (Zimmermann, 1996) as the
upper bound of probability.
The traditional method to build up a model of a situation or
data, is to use a crisp set method (ie a yes-no approach), where
there is no doubt about the values used in the model. There are
not many situations where such an approach is justified. In
engineering an apparatus may measure a certain function very
accurately but it remains questionable how useful this data is and
how representative of the situation, in space - time. Generally, for
most objects and data values there exists a degree of vagueness
or fuzziness in terms of meaning of stages or events, phenomena
or statements (Zimmermann, 1996). In the realm of human
judgement and interpretation particularly, all decisions made are
fuzzy by nature and the linguistic terms used by humans are
fuzzy. An excellent example exists in the area of geology, where,
based on the available subsurface data a three dimensional model
is made. This model must be fuzzy although based on the
available data. If more data becomes available the macro 3D
model may change either drastically or only to a minor extent.
Terms like steeply dipping, coarse grained, strongly jointed and
shallowly plunging are by their nature fuzzy. Figure 8 illustrates
what membership function to describe the plunge of a fold
various terms may have. These are the terms that geologists use
to describe local situations. The only case in which geological
situations might not be regarded as fuzzy, is where there is
complete exposure of the rock mass (ie 100 per cent). In such a
case the features in the rock mass can be regarded as crisp.
However. this is an unrealistic geological situation, as it is only
very rarely that the 3D dimension can be seen definitively. The
common situation is the development of an interpretation of the
3D, based on the available evidence, such as drill hole or seismic
information. Even with this information the interpretation
developed is just that, an interpretation. Neither the seismic or
drill hole data will provide complete coverage and so will exhibit
a degree of fuzziness or vagueness. Through fuzzy logic, an
approach is available to transfer vague linguistic terms such as
those used by geologists, or areas of human judgement,
evaluation and decision making, into a strict mathematical
framework that can be used by engineers.
FIG 8 - The membership functions to characterise various terms used to

describe the plunge of folds.
Fuzzy theory uses: IF THEN rules - these are rules that hold
generally but not universally. A membership function is used to
characterise fuzzy sets. Classical or Boolean logic is based on
truth values or vocabulary operators which are defined using
truth tables. Sets are crisply defined statements or propositions.
The following are a few rules that have been identified through
a literature review and could be incorporated into a rule based
expert system. The rules used will define a membership function,
that can be used to establish fuzzy relations.
General rules
a.
Discontinuity orientations are controlled by factors: eg

history, lithology and topography.
b.
Discontinuity spacing: with increasing development from:

1.
random (log exponential),
2.
log normal, to
3.
normal distributions.
c.
Discontinuity length: a log normal distribution for

mesoscopic discontinuities (Hudson and Priest (1983).
Fracture length may be proportional to layer thickness.
d.
Discontinuity length/displacement: displacement may be

proportional to fault length, the relationship may or may
not be linear, depending on the local rock properties.
Rules deformed areas - rock porosity is known to

increase with deformation
a.
b.
Joint/fold relations - Orientation in relation to folds

depends on the size and type of fold, unit thickness and
competency. Joints are related to the fold (a, b, c: b is
parallel to the fold axis, ab plane is the bedding plane), ie
Hancock (1985).
Areas of faulting.
Fault description:
1.
amplitude - dip and strike or dip direction,
2.
separation geometry,
3.
slip data,
4.
patterns.
Fractures commonly exhibit simple patterns.

Fault related spacing rules:
1.
crack spacing decreases with increased applied stress;
2.
microfracturing
displacement;
3.
microfracturing density is dependent on fault

displacement (Anders and Wiltschko, 1994). Fracture
density decreases in an exponential fashion away from
a fault. Density increases with fault displacement
(Friedman and Logan, 1970); and
4.
spacing is independent of fault type - as the same

spacing is recognised at the same distance from a
normal fault as that observed from a reverse fault.
density
increases
with
fault
c.
Areas of joints/fractures - shear fracture characteristics

depend on lithology.
d.
Rock properties - siffness: stiffer beds, those with >Youngs

modulus, exhibit closer discontinuities. Grain size: spacing
related to grain size, ie closer in finer grained rocks.
e.
Fold/fault relations basic rules:

1.
folds are younger than folded rocks;
2.
faults are younger than rocks they cut;
279
C N WINSOR
f.
3.
an unconformity is younger than the underlying rock

and truncated structure;
4.
metamorphism is younger than rocks it affects; and
5.
intrusive rocks are younger than the host.
Topographic relations - spacing may increase with distance

from the escarpment.
GENERAL METHOD TO PREDICT

DISCONTINUITIES IN OPEN CUT EXPOSURES
IN MULTIDEFORMED TERRAINS
The following discontinuity analysis method is suggested for
future investigations (after Winsor and Fowler, 1997):
1.
Review the regional structural history, geometry and scale

of event. Determine the orientation, spacing and frequency
of macrofold and macrofault zones.
2.
Determine the geometry of any penetrative fabrics and

factors influencing their geometry. Undertake fabric
correlation and assignment of folds and lineations through
regional mapping and fabric correlation.
3.
Identify the geometry of macrofaults and determine their

influence on penetrative fabrics.
4.
Undertake structural analysis (as outlined by Marshak and

Mitra, 1988; Ragan, 1985) and determine the geometry,
scale and timing of macrofolding and related structures.
5.
Make idealised discontinuity predictions in relation to the

macrostructure.
6.
Carefully select positions for scanlines, in relation to

macrostructure.
7.
Undertake scanline surveys and determine geometry and

spacing of discontinuities.
8.
Compare discontinuity predictions with the results of

scanline surveys. If discrepancies exist, review the
macrostructural interpretation.
9.
Document investigation results.
10. Identify the controls exterted on discontinuities and attempt

forward predictions by building a fuzzy-neural network
such as that shown diagrammatically in Figure 9.
The future
An approach is required to incorporate descriptive geology and
inconclusive geological interpretations on the rock mass
characteristics, the regional geology and the controls on
discontinuities, into a mathematical framework suitable for the
mining engineer. The way ahead could involve hand held or
vehicle mounted scanning systems, with direct data input into
computer system. These could incorporate regional geology
influences, establish the factors controlling discontinuity
characteristics, identify fractal and non fractal relationships and
make fuzzy logic - neural network, expert pattern recognition
predictions throughout the rock mass. This would be based on
available data and incorporation of tools to determine the
controlling factors and extrapolate throughout the rock mass.
CONCLUSIONS
This investigation has shown that at the Iron Duke Deposit,
regional and local structural geology assessment can provide an
aid to discontinuity prediction, assisting in the preliminary stages
of site investigations, and there-after improving safety, efficiency
and communication between mine geologists and mining
engineers. As with other areas discontinuity characteristics are
controlled by regional folding, faulting and lithology, although
not always clearly defined. The benefits of incorporating
knowledge concerning the regional structure as part of
discontinuity surveys is apparent. Advantages exist in having a
strong overlap between mining engineers and geologists with
expertise in structural and mine geology, ie a back to the future
approach. Particularly with the development of automatic
scanning systems, there is a need to have a good understanding
of the regional geology and the controls on discontinuities. The
next major step in discontinuity analysis could be the
development of a knowledge based system to assess local
FIG 9 - Fuzzy neural network for discontinutity predictions in variable settings.
280
conditions and determine the regional controls and make rock

mass predictions ahead of mining operations. A fuzzy logic neural network approach is proposed as a way to meet this goal
and as such should be examined more closely.
ACKNOWLEDGEMENTS
BHP Steel Long Products division is acknowledged for allowing
publication of this article. The advice and assistance provided by
Professor Stephen D Priest, School of Engineering, University of
South Australia is appreciated. Alex Kavoukis and Baden Smith
of the Multimedia Services Section, FLC, University of South
Australia are thanked for preparing the diagrams.
REFERENCES
Anders, M H and Wiltschko, D V, 1994. Microfracturing, palaeostress
and the growth of faults, J Struct Geol, 16:795-815.
Ashworth, K L, 1973. The origins of iron ores in the Middleback Ranges,
SA, Unpublished PhD thesis, University of Durham.
Barton, C C and La Pointe, P R, 1995. Fractals in earth science (Plenum
Press).
Bell, T H and Duncan, A C, 1978. A rationalized and unified shorthand
terminology for lineations and fold axes in tectonites,
Tectonophysics, 47:T1-T5.
Chiles, J P, 1988. Fractal and geostatistical methods for modeling of a
fracture network, Mathematical Geology, 20:631-654.
Fietz, G, 1989. The geological structure and slope stability of the Iron
Duke iron ore deposit, unpublished Graduate Diploma Thesis, South
Australian Institute of Technology, 71 p.
Freeman, J A and Skapura, D M, 1991. Neural networks
(Addison-Wesley Publishing Company).
Hammah, R E and Curran, J H, 1998. Fuzzy cluster algorithms for
automatic identification of joint sets, Int Journal Rock Mech Min
Science, 35:889-905.
Hancock, P L, 1985. Brittle microtectonics: principles and practise,
J Struct Geol, 7:437-458.
Hobbs, B E, Means, W D and Williams, P F, 1976. An outline of
structural geology, 571 p (Wiley International: New York).
Kacewice, M, 1998. Fuzzy slope stability method, Mathematical
Geology, 19:757-767.
Kacewice, M, 1990. On the problem of fuzzy searching for hard
workability rocks in open pit mine exploration, Mathematical
Geology, 21:309-318.
Kruse, R, Gebhardt, J and Klawon, F, 1993. Foundations of fuzzy systems,
265 p (J Wiley and Sons).
Marshak, S and Mitra, G, 1988. Basic methods of structural geology, 446
p (Prentice Hall).
Meyers, A G, Walker, D and Priest, S D, 1993. SCANEDIT and

SCANMASTER users manual - limited version. Internal report to
BHP Raw materials research group, unpublished, 32 p.
Nordlund, U, 1996. Formalizing geological knowledge - with an example
of modeling stratigraphy using fuzzy logic, J Sed Pet, 66:689-698.
Parker, A J and Lemon, N M, 1982. Reconstruction of the early
Proterozoic stratigraphy of the Gawler Craton, South Australia,
Journal Geological Society of Australia, 29:221-238.
Priest, S D, 1993. Discontinuity analysis for rock engineering, 473 p
(Chapman and Hall).
Ragan, D M, 1985. Structural geology 3rd ed, 393 p (John Wiley).
Ramsay, W R H and Vandenberg, A H M, 1986. Metallogeny and
tectonic development of the Tasman Foldbelt system in Victoria, Ore
Geology Review, 1:213-257.
Schultz, R A, 1996. Relative scale and the strength and deformability of
rock masses, J Struct Geol, 18:1139-1149.
Stapledon, D H, 1996. Keeping the GEO; Why and How, in
Geomechanics in a Changing World, 7th ANZ conference in
Geomechanics. (Eds: M A Jaksa, W Skaggwa, and D A Cameron),
pp 3-18 (7th ANZ conference in Geomechanics, Adelaide July,
1996).
Theologous, P, 1991. Stratigraphic and structural review of the Iron
Duchess low grade iron ore deposit, Middleback Ranges, South
Australia, 3rd year field project, SAIT.
Thomas, P R and Kor, F H, 1992. Development of a knowledge-based
system for ground investigation in rocks, Eurock 92, pp 159-162.
Winsor, C N, 1985. Interpretation of a set of faults across the hinge and a
limb of a large scale flexure in the Mount Isa fractures related to
folding, J Struct Geol, 7:719-725.
Winsor, C N, 1995. The prediction of rock mass discontinuities in the
Clare Valley Syncline, based on geology interpretation. GSA,
SGTSG, 25-29:177-178.
Winsor, C N and Priest, S D, 1996. The controls exerted on rock
discontinuities - examples from the Adelaide Foldbelt, 7th ANZ
conference in Geomechanics: Geomechanics in a changing world
(Eds: M B Jakasa et al), Adelaide, pp 240-245.
Winsor, C N and Fowler, T J, 1997. Discontinuity analysis in open cut
exposures - examples from selected central Victorian Sites, in
Melbourne).
Winsor, C N, 1998. The controls exerted on discontinuities in the Broken
Hill Lead Lode, Australian Geomechnics, 33:82-90.
Yeates, G, 1980. Middleback Range Iron ore deposits, in Geology of the
Mineral Deposits of Australia and Paupa New Guinea (Ed: F H
Hughes) pp 1045-1048 (The Australasian Institute of Mining and
Zimmermann, H J, 1996. Fuzzy set theory and its applications (Kluwe
Academic Press).
281
Pasminco Century Mine Open Pit Slope Design

A Geotechnical Perspective
A J Dutton1
ABSTRACT
The geotechnical factors controlling the design of the open pit slopes of
the Pasminco Century Mine near Lawn Hill, Queensland are presented
below. Pasminco Century is a stratiform shale hosted zinc deposit in the
Lawn Hill Formation that is overlain in part by the Thorntonia Limestone
Formation. Both the mineralisation and overburden is located within a
competent rock mass such that intact rock strength and discontinuity
spacing is generally sufficient to prevent large-scale slope failure by
breakage through the rock mass. Rather, overall slope stability issues are
dominated by the major discontinuities within the rock mass.
The deposit is located in a structurally complex regime, it consists of a
synclinal basin which is truncated by easterly striking normal faults
through its centre and northern and southern limits and by an
unconformity on its eastern and western margins. Deposit geometry and
slope stability is further complicated by small-scale folding, a pattern of
northeast and southeast striking subvertical faults and north easterly
dipping thrust faults with their associated fault emplaced blocks of Lawn
Hill Formation in the overburden.
The large size of the deposit requires a staged mining approach with
the excavation of both final pit walls around the external sides of each
stage and interim pit walls along the internal boundaries between
adjacent stages. The deposit geometry and major discontinuities control
not only the final pit wall design but also interim wall location and
orientation, excavation sequence and in some stages the excavation
schedule. Two and three-dimensional limit equilibrium stability analyses
were conducted using both deterministic and probabilistic approaches to
determine appropriate pit wall design criteria and to select a suitable
excavation sequence and schedule.
INTRODUCTION
Geotechnical assessment for design of the Pasminco Century
Mine open pit slopes commenced in 1991 with the introduction
of a basic geotechnical logging system and the establishment of a
database for all cored boreholes drilled on the deposit. The level
of geotechnical data collection and input into pit design
increased as the deposit investigation advanced into detailed
feasibility studies. Specific geotechnical boreholes were drilled
both over the deposit and around the margins to obtain rock mass
and discontinuity information in areas of proposed interim and
final pit walls. The borehole information was supplemented with
limited surface and bulk sample excavation mapping and a
program of laboratory testing. The assessment presented below
represents the state of knowledge at the completion of the
feasibility study.
GEOLOGICAL AND GEOTECHNICAL SETTING

The Century deposit is a stratiform shale hosted zinc, lead and
silver deposit within the Proterozoic Lawn Hill Formation.
Figure 1 presents the deposit layout and pertinent geological
features on a plan slice located approximately 150 m below
ground surface. Deposit stratigraphy consists of a black
carbonaceous shale in the lower footwall overlain by a sequence
of interbedded black shale, grey siltstone and brown sideritic
siltstone beds, that contains the 40 50 m thick orebody, this in
turn is overlain by Widdallion (Hangingwall) Sandstone unit and
Thorntonia (Cambrian) Limestone. Mineralisation occurs within
1.
Senior Rock Mechanics Engineer, Golder Associates Pty Ltd, PO

Box 1734, Milton BC Qld 4064.
three ore blocks located between 0 and 350 m below ground

surface. The main northern and southern blocks are separated by
the north dipping Pandoras Fault and bounded on the north by
the south dipping Nikkis Fault and on the south by the north
dipping Magazine Hill Fault. The east and west margins of the
northern and southern blocks terminate on an unconformity at
the base of the limestone unit. The smaller fault bounded eastern
block is located within the Termite Range Fault Zone, a major
regional structure that strikes to the southeast.
The deposit occurs in a structurally complex regime consisting
of a synclinal basin over which smaller scale folding and a
pattern of northeast and southeast striking subvertical normal and
reverse faults and northeast dipping thrust faults have been
superimposed. Bedding dip is generally shallow apart from the
eastern block and the west margin of the northern block. Deposit
geometry is further illustrated in Figure 2 as typical north south
and east west cross-sections. Large blocks of Lawn Hill
Formation material have been emplaced into the limestone by
thrust faulting. Irregular veins of carbonate breccia occur
throughout the Cambrian and Proterozoic units. The carbonate
breccia veins can be either conformable with or cut across
bedding. The vein contacts are often sheared suggesting the
carbonate breccia has intruded along pre-existing structures.
Further information on deposit geology can be found in Waltho
and Andrews (1993) and Waltho, Allnutt and Radojkovic (1993).
Geotechnical characteristics of the different lithological units
are summarised in Table 1. Cambrian limestone is the most
structurally complex unit present. Thrust faulted Lawn Hill
Formation blocks within the limestone are of variable
composition, size and bedding and contact orientation. Some of
the thrust blocks are in excess of 100 m thick and 200 m in plan
dimensions. A weathered zone up to 50 m thick, ranging from
slightly to extremely weathered Proterozoic units and in some
places clay is present below the limestone. Thrust faulting
appears to have removed this weathered zone in some areas.
Bedding partings and joints are present throughout the deposit,
their orientation and surface characteristics varies between units
and location within a given unit. In general the bedding partings
are up to 10 m long, spaced up to 2 m apart with clean, planar,
smooth to rough surfaces. Joints are typically up to 6 m long,
spaced up to 6 m apart with planar to wavy, smooth to rough,
clean surfaces. Laboratory direct shear tests indicate angle of
friction values between 30 and 40.
Faults and sheared zones with persistences in excess of 100 m
and widths in excess of 0.5 m are present in all units. Fault
composition ranges from fragmented limestone, sandy clay,
carbonate breccia and cavities in the limestone to sheared shale,
siltstone, tuff and carbonate breccia in the Proterozoic units.
Laboratory tests indicate fault shear strengths in the order of zero
to 300 kPa cohesion and 19 to 45 angle of friction.
The other dominant major discontinuities present in the Lawn
Hill Formation are sheared zones conformable with bedding
referred to locally as Bedding Shears. The greatest frequency of
bedding shears occurs in the lower 40 m of the hangingwall shale
siltstone. One of the hangingwall bedding shears is shown in
Figure 3. Bedding shears consist of sheared fissile black shale
and tuffaceous clay bands, typically up to 100 mm thick with
persistence in excess of 100 m. Laboratory testing found peak
shear strengths in the range zero to 30 kPa cohesion and 10 to
283
A J DUTTON
FIG 1 - Plan slice through Century deposit at 1050 mRL.
30 angle of friction and residual shear strength angle of friction

values of 9 to 19. Figure 4 shows the distribution of residual
shear strength data.
The groundwater level prior to mining and pumping from
limestone dewatering bores was approximately 30 m below
ground surface.
SLOPE STABILITY ASSESSMENT

The feasibility study concentrated on assessing overall slope
stability for the proposed interim and final pit walls as the
geological and geotechnical models were not sufficiently
advanced to undertake a meaningful assessment of individual
batter stability. Two-dimensional limit equilibrium (program
SLOPE/W) and finite difference numerical stress analysis
(program FLAC) techniques were used to assess the potential for
rock mass failure. Analysis results indicated that rock mass
strength is generally sufficient to prevent overall failure of the
proposed pit walls. Rock mass failure may be possible in a few
isolated narrow zones of fragmented limestone or highly sheared
black carbonaceous shale associated with faults subjected to high
groundwater pressure.
284
These findings dictated that the stability assessment should be

based on identifying and analysing potential failure modes
involving movement along discontinuities. A three-dimensional
geological modelling package was used to determine potential
discontinuity controlled failure modes by visualising the
interaction between the geotechnical model and the proposed pit
walls. Emphasis was placed on studying the location, orientation
and persistence of faults, bedding shears and thrust block
contacts relative to the pit walls. An iterative approach was
adopted to identify potential failure modes and assess overall
slope stability for alternative pit excavation sequences and wall
positions. Two and three-dimensional limit equilibrium analyses
were undertaken using programs SLOPE/W and CLARA to
assess slope stability. Deterministic stability analyses using
laboratory testing derived discontinuity shear strength values
were undertaken using a minimum global factor of safety of 1.3
to design most of the slopes. A minimum factor of safety of 1.5
was used where a slope failure was liable to have a major adverse
economic impact, such as the loss of the haul road or sterilisation
of large quantities of ore. Figure 5 illustrates part of one of the
CLARA models that was used to assess stability of the Stage 5
pit west wall.
PASMINCO CENTURY MINE OPEN PIT SLOPE DESIGN
TABLE 1
Typical geotechnical characteristics of the main rock units.
Rock unit
Thickness (m)
Characteristics
Thorntonia Limestone (CLS)
0 - 430
complexly folded and faulted, contains thrust fault blocks and caverns
very high to extremely high intact rock strength
3 - 4 joint sets of highly variable orientation, persistence and spacing are
present
faults composed of fragmented limestone, sandy clay, carbonate breccia
and cavities
Carbonate Breccia (CBX)
0 - 34
Widdallion Sandstone (HWS)
0 - 150
massive sandstone beds with shale-siltstone interbeds and occasional

thin tuffaceous clay beds
high to very high intact rock strength
bedding partings and 2 - 3 joint sets present
Hangingwall (HWD), Orebody and Footwall

(UFW) Shale-Siltstone
Black Carbonaceous Shale (BCS)
200 - 300
> 250
irregular veins occur throughout the deposit

can be either conformable with or cut across bedding
very high to extremely high intact rock strength
massive with occasional joints
frequently have sheared contacts
interbedded black shale, grey siltstone and brown sideritic siltstone

tuffaceous clay beds common in the hangingwall
high intact rock strength
slightly bedded black shale with thin tuffaceous clay and crushed shale
beds
highly sheared in the vicinity of the major faults
high intact rock strength
1. Intact rock strength categories are based on Australian Standard AS 1726 - 1993.
FIG 2 - Typical cross-sections.
285
A J DUTTON
19.0 - 19.5
18.0 - 18.5
17.0 - 17.5
16.0 - 16.5
15.0 - 15.5
FIG 5 - CLARA slope stability analysis model showing Stage 5 pit west
wall and the potential failure surfaces.
14.0 - 14.5
13.0 - 13.5
12.0 - 12.5
11.0 - 11.5
10.0 - 10.5
9.0 - 9.5
8.0 - 8.5
FIG 3 - Bedding shear composed of tuffaceous clay in hangingwall

shale-siltstone.
FIG 4 - Bedding shear residual strength data.
A simplistic probabilistic assessment approach was also used

where bedding shears controlled slope stability. This was
necessary because of the large variation in shear strength
obtained from laboratory tests on bedding shears as illustrated in
Figure 4. The approach was based on undertaking multiple
three-dimensional limit equilibrium CLARA analyses for each
pit wall, excavation schedule and failure surface geometry case
to determine the bedding shear residual angle of friction value
which produces limiting equilibrium (factor of safety of 1.0).
The probability of failure was then equated with the probability
of occurrence of a bedding shear with a residual shear strength at
or below the value required to produce limiting equilibrium. A
uniform shear strength was assumed for the entire bedding shear.
Pit walls were designed on the basis that the probability of
failure did not exceed five per cent.
286
PIT SLOPE DESIGN

Owing to its large planar extent, the economic extraction of the
deposit requires mining to occur as a series of sequential stages
of pit advance. The position and location of interim stage walls
was based on economic conditions, mining practicalities and
geotechnical constraints. Many different excavation sequences
were evaluated before the sequence illustrated in Figure 6 was
adopted in the feasibility study. Figure 6 shows the final pit
overlaid by the rim of the interim walls of the eight stage pit
shells. Stage 1 mining commences in a relatively shallow
high-grade portion of the centre of the southern block. Mining
then proceeds east to form final pit walls in Stage 2 before
heading north and then west into the northern block, with the
mining completed in the western portion of the southern block.
FIG 6 - Plan of the ultimate pit and pit stage limits.
Geotechnical factors influencing stability and slope design

criteria for interim and final pit walls are summarised in Table 2.
Maximum inter-ramp slope angles, as measured from horizontal,
are given in Table 2. Equipment selection and economic
considerations dictated that interim pit walls be as steep as
possible and that the working bench heights be set at 16 m in
waste and 8 m in ore. Final pit wall batter height was set at 32 m,
produced by double benching, to maximise berm width to permit
equipment access along the final wall berms. Batter inclination is
controlled by joints and bedding partings. Where bedding dips
out of the wall at an inclination steeper than 15 from horizontal
the batters are to be excavated along bedding. Elsewhere batters
were designed at an inclination of 60 from horizontal. The 60
batter inclination was based on reducing the frequency and size

of batter failures, as the frequency of joints dipping out of the
batter significantly increases above this inclination.
Final pit wall design is controlled by the location and
orientation of the deposit bounding faults to the north and south,
bedding orientation to the west and rock mass conditions to the
east. Interim pit wall design is controlled by the location and
orientation of major discontinuities together with the location of
the wall and the excavation schedule. Two of the interim walls,
the Stage 1 pit south wall and Stage 5 pit west wall required
considerable analysis to derive a wall location and excavation
schedule which meet both economic and geotechnical criteria.
287
A J DUTTON
TABLE 2
Geotechnical factors affecting slope stability and wall design criteria.
Max height (m)
Slope angle1 ()
Thrust block bedding orientation and wedge failures

along fault and joint intersections.
130
50
Wedge failures along fault and joint intersections.
130
50
Active-passive wedge failure along Magazine Hill Fault

linked to bedding shears subjected to high groundwater
pressure.
135
50
Pit stage
Wall
Geotechnical factors controlling wall design
North
East
South
West
110
50
North
160
50
East
(final)
Planar and wedge failures along fault and joint

intersections. Localised cavernous ground in CLS.
175
48
South
(final)
Planar failure along Magazine Hill Fault (MF) which

dips 50 - 60 north. Wall located along or behind the MF
footwall.
175
50
North

240
50
East
(final)

160
48
Central
(final)
Planar failure along Pandoras Fault (PF) which dips 40

north. Wall located along or behind the PF footwall.
95
40
West

240
50
East
(final)

190
48
Central
(final)
Planar failure along Pandoras Fault (PF) which dips 35 42 north. Wall located along or behind the PF footwall.
150
35 - 42
West

270
50
East
(final)

225
48
Central
(final)
175
32 - 45
West
Circular failure along bedding shears in either HWD or

UFW. Groundwater depressurisation required. Wall
stability is sensitive to excavation schedule.
290
50
North
(final)
Planar failure along Nikki's Fault (NF) which dips 44 60 south. Wall located along or behind the NF
hangingwall. Fragmented and cavernous CLS and
sheared BCS present which may require a coat of
shotcrete.
255
44 - 48
Central
Active-passive wedge failure along PF and bedding

shears in either HWD or UFW. Groundwater
depressurisation required. Wall stability is sensitive to
excavation schedule.
110
50
West
(final)
Planar sliding along bedding partings and shears in UFW

which dips 30 - 65 east. Individual batters in UFW to be
excavated along bedding.
210
25 - 40
Central
(final)
200
35 - 48
West
(final)

330
25 - 40
South
(final)
Planar failure along Magazine Hill Fault (MF) which

dips 30 - 40 north. Wall located along or behind the MF
footwall.
215
30 - 40
West
(final)

130
25 - 35
1. The maximum inter-ramp slope angle, as measured from horizontal, is given.
288
The potential Stage 1 pit south wall failure mode involved an

active - passive wedge mechanism with sliding down the
Magazine Hill Fault and outwards and upwards along a bedding
shear in the immediate hangingwall or footwall of the orebody.
High groundwater pressures are required along the potential
sliding surfaces to initiate failure. Subhorizontal drain holes were
recommended for the south wall to reduce the groundwater
pressure and thus the risk of failure.
The Stage 5 pit west wall has the potential for overall slope
failure by rotational movement along one or more bedding shears
in either the hangingwall or footwall of the orebody. Two failure
mechanism variants were considered based on the location and
orientation of the lateral and rear release surfaces. The first
variant used Pandoras Fault and Nikkis Fault linked to tension
cracks through the limestone to provide the release surfaces. The
intersection of a northeast striking fault (parallel to the Page
Creek Fault Zone) with a southeast striking fault (parallel to the
Termite Range Fault Zone) provided release surfaces in the
second variant. Stability analyses indicated that a combination of
groundwater depressurisation and advanced waste stripping in
the Stage 6 pit is required to maintain overall slope stability.
Orebody geometry and horizontal distances in excess of 400 m
from the pit wall to the orebody footwall preclude the use of
drain holes. The limestone dewatering bores along the eastern
margin of the deposit terminate at a higher level than the deepest
point of the orebody and consequently cannot fully dewater the
Proterozoic units in the critical area. As such, groundwater
depressurisation will require the installation of pump wells in the
western portion of the orebody. Several different excavation
schedules and wall positions were analysed in order to select a
suitable design that met the maximum risk criterion.
INITIAL MINING EXPERIENCE

Mining of the Stage 1 pit commenced in May 1998. The current
mining sequence contains seven pit stages compared to the eight
stages used in the feasibility study. Minor changes have been
made to the Stages 1, 2 and 3 pits compared to the feasibility
study. The major change has been the elimination of the
feasibility study Stage 4 pit. The current pit Stages 4, 5 and 6 are
approximately equivalent to the pit Stages 5, 6 and 7 shown in
Figure 6. The excavation schedule has also changed as a result of
the pit stage changes. A geotechnical investigation program is
currently under way to determine the suitability of these changes
and the extent of the groundwater depressurisation and advanced
waste stripping requirements.
Geotechnical conditions encountered in the Stage 1 pit south
wall are illustrated in Figure 7. Mining has exposed a higher
level of structural complexity than that indicated in the feasibility
study from the widely spaced resource definition boreholes,
particularly in the north and east walls of the Stage 1 pit. The
western boundary of the Termite Range Fault Zone extends west
into the deposit rather than stopping at the eastern margin as was
previously thought. Increased fracturing, weathering and variable
discontinuity orientations are associated with the Termite Range
Fault. Thrust faulted Lawn Hill Formation blocks are also present
with an increased frequency, extent and degree of contortion than
FIG 7 - Stage 1 pit south and west walls.
was considered in the feasibility study. This increased

complexity has affected the stability of individual batters rather
than the overall slope. The most notable batter instability being
associated with planar and wedge sliding failures along adversely
oriented joints, bedding and bedding shears within thrust blocks
exposed in the north wall of the Stage 1 pit.
Limestone dewatering bores installed off the eastern margin of
the deposit have also depressurised the northeastern part of the
Proterozoic units within the Stage 1 pit. However, groundwater
pressure in the western portion of the south wall has been
unaffected by this pumping. Consequently an array of
subhorizontal drain holes are being installed into the Stage 1 pit
south wall to assist with groundwater depressurisation and
maintenance of slope stability.
Periodic revision of the geotechnical model will be undertaken
using up to date data from in-pit geotechnical mapping by
Pasminco Century Mine Ltd, purpose drilled geotechnical
boreholes and laboratory testing of discontinuity shear strength.
Pit wall design, excavation sequence and schedule will be
optimised by conducting slope stability analyses using the
revised geotechnical model.
ACKNOWLEDGEMENTS
The author gratefully acknowledges Pasminco Century Mine Ltd
for permission to publish this paper and Golder Associates staff
for assistance in the preparation of this paper.
REFERENCES
Waltho, A E and Andrews, S J, 1993. The Century zinc lead deposit,
northwest Queensland, in Proceedings The AusIMM Centenary
Waltho, A E, Allnutt, S L and Radojkovic, A M, 1993. Geology of the
Century zinc deposit, northwest Queensland, in Proceedings World
Zinc 93, pp 111-129 (The Australasian Institute of Mining and
Metallurgy, Melbourne).
289
Modelling Bulk Density The Importance of Getting it Right

I T Lipton1
ABSTRACT
Tonnage and grade are the principal reported components of mineral
resource estimates but tonnage is a derivative of two more fundamental
estimates; volume and bulk density. Typically, most of the effort in a
resource estimation study is directed towards grade and volume, yet poor
application of bulk density data to resource models can result in
significant biases. The impact on tonnage estimates is easily recognised
but poor density estimates may also introduce subtle biases into the grade
estimates.
Methods for measuring bulk density have been discussed previously.
This paper focuses on the manipulation of bulk density data and the
extension of bulk density from essentially point measurements to block
estimates within resource models. Bulk density is a continuous,
geologically controlled, spatial variable and should be treated as such.
Issues such as combining data from multiple sources, sample support and
the interdependence of bulk density and grade estimates must be
resolved. The statistical and geostatistical techniques used for grade
estimation are, under the right circumstances, equally applicable to bulk
density. The importance of bulk density needs to be recognised. With
robust raw data and correct estimation procedures, better estimates of
bulk density can be obtained and both local and global tonnage and grade
estimates can be improved.
INTRODUCTION
Literature on resource estimation is abundant. The great majority
of this published work focusses on the estimation of grade within
a volumetric framework. Almost as a footnote, volumes are
converted to tonnages by the implicit application of bulk density.
This balance is reflected in mining industry practice where,
typically, most of the effort in a resource estimation study is
directed towards grade and volume. Bulk density, a significant
multiplier in the estimation of tonnage, is often given little
attention, yet poor application of bulk density data to resource
models can result in significant biases. The impact on tonnage
estimates is easily recognised but poor density estimates may
also introduce subtle biases into the grade estimates.
Methods for measuring bulk density have been reviewed
previously (Lipton, 1997, in press). Bevan (1993), Dadson
(1968) and Neuss (1998) have discussed the relationship between
bulk density and grade. This paper examines the manipulation of
bulk density data and the extension of bulk density from
essentially point measurements to block estimates within
resource models. The importance of treating bulk density as a
continuous, geologically controlled, spatial variable is discussed
and issues such as combining data from multiple sources, sample
support and the interdependence of bulk density are examined
through a series of real and hypothetical examples.
DEFINITIONS
Density may be measured and expressed in a variety of ways.
The author reviewed many of the methods and compared some of
their strengths and weaknesses in an earlier paper (Lipton, 1997).
The present paper focuses on the application of these
measurements to resource models and hence resource estimates.
The density of a material is defined as the mass per unit
volume. In mining applications we are interested in bulk density;
that is, the density of mineable volumes of rock, inclusive of
naturally occurring pore space or larger voids. The in situ bulk
1.
MAusIMM, Principal Geologist, Mining and Resource Technology

Pty Ltd, Level 3, Kirin Centre, 15 Ogilvie Road, Mt Pleasant WA
6153.
density includes the water naturally contained in the rock. This

measure is important for estimation of the tonnage of material to
be moved during mining. For resource estimates, however, we
require the dry bulk density, which is the dry mass of the
material divided by the in situ volume. Dry bulk density is used
because grades are expressed on a dry mass basis. Specific
gravity (SG) is a measurement very similar to, but not strictly
synonymous with, density. It may differ significantly from bulk
density in rocks with large voids. It is common mining practice
to use the terms SG and density interchangeably but the author
prefers the explicit use of the term bulk density and its variants,
in situ bulk density and dry bulk density. For the remainder of
this paper the term bulk density will be used to refer to dry bulk
density.
VOLUMETRIC SUPPORT
Bulk density is a continuous geologically-controlled, spatial
variable that has a large impact on the estimates of the tonnage
and, in some cases, grade of a resource. Measurements of bulk
density are essentially made at points in 3D space and must be
extrapolated throughout the volume of an orebody in much the
same manner as grade estimates. The level of detail or
sophistication of the estimates of bulk density within the resource
model will depend on the variability of bulk density within the
deposit and the volume, quality and spatial distribution of the
available data. Bulk density data should be subjected to the same
validation and statistical analysis as the grade variables. Extreme
values should be critically assessed to determine whether they
are natural features of the rock or are due to experimental error.
Density measurements are as prone to random or systematic
errors as are assay data and should be evaluated accordingly. As
well as removing erroneous values, it may be necessary to
remove extreme values from the data set, in order to avoid
biasing tonnage estimates.
In common with assay variables sampled at discrete locations,
the variance of measured bulk density values will depend on the
size of the sample. This is an example of Kriges relationship, or
the volume-variance effect, that is seen when samples have
different statistical support. The variance of the bulk density of
small subsamples of core (say, 10 cm pieces) will be higher than
the variance of one metre lengths of core. This is illustrated in
Figure 1, which shows data from a nickel laterite deposit.
Seventy-six one-metre intervals of triple tube HQ-size core were
selected for bulk density measurement by two methods. Firstly,
the caliper method was used on the complete one-metre lengths.
This involved measuring the core diameter and core length,
calculating the core volume and weighing the core after drying.
Secondly, a small subsample of core, usually from 10 cm to
20 cm in length was selected from each metre interval and bulk
density was determined for these subsamples using a water
displacement method, after sealing them with wax. The scatter
plot comparing the two sets of results shows a conditional bias
arising from the much broader spread of values from the
subsamples than from the whole core samples. The plot also
shows an absolute bias, with the subsamples on average returning
higher values than the whole samples. To test for differences
resulting from the two methods of measuring bulk density rather
than differences between the two sets of samples, the bulk
density of a single set of core samples was measured using both
methods. The results were almost identical, demonstrating that
the absolute bias evident in Figure 1 was due to preferential
subsampling of more competent core and not errors in the
measurement technique.
291
I T LIPTON
These examples show that it is important that the differences in

the statistical support of bulk density data are recognised,
particularly where sets of data derived from different sources are
being compared. As with assay data, bulk density data sets with
mixed support should be processed to achieve uniform support.
For some data, it may be appropriate to carry out a variance
adjustment, such as an affine correction. The affine correction
reduces the variance of a distribution without changing its mean,
by squashing all of the values towards the mean. This approach
may be used for distributions that are not strongly skewed. The
data in Figure 2 was adjusted by applying an affine correction of
variance to the core data and an absolute correction to the
geophysical data to account for the moisture content of the drill
hole walls. The corrected data is shown in Figure 3. By applying
these corrections to the larger geophysical data set it was
possible to combine the geophysical and core data for the
purposes of resource estimation.
FIG 1 - Comparison of dry bulk density from measured subsamples

(water displacement method) versus dry bulk density measured from
1 m core (caliper method).
The volume-variance relationship may also be observed when

comparing geophysical logging data with core measurements.
Figure 2 shows a scatter plot of bulk density measured from
one-metre lengths of drill core using the caliper method, plotted
against geophysical logging data sampled at points 10 cm apart
and composited over the same one-metre intervals. The
geophysical data shows a narrower spread than the core data,
resulting in the conditional bias in the scatter plot. The lower
variance of the geophysical data arises because, for any given
metre of drill hole, the gamma rays pass through a volume of
rock that is significantly larger than the volume of the core.
Absolute bias is again observed, in this case because the
geophysical values had not been fully corrected for the moisture
contained in the drill hole wall.
FIG 3 - Same data as in Figure 2, after applying an affine correction of

variance and correction for moisture content to the geophysical data.
INTERPOLATION OF BULK DENSITY
FIG 2 - Scatter plot of dry bulk density of 1 m core samples versus

dry bulk density from 1 m composite geophysical logs.
292
Having validated the density data, removed extreme values and

resolved any differences between data collected by different
methods, the data can then be used to estimate the bulk density
within the resource model. In many deposits that have only small
concentrations of dense minerals, such as many gold deposits,
variations in bulk density within the primary rock may be small
and spatially uncorrelated (random). As with grade estimation,
there is no benefit in generating local estimates that are more
detailed than the data is able to support so, in this case, it is
appropriate to apply the averages of the measured bulk density
values to each rock type.
Weathering and supergene alteration often result in significant
variations in bulk density that show spatial correlation. In many
cases, weathering progressively decreases the bulk density of
rocks due, for example, to the replacement of silicate minerals by
clays and the development of increased pore space. In these
cases, if bulk density is undersampled, it may be appropriate to
divide the resource model into zones that represent the degree of
weathering and apply average values to each zone. Commonly,
weathering effects include remobilisation of the primary rock
components and reprecipitation of minerals in structurally or
chemically controlled locations. This may lead to complex
patterns of bulk density distribution within the weathering profile
that are not adequately modelled by the use of average values.
MODELLING BULK DENSITY THE IMPORTANCE OF GETTING IT RIGHT
From the preceding discussion it should be clear that bulk

density is a spatial variable that may display as much variability
or continuity as is shown by sample grades. As a spatial variable,
it can be analysed using geostatistical techniques and it may be
estimated within a three-dimensional volume using the same
techniques as are used to estimate grade. If structured variograms
can be obtained from the data, bulk density may be estimated by
kriging or other interpolation techniques. In this way geological
variations in an orebody such as, for example, degree of
cementation may be accurately reflected in the resource model.
As with grade estimation, choice of estimation method and
parameters will depend on factors such as data spacing and the
skewness of the distribution.
In deposits in which the element of economic interest occurs in
very small quantities, such as precious metals, there may be little
or no relationship between bulk density and grade. These two
variables can therefore be estimated independently. However, in
many deposits, the bulk density of the ore is directly related to
the mineral assemblage and hence is correlated to the grade. In
this case, grade and bulk density are dependent variables. If they
are treated, incorrectly, as independent variables serious metal
imbalances or biases may be introduced into the resource
estimate.
Dependent relationships between bulk density and grade are
common in massive sulphide deposits. This is expected, since the
ore minerals are often of much higher density than the gangue
minerals in the host rock. This is illustrated by data from a
massive nickel sulphide deposit (Figure 4). The ore zone consists
essentially of an assemblage of pentlandite, pyrrhotite,
chalcopyrite occurring at the base of a serpentinised ultramafic
flow or tectonically mixed with a banded iron formation. Figure
4 shows the correlation between nickel grade and bulk density.
Independent estimation of nickel and bulk density with different
data sets, search parameters or interpolation weights will produce
models in which the relationship between grade and bulk density
observed in the samples is not preserved in the block estimates.
This will lead to local and possibly global bias in the estimates of
nickel metal content.
There are several alternative ways to approach the estimation
of grade and bulk density where they are dependent. These
methods are not discussed in detail here but some of them are
summarised below.
Independent estimation of grade and density

If density and grade measurements are available at all points at
which there is a grade measurement a simple approach that may
be used is to estimate grade and density independently but with
identical interpolation parameters (search neighbourhood,
weights, etc). In this way the grade and density of a block are
estimated using the same set of data points and the same set of
weights. This approach implicitly assumes that the spatial
characteristics of grade and density are identical. Although this is
unlikely to be true, in many cases it will serve as a reasonable
first approximation. In practice, the more usual case is that only a
small proportion of samples with grade data also has density
data. In this case the relationship between grade and density
cannot be properly maintained with this approach and more
sophisticated techniques may be required.
Cokriging
Kriging is an estimation method that allocates weights to the
sample data points surrounding the point, or block, for which the
grade is to be estimated. The weights are derived from a model
of the spatial variability of the variable, known as a
semi-variogram. The weights vary as the sampling pattern
changes, so that kriging is a weighted moving average process.
The kriging system minimises the mean squared estimation error.
Cokriging is a variation of kriging in which two (or more)
variables are used together to estimate each variable in turn. In
addition to the variograms of the individual variables, the
cross-variogram, which describes the spatial cross-covariance of
the two variables, is modelled. Cokriging can be used to estimate
grade and density together, where density is undersampled
compared to grade. With modern geostatistical software and
computer hardware, cokriging is a practical option that should be
considered where grade and density are dependent variables.
Kriging with a trend

There are several kriging methods that allow a well-sampled
variable (ideally an exhaustively sampled variable) to be used to
improve the estimation of a sparsely sampled variable. These
methods include kriging with varying local means and kriging
with an external drift (Goovaerts, 1997). These methods are more
complex than cokriging and some of the conditions that must be
satisfied by the data may not be honoured by grade and density
data. They are mentioned here merely to illustrate the potential
for using well-sampled grade data to improve the local estimates
of sparsely sampled density data.
DERIVATION OF BULK DENSITY FROM

BLOCK GRADES
FIG 4 - Comparison of bulk density estimated from normative minerals

and bulk density measured from drill core from a nickel sulphide deposit.
In mineral deposits that have relatively simple mineral

assemblages and low porosity, the correlation between bulk
density and the grades may be sufficiently strong to permit the
bulk density of ore blocks to be estimated directly from ore block
grades. Formulae for estimating bulk density are commonly
calculated using linear regression (single assay variable) or
multiple linear regression (multiple assay variables, such as Fe,
Pb, Zn). However, where there is sufficient data, it is often
observed that the relationship between density and grade is
non-linear. For example, Bevan (1993) presented plots of specific
gravity against iron and against sulphur, that are best fitted by a
curve. As Bevan demonstrated, a non-linear relationship between
density and assay values is the theoretically expected outcome.
Assays and other chemical analyses are usually expressed on a
weight percentage basis but, because density is expressed in
terms of volume, the relationship between grade and density is a
curve rather than a straight line.
293
I T LIPTON
If the density of a sample or a block is to be estimated from

mineral proportions or assays, and these are expressed on a
percentage mass basis, they should first be converted to a
percentage volume basis. A simple hypothetical example is
presented in Figure 5 to demonstrate this. The example shows
that the errors arising from incorrectly using a mass-weighted
calculation or a linear regression can be significant. Even at more
common grades, the difference between the correct
volume-weighted estimate and the incorrect mass-weighted
estimate is substantial. For example, if the lead content of the
blocks considered in Figure 5 is 25 per cent, the expected bulk
density of the sample is 2.58 t/m3 but the mass-weighted
calculation provides an erroneous prediction of 3.47 t/m3. This
would result in a 35 per cent overestimate of tonnage!
Estimation of bulk density by regression from assay data may

be satisfactory in deposits in which there are only one or two
dense mineral species. Where the assemblage of dense minerals
is more complex, and individual elements occur in more than one
mineral, partitioning of assay grades between mineral species is
necessary. If it is possible to make some realistic assumptions
about the mineral species present in the deposit and their
paragenesis, assay grades can be assigned to an assemblage of
normative minerals. The proportions of the normative minerals in
a sample, converted to a percentage volume basis, can then be
used to estimate the bulk density of the sample. Lipton, Shaw
and Waltho (1999) presented an example of how this method was
successfully used for the Century zinc-lead-silver deposit. At
Century, Zn, Pb, Fe, Mn and S grades were used to estimate the
FIG 5 - Illustration of density calculations.
294
proportions of sphalerite, galena, pyrite, siderite, smithsonite and

cerussite and hence the bulk density of the samples. Comparison
of these results with the results of physical measurements on core
samples and geophysical (density) logging showed that the
normative mineral method produced accurate estimates of bulk
density in this deposit.
A further example of these issues is provided by data from a
magnetite deposit in Western Australia. Figure 6 shows the
correlation between predicted density, estimated by volume
weighting of a set of normative mineral estimates and bulk
density measured on the same intervals of drill core. A few
outliers on the plot indicate that some of the primary ore samples
are probably partially oxidised but the linear correlation between
the estimates and measured values is strong. Mass weighting of
the same normative mineral estimates produces a different set of
bulk density estimates. Figure 7 shows a plot of the
mass-weighted estimates against the volume weighted estimates.
The error is about six per cent for the majority of the data points
(and the majority of the deposit). If the ore tonnage was
overestimated by six per cent, additional waste stripping or
development would be necessary to make up the shortfall and
this would add directly to operating costs.
FIG 6 - Comparison of bulk density estimated from normative minerals and bulk density measured from drill core from a magnetite deposit.
FIG 7 - Comparison of mass-weighted and volume weighted bulk density estimates from a magnetite deposit.
295
I T LIPTON
THE SAMPLE SUPPORT PROBLEM REVISITED

We have already seen how the variance of density data is affected
by the volume in which it is measured and how the density of
mineral or block aggregates must be averaged on a
volume-weighted basis. Grades, on the other hand, are expressed
on a dry mass basis. The geostatistical requirement of
maintaining equal sample support is typically achieved by
compositing samples to uniform sample lengths but this is only
strictly valid where the density of those composites, and hence
the mass, is uniform. In many base metal deposits this is not the
case.
How important is this? Consider a hypothetical set of four
composites of unit volume with the following Pb grades: one per
cent, three per cent, five per cent and 20 per cent. Assuming, as
before, that the Pb occurs as cerrusite and the density of the
gangue is 2.0 t/m3, then the samples have mass of 2.06 t, 2.18 t,
2.29 t and 3.17 t respectively. The Pb contained in the composites
is therefore 0.0206 t, 0.0654 t, 0.115 t and 0.634 t, respectively.
The total metal in all four composites is therefore 0.835 t and the
average grade is 0.835/(2.06 + 2.18 + 2.29 + 3.17) or 8.6 per cent
Pb. The average grade of the four composites, weighted by
volume is only 7.25 per cent, a figure 16 per cent lower than the
correct grade. This is a significant local bias and may be
accompanied by a similar global bias.
For mineral deposits in which there is a large variation in
density, grade estimates should ideally be weighted by sample
mass. This is standard practice when calculating global tonnages
and grades; the global grade is calculated as the
tonnage-weighted average of the individual block grades.
Theoretically, it appears that mass-weighting should also be
implemented during the estimation of individual block grades
because these are estimated by averaging grades of the local
samples. Mass weighting can be readily achieved using the
grade*density product after compositing the samples to a
constant length or volume. Both the grade*density product and
density can then be interpolated as dependent variables into
computer block models. The block grades can then be
back-calculated by division.
The impact of density-weighting the individual block estimates
will vary depending on the variance of the grades of the samples
selected to estimate the block. If the sample grades are all fairly
similar, density weighting will make little difference to the block
estimate. If there is a mixture of high-grade and low-grade
samples, density weighting will increase the average grade of the
block. It might therefore be expected that sediment-hosted
massive sulphide deposits, which tend to have strong lateral
grade continuity, would be less affected by density-weighting
than, for example, Mississippi Valley Type deposits which have
more irregular grade distributions. In a lead deposit examined by
the author, density weighting of the block estimates increased the
average grade of the deposit by 2.5 per cent.
Why, then, is the practice of density weighting uncommon?
There are perhaps three main reasons for this. Firstly, the density
of the drill samples is usually only determined for a small
proportion of the total assayed database. Although the density of
the samples can often be estimated from the sample grades, as
has been discussed previously, the estimates are rarely able to
account for all the local variations in ore and gangue mineral
assemblages and porosity. The density estimates for the samples
may therefore include a high estimation error and it may be
undesirable to pass this error into the estimates of block grades.
Secondly, failure to weight the grades by mass will, where the
correlation between grade and density is positive, result in a
conservative estimate of grades. Thus, in most cases, weighting
grades on a volumetric basis (equal volumetric sample support)
poses little risk to the project. On the other hand, unnecessary
conservatism in resource estimation may contribute to lost
296
opportunities or may hide the effects of mining problems such as

excessive dilution. Conservative or not, poor estimation practices
should be avoided.
Thirdly, there is a suspicion that weighting grades by density
during block estimation and during summation of the blocks to
create a global resource estimate may involve some double
counting, particularly if density and grade are strongly
correlated. This is a question that requires some further
investigation.
In the case of massive sulphide deposits that have been tested
entirely by core drilling, determining the density of every sample
in the deposit may be possible and is desirable (Bevan, 1993).
Robust, inexpensive procedures can be put in place to collect
bulk density data as a routine aspect of core drilling. The
densities can then be used to improve the estimates of both the
tonnage and the grade of the deposit.
CONCLUSION
The importance of bulk density on resource estimates can clearly
be demonstrated on both empirical and theoretical grounds.
Although there is a complex interplay between factors such as
primary mineral assemblage, alteration and porosity, bulk density
is a continuous, geologically controlled, spatial variable. The
examples presented in this paper show that it should be treated as
such and issues such as combining data from multiple sources,
sample support and the interdependence of bulk density and
grade estimates must be resolved. Incorrect density estimation
can lead to substantial local and global errors in tonnage and
grade estimates. In some deposits, errors in tonnage estimates
arising from poor density estimates can easily be larger than
those which are due to the inadequate definition of the geometry
of the orebody.
Where density shows structured variations across the deposit,
the statistical and geostatistical techniques used for grade
estimation may be equally applicable to bulk density. In these
circumstances, the relationship between grade and density is
particularly important and it must be honoured by the estimation
method.
The two aspects of combining bulk density data that are most
commonly overlooked or misunderstood are as follows:
the relationship between grade and density is non-linear
because grades are usually defined on a mass basis and
density is defined on a volume basis;
when averaging grades, the grades should be weighted by
bulk density, a requirement easily achieved by working with
the grade*density product.
Failure to adhere to these principles can easily introduce
significant errors into resource estimates.
Bulk density is the critical third component in resource
estimation after grade and volume. The importance of obtaining
accurate and representative bulk density measurements cannot be
overstated. Biased density measurements are easily translated to
globally biased resource estimates. If the density of the ore is
overestimated, the mine will not achieve its designed
life-of-mine metal production. More importantly, the cost of
production will increase because of the lower tonnage of ore, and
hence lower revenue, per metre of development. If the bulk
density is unknowingly underestimated, the mine may proceed
with a false understanding of ore loss and dilution and may
therefore overlook opportunities to improve profitability.
Reliable bulk density data can often be collected for only a
fraction of the cost of assaying. With proper analysis of the data
and correct estimation procedures, a better model of the variation
in bulk density across the deposit can be obtained, leading
directly to improved estimates of local and global tonnage and
grade.
ACKNOWLEDGEMENTS
The author is indebted to several companies, including
Tanganyika Gold, for providing permission to publish data
presented in this paper.
REFERENCES
Bevan, P A, 1993. The weighting of assays and the importance of both
grade and specific gravity, CIM Bulletin, 86(97):88-90.
Dadson, A S, 1968. Ore estimates and specific gravity, in Ore reserve
estimation and grade control, Special Volume 9, (The Canadian
Institute of Mining and Metallurgy).
Goovaerts, P, 1997. Geostatistics for Natural Resources Evaluation,

(Oxford University Press Inc).
Lipton, I T, 1997. A review of density determination methods for iron ore
deposit evaluation, in Proceedings National Conference on
Ironmaking Resources and Reserves Estimations, pp 51-56, (The
Lipton, I T, (in press). Measurement of Bulk Density for Resource
Estimation, in Mineral Resources and Ore Reserves Estimation
The AusIMM Guide to Good Practice, (The Australasian Institute of
Lipton, I T, Shaw, W J and Waltho, A E, 1999. Characterisation of Ore
Types and Beneficiation Behaviour using Normative Minerals, in
Proceedings PACRIM 99, pp 427 - 433. (The Australasian Institute
297
New Technologies
P K Fullagar
301
Detailed Orebody Mapping Using Borehole

Radar
A Wellington, G Turner,
I Mason and J Hargreaves
315
Application of the SIROLOG Downhole

Geophysical Tool at Callide Coalfields East
Central Queensland
W Nichols
321
Microseismic Monitoring of Shear Zones and

Related Seismic Activity at Broken Hill
A J Morley, J M Murray
and G C Reed
331
Finding More Ore, Further From the Drill

Hole, With DHMMR
J Bishop and R Lewis
337
Nuclear Borehole Logging Techniques

Developed by CSIRO Exploration and Mining
for the Metalliferous Mining Industry
M Borsaru and
J Charbucinski
347

P K Fullagar1
ABSTRACT
Just as geophysics, in the form of 3D seismic and wireline logging, plays
a key role in petroleum reservoir management, so too will geophysics
play an important role in mining of coal, metals, and minerals in the 21st
century. The commercial and social imperatives driving greater use of
geophysics are reduced costs, increased revenues, and enhanced safety. In
short, superior utilisation of capital and management of risks.
In relatively undeformed sedimentary environments it is possible to
adapt petroleum-style 3D reflection seismic to image mines. This has
been demonstrated most impressively in the Witwatersrand.
Anglo-American, for example, completed a $1 million 3D seismic survey
at Western Deeps in order to site a new $300 million shaft with
confidence. Likewise, 3D seismic has gained rapid acceptance at
Australian coal mines in recent years. Encouraged by these successes,
metalliferous mining companies are supporting research in Canada and
Australia, as well as South Africa, to adapt 3D seismic for more highly
structured, metamorphic terranes.
Because the mining industry is far more diverse than the petroleum
industry in terms of commodities and geological environments, 3D
seismic is not always cost-effective. A plethora of other geophysical
techniques can be applied, many of which are employed in mineral
exploration. Mine applications of geophysics differ from traditional
exploration applications in two main ways: time scales and length scales
are shorter, and boreholes are more plentiful. Borehole geophysical
techniques therefore play a more significant role, both for logging and
imaging. Logging systems detail the variations of in situ physical
properties down the borehole at scales measured in centimetres, while
geophysical imaging techniques can map features located tens or even
hundreds of metres from the sensors. While geologists tend to think of
boreholes in terms of chips and core, geophysicists perceive them first
and foremost as access paths for instruments.
Borehole logging has been employed at iron and coal mines for
decades, mainly to accurately define ore boundaries in delineation holes.
In base metal mines, Outokumpu implemented logging widely for
orebody delineation in percussion holes in the 1980s. The benefit was a
direct cost-saving, arising from substitution of core drilling with
percussion drilling plus borehole logging. The practice has since spread
to other companies. INCO, for example, defines nickel boundaries in
blast holes at Sudbury using conductivity logs. The benefits are in the
form of reduced dilution and enhanced ore recovery, flowing from a more
accurate mine model. Density, natural gamma radiation, magnetic
susceptibility, and conductivity are the principal metalliferous mine
logging parameters because they can be recorded in both dry and
water-filled holes.
Sonic velocity is the premier geotechnical logging parameter, given its
sensitivity to rock strength, stress, porosity, and degree of fracturing.
Sonic is recorded routinely in exploration and geotechnical holes at coal
mines. With the advent of slimline dipmeter, full waveform sonic, and
optical and acoustic scanner tools, the role of borehole logging in
geotechnical evaluations is expanding. Monitoring of strain and
micro-seismic activity enhances safety during mine production.
While qualitative interpretation of logs is adequate for stratigraphic
identification or definition of boundaries, a wealth of quantitative
information, eg density, can be derived from properly calibrated,
repeatable geophysical logs. Moreover, geophysical logs can sometimes
serve as surrogates for geochemical assays, and not only for magnetite
and uranium. At Outokumpus Kemi chromite mine, for example,
gamma-gamma logging provides the basis for grade control. By reducing
reliance on assaying, three benefits can be realised: reduced reliance on
core drilling; lower core handling and assaying costs; and shorter
turn-around times.
Conventional downhole EM and borehole magnetics are used for
near-mine exploration, and for ground sterilisation. Inco has enjoyed
considerable success with borehole UTEM in the Sudbury Basin, for
1.
CNTE, PO Box 6088, St Lucia Qld 4067 and Fullagar Geophysics

Pty Ltd, Level 1, 1 Swann Road, Taringa Qld 4068.
example, and Geopeko has applied three-component borehole magnetics

successfully in the Tennant Creek area. Electrical techniques such as
applied potential and magnetometric resistivity (MMR) are also finding
application at mines. For higher resolution, borehole seismic, radio
imaging, and radar have been invoked, with varying degrees of success,
to delineate orebodies and map structures, or to geotechnically
characterise the rock mass and identify hazards. Success mapping nickel
sulphide shoots with borehole radar has been reported by WMC at
Kambalda.
There are no universal geophysical panaceas, and each mine imposes
different geological, logistical, and economic constraints. Geophysics
will not always be cost-effective. However, the greatest single
impediment to expanded use of geophysics at mines has been the low
level of awareness of geophysics on the part of most mine geologists,
engineers, and managers and, equally, the limited understanding of mine
geology and engineering exhibited by the majority of geophysicists. This
myopia will be remembered as a 20th century affliction!
Blue Sky areas for mine geophysics in the next ten years include
integration of geophysical data acquisition with drilling, enhanced grade
estimation and rock mass characterisation, and the incorporation of
geophysical information into mine models using geostatistical
techniques. All these advances will be predicated on an expansion of
petrophysical knowledge.
INTRODUCTION
Geophysical methods can be classified into two broad categories:
borehole logging, for determination of in situ physical properties
in the immediately vicinity of a drill hole; and geophysical
imaging, for mapping features located tens or even hundreds of
metres from the sensors. Imaging as used here encompasses all
the methods applied routinely in exploration, eg seismic,
magnetics, electromagnetics (EM), gravity, as well as the high
resolution techniques, namely ground probing radar, cross-hole
seismic, and radio imaging. Imaging techniques see through
rock; logging data substitute for drill core.
Geophysical techniques are employed extensively in
exploration for virtually all commodities. 3D reflection seismic
and wireline logging have now become an integral component of
petroleum reservoir characterisation (McWhorter and Torguson,
1995) and production monitoring (Key, Pederson and Smith,
1998). Likewise, the role of geophysics is expanding in mining,
and with good reason. Excluding commodity price and sovereign
risk, uncertainty about ore geometry and rock quality are the
principal threats to mine performance. In mine establishment,
major capital expenditures are committed on the basis of sparse
information. Likewise during extraction, local inaccuracies in
mine models can cause unexpected and costly production
shortfalls, through lost ore or bad ground. When suitable
physical contrasts exist, geophysics has the potential to reduce
these risks by mapping the orebody and its environment, from
the surface, from underground development, or from boreholes.
Geophysics is well accepted at coal mines: wireline logging is
recorded routinely (Davies, 1992), 2D seismic and radio imaging
are established methods, and 3D seismic is becoming more
common. However, metalliferous mines as a whole have been
slower to embrace geophysics. This relative tardiness can be
attributed in part to the greater variety, and often complexity, of
geological environments at metalliferous mines. Other factors are
cultural rather than technical (Mutton, 1994), including the
historical divisions between exploration, feasibility, and
production departments. Nevertheless, 3D seismic is proving its
worth at Witwatersrand gold mines, and several major
metalliferous mining companies have already successfully
integrated borehole logging into their mining operations. The
301
P K FULLAGAR
scale of the economic benefit is often difficult to quantify, not

only because companies are often reluctant to share their
commercial information, but also because the benefits are
relative to would have been scenarios and therefore defy
calculation. In million tonne per annum base metal operations the
economic benefit can run to millions of dollars per year (King,
Fullagar and Lamontagne, 1994; Williams, 1996), sufficient to
drive an operation down the cost curve, and maintain its
competitive position.
Geophysical techniques can be applied at almost any scale and
at any stage of a mining operation. Four broad classes of
application can be defined:
orebody delineation, to maximise ore recovery and minimise
dilution;
rock mass characterisation, to guarantee safety and to
optimise mine design;
exploration and ground sterilisation (eg King, 1996;
Kowalczyk, Logan and Campbell, 1996); and
environmental monitoring (eg King and Pesowski, 1993;
Rutley and Fallon, 2000).
In this paper geophysical orebody delineation and rock mass
characterisation will be briefly reviewed, with special emphasis
on metalliferous mines.
Increased utilisation of proven techniques such as 3D seismic
and conventional downhole EM at mines is assured. Likewise,
utilisation of borehole logging is poised to grow, especially as
new tools and interpretational procedures enhance the reliability
of petrophysical grade prediction and rock mass characterisation.
The need to re-access the drill hole, which is the main limitation
of logging, could be alleviated to some extent by the advent of
slimline logging-while-drilling technology. The future of high
resolution in-mine seismic, radar, and radio imaging is less
certain. Substantial funding is required for development of
instrumentation which is both sophisticated and robust, in order
to move beyond technical successes (Wedepohl et al, 1998).
Reliable and cost-effective imaging tools are required in order to
deliver a performance benefit. In the interpretational arena there
is a need for integration of geological, geophysical, and
geostatistical modelling, both predicated on and leading to a
superior understanding of the underlying petrophysics (Fullagar
et al, 1996a).
Slimline borehole logging probes are available from a number

of manufacturers to measure a wide range of physical properties
(eg Fallon, Fullagar and Sheard, 1997; Firth, 1999). Some
systems, such as the Outokumpu OMSLOGG, have stiffened
cable to permit logging of holes oriented upwards (Figure 1).
The principal applications of borehole logging in minerals mines
are:
1.
geological interpretation (hole-to-hole correlation);
2.
ore boundary definition;
3.
grade estimation; and
4.
geotechnical characterisation of the rock mass.
OVERVIEW OF GEOPHYSICAL TECHNIQUES

Rock property determination and borehole logging
The physical properties of rocks are the foundation of
geophysics. Petrophysical properties can be measured either for
discrete rock samples in a laboratory or continuously in situ via
borehole logging. Petrophysical laboratory analysis of
representative ore and host rock samples is highly desirable prior
to collecting geophysical data at a mine site (Fallon, Fullagar and
Sheard, 1997). Electrical and electromagnetic survey design and
equipment selection is especially problematical without
petrophysical orientation, given that conductivity often varies
over several orders of magnitude in metalliferous mines.
Exploration geophysics relies primarily on the existence of
contrasts in rock properties, but in mines absolute properties are
often important. In particular, knowledge of rock density and
strength has always been a necessity at mines, for ore reserve
estimation and mine design respectively. Density can be
measured routinely on samples submitted for geochemical assay.
Sampling for rock strength testing tends to favour more
competent intervals, sometimes raising concerns as to the
representivity of the results. The potential for continuous in situ
density and rock strength information from calibrated borehole
logging is well recognised at coal mines, but has been largely
ignored at metalliferous mines.
302
FIG 1 - Underground OMS-logg density logging at the Tara zinc mine,

Ireland (photo by G N Fallon).
For geological interpretation and ore boundary definition,

qualitative interpretation of logs is often adequate. If, however,
borehole logs are to be used as a basis for quantitative modelling,
be it for ore reserve determination or geomechanical analysis,
petrophysical calibration of probe responses is essential.
Absolute calibration establishes the relationship between
probe readings and physical standards, eg between
gamma-gamma response and density. This is usually performed
at government facilities, such as the GSC radiometric test site in
Ottawa, where holes have been drilled through synthetic
materials with prescribed properties (Killeen, 1986).
Core-based calibration relates a probe response to a property
of the core (from the same location) as determined in a
laboratory, eg sonic velocity to UCS, or conductivity to grade.
Geophysical borehole logs are usually very detailed, with
sampling interval of a few centimetres. Data volumes can
therefore be relatively large, especially when several parameters
are recorded in each hole. Computer-aided interpretation offers a
means for efficient processing of large multi-parameter logging
data sets. Automated interpretation of petrophysical logs is
THE EXPANDING ROLE OF MINE GEOPHYSICS
commonplace in the petroleum and coal industries (eg Baldwin

et al, 1990; Coudert, Frappa and Arias, 1994; Zhang, Salisch,
and McPherson, 1999) and has been addressed in metalliferous
environments in a number of recent studies, eg Urbancic and
Bailey (1988); Kassenaar (1991); Wanstedt (1992); Emilsson
(1993); Nilsson (1995); McCreary and Wanstedt (1995);
Fullagar, Zhou and Fallon (1999). A wide range of techniques
has been examined, including application of simple if tests,

principal component analysis, factor analysis, discriminant
analysis, neural networks, and multi-variate generalisation of
scatterplot interpretation. Some of these auto-interpretation
algorithms are now commercially available, eg LogTrans
(Figure 2).
FIG 2 - Comparison between the geological log and an automated interpretation, based on density and natural gamma logs and geochemical assays,
or a delineation hole at the Yandicoogina pisolitic placer iron deposit, Western Australia. ASR is the alumina/silica ratio. Stratigraphic order was
enforced during auto-interpretation using program LogTrans (Fullagar, Zhou and Fallon, 1999). Iron ore is confined to the GVU and GVL horizons
(per favour Hamersley Iron Pty Ltd).
303
P K FULLAGAR
Data precision is a pre-requisite for any form of quantitative

interpretation. Precision checks are essential at a routine
operational level to ensure that the logging equipment is
functioning normally. This can be achieved via procedures which
allow the operator to monitor the consistency of instrumental
performance (Fallon and Fullagar, 1995). A simple example of a
precision check is zeroing a conductivity probe in air.
Geophysical imaging
A variety of geophysical techniques, including magnetics and
conventional EM, is available for detecting and delineating
features at ranges up to hundreds of metres from the sensors. The
effective radius of investigation and achievable resolution are
dependent on the technique adopted, the survey specifications,
the local rock properties, and ambient noise conditions.
Specifications for mine surveys are normally more stringent than
those for exploration surveys in terms of resolution, timing, noise
suppression, and access (Williams, 1996). Data can be collected
on or above the original ground surface, in open pits, or
underground from drives and boreholes in a variety of survey
configurations: single hole, hole to hole, hole to surface, drive to
drive, etc.
Conventional downhole EM and borehole magnetics are
employed for in-mine exploration and ground sterilisation, eg
King (1996); Jackson, Fallon and Bishop (1996); Kowalczyk,
Logan and Campbell (1996); Turner et al (1996). The advent of
wideband multi-channel acquisition systems such as MIMDAS
has sharpened the effective resolution and expanded the depth of
penetration of EM and electrical methods. The focus in this
paper, however, will be on the higher resolution imaging
techniques required for orebody delineation and rock mass
characterisation.
Seismic is inherently attractive for both exploration and
in-mine production applications because, in principle, it can
provide resolution of a few metres over ranges from tens to
hundreds of metres. In addition, an enormous wealth of seismic
knowledge, expertise, and technology has been developed in the
petroleum industry, much of which can be adapted for mine
applications. The use of surface seismic at both metalliferous and
coal mines is expanding rapidly.
Underground seismic has been applied experimentally
(Greenhalgh and Mason, 1997). Despite technical successes,
seismic underground is relatively slow and expensive, principally
because of the need to mechanically couple both source and
receiver to bedrock. Hydrophones can be used in water-filled
holes, but at the expense of directional information.
Ground penetrating radar (GPR) offers comparable resolution
to seismic, but usually over a limited range. Salt mines are an
exception, where radar is effective over long ranges (Eisenburger,
Sender and Thierbach, 1993). Surface GPR is used to define
shallow ore boundaries for certain alluvial or lateritic deposits.
Underground, radar data can be collected more readily than
seismic, and there is no restriction to water-filled holes for
cross-hole imaging. Massive sulphide orebodies in resistive host
rocks are an excellent target in principle, behaving as almost
perfect radar reflectors. This has provided the rationale for the
on-going borehole radar trials by WMC for delineation of nickel
shoots at Kambalda, Western Australia (Liu et al, 1998).
However, minor amounts of disseminated sulphide in the host
rocks can significantly attenuate radar signals, severely reducing
its effective range (Fullagar and Livelybrooks, 1994).
Reflection techniques require access to a single hole or
roadway only. Access on at least two sides is necessary for
transmission imaging: the spatial distribution of a physical
property (eg velocity, attenuation, resistivity) is mapped by
transmitting signals from one hole or opening to another.
Tomographic techniques similar to those used in medicine are
commonly employed to construct images from the data.
Resolution is generally lower than for reflection surveys.
304
Radio frequency electromagnetic (RFEM) surveys, for

example, are usually undertaken in a cross-hole configuration.
The RFEM method is intermediate in range and resolution
between conventional EM exploration methods and ground
penetrating radar. In principle, therefore, RFEM has the potential
to play a role in both exploration and mining. Conductive zones
between the holes attenuate the radio signals, and massive
sulphide bodies can give rise to radio shadows. The technique
has been implemented successfully in coal and potash mines
(Vozoff et al, 1993; McGaughey and Stolarczyk, 1991), and is
under investigation in metalliferous mines in Australia (Mutton,
1997; Zhou, Fullagar and Fallon, 1998), South Africa
(Wedepohl, 1993; Campbell, 1994), and Canada (Fullagar et al,
1996b; Stevens and Redko, 2000).
OREBODY DELINEATION
Introduction
An accurate knowledge of orebody geometry and grade is
fundamental to mining. The examples below illustrate how
geophysics can contribute to cost-effective and timely orebody
delineation, and hence improve the economic performance of
mines. More precise orebody delineation can also translate into
less environmental impact. Reduced dilution, for example, not
only increases the head grade but also reduces the energy
expended hauling, crushing, and treating waste rock, and
minimises the volume of tailings.
Ore boundary definition

Defining the limits of mineralisation to high accuracy is the most
common application of borehole logging in mines. Geophysical
definition of mineralisation boundaries is cost-effective because
the net cost of drilling and logging percussion or reverse
circulation holes is less (by about $30/m) than the cost of
diamond drilling. The economic benefit may be realised as a
direct drill cost-saving (if the same total meterage is drilled),
and/or as an improvement in mine performance flowing from a
higher number of orebody drill intercepts (if more holes are
drilled for the same net expenditure on drilling). These more
economical drilling techniques do have their limitations, but
these can often be largely overcome by the logging (Fallon,
Fullagar and Sheard, 1997).
In some cases the advent of logging can alter the economics of
a resource, and hence add to ore reserves and mine revenue. At
Zinkgruvan, Sweden, for example, borehole logging was a
crucial component in the modification of a traditional mining
method, undertaken to permit economic extraction of an isolated
ore lens adjacent to the main lode (L Malmstrom, pers comm).
The ore lens was not sufficiently large to support the cost of the
additional development which would normally have been
required for conventional sublevel stoping. Nor was the grade
sufficiently high to tolerate significant dilution. By logging
cablebolt holes drilled from the existing development drift, the
ore boundaries were defined very accurately, permitting precise
placement of charges to minimise overbreak. More than one
petrophysical parameter was required for reliable discrimination
of ore from waste (Wanstedt, 1992): the ore was uniquely
characterised by low susceptibility, high conductivity, and high
density.
Blast holes can be logged to refine charge placement, size, and
sequencing. King, Fullagar and Lamontagne (1994) described
the use of simple conductivity probes to discriminate ore from
waste in blast holes in vertical retreat mining stopes at Stobie
Mine, Sudbury. The economic benefit was approximately
$C20M in 1993, comprised of increased revenue from enhanced
ore recovery, as well as cost-savings from reduced dilution. The
conductivity log is interpreted in a binary fashion: ore for
responses above a previously-determined cutoff value. The

implementation of borehole logging for ore boundary delineation
need not be expensive or complicated: the system used at Stobie
can be carried by one person, and the original purchase price of
the conductivity probes was $C6000.
High resolution geophysical imaging is used for remote ore
boundary definition. Radar is employed in underground tunnels
in the Witwatersrand to accurately map pyritic auriferous reefs at
ranges up to 30 m (Campbell, 1994). Detailed geophysical ore
delineation from underground development is especially
attractive for deeper reefs, given that its position is
under-sampled for any affordable surface drill hole spacing. For
the same reason, borehole radar has been applied by WMC at
Kamblada to delineate nickel shoots (Liu et al, 1998).
Seismic tomography has been performed in a number of
metalliferous mines in Sweden (Gustavsson et al, 1986), Canada
(McGaughey, 1990; Wong, 1997), USA (Thill et al, 1992), and
Australia (Cao and Greenhalgh, 1995; Luo, Hatherly and Fallon,
1998; Fallon, Newland and Nihill, 1999). Cross-hole seismic
reflections, as well as first arrivals, are interpreted in the
Witwatersrand (Wedepohl et al, 1998). The economic rationale
for one specific application of seismic tomography was detailed
by Dyer and Fawcett (1994), in the context of exploration and
delineation of chromite pods at Shurugwi, Zimbabwe. Shooting
detonators from mine drives, their survey successfully defined a
pod, outlined previously from drilling, as a high velocity zone
and also highlighted a low velocity talcose block in the
talc-carbonate host. Tomographic definition was achieved for 30
per cent of the funds and in nine per cent of the time expended
for conventional drill definition. While this comparison takes no
account of the importance of the grade information provided by
the drilling, it nonetheless serves to establish the viability of
seismic tomography for exploration and delineation from drifts.
Ore continuity
Recognition of ore disruptions, due to pinch-outs, fault
displacements, or intrusives, is vital for mine planning.
The deeper the ore, the stronger the case for all-of-mine
imaging, given the prohibitive cost of closely spaced exploration
drilling. In relatively undeformed sedimentary environments,
petroleum style 3D seismic can be applied. In the Witwatersrand
detailed structural models of auriferous reefs have been produced
using 3D seismic (Campbell, 1994). Confidence in the
seismic-based model at South Deep is such that it is used as the
basis for siting mine pillars. Similarly, Anglo-American have
achieved spectacular results with 3D seismic over a Western
Deeps gold mine (Pretorius, Trewick and Irons, 1997).
Stratigraphy and structure were laid bare to depths of ~2 km.
The Ventersdorp Contact Reef (VCR), the uppermost auriferous
unit, lies on a major velocity contrast at the contact between
lavas (~6300 m/s) and quartzites (~5800 m/s), and therefore
gives rise to a strong reflection. Faults with throws of 15 m or
more can be traced at depths of 1 km (Figure 3). An area of
300 km2 was imaged at a cost of $1 million. This allowed a new
shaft, representing an investment of $300 million, to be sited
with confidence.
FIG 3 - 3D seismic definition of Ventersdorp Contact Reef at Vaals Reef Mine, South Africa (after Pretorius, Trewick and Irons, 1997).
305
P K FULLAGAR
Similarly, impressive 3D seismic images of coal mines have

been produced in Australia (Zhou and Hatherly, 2000).
Experimental 3D seismic surveys in Canada, in the Sudbury
Basin (Milkereit et al, 1997), and at Matagami (Adam et al,
1997), have demonstrated the technical feasibility of the method
for near-mine exploration for base metal sulphide orebodies in
Proterzoic and more complex Archean terranes. Conventional 2D
seismic still has its place, and is of course easier to justify
financially (Spencer et al, 1993; Stevenson and Durrheim, 1997).
Vertical seismic profiling (VSP), involving surface shots and
downhole receivers, is appropriate if targets are steeply dipping.
Placer has employed VSP successfully at Getchell, Nevada, to
map near-vertical gold-bearing structures to more than 1 km
depth at ranges in excess of 100 m from the receiver borehole
(P Kowalczyk, pers comm).
Establishing continuity of ore lenses can be crucial during both
delineation and production phases. For conductive ores in
resistive hosts, electrical methods may provide a straightforward
indication of continuity. In 1982 WMC employed
mise-a-la-masse to establish continuity of nickel shoots at Blair,
Western Australia (Williams, 1996). Inco has applied a
cross-hole resistance technique in the footwall at Levack (North
Range, Sudbury) to test the continuity and establish the plunge of
nickel-copper lenses (A King, pers comm). Interest in application
of other electrical techniques in and around mines is also
increasing, including resistivity tomography (Greenhalgh and
Bing, 1997), mise-a-la-masse (Mwenifumbo, 1997) and borehole
magnetometric resistivity (Bishop, 2000).
Radio frequency tomography has been successfully applied for
nickel-sulphide ore boundary definition at Sudbury (Fullagar et
al, 1996b; Stevens and Redko, 2000) and for detection of copper
ore lenses missed by 40 m-spaced evaluation drilling at the CSA
Mine, Cobar, New South Wales (Thomson et al, 1995). In both
of these cases the host lithology was very resistive and the ore
was highly conductive, providing an optimal geoelectrical
environment. More generally, metalliferous mineralisation is a
spatially variable mixture of both economic and non-economic
sulphides. This is epitomised at George Fisher Mine,
Queensland, where shale-hosted polymetallic zinc/lead ore
occurs with pyrite and pyrrhotite gangue in a series of stratiform
massive sulphide lenses (Forrestal, 1990). Radio tomography at a
frequency of 50 kHz clearly outlines the massive sulphide
mineralisation, but the ore boundary is less certain (Fullagar and
Fallon, 1997).
Sphalerite, especially low-iron sphalerite, is a low conductivity
sulphide, with the result that conventional EM is not always
effective for zinc exploration. Nevertheless, radio imaging has
proved effective for defining fault dislocations within the
zinc-mineralised shale-siltstone sequence at Century, Australia
(Figure 4).
Hole-to-hole stratigraphic correlation based on geophysical
logs can define faulting, especially in sedimentary environments.
Structural and stratigraphic interpretation of geophysical logs is
routine in the petroleum and coal industries, and is well
established in metalliferous provinces such as the Hamersley
Basin, both at the iron mines and in exploration (Kerr et al,
1994).
Towards geophysical grade estimation

The amount of core delineation drilling which can be foregone
(in favour of more economical drilling plus geophysical logging)
expands considerably when grade can be reliably inferred from
petrophysical logs.
Geophysical logs have provided a basis for coal quality
assessment for many years. Ash content has been estimated using
a variety of tools, including gamma-gamma and resistivity
(Edwards and Banks, 1978; Campbell, 1994) and
neutron-gamma (Borsaru et al, 1986; Nichols, 2000).
306
FIG 4 - Re-processed cross-hole radio tomogram (520 kHz), Century zinc

mine, Australia, showing fault dislocation of the mineralised
shale-siltstone sequence (after Mutton, 1997).
For some metalliferous ore types there is a close correlation

between petrophysical properties and grade. Natural gamma
activity is routinely logged at uranium deposits for ore
delineation and grade control (Conaway and Killeen, 1978), and
magnetic susceptibility has served as a direct grade indicator at
magnetite mines for decades (Virkkunen and Hattula, 1992).
Conductivity correlates closely with grade for some base metal
sulphide ores, eg at the Enonkoski nickel deposit in Finland
(Hattula and Rekola, 1997).
Nuclear logging techniques such as spectral gamma-gamma
(Killeen and Schock, 1991) and neutron activation (Borsaru,
Charbucinski and Eisler, 1994) can provide a direct indication of
elemental abundance. However, some mining companies are
reluctant to deploy radioactive sources, for safety and logistical
reasons.
For virtually all gold deposits and many base metal deposits,
the abundance of the economic commodity, or its petrophysical
contrast with respect to its host rock, is insufficient for direct
detection. However, in some cases a strong mineralogical
association exists between ore and a mineral which is readily
detectable. In some Witwatersrand mines, for example, a
correlation between uranium content and gold grade permits
prediction of gold grades using natural gamma logging or face
scanning (Campbell, 1994). In other cases multi-parameter grade
estimates have been derived. Nelson and Johnston (1994), for
example, estimated copper grade at an oxide deposit in Arizona
from a linear combination of density (gamma-gamma), natural
gamma, and neutron activation log values. At the Pasminco
polymetallic Rosebery deposit, Tasmania, the ratio of natural
gamma to density correlates with zinc-equivalent grade (Fallon
and Fullagar, 1997).
The chemical accuracy and spatial resolution of petrophysical
grade estimation need not be high in order to be useful. For
example, within the ore intervals it may be possible to reliably
distinguish high grade from low-grade zones on the basis of logs,
without necessarily providing accurate geochemical abundance
estimates within those zones. At Outokumpus Tara lead-zinc
mine in Ireland, for example, gamma-gamma logging of in-fill
percussion delineation holes is used not only to refine contact
geometry, but also to rank the grade of ore blocks within
individual stopes (J Ashton, pers comm). The combined
zinc+lead grade estimates derived from geophysical logs at Tara

are in fact fairly reliable, but are not used for ore reserve
calculations. Similarly, conductivity is a possible copper ore
cutoff indicator at Mt Isa (Fallon, Fullagar, and Zhou, 2000).
If petrophyiscal properties correlate with grade, automatic
grade estimation is achievable using a variety of techniques,
including multi-variate statistics (eg Emilsson, 1993; Fullagar,
Zhou and Fallon, 1999) and neural networks (eg McCreary and
Wanstedt, 1995). The assumption underlying automated
interpretation is that the relationships between petrophysics and
grade established in control holes are valid for other holes some
distance away. It may be necessary to invoke different control
data sets in different sections of a mine.
Petrophysical grade estimation via automated interpretation of
geophysical logs offers attractive benefits:
1.
reduced reliance on core drilling;
2.
lower core handling and assaying costs; and
3.
shorter turn-around times.

Grades estimated from petrophysics are rarely accepted as a
basis for ore reserve estimation, even in mines such as Laisvall
(Sweden) where the correlation between density and lead grade
is well established (Fallon and Fullagar, 1995). A possible
exception Outokumpus Kemi chromite mine, Finland, where
grade estimates based on density (gamma-gamma) logs are
accepted for mine planning purposes (Figure 5). However, the
validity of petrophysical grade estimates for modelling should
not be judged purely in terms of geochemical accuracy because
the sampling volume or support (per unit length of drill hole)
associated with borehole logs is approximately two orders of
magnitude greater than the volume of core. It is possible,
therefore, that ore reserve models based on geophysical logs
could be superior to assay-based models in some cases.
ROCK MASS CHARACTERISATION

Introduction
The strength and integrity of the rock mass is fundamentally
important in mine design and blasting optimisation. Geophysics
could enhance mine economics and safety by providing the
information for more detailed and complete geomechanical
models.
Sonic logging is the premier geophysical tool for rock mass
characterisation, since seismic velocity and attenuation are
sensitive to rock stress, strength, degree of fracturing, porosity,
and the nature of the material occupying the voids. Sonic
velocity can be related to mechanical parameters such as
hardness and uniaxial compressive strength (UCS), eg Ohkubo
and Terasaki (1977), McNally (1990). The effect of an increase
in the number of fractures per unit length on the compressional
velocity has been documented by King, Pandi and Stauffer
(1978). Seismic attenuation is more pronounced in fractured
rock, especially at high frequencies.
Sonic logging enjoys distinct advantages over testing of
individual core samples insofar as it provides a continuous record
of rock character in situ. At minimum, sonic logs can be used to
optimise the selection of core samples for testing and hence
calibration. Ideally, the sample suite should fully and evenly span
the strength range for each rock type. Properly calibrated, sonic
velocity data can provide rock strength information in weak
zones, which are not amenable to core testing due to core
fragmentation and loss. Thus sonic logging offers a means for
overcoming sample selection bias which can be difficult to avoid
in conventional core testing.
Full waveform sonic (FWS) logs allow determination of the
shear wave velocity, Vs, in addition to the compressional
velocity, Vp (Hatherly et al, 1997). In the absence of FWS data,
FIG 5 - Correlation between logged (gamma-gamma) density and chromite content for serpentine-hosted ore, Kemi Mine, Finland (after Talvisto, 1997).
307
P K FULLAGAR
the shear wave velocity can sometimes be estimated from the

compressional velocity and the density of the rock (Entwisle and
McCann, 1990). The dynamic elastic moduli and Poissons ratio
can be calculated from Vp, Vs, and density (eg McCann and
Entwisle, 1992). Figure 6 shows logs of shear, bulk, and Youngs
moduli derived from sonic and density logs for a hole passing
through the Ventersdorp contact reef (VCR) in the South Deep
gold mine, South Africa (Campbell, 1994). The talcose
Westonaria (lava) which directly overlies the VCR is a relatively
soft, weak rock, in contrast to the competent Witwatersrand
quartzites in the footwall. There are significant variations in rock
engineering properties on a fine scale within the rock units,
probably due to local fracturing.
The dynamic and static elastic moduli are almost equal for
high strength rocks, but for weaker rocks large differences can
arise if the high strains (~10-3) imposed during testing cause
permanent deformation (McCann and Entwisle, 1992). Thus it is
not usually appropriate to simply interchange dynamic moduli
for static moduli in rock mechanics modelling programs.
However, it may be possible to develop a conversion from one to
the other using a control suite of core samples (Asten, 1982).
Sonic and density logging are by no means the only
geophysical options for rock mass characterisation. Wade and
Hickinbotham (1997), for example, derive formulae for friction
angle in terms of sonic, natural gamma, and resistivity. Dipmeter
logs, based on resistivity, can indicate joint and fracture
orientations as well as bedding. For highly detailed structural
analysis, increasingly sophisticated acoustic televiewers (ATV)
and borehole scanners are becoming available (eg Elkington,
1996). Placer is making extensive use of ATV data, often
transformed into virtual core, for its geotechnical characterisation
of the Getchell Mine, Nevada (P Kowalczyk, pers comm).
Single-hole radar reflection imaging can also be effective in
resistive environments, for extrapolation of structures beyond the
hole walls (Olsson et al, 1992).
Although dry hole sonic (Hatherly et al, 1992) and resistivity
(Mwenifumbo and Bristow, 1999) logging tools have been
developed, these parameters are not usually recorded above the
standing water level or in holes inclined upwards.
Blasting optimisation
Major savings have been realised in the past in open pit
operations using seismic refraction to define velocity and hence
indicate blastability. Heinen and Dimock (1976) determined the
relationship between velocity and powder factors at the Ely open
pit mine in Nevada. They estimated that a 17 per cent reduction
in drilling and blasting costs was obtained by exploiting the
seismic information. The capital outlay and operating costs were
insignificant in comparison with the cost reductions.
Measurement-while-drilling (MWD) technology, originally
developed to monitor drill performance, constitutes a potential
source of geomechanical information. Scoble, Peck, and
Hendricks (1989) reported success predicting bench geology and
blastability from rotary drill MWD in coal mines. Hardness and
homogeneity parameters were inferred from MWD at the
Viscaria Mine, Sweden (Schunnesson and Holme, 1997). If ore is
mechanically distinct from its host, MWD can be used to define
ore boundaries, and hence guide charge placement, eg at
Zinkgruvan (Schunnesson, 1990a). In practice the estimation of
rock strength and blastability from MWD data involves statistical
analysis of a number of drill performance parameters
(Schunnesson, 1990b). The statistical basis for interpretation
must be developed during a prior control study.
Hazard detection
Seismic velocity increases with an increase in in situ stress. In
order to minimise safety risks posed by rock burst and/or
collapse events, seismic monitoring equipment has been installed
308
in many coal and metalliferous mines, eg McCreary et al (1992);

Maxwell and Young (1993); Luo, Hatherly and McKavanagh
(1998). While it is not usually possible to predict timing of
individual failures, spatial zones of high risk can be defined. In
addition, microseismic monitoring contributes to overall
understanding of the mechanisms for rock failure.
Refraction seismic can be employed on the ground surface for
detection of potential collapse features above old mine workings,
by virtue of their lower velocity (Schwarz, 1990).
Ground penetrating radar has been deployed at a number of
mine sites for detection of cavities, eg for timely identification of
historic mine workings beneath new open pit mines at Kalgoorlie
(Williams, 1996). Similarly, in-seam seismic has been employed
in coal mines, to detect old workings (Hauser, Jackson and
Gagliardi, 1994) or stressed coal (Mason, 1981) in advance of
active mine headings. At Incos Stobie Mine, GPR recorded in
drifts mapped the surface of an ore pass which was advancing
due to erosion of wall rocks (Fullagar and Livelybrooks, 1994).
The inferred range to the ore-pass at closest approach indicated
that the wall had eroded 15 ft beyond its expected position.
The presence of fracture aquifers can be revealed as zones of
anomalously high conductivity and/or dielectric constant in
otherwise resistive formations. In the Witwatersrand, radar is
recorded in development tunnels to map potentially hazardous
water-filled fracture zones (Campbell, 1994).
CONCLUSIONS
Properly applied, geophysics has the potential to enhance the
economic, safety, and environmental performance of mines.
Fundamentally, the benefit of geophysics flows from risk
reduction, via more complete characterisation of the orebody and
its setting. The role of geophysics at mines will expand in the
21st century as mining companies respond to increasing
commercial, moral, and statutory pressures, and as emerging
technologies mature. In some cases, the introduction of new
mining methods, in response to these pressures, will be
predicated on geophysical imaging. Mechanised continuous
mining, for example, is highly intolerant of geological surprises.
3D seismic reduces the risk of unexpected faults or intrusives in
longwall panels at coal mines. Likewise, the success of more
mechanised and continuous sequential grid mining (SGM) in the
deep gold mines of the Witwatersrand will be heavily reliant on
geophysical imaging, both 3D seismic from the surface, to define
mining blocks prior to mine commencement, and detailed
imaging of the unmined blocks from underground development
(Wedepohl et al, 1998).
Increased use of surface 3D seismic for mine planning
purposes is assured, especially in sedimentary environments. In
igneous and metamorphic terranes, downhole EM and magnetics
will continue to play an important role for in-mine exploration
and ground sterilisation at ranges of tens to hundreds of metres
(Hoschke, 1991; King, 1996). The great benefit of these borehole
imaging techniques is their massive expansion of the effective
search radius of the drill hole (Turner et al, 1996). The deeper the
ore, the greater the potential benefit (Watts, 1997).
Utilisation of high resolution in-mine seismic, radio, and radar
imaging, to define ore boundaries to better than 5 m accuracy, is
still sporadic and relatively experimental at metalliferous mines,
both for reflection and transmission surveys. The fundamental
challenge for in-mine imaging is to reduce costs and time scales,
for both acquisition and interpretation, in order to offer a genuine
alternative to extra drilling. One rate limiting factor is that
equipment designed for surface exploration is rarely suitable for
the rigours of underground deployment (Wedepohl et al, 1998).
To a degree, geophysics is in a Catch-22 bind, insofar as the
investment required to re-engineer instrumentation and develop
robust new interpretation strategies will not be forthcoming until
FIG 6 - Dynamic elastic constants derived from sonic and density logs, Witwatersrand Basin, South Africa. Fine scale variations in rock strength are
evident within the weak Westonaria formation, the hangingwall to the auriferous Ventersdorp Contact Reef (after Campbell, 1994).
the viability of in-mine geophysics in a production setting has

been fully established! Technology development often sits
awkwardly with a production mindset, yet if mining companies
do not actively support development of the equipment
themselves, it is unlikely to attract investment from other
quarters, given the highly specialised nature of the application.
In and near mines, the existing geological model is the starting
point for the interpretation of geophysical data. The geophysical
survey will be a failure unless it adds to an already-substantial
understanding of mine geology. The constraints on interpretation
are therefore far tighter than in grass roots exploration. Existing
mine and geophysical modelling software packages are generally
not well adapted for integrated interpretation of geophysical
imaging data, exchange and display of disparate two- and
three-dimensional information. Levett and Logan (1998) have
illustrated one way forward by inputting mine geology
wireframes into magnetic modelling software to expedite
interpretation of borehole magnetic data. McGaughey and Vallee
(1997) have described a more sophisticated common earth
model approach, originating in the petroleum industry, which
offers simultaneous 3D display of a physical property
distribution, geological surfaces, and geophysical data in various
forms. The visualisation engine is linked to modelling and
inversion programs, to expedite quantitative interpretation of the
geophysical data. The joint ownership of the common earth
model by geologists and geophysicists is a key feature, one likely

to enhance communication and maximise the benefit derived
from geophysics. At the same time, the common earth model
reinforces appreciation of the fundamental importance of
petrophysics, which links geology, geophysics, and mine
engineering.
Borehole logging and scanning provide local information
about the borehole wall and the rock immediately surrounding it.
A wide range of slimline borehole tools is now available
(Killeen, 1997). Borehole logging can be regarded as an
alternative to core drilling, and as such provides mine operators
with options to optimise the drilling budget to suit their specific
requirements (Fallon and Fullagar, 1997). The economic benefit
may take a number of different forms, including: substitution of
core drilling with cheaper percussion drilling plus logging;
reduced expenditure and delays for core tests (assays, rock
strength, metallurgy); and enhanced ore recovery and reduced
dilution via accurate definition of ore boundaries in blast holes.
The existing equipment is satisfactory for most in-mine
conditions, and technical specifications will continue to improve.
In logging as in seismic, the mining industry benefits from
developments in the petroleum industry. Logging systems can be
operated by either mine personnel as well as contractors, and the
digital data they generate are well suited to real-time capture in
fully instrumented mines.
309
P K FULLAGAR
There is great scope for improved grade estimation and rock

mass characterisation from logs and scanner data, as
petrophysical understanding and interpretational aids improve.
However, quantitative interpretation relies implicitly on data
quality: regular probe calibrations, routine precision checks, and
adherence to appropriate data acquisition procedures are
essential to achieve high precision and accuracy.
The main limitation of borehole logging is the need to
re-access the hole after drill completion. Logging after drilling is
a logistical complication, involving as it does mobilisation of
personnel and equipment to the drill holes. This effort is futile if
the hole is blocked. Plastic casing and, more so, logging through
drill rods entail extra expense, and can degrade data quality. To
overcome these difficulties, and to shorten the time between
drilling and interpretation, there is a strong incentive to collect
data during drilling. Logging-while-drilling (LWD) technology is
available for a range of parameters in petroleum wells
(McDonald and Ward, 1976; Allen et al, 1989), but only for
navigational measurements in slimline holes, eg to steer in-seam
gas drainage holes. LWD in mines is conceptually attractive, but
it is unclear whether the potential financial return warrants
development of slimline LWD capability for typical mining drill
rigs as yet (Mason and Hatherly, 1997).
Interpretational aids are in a reasonably advanced state for
borehole logging data. However, there is considerable scope for
integration of both data sets and paradigms. Auto-interpretation
algorithms can operate on assays and geophysical logs
simultaneously and can enforce stratigraphic order, thereby
bringing geophysics, geochemistry, and geology together (Figure
2). Properly calibrated borehole logs also have a role to play in
the interpretation of imaging data, both locally, by constraining
models where they are pierced by drill holes, and globally, by
defining physical property ranges. How best to exploit the highly
detailed (~5 cm sampling) borehole logs to constrain models
developed from data with resolution defined in metres or even
tens of metres is not a trivial issue. A great deal of effort has
been devoted to this upscaling problem in the context of
petroleum seismic (Yang et al, 1994), but this is not necessarily
directly transferable to conductivity, for example. A melding of
geostatistical and geophysical paradigms appears to offer
considerable promise, both for addressing the upscaling issue in
particular and, more generally, for joint interpretation of
disparate data sets. For example, Dimitrakopoulos and Kaklis
(1997) have illustrated how geostatistical simulation and radio
tomography can be combined for a case where conductivity
correlates with copper grade.
Lack of mutual awareness between geophysicists on the one
hand and mining geologists, engineers, and managers on the
other has certainly impeded in-mine implementation of
geophysics in the past, and staff geophysicists are still a rarity at
most mines. Awareness is just the beginning, however. Mine
management must also be convinced that the costs of geophysics,
in terms of contractors, capital outlays (if any), and management
effort, are justified in terms of the mine performance benefits
which ensue. The implementation of geophysics must be tailored
at each mine, to deliver the correct information in the appropriate
form at the opportune time for the right price. The challenge for
geophysics is to achieve a level of reliability and productivity to
justify acceptance as an integral component of the mining
operation. This cannot occur in a vacuum: to realise the benefits,
the mines themselves must continue to actively support and
participate in the on-going refinement of geophysical technology.
ACKNOWLEDGEMENTS
This paper is very much an outgrowth of my participation in the
CMTE MM1 (AMIRA P436) research project into applications
of geophysics in metalliferous mines. In particular, I wish to
acknowledge the explicit and implicit contributions of my
colleagues Gary Fallon (MIM, Brisbane), Peter Hatherly
310
(CMTE/CSIRO, Sydney), Binzhong Zhou (CMTE/CSIRO,

Brisbane), and Xun Luo (CSIRO Exploration and Mining,
Brisbane).
This review has been enriched by the results and experiences
of many others, including Stewart Greenhalgh (University of
Adelaide), Alan King (Inco, Sudbury), Peter Kowalczyk (Placer
Dome, Vancouver), John McGaughey (Mira, Montreal), Iain
Mason (University of Sydney), Andrew Mutton (Rio Tinto
Technical Services, Brisbane), Kevin Stevens (Falconbridge,
Sudbury), Tony Watts (Falconbridge, Toronto) and Eric
Wedepohl (CSIR Miningtek, Johannesburg).
Paul Stewart of Minserve, Brisbane, assisted with preparation
of figures.
REFERENCES
Adam, E, Arnold, G, Beaudry, C, Matthews, L, Milkereit, B, Perron, G
and Pineault, R, 1997. Seismic exploration for VMS deposits,
Matagami, Quebec, in Proceedings Exploration 97, Fourth
Decennial International Conference on Mineral Exploration (Ed: A
G Gubins), pp 433-438.
Allen, D, Bergt, D, Best, D, Clark, B, Falconer, I, Hache, J-M, Kienitz,
C, Lesage, M, Rasmus, J, Roulet, C and Wraight, P, 1989. Logging
while drilling, Oilfield Review, 1:4-17.
Asten, M W, 1982. Geophysical borehole logging and the prediction of
static-load rock strength, in Workshop on Geotechnical Logging of
Boreholes, Australian Geomechanics Society, Melbourne, April 21st,
1982.
Baldwin, J L, Bateman, R M and Wheatley, C L, 1990. Application of a
neural network to the problem of mineral identification from well
logs, The Log Analyst, Sept-Oct, pp 279-293.
Bishop, J R, 2000. Finding more ore, further from the drillhole, with
DHMMR, in Proceedings 4th International Mining Geology
Borsaru, M, Ceravolo, C, Charbucinski, J, Eisler, P and Youl, S, 1986.
Ash determination of black coal in exploration boreholes by the
neutron-gamma method, in Borehole Geophysics for Mining and
Geotechnical Applications (Ed: P G Killeen), Geological Survey of
Canada, pp 261-268.
Borsaru, M, Charbucinski, J and Eisler, P L, 1994. Nuclear in-situ
analysis techniques for the mineral and energy resources mining
industries, Proceedings of 9th Pacific Basin Nuclear Conference,
Sydney, Vol 1, pp 391-399.
Campbell, G, 1994. Geophysical contributions to mine development
planning - a risk reduction approach, XVth CMMI Congress, South
African Institute of Mining and Metallurgy, Vol 3, pp 283-325.
Cao, S and Greenhalgh, S A, 1995. High-resolution seismic tomographic
delineation of ore deposits, Exploration Geophysics, 26:315-318.
Conaway, J G and Killeen, P G, 1978. Quantitative uranium
determinations from gamma-ray logs by application of digital time
series analysis, Geophysics, 43:1204-1221.
Coudert, L, Frappa, M and Arias, R, 1994. A statistical method for
litho-facies identification, Journal of Applied Geophysics,
32:257-267.
Davies, A L, 1992. Geophysics at the coal-face, Exploration Geophysics,
23:87-96.
Dimitrakopoulos, R and Kaklis, K, 1997. Joint geostatistical simulation
of crosshole tomographic and assay data for orebody modelling and
mine planning, MINEIT97 (Ed: G Panagiotou) (Balkema:
Rotterdam).
Dyer, B C and Fawcett, A, 1994. The use of tomographic imaging in
mineral exploration, Exploration Mining and Geology, 3:383-387.
Edwards, K W and Banks, K M, 1978. A theoretical approach to the
evaluation of in-situ Coal, CIM Bulletin, April, pp 124-131.
Eisenburger, D, Sender, F and Thierbach, R, 1993. Borehole radar - an
efficient geophysical tool to aid in the planning of salt caverns and
mines: 7th Symposium on Salt, Vol I, pp 279-284 (Elsevier Science
Publishers).
Elkington, P, 1996. The acquisition and analysis of slim acoustic scanner
log data, Australian Society of Exploration Geophysicists Preview,
No 65, pp 21-32.
Emilsson, J, 1993. Geophysical multi-parameter logging techniques

applied to ore exploration in the Skellefte Field, MSc thesis
(unpublished), Lulea University of Technology, Sweden.
Entwisle, D C and McCann, D M, 1990. An assessment of the use of
Christensens equation for the prediction of shear wave velocity and
engineering parameters: in Geological Applications of Wireline Logs
(Eds: A Hurst, M A Lovell and A C Morton), Geological Society
Special Publication No 48, pp 347-354.
Fallon, G N and Fullagar, P K, 1995. Benchmarking study of geophysical
logging in metalliferous mines, Centre for Mining Technology and
Equipment, Brisbane, Report MM1-95/3.
Fallon, G N and Fullagar, P K, 1997. Optimising the drilling budget with
geophysical logging, in Proceedings 3rd International Mine Geology
Fallon, G N, Fullagar, P K and Sheard, S N, 1997. Application of
geophysics in metalliferous mines, Australian Journal of Earth
Science, 44:391-409.
Fallon, G N, Newland, A and Nihill, D, 1999. Geotomography for
metalliferous mining a case study, McArthur River, Australian
Society of Exploration Geophysicists Preview, No 80, pp 8-13.
Fallon, G N, Fullagar, P K and Zhou, B, 2000. Towards grade estimation
via automatic interpretation of geophysical borehole logs,
Exploration Geophysics, 31:236-242.
Firth, D, 1999. Log analysis for mining applications, ReevesWireline
Services.
Forrestal, P J, 1990. Mount Isa and Hilton silver-lead-zinc deposits, in
Geology of the Mineral Deposits of Australia and Papua New
Guinea (Ed: F E Hughes), pp 927-934 (The Australasian Institute of
Fullagar, P K and Livelybrooks, D, 1994. Trial of tunnel radar for cavity
and ore detection in the Sudbury mining camp, Ontario, Proceedings
of 5th International Conference on Ground Penetrating Radar (GPR
94), Kitchener, Ontario, Waterloo Centre for Groundwater
Research, Vol 3, pp 883-894.
Fullagar, P K, Fallon, G N, Hatherly, P J and Emerson, D W, 1996a.
Implementation of geophysics at metalliferous mines: Cooperative
Research Centre for Mining Technology and Equipment, Brisbane,
Australia, Report MM1-96/11.
Fullagar, P K, Zhang, P, Wu, Y and Bertrand, M-J, 1996b. Application of
radio frequency tomography to delineation of nickel sulphide
deposits in the Sudbury Basin, 66th Annual International Meeting,
Society of Exploration Geophysicists, Expanded Abstracts, pp
2065-2068.
Fullagar, P K and Fallon, G N, 1997. Geophysics in metalliferous mines
for orebody delineation and rock mass characterisation, in
Proceedings of Exploration 97, Fourth Decennial International
Conference on Mineral Exploration (Ed: A G Gubins), pp 573-584.
Fullagar, P K, Zhou, B and Fallon, G N, 1999. Automated interpretation
of geophysical borehole logs for orebody delineation and grade
estimation, Mineral Resources Engineering, 8:269-284.
Greenhalgh, S A and Mason, I M, 1997. Seismic imaging with
application to mine layout and development, in Proceedings
Exploration 97, Fourth Decennial International Conference on
Mineral Exploration (Ed: A G Gubins), pp 585-598.
Greenhalgh, S A and Zhou, B, 1997. A synthetic study on crosshole
resistivity imaging using different arrays, Exploration Geophysics,
28:1-5.
Gustavsson, M, Ivansson, S, Moren, P and Pihl, J, 1986. Seismic
borehole tomography measurement system and field study,
Proceedings of the IEEE, 74:339-346.
Hatherly, P J, Bhati, M, Wylie, G and Schulz, E, 1992. Rock mass
assessment for surface mine applications, Final Report on NERDDC
Project 1314, Australian Coal Industry Research Laboratories.
Hatherly, P J, Fallon, G N, Fullagar, P K and Zhou, B, 1997.
Geotechnical characterisation of the rock mass through sonic
logging, Proceedings of CMTE Conference on Mine Automation,
Brisbane, July 1997.
Hattula, A and Rekola, T, 1997. The power and role of geophysics
applied to regional and site-specific mineral explortation and mine
grade control in Outokumpu Base Metals Oy, in Proceedings
Exploration 97, Fourth Decennial International Conference on
Mineral Exploration (Ed: A G Gubins), pp 617-630.
Hauser, K L, Jackson, M J and Gagliardi, J C, 1994. Seismic analysis

applied to cavity detection in a thin coal seam, Fifth Ground Control
Conference for Midwestern US Coal Mines.
Heinen, R H and Dimock, R R, 1976. The use of seismic measurements
to determine the blastibility of rock, Proceedings of 2nd Conference
on Explosives and Blasting Technologies, Kentucky, pp 234-248.
Hoschke, T, 1991. Geophysical discovery and evaluation of the West
Peko copper-gold deposit, Tennant Creek, Exploration Geophysics,
22:485-495.
Jackson, J C, Fallon, G N and Bishop, J R, 1996. DHEM at Isa Mine,
Exploration Geophysics, 27:91-104.
Kassenaar, J D C, 1991. An application of principal components analysis
to borehole geophysical data, in Proceedings 4th International
MGLS/KEGS Symposium on Borehole Geophysics for Minerals,
Geotechnical and Groundwater Applications, Toronto, pp 211-218.
Kerr, T L, OSullivan, P, Podmore, D C, Turner, R and Waters, P, 1994.
Geophysics and iron ore exploration: examples from the Jimlebar
and Shay Gap-Yarrie regions, Western Australia, in Geophysical
signatures of Western Australian Mineral Deposits (Eds: M C
Dentith, K F Frankcombe, S E Ho, J M Shephard, D I Groves and A
Trench), Geology Key Centre, University of Western Australia,
Publication No 26, pp 355-367.
Key, S C, Pederson, S H and Smith, B A, 1998. Adding value to reservoir
management with seismic monitoring technologies, The Leading
Edge, 17:515-519.
Killeen, P G, 1986. A system of deep test holes and calibration facilities
for developing and testing new borehole geophysical techniques,
Proceedings of Borehole Geophysics for Mining and Geotechnical
Applications Symposium, Toronto, 1983, pp 29-46.
Killeen, P G, 1997. Borehole geophysics: exploring the third dimension,
in Proceedings of Exploration 97, Fourth Decennial International
Killeen, P G and Schock, L D, 1991. Borehole assaying with the spectral
gamma-gamma method - some parameters affecting the SGG ratio,
Proceedings of 4th International MGLS/KEGS Symposium on
Borehole Geophysics for Minerals, Geotechnical and Groundwater
Applications, Toronto, pp 399-406.
King, A, 1996. Deep drillhole electromagnetic surveys for nickel/copper
sulphides at Sudbury, Canada, Exploration Geophysics, 27:105-118.
King, A and Pesowski, M, 1993. Environmental applications of surface
and airborne geophysics in mining, Canadian Institute of Mining and
Metallurgy Bulletin, 86:58-67.
King, A, Fullagar, P and Lamontagne, Y, 1994. Borehole geophysics in
exploration, development and production, in Trends, technologies,
and case histories for the modern explorationist (Ed: S E Lawton),
Prospectors and Developers Association, Short Course Proceedings,
pp 239-275.
King, M S, Pandit, B I and Stauffer, M R, 1978. Quality of rock masses
by acoustic borehole logging, in Proceedings 3rd International
Congress, International Association of Engineering Geology, pp
156-164.
Kowalczyk, P L, Logan, K J and Campbell, T J, 1996. Three component
borehole magnetometry in and around magnetic orebodies, in The
Role of In-Mine Geophysics in Resource Evaluation, Pre-Convention
Workshop, 66th Annual International Meeting, Society of
Exploration Geophysicists.
Lappalainen, P and Lehto, J, 1995. Geophysical borehole logging in ore
boundary detection at Outokumpu mines, in Proceedings Third
International Symposium on Mine Mechanisation and Automation,
Colorado School of Mines, Golden, USA.
Levett, J and Logan, K J F, 1998. Geophysics of the Porgera gold mine,
Papua New Guinea, Exploration Geophysics, 29:472-476.
Liu, Q, Osman, N, Manning, P, Hargreaves, J, Mason, I and Turner, G,
1998. Borehole radar reflection characteristics of nickel sulphide ore
shoots, in Proceedings 1998 Australian Mining Technology
Conference, Centre for Mining Technology and Equipment, pp
333-346.
Luo, X, Hatherly, P and Fallon, G, 1998. Seismic tomography
determination of ore boundaries at Scuddles Mine in Australia,
Australian Society of Exploration Geophysicists Preview, No 74, pp
26-27.
Luo, X, Hatherly, P and McKavanagh, B, 1998. Microseismic monitoring
of longwall caving processes at Gordonstone Mine, Australia, in
Advances in Rock Mechanics (Ed: Y Lin) pp 67-79 (World Scientific
Publishing Co).
311
P K FULLAGAR
McCann, D M and Entwisle, D C, 1992. Determination of Youngs

modulus of the rock mass from geophysical well logs, in Geological
Applications of Wireline Logs II (Eds: A Hurst, C M Griffiths and P
F Worthington), Geological Society Special Publication 65, pp
317-325.
McCreary, R, McGaughey, J, Potvin, Y, Ecobichon, D, Hudyma, M,
Kanduth, H and Coulombe, A, 1992. Results from microseismic
monitoring, conventional instrumentation, and tomography surveys
in the creation and thinning of a burst-prone sill pillar, Pure and
Applied Geophysics, 139:349-373.
McCreary, R G and Wanstedt, S, 1995. Applications of multivariate
statistics and pattern recognition to geophysical logging at Noranda,
in Proceedings 6th Mineral and Geotechnical Logging Society
Symposium, Santa Fe, (in press).
McDonald, W J and Ward, C E, 1976. Logging while drilling: a survey of
methods and priorities, Proceedings 17th Annual Logging
Symposium, Society of Professional Well Log Analysts, pp 1-15.
McGaughey, W J, 1990. Mining applications of cross-hole seismic
tomography, PhD thesis (unpublished), Queens University, Canada.
McGaughey, W J and Stolarczyk, L G, 1991. Tomographic inversion of
EM seam-wave absorption in the Prairie Evaporite Formation,
Saskatchewan: 61st Annual Meeting, Society of Exploration
Geophysicists, Expanded Abstracts, pp 403-406.
McGaughey, W J and Vallee, M A, 1997. Ore delineation in three
dimensions, in Proceedings of Exploration 97, Fourth Decennial
International Conference on Mineral Exploration (Ed: A G Gubins),
pp 639-650.
McNally, G H, 1990. The prediction of geotechnical rock properties from
sonic and neutron logs: Exploration Geophysics, 21:65-71.
McWhorter, R and Torguson, B, 1995. Palacios Field - a 3-D case
history: The Leading Edge, 14:1225-1230.
Mason, I M, Buchanan, D J and Booer, A K, 1980. Channel wave
mapping of coal seams in the United Kingdom, Geophysics,
45:1131-1143.
Mason, I M, 1981. Algebraic reconstruction of a two-dimensional
velocity inhomogeneity in the High Hazles seam of Thoresby
Colliery, Geophysics, 46:298-308.
Mason, I M and Hatherly, P J, 1997. Application of
Measurement-While-Drilling (MWD) and Logging-While-Drilling
(LWD) Technology in Minerals Exploration and Mining, AMIRA
Project P477 Final Report.
Maxwell, S C and Young, R P, 1993. Stress change monitoring using
induced microseismicity for sequential passive velocity imaging,
Proceedings Third International Symposium on Rockbursts and
Seismicity in Mines, pp 373-377 (Balkema: Rotterdam).
Milkereit, B, Berrer, E K, Watts, A and Roberts, B, 1997. Development
of 3-D seismic exploration technology for Ni-Cu deposits, Sudbury
Basin, in Proceedings Exploration 97, Fourth Decennial
International Conference on Mineral Exploration (Ed: A G Gubins),
pp 439-448.
Mutton, A J, 1994. Application of downhole geophysical logging to
lithological correlation and resource assessment in base metal
deposits, AMIRA Symposium on the application of borehole logging
to mineral exploration and mining, Perth, WA, Australian Mineral
Industries Research Association, pp 1-36.
Mutton, A J, 1997. The application of geophysics during evaluation of
the Century deposit, in Proceedings Exploration 97, Fourth
Decennial International Conference on Mineral Exploration (Ed: A
G Gubins), pp 599-614.
Mwenifumbo, C J, 1997. Electrical methods for orebody delineation, in
Proceedings Exploration 97, Fourth Decennial International
Mwenifumbo, C J and Bristow, Q, 1999. Field evaluation of a new
borehole resistivity probe using capacitive electrodes, in Proceedings
of Symposium on the Application of Geophysics to Engineering and
Environmental Problems (SAGEEP), Oakland, California, pp
859-867.
Nelson, P H and Johnston, D, 1994. Geophysical and geochemical logs
from a copper oxide deposit, Santa Cruz project, Casa Grande,
Arizona, Geophysics, 59:1827-1838.
Nichols, W, 2000. Application of the SIROLOG downhole geophysical
tool at Callide Coalfields, east central Queensland, in Proceedings
4th International Mining Geology Conference, pp 321-330 (The
312
Nilsson, P, 1995. Pattern recognition techniques applied to borehole

geophysical data in site investigations: MSc thesis (unpublished),
Lulea University of Technology, Sweden.
Ohkubo, T and Terasaki, A, 1977. Physical property and seismic wave
velocity of rock, OYO Technical Note TN-22, Urawa Research
Institute, Japan.
Olsson, O, Falk, L, Forslund, O, Lundmark, L and Sandberg, E, 1992.
Borehole radar applied to the characterization of hydraulically
conductive fracture zones in crystalline rock: Geophysical
Prospecting, 40:109-142.
Pretorius, C C, Trewick, W F and Irons, C, 1997. Application of 3D
seismics to mine planning at Vaal Reefs Gold Mine, Number 10
Shaft, Republic of South Africa, in Proceedings of Exploration 97,
Fourth Decennial International Conference on Mineral Exploration
(Ed: A G Gubins), pp 399-408.
Rutley, A and Fallon, G N, 2000. Electromagnetic surveys for
environmental applications at mining operations - an Argentinean
and Australian perspective, 70th Annual International Meeting,
Society of Exploration Geophysicists, Expanded Abstracts
(submitted).
Schunnesson, H, 1990a. Locating ore boundaries by drill process
monitoring in percussive drilling - a multi-variate approach:
Presented at Professional Development Seminar on Advances in
Mining Equipment Performance Monitoring, McGill University,
Montreal, March 21-23.
Schunnesson, H, 1990b. Drill process monitoring in percussive drilling a multi-variate approach: Licentiate Thesis (unlublished), Lulea
University of Technology, Sweden.
Schunnesson, H and Holme, K, 1997. Drill monitoring for geological
mine planning in the Viscaria copper mine, Sweden, CIM Bulletin,
90:83-89.
Schwarz, S, 1990. Detection of destressed rock and potential collapse
features above old mine workings by the seismic refraction method,
in Geotechnical and environmental geophysics, Investigations in
Geophysics No 5, Society of Exploration Geophysicists, Vol III, pp
281-287.
Scoble, M J, Peck, J and Hendricks, C, 1989. Correlation between rotary
drill performance parameters and borehole geophysical logging,
Mining Science and Technology, 8:301-312.
Spencer, C, Thurlow, G, Wright, J, White, D, Carroll, P, Milkereit, B and
Reed, L, 1993. A vibroseis reflection seismic survey at the Buchans
Mine in central Newfoundland, Geophysics, 58:154-166.
Stevens, K and Redko, G, 2000. In-mine applications of the radio wave
method in the Sudbury Igenous Complex, 70th Annual International
Meeting, Society of Exploration Geophysicists, Expanded Abstracts
(submitted).
Stevenson, F and Durrheim, R J, 1997. Reflection seismics for gold,
platinum and base metal exploration and mining in southern Africa,
in Proceedings Exploration 97, Fourth Decennial International
Talvisto, T, 1997. In-situ ore assays from geophysical g-g logs: Kemi
chromite ore, Outokumpu Chrome Oy, Finland, Poster paper
presented at 59th Conference and Technical Exhibition, European
Association of Geoscientists and Engineers, Geneva, Switzerland.
Thill, R E, Tweeton, D R, Jessop, J A and Roessler, K S, 1992.
Developments in seismic tomography for minesite investigation, in
Rock Mechanics (Eds: Tillerson and Wawersik), pp 1093-1102
(Balkema: Rotterdam).
Thomson, S, Neil, M, Correia, C and Jeffrey, S, 1995. In-mine
applications of radio imaging, Australian Society of Exploration
Geophysicists Preview, No 59, pp 21-25.
Turner, G, Ramsay, R, Wellington, A, Fogarty, S and King, G, 1996.
Drilling 50 m diameter holes with geophysics - a case study of
downhole EM at Kambalda, Western Australia, in The Role of
In-Mine Geophysics in Resource Evaluation, Pre-Convention
Workshop, 66th Annual International Meeting, Society of
Exploration Geophysicists.
Urbancic, T I and Bailey, R C, 1988. Statistical techniques applied to
borehole geophysical data in gold exploration: Geophysical
Prospecting, 36:752-771.
Virkkunen, R and Hattula, A, 1992. Borehole logging at LKAB
Malmberget Fe Mine, in Analys i Borrhl: Samnordiskt Projekt (Eds:
T From, A Hattula, L Malmstrom, N-E Marinder, T Maki and H
Silvennoinen), Gruv Teknik 2000, Mineral Industry Research
Organisation (MITU), Sweden, Report 92:14T.
Vozoff K, Smith, G H, Hatherly, P J and Thomson, S, 1993. An overview

of the radio imaging method in Australian coal mining, First Break,
11:13-21.
Wade, N H and Hickinbotham, A, 1997. Geotechnical strength
parameters from geophysical logs, International Journal of Surface
Mining, Reclamation and Environment, 11:27-32.
Wanstedt, S, 1992. Geophysical logging applied to ore characterisation in
the Zinkgruvan mine, Sweden, Exploration Geophysics, 23:401-406.
Watts, A, 1997. Exploring for nickel in the 90s, or til depth do us part,
in Proceedings of Exploration 97, Fourth Decennial International
Conference on Mineral Exploration (Ed: A G Gubins), pp
1003-1014.
Wedepohl, E, 1993. Radio wave tomography-imaging ore bodies using
radio waves, 3rd Technical Meeting of South African Geophysical
Association, Expanded Abstracts, pp 85-88.
Wedepohl, E, Trickett, J, van Schoor, M, Grodner, M and Schweitzer, J,
1998. A geophysical toolbox for deep level gold mining, South
Africa, in Proceedings 1998 Australian Mining Technology
Conference, Centre for Mining Technology and Equipment, pp
269-291.
Williams, P K, 1996. Using geophysics in underground hard rock mining

- a question of value and vision, in The Role of In-Mine Geophysics
in Resource Evaluation, Pre-Convention Workshop, 66th Annual
International Meeting, Society of Exploration Geophysicists.
Wong, J, 1997. A crosshole seismic survey at the McConnell ore body, in
Proceedings Exploration 97, Fourth Decennial International
Yang, Y S, Thompson, C J and McCarthy, J F, 1994. The generation of
grid block permeabilities from core data: in Proceedings Asia Pacific
Oil and Gas Conference, Society of Petroleum Engineers,
Melbourne, pp 127-134.
Zhang, Y, Salisch, H A and McPherson, J G, 1999. Application of neural
networks to identify lithofacies from well logs, Exploration
Geophysics, 30:45-49.
Zhou, B, Fullagar, P K and Fallon, G N, 1998. Radio frequency
tomography trial at Mt Isa Mine, Exploration Geophysics,
29:675-679.
Zhou, B and Hatherly, P J, 2000. Pushing coal seismic to its limits
through computer-aided interpretation and 3D seismic, Exploration
Geophysics, 31:343-346.
313
Detailed Orebody Mapping Using Borehole Radar

A Wellington1, G Turner2, I Mason3 and J Hargreaves3
KAMBALDA GEOLOGY AND
MINING METHODOLOGY
ABSTRACT
One of the key objectives in production from any mine is the
minimisation of ore loss and dilution. This can only be achieved when
the distribution of ore within the mine is known. The distribution of ore is
usually evaluated from a combination of drill hole information and
underground mapping data. Three-dimensional models of the ore
distribution are then constructed by interpolating grades, contact
locations and confirming these estimates through mapping. While the
interpolation procedure can incorporate an understanding of the local
geological conditions, it is an approximation and thus susceptible to
some degree of ore loss or dilution. Naturally, such problems increase in
areas of sparse geologic information or greater geologic complexity.
Borehole radar offers an approach to decrease the interpolative aspects
of modelling by providing a continuous image of some ore contacts.
Borehole radar works by building a two-dimensional image of an ore
contact from radio waves bounced off the contact from closely spaced
positions along a borehole. This image can be imported into Datamine
(or a similar design package) and incorporated into geological models of
a given ore block.
The results of a trial survey from the Mariners nickel mine at
Kambalda are presented to illustrate the improved level of detail and
accuracy that can be achieved through the incorporation of borehole radar
into orebody delineation practice. Once implemented on a large-scale this
technique has the potential to greatly increase the cost-effectiveness of
long hole stoping at Kambalda, through increased ore recovery and
reduced dilution.
INTRODUCTION
The proper use of large tonnage extraction techniques on
high-grade orebodies offers the ultimate in low-cost mining of
high-grade material. Consequently the application of bulk mining
methods (such as long hole stoping) to narrow, tabular orebodies
has the potential to deliver substantial financial benefits.
However, inferior definition prior to mining, of high-grade ore
zones often precludes realisation of this objective. This difficulty
in interpolation arises from the nature of vein style orebodies,
and can be further complicated by structural dislocation of the
ore syn- or post-formation. Traditionally, structural off-sets can
be difficult to predict in sparsely drilled areas, and in some cases,
level mapping is not sufficient to accurately define the
distribution of ore between levels. In these cases, extraction of
the long hole block between the two levels can result in excessive
dilution (due to thinning of the ore zone) or ore loss (due to
thickening). Both outcomes affect a mines profitability, so every
attempt must be made to accurately define the orebody between
levels prior to long hole extraction.
There are numerous methods of defining the distribution of ore
within a given area, however this paper will focus on a new
method; Borehole Radar. The Borehole Radar method has been
extensively trialed over a number of years at WMCs Kambalda
Nickel Operations, with the data presented herein indicative of
the type of images acquired. The goal of Borehole Radar is to
obtain a more detailed picture of the ore contacts than has been
possible with drilling alone.
1.
MAusIMM, WMC Resources Ltd, Olympic Dam Operations, PO

Box 150, Roxby Downs SA 5725.
2.
MAusIMM, SenseOre Services Pty Ltd, 4/51 Sorrento St, North

Beach WA 6020.
3.
Department of Geology and Geophysics, Sydney University NSW

2006.
The Kambalda nickel sulphide deposits are located in the Yilgarn

Craton of Western Australia. Numerous descriptions of the
regional stratigraphy and local geology have been published
since the Kambalda deposits were first discovered over 30 years
ago (eg Cowden and Roberts, 1990; Stone and Masterman,
1998). Nickel sulphide deposits are generally located at the base
of ultramafic flows (the Silver Lake and Tripod Hill Members),
on the contact between these flows and the basal Lunnon Basalt.
A thin layer of massive sulphide is generally present on the basal
contact, overlain by matrix (40 to 80 per cent sulphide) and/or
disseminated sulphides. Contacts between disseminated ore and
barren ultramafic may be sharp or gradational, with ore
thicknesses rarely exceeding 5 m. All Kambalda deposits have
been deformed post-deposition by numerous phases of
deformation, which has led to complex ore geometry in some
cases. Late-stage felsic intrusives further complicate the
distribution of ore zones in some deposits. Figure 1 is a
schematic block diagram which summarises the main elements
of the ore zone stratigraphy.
In general, the deposits are small (less than five million
tonnes) tabular in form, and high-grade. Consequently, thin vein
style mining methods have been employed to maximise ore
grade. Overhand cut and fill has been the standard method for
extraction of steeply dipping orebodies, with crown pillars
extracted by long hole stoping. More recently, a primary long
hole stoping method has been introduced, where 10 m height
long hole stopes are extracted between strike drives. While this
method does allow the long hole stopes to be positioned in more
stable ground, the opportunities to gain spatial information on the
ore zone, and apply it to future mining are drastically reduced.
ORE ZONE DELINEATION OPTIONS

The challenge is therefore to accurately delineate the ore zone
prior to drilling and blasting of the long hole stopes. Several
options exist. Extensive diamond drilling can be used to define
changes in ore zone thickness. The highest density drilling
coverage currently used routinely at Kambalda is 20 20 m
spaced drilling, but even this close spacing is not always
sufficient to resolve the ore zone thickness to sufficient detail to
offer minimal dilution. To drill out proposed mining areas on
closer spaced grids is prohibitively expensive. In-stope sludge
drilling is also an option, but sludge drilling typically suffers
from smearing of grade, and even discounting this limitation,
each hole can only result in a single point of information (ie the
ore to waste transition).
A third approach was considered worth trialing. The use of
borehole radar to provide a continuous 2D picture of the
ore/waste contact. It was hoped that the ability of radar to
produce a continuous image of the contact rather than just the
single points available from drilling data would assist in
improving the delineation of ore zones.
BOREHOLE RADAR DESCRIPTION

The Arcolab Borehole Radar System was used to collect all of
the data contained in this paper. The radar system consists of two
downhole probes, a transmitter and receiver, which are
assembled in series. These probes transmit data via fibre-optic
315
A WELLINGTON, G TURNER, I MASON and J HARGREAVES3
LEGEND
HANGING WALL
ORE
ULTRAMAFIC
BASALT
SEDIMENT FREE
CONTACT
BASALT ORE BASALT

PINCHOUT
CONTACT
ORE
200
METRES
FIG 1 - Schematic block model of the Kambalda ore environment (modified after Stone and Masterman, 1998). The ore is located on the contact between
the basal ultramafic flow and the Lunnon basalt. The ore tends to form in channels, although it may not be continuous within the channels. Sediments are
also present in the stratigraphy, although they are generally absent form the immediate ore environment. The felsic intrusives depicted here may be
common or absent depending on the deposit.
cables to a PC where the data are recorded and displayed in real

time. A schematic depiction of the radar system is presented in
Figure 2.
The operating principals of Borehole Radar are similar to
seismic reflection in that reflections are produced when the radar
wave encounters a change in rock properties. The relevant
property for radar is the dielectric constant. At the boundary
between two rock types, a change in dielectric constant generally
occurs. The contrast of this property across an interface
determines the magnitude of the reflection. A large change in the
dielectric constant across a boundary results in a strong reflection
(for example, the contact between massive sulphide and basalt).
A profile of the contact location is constructed by taking a
number of shots at regular intervals along the borehole (generally
around 0.25 m intervals).
A series of trials was performed to confirm that the contact of
interest could be mapped from sludge holes drilled in the
footwall to the ore block. These trials established a number of
parameters essential for collection of a useful radar image of the
contact. Firstly, the holes must be as parallel as possible to the
contact under investigation. Holes filled with saline water cannot
be logged successfully, so such holes must first be flushed clean
with fresh water. Additionally the operating range of this
borehole radar system has been established to be between 4 m
and 25 m in Kambalda rocks. These criteria now provide the
guidelines for future radar delineation designs.
CASE STUDY MARINERS NICKEL MINE

The Mariners nickel mine is located around 60 km from
Kambalda in the Widgiemooltha area. The deposit is hosted in an
amphibolite grade metamorphic terrain, and has been extensively
affected by multiple phases of deformation. This deformation has
316
FIG 2 - Schematic of the Borehole Radar system. At each rock interface, a

portion of the incident energy is reflected with the remainder transmitted.
These proportions are dependent on the difference in dielectric constant
between the two rock types.
resulted in enhancement of the ore profile in some areas of the

deposit, and attenuation of the profile in others. Figure 3 is a
longitudinal section of the orebody which demonstrates the
large-scale disruption of the ore surfaces. This structural fabric is
also evident within each ore block, with numerous disruptions to
the ore profile present in each level.
DETAILED OREBODY MAPPING USING BOREHOLE RADAR
200mRL
0mRL
Ore Block
Outlines
-200mRL
0
400
METRES
499900mN
499700mN
499500mN
499300mN
499100mN
-400mRL
FIG 3 - Longitudinal section of the Mariners orebody.

Note the discontinuous nature of the ore surfaces.
This distribution is primarily controlled by structure.
A typical strike drive backs map is presented in Figure 4. The

mapping highlights the discontinuous nature of the orebody, with
many rapid changes from thick to thin profiles. Many thickened
zones have no obvious structural control, which makes
interpretation of these features between levels highly subjective.
Thick massive sulphide zones, ie at 1040 N, may contain more
than 20 per cent of the nickel metal in the stope and may easily
be missed if the thickening commences between levels.
As a result of the structural complexity described above a
technique that could decrease the interpolative aspects of
modelling by providing a continuous image of some ore contacts
would be extremely valuable in this environment. Extraction of
long hole stoping blocks had been problematic in the past, with
significant massive sulphide left on the footwall after blasting in
some cases. Several test holes were drilled to act as radar
platform holes (eg Figure 5). Radar surveys were completed in
each hole, with high quality reflections obtained from the contact
between Lunnon basalt and massive nickel sulphide. Figure 6
shows a sample data set from the program of trials. The contact
reflection is clear and demonstrates the ability of the radar to
FIG 5 - Cross-sectional view of the development of a thickened zone

between levels. This type of ore zone configuration is reasonably
common, and can be difficult to predict in sparsely drilled areas. The
thickening of massive sulphide sections of the ore profile that
accompanies this structural overprint, significantly up grades the
mineralised interval. A hole drilled to act as a platform for borehole radar
is shown. Radar data obtained from this hole enables the thickened zone
to be accurately delineated and incorporated into stope designs.
map subtle variations in the orientation of this contact to sub

metre precision. Additional examples of Borehole Radar data are
available in Turner et al, 2000 and Trickett, Mason and
Stevenson, 1999.
FIG 4 - Backs mapping of the 103 North level at Mariners nickel mine. The mapping illustrates the rapid changes from thick ore profiles to thin ore
profiles. No obvious structures have been mapped around some of the thickened zones, which makes interpretation of continuation up and down dip
difficult.
317
A WELLINGTON, G TURNER, I MASON and J HARGREAVES3
FIG 6 - Sample Borehole Radar data from Mariners nickel mine. The horizontal axis corresponds to the borehole trajectory, with a shot taken every 25 cm.
The vertical axis indicates the distance of features from the hole. The direct arrival is a signal which travels along the shortest path between the radar
transmitter and receiver; ie along the borehole. The reflection from the massive sulphide/basalt contact is clearly visible commencing 8 m from the hole
and gently dipping away from the hole.
APPLICATION OF BOREHOLE RADAR DATA

Collecting data in a single hole can not assist greatly in achieving
an optimal stope design. To achieve this, a large number of holes
must be drilled to gain a reasonably complete coverage of the
contact under investigation. Depending on the complexity of the
geology, holes may have to be drilled on sections as close as 5 m
apart. The radar images obtained in each hole can and have been
converted to strings in 3D space, then imported into Datamine
for incorporation into stope wireframes. This information adds a
considerable level of detail to the geological interpretation, and
results in the most detailed and accurate representation of the
geology currently possible.
For the purpose of illustrating the potential benefits of
incorporating borehole radar into orebody delineation practice in
this environment, Figures 7 through 9 illustrate a hypothetical
example of how Borehole Radar may be applied systematically
to stope delineation on a large-scale. Figure 7 shows the initial
case of two development levels with backs mapping to define the
ore. A fault cuts through the profile, and has resulted in
thickening on one level, but little effect on the other. The
important question is to what extent the ore zone between the
two drives is thickened. A linear interpretation can be obtained
simply by wireframing the two backs maps in Datamine (Figure
8).
To enable borehole radar delineation of the stoping block,
sludge holes would be drilled at 5 m intervals in the footwall,
sub-parallel to the ore zone. In this case, the contact between the
ore and waste is considerably further into the footwall than
indicated by the linear interpretation in Figure 8. Interpretation of
the image produced by the radar leads to a single string in 3D
space. A collection of these strings is then added to the orebody
wireframe around the thickened zone to improve the accuracy in
this region (Figure 9).
318
FIG 7 - This diagram illustrates the starting point of most stope designs;
two strike drives with backs mapping. The red area represents the ore
zones.
Similarly, a thinned zone exists toward the opposite end of the

stope. This zone too would be delineated by a series of Borehole
Radar surveys. In each of these hypothetical (however realistic)
cases, the actual ore profile is significantly different from the
linear interpretation produced by the wireframe. The 3D orebody
model of greatly increased accuracy results from the combination
DETAILED OREBODY MAPPING USING BOREHOLE RADAR
FIG 8 - Wireframing the mapping from the two levels results in this
interpretation of the distribution of the ore between the levels. Note that
the fault has also been wireframed, and that the Radar platform holes
have been planned in the footwall of the ore zone.
$A12 000 per tonne, the total dollar value of the nickel in the
stoping block is $A924 000. If the thickened zone is not
extracted (as might be the case if this zone occurred between
levels), the extracted value is reduced by $A288 000. This drop
in revenue is significant, and may represent the difference
between extracting the block profitably or at a loss.
In this example, definition of the exact location of the footwall
contact through use of borehole radar would allow the following:
accurate definition of thickened zones up and down dip;
definition of thinning zones;
subsequent allocation of any required remnant pillars to the
thinnest zones;
complete extraction of ore within thick zones; and
optimisation of the entire stope (ie extract only the zones that
yield the highest NPV).
It may be that the highest NPV case results when only the
northern 30 m, and southern 10 m of the stope are extracted,
while the thin zone between the 1020 N and the 1000 N is not
mined. This approach will result in extraction of around 87 per
cent of the nickel metal in the stope for around 67 per cent of the
cost of mining the whole stope. Accurate delineation allows this
type of analysis to be completed, with the result being more
cost-effective mining.
CONCLUSIONS
The goal to increase production efficiency through bulk mining
methods has resulted in a requirement to better define stoping
blocks prior to extraction. Borehole Radar has been identified as
a possible method of achieving this and a series of successful
trials have been completed. Excellent quality images of the
massive sulphide/basalt contact can be achieved through use of
this method. Incorporating these data into stope designs results in
a highly accurate representation of the orebody, which can then
be used to plan the optimal extraction of the stope. The accurate
delineation of the ore outlines can lead to increased ore
extraction and/or reduced dilution with each having a significant
impact on the grade and overall economics of the stope. As such,
the utilisation of Borehole Radar data in combination with other
available geological data can play a major role in increasing the
grade, and improving the predicability of long hole mining
operations at Kambalda.
ACKNOWLEDGEMENTS
FIG 9 - This diagram shows the final stope design after the radar surveys
have been completed and the resulting information incorporated into the
original wireframe. The zone of thickening is represented in purple, while
the zone of thinning is represented in yellow.
of mapping and Borehole Radar information. This model can

then be used with confidence to design the stope so all ore is
extracted and minimal dilution is achieved.
FINANCIAL IMPLICATIONS
The financial benefits of this approach to long hole stoping may
be significant. As an example, extraction of a long hole stope
above the strike drive depicted in Figure 4 will be considered
(around 60 m strike length). For simplicity, it is assumed that the
ore profile continues at the mapped thickness over the 10 m
vertical height to the next level. Given approximations of the ore
grade and thickness, the total nickel metal in the block is around
77 tonnes. Of this, 24 tonnes are located in the thickened massive
zone (31 per cent of the total). Assuming a nickel price of
The authors thank WMC for permission to publish this data, and
acknowledge the contributions of the following people; A Hill,
C Allison, D Mapleson, J Battershill, and H Golden. This work
was greatly assisted by the ARCO Geophysical Research
Beneficiation. Recent development work on the borehole radar
has been supported by an Australian Research Council
Collaborative Grant and by WMCs Group Technology.
REFERENCES
Cowden, A C and Roberts, D E, 1990. Komatiite hosted nickel deposits,
Kambalda, in Geology of the Mineral Deposits of Australia and
Papua New Guinea (Ed: F E Hughes) pp 567-581. (The Australasian
Stone, W E and Masterman, E E, 1998. Kambalda nickel deposits, in
(Eds: D A Berkman and D H Mackenzie), pp 347-356 (The
Trickett, J C, Mason, I M and Stevenson, F, 1999, Borehole Radar at an
underground Ventersdorp Contact Reef Site: Presented at the 1999
South African Geophysical Association Conference.
Turner, G S, Mason, I M, Hargreaves, J E and Wellington, A, 2000.
Detailed orebody mapping using borehole radar, Exploration
Geophysics (in press).
319
Application of the SIROLOG Downhole Geophysical Tool at

Callide Coalfields East Central Queensland
W Nichols1
ABSTRACT
Callide Coalfields produces a sub-bituminous, very subhydrous, low
rank, steaming coal with good combustion properties, primarily for
domestic power generation.
Geological input into planning of coal extraction requires expedient
production of detailed structural, lithological, analytical, geophysical and
geotechnical data. The CSIRO-developed SIROLOG borehole logging
software and hardware system allows for detailed spectrometric analysis
of coal/interburden. The gamma-gamma and neutron-gamma SIROLOG
techniques both provide improved prediction of ash content over other
borehole techniques. The neutron-gamma technique also predicts density
and some important ash constituents (especially iron, alumina and silica).
The software/hardware configuration allows for quick, easy interpretation
of coal seam depth and lithology, parting thickness and ash/ash
constituent contents immediately after a borehole has been logged.
Within certain mine areas in the Callide Basin, the iron content of
mined coal is widely variable in both vertical and lateral extent within
and between seams. This variability creates problems in predicting the
slagging behaviour of as-received coal during combustion, especially
where the iron content of the ash exceeds 19 per cent. Using the
SIROLOG tool in co-operation with traditional borehole sample
laboratory analysis results, patterns in this variability can be determined
and related to geological properties and process of formation of the coal
seams. Subsequently, areas of high iron can be predicted in advance of
mining.
1.
MAusIMM, Senior Geologist, Callide Coalfields Pty Ltd, PO Box

144, Biloela Qld 4715.
Since 1992, on-going collection and interpretation of SIROLOG

downhole geophysical log data has significantly contributed to a number
of research projects related to gaining a preliminary understanding of the
iron distribution and formation during the Coal Measures deposition.
Because the probe was able to record variations in coal mineral matter
content, it was found to complement more conventional techniques such
as reflected light petrography, X-Ray Diffraction and SEM techniques
that examine the mineralogy of iron species. Additionally, software has
been developed to allow integration of SIROLOG data into existing
geology databases, enhancing coal quality modelling and daily prediction
of parameters such as raw ash, silica and iron contents that are indicators
of ash fusion propensity.
The correlation of SIROLOG-derived iron geophysical traces between
drillholes and other sample points highlighted lithological and structural
controls in the iron distribution in the sequence previously not detected.
This work has allowed the development of a methodology for predicting
iron in advance of mining and reconciling as-sold coal quality to that
predicted. Predictions are currently being assessed at the coal face as
mining progresses through high-iron zones.
INTRODUCTION
The Callide Coal Measures Basin is located approximately 120
km SW of Gladstone and 450 km NNW of Brisbane in eastern
Central Queensland. The coalfield covers an area of about 18 000
hectares and is located within a NW to SE trending synclinal
basin 22.5 km long by 8 km wide (Figure 1). Present open-cut
mining operations are located in the Dunn Creek, Trap Gully and
The Hut areas in the south, and at Boundary Hill in the NW of
the basin.
FIG 1 - Callide Coalfields location plan.
321
W NICHOLS
At Boundary Hill, the majority of the coal is won from the

Callide Seam Member of the Callide Coal Measures. This unit is
of Late Triassic age and contains alternating sequences of
conglomerate, sandstone, siltstone, mudstone, carbonaceous
shale, and sub-bituminous durainous coal of varying hardness
and thickness. A detailed description of the geology of the area is
given by Biggs et al (1995). Current mining seam thicknesses are
16 - 21 m at Boundary Hill. Seam thicknesses down to and
including 0.3 m can be mined. Figure 2 outlines some typical
geophysical signatures of the various coal seam horizons. Coal
quality parameters vary widely vertically and laterally, eg Total
Moisture (16 - 20 per cent), Raw Ash (8 - 28 per cent), Iron
oxide in Coal Ash (0.5 - 72.0 per cent), and Initial Deformation
Temperature, reducing atmosphere (1080 - 1600+C).
Throughout 1993, a number of these research projects related

to gaining a preliminary understanding of the iron distribution
and formation at Boundary Hill Mine were undertaken:
Dr J Patterson, CSIRO - Division of Coal and Energy
Technology (Patterson and Marvig, 1993) investigated
through SEM techniques the mineralogy of iron species;
Dr J Esterle, CSIRO - Division of Exploration and Mining
(Esterle 1994) examined the vertical and lateral variability in
petrographic composition and iron content in the context of
the geological environment under which the deposit formed;
Dr C R Fielding, University of Queensland (Fielding 1993)
investigated the sedimentological and structural controls on
high iron distribution; and
Dr E Baafi, University of Wollongong (Baafi, 1993) began
preliminary investigations into using indicator kriging and/or
conditional simulation as suitable modelling algorithms for
predicting the behaviour of the iron content on a
seam-by-seam basis.
Prior to July 1993, Auslog performed the field logging portion
of the task. In July 1993, Callide commissioned its own Sirolog
probes and has been using these successfully since then. The
information gained enables updating of the geological quality
databases which in turn allowed rigorous statistical analyses of
available values (by seam).
SIROLOG METHODOLOGY
It has been shown through past research and practice at Callide
that the prompt neutron-gamma method (neutron-capture) can be
employed for the determination of ash and some of its
constituents in coal seams which are intersected by boreholes
(Charbucinski et al, 1986; Biggs 1990, 1991; Borsaru et al, 1991,
1993a, b).
All the boreholes logged are water-filled and the probe is
centralised during logging operations. The logging probe used is
of 70 mm external diameter and is fabricated from aluminium
3.2 mm thick. The scintillation detector in the probe is a 76 38
mm BGO (Bi4Ge3O12) detector. To protect the detector, the
section of the exterior of the probe barrel around it (about 40 cm
in length), is painted with 10B. This reduces the background in
the capture spectrum produced by the thermal neutrons which
interact with the probe, the photomultiplier and the electronic
components.
The neutron source used was initially 5.0 g of 252Cf (source
activity = 118.4 MBq). Owing to the rate of source decay, this
was replaced by a 10 g source in 1997. The source to detector
distance is 15 cm. The capture spectrum is dumped onto the hard
disk on the logging vehicles desktop computer for every 10 cm
logged. The logging speed is 2.5 m/min.
PRINCIPLE OF RAW ASH DETERMINATION
FIG 2 - Long-spaced density geophysical response for borehole C2027.
Historically, Callide has experienced sporadic supply problems

to its customers. When mining from various sections of its
deposits, coal with an unusually high and variable iron oxide in
ash content is encountered. Iron contents above 19 per cent
appear to be a pre-condition, if not probably causation, of
slagging in boilers. A lack of information for mining planning
purposes prompted a major investigation into explaining the iron
oxides formation, occurrence and measurement. The work
program was performed by Technical Services staff in
combination with external academic researchers.
322
The ash content of a coal sample is the weight percentage of

residue after combustion; and is closely related to the mineral
content of coal. The determination of ash in coal seams by the
neutron-gamma method relies on the correlation between ash
content and a combination of two or three of the main
constituents of ash, such as Al, Si and Fe (see Figure 3). Since
the reaction probability of capture-gamma ray emission per
nucleus for Al is much lower than for Si and Fe, the
determination of ash relies mostly on the correlation between ash
content and the sum of Fe and Si contents (Charbuckinski et al,
1986).
The neutron-capture technique can cope with wide variations
in Fe or Si, because their contributions to the ash are measured
separately in the capture spectrum and not through the average
atomic number (Zeq) used in the gamma-gamma method
(Borsaru et al, 1993b).
APPLICATION OF THE SIROLOG DOWNHOLE GEOPHYSICAL TOOL
The logging method described here employs the neutron

capture reaction. The compound nucleus formed in neutron
capture decays almost instantaneously, emitting gamma
radiation. Owing to its high hydrogen content, coal is an
excellent matrix for this technique. The neutrons emitted are
thermalised by colliding with the hydrogen nuclei present in
coal, and they subsequently interact with the nuclei from the coal
matrix. An advantage of this technique is that the major gamma
rays, produced by the main constituents of the mineral matter in
coal (Al, Si, Fe, Ca, Ti, S), have energies above 3 MeV (Table 1).
This level makes the prompt neutron-gamma method less
sensitive to interferences from other neutron interactions
(Charbuckinski et al, 1986).
An example of a full neutron-gamma spectra in a coal seam
(for a high ash, moderate Fe zone) is shown in Figure 4. The
peaks seen in the spectra are from B (0.48 MeV), H (2.22 MeV),
and Si (3.54 and 4.93 MeV). The resolution of the BGO detector
and the counting statistics are insufficient to observe clearly
identifiable peaks corresponding to Fe, Al, Ca, and Ti. Although
individual concentrations cannot be measured, it is possible to set
windows in the spectrum which correspond to combinations of
these elements. Current calibration equations for ash integrate
counts in the 2.6 - 5.18 MeV region.
FIG 3 - Graph of ash Vs Si and Fe for the A1 seam.
PRINCIPLE OF IRON DETERMINATION

The variation in the concentration of Fe in the two coal zones
from the same borehole as displayed in Figure 4 is evident in
Figure 5 through the variation in the intensity of the dominant Fe
peaks at 7.63, 7.64, and 9.30 MeV in the capture spectrum (these
are not separated due to the poor resolution of the scintillation
detector). The two gamma-ray spectra shown were calculated
averaging the count rates over an approximately 2 m thick
interval, one having low ash and Fe and the other high ash and
moderate Fe. Fe values are calculated by integrating counts in the
energy window 7.0 - 8.0 MeV and divided by the count-rate in
the energy window 2.08 - 2.38 MeV (H). This window measures
the variation of the thermal neutron flux surrounding the probe at
the location of the BGO detector (Borsaru et al, 1988).
Iron-sourced gamma rays have higher energies than many
gamma rays released in capture process by other elements
commonly present in coal, and therefore are easily
distinguishable in the neutron-gamma spectrum. Possible
interference could be produced by Cl (7.41, 7.79 MeV) if present
in noticeable quantities (Borsaru et al, 1991). When there is a
large variation in Si (7.20 MeV) and Al (7.69, 7.72 MeV) in the
coal seams, they can also present a source of interference,
although not of significant magnitude.
TABLE 1
Neutron capture data for major components in coal ash.
Capture
Cross-sections
(barn =
10-12 cm)
Major
Gamma Rays
(MeV)
Gamma Ray
Intensity (per
100 neutron
captures)
Aluminium
(26.98)
0.23
7.72
7.69
27.40
4.20
Silicon
(28.09)
0.16
8.47
7.20
6.38
4.93
3.54
2.44
7.80
12.40
62.70
68.00
Chlorine
(35.45)
33.00
7.41
7.79
8.58
10.40
8.55
2.91
Iron
(55.85)
2.55
7.65
7.63
8.88
9.30
22.10
27.20
0.74
4.15
Manganese
(54.94)
13.00
7.06
7.16
7.24
7.27
11.06
5.80
12.10
3.36
Titanium
(47.90)
6.10
6.76
6.56
6.42
4.88
1.38
24.20
4.70
30.10
5.20
69.00
Calcium
(40.08)
0.43
6.42
4.42
1.94
38.90
14.90
72.50
Sulphur
(32.06)
0.52
8.60
7.80
5.42
4.87
3.22
2.93
2.38
0.84
2.70
3.90
59.10
11.50
27.10
22.30
44.50
75.50
Element
(Atomic Mass)
DATA INTERPRETATION AND DISPLAY

Gamma-ray spectra recorded downhole are transferred to a
binary raw data file. Modularised, Windows-based interpretation
software (Winview) has been developed in collaboration with the
CSIRO -Division of Exploration and Mining to allow processing
of this data on desktop or notebook computers (Figure 6).
Options include nomination of variables of interest, eg density,
display, seam/interval selection, calculation of quality parameters
and output of data to printer or ASCII files for transfer to
spreadsheet or other mine planning software. Figure 7 displays
ash and iron ratio trace data transferred to spreadsheet for a
borehole with high iron in ash levels, revealing the variable
nature of the iron distribution and its effect on the ash ratio
values (and subsequent raw ash estimates).
STATISTICAL ANALYSES
Data provided by the SIROLOG probe was used as an adjunct to
existing laboratory analyses to enable interrogation of iron
Modified after Charbucinski et al (1986).
323
W NICHOLS
variations on a seam by seam basis. Basic statistical analyses

have been undertaken by both mine personnel and Dr E Baafi
(Baafi, 1993). He found that there is a significant difference in
the number of samples for each seam (as much as at least twice
SIROLOG iron data compared with the number of analytical
data). This effect is greater for the lower seams where the
number of samples decrease (see Figure 8). In general, both the

mean and median of iron do not vary considerably from seam to
seam, with the A3X having both the highest mean and variability
of iron, while the A5 seam has the lowest mean and variability in
Fe.
FIG 4 - Full neutron-gamma spectra in a coal seam for a high ash, moderate iron zone.
FIG 5 - Intensity of the two dominant iron peaks in the capture spectrum.
324
FIG 6 - Windows-based interpretation software (Winview).
FIG 7 - Ash and iron ratio trace data transferred to spreadsheet.
Figure 9 graphically displays a comparison between laboratory

analyses and SIROLOG estimates, by seam, for ash and iron in
the original calibration cored holes. Baafi (1993) found
significant correlations between analysed iron and silica,
aluminium and SIROLOG iron. The ash estimates from the
neutron gamma SIROLOG system (between the original
calibration ranges of nine to 40 per cent) are highly correlated
with lab analyses (r2 = 0.93 ; rms deviation = 1.94 per cent ash).
More scatter is seen for the iron estimates (r2 =0.79 ; rms
deviation = 0.88 per cent total iron), with Biggs (1991)
accounting for the difference between an analysed and a derived
iron value as being due to one or a combination of the following
factors:
The drill core samples used for calibration are smaller

samples than that logged by the SIROLOG probe and are less
representative (100 mm versus ~ 1000 mm)
Calibration is based on plotting ash and ash analyses versus
probe response to obtain the linear relationships. Australian
black coals commonly contain one per cent carbonate which
is lost upon ignition, and consequently is not accounted for
by the multiple linear relationship.
Sample laboratory error.
Random error.
Histograms were plotted for both the analytical and SIROLOG
estimated iron values for each seam. In general most iron
populations are normally distributed, the SIROLOG estimates
325
W NICHOLS
FIG 8 - Comparison of real data and SIROLOG estimated data.
TABLE 2
Principle iron-bearing minerals identified in Callide coal.
Group
Mineral
Formula
Abundance range
(% vol of mineral matter)
Silicates
Illite, Illite-smectite
Expandable lattice clays
Chlorite
Biotite
K(Al,Fe)2Al(Si3O10)(OH)2H2O
Variable
(Mg,Fe,Al)6(AlSi4O10)(OH)2
K(Mg,Fe)3(AlSi3O10)(OH)2
0-1
0-1
trace
trace
Oxides
Goethite
Haematite
Limonite
Ilmenite
HFeO2 or FeO.OH
Fe2O3
FeO(OH).nH2O
FeTiO3
0 - 20
0 - 10
trace
trace
Carbonates
Siderite
FeCO3
0 - 79
Sulphides
Pyrite
Chalcopyrite
Pyrrhotite
FeS2
CuFeS2
Fe(1-x)S
0-3
trace
trace
Sulphates
Coquimbite
Halotrichite
Natrojarosite
Copiatite
Fe2(SO4)39H2O
FeAl2(SO4).22H2O
NaFe3(SO4)2(OH)6
Fe14O3(SO4)1863H2O
trace
trace
trace
trace
After Biggs (1995).
tend to be more symmetrically distributed than the analytical iron

grades (the analytical values are slightly positively skewed), but
this may be due to the reduced number of data points used
(Baafi, 1993)
CHARACTERISATION OF MINERAL MATTER

Detailed work on mineral matter has been undertaken by the
following:
Ward (1990) who subjected coal samples to low temperature
radio-frequency ashing and XRD techniques;
Faraj (1993); and
326
Patterson and Marvig (1993), whose work involved scanning

electron microscopy and microprobe analyses of coal seam
composites. Results of this work are summarised in Table 2.
The important minerals detected in Callide Basin coals were in
decreasing order, kaolinite, siderite, goethite, quartz and pyrite.
Trace minerals detected were titanium oxide, ilmenite,
pyrrhotite, sphalerite, zircon and chalcopyrite. Calcium and
magnesium appear to be organically bound in the coal. Three
iron containing minerals, siderite, goethite and pyrite account for
the bulk of iron present in the coal samples examined. The
siderite has been found to be partially oxidised to an iron
hydroxide phase, suggested to be cryptocrystalline goethite
FIG 9 - Graph of the comparison between laboratory analyses and SIROLOG estimates
327
W NICHOLS
(FeOOH). While relative amounts of siderite and associated

goethite are not known, these minerals are dominant throughout
and it is clear that changes in their abundance account for the
high and variable iron contents of coal (Patterson and Marvig,
1993).
Additionally, two main forms of siderite with essentially
identical chemical composition were observed by Faraj (1993) in
variable amounts:
a.
siderite nodules formed early in diagenesis; and
b.
cleat/vein and other infillings in coal, presumably formed

much later after consolidation of the coal.
The chemical composition of siderite was essentially the same
regardless of form, locality or stratigraphic depth and, therefore,
is not a useful indicator of changes in depositional environment
or different stages of diagenesis. The average formula
determined by Patterson and Marvig (1993) is Fe0.95 Mn0.02
Ca0.01CO3.
Attempts were made to investigate the correlation of
SIROLOG iron and silica data with mineral matter
determination. However, the simple mineralogy and known
chemistry of the major minerals do allow for good estimates of
siderite and pyrite contents to be obtained from chemical
analyses for inorganic C and pyritic S in the coal. The balance of
total iron could then be expressed as goethite.
IRON FACIES DISTRIBUTION

Fielding (1993) has studied the distribution and origin of the iron
facies from a geological perspective by combining outcrop
observations with detailed chemical analyses of core samples,
SIROLOG iron trace fence diagrams, and a more extensive set of
SIROLOG iron estimates and chemical analyses on a seam by
seam basis.
Contouring of iron in ash was initially conducted on a seam
basis and this revealed a pattern of east-north-east trending highs
and lows, in positions which varied between seams. This trend is
coincident with the dominant orthogonal joint sets (50 and
310) in the mine, as defined in igneous intrusions, faults, joints,
coal cleats and air-photo linears. However, both outcrop
observations and quantitative iron data suggest that the
distribution of iron is too complex to be adequately expressed in
seam by seam maps. Rather, iron minerals occur as an
overlapping network of laterally discontinuous patches and zones
throughout the coal body, across the mine area. To overcome
this, a working section composited iron colour-shaded contour
map was generated. There appears to be some general
association with structural features.
Fielding (1993), in agreement with previous workers, found
that much of the iron is present as siderite, but in two forms. The
first phase of siderite formed during the accumulation of shallow
burial of the formative peat. A second, more widely distributed
phase, formed much later at elevated temperatures and associated
with circulating groundwaters. The second phase overlaps
considerably with the first, but is concentrated along thin beds of
powdered coal (sooty coal); adjacent to clastic partings; along
structural discontinuities; and near igneous intrusions and mafic
dykes (Godfrey 1994).
Fielding (1993) noted that iron estimates from the SIROLOG
probe were only generalising the iron occurrence and could not
differentiate between these different phases. However, the
technique should be used in conjunction with:
aerial mapping of zones of structural disturbance,
isopach mapping of clastic partings,
delineation of areas and seams affected by structuring, and
adjacent to thick clastic partings.
328
COAL PETROLOGY
In conjunction with Fieldings study, a detailed petrographic
study was undertaken by Esterle (1994) to determine whether
high iron occurrence could be related to coal facies distributions
within the seams. This study examined the vertical and lateral
variability in petrographic composition and iron content in the
context of the geological environment under which the deposit
formed. One of the objectives was to determine the relationship
between coal facies distribution and quality, including iron
content and SIROLOG traces.
Esterle (1994) found that, for Boundary Hill, detailed
examination of high peaks on the SIROLOG iron ratio traces
(FERAT) show that they can be correlated for up to 500 m.
Although high iron zones occurring vertically within seams are
often associated spatially with clastic partings (as exemplified by
peaks in the FERAT traces) there is no statistical relationship
between raw ash yield and high iron.
The vertical variability in iron content of coal seams within a
profile is not consistent, even across separate deposits within the
basin. However, correlations can be made between closely
spaced boreholes. As previously mentioned, the SIROLOG
FERAT traces in some boreholes show iron peaks occurring at
1 to 3 m spacings within seams, and between visible clastic
partings, similar to the spacings of petrographic cycle
boundaries.
Previous work by Esterle (1992) had established that, for
Callide Coal, vitrinite content can be reasonably estimated from
volatile matter (on a daf basis). It was hoped that SIROLOG
could be used to estimate macerals. Esterle (1994) conducted a
comparison of iron content with petrographic composition for
borehole L2916. As suggested by the comparisons of iron with
megascopic coal types and raw ash, there is no direct correlation
between iron content and maceral composition. In this borehole,
iron contents are highest in the top and bottom of the A2 to A41
sequence, subjacent to the roof and floor.
CONCLUSIONS
During 1993 research was undertaken on a number of topics
relating to examining the high and variable iron in ash content of
Boundary Hill. Although it was only one of a number of
techniques used, information gained from configuring the
SIROLOG probe to measure in situ raw ash, silica and iron
provided a major adjunct to existing laboratory analyses.
Examination of downhole traces complemented mapping of
highwall exposures, and indeed did highlight that the rates of
spatial variability in iron content are high. Other techniques were
used by researchers to identify two phases of emplacement of the
major iron mineral - siderite. One phase is thought to occur
syngenetically-epigenetically with peat accumulation and early
burial history, whereas the other is post-coalification and
associated with intrusive volcanics and groundwater movement.
This variability creates difficulties in predicting zones of high
iron in advance of mining.
Variations between iron predictions and as-dispatched quality
are thought to be due to blending operations and the difficulties
in accurately representing reality using grid-based modelling
algorithims. Detailed investigations of mineral and maceral
associations by Esterle (1994) suggest that a cyclicity in
petrographic variation within seam profiles occurs, and this can
be related to episodic flooding with water (both laden with, and
devoid of, sediment) during the peat accumulation of the seams.
These flooding events control the mineable geometry of the
deposits by means of simple seam splitting and increase in raw
ash contents, as shown in SIROLOG fence diagrams.
The scale of vertical variability in iron parallels the cyclicity in
maceral composition in the northern end of the mine, but not the
southern (Esterle 1994). Seam splitting and thinning is also more
common in the northern area. However, work by Fielding (1993)

suggests that much of the iron mineralisation is secondary and
distributed relative to the structural grain of the geology, rather
than the sedimentological features. The proportions of each
phase are as yet unknown. SIROLOG is a unsuitable tool in this
regard, being only capable of measuring total element (eg iron)
concentrations.
At the highwall scale, iron mineralisation decreases rapidly
(within a few hundred metres) away from dykes, faults and
joints. It is suggested that in addition to following vertical
features such as faults/joints and cleat, secondary iron
mineralisation in these areas of high iron have followed
horizontal discontinuities within seams. These discontinuities
were created by clastic partings, boundaries between
petrographic cycles, small scale erosive features and fracture
zones caused by compressive deformation (Esterle, 1994;
Godfrey, 1994).
The use of SIROLOG has contributed significantly to this iron
study in three areas. The system was firstly utilised as a source of
routine downhole geophysical data, providing detailed
information on seam thickness and correlation, and the presence
and thickness of clastic partings. Additionally it significantly
enhanced the deposits coal quality database by the ability of
on-site technical personnel to generate satisfactorily accurate raw
coal quality estimates of in situ ash, silica and total iron. Finally,
relevant data was transferred to sophisticated mine planning
software which enabled a series of fence diagrams and colour
shaded contour maps to be generated, so as to investigate the
vertical and lateral distribution of iron facies across the deposit.
Work is continuing in examining in more detail the correlation
of SIROLOG iron traces between drillholes and other coal
quality parameters such as macerals, mineral matter and ash
fusion temperatures
REFERENCES
Baafi, E, 1993. Basic Data Analysis of Iron Values Callide Coalfields Boundary Hill Mine, pp 1-18, unpublished report (Dept Civil and
Mining Engineering, University of Wollongong).
Biggs, M S, 1990. The Use of New Techniques in Computerised Data
Recording and Downhole Geophysical Logging using SIROLOG
During Drilling Programs at Callide Coalfields, in Proceedings of
the Bowen Basin Symposium 1990, pp 163-180, (Ed: J W Beeston),
Geol Soc Aust (Qld), Mackay, September 1990.
Biggs, M S, 1991. The Application of Neutron-Gamma Sirolog to
Estimate Iron Content in Coal and Implication for Estimating Ash
Fusion Characteristics, in Proceedings Queensland Coal Symposium,
Melbourne).
Biggs, M S, Burgess, A W and Patrick, R B, 1995. Callide Basin, in

Geology of Australain Coal Basins, (Eds: C R Ward, H J Harrington,
C W Mallet and J W Beeston), pp 471-488, Geological Society of
Australia Coal Geology Group Special Publication.
Borsaru, M, Charbucinski, J, Huppert, P, Youl, S F and Eisler, P L, 1988.
Coal Ash Determination in Dry Boreholes by the Neutron Capture
Technique, Nuclear Geophysics, 2(4):201-206.
Borsaru, M, Eisler, P L and Youl, S F, 1991. Neutron-Gamma Logging in
Coal Seams of Variable Iron Content, Nuclear Geophysics,
5(1/2):117-122.
Borsaru, M, Biggs, M S and Nichols, W J F, 1993a. Neutron-Gamma
Logging for Iron in Coal and Implications for Estimating the Ash
Fusion Characteristics at Callide Mine, Nuclear Geophysics,
7(4):539-545.
Borsaru, M, Millitz, P and Ceravolo, C, 1993b. Comparison Between the
Neutron-Gamma and Gamma- Gamma Techniques for Ash
Prediction in 140mm Diameter Quality Holes at the Callide Mine,
Nuclear Geophysics, 7(1):125-132.
Charbucinski, J, Youl S F, Eisler P L and Borsaru, M, 1986. Prompt
Neutron Gamma Logging for Coal Ash in Water Filled Boreholes,
Geophysics, 51(5):1110-1118.
Esterle, J S, 1992. Maceral Reflectance Variability and Recognition in
Callide Coals - Report IR 174, pp 1-30, CSIRO Division of
Geomechanics, unpublished report, (December 1992).
Esterle, J S, 1994. Petrographic Variation in Coal Seams at Boundary Hill
Mine, Callide Coalfields, and its relation to Iron Content - Report
5A, pp 1-45, CSIRO Division of Exploration and Mining,
unpublished report (Jan).
Fielding, C R, 1993. Origin and Distribution of Iron-Bearing Minerals in
Coal Seams at the Boundary Hill Mine, pp 1-12, Dept of Earth
Sciences, University of Queensland, unpublished report, (Nov).
Faraj, B S M, 1993. Investigations into Iron Mineralisation in Boundary
Hill Coal, pp 1-3, Dep of Earth Sciences, University of Queensland,
unpublished report, (25 May 1993).
Godfrey, N H, 1994. Brief report on Trial Highwall Mapping of Geology
at Boundary Hill Pit Box Cut, Callide Coal Mine, pp 118, Gary
Leblang and Associates, unpublished report to Callide Coalfields.
Patterson, J H and Marvig, P, 1993. Characterisation of Iron Containing
Minerals in Callide Basin Coals, Report CET/IR184R, pp 1-10,
CSIRO - Division of Coal and Energy Technology, unpublished
report, (December).
Ward, C R, 1990. Mineral Matter Analysis of Coal Samples from Callide,
Queensland - Report R685, pp 1-16, Unisearch, University of New
South Wales, unpublished report.
329
Microseismic Monitoring of Shear Zones and Related Seismic

Activity at Broken Hill
A J Morley1, J M Murray1 and G C Reed1
INTRODUCTION
The Broken Hill lead-zinc-silver orebody is situated in far
western New South Wales, Australia at latitude 3158S,
longitude 14128E, approximately 1200 km west of Sydney
(Figure 1).
The deposit lies within the Willyama Supergroup in the
Broken Hill Block; one of several inliers of Early to Middle
Proterozoic basement rocks in western New South Wales and
eastern South Australia (Willis et al, 1983). The orebody is
hosted by the Broken Hill Group and characterised by high-grade
(amphibolite to granulite facies) metamorphic mineral
assemblages overprinting the sedimentary protolith. Several
periods of deformation are evident, coincident with a complex
structural and metamorphic history.
The Broken Hill orebody has a strike length of 7 km, is
comprised of a stacked series of lead-rich and zinc-rich lenses
and, prior to mining, contained 300 million tonnes of ore grading
over 15 per cent combined lead and zinc (Haydon and
McConachy, 1987).
Pasmincos Southern Operation, an amalgamation of workings
accessed by the New Broken Hill Consolidated (NBHC), Zinc
Corporation (ZC) and Southern Cross shafts, is the only active
underground mine on the main Broken Hill orebody at present.
550000E
GAusIMM, Mine Geology Department, Pasminco Broken Hill Mine,

PO Box 460, Broken Hill NSW 2880.
540000E
1.
560000E
ABSTRACT
During more than 100 years continuous mining of the lead-zinc orebody
at Broken Hill, an immense amount of geological data has been collected
about shear zones related to the orebody. The challenge to current mine
geologists is to synthesise and interpret these data sets in a way that is
applicable to production demands. Precise knowledge of the locations of
major shear zones is important not only for the determination of ground
conditions but also to develop an understanding of ore remobilisation
along these structures.
Traditionally, geological information has been collected from drill core
and back and wall mapping underground, and interpreted on sections that
are approximately 20 m apart. The recent introduction of Mapteks
Vulcan software has enabled the shear zones to be modelled in three
dimensions.
The new three-dimensional shear models have been combined with
seismic data from the Southern Cross area. The seismic data is collected
from a microseismic monitoring system installed in the Pasminco
Southern Operations in August 1996 (Rauert and Tully, 1998). The
addition of seismic information to the shear zone models has led to a
greater degree of certainty about the shear zone locations and has
provided information about shear zones in areas yet to be drilled.
6470000N
Broken
Hill
Potosi
2K
Fitzpatrick
North Mine Shafts
Southern
Cross
Shaft
Southern Mining Operations

Broken Hill Orebody
6480000N
Mineralization
NBHC Service &

Haulage Shaft
6480000N
Operating Shafts
Non Operating Shafts
City of Broken Hill
560000E
550000E
540000E
Kilometres
FIG 1 - Location of the Pasminco Broken Hill Mine Southern Mine Leases.
331
A J MORLEY, J M MURRAY and G C REED
The dominant mining method used at Broken Hill is longhole

open stoping. Mining has reached a depth of 1.2 km from the
surface and a final depth of 1.4 km will be achieved before the
mine is closed.
Approximately 2.8 Mtpa are mined and the March 1999
reserve estimate stands at 17.9 Mt at 4.0 per cent lead, 7.5 per
cent zinc and 41 g/t silver (Lutherborrow, 1999).
The Southern Cross area of the mine is becoming increasingly
important to mine production as the famous argentiferous Lead
Lodes are depleted. Southern Cross is situated at the southern
extremities of the orebody (Figure 1) and encompasses the
southernmost ends of the zinc-rich A, B and C lodes, as well as
the shear-zone bounded South Eastern A Lode (SEAL).
PREVIOUS WORK
The geology of the Broken Hill orebody has been described by
numerous authors and will not be discussed in detail here. Recent
key reference papers include Haydon and McConachy (1987),
Mackenzie and Davies (1990) and Morland and Webster (1998).
Structural geology
The structural geology of the Broken Hill ore deposit and
surrounding area is complex, and remains in dispute. Archibald
(1978), Marjoribanks et al (1980) and Hobbs et al (1984) have
described the regional structural geology of the Broken Hill
Block. Numerous papers and theses have focussed on the
structure of the Broken Hill orebody itself, including the early
general reviews of Andrews (1922) and Kenny (1932). In a major
geological review of the Broken Hill region, Gustafson (1939)
described a zone of attenuation on the eastern side of the orebody
named the Main Shear.
The structural control on the orebody has been reviewed by
Hobbs et al (1968), Both and Rutland (1976) and more recently
by Webster (1993, 1994). Laing, Marjoribanks and Rutland
(1978) used graded bedding as facing indicator to propose that
the mine sequence occurred on the inverted limb of a regional
nappe structure. Other recent hypotheses include Findlay (1994)
who described the Broken Hill orebody as a large-scale
boudinage-hosted deposit and White (1994) who refuted the
presence of a regional nappe structure and claimed that later
phases of folding were the result of regional thrust shearing.
Mine geology
Mine geologists at Broken Hill have traditionally used the
information gained from drillcore and underground mapping to
predict the position of shear zones in the vicinity of the ore. The
shear zones are then interpreted on sections that are
approximately 20 m apart. A three-dimensional model is created
from these sections using Mapteks Vulcan software. In areas of
ample drilling and development, the position of these structures
is accurately constrained.
The depletion of the more accessible parts of the original
resource means that new drives and stopes are being developed in
areas with little existing information. Drilling is often
inadequately spaced to predict the presence of shear zones in the
footwall. This has caused problems with ground support in recent
access development in the Southern Cross area in particular. The
challenge is to use different techniques to predict the presence
and intensity of these shear zones, and to divert planned
development away from the most sheared areas.
Microseismic monitoring system

A microseismic monitoring system devised by Integrated
Seismic System International (ISS) was installed at the Pasminco
332
Southern Operations in August 1996 (Rauert and Tully, 1998).

The system was later upgraded and expanded to include the
NBHC/ZC areas (Tully, 1997). Muliadi (1997) used the
monitoring system to model principal stress direction regimes
and failure criteria in the Southern A Lode. Microseismic data
from the system have also been analysed by ISS personnel to
interpret and explain events of unusually large magnitude
(Siggins, 1997).
MICROSEISMIC SYSTEM DESCRIPTION

The Southern Cross microseismic monitoring system comprises
six triaxial geophone sites scattered at an approximately regular
spacing (350 m vertically, 500 m along strike, 400 m across
strike) throughout the Southern Cross mine area. Each site
consists of a geophone; power supply; and processing
seismometer (Figure 2), which digitises and saves the seismic
information, then processes the waveforms. This information is
periodically sent to a network multiplexer, which combines the
signals from all geophones and transmits the information to a
fibre optic line driver.
The signal is broadcast via optic fibre to the surface where it is
decoded by a fibre optic concentrator. From there it is transmitted
to a Silicon Graphics Indy Unix computer workstation, to be
processed by the Rock Mechanics Engineer. The seismic data
can then be exported to Vulcan software for geological modelling
(Rauert and Tully, 1998).
The Southern Cross system was expanded and updated to
include the NBHC/ZC area in 1999. An additional eight
monitoring sites were installed (Tully, 1997). Geophone receivers
were replaced by accelerometers in the expanded system to limit
the amount of resonance in the signal. The array has been tested
using an explosive charge at a known position. In the Southern
Cross area the position of each recorded event is within 15 m
accuracy of the seismic source (Rauert and Tully, 1998).
INFORMATION/RAW DATA
Introduction
Microseisms are defined as small motions of the earth that are
relatively short-lived (1.0 - 9.0 s). They are unrelated to
earthquakes and are usually induced by stress changes such as
mining (Bates and Jackson, 1987). More than 5700 microseismic
events have been recorded since the microseismic monitoring
system was installed (Tully, 1997).
Raw data
Each microseismic event generates an analogue signal which is
converted to a digital waveform at the surface by the monitoring
system. Each geophone or accelerometer site records
P (compressional) and S (shear) waveforms in three dimensions.
The microseismic monitoring system automatically interprets the
best fit to determine the relative timing of P and S events. This
interpretation is then validated manually by the rock mechanics
engineer. The system uses the timing between P and S events to
calculate a sphere about each point at which the microseism may
have occurred, known as the location sphere. The position of the
seismic event is constrained by determining the point at which
location spheres from at least three different recording stations
meet. A confidence level and location pick error are calculated
by the system to determine the accuracy of the data. An
acceptable location pick error is ten per cent of the distance
between microseismic event and geophone/accelerometer sites.
The magnitude of the microseismic event is calculated by the
monitoring system after the event location is accurately
determined.
MICROSEISMIC MONITORING OF SHEAR ZONES AND RELATED SEISMIC ACTIVITY
FOM -Southern Cross

NBHC TELEPHONE
EXCHANGE
SOUTHERN CROSS
SHAFT
NBHC OFFICES
ROCK MECHANICS
OFFICE (SGI)
HARDCOPY
(PRINTERS)
MINE DESIGN
WORKSTATIONS
(VULCAN)
M.C.C.
PASMINCO
BROKEN HILL
MINE
MUX
SOUTHERN CROSS
TELEPHONE
EXCHANGE
LAN
FOM -13L NBHC
LEGEND
ACCEL
16 LEVEL
PS
GEO
PS
16 LEVEL
MS
CABLE
VIA SHAFT
GEO
17 LEVEL
FO
FOM
17 LEVEL
FO
PS
19 LEVEL SHIFT
BOSSES CUDDY
ACCELEROMETER TRANSDUCER
MULTIPLE CH. FIBRE OPTIC

MULTIPLEXER
FIBRE OPTIC CABLE
MS
LAN
EXISTING LOCAL AREA NETWORK
LA
RS 232 LINE AMP

19- 20 LEVEL
RAMP
13 LEVEL
FUTURE SITE
PS
13 LEVEL
PLAT
ACCEL
CABLE
VIA SHAFT
MS
M.S.
GEO
GEO
21- 22 L EXPO DRIVE

FUTURE SITES
15 LEVEL
- SOUTH
15 LEVEL
15 LEVEL
- NORTH
18 LEVEL
RADIO
CUDDY
24V DC
ACCEL
FOM(8CH)
I.S.S. 16 CH.MULTIPLEXER
ACCEL
20-21LRAMP
ACCEL
MUX
PS
10 LEVEL
- SOUTH
10 LEVEL
C.CUDDY
I.S.S.MULTIPLE SEISMOMETER
19 LEVEL
21 LEVEL
10 LEVEL
-RAMP AREA
MS
I.S.S. PROCESSING
SEISMOMETER
PS
2 PAIR TX/RX
CABLE
VIA RAMPS
HAULAGE
SHAFT
GEOPHONE TRANSDUCER
19 L M/BAY
GEO
SERVICE
SHAFT
PS
ACCEL
18 LEVEL
- W.A.Lode
ACCEL
21 L. ELECT.CUDDY
FIG 2 - Microseismic monitoring system network configuration (after Rauert and Tully, 1998).
Raw data in a practical form

Each microseismic event is assigned a location and magnitude.
These are displayed graphically as co-ordinates in
three-dimensional space, and coloured according to a legend
based on the Richter Scale magnitude of the occurrence. The
data is then exported to Vulcan and assigned to a specific Vulcan
layer based on the year of occurrence. These can be viewed
separately by mine geologists to determine broad differences in
seismic activity over time.
Events that are not recorded by three or more sites are deleted
before they can be processed. This filters any signals which are
due to localised events, such as passing trucks. The distinctive
waveforms resulting from stope and development firing are also
excluded from the data set.
A 24-hour period may typically include 30 to 40 events
measuring less than one on the Richter scale. These events often
increase in magnitude and frequency when large stopes are fired.
Microseismic activity may intensify for several weeks following
a mass blast.
INTERPRETATION OF MICROSEISMIC DATA

AND COMPARISON WITH TRADITIONAL
GEOLOGICAL INTERPRETATIONS
The importance of accurate shear zone interpretations
Mine planning and production staff have always looked to
geologists for practical information about the orebody. Emphasis
has recently been placed on shear zone interpretations by mine
geologists. This is due to a number of significant stope failures

and development problems, which are generally concentrated in
the Southern Cross area of the mine. The mine is increasingly
dependant on more marginal ore areas, so the need for accurate
stoping is growing. It is critical to halt occurrences of stope
dilution from hanging wall failure to maintain higher metal
grades.
Interpretation of raw data

The microseismic data points do not always occur exactly within
the shear zones. The centres of events are often focussed instead
in the more competent rocks along the edges of shear zones
(Figure 3). It is here that sufficient stress is built up to cause a
seismic event. However, in areas where two major shear zones
intersect, there is a concentration of seismic points of all
magnitudes (Figure 4). Such zones produce the worst ground
conditions and occur consistently throughout the Southern Cross
area.
Comparison of microseismic data with geological

interpretations
The major shear zones in the mine are interpreted from drillcore
and underground mapping. These interpretations are then
digitised and modelled in three dimensions using the Vulcan
software.
Existing Vulcan three-dimensional shear zone models derived
from drillcore analysis and underground mapping are compared
with predicted shear zone locations from the microseismic
monitoring system. In areas of extensive drilling and geological
333
FIG 3 - Location of microseismic events along edges of interpreted shear zone position (geological cross-section 119S).
9500RL
Legend:
Point magnitude
on Richter Scale:
9400RL
-5 to -1
-1 to 0
0 to 1
1 to 2
Interpreted shear
zone
9300RL
9200RL
300mW
400mW
500mW
600mW
FIG 4 - Cluster of microseismic points at the intersection of two shear zones (geological cross-section 107S).
334
MICROSEISMIC MONITORING OF SHEAR ZONES AND RELATED SEISMIC ACTIVITY
mapping, this comparison shows that there is a good correlation

between the microseismic data points and the interpreted shear
zones. This leads to a greater degree of certainty in the final
interpretation of the shear zone positions in more sparsely drilled
areas from the monitoring system. Microseismic data also
provides preliminary information about shear zones in areas that
are yet to be drilled.
DISCUSSION
Introduction
The microseismic system was originally installed in the mine to
create a database of seismic events to be used by the geotechnical
staff. It was envisaged that this database would be most useful in
determining both background and mining-related seismicity.
Improved safety
The system is used to regularly monitor ground conditions and
rock strain and to determine safety conditions for the
underground workers. Regions of high-risk are identified and
access is restricted to these areas until microseismic activity has
subsided. Seismic activity is discussed at production meetings
and the data is made available to the Mine Control Centre which
is in continuous contact with all underground employees.
Practical application of the microseismic system to

geological interpretation
The application of the system to mine geology has only recently
been recognised. As new development is pushed into areas that
contain sparse drill hole information, more unorthodox forms of
data have become useful tools in the interpretation of the shear

zone positions.
As a stand-alone method, the microseismic system is not
entirely reliable in predicting the precise position of shear zones.
However, in combination with the information taken from
drillcore and wall mapping, more accurate results can be
achieved. In cases where the approximate position of the shear
zones is known, it is possible to extrapolate shear zone
extensions with a greater degree of confidence using the seismic
event locations. Recent experience has shown that areas of
consistently large clusters of microseismic data points on
sections throughout the mine represent very poor ground
conditions that should be avoided in mine designs if possible.
Interpretation of the microseismic data points has led to the
discovery of large-scale structures not previously recognised by
geologists through drillcore interpretation and mapping. These
include a prominent flat-lying feature known informally by mine
geologists as the Flat Fault (Figure 5). The exact nature of this
structure is as yet unknown, but it is implicated in the 1998
failure of the SEAL 104-107 stope, as well as instability in the
SAL4 and SAL5 stopes.
Exploration potential
The microseismic system also has potential as an exploration
tool within the near-mine area. Ore is commonly remobilised
along large-scale shear zones, such as the Main, Central and
Dropper Shears. Determining the existence and location of
similar shear zones in relation to the Broken Hill orebody may
lead to further ore discoveries. Application of microseismic
technology will play an important role in determining future
exploration targets.
9600RL
Legend:
Point magnitude
on Richter Scale:
9500RL
-5 to -1
-1 to 0
0 to 1
1 to 2
Interpreted shear
zone
9400RL
Interpreted Flat
Fault
9300RL
200mW
300mW
400mW
500mW
600mW
FIG 5 - Cluster of microseismic points that indicate the presence of a flat structure, known as the Flat Fault (geological cross-section 103S).
335
Limitations
Data collected from the microseismic monitoring system is
subject to several limitations. Major seismic events are often
initiated by nearby mining activity. Some areas that appear to be
free from microseismic points in pre-mining data become quite
seismically active once mining has commenced. This becomes
problematic when mining development intersects shear zones
which were not seismically active at the time of mine design and
planning. The ground conditions in these areas can be worse than
predicted. In addition, the monitoring system at Broken Hill
operates in a very demanding environment. As a result the
reliability of the system can be poor as there are often problems
with power supply fluctuations. If less than three geophones
record an event, an accurate position of the epicentre can not be
determined.
Geological features such as major faults or unusual rock mass
characteristics can alter the microseismic signal received by the
geophones or accelerometer. This can usually be compensated
for during the initial calibration of the system.
Another intrinsic limitation is that the frequency of
microseismic events is often higher than that of the geophones
resulting in resonance in the signal (Rauert and Tully, 1998).
This problem has been overcome by using accelerometers
instead of geophones in the NBHC/ZC area of the mine, however
geophones are still in use in the Southern Cross area of the mine.
CONCLUSIONS
Accuracy checks on shear zone interpretations
Addition of the microseismic system to geological interpretations
is still at an early stage. However it is already invaluable in the
refinement of existing shear zone models, and has been used
extensively to validate shear zone interpretations primarily based
on drillcore and mapping. The system has also proved very
useful in the extrapolation of shear zone interpretation into
poorly drilled areas.
Increased safety factors

The microseismic monitoring system has substantially increased
levels of safety involved in mining of the Broken Hill orebody. It
provides direct feedback on recent seismic events, and when
correlated with shear zone interpretations, the microseismic data
enable mine geologists to predict which areas will have the worst
ground conditions. Development and stoping can be designed to
avoid dangerous areas, so that underground mine staff are less
exposed to working in sheared and unstable rocks.
ACKNOWLEDGEMENTS
The authors sincerely thank Neil Rauert and Kim Tully for their
advice and for permission to reproduce figures from their work.
Many thanks also to Helen Degeling, Giorgio Dallarmi and J J
Reed for their helpful comments. The contributions of all the
mine geologists that have worked underground at Broken Hill
throughout the life of the Line of Lode are gratefully
acknowledged. This paper is published with the permission of
Pasminco Broken Hill Mine.
REFERENCES
Andrews, E C, 1922. The geology of the Broken Hill District: Memoirs
of the Geological Survey of NSW, Geology, No 8.
Archibald, N J, 1978. Report on the 1977-78 phase of the Broken Hill
Lode Project. CRAE internal report (unpublished).
336
Bates, R L and Jackson, J A, 1987. Glossary of Geology (Third Edition).

p 422 (American Geological Institute: Alexandra, Virginia).
Both, R A and Rutland, R W R, 1976. The problem of identifying and
interpreting stratiform ore bodies in highly metamorphosed terrains:
The Broken Hill example, in Handbook of Stratabound and
Stratiform Ore Deposits (Ed: KH Wolf), Vol 4, pp 260-325.
Findlay, D, 1994. Geological Note. Boudinage, a reinterpretation of the
structural control on the mineralisation at Broken Hill, Australian
Journal of Earth Sciences, 41:387-390.
Gustafson, J K, 1939. Geological investigation in Broken Hill, Final
Report Central Geological Survey, 1939 (unpublished).
Haydon, R C and McConachy, G W, 1987. The stratigraphic setting of
Pb-Zn-Ag mineralisation at Broken Hill, Economic Geology,
84:1235-1261.
Hobbs, B E, Ranson, D M, Vernon, R H and Williams, P F, 1968. The
Broken Hill orebody, Australia. A review of recent work,
Mineralium Deposita, 3:293-316.
Hobbs, B E, Archibald, N J, Etheridge, M E and Wall, V J, 1984.
Tectonic history of the Broken Hill Block, Australia, in Precambrian
Tectonics Illustrated (Eds: A Krner and R Geiling) pp 353-368
(E Schweiz Verlags: Stuttgart).
Kenny, E J, 1932. The Broken Hill lode - its geological structure, Proc
Australas Inst Min Metall, 87:217-245.
Laing, W P, Marjoribanks, R W and Rutland, R W R, 1978. Structure of
the Broken Hill Mine area and its significance for the genesis of the
orebodies, Economic Geology, 73:1112-1136.
Lutherborrow, C H, 1999. Pasminco Broken Hill Mine resource and
reserve estimate at 31 March 1999. Pasminco internal company
report (unpublished).
Marjoribanks, R W, Rutland, R W R, Glen, R A and Laing, W P, 1980.
The structure and tectonic evolution of the Broken Hill Region,
Australia, Precambrian Research, 13:209-240.
Mackenzie, D H and Davies, R H, 1990. Broken Hill lead-silver-zinc
deposit at ZC Mines, in Geology of the Mineral Deposits of Australia
and Papua New Guinea (Ed: F E Hughes) pp 1079-1084. (The
Morland, R and Webster, A E, 1998. Broken Hill Lead-Silver-Zinc
Deposit, in Geology of Australian and Papua New Guinean Mineral
Deposits pp 619-626 (Eds: D Berkman and D Mackenzie) (The
Muliadi, 1997. The identification and behaviour of highly stressed rock
in the Broken Hill Mines. Master of Engineering (unpublished) in
Geological Engineering Thesis, Department of Civil and Geological
Engineering, Faculty of Engineering, Royal Melbourne Institute of
Technology.
Rauert, N S and Tully, K P, 1998. Integration of a Microseismic
Monitoring System in Mining the Pasminco Broken Hill Southern
Cross Area, in Proceedings Seventh Underground Operators
Conference, pp 115 - 120 (The Australasian Institute of Mining and
Siggins, A F, 1997. Analysis of precursors to the seismic event of July
26, 1997. Confidential report to Pasminco Broken Hill. ISS Pacific
Pty Ltd, Melbourne (unpublished).
Tully, K P, 1997. Pasminco Broken Hill Mine application for capital
expenditure. Pasminco internal company memo (unpublished).
Webster, A E, 1993. A structural interpretation of the Broken Hill
orebody as suggested by the internal features and macroscopic
geometry of the mineralisation, Geol Soc Aust Abstracts No 37, 12th
Geological Convention, Perth, September 1994.
Webster, A E, 1994. The Structure and Stratification of Lead Lode,
Southern Operations, Broken Hill, NSW, Australia. MSc Thesis
(unpublished), James Cook University of North Queensland.
Willis, I L, Brown, R E, Stroud, W J and Stevens, B P J, 1983. The Early
Proterozoic Willyama Supergroup: stratigraphic subdivision and
interpretation of high to low-grade metamorphic rocks in the Broken
Hill Block, New South Wales, Journal of the Geological Society of
Australia, 30:195-224.
White, S H, 1994. Structural study of the shear zones in the Broken Hill
area. Pasminco Exploration Broken Hill internal report, PX1028
(unpublished).
Finding More Ore, Further From the Drill Hole, With DHMMR
J Bishop1 and R Lewis2
ABSTRACT
Drill hole magnetometric resistivity (DHMMR) is a relatively recent
addition to the explorationists suite of techniques for finding further
resources in the vicinity of a drill hole. The method relies on gathering of
an applied current into zones which are less resistive than the country
rocks and it was first applied to exploration programs looking for weakly
conductive sulphides, such as, for example, carbonate-hosted zinc
deposits. More recently, it has been applied to a wider range of deposit
styles, including highly conductive nickel sulphides. It has been found
that the technique is capable of not only detecting mineralisation not seen
by conventional electromagnetic (EM) techniques, but it can detect at a
greater distance. Also, whilst EM is superior at defining tabular-shaped
conductors, DHMMR may be more effective at finding cigar shaped
sulphides. This paper presents a number of examples illustrating these
findings.
INTRODUCTION
Down-hole geophysical logging techniques are now a
well-established part of many in-mine and near-mine exploration
and development programs. A relatively new addition to the suite
of available methods is drill hole magnetometric resistivity
(DHMMR)
This technique was originally developed to detect poorly
conducting targets such as occur in many zinc deposits (eg
Bishop et al, 1997) since the method merely requires a target
with a conductivity contrast rather than one with an absolute high
conductivity. However, it has since been successfully applied to
exploration programs where moderate to good to excellent
conductors were expected. Apart from a wider range of
conductivities, DHMMR will also detect rod and ribbon shaped
bodies at significant distances from the hole, which may only
weakly, if at all, respond to other electrical techniques.
The DHMMR method has been successfully tried at a number
of mines encompassing a variety of commodities. This paper
presents data from surveys at gold, silver-lead-zinc, and nickel
mines. These examples demonstrate the methods ability to
detect mineralisation insufficiently conductive or the wrong
style for drill hole electromagnetics (DHEM); to detect
sulphides at a significant distance from the hole and to place
them unambiguously on one specific side of the hole.
THE DHMMR METHOD

The target zone is energised with a low frequency square wave
impressed into the earth via a grounded dipole which is laid out
in a U shape with the holes to be surveyed within the U. Dipole
lengths are typically 500 - 2000 m along strike and several holes
can be surveyed from the one dipole (Figure 1a). The electrodes
may be on the surface and/or downhole. A standard IP
transmitter is used to produce the required current, which ideally
will be several amps.
The earth return current seeks the path of least resistance
between the two dipole electrodes and thus any relatively
conductive zone such as a disseminated sulphide deposit will be
preferentially energised. The increased current density in the
target has an associated magnetic field (B) and this is monitored
1.
MAusIMM, Applied Geophysical Research Pty Ltd, PO Box 974,

Sandy Bay Tas 7006.
2.
Applied Geophysical Research Pty Ltd, PO Box 974, Sandy Bay

Tas 7006.
FIG 1 - (a) DHMMR layout. Current in the ground is defined to

flow from south to north and upwards is defined as positive.
(b) For east-facing drill holes, a body above a hole will produce a
positive response and a negative response if below the hole.
The opposite applies for west-facing holes.
with a standard DHEM probe which records a voltage

proportional to the time differential of the magnetic field (ie
dB/dt). The recorded dB/dt voltage (usually in microvolts) is
integrated to the magnetic field B (in picoTeslas per amp) and
corrected for the effect of the magnetic field due to the current
flowing in the wire and for the fields caused by the electrodes
and the background currents in the target host rocks.
Polarity definitions are particularly important in DHMMR. If
the current in the ground is defined as flowing from south to
north, then a conductor beneath a drill hole with an easterly
azimuth will produce a negative response and a positive response
if above the hole. The opposite will apply for west-facing holes
(Figure 1b).
The phase difference between the transmitted current and the
recorded voltage is also recorded (in milliradians). This is an
induced polarisation (IP) parameter which can be used
qualitatively in an interpretation, but it has not yet been
incorporated into the modelling.
Example A: Weakly conducting lead-zinc

mineralisation
The Potosi open cut is located about 2 km north of the Broken
Hill orebody on the Northern Leases. It had a total resource
approaching 1 Mt at 9.1 per cent Zn, 2.3 per cent Pb and 28 g/t
Ag with little other sulphide. The mineralisation, mostly
sphalerite and galena with minor chalcopyrite and pyrrhotite,
occurs largely within the Freyers Metasediments (Unit 4.5),
337
J BISHOP and R LEWIS
stratigraphically below the Hores Gneiss (Unit 4.7) which hosts

the main Broken Hill orebody (Morland and Leevers, 1998).
Although Unit 4.7 mineralisation is generally quite conductive,
Unit 4.5 is much less so. Near-mine exploration has focussed on
a north plunging zone of Potosi mineralisation which lies close to
the contact with the overlying Unit 4.6 amphibolite.
Figure 2a shows the profile from a DHEM survey carried out
down DDH 3153 which is located approximately 700 m to the
grid north of the open cut. There are no features of interest,
including from that section where the hole lies within the target
Freyers Metasediment horizon. In contrast, the DHMMR survey
(Figure 2b) shows a broad, well-defined low indicating (for this
geometry) a conductor, or conductors, below the hole. Figure 2c
shows the modelled DHMMR conductors in good agreement
with the intersected sulphides. Following this result, DHMMR
became the preferred downhole exploration technique on the
Northern Leases.
Example B: Detection at a distance

This example also comes from the Potosi mine area. Section 21S,
lies along strike, ~3300 m, to the north of example A and has a
similar geological cross-section (Figure 3a). DDH 3230 was
surveyed with DHMMR and some low amplitude but
well-defined responses pointed to a number of below-hole
conductors. The interpreted conductors from the DHMMR data
have been superimposed onto the geological cross-section
(Figure 3a).
A follow up hole, 3458, was targeted on the deepest
interpreted conductor, which is about 145 m from 3230. DDH
3458 intersected two zones of patchy mineralisation: one
between 347 m and 352 m, and another between 363 m and 371
m. The deeper zone coincides with the target DHMMR
conductor. DDH 3458 was then itself surveyed with DHMMR.
The positive response in the east-facing 3458 (Figure 3b)
indicates that the bulk of the mineralisation lies above the hole
and the modelling defined two adjacent zones of mineralisation
extending away from, and above, 3458 (Figure 3c).
DDH 3466 was drilled to test for the conductors interpreted to
lie above 3458. This hole intersected 3.6 m at 12.3 per cent Pb,
12.3 per cent Zn and 4.5 per cent Fe from 319 m, plus 6.0 m at
4.0 per cent Pb, 7.3 per cent Zn and 4.5 per cent Fe from 341 m.
Thus the initial DHMMR survey detected economic grade
lead-zinc mineralisation at a distance of more than 100 m from
the drill hole.
Example C: Sulphide-hosted gold

DHMMR has been used to explore for extensions to shear-hosted
mineralisation at the Wiluna Gold Mine in the Yilgarn Craton,
WA. Most of the gold is refractory within disseminated sulphides
averaging five per cent pyrite with some arsenopyrite. The ore
conductivity is low but at least ten times greater than the host
rock. Thus DHMMR was chosen as the only method likely to be
able to find weakly conductive ore at depths of 1 km or more,
below a conductive overburden, and at significant distances from
the drill holes (Meyers et al, 2000).
The electrodes for the energising dipole were located in two
drill holes which intersected the host shear at a downhole depth
of 925 m for the southern electrode and 870 m for the northern
electrode. Drill hole WD494 was surveyed over the interval
200 m to 1520 m using a three component probe. The results
FIG 2 - Potosi mine, section 31.75S. (a) DHEM results from DDH 3153 showing no response from the underlying sulphides intersected in 3147 and 3284.
(b) DHMMR profile from 3153X showing a clear response from the underlying mineralisation. (c) Geological cross-section of section line 31.75S of the
Broken Hill Northern Leases with the interpreted conductors from the DHMMR survey of 3153X. DDH 3147 intersected ~9 m @ 0.9 per cent Pb + 10.2
per cent Zn and follow up hole 3284 intersected ~1 m @ 2.0 per cent Pb + 4.8 per cent Zn (after Bishop et al, 1997).
338
FINDING MORE ORE, FURTHER FROM THE DRILL HOLE, WITH DHMMR
FIG 3A - Potosi mine, section 21S. Interpreted geological cross-section for

section line 21S of the Broken Hill Northern Leases when only holes
3228 and 3230 had been drilled, with modelled conductors from the
DHMMR survey of these two holes. The deepest conductor, at
~9920RL, was targeted for follow up drilling with 3458.
FIG 3C - Potosi mine, section 21S. Revised geological cross-section

showing 3230, 3458 and 3466 with superimposed model conductors
from the DHMMR surveys of 3230 and 3458. (The responses from
these model conductors are compared with the field data in Figure 3b.)
DDH 3466 was targeted on the off-hole conductors above 3458 and it
intersected 3.6 m at 12.3 per cent Pb, 12.3 per cent Zn and
4.5 per cent Fe from 319 m, plus 6.0 m at 4.0 per cent Pb,
7.3 per cent Zn and per cent Fe from 341 m.
required to produce the broad low in the axial profile (#2 in

Figure 4b). This body, which varies in distance from 120 m to
220 m from the hole, was interpreted to be near-vertical with a
southerly plunge. Superimposed on the geological cross-section,
it lies within the shear zone and part of this conductor has been
already tested by DDH WD424, which intersected approximately
20 m (true width) averaging nearly 1.g/t Au (Figure 4c).
Thus the DHMMR has identified a previously defined zone of
mineralisation and confirmed the southerly plunge, at distances
of 100 m to 200 m from the drill hole.
Example D: Shape dependent responses
FIG 3B - Potosi mine, section 21S. Field and model DHMMR B field data
for 3230 and 3458. The results from 3458 suggest that the bulk of the
mineralisation lies above this hole.
show no obvious response from the upper gold horizon, but there
are low amplitude, high frequency responses in all three
components associated with the lower horizon which averages
around two per cent total sulphide (Figure 4a). Of more interest
however, are the larger and broader responses which suggest the
presence of a significant conductor (mineralisation?) located
off-hole.
The axial data was modelled using a 2D program (Lewis,
1998) and the cross-component data was qualitatively
interpreted. The modelling produced a number of small bodies,
mostly close to the drill hole (Figure 4b). One exception was an
above-hole body at a down hole depth of ~1000 m which was
A nickel sulphide prospect on the Perseverance Fault in the

Agnew-Wiluna greenstone belt, WA, has been drilled in some
detail and DHEM was successfully used to help define the extent
of the massive sulphides. As well as the massive sulphides, some
stringer, or matrix, sulphides were known to occur above the
massive sulphides, close to a shallowly dipping thrust zone
which had emplaced felsic volcanics over the ultramafics. Figure
5a presents the axial profile DHEM data for DDH 509 and
Figure 5b shows the good agreement between the interpreted
models and the geological cross-section.
A DHMMR survey was trialled to see if it could add to the
geological picture. A survey down DDH 509 gave little response
from the underlying massive sulphides, but a broad positive
response centred at ~135 m down the hole, suggested the
presence of a significant conductor above the hole (Figure 5c).
This was tested by DDH 552 which intersected 14 m of 2.9 per
cent Ni. This mineralisation can be seen (in retrospect) as the
source of the small, rapidly decaying response at ~140 m in the
DHEM profile in Figure 5a. This response only occurred in loops
offset from the mineralisation (Bishop et al, 2000).
339
FIG 4A - Wiluna gold mine. Gold assays, sulphide percentages, plus DHMMR profiles for WD494 (note the different scales for the three components).
The upper gold horizon has no obvious associated DHMMR response, but the lower horizon shows a low amplitude, high frequency ripple. The main
MMR response, at ~1050 m, is produced by an off-hole conductor.
FIG 4B - Wiluna gold mine. 2D Modelling of the axial data from WD494.
Numerous small bodies, mostly close to the hole were required to match
the observed data. One important exception is body no. 2, which has a
depth extent of ~200m and lies between 120 m and 220 m from the hole.
340
FIG 4C - Wiluna gold mine. Body no 2 from Figure 4b superimposed

onto the geological cross-section. The body lies within the host shear and
has already been tested by DDH WD424, which intersected 42.5 m with
an average grade of 0.83 g/t gold.
FIG 5 - WA nickel prospect. (a) Model and observed DHEM data for DDH509. (b) DHEM conductors superimposed on the geological cross-section.
(c) DHMMR data for DDH509 and the follow up hole 552. The positive response at ~135 m in the 509 profile indicates a significant above-hole
conductor. This was tested by 552 which intersected 14 m of 2.9 per cent Ni. DHMMR surveying of 552 showed sharp, near-symmetrical (either side
of zero) responses which suggest that 552 has intersected close to the centre of the source. (d) Revised geological cross-section for DDHs 509 and 552
showing the interpreted DHEM and DHMMR conductors.
341
Figure 5d shows the revised geological cross-section with the

DHEM and DHMMR interpreted conductors superimposed.
Thus the DHEM has responded well to the sheet-like massive
sulphides but not to the overlying cigar-shaped stringer
sulphides. On the other hand, the DHMMR has responded to the
stringer sulphides, but not to the massive sulphides.
These two sets of data show that EM is applicable for finding
conductors of large areal extent, but that for bodies with large
cross-sectional area relative to their long section DHMMR is
probably a more appropriate method. Thus both techniques were
required here to detect the two different styles of mineralisation.
Example E: Unambiguous location of mineralisation

Massive Ni-Cu sulphides at the Nkomati mine in Mpumalanga,
South Africa occur within the trough-shaped Uitkomst
ultramafic/mafic complex. These are overlain by a number of
other, more disseminated, sulphide layers, some of which contain
significant amounts of pyrrhotite (Woolfe, 1996). Thus the target
sulphides are shielded by zones of conductive and magnetic
mineralisation. Figure 6a shows a plan view of the massive
sulphide body within the complex and the location of four drill
holes used for the downhole geophysical testwork. Figure 6b
presents a simplified cross-section of the complex with the
massive sulphides at the base of the complex and the same four
vertical drill holes.
A large-scale drilling program was undertaken to look for
further occurrences of massive sulphide and various downhole
geophysical techniques were employed to help find further ore,
which was expected to occur at around the same reduced level
(RL) as the defined resources. One problem, for all techniques, is
the presence of resolving the deeper massive sulphides from the
overlying disseminated sulphides which are generally thicker and
still quite conductive and magnetic. All holes were vertical and
this meant that for the DHEM surveys, both frequency and time
domain, only axial data could be usefully recorded and thus the
interpretations had an axial symmetry; ie conductors could be
interpreted with respect to depth and dip, but not azimuth (Nyoni
and Bishop, 2000).
FIG 6B - Nkomati nickel mine. Simplified cross-section showing the

massive sulphide orebody at the base of the complex, together with the
four vertical drill holes used for the downhole geophysics.
FIG 6A - Nkomati nickel mine. Plan view showing the location of the massive sulphide body within the mafic/ultramafic complex and
the fence of four holes used for the downhole geophysics.
342
FIG 6CI - Observed and model B field data from the along strike dipole, E3. The broad positive response in Sh74 and the sharp negative one in Sh68,
at ~960 m RL, are due to the same conductor, which is located between the two holes at the massive sulphide level. The slightly shallower negative
response in Sh160 plus the positive in Sh157 are due to separate sources in the overlying sulphide horizon.
343
FIG 6CII - The deep negative-going responses from both holes are at massive sulphide level and are due to conductors located to the north-west of the
drill holes.
DHMMR surveys were conducted with a dipole located both

along the complex (E3 in Figure 6a) and across the complex (N2
in Figure 6a). It was hoped the latter would better distinguish the
(presumably) more conductive, short strike length massive
sulphides from the overlying long strike length disseminated
sulphides. The results were certainly different (Figure 6c) and
complementary rather than comparative.
The interpretations of both sets of data were unambiguous and
contrary to expectation, with both showing good conductors, at
the target depth, on the opposite side of the drill holes to the
known massive sulphides (Figure 6d). Thus the DHMMR
suggests that there are better zones of more conductive sulphides
lying away from the known ore, on the northwestern side of the
344
drill holes. These are yet to be confirmed by drilling which will

take place from underground.
CONCLUSIONS
DHMMR is another tool for finding ore off-hole. It is capable of
detecting sulphides too resistive for DHEM and of seeing further
from the hole than DHEM (and much further than IP). It will not
define the extent of thin tabular bodies, but rather will locate the
most conductive zones of such bodies. Thus the method may
replace DHEM where the sulphides are only weakly conductive
or act as a complementary technique where both tabular bodies
and elongate lenses occur.
FIG 6D - Nkomati nickel mine. A plan view of the massive sulphide body with the interpreted massive sulphide level DHMMR conductors. The
DHMMR has not defined the extent of the known mineralisation (which was confirmed by DHEM), but it does suggest the presence of further
mineralisation to the west and north of the presently defined massive sulphide boundary.
ACKNOWLEDGEMENTS
Data presented in this paper was published with the kind
permission of Anglovaal Minerals, Pasminco Mining Broken
Hill, WMC Resources and Wiluna Gold Mines. Pongratz
Publications produced the figures.
REFERENCES
Bishop, J R, Hatch, M, Asten, M, Carroll, N and MacInnes, S, 1997.
Finding sphalerite at Broken Hill with drill hole magnetometric
resistivity. Exploration Geophysics, 28:6-10.
Bishop, J R, Lewis, R J G and Stolz, E, 2000. Horses for (conductive)
courses: DHEM and DHMMR. Exploration Geophysics,
31:192-199.
Lewis, R J G, 1998. The art of DHMMR. AGR Pty Ltd publication.

Meyers, J, Cooper, M, Bishop, J R and Hatch, M, 2000. Downhole
magnetometric resistivity surveying for refractory gold ore at Wiluna
gold mine, Western Australia. Presented at the ASEG conference,
Perth, WA.
Morland, R and Leevers, P R, 1998. Potosi zinc-lead-silver deposit,
Broken Hill in Geology of Australian and Papua New Guinean
Mineral Deposits (Eds: D A Berkman and D H Mackenzie), pp
615-618, (The Australasian Institute of Mining and Metallurgy:
Melbourne).
Nyoni, M and Bishop, J R, 2000. Geophysical Surveys at the Nkomati
Mine, Mpumalanga, South Africa. Presented at the ASEG
conference, Perth, WA (Exploration Geophysics, in press).
Woolfe, J A S, 1996, The Nkomati Joint Venture a nickel mine in the
making, Geological Society of South Africa Geobulletin, 39:5-7.
345
Nuclear Borehole Logging Techniques Developed by CSIRO

Exploration and Mining for the Metalliferous Mining Industry
M Borsaru1 and J Charbucinski1
ABSTRACT
Commonwealth Scientific and Industrial Research Organisation
(CSIRO), Exploration and Mining, has developed nuclear spectrometric
techniques for borehole logging. The paper reviews the latest
developments of this technology for the metalliferous mining industries.
The SIROLOG logging system developed by CSIRO Exploration and
Mining has been commercialised and is available to the mining industry.
INTRODUCTION
Geophysical techniques are well established in the resource
industries like oil, gas, uranium, coal and minerals. Nuclear
borehole logging, which represents a subset of this group, has
been widely used in the oil, gas, uranium and coal industries for
a long time. Nuclear borehole logging tools encompass three
types: neutron source and gamma-ray detector, gamma-ray
source and gamma-ray detector and finally, gamma-ray detector
with no source. Owing to the deep penetration of neutrons and
gamma rays, nuclear techniques are suitable for borehole logging
applications and they are making inroads in the metalliferous
mining industry.
Many holes drilled in the coal and metalliferous mining
industries are cored and a chemical analysis performed. This
analysis typically is the only source of information from a
borehole. However, this does not make nuclear borehole logging
redundant. The benefits that can be derived from nuclear logging
are:
1.
it samples a much larger volume of the material

surrounding the borehole than the core sample and
therefore provides better sampling statistics especially in
heterogeneous deposits;
2.
the results are instantaneous; and
3.
the cost of drilling open holes is cheaper than the cost of

cored holes.
Considering that the full information provided by the
laboratory analysis is not always needed and that in some
mineral deposits the core cannot be fully recovered, nuclear
logging and laboratory analysis of the core are complementary.
CSIRO Exploration and Mining has developed the
spectrometric nuclear logging system, SIROLOG, based on
natural-gamma, gamma-gamma and neutron-gamma techniques.
In this system the whole gamma ray spectrum is recorded for
each preset logging interval. By recording and analysing the
whole gamma-ray energy spectrum from a spectrometric
measurement one can extract more information from the logging
data. A dedicated software package for spectrometric data
analysis and interpretation has also been developed. The data
interpretation and the source/detector configuration are usually
specific to the type of application, that is, tuned for each specific
application.
INSTRUMENTATION
Gain stabilisation is an essential part of a spectrometric system.
SIROLOG has been upgraded to a fully digital system. Pulses
produced by the gamma-ray detector are processed by a
1.
CSIRO Exploration and Mining, PO Box 883, Kenmore Qld 4069.
microprocessor incorporated in the probe and transmitted to the

uphole computer. The digital transmission of the signal is noise
proof and is not susceptible to attenuation in the cable. The
whole system consists of the probe, winch and a laptop
computer. When low activity sources are used and a source
transporter is not required, the system is portable and does not
require a dedicated logging vehicle. This makes SIROLOG
suitable for logging in areas where the only access is by
helicopter.
The standard logging probe has a diameter of 60 mm and can
accommodate a scintillation detector of 37 mm diameter. The
length of the probe is about 2 m. When logging large diameter
boreholes it is advantageous to use larger volume detectors,
which are more efficient for gamma ray detection. Larger
diameter probes are constructed for these applications.
Scintillation detectors are used in the SIROLOG system. The
most common type used in the gamma-gamma tool is NaI(Tl),
while BGO (bismuth germanate) is the preferred scintillator for
natural-gamma and neutron-gamma logging.
APPLICATIONS
Iron ore mining industry
The three nuclear borehole logging techniques, prompt gamma
neutron activation, gamma-gamma and natural gamma have
found applications in the iron ore mining industry.
One of the features of nuclear logging is that it samples a
much larger volume than coring and subsequent chemical
analysis. The following example shows the weight of ore
sampled by the conventional sampling and nuclear logging for a
typical production bench. The bench consists of a 10 m regular
grid of holes to a depth of 20 m intersecting ore of density
2.8 t/m3. Consequently, each hole represents an ore-block
weighing 5600 t. In conventional sampling, a sample of around
4 kg taken manually from the cone of drill cuttings represents the
5600 t ore-block, giving a sample to ore-block ratio of
approximately 1:1 400 000. Both gamma-gamma and prompt
gamma neutron activation techniques have been used
successfully for borehole logging in the iron ore mining industry.
The sampling radius of gamma-gamma and neutron-gamma are
150 mm and 300 mm respectively. For a 300 mm diameter blast
hole, the masses of ore sampled by the gamma-gamma and
neutron-gamma tools are approximately 12 and 32 t respectively.
Consequently, the sample to ore-block ratios for these two
techniques are about 1:467 and 1:175 which is much more
favourable than for the conventional method.
Prompt gamma neutron activation logging has been used
successfully for the determination of iron grade and silica
content in iron ore (Eisler et al, 1977; Charbucinski, 1991;
Charbucinski, 1993). Neutron-gamma is a specific method; its
response is characteristic to elemental composition of the ore.
The neutron-gamma reactions produced by thermal neutrons
produce gamma-rays specific to the chemical elements found in
the ore. Both iron and silicon have relatively large
neutron-capture cross-sections and produce easily identifiable
gamma-rays used to determine their concentration in the ore.
Figure 1 shows a neutron gamma spectrum recorded by the
prompt gamma neutron activation probe with a BGO detector in
347
M BORSARU and J CHARBUCINSKI
FIG 1 - Spectrum collected with the prompt gamma neutron activation probe in iron ore.
an iron ore deposit. The iron and silica peaks are evident in the
spectrum. The accuracy for the determination of iron in 280 mm
blastholes was 0.54 per cent Fe and for the determination of
average silica content was 0.65 per cent Si per blasthole.
The spectrometric gamma-gamma technique enabled the
simultaneous measurement of iron grade, density and borehole
diameter on a stratigraphic basis in wide (310 and 380 mm) dry
blast holes in iron ore deposits (Eisler et al, 1987). The primary
gamma-ray source used was 60Co. The accuracy for iron
determination in 1.5 m split intervals was 1.6 per cent Fe in ore
bearing at least 55 per cent Fe. For ore of wider range in grade,
the accuracy was 1.9 per cent Fe, where ore grades ranged from
35 to 69 per cent Fe, whilst the accuracy was three per cent Fe
where the minimum grade was nine per cent Fe. Total blast hole
grade, over an average length of 15 m, was determined with the
accuracy ranging from 0.9 to 1.3 per cent Fe, depending on the
quality and location of the ore. The accuracy for density
measurements was 0.18 and 0.30 t/m3 in blast holes of 380 and
310 mm diameter respectively. The accuracies for the hole
diameter determinations were 10 and 5 mm for hole diameter
ranging from 310 to 410 mm and from 390 to 500 mm
respectively.
Natural-gamma is used for delineating the iron orebody based
on the big difference in natural gamma radiation between the
iron ore (low in natural gamma radiation) and the shaly rock. It
can also provide a means for determining alumina contamination
of iron ore based on the correlation between alumina and the
kaolinitic material of the ore matrix (Charbucinski et al, 1991)
Manganese
Both prompt neutron gamma activation and neutron activation
techniques have been used for the determination of the
manganese content of manganese ore (Aylmer et al, 1984). 252Cf
neutron sources were used for both techniques. The field trials
were carried out in dry, shallow (15 m deep) boreholes of 150
348
mm diameter. In spite of the difficulties in sampling for

calibration, both of the quantitative techniques performed
accurately given the wide range of ore grades and ore types
encountered in the deposits where the field trials took place. The
root-mean-square (rms) deviation between manganese grade
based on regression analysis of logging data and laboratory
assays was approximately 3.4 per cent Mn for ores ranging in
grade from ten per cent to 50 per cent Mn for both techniques.
However, the prompt gamma neutron activation was less
dependent than the neutron activation on chemical differences
between the mineralised ores at the different deposits.
Nickel logging
Preliminary investigation on nickel logging by CSIRO showed
that spectrometric logging could be beneficial to the nickel
mining industry. Prompt gamma neutron activation was found to
be the only quantitative technique to this application. On the
other hand, although it was not possible to interpret the
spectrometric gamma-gamma probe quantitatively, because the
iron probe response is similar to the nickel response, the logs
obtained using this probe delineated the mineralised zones
intersected by the boreholes.
RECENT DEVELOPMENTS USING ULTRA-LOW

RADIATION INTENSITY GAMMA-RAY SOURCES
Although nuclear techniques are widely used in the mining
industry, some mines are still reluctant to use them because of
the extra care required when working with radioactive sources.
Work has been carried out over the last years to develop
environmentally friendly techniques for in situ analysis using
ultra-low radiation intensity gamma-ray sources. Equipment for
in situ analysis using low activity sources significantly simplifies
safety procedures and reduces to a minimum the source radiation
risk. Logging systems using very low activity sources are much
more likely to be accepted by the mining industry.
NUCLEAR BOREHOLE LOGGING TECHNIQUES
Two source/detector configurations have been developed

(Charbucinski et al, 1997). One configuration, used in the LOW
ACTIVITY TOOL, comprises a 1.8 MBq 137Cs gamma-ray
source placed along the axial centreline of the detector. The other
configuration, used in the ZERO PROBE, comprises three 0.36
MBq 137Cs gamma-ray sources placed circumferentially around a
cylindrical iron/lead shield.
Iron ore
The Low Activity Tool was tested in an iron ore deposit (Borsaru
et al, 1995). The probe proved suitable for delineation of the
orebody and also for predicting its grade. The iron ore region can
be delineated from the count rates recorded in different energy
windows of the backscattered spectrum, or the ratio of counts
recorded in two windows selected in the high and low energy
regions of the spectrum (Pz ratio). The low energy region of the
spectrum is sensitive to the average atomic number Zeq of the
matrix and it can therefore be used to delineate the iron ore (high
Zeq) from the host rock (low Zeq). Figure 2 shows the delineation
of a shale band, approximately 1 m thick, intersected by the
borehole in an iron ore deposit.
Pb-Zn ores
Both single source and three source configurations were tested
for orebody delineation and grade control of Pb-Zn ore
(Charbucinski et al, 1997; Almasoumi et al, 1998).
Lead grade is determined from the 80 keV K X-ray peak
excited by the multiscattered gamma-rays. The probes were field
tested in two cored holes in a Zn ore deposit, reamed later to a
diameter of 142 mm. The holes were water-filled. Figure 3 shows
spectra collected in a borehole over the intervals 93.95 95.45
and 106 107.2 m, with corresponding Pb assays of six per cent
and 0.04 per cent Pb respectively. A strong Pb peak is visible at
80 keV energy for the interval containing high lead. The lead
peak is absent in the other spectrum. The rms deviation given by
the regression equation was 0.3 per cent Pb for both probes and
the standard deviation of the population was 1.7 per cent Pb.
FIG 2 - The delineation of a shale band in an iron ore deposit with the
Low Activity Tool.
FIG 3 - Spectra collected with the Low Activity Tool over two intervals with Pb assays of six per cent and 0.04 per cent.
349
M BORSARU and J CHARBUCINSKI
The gamma-gamma probe is not able to measure the

concentration of zinc directly. The probes response is related to
the overall contributions given by the major components with
high atomic number which are present in the Pb-Zn ore, eg Pb,
Zn, Fe and Mn. Because Pb concentration can be measured
directly, the determination of Zn is possible if the Fe and Mn
concentrations in the ore are constant, or can be estimated in a
different way. The concentrations of Fe and Mn were estimated
in this work from statistical information obtained from a
geological database. The rms deviation for the determination of
per cent Zn was 2.4 per cent Zn and the correlation coefficient
was 0.85. The standard deviation of the population of 72 samples
used was 4.55 per cent Zn.
CONCLUSIONS
Nuclear techniques are more widely used in the oil industry than
in the mineral industry. One explanation to this fact is that
nuclear logging provide vital information for the oil industry
which can not be obtained by any other means. In the mineral
mining industry, core recovery and subsequent chemical analysis
can provide all the information required, although at a higher
cost. Also, the cost of logging in the oil industry constitutes only
a small fraction of the cost of drilling deep wells. This is not the
case when drilling much shallower, cheaper holes common in the
mineral industry. As a percentage of the cost of drilling, logging
costs are much higher for the mineral industry.
There are important benefits from using nuclear logging:
1.
it samples a much larger volume of rock than the core

samples;
2.
it provides results in real time; and
3.
it saves money by providing information from open holes,

which are cheaper to drill than the cored holes.
The spectrometric SIROLOG system for in situ analysis
developed by CSIRO has proved itself in the Australian mining
industry. The new fully digitised, portable systems using
ultra-low radiation intensity gamma-ray sources will make the
system even more competitive.
350
It is very likely that nuclear techniques will contribute much

more to the mineral industry in the future, especially at
development and mine production stages.
REFERENCES
Almasoumi A, Borsaru, M and Charbucinski, J, 1998. Determination of
the lead concentration of Pb-Zn ores in laboratory boreholes using
gamma-gamma techniques with very low activity sources, Applied
Radiation and Isotopes, 49:125-131.
Aylmer, J A, Charbucinski, J, Eisler, P L and Youl, S F, 1984.
Quantitative borehole logging of manganese ore by prompt
neutron-gamma and neutron activation methods. SPWLA 25th
Annual Logging Symposium, p D, (New Orleans).
Borsaru, M, Ceravolo, C and Tchen, T, 1995. The application of the low
activity borehole logging tool to the iron ore mining industry,
Nuclear Geophysics, 9:55-62.
Charbucinski, J, 1991. In situ assaying of iron ore in blast holes for iron
ore and silica content, in Proceedings IAEA CRP Meeting on
Nuclear Techniques in Exploration and Exploitation of Natural
Resources, (Debrecen).
Charbucinski, J, Millitz, P and Ceravolo, C, 1991. in situ assaying of iron
ore in blast holes for alumina content, CSIRO, Division of
Geomechanics Internal Report (Newseries) No 61.
Charbucinski, J, 1993, Comparison of Spectrometric neutron-gamma and
gamma-gamma techniques for in situ assaying for iron grade in large
diameter production holes, Nuclear Geophysics, 7:133-141.
Charbucinski, J, Borsaru, M and Gladwin, M, 1997. Ultra-low radiation
intensity spectrometric probe for orebody delineation and grade
control of Pb-Zn ore, in Proceedings of Exploration 97: the Fourth
Decenial International Conference on Mineral Exploration,
September 1997, pp 631-638, Toronto, Canada.
Eisler, P L, Huppert, P, Mathew, P J, Wylie, A W and Youl, S F, 1977.
Use of neutron capture gamma radiation for determining grade of
iron ore in blast holes and exploration holes, in Proceedings of IAEA
Symposiom on Nuclear Techniques and Mineral Resources, 215 p,
(Vienna).
Eisler, P L, Charbucinski, J, Borsaru, M and Indrans, A P, 1987.
Spectrometric gamma-gamma borehole methods for iron ore mining,
in Proceedings 2nd International Symposium on Borehole
Geophysics for Minerals, Geotechnical and Groundwater
Applications, 189 p, (Golden, Colorado).
Publications published by
The AusIMM
THE AUSTRALASIAN INSTITUTE OF MINING AND METALLURGY

Level 3, 15 - 31 Pelham Street, Carlton Vic 3053
MONOGRAPH SERIES
1
Detrital Heavy Minerals in Natural Accumulates
George Baker
1962
2.
Research in Chemical and Extraction Metallurgy
Ed: J T Woodcock,
A E Jenkins
and G M Willis
1967
3.
Broken Hill Mines - 1968
Ed: M Kadmanovich
and J T Woodcock
1968
4.
Economic Geology of New Zealand
Ed: G J Williams
1974
5.
Economic Geology of Australia and Papua New Guinea - 1 Metals
Ed: C L Knight
1975
Economic Geology of Australia and Papua New Guinea - 2 Coal
Ed: D M Traves
and D King
1975
7.
Economic Geology of Australia and Papua New Guinea - 3 Petroleum
Ed: R B Leslie
H J Evans
and C L Knight
1976
8.
Economic Geology of Australia and Papua New Guinea 4 Industrial Minerals and Rocks
Ed: C L Knight
1976
9.
Field Geologists Manual

1st Edition
2nd Edition
3rd Edition
Ed: D A Berkman
and W Ryall
Ed: D A Berkman
1976
1982
1989
10
Mining and Metallurgical Practices in Australasia

(the Sir Maurice Mawby Memorial Volume)
Ed: J T Woodcock
1980
11.
Victorias Brown Coal - A Huge Fortune in Chancery

(the Sir Willis Connolly Memorial Volume)
Ed: J T Woodcock
1984
12.
Australasian Coal Mining Practice

1st Edition
2nd Edition
Ed: C H Martin
Ed: C H Martin
and A J Hargraves
1986
1993
13.
Mineral Deposits of New Zealand
Ed: Dr D Kear
1989
14.
Geology of the Mineral Deposits of Australia and Papua New Guinea
Ed: F E Hughes
1990
15.
The Rocks Speak
H King
1989
16.
Hidden Gold - The Central Norseman Story
J D Campbell
1990
17.
Geological Aspects of the Discovery of

Some Important Mineral Deposits in Australia
K R Glasson
and J H Rattigan
1990
18.
Down Under - Mineral Heritage in Australasia
Sir Arvi Parbo
1992
19.
Australasian Mining and Metallurgy

(the Sir Maurice Mawby Memorial Volume)
Ed: J T Woodcock
and K Hamilton
1993
20.
Cost Estimation Handbook for the

Australian Mining Industry
Ed: M Noakes
and T Lanz
1993
21.
History of Coal Mining in Australia

(the Con Martin Memorial Volume)
Ed: A J Hargraves,
R J Kininmonth,
C H Martin
and S M C Saywell
1993
22.
Ed: D Berkman
and D Mackenzie
1998
Copies of all books currently in print can be obtained from

The Institute office - Tel (03) 9662 3166
Key:
Out of print
PUBLICATIONS OF THE AUSTRALASIAN INSTITUTE

OF MINING AND METALLURGY
CONFERENCE, SYMPOSIUM AND MISCELLANEOUS
1972
1/72
1973
SI
*
*
Annual Conference, Newcastle
1/73
S2
Subsidence, Illawarra
2/73
S3
Mine Filling, North West Queensland
3/73
S4
Transportation, Sydney
4/73
S8
Mine Fires, Southern Queensland
Annual Conference, Western Australia
1/74
S5
Support in Pillar Extraction, Illawarra
2/74
S6
Recent Technical and Social Advances in the North Australian Minerals Industry,
North West Queensland
3/74
S7
Pellets and Granules, Newcastle and District
Annual Conference, Southern and Central Queensland
2/72
5/73
1974
4/74
1975
1/75
S9
People and the Mining Industry - The Future, Broken Hill
2/75
S 10
Occupational Safety in Mines, Southern Queensland
3/75
S 11
Australian Black Coal, lllawarra
Annual Conference, South Australia
S13
Landscaping and Land Use Planning as Related to Mining Operations, Adelaide
Design and Construction of Tunnels and Shafts, Melbourne - 2nd Australian

Tunnelling Conference
4/75
1976
1/76
2/76
3/76
S 14
Thick Seam Mining by Underground Methods, Central Queensland
4/76
S 15
Sampling Practices in the Mineral Industries, Melbourne
AnnualConference, lllawara
5/76
1977
1 /77
S 16
Apcom 77, Brisbane
2/77
S 18
Coal Borehole Evaluation, Southern Queensland
3/77
S 17
Underground Operators Conference, Broken Hill
Annual Conference, Hobart
4/77
1978
1979
1 /78
S 19
Mill Operators Conference, North West Queensland
2/78
S20
Rock Breaking Equipment and Techniques, Melbourne
3/78
International Resource Management, Canberra
4/78
Annual Conference, Townsville
1/79
S21
Utilisation of Steelplant Slags, lllawarra
2/79
S22
Estimation and Statement of Mineral Reserves, Sydney
Annual Conference, Perth
3/79
1980
Project Evaluation and Management, Melbourne
1/80
Annual Conference, New Zealand
2/80
S23
Australia/Japan Extractive Metallurgy Symposium, Sydney
3/80
S/24
Occurrence, Prediction and Control of Outbursts in Coal Mines, Southern

Queensland
4/80
S25
Management in the Mining Industry, Melbourne
Copies of all books currently in print can be obtained from The Institute office
Tel (03) 9662 3166 or Fax (03) 9662 3662
* = Out of print
The S numbers in the third column refer to an older identifying number for Symposia, the numbers preceeding the S number signify the new
publication ordering number.
1981
I/81
S26
2/81
3/81
S27
4/81
1982
Annual Conference, Sydney

Strip Mining 45 Metres and Beyond, Central Queensland
1/82
S29
Off Highway Truck Haulage Conference, Newman
2/82
S30
Mill Operators Conference, North West Queensland
3/82
S31
Underground Operators Conference, West Coast Tasmania
Annual Conference, Melbourne
5/82
S32
Carbon-ln-Pulp Technology for the Extraction of Gold, Perth and Kalgoorlie,

(Reprinted 1988)
6/82
S33
Seam Gas Drainage with Particular Reference to the Working Seam, Wollongong
1/83
S34
Computers in Mining, Southern Queensland
Annual Conference, Broken Hill
3/83
S35
Project Development Symposium, Sydney
4/83
S37
Ventilation of Coal Mines, Wollongong
5/83
S40
Principles of Mineral Flotation (The Wark Symposium), Adelaide
1/84
S36
Metallurgy Symposium, Melbourne
2/84
S38
Coal and Mineral Sizing, Wollongong
Annual Conference, Darwin
4/84
S39
1/85
S41
Smelting and Refining Operators Symposium, North Queensland
2/85
S42
Underground Operators Conference, Kalgoorlie
Annual Conference, Brisbane
Scientific and Technological Developments in Extractive Metallurgy (G K

Williams Memorial Volume), Melbourne
l3th Congress The Council of Mining and Metallurgical Institutions, Singapore, 6

Volumes
4/85
1987
Ignitions, Explosions and Fires, Wollongong
3/85
1986
Fourth Australian Tunnelling Conference
*
*
3/84
1985
S28
2/83
1984
International Blast Furnace Hearth and Raceway Symposium, Newcastle
5/81
4/82
1983
S43
1/86
Gold Mining, Metallurgy and Geology, Kalgoorlie
2/86
S44
Selective, Open Pit Gold Mining Seminar, Perth
3/86
S45
Ground Movement and Control Related to Coal Mining, Wollongong
4/86
S46
Australia: A World Source of Illmenite, Rutile, Monazite and Zircon Conference,

Perth
5/86
S47
Second Project Development Symposium, Sydney
6/86
S48
Large Open Pit Mining Conference, Newman
7/86
S49
Education and Research for the Mineral Industry for the Future, Melbourne
8/86
The AuslMM 10 Year lndex
1/87
Vl Australian Tunnelling Conference: Bore or Blast, Melbourne
Risk and Survival Seminar, Canberra
Annual Conference, Newcastle: Coal Power 87
Research and Development in Extractive Metallurgy, Adelaide
Leslie Bradford Golden Jubilee Oration
Mining and Environment: A Professional Approach, Brisbane
Pacrim 87, Gold Coast, Queensland
Dense Medium Operators Conference, Brisbane
2/87
S50
3/87
4/87
S51
5/87
6/87
S52
7/87
8/87
S53
Tel (03) 9662 3166 or Fax (03) 9662 3662
* = Out of print
9/87
S54
Equipment in the Minerals Industry: Exploration Mining and Processing

Conference, Kalgoorlie
10/87
S55
Resources and Reserves, Sydney
South Australias Mining Heritage
S56
21st Century Higher Production Coal Mining Systems Symposium, Wollongong
The Second International Conference on Prospecting in Arid Terrain, Perth
11/87
1988
1/88
2/88
3/88
S57
Third Mill Operators Conference, Cobar
4/88
S58
Underground Operators Conference, Mount Isa
5/88
Fourth International Mine Ventilation Congress, Brisbane, (Proceedings and

Addendum volume)
6/88
Annual Conference, Sydney: Minerals and Exploration at the Crossroads: The

International Outreach
7/88
S59
Second AuslMM Mineral Heritage Seminar, Sydney
8/88
S60
Economics and Practice of Heap Leaching in Gold Mining Workshop, Cairns
Third International Mine Water Congress, Melbourne
9/88
10/88
1989
Explosives in Mining Workshop, Melbourne
1/89
S61
Mineralogy and Petrology, Sydney, February
2/89
Second Large Open Pit Mining Conference, Latrobe Valley Vic
3/89
NQ Gold 89 Conference, Townsville Qld
4/89
Annual Conference, Perth-Kalgoorlie: Education, Training and Professional

Development; Industrial Minerals; Project Development/Processing
5/89
Mineral Fuel Alternatives and the Greenhouse Effect, July 1989
6/89
7/89
Non-ferrous Smelting Symposium: 100 Years of Smelting and Refining

Operations in Port Pirie, SA September 1989
*
89
1990
1991
1 /90
Dewatering Technology and Practice Conference, Brisbane October 1989

MINVAL 89, Mining and Petroleum Valuation 1989, Sydney September 1989
Ore Reserve Estimates - The Impact on Miners and Financiers, Melbourne, March
1990
2/90
Annual Conference, The Mineral Industry in New Zealand, Rotorua New

Zealand, March 1990
3/90
Pacific Rim Congress, Gold Coast Qld, May 1990
4/90
Mining Industry Capital and Operating Cost Estimation Conference, Sydney, June
1990
5/90
Third International Symposium on Rock Fragmentation by Blasting, Brisbane,

August 1990
6/90
Sir Edgeworth David Memorial Oration, May 1990
7/90
Mine Geologists Conference, Mount Isa, October 1990
1/91
Fourth Mill Operators Conference, Burnie Tas, March 1991
2/91
World Gold 91, Cairns Qld, April 1991
3/91
4/91
Mining Industry Optimisation Conference, Sydney, June l991

*
5/91
PNG Geology, Exploration and Mining Conference, Rabaul, June 1991

Qld Coal Symposium, Brisbane, August 1991
6/91
Reliability Production and Control in Coal Mines, Wollongong, September 1991
7/91
Fifth AuslMM Extractive Metallurgy Conference, Perth, October 1991
Tel (03) 9662 3166 or Fax (03) 9662 3662
* = Out of print
1992
1/92
Enviromine Australia, Sydney NSW, March 1992
2/92
The AuslMM Annual Conference, The State-of-the-Art - A Product of 100 Years

of Learning, Broken Hill NSW, May 1992
3/92
Energy, Economics and Environment Gippsland Basin Symposium, Melbourne,

June 1992
4/92
Arnold Black Mineral Heritage Oration
5/92
The Man from ASARCO: a life and times of Julius Kruttschnitt
6/92
5th Underground Operators Conference, Ballarat, July 1992
7/92
11th International Conference on Ground Control in Mining, Wollongong, July

1992
8/92
Third Large Open Pit Mining Conference, Mackay, August 1992
9/92
Extractive Metallurgy of Gold and Base Metals Conference, Kalgoorlie, October

1992
10/92
11/92
1993
1994
1995
Sampling Practices in the Minerals Industry, Mount Isa, November 1992

*
Rehabilitate Victoria, Latrobe Valley, November 1992
1/93
Mining People - A Century
2/93
The AuslMM Centenary Conference, Adelaide, March 1993
3/93
XVIII International Mineral Processing Congress, Sydney, May 1993
4/93
Narrow Vein Mining Seminar, Bendigo, June 1993
5/93
International Mining Geology Conference, Kalgoorlie, July 1993
6/93
Vlll Australian Tunnelling Conference, Sydney, August 1993
7/93
World Zinc 93 - International Symposium, Hobart, October 1993
1/94
1994 AuslMM Student Conference, Brisbane, April 1994
2/94
PNG Geology, Exploration and Mining Conference, Lae, PNG, lune 1994
3/94
No Two The Same by Bert Mason
4/94
Sixth Extractive Metallurgy Conference, Brisbane July 1994
5/94
1994 AuslMM Annual Conference, Darwin, August 1994
6/94
4th Large Open Pit Mining Conference, Perth, September 1994
7/94
Recent Trends in Heap Leaching, Bendigo, September 1994
8/94
Maintenance in the Mining and Metallurgical Industries,Wollongong, October

1994
9/94
Fifth Mill Operators Conference, Roxby Downs, October 1994
10/94
Mineral Valuation Methodologies 1994, Sydney, October 1994
11/94
Victorian Mining Week Conference, Melbourne, October 1994
1/95
1995 AuslMM Annual Conference, Newcastle, March 1995
2/95
Sir Maurice Mawby Memorial Oration
3/95
Worlds Best Practice in Mining and Processing Conference, Sydney, May 1995
4/95
APCOM XXV 1995 Conference, Brisbane, July 1995
5/95
Mineral Valuation Methodologies 1994, Sydney, October 1994 (revised)
6/95
EXPLO 95 Conference, Brisbane, September 1995
7/95
Underground Operators Conference, Kalgoorlie, November 1995
8/95
Young Professionals Conference, Mt Isa, October 1995
9/95
10/95
PACRIM 95 Congress, Auckland, New Zealand, November 1995

Ethics, Liability and the Technical Expert, Sydney, December 1995
Tel (03) 9662 3166 or Fax (03) 9662 3662
* = Out of print
1996
1/96
1996 AusIMM Annual Conference, Perth, March 1996
1a/96
1996 AusIMM Annual Conference Supplementary Volume, Perth, March 1996
2/96
Ethics, Liability and the Technical Expert, Sydney, March 1996
3/96
Entrepreneurs and Partners, Sydney, July 1996
4/96
Contract Operators Conference, Kalgoorlie, October 1996
5/96
Asia/Pacific Mining Communications Summit, Singapore, November 1996 Withdrawn
6/96
1997
1998
1/97
1997 AusIMM Annual Conference, Ballarat, March 1997
2/97
World Gold 97 Conference, Singapore, September 1997
3/97
Sixth Mill Operators Conference, Madang, PNG, October 1997
4/97
Gem 97, Madang, PNG, October 1997
5/97
Contract Operators Conference, Brisbane, Qld, October 1997
6/97
Third International Mining Geology Conference, Launceston, Tas, November

1997
7/97
Mindev 97 - The International Conference on Mine Project Development, Sydney,

November 1997
8/97
1997 AusIMM Travelling Technology Forum, Singleton, NSW, March 1997
1/98
MINEFILL 98 - The Sixth International Symposium on Mining with Backfill,

Brisbane, Qld, April 1998
2/98
1999
2000
Nickel 96, Kalgoorlie, November 1996
AusIMM98 - The Mining Cycle, Mount Isa, Qld, April 1998
3/98
Seventh Underground Operators Conference, Townsville, Qld, June/July 1998
4/98
Mine to Mill Conference, Brisbane, Qld, October 1998
5/98
Third Regional APCOM - Computer Applications in the Minerals Industries

International Symposium, Kalgoorlie, WA, December 1998
1/99
10th Australian Tunnelling Conference, Melbourne, Vic, March 1999
1a/99
10th Australian Tunnelling Conference Keynote Addresses and AsiaPacific

Forum, Melbourne, Vic, March 1999
2/99
Students and Young Professionals Conference, Perth, WA, July 1999
3/99
ICARISM 99 Conference, Perth, WA, September 1999
4/99
PACRIM 99 Congress, Bali, October 1999
5/99
Explo 99 Conference, Kalgoorlie, WA, November 1999
1/2000
Southern Africa - Australia Mineral Sector Synergies Symposium, Canberra,

ACT, March 2000
2/2000
After 2000 - The Future of Mining, Sydney, NSW, April 2000
3/2000
4th International Mining Geology Conference, Coolum, Qld, May 2000
Tel (03) 9662 3166 or Fax (03) 9662 3662
* = Out of print
SPECTRUM SERIES
1.
Making the Mount Isa Mine, 1923 - 1933
Don Berkman
1996
2.
History of Drilling
Graham McGogggan
1996
3.
The Cobar Mineral Field - A 1996 Perspective
Warren Cook
Andrew Ford
Julian McDermott
Peter Standish
Craig Stegman and
Therese Stegman
1996
4.
Towards 2000 - Resource to Reserve Inputs Seminar - Melbourne, Vic
1997
5.
Towards 2000 - National Conference on Ironmaking Resources and Reserves

Estimation, Perth, WA
1997
6.
Towards 2000 - The Resource Database Towards 2000 - Wollongong, NSW
1997
7.
Towards 2000 - Ore Reserves and Finance - Sydney, NSW
1998
8.
Towards 2000 - Assessment of Reserves in Low Rank Coals - Morwell, Vic
1997
9.
Towards 2000 - Ore Reserve Reconciliation Workshop - Darwin, NT
1997
10.
Towards 2000 - Gold and Nickel Ore Reserve Estimation Practice Seminar
1997
11.
Towards 2000 - Resource/Reserves Estimation Practice in the Central West

New South Wales Mining Industry, Cobar, NSW
1998
Tel (03) 9662 3166 or Fax (03) 9662 3662

Yhd-33 3620 The Jorc Code - 2.6.6

Загружено:

Сведения о документе

Оригинальное название

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

Yhd-33 3620 The Jorc Code - 2.6.6

Загружено:

Авторское право:

Доступные форматы

Fourth International Mining

The Australasian Institute of Mining and Metallurgy

The Australasian Institute of Mining and Metallurgy 2000

Desktop published by:

AusIMM Central Services

Dynamic Links Between Geology and the Mining Process

T C McCuaig, J Vann and

Sampling and Databases

Technical Data Management at Porgera

Sampling Practice at the Vera Nancy Gold Mine

Leachwell Versus Fire Assay: Comments from Assay Quality Assurance

Controls on High-Grade Gold Distribution at Vera Nancy Mine

Geological Modelling and Grade Control in a Narrow Vein, High-Grade

Mine Geology Practices at the Sunrise Open Pit

E Haren and P Williams

The Spatial Distribution of Grade

The Bunyip Lateritic Nickel-Cobalt Deposit, Cawse Nickel-Cobalt

A Bywater and S M Denn

Cadia Hill Gold Mine One Year Down the Track

The Geita and Kukuluma Mineralised Trends, Lake Victoria Goldfield,

Kunwarara Magnesite Deposit

Current Geological Understanding of Telfer Gold Mine

Geology and Structure of the Morning Star Mine, Mt Magnet, WA

Resource Evaluation of Nuggety Slate-Hosted Gold-Quartz Reefs

Multiple Indicator Kriging Is it Suited to My Deposit?

J Vann, D Guibal and

Mining Bench Height Evaluation for the Wallaby Resource

I M Glacken, M Nopp and

Ore Definition at the Henty Gold Mine Tasmania

Computer-Based Resource Estimation in Accordance with the

J Duke and P Hanna

Mining Grade Control Past, Present and Future

Evolution of Grade Control at KCGM

V M OBrien and T Cutts

Ore Characterisation for Mine to Mill Fragmentation

A Scott, J Segui and

Upgrade Ability and Geology of Cawse Nickel Ore

The Interaction Between Geology, Mining and Metallurgy at Stawell

Resource and Reserves

Geotechnical and Structural

Pasminco Century Mine Open Pit Slope Design A Geotechnical

Modelling Bulk Density The Importance of Getting it Right

The Expanding Role of Mine Geophysics

Detailed Orebody Mapping Using Borehole Radar

Application of the SIROLOG Downhole Geophysical Tool at Callide

Microseismic Monitoring of Shear Zones and Related Seismic Activity at

J Bishop and R Lewis

Nuclear Borehole Logging Techniques Developed by CSIRO Exploration

Role of the Mine Geologist

Dynamic Links Between Geology and the

T C McCuaig, J Vann and

The Mine Geologist in a Business Perspective

customer/supplier relationships, and

MAusIMM, General Manager Resource Development, Normandy

4th International Mining Geology Conference

On reflection, the concept of a business perspective for a mine

Coolum, Qld, 14 - 17 May 2000

In a mine, many people visualise the process of mining

In this model, Department A is larger than all the others and

FIG 2 - Schematic representation of the Linear Model.

Model 3 Functional Teams

FIG 1 - Schematic representation of the Centralist Model.

The preferred model (three) suggests that functional teams and