CIM JOURNAL

Editors International Advisory Board

Agus Sasmito Maintenance, Engineering and Reliability Members
Martin Grenon Rock Engineering Dr. Saiied Aminossadati, Senior Lecturer, University of
Queensland, Australia
Ali Madiseh Underground Mining
Dr. Juliette Biao Koudenoukpo, Director and Regional
Steve McCutcheon Geological
Representative, Africa Office, UN Environment, Kenya
FOUNDING EDITOR Chuck Edwards Dr. Mark Board, Vice President – Technical Services, Hecla
Limited, USA
ASSOCIATE EDITORS
Mr. Vitor Correia, President, European Federation of
Donna Beneteau Underground Mining Geologists, Portugal
Erin Bobicki Canadian Mineral Processors Dr. Éva Hartai, Professor, University of Miskolc; Editor,
James Budac Metallurgy and Materials European Geologist Journal, Hungary
Gabriel Castillo Environmental & Social Responsibility Dr. Pierre-Yves Le Meur, Professor, Université Paul Valéry,
Site Saint-Charles, France and Pacific New Studies Centre,
Loy Chunpongtong Surface Mining New Caledonia
Hendrik Falck Geological Dr. Leon Lorenzen, Lorenzen Consultants Pty Ltd, Mintrex,
Édith Garneau Environmental & Social Responsibility University of Stellenbosch, Australia
Dr. Baisheng Nie, Professor, China University of Mining and
EDITORIAL SERVICE Purple Rock Inc., www.purplerock.ca Technology, China
EDITORIAL COORDINATOR Janice Burke, jburke@cim.org
Mr. Bjorn Nielsen, VP Dry Comminution & Sorting, Outotec,
PUBLISHER CIM, cim@cim.org Australia
Dr. Bob Ring, Principal Consultant, ANSTO Minerals,
Australian Nuclear Science and Technology Organisation,
Australia
Dr. A. J. S. (Sam) Spearing, Director, Western Australian
School of Mines, Curtin University, Australia
Mr. Barry Vayler, Analysis & Improvement Specialist, BHP,
Australia

© 2019 Canadian Institute of Mining, Metallurgy and Petroleum

It is a condition of publication that manuscripts submitted to this
journal have not been published and will not be simultaneously
submitted or published elsewhere. By submitting a manuscript, the
authors agree that the copyright for their article is transferred to the
Canadian Institute of Mining, Metallurgy and Petroleum if and when
the article is accepted for publication. However, assignment of
copyright is not required from authors who work for organizations
which do not permit such an assignment. The copyright covers the
reprints, photographic reproductions, microfilm, or any other
reproductions of similar nature and translations. No part of this
publication may be reproduced, stored in a retrieval system or
transmitted in any form or by any means, electronic, electrostatic,
magnetic tape, mechanical, photocopying, recording or otherwise,
without permission in writing from the copyright holder.
Permission for other use
The copyright owner’s consent does not extend to copying for
general distribution, for promotion, for creating new works or for
resale. Specific written permission must be obtained from the
publisher for such copying.

The Institute, as a body, is not responsible for statements made or opinions advanced either in articles or in any discussion appearing in
its publications.
ISSN 1923-6026
Instructions for authors are available on the CIM website

Published quarterly by the Canadian Institute of Mining, Subscription rates

Metallurgy and Petroleum Online version included in CIM Membership ($197/yr).
1250 – 3500 de Maisonneuve Blvd. West Print version for institutions or agencies – Canada: $275/yr (AB,
Westmount, QC H3Z 3C1 BC, MB, NT, NU, SK, YT add 5% GST; ON add 13% HST; QC
Tel: 514.939.2710; Fax: 514.939.2714 add 5% GST + 9.975% PST; NB, NL, NS, PE add 15% HST).
www.cim.org Print version for institutions or agencies – USA/International:
US$325/yr. Online access to single copy: $50.
Layout & design: www.clocommunications.com
CIM JOURNAL
Volume 10, Number 3, 2019
Contents

115 Automation technology to increase productivity and reduce energy consumption in deep
underground mining operations
K. Moreau, R. Bose, H. Shang and J. A. Scott

125 Simulating the effect of LHD operations on production rates and ventilation costs in a
sublevel cave underground mine
B. Skawina, A. Salama, and J. Greberg

133 Reconciling mineral reserves at the well-to-well in-situ copper leaching operation at San
Manuel mine, Arizona, USA
G. A. Sutton

143 Using piston–die press tests for preliminary sizing and selection of a high-pressure
grinding roll
F. Wang, B. Klein, and P. Rosario
 CALL FOR
SUBMISSIONS
APPEL À
CONTRIBUTIONS
Join forces with some of the brightest minds and share your
expertise by submitting a Technical Paper to CIM Journal.
Rejoignez une équipe de scientifiques brillants et partagez
vos connaissances en envoyant un document technique au
CIM Journal.

www.cim.org
 MAINTENANCE, ENGINEERING AND RELIABILITY 115

Automation technology to increase productivity and reduce energy

consumption in deep underground mining operations
K. Moreau
Bharti School of Engineering, Laurentian University, Sudbury, Ontario, Canada

R. Bose
Glencore – XPS, Falconbridge, Ontario, Canada

H. Shang and J. A. Scott

Bharti School of Engineering, Laurentian University, Sudbury, Ontario, Canada

https://doi.org/10.15834/cimj.2019.11

ABSTRACT Trials of automated technology at underground mine sites around the world have reported
increased production and the potential for lower energy consumption. Autonomous mining eliminates a
range of delays related to operator downtime and travel to the workface and facilitates continuous oper-
ation in conditions that are unsafe for workers. Automation also introduces complexities and safety haz-
ards arising from equipment–personnel interactions; automated processes must be isolated from other
stope-cycle–related activities such as production drilling, development, loading, and backfilling, poten-
tially requiring changes to the mine design to maintain an efficient and safe operation. This paper reviews
and discusses the implementation of automated equipment in deep underground mining operations and
the challenges and advantages associated with the technology.
■ KEYWORDS Automated equipment, Emissions reduction, Energy, Productivity, Sustainability, Technology,
Underground mines

RÉSUMÉ Des essais de technologies automatisées dans des sites miniers souterrains aux quatre coins du
monde ont indiqué une production accrue et la possibilité de réduire la consommation d’énergie. L’exploi-
tation minière autonome élimine tout un éventail de retards liés aux temps d’immobilisation des opéra-
teurs et aux déplacements vers le front de taille, et facilite l’exploitation continue dans des conditions qui
s’avèrent peu sécurisées pour les travailleurs. L’automatisation introduit également des complexités et des
risques pour la sécurité, qui surviennent en raison des interactions entre l’équipement et le personnel. On
recommande d’isoler les procédés automatisés des autres activités liées au cycle des chantiers telles que
le forage de production, le développement, le chargement et le remblai, ce qui pourrait entraîner des chan-
gements au niveau de la conception de la mine afin de préserver rendement et sécurité dans l’exploitation.
Cet article examine et aborde la mise en œuvre de l’équipement automatisé dans des exploitations
minières souterraines profondes, ainsi que les enjeux et les avantages associés à cette technologie.
■ MOTS CLÉS durabilité, énergie, équipement automatisé, mines souterraines, productivité, réduction des
émissions, technologie

INTRODUCTION
As the mining industry continues to exhaust easily
accessible orebodies, the need for efficient, deep (typically
below 2,500 m) mine operations increases (Atlas Copco,
2013a); however, the mining process can become econom-
ically infeasible at great depths because operation costs
drastically increase. This has resulted in investigations into
automating mining operations as the industry strives for
improved productivity (Schunnesson, Gustafson, & Kumar,
2009) and ways to remove personnel from hazardous work
areas.
Advancements in autonomous mining equipment and
information technology communications infrastructure
have enabled the transition from manually operated equip-
ment and line-of-sight remote control systems to semi-
autonomous operations (Gustafson, 2011). Mine sites
around the world are trialling and implementing automated

CIM Journal | Vol. 10, No. 3, 2019

116 K. Moreau, R. Bose, H. Shang, and J. A. Scott
drills, trucks, and load-haul-dump machines (LHDs). The
intent is to increase the efficiency of removing and trans-
porting ore while also providing safer working conditions
(Chadwick, 2010).
Reports show the potential of automation technology to
increase productivity and reduce energy use (Schunnesson
et al., 2009; Chadwick, 2010; Paraszczak, Gustafson, &
Schunnesson, 2015). In particular, it has been found that
operating autonomously from the surface between manned
shifts provides an additional 3–4 h of machine productivity
per day (Schunnesson et al., 2009); however, it is difficult
to design the mine layout to exclude underground personnel
from automated areas during regular shift time. Further
research and trials are needed to determine optimal mine
layouts that isolate automated equipment in a way that does
not impact other underground operations.
BENEFITS OF AUTOMATED MINING
The many benefits associated with automated mining
can have a strong influence on the feasibility of deep mine
projects. Glencore’s Kidd Creek mine (Timmins, Ontario,
Canada), which extends 2,927 m below surface (Duddu,
2013), has operated using autonomous LHDs since 2012.
The aim was to increase safety and productivity as the mine
developed deeper underground (Kelly, 2017). Glencore has
also announced their Onaping Depth project (Sudbury,
Ontario, Canada), which will be more than 2,500 m under-
ground (White, 2018), significantly deeper than their cur-
rent Nickel Rim South mine, where operations extend from
1,100 to 1,800 m deep (Glencore, 2016; White, 2018).
Glencore has committed to an all-electric fleet for Onaping
Depth to facilitate decreased ventilation requirements, and
will also look to use automated technology to remove min-
ers from underground hazards (Tollinsky, 2018). Auto-
mated and electric equipment have the ability to lower
operating costs, increase productivity, and provide a sus-
tainable future through improved safety, decreased diesel
usage, and overall reduced energy consumption due to
lower ventilation demands.
Safety
A fundamental supporting argument for transitioning
from manual to automated mining equipment is that
automation results in increased safety and improved work-
ing conditions for operators (Paraszczak et al., 2015).
Automation allows machines to be operated from a control
station on the surface, which alleviates the need for opera-
tors to spend time underground in dangerous work areas
with poor air quality (Chadwick, 2005).
Bugarski et al. (2004) reported that underground work-
ers are exposed to the highest levels of diesel particulate
matter (DPM) of all workers in the industrial sector. Diesel
emissions control is primarily focused on regulating the
concentration of nitrogen oxides (NOx) and elemental car-
bon (EC), which is a basic fraction of DPM derived from
CIM Journal | Vol. 10, No. 3, 2019
fuel and lubricating oil (Majewski, 2016). Underground
mining and maintenance workers, including tunnel work-
ers, experience levels of EC and NOx that are four to five
times higher than workers in other industrial sectors, which
is a major concern for mining companies and their employ-
ees (World Health Organization, 2014). In 2012, DPM was
categorized as a Group 1 carcinogen by the World Health
Organization due to evidence linking exposure to an
increased risk of developing lung cancer (World Health
Organization, 2012).
Due to hazardous emissions from diesel engines,
research into alternative fuel sources, such as biofuels and
battery-electric vehicles (BEVs), has become a top priority.
There are high capital costs associated with implementing
BEVs (20–30% higher than diesel equipment; Paraszczak,
Laflamme, & Fytas, 2013); however, alternative fuels such
as biodiesel can be used as an interim solution to improve
air quality underground. While companies transition their
existing diesel-operated fleet to BEVs, some are also using
fuels with higher blends of biodiesel as a means of limiting
total carbon emissions (Validakis, 2012). Since 2012,
Macassa mine (Kirkland Lake, Ontario, Canada) has been
operating machines using biodiesel blended fuel (50%
blend in the summer; 20% blend in the winter). Emissions
tests performed on their equipment revealed a reduction in
average CO and NOx emissions from 2.36 to 1.85 ppm and
0.15 to 0.085 ppm, respectively, when compared to their
diesel equipment counterparts (Kirkland Lake Gold Inc.,
2015).
Workplace accidents are also a safety concern. Studies
have shown that there are a high number of pinch point and
fall incidents when climbing in and out of equipment, as
well as injuries resulting from collisions between heavy
mobile machinery (Caterpillar Global Mining, 2008b).
With automation, the operator is not working near (or in)
the machine, reducing the potential for these types of
injuries. Automated mining equipment also travel desig-
nated pathways with higher accuracy than manually oper-
ated equipment, preventing interequipment collisions
(Caterpillar Global Mining, 2008b).
To eliminate the possibility of autonomous vehicles
coming into contact with work personnel, automated equip-
ment should be located within isolated work areas equipped
with light barriers (Dragt, Camisani-Calzolari, & Craig,
2005; Caterpillar Global Mining, 2015). Any object or per-
son that crosses the barrier will cause the automated equip-
ment to immediately shut down (Gustafson, 2011; Atlas
Copco, 2015b).
Productivity and utilization
A four-month trial at BHP Billiton’s Olympic Dam oper-
ation (South Australia) using Caterpillar’s MINEGEM
automation technology found that an autonomous LHD
worked an additional 1.9 h/shift compared to a manually
operated LHD typically operating 9–10 h/shift (McHugh,
Automation technology to increase productivity and reduce energy consumption in deep underground mining operations 117
2004). Most of this utilization increase came from the abil-
ity to operate between shifts during the blast clearance win-
dow; the increase led to production increases of up to 30%
(Sandvik Mining, 2017). There was also an increase in
LHD utilization due to eliminating the time it takes to travel
to and from the surface, a factor of major significance in
ultra-deep mines (Chadwick, 2010).
When tele-remote trams were implemented in the early
2000s, tram speeds were lower compared to manual opera-
tions, resulting in increased cycle times (Roberts et al.,
2000; Dragt et al., 2005); however, with advancements in
automation technology, cycle times have decreased. In
2017, an operator of automated LHDs at a Canadian under-
ground mine site reported that shorter cycle times con-
tributed to increased production levels over a four-month
period (Hudbay Minerals operator, personal communica-
tion, 2017. Cycle times of automated mining equipment are
now lower compared to manually operated machines due to
optimum tram speeds and removal of the need for the oper-
ator to climb in and out of the machine at the entrance of a
stope (Roberts et al., 2000; Schunnesson et al., 2009).
In 2017, a Canadian mine site began trialling automated
LHDs with the goal of achieving a production rate of
250 tph. During the trial, the haulage hours were increased
from 8.4 h/day to 10.8 h/day, which increased the average
tonnes moved per day. The hourly production rate also
increased from 161.4 to 217.5 tph using the automated
LHD. The initial target of 250 tph was not, on average,
achieved throughout the 13-day trial, but peak production
did surpass this target on several occasions. The belief is
that further familiarization with the technology will allow
the mine site to reach their target production on a more con-
sistent basis.
Maintenance
To optimize an autonomous mine, there must be a strong
focus on equipment maintenance. Because work personnel
are no longer located near the equipment where they can
observe and report the need for maintenance actions, a
maintenance strategy is required to prevent breakdowns
and lost production (Gustafson, 2013; Paraszczak et al.,
2015). Automation requires more experienced and skilled
personnel for proper maintenance, but this cost is offset by
decreased vehicle collisions, optimized driving, increased
tire life, and reduced consumption of spare parts (Bellamy
& Pravica, 2011).
Operations using autonomous vehicles are expected to
experience reduced machine downtime and maintenance
costs compared to manual operations. Equipment operated
manually and through line-of-sight remote control is sub-
ject to wear and tear such as damage from collisions with
drift walls, excess tire wear, and overworked engines due to
operator error and poor visibility or blind spots (Dragt et
al., 2005; Larsson, 2011; Gustafson, Schunnesson, &
Kumar, 2013). Automation can help reduce stress and strain
CIM Journal | Vol. 10, No. 3, 2019
on working machines by optimizing usage. The precision
of the navigation system allows the machine to follow an
exact pathway through narrow mine drifts, reducing dam-
age from collisions with walls and other equipment
(Gustafson, 2013; Paraszczak et al., 2015). The system can
also respond precisely to the topography of the under-
ground environment (Gustafson, 2011). For instance, auto-
mated machines will shift gears at optimum times, thereby
improving fuel efficiency and extending the equipment life
expectancy by ensuring engines are not over-revved or
overheated (Gustafson, 2011; Gustafson et al., 2013).
Environmental impact
There has been pressure in recent years to create mining
methods that result in a more sustainable operation (Cater-
pillar Global Mining, 2008c). It is generally suggested that
automation will allow mines to operate with less equipment
and burn less fuel, thereby helping reduce costs and envi-
ronmental impact (Caterpillar Global Mining, 2008c). With
each machine operating at a higher production rate, mine
sites can reduce the amount of equipment they use, and ulti-
mately reduce their greenhouse gas emissions (GHG) over
the life of the mine (Paraszczak, 2014). GHG and the site’s
carbon footprint are also reduced due to fuel efficiencies
gained from optimized driving regimes (Bellamy & Prav-
ica, 2011; McNab & Garcia-Vasquez, 2011). An improve-
ment in efficiency is also economically significant because
diesel fuel can amount to as much as 30% of total operating
costs (Bellamy & Pravica, 2011).
During the automated LHD trial at the previously men-
tioned Canadian mine site, the fuel efficiency of the auto-
mated LHD was compared to diesel equipment, with the
goal of lowering costs and reducing CO2 emissions. The
improved utilization of equipment throughout the day
should result in increased daily fuel consumption and CO2
production, but total CO2 emissions per tonne of ore pro-
duced should be lower due to greater efficiency in ore pro-
duction. Data monitoring technology incorporated with the
automation package was able to monitor fuel usage
throughout the 13-day trial. The recorded consumption of
fuel per 100 t of ore moved was on average 30% less
throughout the trial for the autonomous LHD compared to
a manually operated LHD. Furthermore, CO2 emissions
were reduced by 32% per 100 t of ore moved. This trial
used a small sample size and more extensive long-term data
will be required for a more comprehensive analysis, but the
initial findings demonstrate improved fuel efficiency and
reduced environmental impact, two key motivations for
implementing automated technology in underground
mines.
Further improvements to GHG emissions might arise
from a shift to electrification (Ewing, 2016). Transitioning
from diesel fuel to battery-electric equipment allows mine
sites to operate equipment without contributing to mine air
pollution, eliminate DPM from the underground work envi-
118 K. Moreau, R. Bose, H. Shang, and J. A. Scott

ronment (Jacobs, Hodkiewicz, & Bräunl, 2015), and reduce Because ventilation regulations in Ontario, Canada, require
costs by lowering ventilation demands (Jacobs, 2013; supplying 100 cfm/HP of diesel-powered equipment, there
Ewing, 2016; Jaderblom, 2017). Goldcorp Inc. has pro- is potential to significantly reduce ventilation requirements
jected that the shift from diesel to battery-electric at Borden by operating battery-powered equipment (Campbell, See-
Lake (Chapleau, Ontario, Canada) will eliminate the need ber, & Wywrot, 2003).
for 3 million litres of diesel fuel and 33,000 MWh of elec- According to the United States Environmental Protec-
tricity per year, and reduce annual CO2 emissions by tion Agency (2018), CO2 emissions for underground heavy
7,000 t (Braul, 2018). equipment are 700 kg/MWh of electrical power consumed
The three current methods of electrification in under- and 2.6 kg/L of the diesel fuel used. Energy consumption
ground mining are battery-electric motors, overhead power by underground mining totals 109 MWh/year, or 30% of the
lines, and umbilical trailing cables connected to the mine’s total energy consumed by the mining industry within the
electric infrastructure (Paraszczak et al., 2013; Jacobs et al., United States (Holmberg, Kivikytö-Reponen, Härkisaari,
2015); however, due to the limited range of trailing cables Valtonen, & Erdemir, 2017). With ventilation and haulage
and installed overhead power lines, battery-electric equip- contributing up to 40% and 25% of total energy consump-
ment provides the most flexibility for underground opera- tion, respectively, there is potential for significant reduc-
tions (Paraszczak et al., 2013). Battery-electric motors tions to CO2 emissions through the use of automated and
allow for travel throughout the entirety of the mine; how- electric equipment underground (Vergne, 2008; Holmberg
ever, challenges with battery life and recharge times need to et al., 2017).
be overcome before optimal operation can be achieved
(Jacobs et al., 2015). There is also the possibility of strate- NAVIGATION TECHNIQUES
gic implementation of multiple technologies in specific According to one report, existing navigation techniques
areas of the mine. Overhead power cables would be most for autonomous vehicles can be divided into four categories
effective in areas with frequent or repetitive travel, such as (Dragt et al., 2005):
within the ramp or on haulage levels, whereas battery-elec- • Absolute navigation – the vehicle is “taught” and
tric equipment would be more practical for temporary drift replays the layout of the underground work area.
or stope operations (Paraszczak et al., 2013). • Reactive navigation – the vehicle detects the environ-
The Macassa mine site, which implemented battery- ment surrounding the vehicle using sensors located on
electric equipment, found that their 18 t haulage trucks had the machine.
a 1–2 h battery life and required 1.5–2 h to recharge, • Absolute navigation and dead reckoning – the vehicle
whereas 3 yd3 (~2 m3) LHDs had a 1.5–2.5 h battery life locates itself within the work area.
and required 1.5–2 h to recharge (S. Ross, personal commu- • Reactive navigation and dead reckoning – the vehicle
nication, 2018). To overcome the constraints of battery life navigates itself through designated pathways.
and recharge times, revisions to procedures and mine Absolute navigation uses a real-world coordinate system
designs might be required to facilitate efficient battery to determine the position of an autonomous vehicle in the
changes during work shifts, and additional batteries will underground work area (Dragt et al., 2005). An operator
need to be included within capital and maintenance costs. initially drives the vehicle along the desired route to
The breakdown of the equipment fleet at Macassa mine is “teach” the machine the predefined pathway to follow
listed in Table 1, with the battery equipment listed using a (Mäkelä, 2001b). Lasers and cameras mounted on the
horsepower (HP) equivalent to their diesel counterpart. machine build a representation of the underground environ-
ment so that, when operating autonomously, the vehicle can
locate itself (Larsson, Appelgren, & Marshall, n.d.); how-
Table 1. Breakdown of the equipment fleet at Macassa mine (Kirkland
Lake, Ontario, Canada; S. Ross, personal communication, 2018) ever, during operation, obstacles that were not present dur-
Equipment Fuel type Horsepower (HP) Portion of ing the teaching phase will not be detected by the machine,
equivalent equipment which presents safety concerns and potential for machine
fleet (%)
damage (Gustafson, 2011).
Load-haul-dump Electric 250 5
machines Battery 3,455 63
Reactive navigation allows the vehicle to detect and
Diesel 1,805 33 react to its immediate environment, including any physical
Total 5,510 changes from the time the machine was taught the route
Haulage trucks Battery 2,200 85 (Roberts et al., 2000). Underground autonomous vehicles
Diesel 400 15 are equipped with lasers and cameras to identify drift walls,
Total 2,600
other vehicles, and obstacles located within the vicinity of
Combined Electric 250 3
the machine (Gustafson, 2011).
Battery 5,655 70
Diesel 2,205 27
The dead reckoning method measures the motion of the
Total 8,110 autonomous vehicle to determine its change in position
from the initial starting point by estimating the speed vector

CIM Journal | Vol. 10, No. 3, 2019

Automation technology to increase productivity and reduce energy consumption in deep underground mining operations 119

based on wheel rotations and determining the directional
trajectory of the machine (Mäkelä, 2001a; Dragt et al.,
2005). The most common instruments used for dead reck-
oning are the odometer, which measures the rotation of the
wheels to determine the distance travelled, and gyroscopes,
which measure the orientation of the vehicle (Mäkelä,
2001b; Gustafson, 2011).
NETWORK INFRASTRUCTURE
The successful operation of autonomous equipment in
underground mining also requires specific communication
infrastructure. Autonomous technology can connect to the
mine’s network backbone if it meets the performance
requirements listed in Table 2.
Table 2. Backbone network performance requirements to support
autonomous technology; compiled data from Sandvik Mining &
Construction (2017) and Atlas Copco (2017)
Parameter Requirement Notes
Latency 50 ms Video quality will degrade above
100 ms
Jitter ±20 ms Delay can only vary a small amount
Bandwidth 10 Mbits/s Includes control/safety signals and
per machine two video streams
The mine site is responsible for the installation and con-
figuration of the required backbone network, which
includes fibre optic cable connecting the operator’s station
at the surface to the autonomous work area underground
(Sandvik Mining & Construction, 2017). Table 3 lists the
minimum requirements that the network infrastructure
should support.
Communication links between the automated machine
and the control system are made via internet protocol (IP)–
based wireless (Wi-Fi™) and nonwireless (Ethernet) com-
munication (Sandvik Mining & Construction, 2017). Access
points (Wi-Fi routers) should be installed in the autonomous
work area in an arrangement that ensures the autonomous
vehicle remains within the line of sight of at least one access
point at all times. Access points are connected via Ethernet
cable, and the wireless signal strength must be maintained
above –60 dBm (Atlas Copco, 2015a). Table 4 lists the
design limitations for the autonomous work area.
There is also potential to use telecommunication or
long-term evolution (LTE) to implement automation tech-
nology. At Agnico Eagle’s LaRonde mine (Quebec,
Canada), an LTE network was installed to allow for high-
speed wireless communication via mobile devices
because it offered better coverage over a larger area com-
pared to Wi-Fi and is subject to less interference (Rolfe,
2018). The technology is being used for data sharing and
communication among workers and supervisors, and has
additional potential for future autonomous operations
(Rolfe, 2018).
ADVANCES IN MINE AUTOMATION
Mining automation is undergoing rapid development, as
older technologies such as line-of-sight remote control and
tele-remote operations are being replaced with semi-
autonomous technologies. This progression is discussed in
the following sections.
Radio remote control
The first level of automation introduced to the mining
process was remote machine operation while maintaining
an open line of sight (Gustafson, 2011). This allowed for
entrance into dangerous work areas with unsupported
ground or poor ventilation (Poole, 1999). Radio remote
control is typically used with LHDs, but reduced productiv-
ity is expected due to poor visibility and the operator
repeatedly dismounting and mounting the machine
(Roberts et al., 2000; Ghodratia, Hoseiniea, & Garmabakia,
2015).
Tele-remote operations
Tele-remote operation was the next advancement;
machines are operated from a control station on the surface
via cameras, sensors, and positioning data (McNab & Gar-
cia-Vasquez, 2011; Ghodratia et al., 2015). The operator is
in full command of the machine at all times using controls
similar to those located in manually operated equipment
(Gustafson, 2011; Ghodratia et al., 2015). Productivity is
gained through the elimination of travel time and the ability
to operate through the blast window (Chadwick, 2010;
Sandvik Mining, 2017); however, a worker can operate
only one machine at a time (Gustafson, 2011).

Table 3. Minimum standards and protocols that the network infrastructure of the mine should support to facilitate autonomous technology; data from Atlas
Copco (2017)
Standards and protocols Description Application
IEEE 802.3u 10/100 Tx-Base Ethernet LAN communications, control, and safety
IEEE 802.1Q &p VLAN, Layer-2 QoS Network segmentation and traffic priority; standard requirement for PROFINET and MODBUS/TCP
IEEE 802.11 a/b/g Wireless LAN Mobility; tele-remote control and autonomous operation
IEEE 802.1AB LLDP Layer-2 discovery; standard requirement for PROFINET and topology-aware systems
RFC 1441, 1452 SNMPv2 Network management; standard requirement for PROFINET and network monitoring
IEC 61158, 61784 PROFINET-IO Safety communications; safety barriers and loaders
MODICON 1979 MODBUS TCP/IP (or UDP) Safety communications; emergency system stop

CIM Journal | Vol. 10, No. 3, 2019

120 K. Moreau, R. Bose, H. Shang, and J. A. Scott

Cat® MineStar™ Command technology (Caterpillar Inc.,

Table 4. Cable and signal limitations for communication infrastructure in
the autonomous work area (Atlas Copco, 2015a)
2018a). Hard-Line Solutions claims to be able to configure
Component Limitation (m)
their automation technology to any make or model of
Ethernet cable < 90 machine within a mine’s existing fleet (Hard-Line Solu-
Directional antenna ~100 tions, 2018). Table 5 summarizes five major companies
Omnidirectional antenna ~40 offering automation technology for LHDs; Table 6 is a
more detailed description of available equipment and sizes.

Semi-autonomous operations CASE STUDIES

Most mine sites progressing toward increased automa-
tion are currently using semi-autonomous machines, in
which some functions, such as drilling and loading, are
controlled remotely from an operator’s station on the sur-
face, whereas others are fully automated (Chadwick, 2010;
Ghodratia et al., 2015). For example, a semi-autonomous
LHD requires operator intervention via remote operation
for bucket loading, but the machine is fully automated dur-
ing haulage and dumping actions (Chadwick, 2010;
Gustafson, 2011; Ghodratia et al., 2015). Typical on-sur-
face operating stations provided by Epiroc and Sandvik are
shown in Figure 1.
Automated LHD operations
In 2007, LKAB’s Malmberget iron ore mine (Malmber-
get, Sweden) partnered with Caterpillar to automate the
R2900G LHD in their underground operations and it was
reported that there was a 10–20% increase in production
after transitioning to automation (Caterpillar Global Min-
ing, 2008a; Schunnesson et al., 2009). This production
increase was partially attributed to the sublevel caving min-
ing method. The large, slightly inclined orebody is consid-
ered to present ideal conditions for automated LHD

Fully autonomous
operations
Achieving fully autonomous
operations requires technology to
carry out drilling and loading
functions, thereby removing the
need for any direct operator
interaction (Gustafson, 2011;
Ghodratia et al., 2015). Loading
assistance technology for LHDs
has been developed by several
equipment manufacturers, result-
ing in fully automated cycles that
eliminate the need for remote-
control operation (Sandvik Min-
ing, 2016). These advancements
also allow one operator on the
surface to oversee multiple
machines and monitor overall
productivity (Gustafson, 2011;
Ghodratia et al., 2015; Sandvik
Mining, 2016).

AVAILABLE
TECHNOLOGY
Underground mining equip-
ment with technologies to
improve safety and production is
supplied by several companies.
For example, Caterpillar Inc. has
announced a project to retrofit
the Komatsu 930E mining truck Figure 1. Typical on-surface operating stations for semi-autonomous load-haul-dump machines provided by
for autonomous driving using a) Epiroc and b) Sandvik (courtesy of Hudbay Minerals)

CIM Journal | Vol. 10, No. 3, 2019

Automation technology to increase productivity and reduce energy consumption in deep underground mining operations 121

operations due to the designed repetitiveness and low vari-
ability of the layout and activities associated with sublevel
cave mines (Caterpillar Global Mining, 2008a; Gustafson,
2011; The Redpath Group, 2012).
Codelco’s Andina copper mine (Santiago, Chile) used
Epiroc’s ST-14 scooptram over a one-year period (Atlas
Copco, 2013a). The average ore production using manually
operated LHDs was 44,850 t per month; during the trial, the
average rate increased to 80,000 t per month using the ST-14
and reached a peak of 133,000 t in one month (Atlas Copco,
2013a). The ST-14 operators were located 80 km away, far
from dangerous work conditions (Atlas Copco, 2013a).

Table 5. Mining equipment suppliers currently offering automation technology

Company Automation technology Headquarters
Caterpillar Inc. (2018b) Cat® MineStar™ Command Illinois, USA
Epiroc (2018) Scooptram automation rig control system Nacka, Sweden
Hard-Line Solutions (2018) Teleop™ Ontario, Canada
Komatsu Limited (2018) Autonomous haulage vehicle with Cat® MineStar™ Command Tokyo, Japan
Sandvik Group (2018) AutoMine® Stockholm, Sweden

Table 6. Specifications of currently available automated equipment based on suppliers’ websites; LHD, load-haul-dump machine
Supplier Equipment type Model name Size Description
Caterpillar Inc. Dozer D11T 229,800 lbs (~104 t) Semi-autonomous technology; one operator can control up to
(2018b) three machines at once
Loader 988K 4.7–13 m3 bucket size Semi-autonomous technology; move material safely and efficiently
R1600H 4.2–5.9 m3 bucket size
R1700G 4.6–8.8 m3 bucket size
R2900G 6.3–8.9 m3 bucket size
Haulage truck 793F 250 tons (~223 t) Autonomous haulage to dump points; reports maintenance issues
789D 200 tons (~179 t)
Drill MD6250 152–250 mm hole diameter Autonomous drill; more accuracy and one operator can control
up to three drills at once
Epiroc (2018) LHD ST-7 7 t load capacity Loaders capable of equiping scooptram automation, a semi-
ST-14 14 t autonomous technology improving safety and performance
ST-18 18 t
Haulage truck MT-42 42 t To be automated in the near future
MT-65 65 t
Drill PV-235 12 m hole depth Hole diameter 171–270 mm; automated rotary blasthole drill
PV-271 19.2 m
PV-275 59.4 m
PV-311 19.8 m Hole diameter 270–311 mm; automated rotary blasthole drill
PV-316 91.4 m
PV-351 19.8 m Hole diameter 207–406 mm; automated rotary blasthole drill
Simba S7 51–89 mm hole diameter High-precision, automated, longhole drill
Simba M4 51–178 mm Automated hydraulic production drill rig
Simba E6-W 89–254 mm Automated longhole production drill for large size drift
Boomer S2 43–64 mm Automated, two-boom–face drilling rig
Komatsu Limited Haulage truck 830E-1AC 245 tons (~222 t) Autonomous haulage vehicle design by Komatsu
(2018) 930E-4SE 320 tons (~290 t) Automation system to be designed by Caterpillar
Sandvik Group LHD LH-410 10 t Mass mining loaders; AutoMine onboard package; AutoMine
(2018) LH-514 14 t loading readiness
LH-517 17 t
LH-621 21 t
Haulage truck TH-551i 51 t Articulated underground dump trucks; AutoMine trucking onboard
TH-663i 63 t option allows autonomous haulage for transfer level and decline
ramp applications
Drill DR-412i 216–311 mm Rotary blasthole surface drilling rig
DU-412i 89–216 mm Articulated in-the-hole longhole production drill
DD-422i 43–64 mm Development drill rig
DT-1121i 43–64 mm Tunnelling jumbo drill; two-boom
DT-1131i 45–64 mm Tunnelling jumbo drill; three-boom; 183 m2 coverage
DT-1231i 45–64 mm Tunnelling jumbo drill; three-boom; 211 m2 coverage
DT-1331i 45–64 mm Tunnelling jumbo drill; three-boom; 232 m2 coverage

CIM Journal | Vol. 10, No. 3, 2019

122 K. Moreau, R. Bose, H. Shang, and J. A. Scott
Automated haulage trucks
From September 2015 to January 2017, Fortescue
Metals Group Ltd’s 54 autonomous trucks safely hauled
240 million tons (~218 million tonnes) of ore at their
Solomon Hub iron ore mine (Mount Sheila, Australia)—
a 20% increase in productivity compared to their manual
operations (Brown, 2017). There are similar reports from
other mining companies using autonomous vehicles,
including Rio Tinto and BHP Billiton. In 2008, Rio Tinto
began trialling Komatsu Limited’s autonomous haulage
system (AHS) technology at five sites across the Pilbara
region of Western Australia (Mining Journal, 2018); by
2017, the autonomous fleet had expanded to 80 haulage
trucks and it is expected to grow to more than 140 by
2020 (Mining Journal, 2018). Rio Tinto stated that, in
2017, each autonomous truck operated an average of 700
more hours than manually operated trucks, and unit costs
were lowered by approximately 15% (Mining Journal,
2018).
Automated drilling operations
A 10–20% increase in productivity reported at the
Malmberget mine is the result of automated Caterpillar and
Epiroc drill rigs (Caterpillar Global Mining, 2008a; Atlas
Copco, 2013). Caterpillar deployed five electric-powered,
automated production rigs that each drill 10,000 m of
11.5 cm diameter hole per month, with an average depth of
50 m per hole (Caterpillar Global Mining, 2008a). In 2013,
the mine introduced six of Epiroc’s Simba WL6 C rigs for
fully automated upward drilling (Atlas Copco, 2013). They
drilled 115 mm diameter holes with depths of 30–47 m and
it is expected that the new automated fleet will lead to a
20% increase in productivity (Atlas Copco, 2013).
BHP Billiton has been using Epiroc’s rig control system
within their Yandi mine (Pilbara, Australia) since 2015. As
of November 2017, a Pit Viper 271 drill had operated
autonomously for 15,000 h and drilled more than 1 mil-
lion metres (Turner, 2017). The mine reported a 20%
increase in the use of their drilling operations due to a 16%
speed increase per drill cycle, as well as improved devia-
tion metrics through drilling, levelling, tramming, and de-
levelling operations (Turner, 2017). The rigs were also
capable of working 11.5 h of a 12-h shift compared to 8.5 h
with a manually operated drill (Turner, 2017).
CHALLENGES FOR AUTOMATED MINING
There are many benefits associated with the introduction
of automated vehicles and equipment within the under-
ground workplace, but there are challenges to overcome to
optimize autonomous operations and reach the full poten-
tial of the technology. The condition of haulage roads is one
of the most important factors associated with production
(Caterpillar Global Mining, 2008d). An operator from a
Canadian underground mine site claimed that when road-
ways were not maintained, the autonomous vehicle was not
CIM Journal | Vol. 10, No. 3, 2019
able to travel at peak velocity; therefore, the vehicle was
unable to realize the projected reduced cycle times. They
acknowledged that because the operator is located on the
surface, there is an absence of work personnel underground
to perform roadway maintenance and that work procedures
need to be developed to address this issue (Hudbay Miner-
als operator, personal communication, 2017).
With fewer personnel underground and the equipment
operator located on the surface, a new challenge is equip-
ment maintenance, which requires additional maintenance
personnel with specialized technical expertise (Gustafson
et al., 2013). The reliability of mining equipment has his-
torically been low and the time between equipment failures
will need to be increased to accomplish desired production
increases (Schunnesson et al., 2009). An important compo-
nent of developing automated technology will, therefore,
be to reduce the frequency of scheduled maintenance. This
may be aided by less wear and tear through use of automa-
tion, but scheduling corrective maintenance appropriately
will be vital to limit equipment downtime and maintain
high production rates (Gustafson et al., 2013; Paraszczak et
al., 2015).
Infrastructure and layout can affect the performance of
automated equipment in the underground mine environ-
ment, but the required infrastructure may not coincide well
with a mine’s existing infrastructure and layout (Ghodratia
et al., 2015). It is also important to design the autonomous
work area properly (Ghodratia et al., 2015) to isolate
autonomous operations from other areas of the mine, with-
out hindering the productivity of operations that have yet to
be automated (Gustafson, 2013).
CONCLUSIONS
From improved use of equipment to the ability to oper-
ate between shifts, mine sites have the opportunity to
increase machine productivity with near–real-time data
reporting through mining automation; however, to realize
the maximum potential of autonomous technology, the
challenges of mine design, maintenance planning, and
scheduled roadway maintenance will need to be addressed
through innovation by mine engineers.
There is potential to conduct further research to quantify
the environmental impact of autonomous operations in
underground mines. Past automation trials have demon-
strated promising higher efficiencies, lower fuel consump-
tion, and lower energy consumption due to decreased
ventilation demands. An improved impact on the environ-
ment will be an important justification for the transition to
autonomous vehicles and the sustainability of the mining
industry.
ACKNOWLEDGMENTS
The authors thank Sudbury Integrated Nickel Opera-
tions, Mitacs, Laurentian University, and the Goodman
School of Mines for their continued support of our
Automation technology to increase productivity and reduce energy consumption in deep underground mining operations 123
research. The authors also thank the Canadian Institute of
Mining, Metallurgy and Petroleum and their volunteer peer
reviewers for their time and contributions to the publication
of this paper.
Paper reviewed and approved for publication by the
Maintenance, Engineering and Reliability Society of CIM.
Kyle Moreau graduated from Western University, Canada, with a Bachelor
of Engineering Science (BESc) in Chemical Engineering and is currently
pursuing a PhD at Laurentian University, Canada, in the Bharti School of
Engineering under the supervision of Dr. Scott and Dr. Shang. He is a part of
the ONGEN research group and is working on automated deep mining.
REFERENCES
Atlas Copco. (2013a). Autonomy in the Andes - Chile’s copper giant
Codelco puts Scooptram ST14 to the test. Mining & Construction, 2013
(2), 34–35.
Atlas Copco (2013b, August 20). Northern Star shines bright. Mining &
Construction. Retrieved from https://miningandconstruction.com/min-
ing/northern-star-shines-bright-2524/
Atlas Copco (2015a). Network system - scooptram automation. Unpub-
lished internal report.
Atlas Copco (2015b). Safety - scooptram automation. Unpublished
internal report.
Atlas Copco (2017). Scooptram automation - communication require-
ments. Unpublished internal report.
Bellamy, D., & Pravica, L. (2011). Assessing the impact of driverless
haul trucks in Australian surface mining. Resources Policy, 36(2), 149–
158. https://doi.org/10.1016/j.resourpol.2010.09.002
Braul, P. (2018, April 23). Charging ahead, CIM Magazine, 13(2), 40–
43.
Brown, B. (2017). Autonomous mining is coming to the U.S. Fox Busi-
ness. Retrieved from https://www.foxbusiness.com/features/autonomous
-mining-is-coming-to-the-u-s
Bugarksi, A., Schnakenberg, G., Noll, J., Mischler, S., Crum, M., &
Anderson, R. (2004). Evaluation of diesel particulate filter systems and
biodiesel blends in an underground. Society for Mining, Metallurgy and
Exploration, 318, 207–218.
Campbell, H., Seeber, C., & Wywrot, J. (2003). Ventilation Part 1.
Canadian Mining Journal, 124(5), 24. Retrieved from www.canadian-
miningjournal.com/features/ventilation-part-1
Caterpillar Global Mining. (2008a). Automation keeping underground
workers safe at LKAB Malmberget Mine. Viewpoint: Perspectives on
Global Mining, 3, 14–21.
Caterpillar Global Mining. (2008b). Building the technologies for the
mine sites of the future. Viewpoint: Perspectives on Modern Mining, 4,
38–44.
Caterpillar Global Mining. (2008c). Improving productivity through
technology integration. Viewpoint: Perspectives on Modern Mining, 3,
2–7.
Caterpillar Global Mining. (2008d). Mining fleet achieves 100,000
hours and beyond. Viewpoint: Perspectives on Modern Mining, 4, 2–9.
Caterpillar Global Mining. (2015). Change management makes the dif-
ference. Viewpoint: Perspectives on Modern Mining, 11, 18–23.
CIM Journal | Vol. 10, No. 3, 2019
Ron Bose is a professional engineer and a certified energy manager. As a
member of the innovation and technology team at Glencore’s Sudbury
Integrated Nickel Operations, he implements initiatives to improve
productivity through better resource management and nurturing the creation
of information from data. His mantra is People ← Process ← Technology.
Helen Shang completed her PhD in the department of chemical engineering,
University of Alberta, Canada, where she carried out research on advanced
control methods. Since 2002, she has been a professor, teaching process
control and other engineering courses at Laurentian University, where she
also conducts significant industrial-based research in applied process
modelling and control.
John A. Scott is a professor and chemical engineer at Laurentian University.
He is head of the ONGEN (ongen.ca) research group and leads a wide range
of R&D projects centred on industrial economic sustainability through
assessment and introduction of new practices, processes, and technologies.
jascott@laurentian.ca
Caterpillar Inc. (2018a). Command for hauling. Retrieved from
https://www.cat.com/en_US/by-industry/mining/surface-mining/sur-
face-technology/command/command-for-hauling.html
Caterpillar Inc. (2018b). Underground - hard rock equipment. Retrieved
from https://www.cat.com/en_US/products/new/equipment/under-
ground-hard-rock.html
Chadwick, J. (2005). Rise of the machines. Mining Magazine, 15–17.
Chadwick, J. (2010). The autonomous mine. International Mining, 6,
46–57.
Dragt, B. J., Camisani-Calzolari, F. R., & Craig, I. K. (2005). An
overview of the automation of load-haul-dump vehicles in an under-
ground mining environment. IFAC Proceedings Volumes, 38(1), 37–48.
https://doi.org/10.3182/20050703-6-CZ-1902.01389
Duddu, P. (2013, September 11). The top ten deepest mines in the world.
Mining Technology. Retrieved from https://www.mining-technology
.com/features/feature-top-ten-deepest-mines-world-south-africa/
Dyson, N. (2017, July 10). BHP to double autonomous trucks at Jimble-
bar. Mining Magazine. Retrieved from https://www.miningmagazine.com
/innovation/news/1331400/bhp-to-double-autonomous-trucks-at-jimblebar
Epiroc. (2018). Mining and construction products and equipment.
Retrieved from https://www.epiroc.com/en-ca/products
Ewing, I. (2016, July 22). The electric underground. CIM Magazine.
Retrieved from http://magazine.cim.org/en/projects/the-electric-under-
ground/
Ghodratia, B., Hoseiniea, S. H., & Garmabakia, A. (2015). Reliability
considerations in automated mining systems. International Journal of
Mining, Reclamation and Environment, 29(5), 404–418.
Glencore. (2016). Sudbury operations - overview. Presentation to
Greater City of Sudbury. Retrieved from https://agendasonline.greater-
sudbury.ca/index.cfm?pg=feed&action=file&attachm
ent=17056.pdf
Gustafson, A. (2011). Automation of load haul dump machines
(Research report). Luleå University of Technology, Luleå, Sweden.
Gustafson, A. (2013). Automation of load haul dump machines - com-
parative performance analysis and maintenance analysis (Doctoral dis-
sertation). Luleå University of Technology, Luleå, Sweden.
Gustafson, A., Schunnesson, H., & Kumar, U. (2013). Reliability analy-
sis and comparison between automatic and manual load haul dump
machines. Quality and Reliability Engineering International, 31(3),
523–531. https://doi.org/10.1002/qre.1610
124 K. Moreau, R. Bose, H. Shang, and J. A. Scott
Hard-Line Solutions. (2018). Company Overview. Retrieved from
http://www.hard-line.com/company/
Holmberg, K., Kivikytö-Reponen, P., Härkisaari, P., Valtonen, K., &
Erdemir, A. (2017). Global energy consumption due to friction and wear
in the mining industry. Tribology International, 115(November), 116–
139. https://doi.org/10.1016/j.triboint.2017.05.010
Jacobs, W. (2013). Electric LHDs in underground hard rock mining: A
cost/benefit analysis (Final year project thesis). Univeristy of Western
Australia, Perth, Australia.
Jacobs, W., Hodkiewicz, M. R., & Bräunl, T. (2015). A cost-benefit
analysis of electric loaders to reduce diesel emissions in underground
hard rock mines. IEEE Transactions on Industry Applications, 51(3),
2565–2573. https://doi.org/10.1109/TIA.2014.2372046
Jaderblom, N. (2017). From diesel to battery power in underground
mines: A pilot study of diesel free LHDs (Degree project). Luleå Univer-
sity of Technology, Luleå, Sweden.
Kelly, L. (2017, May 23). Innovation upping safety game underground
- Northern mines using technology to keep workers safe. Northern
Ontario Business. Retrieved from https://www.northernontariobusi-
ness.com/industry-news/mining/innovation-upping-safety-game-under-
ground-618691
Kirkland Lake Gold Inc. (2015). Battery powered mobile equipment.
Unpublished internal report.
Komatsu Limited. (2018). Construction Mining and Utility Equip-
ment. Retrieved from https://home.komatsu/en/products/construction-
machine/#anc01
Larsson, J. (2011). Unmanned operation of load-haul-dump vehicles in
mining environments (Doctoral dissertation). Örebro University, Öre-
bro, Sweden.
Larsson, J., Appelgren, J., & Marshall, J. (n.d.). Next Generation Sys-
tem for Unmanned LHD Operation in Underground Mines. Örebro
Sweden: Atlas Copco Rock Drills AB. & Brampton, ON: MDA Inc.
Majewski, W. A. (2018). DieselNet Technology Guide. Retrieved from
https://www.dieselnet.com/tech/dpm.php
Mäkelä, H. (2001a). Outdoor navigation of mobile robots (Doctoral dis-
sertation). Helsinki University of Technology, Espoo, Finland.
Mäkelä, H. (2001b). Overview of LHD navigation without artificial
beacons. Robotics and Autonomous Systems, 36(1), 21–35.
https://doi.org/10.1016/S0921-8890(01)00115-4
McHugh, C. (2004). Introduction of autonomous loaders to Olympic
Dam operations, Australia. Paper presented at the 4th Int. Conference
and Exhibition on Mass. Santiago, Chile.
McNab, K., & Garcia-Vasquez, M. (2011). Autonomous and remote
operation technologies in Australian mining. Centre for Social Respon-
sibility in Mining (CSRM) - Sustainable Minerals Institute, University
of Queensland, Brisbane City, Australia.
Mining Journal (2018, January 29). Rio’s autonomous fleet hauls 1 bil-
lion tonnes. Mining Journal. Retrieved from https://www.mining-jour-
nal.com/investment/news/1311213/rios-autonomous-fleet-hauls-billion
-tonnes
Paraszczak, J. (2014). Maximization of productivity of autonomous
equipment in underground mines. Mining Engineering, 66(6), 24–41.
Paraszczak, J., Gustafson, A., & Schunnesson, H. (2015). Technical and
operational aspects of autonomous LHD application in metal mines.
International Journal of Mining, Reclamation and Environment, 29(5),
391–403.
©2019 Canadian Institute of Mining, Metallurgy and Petroleum. All rights reserved.
Paraszczak, J., Laflamme, M., & Fytas, K. (2013). Electric load-haul-
dump machines: real alternative for diesels? Canadian Institute of Min-
ing, Metallurgy and Petroleum, 4(1), 13–19.
Poole, R. (1999). Load-haul-dump machine automation at Inco’s
Ontario division (Master’s thesis). McGill University, Montreal, Canada.
Roberts, J. M., Duff, E. S., Corke, P. I., Sikka, P., Winstanley, G. J., &
Cunningham, J. (2000). Autonomous control of underground mining
vehicles using reactive navigation. In Proceedings of the IEEE Interna-
tional Conference on Robotics & Automation, , San Fransico, CA, Vol-
ume 4, 3790–3795. https://doi.org/10.1109/ROBOT.2000.845322
Rolfe, K. (2018, March 29). Agnico Eagles gets LTE underground at
LaRonde. CIM Magazine. Retrieved from http://magazine.cim.org/en/
news/2018/agnico-eagle-gets-lte-underground-at-laronde/
Sandvik Group. (2018). Mining and Rock Excavation - Products and
Equipment. Retrieved from https://www.rocktechnology.sandvik
/en/products/
Sandvik Mining. (2016, September 26). Sandvik enables automation
that mines are ready for. Mining.com. Retrieved from http://www.min-
ing.com/web/sandvik-enables-automation-that-mines-are-ready-for/
Sandvik Mining. (2017, July 7). Sandvik intelligent trucks automate
haulage with AutoMine® [News release]. Retrieved from
https://www.home.sandvik/en/news-and-
media/newslist/news/2017/07/sandvik-intelligent-trucks-automate-
haulage-with-automine/
Sandvik Mining & Construction. (2017). AutoMine loading - lite system
proposal. Unpublished internal report.
Schunnesson, H., Gustafson, A., & Kumar, U. (2009). Performance of
automated LHD: A review. Luleå University of Technology, Luleå,
Sweden.
The Redpath Group. (2012, June 7). Keeping up with the caving.
Retrieved from http://www.redpathmining.com/wp-content/uploads
/2012/07/Block-Caving_Mining-Magazine-June-2012.pdf
Tollinsky, N. (2018, February 21). Glencore, Vale approve spending for
Sudbury. Sudbury Mining Solutions Journal. Retrieved from
http://www.sudburyminingsolutions.com/glencore-vale-approve-spend-
ing-for-sudbury.html
Turner, J. (2017, November 28). Deep impact: Atlas Copco takes auto-
mated drilling to the next level. Mining Technology. Retrieved from
https://www.mining-technology.com/features/deep-impact-atlas-copco-
takes-automated-drilling-next-level/
United States Environmental Protection Agency (2018). Energy and the
Environment. Retrieved from https://www.epa.gov/energy/greenhouse-
gases-equivalencies-calculator-calculations-and-references
Validakis, V. (2012, October 26). Biodiesel and its future in mining.
Australian Mining. Retrieved from https://www.australianmining.com.au
/features/biodiesel-and-its-future-in-mining/
Vergne, J. D. (2008). Hard rock miner’s handbook. Edmonton, Canada:
Stantec Consulting Ltd.
White, E. (2018, April 5). ‘Deep will be the new norm’ — Glencore
spends $1B to find new ore beneath Sudbury. CBC News. Retrieved
from https://www.cbc.ca/news/canada/sudbury/onaping-depth-sudbury-
mining-1.4604208
World Health Organization. (2012). Diesel Engine Exhaust Carcino-
genic. Lyon, France: Internation Agency for Research on Cancer.
World Health Organization. (2014). Diesel and gasoline engine
exhausts and some nitroarenes. Lyon, France: International Agency for
Research on Cancer.
 MAINTENANCE, ENGINEERING AND RELIABILITY 125

Simulating the effect of LHD operations on production rates

and ventilation costs in a sublevel cave underground mine
B. Skawina
Department of Civil, Environmental and Natural Resources Engineering,
Division of Mining and Geotechnical Engineering, Luleå University of Technology, Luleå, Sweden

A. Salama
Department of Chemical and Mining Engineering, College of Engineering and Technology,
University of Dar es Salaam, Dar es Salaam, Tanzania

J. Greberg
Department of Civil, Environmental and Natural Resources Engineering,
Division of Mining and Geotechnical Engineering, Luleå University of Technology, Luleå, Sweden

https://doi.org/10.15834/cimj.2019.12

ABSTRACT Load-haul-dump machines (LHDs) are typically used in underground metal mining opera-
tions. Delays or inappropriate use of LHDs can result in production loss. Optimized LHD use is especially
crucial in larger mines because longer travel distances increase heat, dust, and gas emissions, which in turn
increase ventilation costs. This study, conducted at an existing Swedish sublevel cave underground mine,
used discrete event simulation and the AutoMod™ DES tool to determine the ventilation costs related to
using too many diesel LHDs in a production area with reduced ore pass availability. When fewer ore
passes are available, ventilation costs related to operating LHDs in the production area were found to
increase by as much as 200, 224, and 306% for one, three, and six LHDs in operation, respectively.
■ KEYWORDS Discrete event simulation (DES), Loading operations, Production rate, Ventilation costs

RÉSUMÉ Dans les exploitations souterraines d’extraction des métaux, on utilise généralement des char-
geurs-transporteurs (LHD, de l’anglais Load-haul-dump machines). Les retards ou l’utilisation abusive
des LHD peuvent entraîner une perte de production. L’utilisation optimisée des LHD est particulièrement
importante dans les grandes mines en raison des longues distances parcourues, lesquelles augmentent la
chaleur, la poussière et les émissions de gaz qui, à leur tour, décuplent les coûts de l’aérage. Cette étude,
menée en Suède dans une mine souterraine utilisant la méthode d’exploitation en sous étages avec fou-
droyage, s’est basée sur la simulation à événements discrets (SED) et l’outil de SED AutoMod™ pour
déterminer les coûts de l’aérage liés à l’utilisation d’un nombre trop élevé de LHD alimentés au diesel
dans la zone de production, avec une disponibilité réduite de la cheminée à minerai. Lorsque les mines
disposent de moins de cheminées à minerai, l’étude montre que les coûts de l’aérage liés à l’exploitation
des LHD dans la zone de production augmentent respectivement de 200, 224 et 306 % pour un, trois et
six LHD en fonctionnement.
■ MOTS CLÉS coûts de l’aérage, opérations de chargement, simulation à événements discrets (SED), taux
de production

INTRODUCTION
Transporting material from the mine face to the dumping
location requires flexible loading and hauling equipment
that is able to cope with sharp curves and a complex mining
environment. Load-haul-dump machines (LHDs) are
designed for such an environment. Models are available to
suit specific features of the working environment such as
small, large, or narrow open stopes (Atlas Copco Rock
Drills AB, 2007). When carrying a full load, haulage speed
is higher for LHDs than for front-end loaders with the same
engine size. Also, payloads are approximately 50% higher
for LHDs than for front-end loaders (Atlas Copco Rock
Drills AB, 2007). LHDs are capable of climbing steep

CIM Journal | Vol. 10, No. 3, 2019

126 B. Skawina, A. Salama, and J. Greberg
grades effectively at up to 20% and ideally at up to 14%
(Bloss, Harvey, Grant, & Routley, 2011). The haulage dis-
tance between stope and ore pass is usually between 50 m
and 400 m. LHDs are efficient for tramming distances up to
200 m (Bloss et al., 2011), whereas when loading trucks,
the LHDs work efficiently when the distance is less than
100 m (Atlas Copco Rock Drills AB, 2007). The largest
diesel-powered LHDs, with drive motors of up to 350 kW,
can handle a payload of 21 t (Wennmohs, 2014). Most min-
ing equipment runs on diesel fuel, which provides flexibil-
ity, high durability, and high efficiency (Pronk, Coble, &
Stewart, 2009); however, other alternatives such as electric,
hybrid diesel/electric, or hydrogen fuel cell machines are
increasingly being developed and used. This change is due
to diesel-powered machines accounting for a significant
portion of the ventilation costs in the mine (Chadwick,
2008; Paraszczak, Laflamme, & Fytas, 2013). Currently, as
part of the European sustainable intelligent mining systems
(SIMS) project, Epiroc AB is planning to demonstrate tech-
nology for a diesel-free underground mine at Agnico
Eagle’s Kittlä mine in Finland using battery-powered
mobile equipment, with the goal of achieving the same pro-
duction as could be realized using a fleet of similar diesel-
powered equipment in the same environmental conditions
(SIMS, 2018).
The purpose of this study, based on operations at an exist-
ing Swedish mine, is to determine the changes in production
rate and ventilation costs related to the use of additional diesel
LHDs when an ore pass is not available. The AutoMod™ dis-
crete event simulation (DES) tool is used to simulate the
underground loading operation at one of the production lev-
els. The ventilation costs related to operating the LHDs in the
studied production area are calculated based on the technical
and operating parameters of a 21 t LHD (motor power, oper-
ating times, production rates), the ventilation system of the
studied production area (flow rate capacity, ventilation
requirements), and electricity costs.
Ventilation in underground mines
Ventilation accounts for a large part of the power costs
for a mine. According to de la Vergne (2003), ventilation is
responsible for one-third of the power costs for a mine;
Paraszczak et al. (2013) report that the costs are higher (up
to 40%). The ventilation requirements (diesel exhaust and
noxious gases) in Sweden include regulations for nitrogen
dioxide (NO2) and carbon monoxide (CO) content (Work-
place exposure to dusts and aerosols – diesel exhaust,
2017). In this study, the ventilation requirements and costs
related to LHDs are analyzed. When many diesel machines
are in operation, high amounts of diesel exhaust are gener-
ated (Pronk et al., 2009). If the working areas are not ven-
tilated properly, workers will be at risk of exposure to
noxious gases and carcinogenic agents (Pronk et al., 2009).
There is a need to provide oxygen to personnel and for
combustion, but also to provide the necessary volume of air
CIM Journal | Vol. 10, No. 3, 2019
to reduce fumes from diesel engines, remove diesel partic-
ulate matter (DPM) and blasting fumes from the air, regu-
late heat, and improve visibility (Tuck, 2011). One solution
is to replace older mining equipment with newer equipment
that emit less noxious gases (thus improving health and
safety for the personnel) and produce less DPM, dust, and
heat, while at the same time reducing energy consumption.
An alternative is to use electric vehicles in the system or
create an entry zone where personnel access is restricted to
limit human exposure to poor-quality air.
Discrete event simulation
Discrete event simulation (DES) is usually applied to
dynamic, complex systems, and many recent studies of
mining operations have used this technique. Yuriy, Runci-
man, and Vayenas (2013) presented and discussed the Vale
Canada experience modelling the functionality, capability,
and user friendliness of underground mining operations
using the WITNESS, Quest, SimMine®, ProModel®, and
AutoMod simulation tools. Bailey, Olsson, and Glassock
(2012) used Arena® software to model the face handling
system at Olympic Dam, Australia. Greberg and Sundqvist
(2011) simulated the mine planning process at the Newcrest
Cadia East underground operation in Australia to optimize
the development of panels. Usmani, Szymanski, and Apel
(2014) used a simulation to optimize the extraction level of
a block cave mine. Greberg, Salama, Gustafson, and Skaw-
ina (2016) used the AutoMod simulation model to analyze
the costs of using haul trucks to transport ore and waste up
the ramp to the existing crusher.
A detailed, accurate simulation model for large, complex
systems requires a large dataset, statistical distributions of
the data, and carefully chosen simulation software. In this
study, the simulation tool AutoMod (Applied Materials,
2019) was used to model the studied production area. The
AutoMod tool is based on DES and was used because it has
the necessary capabilities and flexibility to model various
custom-made transportation systems and has built-in
debugging and tracing features that make the verification
and validation process easier. A verified and validated sim-
ulation model provides results similar to those seen in an
actual operating system.
CASE STUDY
This study is based on data and information obtained
from an underground mine located in Sweden. The mine
uses sublevel caving methods to extract iron ore. The ore
reserves are distributed over 20 large and small orebodies,
with 14 orebodies currently in production. Depending on
the type of orebody, the interval between the sublevels
varies from 20 to 30 m, with drift spacing of 22.5 m. The
orebodies are divided into blocks and then further divided
into production areas spread over a 5 km long and 2.5 km
wide area, each with its own group of ore passes. The ore
passes are approximately 300 m long, have a diameter of
Simulating the effect of LHD operations on production rates and ventilation costs in a sublevel cave underground mine
approximately 3 m and dip approximately 60°. The drifts
are approximately 5 m high and 6.5 m wide. The sequenc-
ing of drifts to be extracted is based on rock stress–related
factors, the ancillary activities of the production drifts, dis-
tances to the ore passes (depending on the production rate
requirement), and the amount of ore left in each of the pro-
duction drifts. The number of ore passes in the production
areas varies between one and four depending on the size of
the production area. Figure 1 presents the orebody outlines,
ore passes, and production drifts used for hauling the ore
from the face to one of the ore pass locations.
The current loading practice is based on the use of man-
ually operated diesel LHDs with a 21 t bucket capacity. In
smaller production areas, there is usually one LHD in oper-
ation. In larger production areas, two LHDs can be used
simultaneously. The ore is transported from the draw points
to the ore passes. From the ore passes, the ore is collected
on the main haulage level by trucks and then hauled to the
crusher, where the rock is fragmented and further trans-
ported via conveyor to the hoisting system and to the sur-
face.
The ventilation system in the mine is automated and
consists of 4 inflow and 12 outflow ventilation shafts. The
studied production area is located in the eastern part of the
mine at level 996 (Figure 2). The inflow ventilation shafts
cover different regions and the evacuation lifts are located
behind these shafts. Outflow shafts cover the old mine
workings and some of them have been sealed with concrete
and rock to prevent airflow leakage. In the winter, the air is
heated with an oil heater to a temperature of approximately
+1°C. The first set of fans pushes the air through the venti-
lation shaft to level 800. From level 800, a second set of
fans push the air down to level 1,250. In the central region
on level 800, two 2 MW fans operating at a pressure of
1,500 Pa distribute the air to other parts of the mine.
Figure 1. Studied production area at an underground mine in Sweden (plan view)
CIM Journal | Vol. 10, No. 3, 2019
127
Simulation settings for the predefined production
area
The production area shown in Figure 1 was modelled
using AutoMod simulation software. In the simulation, the
LHDs were ordered to travel to one of the production drifts
and then assigned to the closest available ore pass. If all ore
passes were unavailable, the LHD operators waited in the
production drift until an ore pass was available, which
affected the LHD cycle time. Each simulation ended when
there was no longer any material left to be loaded, that is,
when the production level was mined out. It was assumed
that there was enough blasted material at the face to load
the LHDs at all times. The LHDs were scheduled to operate
in the simulated production area for 15 h/day. During oper-
ations, the LHDs encountered delays, thus operating time
was reduced, whenever production disturbances or vehicle
breakdowns took place. Whenever a vehicle broke down, it
stopped at the location of the breakdown. Work was per-
formed seven days a week and during breaks, the LHDs
were sent to the closest parking space.
Input data
For the model, the total theoretical volume of ore avail-
able in the studied production area was 17.43 Mt. The LHD
performance data were collected from the mine on several
occasions via video recordings, documentation, and time
studies. These data consisted of time spent by the LHD at
the face (13–429 s), time spent by the LHD at the ore pass
(10–12 s), travelling speed (an average of 17.4 km/h),
bucket weights (19.32 ± 3.4 t), and turning times (6–8 s).
The data variations were fitted to probability distributions
using the statistical tool EasyFit® (MathWave, 2019) to
characterize the uncertainty and randomness of the opera-
tion. It was assumed that LHDs would be available to oper-
ate 90% of the total time and production disturbances were
128 B. Skawina, A. Salama, and J. Greberg

Level
Legend:
Hoist
Exhaust system
Fresh air intake
Studied production area
Main level
Regional border
Exhaust fans
Exhaust fan station
Secondary fans
Primary fan station
Rock pile

Figure 2. Layout of the mine ventilation system in the region of the studied production area

estimated to account for 20% of the total time. The 21 t because a larger number of machines would result in exten-
diesel LHD used in the study has a drive power of 350 kW sive waiting times and would be highly unpractical. For all
and a vehicle weight of 56.8 t. Based on the information scenarios, LHD production rates and operating times were
obtained from the mine, the ventilation in the studied pro- obtained from simulation analysis. The typical resistance
duction area has a flow-rate capacity of 250 m3/s, consum- was calculated based on the ventilation capacity of the stud-
ing 2,661 kW of power at a typical resistance of ied production area. The ventilation costs related to operat-
0.1 Ns2/m8. The ventilation requirements of the studied
production area are set at 0.05 m3/s/kW. The assumed elec- Table 1. Number and location of operational ore passes for 15 simulated
tricity cost was set to US$0.12/kWh (United States Energy operation scenarios in the studied production area; FL: far-left ore pass,
CL: centre-left ore pass, CR: centre-right ore pass, FR: far-right ore pass
Information Administration, 2017).
Scenario No. of Ore pass
operational operational states
Scenarios ore passes (1: operating, 0: not operating)
Fifteen combinations of ore pass availability and loca- FL CL CR FR
tion were analyzed and are termed scenarios in this study 1 4 1 1 1 1
(Table 1). For each scenario, the number of operational ore 2 3 1 1 1 0
passes varied between one and four. In Table 1, FL stands 3 3 1 1 0 1
4 3 1 0 1 1
for the ore pass located in the far-left side of the studied
5 3 0 1 1 1
production area shown in Figure 1, CL stands for the ore
6 2 1 1 0 0
passes located in the centre-left position, and CR and FR 7 2 0 1 1 0
stand for centre-right and far-right, respectively. Each sce- 8 2 0 0 1 1
nario was tested with one, three, and six operating LHDs 9 2 1 0 0 1
with the aim of determining the incurred ventilation costs 10 2 1 0 1 0
due to longer haul times resulting from ore pass unavail- 11 2 0 1 0 1
12 1 1 0 0 0
ability when additional LHDs are used. Varying the number
13 1 0 1 0 0
of LHDs was done to observe the variations in energy and
14 1 0 0 1 0
ventilation costs in the current infrastructure in case of ore 15 1 0 0 0 1
pass unavailability. A maximum of six LHDs was modelled

CIM Journal | Vol. 10, No. 3, 2019

Simulating the effect of LHD operations on production rates and ventilation costs in a sublevel cave underground mine
ing the LHDs in the studied production area were obtained
by calculating the airflow rate for loaders, then converting
the flow rate into an air power and further multiplying the
air power by operating times and electricity costs.
Model verification and validation
Verification and validation tests of the model are neces-
sary to increase the level of confidence, credibility, and
probability of model correctness (Sargent, 2011). Verifica-
tion tests ensure that a conceptual model is accurately trans-
lated into a simulation model, whereas the validation tests
ensure that the model reflects the real system to a
given/specified level of detail (Sargent, 2011).
The results of the simulations run in this study were
tested for correctness and credibility. Simulation model con-
sistency with the conceptual model (verification) was
ensured by having the specialists and experts involved in the
study test the model behaviour and use a debugging feature
called an internal run controller. Specialists also assisted
with the comparison of the scenarios to the real system (val-
idation). Validation was ensured by comparing the produc-
tion data from the operating mine with the results obtained
from the simulation. In addition to validation, extreme con-
dition tests, degenerate tests, traces, and internal validity
were used to test the credibility of the model. Internal valid-
ity was tested by running an additional five replications for
each set of scenarios in each production area to ensure that
the model’s results did not significantly differ.
RESULTS AND DISCUSSION
The simulation was conducted for 15 scenarios, each
with a different number (1–4) and arrangement of opera-
tional ore passes. For each scenario, runs were completed
with one, three, and six LHDs in operation (one run for
each number of LHDs for a
total of three runs per sce-
nario). The production rate and
ventilation costs for each run
are presented in Figure 3. The
ventilation costs calculated in
this study are intended for
comparative use only and do
not represent realistic cost val-
ues. The production rate is
defined as total tonnes loaded
divided by the total number of
days required to finish produc-
tion.
The production rate varied
depending on the scenario,
with a general trend towards a
lower rate as the number of
available ore passes decreased. Figure 3. Production rate (PR) and ventilation costs (VC) for the studied production area during scenarios
For each scenario, an increase with varying ore pass availability and location and one, three, or six operating load-haul-dump (LHD)
in the number of LHDs in machines
CIM Journal | Vol. 10, No. 3, 2019
129
operation resulted in an increase in the production rate. The
highest production rate, which occurred when six LHDs
were in operation and all ore passes were available (sce-
nario 1), was 20,350 t/day, whereas the lowest production
rate, 2,280 t/day, occurred when only one LHD was in oper-
ation (scenario 15). In areas affected by an ore pass avail-
ability issue, the production rate decreased by as much as
2,259, 6,920, and 13,642 t/day for one, three, and six LHDs
in operation, respectively. The ventilation cost increased as
the number of LHDs in operation increased. For one and
three LHDs in operation, the ventilation cost averaged
US$0.12/kt/LHD and US$0.43/kt/LHD, respectively. The
difference between the highest and lowest ventilation costs
across all 15 scenarios was US$0.09/kt/LHD with one LHD
operating and US$0.36/kt/LHD with three LHDs operating.
In cases where six LHDs were used, the ventilation cost
increased to an average of US$1.21/kt/LHD, and the range
across scenarios was US$1.47/kt/LHD. Overall, when less
ore passes are operational and more LHDs are in use, ven-
tilation costs rise by an average of US$1.09/kt/LHD.
In Table 2, the variation in the production rate and ven-
tilation costs are shown as percentages, where a production
rate of 100% and a ventilation cost of 100% are defined as
the values recorded for scenario 1 (four ore passes in oper-
ation). All other percentages are calculated as (scenario X /
scenario 1) × 100, where scenario X is one of scenarios
2–15.
Based on the results in Table 2, the production rate
decreases by as much as 50, 55, and 67% for one, three, and
six LHDs in operation, respectively, when ore pass avail-
ability is reduced. The results also show that the ventilation
cost increases by up to 200, 224, and 306% for one, three,
and six LHDs in operation, respectively, when fewer ore
passes are available.
130 B. Skawina, A. Salama, and J. Greberg

Table 2. Production rates and ventilation costs for one, three, or six load-haul-dump (LHD) machines operating during 15 simulation scenarios with varied ore
pass availability, expressed as a percentage of the values recorded during scenario 1.
Scenario 1 LHD 3 LHDs 6 LHDs
Production rate Ventilation cost Production rate Ventilation cost Production rate Ventilation cost
(%) (%) (%) (%) (%) (%)
Four ore passes in operation
1 100 100 100 100 100 100
Three ore passes in operation
2 93 93 108 87 116 93
3 91 109 93 106 80 123
4 92 109 92 108 81 124
5 85 117 82 121 82 122
Two ore passes in operation
6 85 117 68 147 66 153
7 80 125 77 130 73 138
8 68 146 64 157 63 160
9 75 133 67 149 58 172
10 86 117 74 135 69 146
11 79 126 69 145 66 152
One ore pass in operation
12 53 188 46 216 35 287
13 67 150 59 168 46 220
14 65 154 59 168 42 243
15 50 200 45 224 33 306

In the event that an ore pass is not available, buying new
LHDs is not a likely scenario, and often LHDs from other
production areas are relocated to the affected production
area to compensate for the longer travelling distances and
maintain production targets. Increasing the number of
LHDs in a production area with reduced ore pass availabil-
ity decreases the production rate and increases ventilation
costs. This can be illustrated using the production rates for
one, three, and six LHDs in scenario 1. The production
rates for the separate production areas are based on the pro-
duction rates achieved in scenario 1 and multiplied by the
number of production areas used. Comparing three LHDs
operating in three separate production areas (13,614 t/day)
with the same three LHDs operating in a single production
area (12,499 t/day) results in a difference of 1,115 t/day.
Similarly, six LHDs operating in six separate production
areas yields 27,228 t/day, whereas the same six LHDs oper-
ating in a single production area yields 20,349 t/day, a dif-
ference of 6,879 t/day. If the same comparison is made
using the results of scenario 15, in which only one ore pass
is operational, the differences are even higher: 8,035 t/day
for three LHDs and 20,521 t/day for six LHDs. Similarly,
the ventilation cost for three LHDs working in three sepa-
rate production areas with similar conditions to scenario 1
is US$0.27/kt, whereas the ventilation cost for three LHDs
working in a single production area is US$0.29/kt. The ven-
tilation cost for six LHDs working in six separate produc-
tion areas with similar conditions to scenario 1 is
US$0.54/kt, whereas if six LHDs are in operation in a sin-
gle area, the cost is US$0.71/kt. If a similar comparison is
made using scenario 15, in which only one ore pass is oper-
ational, the difference in cost is even higher: US$0.39/kt for
three LHDs in operation and US$1.64/kt for six LHDs in
operation.
CONCLUSIONS
In this study, scenarios representing variations in ore
pass availability with different numbers of operating LHDs
were analyzed to determine the additional ventilation costs
related to LHD operation when one or more ore passes are
not available. The results indicated the following:
• Increasing the number of LHDs in the areas affected by
disturbance (e.g., ore pass availability) causes LHD pro-
duction rates to decrease and ventilation costs to increase.
• The production rate decreased by as much as 50%
(2,259 t/day), 55% (6,920 t/day), and 67% (13,642 t/day)
for one, three, and six operating LHDs, respectively,
when the number of ore passes was reduced from four to
one.
• The ventilation cost increased by as much as 200%
(US$0.09/kt/LHD), 224% (US$0.36/kt/LHD), and
306% (US$1.47/kt/LHD) for one, three, and six operat-
ing LHDs, respectively, when ore pass availability was
reduced from four to one.
• Increasing the number of LHDs operating in an area
affected by a disturbance (e.g., ore pass availability)
generated higher ventilation costs and decreased the
potential production rate. One, three, and six LHDs
operating in separate production areas produced an addi-
tional 1,115, 8,035, and 20,521 t/day, respectively, than
did the same number of LHDs operating within a single
production area.

CIM Journal | Vol. 10, No. 3, 2019

Simulating the effect of LHD operations on production rates and ventilation costs in a sublevel cave underground mine
In the event that an ore pass is not available, buying new
LHDs is not a likely scenario, and often LHDs from other
production areas are relocated to the affected production
area to compensate for the longer travelling distances and
maintain production targets. This study shows that increas-
ing the number of LHDs in a production area with reduced
ore pass availability decreases the production rate and
increases ventilation costs.
ACKNOWLEDGMENTS
The authors thank the reviewers for comments that
greatly improved the paper.
Paper reviewed and approved for publication by the
Maintenance, Engineering and Reliability Society of CIM.
REFERENCES
Applied Materials. (2019). Applied SmartFactory® Simulation. Retrieved
from http://www.appliedmaterials.com/global-services/automation-soft-
ware/automod
Bailey, M., Olsson, B., & Glassock C. (2012). Underground Ore Han-
dling Systems. Retrieved from https://ausimm.com/product/under-
ground-ore-handling-systems/
Bloss, M., Harvey, P., Grant, D., & Routley, C. (2011). Underground ore
movement. In P. Darling (Ed.), SME Mining Engineering Handbook.
USA: Society for Mining, Metallurgy and Exploration, Inc, 1271–1294.
Chadwick, J. (2008). Moving ore efficiently. International Mining,
June, 38–46.
de la Vergne, J. N. (2003). The hard rock miner’s handbook (3rd ed.).
North Bay, ON: McIntosh Engineering.
Atlas Copco Rock Drills AB. (2007). Loading and haulage in under-
ground mining. Sweden: Author
Greberg, J., Salama, A., Gustafson, A., & Skawina, B. (2016). Alterna-
tive process flow for underground mining operations: Analysis of con-
ceptual transport methods using discrete event simulation. Minerals,
6(3), 65. https://doi.org/10.3390/min6030065
Greberg, J., & Sundqvist, F. (2011). Simulation as a tool for mine plan-
ning. Proceedings of the 2nd International Future Mining Conference,
Melbourne, Australia, 273–278.
MathWave. (2019). EasyFit. Retrieved from http://www.mathwave.com
/easyfit-distribution-fitting.html
Paraszczak, J., Laflamme, M., & Fytas, K. (2013). Electric load-haul-
dump machines: Real alternative for diesel? CIM Journal, 4(1), 13–19.
Pronk, A., Coble, J., & Stewart, P. A. (2009). Occupational exposure to
diesel engine exhaust: A literature review. Journal of Exposure Science
and Environmental Epidemiology, 19, 443–457. https://doi.org/10.1038
/jes.2009.21
©2019 Canadian Institute of Mining, Metallurgy and Petroleum. All rights reserved.
131
Bartlomiej Skawina is a PhD student at Luleå University of Technology
(LTU), Sweden. He has a MEng in civil engineering, specializing in mining
and geotechnical engineering. Currently, he works for the sustainable
intelligent mining systems (SIMS) project, WP 4 Integrated process control.
bart.skawina@ltu.se
Abubakary Salama has a PhD in Rock Engineering from LTU, a MSc in
Mining Engineering from the École des mines de Paris, France, and a BSc in
Mining Engineering from the University of Dar es Salaam, Tanzania, where
he is a lecturer. His research interest is production simulation and
optimization for deep mines. He is also involved in several deep mine
projects at LTU.
Jenny Greberg is an associate professor at LTU. Her current research
focuses on mine production optimization. She is also programme director
for the Strategic Innovation Programme for the Swedish Mining and
Metal Producing Industry (SIP STRIM). She has experience in the
aerospace industry and in mining consulting, as well as extensive
experience in research and innovation management in Swedish and
European organizations.
Workplace exposure to dusts and aerosols – diesel exhaust. (2017, May
30). OSHWiki. Retrieved from http://oshwiki.eu/wiki/Workplace_expo
sure_to_dusts_and_aerosols_-_diesel_exhaust
Sargent, R. G. (2011). Verification and validation of simulation models.
Proceedings of the 2011 Winter Simulation Conference, Syracuse Uni-
versity, New York, NY, 183–198. https://doi.org/10.1109/WSC.2011.
6147750
SIMS (Sustainable Intelligent Mining Systems). (2018). Horizon 2020
funded project. Retrieved from http://simsmining.eu
Tuck, M. (2011). Ventilation. In P. Darling (Ed.), SME Mining Engi-
neering Handbook (3rd ed). USA: Society for Mining, Metallurgy and
Exploration, Inc.
United States Energy Information Administration. (2017). Annual
energy outlook 2017 (DOE/EIA-0484). Washington DC: author.
Retrieved from https://www.eia.gov/outlooks/ieo/
Usmani, T., Szymanski, J., & Apel, D. (2014). Optimisation of LHDs oper-
ation at the extraction level in a block cave project. International
Journal of Mining and Mineral Engineering, 5(2), 138–151. https://doi.org
/10.1504/IJMME.2014.060205
Wennmohs, K. H. (2014). New standards in underground loading and
transport technology: electrically powered mining trucks in interaction
with electrically powered LHD loaders. Mining Report, 150(1–2), 48–
53. https://doi.org/10.1002/mire.201400001
Yuriy, G., Runciman, N., & Vayenas, N. (2013). Discrete-event simula-
tion in underground mining – applicability and experience at Vale
Canada operations. Paper presented at the 23rd World Mining Congress
and Expo, Montreal, QC.
 GEOLOGY 133

Reconciling mineral reserves at the well-to-well in-situ copper

leaching operation at San Manuel mine, Arizona, USA
G. A. Sutton
Sutton Geological Services Ltd., Vancouver, British Columbia, Canada

https://doi.org/10.15834/cimj.2019.9

ABSTRACT At the San Manuel mine (Arizona, USA), in-situ copper leaching of a copper porphyry
deposit was carried out from 1986 to 1999 during underground and open-pit mining. In 1993–1994, well-
field testing of the supergene zone, composed primarily of chrysocolla and copper-rich clay minerals, led
to estimates that 52% of the acid-soluble copper (ASCu) could be recovered during in-situ leaching;
however, the actual recovery from zone 6, which had the longest leach history, was approximately 42%.
Comparing drillcore assay data from holes drilled before and after leaching also indicated an approxi-
mate recovery of 42%. The discrepancy between the predicted (52%) and actual (42%) recovery is prob-
ably related to several factors, including difficulty maintaining fluid saturation, channelling effects, and
gypsum precipitation.
■ KEYWORDS Copper oxides, in-situ leaching, Modelling, Porphyry, Solvent extraction and electrowinning
(SXEW), Well-to-well

RÉSUMÉ À la mine San Manuel (Arizona, États-Unis), on a procédé de 1986 à 1999 à la lixiviation in-
situ de cuivre d’un gisement porphyre cuprifère pendant l’exploitation minière souterraine et à ciel
ouvert. En 1993 et 1994, des essais effectués dans des champs de captage de la zone supergène, compo-
sée principalement de chrysocolle et de minéraux argileux cuprifères, ont abouti à des estimations selon
lesquelles on peut récupérer 52 % de cuivre soluble dans l’acide (ASCu) pendant la lixiviation in-situ ;
toutefois, la récupération réelle dans la zone 6, qui affiche l’histoire la plus ancienne en matière de lixi-
viation, était d’environ 42 %. La comparaison des données d’analyses de forage obtenues à partir de
trous forés avant et après la lixiviation indiquait également une récupération approximative de 42 %. La
différence entre la récupération prévue (52 %) et réelle (42 %) réside probablement dans plusieurs fac-
teurs, notamment la difficulté à maintenir la saturation des fluides, les effets de canalisation et la préci-
pitation du gypse.
■ MOTS CLÉS d’un puits à l’autre, extraction par solvant et extraction électrolytique (SXEW), lixiviation in-
situ, modélisation, oxydes de cuivre, porphyre

INTRODUCTION
in-situ copper leaching at the San Manuel mine, Ari-
zona, USA (Figure 1), began in 1986. It was recognized
that some of the copper oxide resource was not economi-
cally recoverable by conventional open-pit and heap leach
methods using solvent extraction and electrowinning
(SXEW) and a high stripping ratio would be required to
access part of the resource. At the time, in-situ leaching
was becoming popular in the uranium industry, but the
approach was virtually unknown in the copper industry,
especially at a large scale.
From 1986 to 1989, the San Manuel mine underwent
considerable technical changes related to the design, devel-
opment, and implementation of in-situ well construction
methods, including the formulation of suitable well pattern
arrays and the development of extensive hydrogeological
and geochemical knowledge of the mineral deposit (Wiley,
Ramey, & Rex, 1994). Surface access was limited due to
the active open-pit operation; therefore, in-situ leaching
was applied in the areas that were available. Initially, this
resulted in injection wells developed to build up solution in
the dewatered area of the open-pit operation envelope.
Pregnant leach solution (PLS) was collected via gravity
flow from an abandoned portion of the block cave under-
ground mine.
In 1989, the first well-to-well pattern was drilled to test
the feasibility of extracting the PLS from the surface. Some

CIM Journal | Vol. 10, No. 3, 2019

134 G. A. Sutton
of the upper benches of the open-pit operation were
selected because the system could operate for up to a year
before open-pit stripping had to remove some of these
wells. By 1994, an increased number of benches had been
deemed available for use, allowing the expansion of the in-
situ leaching operation, which was transformed into a com-
bination of well-to-well recovery from surface and
collection from the underground workings.
In January 1995, the last load was hauled from the open-
pit operation to the heap leach pad. This permitted a rapid
expansion of the in-situ operation, focused primarily on
increasing the well-to-well extraction method. During the
next four years, more than 1,000 wells were put into oper-
ation. At that point, the injected raffinate had an average pH
of 1.3 and contained 0.05 g/L copper in solution. The com-
position of the PLS produced by the extraction wells varied
considerably, with a pH of 1.5 to 3.5 and a copper content
of 0.20 g/L to 3.0 g/L. Although extraction from the under-
ground workings continued as some solution migrated
downwards, more than two-thirds of the copper was
extracted from pump wells on the surface.
The final phase of development was the west side
expansion, which put wells on the west side of the deposit
for the first time. The PLS was expected to be recovered
equally from the underground workings and from well-to-
well pumping. Construction of the expansion began in late
1998 and well spacing was increased from 12.1 to 15.2 m
for the first time. Low copper prices forced the closure of
the San Manuel mine operations, including the under-
ground block caving operations, in 1999. The in-situ oper-
ation closed at the same time, even though it was producing
close to 18,000 t (40 million lb) of copper annually and had
an operating cost of only US$0.18/kg ($0.40/lb). The in-
situ operation had the potential to continue as a standalone
operation due to its low operating cost.
STUDY
The topic of this study is the reconciliation of mineral
reserves—comparing the expected copper recovery to the
actual amount recovered. Both measurements have associated
levels of accuracy due to error margins in their determination.
Numerous examples of standard mining reconciliations
exist; however, this study discusses some of the basic
parameters for determining mineral reserves for an in-situ
operation. The San Manuel production data were examined
to demonstrate how accurate the estimates of those mineral
reserves were to the actual amount of copper recovered.
Deposit geology and geotechnical description
The San Manuel ore deposit has been cited as one of the
classic copper porphyry systems (Lowell & Guilbert, 1970;
Guilbert & Park, 1986). Copper porphyry systems have a
few key characteristics that make them viable candidates for
in-situ leaching, including closely spaced fractures and cop-
per mineralization on the fracture surfaces. Copper leaching
CIM Journal | Vol. 10, No. 3, 2019
is typically restricted to
copper oxide minerals,
which are commonly
best developed in the
supergene zone of a cop-
per porphyry deposit, if
the deposit has been
altered and the oxidation
products preserved.
The hostrocks at San
Manuel are a Precam-
brian quartz monzonite
and a granodiorite por- Figure 1. Location of the San Manuel
phyry formed during the mine in southeastern Arizona, USA
Laramide orogeny. Both
rock types are fractured and faulted by basin and range tec-
tonics, weathered and altered by leaching, and mineralized
by supergene precipitation. The in-situ leaching operation at
San Manuel focused on this supergene zone. The readily
leachable mineralogical suite consisted largely of chryso-
colla acting as a filled fracture and as a coating accompany-
ing clays on altered feldspars (Wiley et al., 1994).
Because the amount of fracturing impacts the ability of
fluids to migrate through the rock mass and subsequently
leach the copper, a well-fractured rock mass is important
for any successful in-situ leaching operation. Diamond drill
core logging within the supergene zone of San Manuel had
a rock quality designation (RQD) averaging below 10%. In
fact, an intact core was the exception rather than the rule at
San Manuel, with many core boxes containing only a few
pieces of core greater than 10 cm (4 in.) in length.
Systematic measurement of the longest piece of core can
be used to indicate the density of fracturing and, therefore,
the degree of fluid flow through the rock mass. In more
than 10,000 core intervals logged during the open-pit study
at San Manuel, 95% of the intact core was less than 30 cm
(12 in.) in length. This estimate should be qualified because
it is from the entire drilling campaign and includes substan-
tial waste intervals. An example of typical core quality is
drillhole A-17, which encountered approximately 122 m of
oxide ore with typical San Manual characteristics; data
from 30 m of the interval are presented in Table 1. Another
measurement that relates to the amount of fracturing pre-
sent is the average size of the core fragments. As seen in
Table 1, the typical average size of fragments is less than
2.5 cm (1 in.) in length.
Copper porphyry deposits tend to be composed of highly
fractured rock due to the manner of orebody emplacement.
At San Manuel this was enhanced by fracturing due to the
underground block cave mining method used beneath and
adjacent to the open pit. This block caving had three major
impacts: additional fault surfaces were created with displace-
ments on the order of metres to tens of metres; existing faults
that already had brecciated zones up to 1 m thick experienced
additional displacement; and the rock mass near the under-
Reconciling mineral reserves at the well-to-well in-situ copper leaching operation at San Manuel mine, Arizona, USA 135

Table 1. Core data from an interval of drillhole A-17 with typical San Manuel deposit characteristics; RQD (rock quality designation)
From (m) To (m) Mean fragment size (cm) Maximum fragment size (cm) RQD (%)
121.92 123.44 1.3 16.5 0
123.44 124.66 1.3 10.2 0
124.66 126.19 1.3 7.6 0
126.19 127.71 1.3 8.9 0
127.71 129.24 1.3 10.2 0
129.24 130.76 2.0 12.7 8
130.76 132.28 2.5 12.7 0
132.28 133.50 1.3 8.9 0
133.50 134.72 1.3 8.9 0
134.72 136.09 1.3 8.9 0
136.09 137.62 0.5 7.6 0
137.62 139.14 1.3 6.4 0
139.14 140.67 1.3 6.4 0
140.67 142.19 0.5 7.6 0
142.19 142.95 0.5 5.1 0
142.95 144.48 0.8 9.5 0
144.48 145.39 1.0 5.7 0
145.39 146.91 1.3 25.4 12
146.91 148.44 2.5 22.9 8
148.44 149.96 2.5 12.1 0
149.96 151.49 5.1 10.2 0
151.49 152.40 5.1 12.7 0

ground operation was fragmented into “cave breccia,” a term
used to describe an unconsolidated breccia material. The
impacts of block caving on the rock mass likely make San
Manuel unique to most other existing or potential operations
considering the in-situ leach extraction technique; this should
be kept in mind during the remaining discussion.
Mineral reserves
The San Manuel ore reserves published in 1999 (BHP
Technical Oxide Planning and Projects Group, 1999) fol-
lowed the Australasian Code for Reporting of Exploration
Results, Mineral Resources and Ore Reserves (JORC code,
which can be found at http://www.jorc.org/) because the
mine was owned by an Australian company. It included a
comprehensive ore reserve report for San Manuel for the
years 1996 to 1999. Due to the uniqueness of in-situ leach-
ing, there was very little previously published material that
could be used for comparison.
During the 1990s, the proportion of copper extracted at
San Manuel via the in-situ method increased. Plans to ramp
up in-situ production following the closure of the open pit in
1995 required a better understanding of the potential copper
recovery using the in-situ method. In 1993 and 1994, well-
fields were tested to more accurately determine the amount
of recoverable copper. Ramey, Wiley, and Rex (1993) and
the BHP Technical Oxide Planning and Projects Group
(1999) determined that the fluid would reach 70% of the
leachable copper minerals (sometimes referred to as the
sweep factor or sweep efficiency) and that 75% of the leach-
able copper minerals would be recovered. This suggested an
overall expected recovery of 70% × 75% = 52% of the acid-
soluble copper (ASCu) within the supergene zone, which
was composed primarily of chrysocolla and copper-rich clay
minerals. From 1995 to 1999, the reserve estimates of the
in-situ leaching operation are based on this recovery for-
mula.
Fluid saturation
The operating conditions of an in-situ leaching opera-
tion will affect the amount of saturation and the likely
sweep efficiency to be used to determine recovery. Heap
leaching, which is the mainstay of copper oxide extraction,
operates only in partially saturated conditions (Rucker et
al., 2015). in-situ leaching may need to operate in saturated
or unsaturated conditions depending on the local situation.
San Manuel operated within the dewatered area of the
open-pit and underground workings. Early attempts at
using in-situ leaching at San Manuel resulted in little bar-
ren leach solution being returned after injection into the
ground. It was concluded that the solution had to build up
in the rock mass, thereby producing somewhat saturated
conditions, before fluid would be available for collection.
This was confirmed as San Manuel operations progressed
and a nearly saturated zone built up under leach. Newly
created wells adjacent to the existing wellfields encoun-
tered some level of saturation, indicating that the solution
was migrating beyond the active wellfield; however, there
were difficulties maintaining fluid levels near the surface.
Even though the orebody was exposed at the surface, in
some parts of the open pit it was not feasible to maintain
fluid levels all the way, or even close, to the surface
because slope failure due to seepage out of the pit walls

CIM Journal | Vol. 10, No. 3, 2019

136
was a concern. In addition, the lack of automated injection
wells caused fluid levels to rise and fall daily, resulting in
unsaturated conditions near the surface.
Figure 2 shows the cross section of a controlled-source
audio-frequency magnetotelluric (CSAMT) survey per-
formed within the San Manuel open pit (Carlson, Zonge,
Ring, & Rex, 2000), which was used to track the fluid sat-
uration of the wellfields. The vertical axis topography
shows the open pit outline and the 20 ohm-m contour was
considered the limit of fluid saturation. The survey indi-
cated that a substantial portion of the wellfield was satu-
rated, but a few areas were only partially saturated—near
the open-pit surface, the upper pit benches, and the base of
the pit slope. Note that the extended saturation zone (areas
greater than 20 ohm-m) from the base of the pit to the
1,200 ft. level matches the old underground workings,
where much of the solution migrated.
Well spacing
Well spacing can lead to differences in recovery rates.
Determining optimum well spacing is a complex endeav-
our that ultimately must be deposit specific; factors like
capital costs, fluid flow, fracture intensity, grades, and
Figure 2. Cross section of a controlled source audio-frequency magnetotelluric survey
performed within the San Manuel open pit (from Carlson et al., 2000)
CIM Journal | Vol. 10, No. 3, 2019
G. A. Sutton
copper recovery must be considered. During the 1990s,
attempts were made to maintain a 12.1 m corner-to-corner
well spacing at San Manuel. Open-pit benches at San
Manuel were typically 18.3 m in height and approxi-
mately 19.8 m from toe to crest with some variability. The
availability of pit benches and the necessity of keeping a
set distance between the wells and the toe and crest loca-
tions typically resulted in a double row of wells on each
bench with the rows spaced approximately 6.1 to 12.1 m
apart. The distance to the adjacent row on the next bench
was approximately 12.1 to 18.3 m. This spacing was
determined early in the operation through trial and error.
In the last year of operations, the spacing was increased to
15.2 m to reduce capital costs and because the particular
zone that was being developed was anticipated to have a
large underground collection component. Underground
collection normally resulted in higher grades due to the
longer flow paths of more than 100 m through the rock
mass. It was expected that the expanded well spacing
would not impact production rates but would result in a
significant reduction in the capital cost.
Given the limitations at San Manuel, including the pit
bench limitations and complex faulting, it was necessary to
keep well spacing at or below 15.2 m. Pro-
jects such as Taseko Mines Ltd.’s Florence
(Florence, Arizona) and Excelsior Min-
ing’s Gunnison (105 km east of Tucson,
Arizona) proposed 21.6 m corner-to-cor-
ner well spacing, which might be func-
tional if these projects encounter less
complex environments and require sub-
stantially lower capital costs in order to be
more economic while still maintaining
high sweep efficiency.
Channelling effects
One of the concerns with in-situ leach-
ing is the possibility of fluid becoming
channelized and either not reaching or
insufficiently reaching all available frac-
ture surfaces. This problem is also faced
by heap leach pads. In the in-situ case,
however, it is difficult to determine the
scale of channelling unless the leached
rock mass can be examined. Fortunately,
in the case of San Manuel, a handful of
diamond drill holes cored within the
leached rock mass could be analyzed.
Diamond drill hole SMO9607 (drilled
May 6, 1996, in zone 6, on bench 2460) is
representative of the effects of leaching
after the area had been under leach for
approximately 3 years. It is possible that
the leach solution had been in the area or
the adjacent area for several years prior;
Reconciling mineral reserves at the well-to-well in-situ copper leaching operation at San Manuel mine, Arizona, USA 137

this was corroborated by other drillcore

that exhibited similar characteristics. It is
estimated that by the time drillhole
SMO9607 was drilled, the in-situ leaching
operation had recovered approximately
35% of the ASCu; however, it should be
noted that underground collection was
allocated to zones based on operational
experience and precisely which bench and
zone it came from was not known.
Figure 3 shows the ratio of acid solu-
ble copper grade to total copper grade
(ASCu/TCu), which gives a good repre-
sentation at a macro scale of the effects
of leaching on the rock mass. As previ-
ously noted, the orebody is a copper por-
phyry, which has a relatively high level Figure 3. Grade ratio of acid soluble copper (ASCu) to total copper (TCu) in drillhole SMO9607
of homogeneity with regard to rock type at the San Manuel mine
and copper grade. Figure 3 shows the
effects within the supergene zone, where the copper ore fluid flow. This was a potential cause of channelling in the
consists primarily of chrysocolla with some copper-rich operation.
clays. No sulfide minerals were observed in this interval. If leaching is occurring only in select areas where solu-
The typical ASCu/TCu ratio of unleached materials was tion has been flowing through the rock mass for extended
75–80% in the supergene zone of the deposit. It can thus periods of up to several years, as Figure 3 suggests, then the
be inferred that copper has been leached wherever the ratio cause of the channelling needs to be considered. Although
falls significantly below 70% within the supergene zone. this paper does not comprehensively address the exact
Additionally, physical characteristics noted in the core log cause of channelling, a number of observations are shared
identifies the areas with low ASCu/TCu ratios as portions here. As demonstrated by the CSAMT survey, the entire
that show the effects of leaching, whereas the high ratio rock mass was not saturated. There were numerous fault
areas do not show leaching effects. Leaching effects show surfaces in the orebody that could have acted as conduits or
on the core as gypsum precipitates and powdery white barriers to fluid flow. The faults are commonly up to 1 m
residue. The following two paragraphs relay specific mea- thick and composed of brecciated material with clay. The
surements and comments from two sections of core. wellfields are delineated by zones that are defined by fault
The interval from 86.9 to 140.2 m shows strong leaching boundaries. As was described in the core interval from
effects. The average grade is 0.45% TCu and 0.13% ASCu, 140.2 to 172.2 m, there was a considerable amount of fine-
which is a 32.2% ASCu/TCu ratio. The entire interval has grained material present, which could have inhibited fluid
been logged, with the rock type being a monzonite por- flow. It is also possible that activating pump wells too early
phyry displaying fairly typical copper porphyry character- would not give the injection wells enough time to saturate
istics and not containing any significant fault material. the rock mass. Once preferred flow paths are established it
Common core descriptions include strong leaching effects may be difficult to correct the situation.
and moderate quantities of gypsum. This interval has fairly Original grades in the supergene zone of the San Manuel
typical copper porphyry characteristics and does not con- deposit were on the order of 0.50–0.80% ASCu, based on
tain any significant fault material. the grade of unleached rock and the expectation that the
The interval from 140.2 to 172.2 m shows only limited original grade would have been similar in the leached areas.
leaching effects. The average grade is 0.59% TCu and Figure 4 shows the ASCu grade for drillhole SMO9607 as
0.35% ASCu, which is a 67.6% ASCu/TCu ratio. For the an illustration of the subsequent reduction after leaching. In
entire interval, the rock type is a monzonite porphyry. Com- areas of strong leaching, the ASCu is nearly depleted
mon core descriptions include little to no leaching effects (approximately 0.03%).
and little to no gypsum present. One of the bigger differ- The amount of copper that appears to have been recov-
ences from the previous interval is that the presence of ered in some of the areas that have been well leached should
gouge and breccia is recorded several times. The phrase be determined. A good example is the interval of drillhole
“gouge and breccia” was commonly used at San Manuel SMO9607 from 109.7 to 112.8 m. Samples and assays were
when the rock mass had been considerably brecciated by taken at 1.5 m increments; therefore, this example represents
faulting and block caving. The elevated amount of fine- two separate assays that essentially produced the same
grained material in the breccia may have been a barrier to results. Both samples included the presence of gypsum in the

CIM Journal | Vol. 10, No. 3, 2019

138 G. A. Sutton

holes drilled in the deposit after the initia-

tion of leaching.

Effect of gypsum precipitation

(selenite crystals)
The presence of gypsum in the form of
selenite crystals was prevalent in rock
chip samples from well drilling. Because
well drilling was normally done immedi-
ately adjacent to existing wellfields, leach
solution was commonly present in the
areas being developed for new wells.
Also, during the 1997 core drilling cam-
paign, several holes were drilled in active
leaching areas. After drying the core from
Figure 4. Acid soluble copper (ASCu) grade of core from drillhole SMO9607 at the San Manuel these holes overnight, the surface of the
mine core was sometimes coated with selenite
crystals the next day.
highly fractured rock, which has almost the consistency of The exact effect of selenite on production from the in-
gravel. The interval was composed of the monzonite por- situ operation was not clear. Were the crystals present
phyry, which hosted the mineralization, and strong leaching and, if so, were they the observed size prior to drilling, or
effects were seen. The total copper (TCu) grade based on did crystal growth occur after rock chips and core were
assays of the interval (Figure 5) was 0.168% and the ASCu removed from the ground? It seemed that both scenarios
grade (Figure 4) was 0.035%. It should be noted that this occurred. Rock chips from well drilling showed the pres-
interval was entirely within the supergene zone and no cop- ence of selenite crystals as soon as they were recovered.
per sulfide minerals were present. This zone showed exten- Because only a few minutes of time had passed, not
sive leaching and may have had leach solution flowing enough time for significant crystal growth, these crystals
through it for three or more years; however, the ASCu grade developed underground. In the case of core drilling, crys-
was not 0% here or in other results from this hole, nor in tals grew on the cores overnight when exposed to the air.
numerous other holes. If the residual ASCu grade of 0.035% As leaching occurs, material is dissolved and removed
is close to the ultimate limit of depletion and typical starting from the rock mass, opening up pathways for the solution to
grades in the orebody are 0.40% and higher, then a recovery travel; gypsum may have partially filled these new open
of 90% of the ASCu can be presumed if the leach solution spaces. Records illustrate, however, that many wellfields
reaches the rock mass in an area. It should be noted that this experienced decreasing flow rates. This may have been
presumption is based only on small-scale observations caused by gypsum plugging up the rock mass, although no
because this degree of leaching was not found over complete scientific study has been done on this issue. Several other fac-
core intervals and instead only in a small number of core tors may have caused decreased flow rates, such as ground
subsidence shearing wells off, pit bench
failures, and seepage from pit walls. The
presence of gypsum in drillcore and the
geochemical modelling completed on the
Florence project showed that the saturation
concentration of gypsum was reached in
most leach tests (Sinclair, 2015). This issue
should be considered for future in-situ
leaching operations because it might have a
significant impact on the sweep efficiency
used in reserve calculations.

Other factors
A number of other factors should be
considered with respect to production rec-
onciliation. The factors that had perhaps
the largest potential effect on reconcilia-
Figure 5. Total copper (TCu) grade of core from drillhole SMO9607 at the San Manuel mine tion at San Manuel were the ASCu assay-

CIM Journal | Vol. 10, No. 3, 2019

Reconciling mineral reserves at the well-to-well in-situ copper leaching operation at San Manuel mine, Arizona, USA
ing, the challenge of attributing production to the correct
wellfields, and the use of the block modelling method.
The ASCu analysis method used by the San Manuel lab-
oratory would not meet today’s best practice standards.
The San Manuel method consisted of measuring a specific
quantity of pulverized sample, placing it in a sulfuric acid
bath for a set period of time, rinsing, and performing
atomic absorption analysis. To the best of the author’s
knowledge, all assays were completed by the internal San
Manuel laboratory. A small number of assays were sent to
external labs for comparison, where “check samples” or
duplicate samples were assayed to validate and verify.
During an internal 1997 core drilling campaign, check
samples, duplicate samples, and third-party lab assays
were implemented and it was found that there were some
discrepancies in the results from the San Manuel lab. Fur-
thermore, during the Gunnison project prefeasibility study,
M3 Engineering (2014) noted that the San Manuel labora-
tory was a non-independent lab and that because they were
not reproduced during re-assaying done in 2014, the origi-
nal results were not used in their study. The Florence pro-
ject prefeasibility study (M3 Engineering, 2013) includes a
detailed explanation of the “San Manuel” method for
ASCu assays. During the 1990s, there were several ASCu
assaying methods used in Arizona. Although all were rela-
tively similar, each did have its idiosyncrasies. Even
though the assay quality would not meet today’s standards,
the San Manuel laboratory produced consistent internal
assays for the mine site and recoveries from the heap leach
pad met with those expectations (i.e., 85–87% of ASCu).
Attributing copper production to the correct block model
areas was complicated. The challenge with any reconcilia-
tion effort is trying to compare what was produced to where
it came from. Just as an open-stope mining operation will be
in production in multiple zones with differing geology and
grades at the same time, in-situ leaching draws from areas
with differing geology and grades, which leads to the chal-
lenge of determining the correct location from which the
copper is leached. At San Manuel, there were a couple of fac-
tors to consider. First, a portion of the copper production
came from collecting PLS from the old underground work-
ings. Because the distance from the bottom of each well to
the workings was more than 100 m, partially saturated con-
ditions were encountered, and fluid migration could be con-
trolled by unknown fault surfaces, it was never known
precisely from where production had originated. Opera-
tionally, increases in flow rate when new zones were put
online gave some indication of the source, but not with a high
degree of accuracy. The second issue was that there was a
gradient. The orebody was inclined, and solution migrated
toward the bottom of the pit. Because of this, it was expected
that the solution being pumped probably originated from the
adjacent injection wells and, in part, from the injection wells
in the bench above. The amount of solution coming from
each location could not be determined exactly.
CIM Journal | Vol. 10, No. 3, 2019
139
Developing the 3D block model for the resource was
more challenging than typical block models. In 1995, the
first resource block model was created specifically for the
in-situ operation. Prior to that, the model had been designed
for open pit operations. An immediate challenge in creating
the 1995 block model was that leaching had already taken
place and creating a 3D block model that was representa-
tive of the current conditions was not possible; therefore,
the 1995 model relied on drilling data that were collected
prior to any in-situ leaching in the area or, where possible,
more recent drilling data if the area had not been affected
by the in-situ leaching. Additionally, ground subsidence
had occurred since the original drilling campaigns, causing
the ground level to drop. Drillhole collars had to be manu-
ally adjusted to reflect the new ground level but ground
subsidence had not been entirely vertical, resulting in a less
than ideal drillhole dataset.
Copper production from zone 6
Of the 16 named zones at San Manuel, zone 6 is the best
one to use to compare production history with the reserves
for reconciliation because it was in production for the longest
period of time and was near its economic limit at the time of
mine closure. Based on the mine economics, a resultant PLS
grade of 0.20 g/L copper was determined as the cut-off grade
for economic processing. Keeping in mind that it was diffi-
cult to attribute production to exact wellfields, looking at an
entire zone does give a good indication of the potential total
recovery. Zone 6 was, however, below the average grade for
the deposit, so it might not be representative of recoveries
everywhere in the deposit. Recovery might be tied to starting
grade for this type of operation, with higher starting grades
resulting in higher recovery rates. This has not yet been stud-
ied in sufficient detail to be conclusive. For reconciliation
purposes, wellfield production was reported by the zone and
open-pit bench where the leaching operation was located.
Production statistics for zone 6, including the starting ASCu
grade from the block model, the available ASCu (in kg), the
copper production attributed to each bench, and the calcu-
lated recovery of ASCu are listed in Table 2. Note that this
zone was still in production at the time of mine closure and
would likely have approached 50% recovery by the time eco-
nomic limits were reached.
Reconciliation
Reconciling the copper recovered from an in-situ leach
operation is complex and there are many uncertainties.
Even with good operational data, it is difficult to exactly
determine the origin of the dissolved copper. Looking at a
larger scale is a better approach because solutions can
migrate along unknown paths and dissolved copper can
come from anywhere along that flow path. Because there
have been few in-situ copper leaching operations in the
world and even fewer that have used the well-to-well
recovery method, there is insufficient information available
140 G. A. Sutton

well-to-well in-situ leach operation, the comparison is

Table 2. Production data from zone 6 at San Manuel mine; ASCu (acid
soluble copper)
worth mentioning.
Bench Block model Available Cu ASCu
Hole ARM158, drilled prior to leaching, had an average
ASCu grade ASCu production recovery grade of 0.53% ASCu from 0 to 213.4 m. SMO9607, drilled
(%) (kg) (kg) (%) after approximately 3 years of leaching, had an average
2340 0.423 15,182,041 7,121,400 46.9
grade of 0.31% ASCu from 0 to 213.4 m. The difference in
2400 0.469 12,283,126 5,216,308 42.5
2460 0.480 11,633,005 4,082,328 35.1
grade is a reduction of 42%. This area continued to be
2520 0.479 14,580,695 6,486,366 44.5 leached for three additional years after hole SMO9607 was
2580 0.398 11,199,172 4,762,716 42.5 drilled, although at a much lower production rate. Produc-
2640 0.375 11,094,283 5,669,900 51.1 tion data suggests that approximately 35% of the copper
2700 0.341 18,892,391 7,779,103 41.2 had been removed by the time SMO9607 was drilled and
2760 0.347 9,487,860 3,764,814 39.7 that a further 7% was recovered after the hole was drilled.
2820 0.342 7,440,730 2,449,397 32.9
Although this is not a highly accurate method of compari-
2880 0.336 2,538,315 430,912 17.0
2940 0.312 571,011 165,561 29.0
son, it does indicate an ASCu recovery in the 50% range at
Total – 114,902,618 47,928,799 – the San Manuel mine.
Average 0.406 – – 41.7
DISCUSSION
Although the reconciliation indicates that San Manuel
in the literature regarding what kind of recovery could be was on track to recover approximately 50% of the ASCu
expected. Most research has relied on laboratory testing to from one of its zones, one should not conclude that other
infer potential recovery rates. San Manuel was one of the operations will experience the same results. Operations in
largest scale well-to-well in-situ copper leaching operations higher grade zones at San Manuel appear to have achieved
in the world; even so, it only partially leached the orebody higher recovery rates. Other deposits may not have the
and mining approached economic limits only in a limited amount of faulting that existed at San Manuel and certainly
area. As discussed above, zone 6 had the longest opera- few deposits will have to deal with the effects of ground
tional history of any zone brought online at the operation. subsidence from block caving operations. Channelling
The operational data suggest that approximately 42% of the appears to have been a significant issue at San Manuel, as
ASCu was recovered and that 50% recovery could have indicated by the limited data available. This might not be as
been reached if mining had continued. The grade at zone 6 significant an issue at other operations, particularly if they
was slightly lower than the average grade of the deposit, so are working in saturated conditions or refrain from pump-
it may not reflect the recovery for the entire operation. With ing until the injection wells have saturated the rock mass.
respect to reconciliation, the reserves were based on 52% The recoverable ASCu was estimated at 75% for the min-
recovery of the ASCu and production came close to sup- eral reserves at San Manuel and an expected sweep effi-
porting that figure; therefore, on a macro scale, it seems ciency of 70% was used in the reserve calculation. The
that production would have eventually matched the challenges maintaining fluid saturation and the apparent
expected recovery. The two main components that were channelling effects meant that 70% sweep efficiency was
used to determine the expected 52% recovery rate were a probably an overestimate. In reality, at least one of the main
sweep factor of 70% and an ASCu recovery of 75%. The zones in operation might have been achieving a sweep effi-
post-leaching core hole data suggest that the ASCu recov- ciency of only 55%. It is possible that other operations could
ery was closer to 90% based on short core intervals; this achieve sweep efficiencies of 90% or greater, particularly if
results in a calculated sweep factor of closer to 55%. they did not encounter the complications that occurred at
A second way to test the reconciliation would be to drill San Manuel. Whereas 90% ASCu recovery was achievable
core holes in the same location before and after leaching. on the small scale—in a 3 m length of core—the main rea-
This would seem impractical for the entire deposit because son that 90% did not seem to be recoverable at the deposit
it would require twice the number of holes drilled; how- scale was that the fluid flow was not able to reach all of the
ever, at San Manuel there was one instance of a second hole fracture surfaces where copper mineralization occurred.
(SMO9607) being collared in the leach area 18.3 m away This issue is related to sweep efficiency, not leachability.
from an original hole (ARM158) collared prior to leaching. The fact that 90% could be leachable on a small scale should
The distance apart makes them a less-than-perfect example; indicate that it can be applied to the larger scale.
however, being a copper porphyry deposit the grades did If mineral reserves were to be reported at San Manuel
not tend to change that much over these distances. For com- today, it would probably be more accurate to use a sweep
parison, the typical range value found in the variography efficiency of 55% and a leachability factor of 90% to pro-
was 76.2 m. This method of comparison is even less accu- duce an overall ASCu recovery of 50%. This would be a
rate than the well production method. Nevertheless, since positive outcome for future projects planning on using this
few data exist regarding the reconciliation of reserves at a extraction method because they are unlikely to encounter

CIM Journal | Vol. 10, No. 3, 2019

Reconciling mineral reserves at the well-to-well in-situ copper leaching operation at San Manuel mine, Arizona, USA
the extent of issues that San Manuel faced with faulting and
the effects of block caving and thus will likely have a much
higher sweep efficiency.
Reconciliation will continue to be challenging for future
in-situ leaching operations. A number of steps should be
taken to better determine the location that the copper is
being recovered from: Real-time flow rates should be deter-
mined for each recovery and injection well; copper grades
in the injected leach solution and PLS grades from each
recovery well should be measured at a minimum of daily
using an automated system; and hydrogeological models
should be created to estimate the fluid flow paths and vol-
ume of fluid moving from each injection well to each
recovery well. These analyses will result in a more accurate
representation of the source of the copper being collected.
It is expected that good sampling and assaying techniques
and better 3D block modelling practices will also improve
understanding and results.
CONCLUSIONS
in-situ leaching operations are not yet common in the cop-
per industry; however, the positive economic results from
San Manuel, coupled with the currently anticipated start up
of two advanced-stage projects in Arizona, may mark the
beginning of the widespread use of this environmentally
friendly, low capital cost, feasible extraction method. Being
able to accurately reconcile the mineral reserves with copper
production is, therefore, of increasing importance.
From 1993 to 1994, wellfields at San Manuel were
tested to determine the amount of copper that could be
recovered from in-situ leaching of the supergene zone,
which was composed primarily of chrysocolla and copper-
rich clay minerals. It was determined that the leach fluid
could reach 70% of the copper minerals and that 75% could
be leached for an ASCu recovery of 52%. For future pro-
jects or operations, a pilot test might not be feasible. Stud-
ies using flow modelling and fracture analysis have been
used to estimate the sweep efficiency, as in the case of the
Florence and Gunnison projects.
REFERENCES
BHP Technical Oxide Planning and Projects Group. (1999). San Manuel
operations oxide ore reserves report. Unpublished internal report.
Carlson, N., Zonge, K., Ring, G., & Rex, M. (2000). Fluid flow map-
ping at a copper leaching operation in Arizona. The Leading Edge,
19(7), 752–755. https://doi.org/10.1190/1.1438710
Guilbert, J. M., & Park, C. F. (1986). The geology of ore deposits. New
York: W. H. Freeman and Company.
Lowell, J. D., & Guilbert, J. M. (1970). Lateral and vertical alteration-
mineralization zoning in porphyry copper ore deposits. Economic Geol-
ogy, 65(4), 373–408. https://doi.org/10.2113/gsecongeo.65.4.373
M3 Engineering. (2013). Florence Copper Project NI 43-101 technical
report prefeasibility study, revision 1. (Technical report)
©2019 Canadian Institute of Mining, Metallurgy and Petroleum. All rights reserved.
141
Determining the amount of recovered copper in a dis-
crete area of rock mass can be done by post-leach core
drilling or a combination of flow modelling using daily
flow rate collection and solution grades. San Manuel was
able to use both methods. The final results of tabulating
many years of daily flow rates and grades from the zone 6
wellfield showed that approximately 42% of the ASCu was
recovered, indicating that the sweep efficiency was closer
to 55%. Comparison of drillcore assay data from two
closely situated holes, one drilled before (ARM158) and
the other after (SMO9607) leaching, also indicates that
approximately 42% of the ASCu was recovered.
The discrepancy between the predicted (52%) and actual
(42%) recovery is likely related to a combination of factors,
including difficulty maintaining fluid saturation, channelling
effects, and gypsum precipitation. These recovery rates do not
necessarily indicate what other projects may achieve because
each deposit will have its own unique characteristics.
ACKNOWLEDGMENTS
The author acknowledges the following individuals at
the San Manuel in-situ operation: Jim Spencer, structural
geologist; Greg Mazur, hydrogeologist; George Ring,
senior hydrogeologist; and Martin Rex, technical work
group leader.
An earlier draft of this paper was published in the
Proceedings of the 58th Annual Conference of
Metallurgists (COM 2019) hosting Copper 2019 prior to
undergoing the CIM Journal peer-review process.
Paper reviewed and approved for publication by the
Geological Society of CIM.
Gary Sutton was the competent person for declaring reserves at the San
Manuel in-situ leaching operation. His work included 3D geological
modelling, locating boreholes for the wellfields, and submitting requests for
capital expenditures. He now consults on a number of in-situ leaching
projects worldwide. sutton.gary@gmail.com
M3 Engineering. (2014). Gunnison Copper Project NI 43-101 technical
report prefeasibility study. (Technical report).
Ramey, D. S., Wiley, K. L., & Rex, M. J. (1993). In-situ resource recov-
ery testing at San Manuel. Presented at SME annual meeting in Reno,
NV. Preprint Number 93-229.
Rucker, D. F., Baldyga, C. A., Cubbage, B., & Milczarek, M. (2015).
Investigating leaching alternatives for heterogeneous heap leach pads.
Proceedings of Heap Leach Solutions 2015, Reno, NV.
Sinclair, L. K. (2015). Test work and geochemical modeling for copper
in-situ leaching. Presented at the 2015 Conference of Metallurgists,
Montreal, QC.
Wiley, K. L., Ramey, D. S., & Rex, M. J. (1994). in-situ leaching well-
field design at San Manuel. Mining Engineering, August, 991–994.
 MINERAL PROCESSING 143

Using piston–die press tests for preliminary sizing and selection

of a high-pressure grinding roll
F. Wang
Norman B. Keevil Institute of Mining Engineering, The University of British Columbia,
Vancouver, British Columbia, Canada
and
Hatch, Vancouver, British Columbia, Canada

B. Klein
Norman B. Keevil Institute of Mining Engineering, The University of British Columbia,
Vancouver, British Columbia, Canada

P. Rosario
Hatch, Vancouver, British Columbia, Canada

https://doi.org/10.15834/cimj.2019.10

ABSTRACT Industry practice for high-pressure grinding roll (HPGR) assessment and sizing relies on
tests using pilot-scale machines with 0.7–1.0 m roll diameters. Typically 1–10 t of sample are required,
depending on the level of study and resource complexity. This is challenging during early-stage devel-
opment, which hinders incorporating HPGR into designs. With support from the Canadian Mining Indus-
try Research Organization, researchers at the Norman B. Keevil Institute of Mining Engineering at The
University of British Columbia developed bench-scale piston–die press tests (PPTs) that correlate to
pilot- and full-scale HPGRs. This paper describes a new regression model to estimate the HPGR-specific
throughput rate using both PPT results and an HPGR database. A case study is presented to show how
the current PPT method is used for engineering studies.
■ KEYWORDS Empirical modelling, High-pressure grinding roll (HPGR), Piston–die press test (PPT), Sizing
and scale up

RÉSUMÉ La pratique en vigueur dans l’industrie concernant l’évaluation et le calibrage d’un cylindre de
broyage à haute pression (HPGR, de l’anglais high-pressure grinding roll) dépend d’essais menés à l’aide
de machines à l’échelle pilote dont les diamètres de bobine sont de 0,7 à 0,10 mètre. On a généralement
besoin de 1 à 10 tonnes d’échantillons en fonction du niveau de complexité de l’étude et des ressources.
C’est une pratique difficile durant les premières phases de développement, qui empêche l’intégration du
HPGR dans les plans. Bénéficiant du soutien de la Canadian Mining Industry Research Organization
(CAMIRO, l’organisation de recherche de l’industrie minière canadienne), des chercheurs du Norman
B. Keevil Institute of Mining Engineering (l’institut Norman B. Keevil de génie minier) de l’université de
la Colombie-Britannique (UBC) ont mis au point des essais en laboratoire sur des presses à matricer à pis-
ton (PPT, de l’anglais piston–die press tests), qui mettent en corrélation les HPGR aux échelles pilote et
réelle. Cet article décrit un nouveau modèle de régression qui évalue le taux de production spécifique aux
HPGR à l’aide des résultats tirés des PPT et d’une base de données dédiée aux HPGR. Une étude de cas
est présentée afin de décrire l’utilisation de la méthode actuelle des PPT dans les études techniques.
■ MOTS CLÉS calibrage et mise à l’échelle, cylindre de broyage à haute pression (HPGR), essai sur presses à
matricer à piston (PPT), modélisation empirique

INTRODUCTION
It is widely accepted that high-pressure grinding rolls ern comminution flowsheet design. Table 1 summarizes
(HPGRs) are a viable, energy-efficient solution for mod- select examples of HPGR applications in the mining

CIM Journal | Vol. 10, No. 3, 2019

144 F. Wang, B. Klein, and P. Rosario

Table 1. Examples of high-pressure grinding roll (HPGR) applications in mining industry

Operation/project Throughput HPGR dimensions Number of Power Status
(kt/d) (diameter × width; m) HPGR units (MW)
Cerro Verde 1 & 2, Peru (Cu-Mo) 360 2.4 × 1.65 12 2 × 2.5 Operating
Boddington, Australia (Au) 100 2.4 × 1.65 4 2 × 2.8 Operating
Morenci Metcalf, USA (Cu) 63.5 3.0 × 2.00 1 2 × 5.7 Operating
Sierra Gorda, Chile (Cu-Mo) 110 2.4 × 1.65 4 2 × 2.8 Operating
Mogalakwena, South Africa (Pt) 20 2.2 × 1.60 1 2 × 2.8 Operating
Los Colorados, Chile (Fe) 45 1.8 × 1.70 1 2 × 1.85 Operating
Côté, Canada (Au) 32 2.4 × 1.65 1 2 × 3.0 Feasibility

industry. The main benefits of HPGRs are energy savings, Previous works do not estimate the specific throughput
which leads to lower operating costs, and stable operating constant (ṁ or “m-dot”), an important HPGR sizing param-
characteristics that can better cope with variations in ore eter defined as the expected throughput capacity (Q) for an
hardness compared to the conventional autogenous (AG) HPGR unit with rolls 1 m in diameter and 1 m in width run-
or semi-autogenous (SAG) mill-based comminution cir- ning at a peripheral speed of 1 m/s (Klymowsky et al.,
cuits. 2002). To date, determining ṁ has relied on the pilot test
Current industry practice for HPGR assessment and sizing program or benchmarking data from other HPGR opera-
relies on a series of tests using pilot-scale machines with roll tions. This work adds a new set of regression models to the
diameters of 0.7–1.0 m. Large samples are required for such existing PPT method, which allows practitioners to esti-
pilot test programs, typically 1–10 t, depending on the level of mate the HPGR throughput capacity using a combination of
study and complexity of the resource. Collecting large rep- PPT results and the HPGR database. The objective of the
resentative samples during early-stage development is chal- study is to demonstrate the application of the current PPT
lenging and expensive. Furthermore, meaningful method to engineering studies, particularly preliminary
geometallurgical studies to characterize the responses of dif- HPGR sizing and selection.
ferent ore types to high-pressure comminution using pilot-
scale HPGRs are not practical. METHODS
With support from the Canadian Mining Industry Figure 1 outlines the overall testing program, which
Research Organization (CAMIRO), researchers at the Nor- included conducting a direct comparison of HPGR pilot
man B. Keevil (NBK) Institute of Mining Engineering at testing and PPTs on a composite sample, followed by a
The University of British Columbia (UBC) developed a variability assessment based on individual PPTs using sam-
suite of bench-scale piston–die press tests (PPTs) that cor- ples representing a range of ore types. PPT results were cal-
relate the energy and size reduction in pilot- and full-scale ibrated according to the UBC HPGR PPT database for
HPGRs (Davaanyam, 2015; Davaanyam, Klein, & Nadol- specific pressing force, particle size reduction ratio, and
ski, 2015; McClintock, 2018). The initial development of normalized product size distribution. The power draw and
HPGRs originated from fundamental research conducted specific throughput constant parameters were estimated
by Schönert (1988), who studied interparticle breakage using empirical equations derived from the HPGR
behaviour using PPTs. Table 2 compares PPTs and HPGR database. Subsequently, the models were used to predict
pilot testing and demonstrates the merits of the PPT net-specific energy consumption, product size distribution,
method. and specific throughput constant at varying HPGR specific
pressing forces. Lastly, HPGR sizing and selection calcula-
Table 2. Comparison of piston–die press testing and high-pressure grinding
tions were conducted based on the results of the PPTs and
roll (HPGR) pilot testing HPGR pilot tests.
Piston–die HPGR pilot
press testing testing Materials
Description Four tests under Three–four single-pass
different pressing tests under different
All materials tested in the study were porphyry ores
forces, open circuit pressing forces, open circuit from a copper deposit in British Columbia. Table 3 summa-
Sample size (kg) ~5 ~1,000 rizes the hardness characteristics of the composite sample
Maximum feed 12.5 32 and variability samples prepared for the experiment. The
size (mm)
A × b value, an indicator of ore competency derived from
Application Determine energy Accurate
and size reduction preliminary the JK drop weight test, ranged from 23.0 to 40.0, with a
relationship sizing median value of 31.7. The samples are classified as moder-
Method to obtain Empirical modelling Direct ately hard to very hard (competent) with respect to
sizing parameters predictions measurement
AG/SAG milling.

CIM Journal | Vol. 10, No. 3, 2019

 Using piston–die press tests for preliminary sizing and selection of a high-pressure grinding roll 145

Piston–die press testing

The instrumented MTS hydraulic Variability Composite
press used for the PPT program (shown samples sample
in Figure 2) can apply a maximum force
of 1,400 kN. The test apparatus is
located at the coal and mineral process- Piston–die press HPGR pilot
ing laboratory (CMPL) at the NBK tests tests
Institute of Mining Engineering. All
PPTs were conducted with an 86 mm
diameter die using feed material with a
Database HPGR sizing
top size of 12.5 mm. Davaanyam (2015) Comparison
calibrations parameters
outlined the main considerations con-
cerning the design and selection of the
piston–die apparatus, including applied
pressure, ratio of piston diameter to feed Preliminary
top size, material bed height corre- HPGR sizing
sponding to the HPGR operating gap,
and quantity of material for a represen- Figure 1. Flow chart of overall high-pressure grinding roll (HPGR) and piston–die testing program
tative sample. The geometric constraints
of the piston–die apparatus should be Table 3. Hardness characteristics of samples prepared for high-pressure grinding roll and piston–die
similar to those observed in the HPGR press testing
(Davaanyam, 2015). No. of No. of JK parameter A × b
All sample materials were prepared to composite variability Range Median Standard 80th
samples samples Deviation percentile
100% passing 12.5 mm via stage crush- 1 6 23–40 31.7 4.3 29.2
ing and screening in a reverse closed
arrangement. The moisture of the sam-
ples was adjusted to 2.5% and then the samples were
homogenized. The packed-bed bulk density of the sample
was measured to calculate the target weight of 240 mL sub-
samples. Each subsample was obtained using a riffle split-
ter and was within ±10 g of the target weight.
Subsamples were pressed in the PPT apparatus at four pis-
ton pressures (Ppiston) representing four energy levels: 1,399 kN
(240 MPa), 1,100 kN (189 MPa), 800 kN (138 MPa), and 500 kN
(86 MPa). The total energy input was determined by numer-
ical integration of the recorded force–displacement curve.
Prior to testing, the mechanical strain of the PPT system was
measured and this value was subtracted from test measure-
ments. The specific energy input to the sample was calcu-
lated by dividing the total energy input by the mass of the
sample contained in the die. For each PPT, the particle size
distribution (PSD) of the feed and product was determined
through a combination of wet and dry sieving. The relation-
ship of specific energy consumption to particle size reduc-
tion was determined for each sample.

HPGR pilot testing

A Köppern HPGR located at the CMPL was used to con-
firm the accuracy of the PPT predictions. The Köppern
pilot unit operates at a 32 mm feed top size and has a roll
diameter of 0.75 m and roll width of 0.22 m. Pilot testing
was carried out on the composite sample at select specific
pressing forces. The pilot test program evaluates the influ-
ence of specific pressing forces on HPGR comminution and Figure 2. MTS piston–die press testing apparatus at the Norman B. Keevil
provides information about the net specific energy con- Institute of Mining Engineering, The University of British Columbia.

CIM Journal | Vol. 10, No. 3, 2019

146 F. Wang, B. Klein, and P. Rosario
sumption, specific throughput constant (ṁ), and general
suitability of the ore for interparticle comminution. It
should be noted that tests on a pilot unit are generally
accepted by all HPGR suppliers and can be used to calcu-
late all the necessary process parameters for HPGR sizing.
MODELLING KEY SIZING PARAMETERS
Key parameters for sizing and scale up of an HPGR unit
are specific pressing force (Fsp), net specific energy consump-
tion (Esp), specific throughput constant (ṁ), and product
PSDs. These parameters are usually determined from the
HPGR pilot program. To model these parameters, PPT results
were calibrated to multivariable linear regression models
developed from previous testwork conducted using UBC’s
pilot HPGR unit and their MTS PPT unit. The current UBC
HPGR PPT database consists of more than 200 pilot-scale
HPGR tests covering more than 20 geological ore types.
Specific pressing force model
The empirical model used to calibrate the specific press-
ing force is expressed by equation 1 (Davaanyam, 2015).
Ppiston= 5.53 + 53.3Fsp + 24.3w − 86.2ρbulk +
(1)
HPGR
F50
HPGR − 44.4 piston
13.1F50 F50piston + 2.98P1mm
where Ppiston is the PPT piston pressure in MPa, Fsp is the
HPGR specific pressing force in N/mm2, w is the HPGR
feed moisture content in %, ρbulk is the feed bulk density
HPGR
in g/cm2, F50 is the 50% passing size of the HPGR feed
piston is the 50% passing size of the PPT feed in mm,
in mm, F50
piston is the fraction passing 1 mm in the PPT feed in %.
and P1mm
The required PPT piston pressure can be estimated using
the HPGR specific pressing force and other relevant mate-
rial properties. The specific energy consumption at the cor-
responding specific pressing force can then be determined
directly from the PPT specific energy calculation (total
energy input divided by sample mass).
Size reduction and size distribution model
The size reduction achieved in the HPGR can be cali-
brated using the reduction ratio model in equation 2
(Davaanyam, 2015).
HPGR
F50
RRHPGR = 1.86 + 1.41RRpiston+2.31 F piston −
50
(2)
HPGR − 1.02w
0.41F50
where RRHPGR is the size reduction ratio, based on 50%
passing feed size (F50) to 50% passing the full product size
(P50) for the HPGR testing, and RRpiston is the size reduc-
tion ratio (F50/P50) for PPTs. The HPGR full product refers
to the combined centre and edge product based on the mass
proportion observed during the HPGR test.
CIM Journal | Vol. 10, No. 3, 2019
Researchers have shown that the PSD of HPGR and PPT
products for samples processed at different pressing force
settings are self-similar and can therefore be normalized by
dividing by their median particle size (Fuerstenau, Shukla,
& Kapur, 1991; Fuerstenau, Gutsche, & Kapur, 1996;
Davaanyam, 2015). The normalized piston–die press prod-
uct PSD is fitted to equation 3, as suggested by Lim, Voigt,
and Weller (1996), using the size reduction ratio informa-
tion obtained in equation 2. The normalized curves for the
PPT products can then be converted to the predicted prod-
uct PSDs of the HPGR.
x +n
{ [
F –––– = 100 1 − exp −A –––– ( X50 ) ]}
x x m –––
(X50
) X50
( ) (3)
x x
where F ––––
x ( ) –––
X50 is the cumulative percent passing at X50 size;
––– is the dimensionless size normalized by dividing by the
X50
median particle size of the product; and A, m, n are fitted
model parameters.
Power and throughput model
The HPGR power draw (P) can be estimated by equa-
tion 4 (Schönert, 1988; Klymowsky, Patzelt, Knechtz, &
Burchardt, 2002) using the roll peripheral speed (u), applied
grinding force (F), and force reaction angle (β). As illus-
trated in Figure 3, β is approximately half the compression
angle alpha (α) and is defined as the position at which com-
pression between the rolls starts (Klymowsky et al., 2002).
α
P = ωT = 2Fusinβ = 2Fusin (––) (4)
2
where P is the power draw in kW, ω is the roll radial veloc-
ity of the HPGR in 1/s, T is the roll torque in kN·m, F is the
grinding force in kN, and u is the roll peripheral speed in
m/s.
For the purpose of predicting power draw, an empirical
model for estimating α was established based on the UBC
HPGR PPT database. The influencing variables for the α
determination are the specific pressing force (Fsp), feed
moisture content (w), and feed size (F80 and F50; equation
5). The HPGR power draw can be also expressed as a func-
tion of the feed material properties and applied pressure,
and can be estimated independently of the PPT results.
α = 9.024 − 0.813Fsp + 0.199w + 0.108F80 − 0.036F50 (5)
As expressed in equation 6 (von Seebach & Knobloch,
1987), the specific throughput constant (ṁ) is defined as the
expected throughput capacity (Q) for an HPGR unit with
rolls 1 m in diameter and 1 m in width running at a periph-
eral speed of 1 m/s (Klymowsky et al., 2002). Equation 6
can then be rewritten as a function of power draw and spe-
cific energy consumption (equation 7), allowing the spe-
 Using piston–die press tests for preliminary sizing and selection of a high-pressure grinding roll 147

cific throughput constant to be estimated based on PPT in m, L is the roll width in m, Pnet is the HPGR net power
results and the HPGR database. draw in kW, Esp is the net specific energy consumption, and
u is the HPGR peripheral roll speed in m/s.
Q Figures 4 and 5 compare power draw and HPGR
ṁ = –––– (6)
DLu throughput capacity, respectively, measured during the
HPGR pilot tests to predicted values obtained using the
netP empirical models. Clearly, the measured power draw values
––––
Q Esp from the pilot tests are strongly correlated to those pre-
ṁ = –––– = –––––– (7)
DLu DLu dicted by the model. Measured and predicted throughput
capacity are also correlated, though to a lesser degree. The
authors believe that the model accuracy of the throughput
where ṁ is the specific throughput constant in ts/hm3, Q is capacity prediction can still be considered appropriate for
the HPGR throughput capacity in t/h, D is the roll diameter estimating HPGR performance during preliminary sizing.

Figure 3. Simplified schematic of forces in a high-pressure grinding roll (HPGR) unit (modified from Torres & Casali, 2009); α: compression angle (°),
β: force reaction angle (°), F: pressing force (kN), T: HPGR torque (kN·m), u: roll speed (m/s)

50
y = 1.0027x
R² = 0.5166
40 p < 0.0001
Predicted throughput capacity (t/h)

30

20

10

0
0 10 20 30 40 50
Measured throughput capacity (t/h)

Figure 4. Linear relationship between total power draw values measured Figure 5. Linear relationship between throughput capacity values
during high-pressure grinding roll pilot testing and power draw values measured during high-pressure grinding roll pilot testing and throughput
predicted by empirical modelling capacity values predicted by empirical modelling

CIM Journal | Vol. 10, No. 3, 2019

148 F. Wang, B. Klein, and P. Rosario
RESULTS AND DISCUSSION
Direct comparison of PPTs and HPGR pilot tests
Figure 6a compares the net specific energy consumption
predicted from PPTs of the composite sample to that mea-
sured during HPGR pilot tests at specific pressing forces.
As expected, the net specific energy consumption increases
linearly with an increase in the specific pressing force. The
PPTs overpredicted the net specific energy consumption at
the lowest specific pressing force, but the predicted and
measured values converge with increasing specific pressing
force. The reproducibility of the PPT method was examined
by repeated testing using subsamples prepared from the
composite. The 95% confidence intervals were calculated
and plotted, indicating an average of 8% error in the PPT
predictions of net specific energy consumption. Figure 6b
shows that the predicted net specific energy consumption
fits reasonably well with the measured HPGR pilot results;
the plotted data fall within ±20% of the 1:1 correlation line.
Figure 7 compares the product sizes (P80, P50, and per-
cent passing 6 mm) of the composite sample measured dur-
ing HPGR pilot tests to those predicted from the PPTs. The
product size prediction was within ±10% of the measured
results. The percent passing 6 mm from the HPGR product
PSD was used to estimate the recirculating load for a closed
HPGR circuit with a 6 mm screen, assuming a 90% screen
efficiency. For the tested specific pressing force range, a
higher specific pressing force provides a greater energy
input to the sample, which results in a smaller product size
(i.e., a greater percentage passing 6 mm) and thus a smaller
recirculating load.
The circuit-specific energy requirements for an HPGR
operation in closed circuit with a 6 mm screen at different
specific pressing force settings were determined using the
estimated recirculating load, assuming a 95% efficiency in
the calculation to account for HPGR drivetrain energy loss.
Figure 8a compares the total circuit-specific energy pre-
dicted from PPTs to that determined from HPGR pilot tests.
Similar to the observation made regarding the net specific
energy consumption, the PPTs also appear to overpredict the
total circuit-specific energy required at the lowest specific
pressing force. To assess the overall throughput capacity of
an HPGR operation in a closed circuit, the HPGR-specific
throughput constant and the corresponding recirculating
load should be considered. This can be determined from the
circuit-specific throughput constant (ṁc), which is equal to
the specific throughput constant divided by 1 plus the recir-
culating load ratio as expressed in equation 8.
ṁ
ṁc = ––––– (8)
1 + CL
where ṁc is the circuit-specific throughput constant in
ts/hm3, ṁ is the specific throughput constant in ts/hm3, and
CL is the recirculating load in %.
CIM Journal | Vol. 10, No. 3, 2019
Figure 8b compares the specific throughput constant
predicted from PPTs to those measured from HPGR pilot
tests, and the circuit-specific throughput constant predicted
from PPTs to those determined from HPGR pilot tests and
equation 8. The PPTs were found to consistently underpre-
dict the specific throughput constant at a given specific
pressing force; the accuracy for throughput prediction lies
within ±20%.
PPT variability
Figure 9a compares the net specific energy consump-
tion predicted from the PPTs performed on the variabil-
ity samples to those measured from the HPGR pilot tests
on the composite sample. The variability samples
required slightly higher specific energy consumption
compared to the composite sample at the same specific
pressing force. The results might suggest at first glance
that the variability samples are harder than the compos-
ite sample; however, plotting the predicted product par-
ticle size (percent passing 6 mm) and the estimated
recirculating load against the HPGR pilot test results
(Figure 9b) does not present conclusive evidence that the
variability samples are truly harder. The predicted prod-
uct particle size of the variability samples is smaller
(i.e., more material passes 6 mm) than that obtained
from the HPGR pilot tests, most likely due to the higher
specific energy input in the PPTs.
Figure 10 compares the total circuit-specific energy con-
sumption calculated from the PPTs on the variability sam-
ples to that measured during HPGR pilot testing of the
composite sample. Visually, the total circuit-specific energy
consumption from the HPGR pilot tests falls into the pre-
diction envelope of the PPTs.
The circuit-specific energy consumption varied little
among the six lithological units tested using the variability
samples. Table 4 presents the measured variation in circuit-
specific energy consumption of the variability samples at a
specific pressing force of 4.5 N/mm2; an average value of
3.87 kWh/t was calculated with a coefficient of variation of
3%.
The database prediction of specific throughput appears
to be moderately conservative compared to the HPGR
pilot testing results (Figure 11). Typically, a higher spe-
cific pressing force corresponds to a smaller operating gap
and results in reduced specific throughput; however, the
Table 4. Variation in circuit-specific energy consumption (at Fsp of 4.5 N/mm2)
required for high-pressure grinding roll (HPGR) comminution of six samples
representing different ore types
Description Unit Value
Number of samples – 6
Mean kWh/t 3.87
Standard deviation – 0.12
Coefficient of variation % 3.2
 Using piston–die press tests for preliminary sizing and selection of a high-pressure grinding roll 149

Figure 6. Comparison of net specific energy consumption for piston–die press tests (PPTs) and high-pressure grinding roll (HPGR) pilot tests on a
composite sample: a) Predicted (PPTs) and measured (HPGR pilot tests) net specific energy consumption at select specific pressing forces (error bars
represent 95% confidence intervals) and b) correlation of predicted (PPTs) and measured (HPGR pilot tests) net specific energy consumption

CIM Journal | Vol. 10, No. 3, 2019

150 F. Wang, B. Klein, and P. Rosario

Figure 7. Effect of select specific pressing forces on the products of high-pressure grinding roll (HPGR) pilot tests and piston–die press tests (PPTs)
using a composite sample (error bars represent 95% confidence intervals); a) predicted 50% (P50) and 80% (P80) passing size of products; b) percent of
product passing 6 mm and estimated recirculating load

CIM Journal | Vol. 10, No. 3, 2019

 Using piston–die press tests for preliminary sizing and selection of a high-pressure grinding roll 151

Figure 8. Comparison of predicted (piston–die press test; PPT) and measured (high-pressure grinding roll [HPGR] pilot test) a) specific energy
consumption and b) specific throughput at select specific pressing forces using composite samples; error bars represent 95% confidence intervals

CIM Journal | Vol. 10, No. 3, 2019

152 F. Wang, B. Klein, and P. Rosario

Figure 9. Comparison of a) net specific energy consumption and b) percent passing 6 mm at select specific pressing forces determined from piston–die
press tests (PPTs) on variability samples and high-pressure grinding roll (HPGR) pilot tests on a composite sample; error bars represent maximum and
minimum values

CIM Journal | Vol. 10, No. 3, 2019

 Using piston–die press tests for preliminary sizing and selection of a high-pressure grinding roll 153

Figure 10. Comparison of predicted (piston–die press test; PPT) and measured (high-pressure grinding roll [HPGR] pilot test) net specific energy (SE)
consumption and circuit-specific energy consumption at select specific pressing forces using variability samples for the PPTs; error bars represent
maximum and minimum values

Figure 11. Comparison of the specific throughput constant and circuit-specific throughput constant at select specific pressing forces determined from
piston–die press tests (PPTs) on variability samples and high-pressure grinding roll (HPGR) pilot testing on a composite sample; error bars represent
maximum and minimum values

CIM Journal | Vol. 10, No. 3, 2019

154 F. Wang, B. Klein, and P. Rosario

product particle size decreases in response to the higher spe- CONCLUSIONS

cific energy input and higher specific pressing force. Over-
predicting the specific energy input at a given specific
pressing force will likely result in an underprediction of spe-
cific throughput. The circuit-specific throughput was
marginally affected by the specific pressing force settings
(Figure 11). Despite some discrepancies in the specific
throughput, excellent predictions of circuit-specific through-
put resulted from the PPTs using the variability samples.
Preliminary sizing and selection
Two calculations were independently performed for pre-
liminary HPGR sizing based on the results of PPTs and
HPGR pilot tests (summarized in Table 5). The higher sizing
parameters arising from the PPT variability assessment were
used to ensure a sufficient margin of safety for the design.
Ultimately, a 2.6 m × 1.7 m HPGR with an installed grind-
ing force of 17,000 kN and a total motor power of 7,400 kW
(two 3,700 kW motors) was recommended for a
42,000 t/day processing plant with a HPGR unit in a tertiary
crushing application. A nominal specific pressing force of
3.5–4.0 N/mm2 was selected.
Piston–die press tests were carried out on ore samples
from a copper porphyry deposit in British Columbia to
determine the suitability of empirical modelling based on
PPTs for HPGR sizing and selection. The results show a
low variability among the tested samples with respect to
high-pressure comminution. Reasonable agreement was
observed between the HPGR pilot test measurements and
empirical model predictions of the specific pressing force,
specific energy consumption, specific throughput con-
stant, and product size distribution. The results indicate
that standalone PPT models are applicable to the HPGR
evaluation and preliminary sizing, provided that the ore
types and projected operating conditions are within the
ranges represented in the UBC HPGR PPT database. The
current UBC HPGR PPT database is limited to tertiary-
and quaternary-stage crushing applications, thus the
developed PPT models might not be suitable for HPGR
fine grinding or iron-ore pellet-feed preparation.
The case study provided in this paper demonstrates the
application of the PPT method to preliminary sizing and
selection of HPGRs for tertiary-stage crushing. The sim-

Table 5. High-pressure grinding roll (HPGR) sizing and selection based on piston–die press test (PPT) and HPGR pilot test results
Description Unit PPTs HPGR pilot tests
Basic data
Ore specific gravity – 2.80 2.80
Daily throughput, dry t/d 42,000 42,000
HPGR availability % 92 92
Hourly throughput, dry t/h 1,902 1,902
Feed moisture content (w) % 3.0 3.0
Sizing parameters
Specific pressing force (Fsp) N/mm2 2.5 3.5 4.5 3.1 4.1 5.0
Net specific energy consumption (Esp) kWh/t 1.8 2.3 2.8 1.8 2.3 2.9
Specific throughput constant (ṁ) ts/hm3 220 218 207 253 247 237
HPGR product passing 6 mm % 69 75 79 68 74 77
HPGR sizing calculation
Hourly HPGR circuit throughput, wet t/h 1,961 1,961
Estimated recirculating load (90% screen efficiency) % 61 49 41 64 51 45
Total circuit-specific energy consumption kWh/t 3.0 3.6 4.1 3.0 3.7 4.4
Total circuit-specific throughput ts/hm3 137 146 147 154 164 164
HPGR roll diameter (estimate) m 2.6 2.6 2.6 2.6 2.6 2.6
HPGR roll width (estimate) m 1.75 1.75 1.75 1.75 1.75 1.75
Roll peripheral speed – required m/s 3.1 2.9 2.9 2.8 2.6 2.6
Roll peripheral speed – max m/s 3.4 3.4 3.4 3.4 3.4 3.4
HPGR total throughput – max t/h 3,411 3,373 3,200 3,913 3,830 3,678
HPGR circuit throughput – max t/h 2,124 2,263 2,272 2,390 2,534 2,539
Throughput margin % 8 15 16 22 29 29
HPGR required power kW 5,812 6,974 8,078 5,928 7,288 8,675
HPGR sizes and quantities
Nominal specific pressure N/mm2 3.5–3.75 3.5–4.0
Number of HPGR units # 1 1
HPGR roll diameter m 2.60 2.60
HPGR roll width m 1.75 1.75
HPGR operating power kW 6,955–7,238 6,494–7,210
Total installed motor power (2 motors per HPGR unit) kW 7,400 (2 × 3,700) 7,400 (2 × 3,700)

CIM Journal | Vol. 10, No. 3, 2019

 Using piston–die press tests for preliminary sizing and selection of a high-pressure grinding roll 155

plicity and accuracy of the PPT models, and the small sam-
ple size required for testing, make this an ideal method for
geometallurgical programs, allowing HPGR performance
to be predicted for varying lithologies or ore types across a
deposit.
The following recommendations with respect to the
application of PPTs are made:
• The empirical models developed and presented here are
valid for the parameter ranges used in the UBC HPGR
PPT database. Continuous effort is required to expand
the database and improve the models.
• The standalone PPT method is suitable for the scoping
and preliminary economic assessment levels of study.
• For sizing and selection of an HPGR at the prefeasibility
study level, the direct calibration method developed by
Davaanyam (2015) is recommended because it is more
accurate. This direct calibration method involves using a
limited number of HPGR pilot tests and PPTs to estab-
lish ore-specific calibration models instead of relying on
the database.
• A full pilot program is recommended for feasibility level
studies to evaluate the effects of other process variables
such as feed top size, roll speed, and feed moisture, and
to acquire the performance guarantee from the vendor.
ACKNOWLEDGMENTS
The authors acknowledge Köppern for allowing them to
use the pilot machine for this research. The authors thank
the mining company who provided the composite sample
and variability samples for testing. Special thanks to the
CIM peer reviewers for their thorough reviews, which
improved the quality of this paper.

An earlier draft of this paper was published in the

NOMENCLATURE Proceedings of the 2019 Annual Meeting of the Canadian
α HPGR compression angle (°) Mineral Processors, prior to undergoing the CIM Journal
β HPGR force reaction angle (°) peer-review process.
D HPGR roll diameter (m)
Esp Net specific energy consumption (kWh/t) Paper reviewed and approved for publication by the
F Pressing or grinding force (kN) Canadian Mineral Processors Society of CIM.
Fsp Specific pressing force (N/mm2)
F50 50% passing size of feed (mm) Fisher Wang is a professional engineer and PhD candidate in mineral
F80 80% passing size of feed (mm) processing at the Norman B. Keevil Institute of Mining Engineering at The
L HPGR roll width (m) University of British Columbia (UBC). His research focus is HPGR
modelling and simulation. He is currently the main operator of the pilot
ṁ HPGR specific throughput constant (ts/m3h)
HPGR facility at UBC and is responsible for all aspects of HPGR testing.
ṁc HPGR circuit-specific throughput constant (ts/m3h) fwang@mining.ubc.ca
P HPGR total power draw (kW)
Pno-load HPGR no-load power draw (kW) Bern Klein is a professional engineer and professor of mineral processing at
Pnet HPGR net power draw (kW) the Norman B. Keevil Institute of Mining Engineering at UBC. Since joining
Ppiston PPT piston pressure (MPa) UBC in 1997, He has focused much of his research on novel comminution
technologies including HPGRs and high-speed stirred mills, as well as
P piston1mm fraction passing 1 mm in the PPT feed (%)
preconcentration technologies such as sensor-based ore sorting. His research
P50 50% passing size of feed (mm) builds upon 8 years of industrial experience, during which he was the project
P80 80% passing size of feed (mm) manager responsible for the design of mineral processes for more than 300
ρbulk HPGR feed bulk density (g/cm3) mining projects.
Q HPGR throughput capacity (t/h)
RR Size reduction ratio = F50/P50 Persio Rosario is a professional engineer, director–comminution at Hatch
Engineering North America, and adjunct professor of mineral processing at
T HPGR torque (kN·m)
the Norman B. Keevil Institute of Mining Engineering at UBC. He has more
u HPGR roll peripheral speed (m/s) than 25 years of combined experience in operations, engineering, and
w HPGR feed moisture content (%) management in the mining and mineral processing industry. He specializes
ω HPGR roll radial velocity (1/s) in comminution, with an emphasis on HPGR technology and innovative
comminution circuit design.

CIM Journal | Vol. 10, No. 3, 2019

156 F. Wang, B. Klein, and P. Rosario
REFERENCES
Davaanyam, Z. (2015). Piston press test procedures for predicting
energy-size reduction of high pressure grinding (Doctoral dissertation).
The University of British Columbia.
Davaanyam, Z., Klein, B., & Nadolski, S. (2015). Using piston press
tests for determining optimal energy input for an HPGR operation. Pro-
ceedings of the 6th International Conference on Semi-Autogenous and
High Pressure Grinding Technology, Vancouver, BC.
Fuerstenau, D. W., Gutsche, O., & Kapur, P. C. (1996). Confined parti-
cle bed comminution under compressive loads. International Journal of
Mineral Processing, 44–45(March), 521–537. https://doi.org/10.1016/
0301-7516(95)00063-1
Fuerstenau, D. W., Shukla, A., & Kapur, P. C. (1991). Energy consump-
tion and product size distributions in choke-fed, high-compression roll
mills. International Journal of Mineral Processing, 32(1–2), 59–79.
https://doi.org/10.1016/0301-7516(91)90019-F
Klymowsky, R., Patzelt, N., Knechtz, J., & Burchardt, E. (2002). Selec-
tion and sizing of high pressure grinding rolls. In A. L. Mular, D. J. Bar-
ratt, & D. N. Halbe (Eds.), Mineral Processing Plant Design, Practice,
and Control Proceedings, Volume 1: Vancouver, BC.
©2019 Canadian Institute of Mining, Metallurgy and Petroleum. All rights reserved.
Lim, I. L., Voigt, W., & Weller, K. R. (1996). Product size distribution
and energy expenditure in grinding minerals and ores in high pressure
rolls. International Journal of Mineral Processing, 44–45(March), 539–
559. https://doi.org/10.1016/0301-7516(95)00064-X
McClintock, M. (2018). Validation of methodologies for sizing high
pressure grinding rollers (Masters thesis). The University of British
Columbia.
Schönert, K. (1988). A first survey of grinding with high-compression
roller mills. International Journal of Mineral Processing, 22(1–4), 401–
412. https://doi.org/10.1016/0301-7516(88)90075-0
Torres, M., & Casali, A. (2009). A novel approach for the modelling of
high-pressure grinding rolls. Minerals Engineering, 22(13), 1137–1146.
https://doi.org/10.1016/j.mineng.2009.04.011
von Seebach, M., & Knobloch, O. R. (1987). High pressure grinding
rolls in industrial application. Paper presented at the 1987 SME Annual
Meeting, Denver, CO.
CIM JOURNAL
Volume 10, Number 3, 2019
Contents

Automation technology to increase 115 K. Moreau, R. Bose,

productivity and reduce energy H. Shang and J. A. Scott
consumption in deep underground
mining operations

Simulating the effect of LHD operations 125 B. Skawina, A. Salama,

on production rates and ventilation and J. Greberg
costs in a sublevel cave underground mine

Reconciling mineral reserves at the 133 G. A. Sutton

well-to-well in-situ copper leaching operation
at San Manuel mine, Arizona, USA

Using piston–die press tests for preliminary 143 F. Wang, B. Klein, and P. Rosario
sizing and selection of a high-pressure
grinding roll

Published by the Publié par l’Institut

Canadian Institute of canadien des mines,
Mining, Metallurgy de la métallurgie
and Petroleum et du pétrole