Вы находитесь на странице: 1из 15

Engineering Geology 87 (2006) 60 74

www.elsevier.com/locate/enggeo

Prediction of fragmentation and yield curves with


reference to armourstone production
John-Paul Latham a,, Jan Van Meulen b , Sebastien Dupray c
a

Department of Earth Science and Engineering, Imperial College London, London SW7 2AZ, United Kingdom
b
Boskalis, PO Box 43, 3350 AA Papendrecht, Holland, The Netherlands
c
CETE de Lyon, LRPC, Groupe Mcanique des roches, 69674 Bron Cedex 01, France
Received 8 March 2006; received in revised form 16 May 2006; accepted 23 May 2006
Available online 1 August 2006

Abstract
Armourstone production involves aspects of blast design and yield prediction. How they differ from methods drawn from
experience in mining and aggregates blasting operations is examined. A number of possible blast fragmentation models and
associated prediction methods are described, several being outlined in full. Their applicability to armourstone production and yield
curve prediction is discussed by comparing model results based on a hypothetical armourstone blast design in a rock mass with
realistic properties for an armourstone quarry. It is suggested that appropriate models for armourstone yield prediction will require
some form of an in-situ block size distribution assessment. Such approaches rule out the standard application of the KuzRam
model. The recently reported Swebrec function and associated prediction model, developed by the Swedish Blasting Research
Centre, provides a promising replacement for the RosinRammler based models for representing armourstone blast yield curves.
2006 Elsevier B.V. All rights reserved.
Keywords: Armourstone; Yield curve; Blasting; Fragmentation; Model; Quarry

1. Engineering context
Efficient production of construction materials and the
quest for improved quarrying techniques have a strong
link with civil engineering, engineering geology and
rock mechanics as explained in a companion paper
(Latham et al., 2006). The CIRIA/CUR (1991) rock
manual on coastal and shoreline engineering included
brief references to some possible methods for armourstone production and yield curve prediction. In its second
Corresponding author. Tel.: +44 2 7594 7327.
E-mail addresses: j.p.latham@imperial.ac.uk (J.-P. Latham),
j.a.vanmeulen@boskalis.nl (J. Van Meulen),
sebastien.dupray@equipement.gouv.fr (S. Dupray).
0013-7952/$ - see front matter 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.enggeo.2006.05.005

edition (CIRIA, CUR, CETMEF, 2007) it has sought to


update and expand on the special problems faced by
practicing engineers wishing to plan on the basis of
predicted armourstone yields, but such a publication
cannot include a satisfactory discussion of the related
research, much of which is very recent. Engineers
working for the first time on an armourstone project,
face many potential pitfalls. Most will therefore wish to
obtain maximum assistance from geologists for those
aspects of blasting that the production engineer cannot
control. Understandably, research literature on rock
excavation by blasting is spread amongst the mining
and rock mechanics journals. This paper is thus a response
to the need to describe and compare the most promising
prediction techniques often applied to higher energy

J.-P. Latham et al. / Engineering Geology 87 (2006) 6074

aggregates and mine production methods and see how


they fare when applied to the low energy fragmentation
blasts associated with armourstone production as is the
concern of civil engineers.
The quantification of the percentages of blocks
bounded by joints or bedding planes within a rock mass
and of sufficient size to be useful in breakwater armour
layers was addressed in a recent companion paper
(Latham et al., 2006). The prediction of the in-situ block
size distribution (IBSD) was presented as a vital first
step towards better prediction of the blasted block size
distribution (BBSD), commonly termed the yield curve
or fragmentation curve in quarrying and mining. The
motivation for this study of the factors governing the
yield curves of armourstone quarries supplying breakwater projects comes from two sources. First, the
production engineer needs tools (i.e. BBSD models) to
help achieve the blasting objectives and there are not
many of these in existence that were designed for low
fragmentation blasting. Second, the breakwater designer is fully aware that a much more cost effective design
can be specified given an accurate prediction of the
quarry yield.
The Espevik quarry shown in Fig. 1 excavates a
slightly metamorphosed granite gneiss. It was originally
opened with the intention of fully exploiting its armourstone potential while having the option of selling the
undersize as aggregates. The type of IBSD and BBSD
analysis methods discussed later in this paper and its
companion were vital elements in winning the case for
the investment to go ahead and open the quarry. Special

61

armourstone blasting techniques were developed with


low benches of 10 m, 4.5 m burden and 3.0 m spacing.
Based on a BondRam analysis (described later), the
predicted yield was greater than 50% exceeding 1 t
assuming a specific charge of 0.23 kg/m3. The actual
average production over several weeks of armourstone
blasts generated the following: 10 t, 83% passing; 5 t,
60% passing, 3 t, 49% passing; 1 t 40% passing an even
higher percentage of armourstone than predicted.
In this paper, we introduce the subject of rock blasting
in sufficient detail to present the main differences between
various BBSD prediction models potentially suited to the
range of low energy blasts associated with armourstone
production. We begin with a brief introduction to those
blasting factors that concern armourstone and aggregates
production. The fragmentation process is briefly described to better understand the basic differences between
aggregates and armourstone blast design and practical
measures often found useful to maximise the yield of
armourstone are listed. Starting with the widely employed
KuzRam model and ending with a model based on a new
three-parameter function, the Swebrec function, for the
yield curve which replaces the RosinRammler equation,
various models are presented. Practical methods for the
assessment of blastpiles (muckpiles) to check and feedback for further calibration of predictions are also given.
The differences between various models predicting the
BBSD curves for a given hypothetical armourstone blast
design and a given IBSD assessment provide the basis of a
discussion on suitability of approaches for armourstone
blast yield prediction.

Fig. 1. The Espevik Quarry in Norway, as seen during loading of a 20,000 t barge in January 1992, see text.

62

J.-P. Latham et al. / Engineering Geology 87 (2006) 6074

2. Factors affecting blasting for armourstone and


aggregates

blasting for improved or reduced yields of heavy


blocks in dedicated quarries.

Certain aspects of armourstone production require


attention to details that are not usually emphasised in
the extensive literature on blasting, e.g. Persson et al.
(1993), JKMRC (1996), Jimeno et al. (1997). The
focus of armourstone production is on larger blocks
than for normal fragmentation blasting. The aim of
any blast is to produce rock of the size and form that
will facilitate subsequent operations such as crushing
and lead to minimum overall costs. The blast design is
a significant process in securing desired fragmentation
but there are many difficulties to overcome, not least
because there are many factors affecting fragmentation
beyond the control of the blast engineer. These factors
were investigated by Lilly (1986) and Lizotte and
Scoble (1994) and considered in the formulation of a
blastability index by Latham and Lu (1999).
Uncontrollable factors: geological characteristics of
the rock mass or effects of rainfall

The economics of the second case are constrained by a


need to produce, as far as is possible, only the material
demanded by the design. This may require that secondary
breakage is embraced fully as a means of production when
setting the blasting objectives. (This is the theme of the
final section of the companion paper, Latham et al., 2006).

discontinuity spacings and orientation (bedding,


joints, faults, cohesion across planes). In-situ block
size distributions can be assessed using the techniques
described elsewhere (e.g. Latham et al., 2006)
strength and elasticity (rock type, weathering
characteristics)
density, porosity, permeability
presence of water in blastholes, fractures and joints
spatial variations of geology and rock types in general
Controllable factors:
properties and detonation methods of the explosives
used, including delay timings
blast design (configuration and drilling pattern).
Successful blasting engineers work to clearly defined
objectives such as: the required size distribution results,
ease of blastpile digging, minimum disruption to next
blast, etc. They apply theoretical understanding of the
rock fragmentation process and rock characteristics,
knowledge of the effects of using different explosives
and detonation techniques, the environmental constraints
and lastly, experience and expertise in combining these
which may include the assistance of blasting software.
The most important fragmentation objectives for
armourstone blasts are:
blasting for improved yields of heavy blocks in specially set aside faces of aggregates quarries

2.1. Fragmentation processes


The way in which in-situ bedding, jointing and other
discontinuities slice up the natural rock mass into blocks
of predefined shape distributions and size distributions
prior to blasting is illustrated in virtually every exposure
of rock. The concentrated release of energy from explosives detonated in confined blastholes, transforms the
IBSD to a BBSD of finer material (Fig. 2). To summarise
the consensus of blasting research based on references
such as those mentioned above; the sudden very high
phase transformation pressures in the blast detonation
causes shock wave transmission, compressive crushing
near the borehole walls, radial tensile fracturing and
slabbing tensile cracking at free faces. Fracturing and
fragmentation is accompanied by detonation gas flow
into cracks, extending them further. The explosive gas,
assisted by gravity, heaves the blocks away from the
face and into the blastpile. The ability to achieve a desired
BBSD depends on knowledge of the IBSD, the strength
and persistence of the natural geological flaws and:
other uncontrollable factors such as strength, elasticity and density that contribute to the inherent ease of
breakage or blastability of the rock,
blast energy mobilised through the blast design.
2.2. Comparison of armourstone and aggregates blast
design
Design of an aggregates blast aims to minimise
excess oversize (and expensive secondary breakage)
keeping the average BBSD to < 10% exceeding about
3 t while ensuring not too much rock is reduced to
useless fines by excessive blast energy. Typically, a
specific charge for ANFO explosives of 0.4 to 0.7 kg/
m3 (kg of explosive per cubic metre of in-situ rock) is
used in a two or three row shot with delays to achieve
sufficient breakage. Blasting engineers working on
armourstone operations for the first time will soon
become aware of the fundamental differences in
fragmentation results compared with aggregates blasts.

J.-P. Latham et al. / Engineering Geology 87 (2006) 6074

63

Fig. 2. Illustration of theoretical scenarios for an aggregates blast and an armourstone blast applied to the same rock mass. IBSD and BBSD are
represented by RosinRammler curves.

Fig. 2 is a schematic diagram showing the essential


differences in yield curves. RosinRammler coefficients were used to illustrate blast curves corresponding
with typical results from the aggregates industry and
breakwater contractors. The theoretical curves are
defined later in the text. In terms of practical blast
design, many simple steps to improve armourstone
production are discussed below. The most fundamental
for an armourstone blast, is a low specific charge:
0.2 kg/m3 is often used and even lower values may help
achieve the objectives. For a rapid insight into the
practical and technical factors governing production in
an armourstone quarry dedicated to producing breakwater materials, see Van Meulen (1998).
2.3. Suggestions for improving the yields
of armourstone

2. The spacing to burden ratio should generally be less


than or equal to 1 with burden larger than the discontinuity spacing in a jointed rock mass. Ratios as low
as 0.5, more typically associated with pre-splitting, have
been successfully applied to armourstone operations.
3. If the bench is either too high or too low, armourstone
production will be poor. For an initial estimate, bench
height could be selected as two to three times the
burden. In planning bench levels, the rock mass from
which most armourstones might be produced, such as
thickly bedded layers, should be located nearly at the top
of the bench alongside the stemming section of the
holes.
4. A large stemming length, larger than the burden, is
usually recommended.
5. A small blasthole diameter of less than 100 mm is
recommended.

Generally, the proportion of armour-sized blocks in


the blast increases with increasing intact tensile strength,
increasing Young's Modulus and increasing discontinuity spacing. Normal blasting practice (e.g. for aggregates and ores) aims to achieve high fragmentation
blasts. By contrast, greater percentages of armourstone
can be achieved by adjusting common practice through
consideration of the following list based on the original
research by Wang et al. (1991). Note, blast terminology
is provided in Fig. 3.
1. A low specific charge. Generally, a specific charge as
low as 0.11 to 0.25 kg/m3 can be used. If possible, the
explosive used should have lower velocity of detonation, (VOD). For such low specific charges, maintaining high drilling accuracy is more critical to avoid
insufficient rock break out.

Fig. 3. Geometric blast design parameters.

64

J.-P. Latham et al. / Engineering Geology 87 (2006) 6074

6. One row of holes is found to be better than multi-rows.


If permitted, holes should be fired instantaneously
rather than using inter-hole delaying, but may cause
high undesired ground vibration.
7. A bottom charge of high energy concentration is
needed for the bottom to cleanly break away.
8. A decoupled column charge of ANFO packed in
plastic cartridges (sausages) is often effective when a
3003000 kg mass range is the main product size
required, the explosives are then evenly distributed
giving quite even fragmentation.
9. A decked charge, to break up the continuity of explosives, will be necessary in most situations when
armourstone greater than 3 t is required. The material
for decking can be either air or aggregates.
The most common objective of an armourstone bench
blast is to achieve a BBSD with the maximum percentage
of the largest blocks possible. Such a blast is to cause the
minimum of new fractures while having sufficient energy
concentration to fully loosen the in-situ blocks and bring
the rock face down cleanly. The best achievable BBSD
curve will lie close to and just to the left of the upper part
of the IBSD curve, spreading out considerably at lower
sizes. Where the mean discontinuity spacing gives vast
in-situ blocks, blast design must ensure sufficient
breakage to limit the proportion of blocks above 20 t,
which is about the limit for practical handling.
It is interesting to note that a lower percentage of
armourstone recovery can often be more economical,
even though more rock is eventually excavated and
therefore more is left behind as over-production because of poorer rates of production. Excavation of the
blastpile and keeping good faces and toes becomes more
difficult the greater the yields of heavy armourstone and
the lower the specific charge. The rate of output from
excavators, loaders and selection plant also becomes
slower.
3. Prediction of yield curves
It is not uncommon for disputes to develop over
liability for unforeseen materials production costs where
contracts are based upon dedicated armourstone quarries.
Average yield curves derived from back analysis that
represent results of materials tonnages supplied to a
breakwater project as different classes of stone (e.g. core,
underlayer, various armour classes etc.), often reveal that
in practice, the final percentage of armourstone blocks
(e.g. >3 t) was less than the predicted curves suggested.
Prediction of blasted block size distributions, BBSDs
(fragmentation curves, yield curves) is the subject of

significant research effort as the possible error in prediction remains very high. Accuracy is limited because
the geological conditions cannot easily be determined for
every blast and the implementation of the blast design
may suffer from practical constraints. For dedicated
armourstone quarries, early prediction of quarry yield
curves, whether by trial blasts, or by scanline and borehole
discontinuity surveys together with blast modelling, plays
a vital part in breakwater design optimization. Described
below are four approaches to fragmentation prediction:
KuzRam model, implemented in many software
applications
BondRam models
EBT model
KuznetsovCunninghamOuchterlony (KCO)
models
3.1. KuzRam model
Cunningham brought Kuznetsov's (1973) work up to
date, introducing the KuzRam Model in 1983. Later
revisions to KuzRam, Cunningham (1987), included
improved estimation of the rock mass factor A based on
Lilly's (1986) blastability index. There are four important equations that by simple substitution of parameters, give the BBSD curve. The use of the KuzRam,
or similar models, requires caution. Factors of recognised importance such as detonation delay timing are
not included in KuzRam (a large literature on timing
effects exists, some indications are given by Chung and
Katsabanis (2000)) while the effect of rock mass
structure, and the burden to spacing ratio needs careful
consideration (Konya and Walter, 1990).
(i) RosinRammler Equation: is the cumulative form
of the Weibull distribution and provides the basic shape
of the BBSD to be expected, in terms of the 50% passing
sieve size in the blastpile, Db50 and RosinRammler
uniformity index for sizes, nRRD, giving the fraction
passing, y, corresponding to a certain sieve size Dy
(Rosin and Rammler, 1933).
y 1expf0:693Dy =Db50 nRRD g

After Db50 and nRRD have been determined from


Eqs. (2) and (3) below, substitution of Dy values will
return fraction passing values from which the complete
BBSD curve can be deduced. For a BBSD prediction
focused on armourstone sizes of say, 0.1 m to between 1
and 2 m, the RosinRammler equation is considered the
most attractive simple choice. It should be noted that
where data from sieved or photo-analysed blastpiles
deviate surprisingly from the RosinRammler fitting

J.-P. Latham et al. / Engineering Geology 87 (2006) 6074

function near the maximum sizes, this could be due to


the inherently poor sampling of the coarsest fraction
which can throw the measured results out from the
average production in question. Shortcomings of the
RosinRammler equation include:
It has been reported to sometimes give a poor fit to
blastpiles with high yields of armourstone sizes
(Lizotte and Scoble, 1994).
It fails to give a clear maximum size because the
function is asymptotic to the 100% passing value.
It is commonly unable to describe with reasonable
accuracy the fines content below sizes of about
50 mm in a blast (Ouchterlony, 2005a); of particular
concern for predicting the detailed nature of the
quarry run and the resultant behaviour of core
materials derived from the quarry.
(ii) Kuznetsov Equation: gives Db50 (in m not cm) as
a function of (A, V, Q, E), which locates the position of
one point on the BBSD curve. Essentially, this suggests
that average size is controlled by specific charge
Db50 0:01d AV =Q0:8 d Q0:167 d E=1150:633

where:
A

V
E
Q/V

= rock factor; = 1 for extremely weak rock, A = 7


for medium rock, A = 10 for hard, highly fissured
rock; A = 13 for hard, weakly fissured rock.
Several schemes similar to rock mass rating
systems now exist for improved estimation of A.
For example, Lilly's (1986) original blastability
algorithm was adopted by Cunningham (1987).
= charge concentration per blast hole (kg); to
calculate, consider borehole volume filled and
density of explosive.
volume of rock broken per blast hole (m3)
relative weight strength of explosive (ANFO =
100, TNT = 115);
Specific Charge (kg/m3), a general measure of
explosive power in the blast

Spathis (2004) pointed out an implicit assumption in


Cunningham's KuzRam application of Kuznetsov's
original equation. The assumption is increasingly invalid
for lower nRRD values typical of armourstone blasts
because the mean size differs more significantly from the
median size as nRRD decreases. Spathis plotted the
correction needed as a function of nRRD which indicates
that for nRRD as low as 0.8, the characteristic size would be
1.8 times too large if Eq. (2) is used without the correction.

65

(iii) Cunningham's uniformity index algorithm:


Cunningham (1987) developed an empirical equation
that determines nRRD i.e. the steepness of the BBSD
curve, as a function of blast design geometry with
terms independent of those in Eqs. (1) and (2). Note,
there is no significant body of evidence from
physically measured sieved distributions to support
this equation, although it remains widely used as a
tool. The first term typically takes a value of 1.5 and
is a base term about which the other terms for hole
patterns, drill deviation, different column and base
charges and charged proportion of bench, are all
correction terms.
nRRD 2:214B=dd f0:51 S=Bg0:5 d 1W =B
d absBCLCCL=L 0:10:1 d L=H
3
where
d
B
S
BCL
abs
CCL
L
H
W

= borehole diameter (mm)


= burden (m),
= spacing (m),
= bottom charge length (m),
= absolute value of
= column charge length (m),
= total charge length above grade (m),
= bench height or hole depth (m).
= standard deviation of drilling error (m),

(iv) Rock factor A: This section provides a guide to


setting parameters of the rock mass from which the rock
mass factor A, needed for the KuzRam and KCO
models, can be estimated. It would be a very rare rock
mass that could achieve A-values above 14 and for rock
to be considered for armourstone, it is considered likely
that A would fall in the range of 914.
A 0:06RMD JF RDI HF

where
RMD = Rock mass description = 10 if powdery or
friable, = JF if vertically jointed, = 50 if massive rock
JF = Joint Factor = Joint Plane Spacing term (JPS)
+ Joint Plane Angle term (JPA)
JPS = 10 if average Principal Mean Spacing, PMS
(e.g. cube root of product of three principal mean
spacings) < 0.1 m, 20 if average PMS is within range
0.1 m to 1 m, 50 if average PMS > 1 m.
JPA = 20 if dipping out of face, 30 if striking
perpendicular to face, 40 if dipping into face
RDI = Rock Density Influence = 0.025r(kg/m3) 50

66

J.-P. Latham et al. / Engineering Geology 87 (2006) 6074

HF = Hardness factor = E / 3 if E < 50, or UCS/5


if > 50, depending on uniaxial compressive strength
UCS (MPa) or Young's Modulus E (GPa).

mined by grinding experiments. Such index values may


be misleading if used directly in blast models without a
correction factor.

3.2. BondRam models

3.2.2. BRM(B)
Chung and Katsabanis (2000) demonstrated that Eq.
(3) gave nRRD values consistently too high compared to
results they studied from sieved blastpiles of small scale
blasts. They suggested linking Db50 determined from
Kuznetsov's Eq. (2) with Db80 determined from Bond's
theory, as a means to obtain nRRD in the RosRam equation, Eq. (1), thus providing an alternative to Cunningham's Eq. (3). In so doing, nRRD, as given analytically by
0.842 / (lnDb80 lnDb50) together with Db50 given from
Kuznetsov's equation, were found to provide RosRam
coefficients in Eq. (1) that generated final BBSD
prediction curves fitting closer to field data. This Bond
Ram approach was proposed by Chung and Katsabanis
(2000). However, they assumed the 80% passing in-situ
size to be infinite which is an unnecessary restriction
when an estimated IBSD curve, suggesting Di80 perhaps
of 1 to 2 m, has been derived. Here, the use of realistic (not
infinite) Di80 in Eq. (5) is suggested. The consequent
increase in nRRD values obtained from using realistic insitu sizes compared with applying their assumption can be
quite significant. This may provide a partial explanation
for the low nRRD values of the two sieve-measured fullscale quarry examples quoted by Chung and Katsabanis
which gave predicted nRRD of 0.77 and 0.73 when the
measured values were 0.81 and 0.85 respectively. It is
worth noting that when using this approach, the uniformity index is no longer given as a function of blast
design geometry as implied by Eq. (3). It would appear to
be a promising yield prediction approach for armourstone
production and is termed BRM(B) in this paper.
It should be pointed out that to produce more accurate
BondRam predictions, further calibration of an appropriate value for Wi is recommended for quarry bench
blasting of armourstone. Da Gama (1983) suggested the
use of Eqs. (6) and (7), a relation from empirical studies on
a small data set of blasts in a basalt quarry. From a back
analysis of case histories, results presented in Lu and
Latham (1998) suggested a somewhat lower range of
values e.g. Wi = 6.7 1.1 kW h/t for one particular Carboniferous limestone quarry and Wi = 10 4 kW h/t for
host rock from various ore mining blasts. Lower values of
Wi imply greater ease of blasting into small pieces. Blasting engineers wishing to adopt the Bond equation for
blasting are advised to consult recent research, e.g.
Kariman et al. (2001) to constrain the wide choice from
the high values suggested by Da Gama for basalt
(25 kW h/t) and the significantly lower value of

Da Gama (1983) applied Bond's Third Theory of


comminution to blasting using Bond's relation, (Eq.
(5) below) to fix the 80% passing size in the blast.
Bond's relation was applied together with the Rosin
Rammler Eq. (1), and Cunningham's uniformity
coefficient in Eq. (3), by Wang et al. (1992a,b). They
called this combined approach the BondRam model.
It is termed BRM(A) in this paper while another more
recent approach by Chung and Katsabanis (2000) is
termed BRM(B).
3.2.1. BRM(A)
Bond equation: based on Bond's third theory of
Comminution, the reduction in the 80% passing size
during blasting is expressed in terms of the blast energy
qei and a material property, Wi as follows:
qei 10dWi d f1=MDb80 1=MDi80 g

To apply BRM(A), qei and Wi values together with


Di80 from IBSD information are substituted in Eq. (5) and
Db80 is determined. Substituting y = 0.8 and Dy = Db80,
together with nRRD, determined from Eq. (3), in Eq. (1),
then gives Db50, from which the complete BBSD curve
of RosinRammler form can be deduced where:
Db80 and Di80 are the 80% passing sieve sizes, after
blasting and in-situ respectively (in microns)
qei is the energy required for fragmentation and is a
function of (E, V, Q, r). It can be estimated from

qei 0:00365 E Q=V =qr

where E = weight strength of explosive (%) relative


to ANFO and r = rock density in t/m3
Wi (in kW h/t) is analogous to Bond's Work Index for
grinding but is here calibrated for blasting (Da Gama,
1983) as follows:
Wi 15:42 27:35Di50 =B

where B = burden (m), Di50 = 50% passing in-situ


block size (m) and the empirical fit coefficients have
the appropriate units.
Note: in grinding, work index values are known from
tables for grinding of different ores, or they are deter-

J.-P. Latham et al. / Engineering Geology 87 (2006) 6074

67

Fig. 4. Use of a three point method to characterise fragmentation and demonstrate the decrease in D50 with increasing specific charge data from
quarry analysis by J Van Meulen.

10 kW h/t as suggested above and recently by Chung and


Katsabanis (2000), or calibrate their own case-specific Wi
that provides a consistently good fit to observed fragmentation in the quarry.
3.3. EBT Model
Lu and Latham (1998) developed an energy-blocktransition (EBT) model for BBSD prediction based on
relating the area between the IBSD and BBSD curves to
the energy consumed in transforming bigger blocks to
smaller ones. First, the IBSD curve is predicted (see
methods suggested in Latham et al., 2006) giving the
mean in-situ block size, kai. A measure of the intrinsic
blastability of the rock mass, known as the EBTcoefficient, Bi must then be obtained, together with the
energy input, qei, see Eq. (6). Procedures for doing so
are described in Latham and Lu (1999) where a
blastability designation BD 10 / Bi was proposed for
obtaining Bi. Alternatively, they recommended that
when procedures for deriving BD needed to be made
much simpler, the schemes for calculating rock factor A
could be used, where Bi = 10A / 13. The mean blasted
block size, kab can then be obtained from the EBT
model, Eq. (8) as follows:
qei

Bi

kai kab

kai kab 0:5

A predicted BBSD curve can then be obtained.


Further to the discussion by Spathis, 2004, it should be
noted that ka is the mean and as such will not be well
approximated by the median for low uniformity indices.
The reader can find relations relating median and mean
for different distributions in statistics texts but the
Spathis paper is an ideal starting point.

3.4. KCO model


In recognition of the poorer fit in the fines region of
the two-coefficient RosRam and power law equations, more complex equations with four or five curve
fitting coefficients have been introduced. These curve
shapes can overcome the underestimate of fines often
found with RosinRammler curves and are designed
to account for more complex combinations of
breakage mechanisms such as fine-scale crushing
near the borehole, fines development occurring along
propagating branching cracks, and the coarser fragmentation by tensile cracking (Djordjevic, 1999;
Kanchitbotla et al., 1999). More recently, and with
no reduction in curve fitting accuracy compared with
the four- and five-coefficient equations, Ouchterlony
(2005b) proposed a 3-parameter cumulative size
distribution function, termed the Swebrec function,
given here as Eq. (9)
y 1=f1 lnDbmax =Dy =lnDbmax =Db50 b g

where Db50 is given from Eq. (2), (where the rock mass
factor A found from Eq. (4) is required), and b is called
the curve undulation parameter. Ouchterlony suggests
Dbmax which is the upper limit to the blasted fragment
sizes can be taken as equal to the largest in-situ block
size, Di100 or either the burden or spacing if smaller than
Di100. When introducing the correct blast parameters
into Eq. (9), the equation becomes a BBSD prediction
model. The KCO (KuznetsovCunninghamOuchterlony) model was proposed as a suitable name for the
model. Ouchterlony has proposed two methods for
predicting the value for b.
The first is to adopt Cunningham's nRRD from Eq. (3)
but to also introduce an effect recognised by Ouchterlony, that the size distribution's slope at the Db50 point is

68

J.-P. Latham et al. / Engineering Geology 87 (2006) 6074

also dependent on Db50 itself. A good approximation for


b was found to be:
b nRRD d 2d ln2d lnDbmax =Db50

10

The second is to use an empirical equation derived


from sieved results from several full-scale blasts where
Db50 is in mm, (Ouchterlony, 2005a):
b 0:5D0:25
b50 d lnDbmax =Db50

11

Ouchterlony (2005b) shows how the function presented in Eq. (9) fits BBSD sieving results from a wide
range of rock types and blast conditions remarkably well
and plugs into the KuzRam model with ease,
improving predictive capability in the fines range and
the cut-off at the upper limit, especially if a good Di100
estimate can be substituted for Dbmax. Ouchterlony suggests that Dbmax could be set at the minimum of burden,
spacing or in-situ maximum size.
It is suggested that the KCO model offers great
potential to improve on the KuzRam model in most
bench blasting operations. Its suitability for armourstone
blasts also looks quite promising. For armourstone blast
prediction, as with all prediction models, it should be
applied with caution, especially as it has been developed
for blasts with relatively higher specific charges and burden to spacing ratios than is common for armourstone
blasts. It should also be noted that many unconventional
blasting methods such as decoupling and simultaneous
detonation are used for armourstone blasts. Accuracy of
the KCO model and the function given by Eq. (9) has not
been examined as thoroughly in the 80100% passing size
range (where it is most critical for armourstone prediction),
as it has for the medium and smaller sizes considered more
important for productivity in high fragmentation blasts.

material volumes are logged during production through


the selection plant (e.g. trommel screen). Provided the
coarsest proportion from the blast can be estimated, for
example by counting blocks in heavy grading classes or as
described above, a curve based on assessment at three
points can be drawn. In Fig. 4, three important points on
the yield curve were used to chart the change in BBSD
while reducing specific charge. Screen analysis of fullscale production blasts, clearly more reliable than image
analysis for assessments, were reported in Stagg and
Otterness (1995) and in Ouchterlony (2005a,b).
4.2. Image analysis
A discussion on predicted fragmentation would be
incomplete without some mention of assessment methods
to examine actual fragmentation. Automated image analysis methods are becoming more widespread for determining blastpile size distributions in mining and quarrying
operations. Digital photos taken while piles are being
loaded, (so as to represent the full depth of the pile) and
taken from above loaded trucks, provide input that readily
available image analysis software will convert into size
distributions using sophisticated correction algorithms. A
blind trial of various commercial image analysis software
packages (Latham et al., 2003) gives a snapshot of their
performance. Fig. 5 shows images with known size distributions of the type often used to calibrate image analysis
software. Franklin and Katsabanis (1996) compiled a
monograph of papers and references to such methods.
At least half a dozen commercial automated sizing
systems are now in widespread use, not only for blast
yield assessment, but also for production control of
processed minerals. There is potential for wider use of
such systems in quality control of gradings, e.g. barge
deliveries of light gradings.

4. Assessment of mass distributions


4.3. Photo-scanline methods
4.1. Direct screening and block measurement methods
It may sometimes be practical to count the number of
blocks N in the entire potential armourstone oversize
material in a blast, and to perform measurements of block
dimensions from a representative sample of say N / 5
blocks. The sizes can be converted to masses using shape
factors based on blockiness concepts (e.g. see Gauss and
Latham, 1995). Knowing the total rock mass in the blast
and estimating the total mass in the oversize, the upper
part of the BBSD can be plotted, and may be merged with
photo-scanline or image analysis results.
In a production with no crushing, it is possible to assess
the proportions in a blast if it is all processed. The sorted

An alternative manual photographic method (Lu and


Latham, 1996) that is simple and can be undertaken
without software is to superimpose scanlines directly on
the scaled photographs. The method was employed by
McKibbins (1996) to assess the armourstone blast
(4000 t) shown in Fig. 6. Many scanlines are drawn
on each photo with directions chosen to minimise bias.
Care is needed to correct for perspective distortion. A
single length distribution from measurements of segment lengths defined by intersections between the particle edges is created from all the photos making up a
representative sample. It is invariably found that the
cumulative form of this length distribution has a Rosin

J.-P. Latham et al. / Engineering Geology 87 (2006) 6074

69

Fig. 5. Typical size distributions with similar appearance (representative of gradations in blastpiles if scale divisions = 1 m). P44: nRRD = 0.7,
D63.2 = 800 mm, D50 460 mm, P41: nRRD = 0.9, D63.2 = 350 mm, D50 240 mm. The same distributions are shown in Fig. 2 Note, for high
fragmentation blast geometry in Eq. (3), the KuzRam model often predicts nRRD > 1.0. With low nRRD laboratory piles it is very difficult to make up
a sample big enough to properly represent the larger sizes present in a perfect RosinRammler distribution in effect, even the artificially made up
piles have a partly bimodal distribution due to large size censoring.

Rammler form. The best fit photo-scanline Rosin


Rammler parameters nRRDp, D63.2p, for uniformity and
characteristic length can be obtained from a linearized
plot. To convert the RosinRammler to a linear form,
substitute the left hand side of Eq. (12) as the variable Y
and logDp as the variable X and apply linear regression
of Y on X to obtain the gradient and intercept which give
nRRDp and D63.2p.
logln1=1y
nRRDp d logDp nRRDp d logD63:2p

12

The calibration equations to convert from segment


length distribution coefficients to nRRD and D63.2 are:
D63:2 1:119D63:2p

13

nRRD 1:096nRRDp 0:175

14

As for any assessment of blastpiles that only sample


the surface-visible blocks, the results are likely to give
coarser BBSD predictions than is representative of the

entire pile. Taking many sample photographs during


blastpile loading is preferable.
5. Comparison of BBSD prediction models
The purpose here is not to explore the range of validity
of each model, and compare it with well documented case
histories. (With the equations given above, the reader can
simply implement the model formulae and compare yield
curve results for various local quarry and blasting conditions using a spreadsheet). Instead, a hypothetical case
of one potentially reasonable armourstone blast design,
applied to a hypothetical rock mass viable for armourstone is considered to illustrate some of the models discussed above. The chosen hypothetical rock mass is
equivalent to a widely jointed but not especially massive
rock mass typical of a competent limestone of density
2.7 t/m3. It is given a rock factor A of 10 with an IBSD
analysis as given in Table 1. A somewhat less steep IBSD
might be appropriate for a more massive rock mass with
locally disturbed fractured areas.

Fig. 6. Photographic image of a blastpile in the Hulands Quarry, Co Durham UK, used for photo-scanline analysis. The surface of the blast pile
contains many blocks from the stemming section and is somewhat coarser than the material beneath.

70

J.-P. Latham et al. / Engineering Geology 87 (2006) 6074

Table 1
Assumed IBSD for illustration

Table 3
Parameters for KuzRam models

Fraction passing

Mass (kg)

D sieve (m)

0.1
0.3
0.5
0.7
0.8
0.9
0.95
1

710
1795
3423
5902
7933
10,892
14,105
23,242

0.763
1.039
1.288
1.545
1.705
1.895
2.066
2.440

Example applications of the KuzRam model are


shown for a suggested armourstone blast design with
parameters as shown in Table 2 and the results of the
model in terms of RosRam yield parameters are as
shown in Table 3. Interestingly, for these chosen charge
length to borehole length ratios, burden and spacing values, it appears that Cunningham's Eq. (3) is in this case
successful in the sense that it gives a reasonably low
uniformity index in the same range commonly observed
for armourstone blasts. In applying KuzRam to unconventional blasts, it is often difficult to judge how best to
convert the subtleties of special armourstone techniques
such as air-decking, decoupling, delay timings into model
parameters. In selecting the values given in Table 3, the
blast assumes a fully coupled long base charge with no
column charge as such, but a long stemming length. To
address the idea that the upper rock mass blocks are
simply liberated in armourstone blasts, McKibbins (1996)
attempted a BBSD prediction by summing block sizes
given by the IBSD of the upper stemmed part of the blast
(assumed to be liberated and untransformed in the blast),
with those block sizes transformed by a blast model for the
lower charged part of the hole. This composite model
approach gave less satisfactory results than the Bond
Ram approach.
Table 2
Suggested armourstone blast parameters
KuzRam model input parameters

Suggested armourstone blast

Rock factor A ()
Specific charge Q / V (kg/m3)
Spacing / burden ()
Borehole diameter (m)
Burden (m)
Spacing (m)
Charged column length (m)
Bench height (m)
No. of holes
Volume of rock blasted (m3)
Explosive weight strength
Charged explosive (kg)
St. dev. of drill error (m)

10
0.266
0.61
0.082
4.1
2.5
9
15
10
1538
100
404
0.1

RosRam
coefficients

KuzRam SFB
(KR)

Shifted KuzRam SFB


(sKR)

Vb50 (m3)
Mb50 (kg)
Db50 (m)
nRRM ()
nRRD ()

0.386
1013
0.859
0.265
0.795

0.0763
206
0.505
0.265
0.795

A very important observation is that because of the low


value of nRRM for many armourstone blasts used on
breakwater projects (nRRD typically from 0.7 to 0.9, see
Latham et al., 2006), a significant shift in sizes is predicted
with the Spathis correction. Within this range of nRRM, the
routine application of KuzRam gives yield sizes that are
a factor of about 1.8 too large because of significant
differences between mean and median for such wide
distributions (see Spathis, 2004). The shift is shown in
Fig. 7, however both yield prediction curves appear
unrealistic when compared with the reasonable IBSD
curve we have assumed for the rock mass.
To apply the standard BondRam model BRM(A), the
blast in Table 1 uses the same uniformity index as in the
KuzRam model. The Bond equation requires input from
IBSD at 80% passing together with the blast input energy
and importantly, the Work Index. Two example values,
Wi = 24 and Wi = 10 kW h/t are examined in Fig. 8, the
former derived from Da Gama's calibration, the latter
being closer to expected values for a blast in competent
limestone, see Table 4.
When adopting the BondRam model, sBRM(B),
the blast geometry approach to finding nRRM via Eq. (3)
is discarded. In our hypothetical example, both of the
sBRM(B) slopes shown in Fig. 8 have higher
uniformity than were found using BRM(A) and Eq.
(3), which is opposite to the original finding of Chung
and Katsabanis (2000) where higher specific charge
blasts (0.8 kg/m3) were considered. Substituting a
lower Work Index shifts the 80% passing value
fractionally more. Because in both cases the Mb50 is
pinned by Eq. (2) it therefore gives a slightly steeper
yield curve. Note, to obtain Mb50 the Spathis correction
has been applied. Without it, the results for BRM(B)
would appear impossibly steep for this example. It is
interesting to note that although the Mi80 value has been
used in the analysis, there is an unsatisfactory lack of
convergence of the initial top mass Mi100 and the final
top mass Mb100 for all but the sBRM(A) Wi10 curve.
This effect would be less pronounced and the BBSD
results more generally acceptable had the chosen IBSD
been given a less steep curve.

J.-P. Latham et al. / Engineering Geology 87 (2006) 6074

71

Fig. 7. Application of the KuzRam model to the hypothetical blast and rock mass parameters (Table 2), showing how the correction identified by
Spathis (2004) results in a major shift in the final predicted yield curve (sKR). Neither result appears compatible with the suggested IBSD curve.

The Swebrec function and KCO model removes this


problem by setting the before and after top masses
equal but the uncertainty at the 100% value is always
quite large. The KCO methodology takes no further
notice of the predicted IBSD curve's actual form except
in a rather subjective manner through the JPS score for
the rock factor A. The two Swebrec function curves
take two quite different paths between Mb50 and Mb100
depending upon whether the new empirical KCO or the
first KCO model is used to obtain the undulation
parameter b (see Fig. 9 and Table 5). Note that in both
cases the Spathis correction has been applied to obtain
Mb50. The data set used by Ouchterlony (2005b) to
calibrate the empirical relation Eq. (11) may not be
sufficient to represent low energy blasts typical of
armourstone with the same level of confidence as for
aggregates and mine blasts. Speculating, an undulation
parameter giving a yield curve about half way between

the two shown would seem more reasonable for such


an armourstone blast.
The wide variation between prediction model results
shown in Figs. 79, for one set of blast design data, is
testimony to the difficulty of BBSD prediction, especially for the case of armourstone production.
6. Discussion
Experience to date does not point to a single best
prediction method for armourstone. The best practice is
somewhat clearer for prediction in higher fragmentation blasts for mines and aggregates quarries. This is
because the number of documented studies with high
accuracy in the blastpile assessment (accuracies
associated with sieving a sample of the full-scale
blast or a well controlled image analysis campaign
using several magnifications), together with detailed

Fig. 8. Application of two different BondRam models showing results broadly compatible with the IBSD, and a large dependence on the Work Index
value (Wi) assumed, see text.

72

J.-P. Latham et al. / Engineering Geology 87 (2006) 6074

Table 4
Parameters for BondRam models

Table 5
Parameters for KCO models

BondRam
SFB BRM SFB BRM SFB a BRM SFBa
model parameters (A) Wi10
(A) Wi24
(B) Wi10
BRM(B)
and results
Wi24

Swebrec function inputs

Work index
(kW h/t)
Specific charge
(kg/m3)
Vi80 (m)
Mi80 (kg)
Di80 (m)
Vb50 (m3)
Mb50 (kg)
Db50 (m)
nRRM ()
nRRD ()
a

10

24

10

24

0.266

0.266

0.266

0.266

2.938
7933
1.705
0.0424
114.463
0.4134
0.265
0.795

2.938
7933
1.705
0.0123
33.21
0.2737
0.265
0.795

2.938
7933
1.705
0.0764
206.2
0.5051
0.353
1.058

2.938
7933
1.705
0.0764
206.2
0.5051
0.327
0.980

Note, the Spathis (2004) correction has been applied.

IBSD and rock mass analysis, has been growing.


However, even the large amount of case studies
reported remains a relatively small database if all the
blast design variables are to be investigated. Field data
in the literature from low energy blasts, where the
objective is often simply to liberate in-situ blocks for
use as armourstone, are much scarcer.
If a reasonably confident estimate of rock mass factor
A can be made, but discontinuity spacing is poorly
known, the KuzRam model will provide a complete
prediction curve but with a value for nRRD that has a
reputation in general blast designs for being too high,
thus generally under-predicting the amount of fines
produced.
To take advantage of the known importance of the insitu discontinuities, it is invariably worth the investment
in IBSD data and at least, to estimate the maximum and
typical in-situ block volumes. Drawing upon the IBSD

Vb50 (m )
Mb50 (kg)
Db50 (m)
nRRM ()
nRRD ()
b ()
Dbmax (m)

KCO Eq. (10)

New KCO Eq. (11)

0.0763
206
0.505
0.265
0.795
1.88
2.44

0.0763
206
0.505

3.73
2.44

methods and conclusions of (Latham et al., 2006), the


weighted joint density method of Palmstrm (2001)
using drill core data appears to be a suitable first
approach in poorly exposed green field sites when
scanline surveys and photographic face mapping are
impossible.
If a thorough site investigation can reveal the essential
variations of the in-situ rock mass properties, the IBSD
curve giving 100, 80 and 50% passing values will help the
blast prediction considerably. The BondRam and EBT
models make good use of the whole IBSD and if the work
index Wi or Bi is well calibrated for the rock mass in
question, these approaches look promising as they do not
rely on an accurate determination of maximum in-situ size.
The BondRam model focuses on the 80% passing sizes,
which in practice, have great significance for armourstone
production. Also, Mb80 can potentially be determined with
more accuracy than Mb100 during early assessment stages
of the actual production, from which further calibration and
refinement of models can take place.
The KCO model approach appears to be suitable for
predicting the smaller sizes (below 50 mm) of any blast
which has great implications for quarry waste in breakwater projects as discussed in Latham et al. (2006). It
also appears that if IBSD analysis methods are used to

Fig. 9. Swebrec function and application of two forms of the KCO model (Ouchterlony, 2005a,b) showing compatibility with the IBSD, see text.

J.-P. Latham et al. / Engineering Geology 87 (2006) 6074

provide a reasonably confident estimate of Di100 = Db100 = Dbmax it may work well for armourstone blasts. It
is suggested that there may be advantages to resetting
the Swebrec function so that it operates with a Db90 or
Db95 for the input parameter, because of the increasing
lack of confidence with the determination of the IBSD
and thus BBSD as they approach the 100% value.
However, this raises the further problem of how to relate
Di95 to Db95.
At present, with the KCO model one must chose from
two approaches offered for setting the undulation
parameter b. It has been seen how each one can give
very different proportions of large blocks for the part of
the curve between Db100 and Db50. Future research results
to test the simpler empirical Eq. (11) and the successful
setting of objective values for rock mass factor A and
Dbmax will help evaluate the use of the three parameters in
the Swebrec function and whether the new KCO model is
indeed the best on offer for armourstone blast prediction.
8. Concluding remarks
The KuzRam model is not appropriate for prediction of BBSD for armourstone blasts because no reference to the IBSD is given to constrain the location of the
RosinRammler curve. Typical armourstone blasts have
lower uniformity coefficients than high fragmentation
blasts and therefore it is vital that the correction identified by Spathis (2004) is applied to all uses of the
Kusnetsov equation in BBSD models for armourstone
production.
Blastability approaches such as the BondRam and
EBT models have the advantage of using IBSD information in the relationship that governs the location of
the BBSD curve, however, the intrinsic blastability coefficients suggested for use with different rock masses
remain poorly calibrated for these blastability models.
The introduction of the Swebrec function, and the
KCO model by Ouchterlony (2005a) has advanced
considerably our ability to predict the fines content in
routine blasts. In allowing the slope of the curve at Db50
to be a function of Db50 the curve takes a more realistic
path. For armourstone blasts, yield curve prediction with
the KCO model looks promising but now requires
further validation with case histories.
We are a step closer to making predictions for armourstone blast yield curves with the confidence needed
by practitioners. A summary of experience (see Table 3,
Latham et al., 2006) learned by breakwater contractors
working with production engineers when opening
armourstone quarries, will continue for some years to
be a primary source from which to predict yield curves.

73

Acknowledgements
This paper extends the content of work presented in
the Rock Manual (CIRIA/CUR/CETMEF, 2007) and is
printed with kind permission of CIRIA/CUR/CETMEF.
The authors are grateful for comments provided during
review and the motivation provided by the Rock Manual
team.
References
CIRIA/CUR. 1991. Manual on the use of rock in coastal and
shoreline engineering. CIRIA special publication 83, CUR report
154, 607p.
CIRIA, CUR, CETMEF, 2007. The Rock Manual. The Use of Rock in
Hydraulic Engineering (Second Edition). C683. CIRIA, London.
Chung, S.H., Katsabanis, P.D., 2000. Fragmentation prediction using
improved fragmentation formulae. FRAGBLAST The International Journal for Blasting and Fragmentation 4 (2), 198207.
Cunningham, C.V.B., 1987. Fragmentation estimations and the Kuz
Ram model four years on. In: Fourney, W.L., Dick, R.D. (Eds.),
Proc. Second Int. Symp. on Rock Fragmentation by Blasting.
SEM, Bethel CT, pp. 475487.
Da Gama, D.C., 1983. Use of comminution theory to predict fragmentation
of jointed rock mass subjected to blasting. Proc. First Int. Symp. on
Rock Frag. by Blasting, Lulea, Sweden, August 1, pp. 563579.
Djordjevic, N., 1999. Two-component model of blast fragmentation.
Proc 6th Int Symp on Rock Fragmentation by Blasting, Symposium series S21. SAIMM, Johannesburg, pp. 213219.
Franklin, J.A., Katsabanis, T. (Eds.), 1996. Measurement of Blast
Fragmentation. AA Balkema, Rotterdam, p. 315.
Gauss, G.A., Latham, J.-P., 1995. The on-site quality control of the asproduced properties of armourstone during construction of a rock
revetment in South Devon. Geotechnical and Geological Engineering 13, 2949.
Jimeno, C.L., Jimeno, E.L., Carcedo, F.J.A., 1997. Drilling and
Blasting of Rocks. AA Balkema, Rotterdam.
JKMRC, 1996. Open Pit Blast Design Analysis and Optimisation.
Julius Kruttchnitt Mineral Research centre, Indooroopilly, Australia, p. 338.
Kanchitbotla, S.S., Valery, W., Morell, S., 1999. Modelling fines in
blast fragmentation and its impact on crushing and grinding. In:
Workman-Davies, C. (Ed.), Proc Explo 1999 Conference.
AusIMM, Carlton, VIC, pp. 137144.
Kariman, A., Ozkan, S.G., Sul, O.L., Demirci, A., 2001. Estimation of
the powder factor in bench blasting from the Bond work index.
Transactions of the Institution of Mining and Metallurgy (Sect A:
Min. Technol.) 110, A114A118.
Konya, C.J., Walter, E.J., 1990. Surface Blast Design. Prentice hall,
Englewood Cliffs, New Jersey, p. 303.
Kuznetsov, V.M., 1973. The mean diameter of fragments formed by
blasting rock. Soviet Mining Sciences 9 (2), 144148.
Latham, J.-P., Lu, P., 1999. Development of an assessment system of
blastability for rock masses. International Journal of Rock
Mechanics and Mining Sciences 36, 4155.
Latham, J.-P., Kemeny, J., Maerz, N., Noy, M., Schleifer, J., Tose, S.,
2003. A blind comparison between results of four image analysis
systems using a photo-library of piles of sieved fragments.
FRAGBLAST The International Journal for Blasting and
Fragmentation 7 (2), 105132.

74

J.-P. Latham et al. / Engineering Geology 87 (2006) 6074

Latham, J.-P., Van Meulen, J.A., Dupray, S., 2006. Prediction of in-situ
block size distributions with reference to armourstone for breakwaters. Engineering Geology 86, 1836.
Lilly, P.A., 1986. An empirical method for assessing rock mass
blastability. In: DAVIDSON (ed.) Proc AusIMM/I.E. Aust. Newman Combined Group Large Open Pit Mining Conference.,
Victoria, 4144, 8992.
Lizotte, Y.C., Scoble, M.J., 1994. Geological control over blast fragmentation. CIM Bulletin 87 (983), 5771.
Lu, P., Latham, J.-P., 1996. Estimation of blasted block size distribution
of a blastpile combining photo-scanline technique with RosRam
and Schuhmann models. In: Guo, Y., Golosinski, T.S. (Eds.),
Proceedings of the '96 Int. Symposium on Mining Science and
Technology, Jiangsu, China. Balkema, Rotterdam, pp. 683688.
Lu, P., Latham, J.-P., 1998. A model for the transition of block sizes
during blasting. FRAGBLAST The International Journal for
Fragmentation and Blasting 2, 341368.
McKibbins, L.D. 1996. An assessment of armourstone potential at
Hulands Quarry. MSc Thesis, London University, Queen Mary and
Westfield College p 222.
Ouchterlony, F., 2005a. What Does the Fragment Size Distribution of
Blasted Rock Look Like? In: Holmberg, R. (Ed.), I Proc 3rd EFEE
World Conference on Explosives and Blasting. European Federation of Explosives Engineers, UK, pp. 189199.
Ouchterlony, F., 2005b. The Swebrec function: linking fragmentation
by blasting and crushing. Mining Technology (Trans. Inst. Min.
Metal A) 114, A29A44.
Palmstrm, A., 2001. Measurement and characterization of rock mass
jointing. In: Sharma, V.K., Saxena, K.R. (Eds.), In-situ Characterisation of Rocks. AA Balkema, pp. 4997.

Persson, P.A., Holberg, R., Lee, J., 1993. Rock Blasting and Explosives
Engineering. CRC Press, Boca Raton, Florida, USA.
Rosin, P., Rammler, E., 1933. The laws governing the fineness of
powdered coal. Journal of the Institute of Fuel 7, 2936.
Spathis, A.T., 2004. A correction relating to the analysis of the original
KuzRam model. FRAGBLAST The International Journal for
Blasting and Fragmentation 8, 201205.
Stagg, M.S., Otterness, R.E., 1995. Screen analysis of full-scale production blasts. Proc. Eleventh Annual Symposium on Explosives
and Blasting Research. International Society of Explosives
engineers, Nashville Tennessee.
Van Meulen, J.A., 1998. Opening and operation of the Pasir Panjang
quarry: a dedicated armourstone quarry in the Malaysian jungle.
In: Latham, J.-P. (Ed.), Advances in Aggregates and Armourstone
Evaluation. Engineering Geology Special Publication, vol. 13.
Geological Society, London, pp. 107120.
Wang, H., Latham, J.-P., Poole, A.B., 1991. Blast design for armourstone production. J. Quarry Management. Part I (July), 1721, Part
II (Aug), 1922.
Wang, H., Latham, J.-P., Matheson, G.D., 1992a. Design of fragmentation blasting in surface rock excavation. In: Hudson, J.A. (Ed.),
Proceedings of the ISRM Symposium: EUROCK '92. Balkema,
Rotterdam, pp. 233238. Chapter 41.
Wang, H., Latham, J.-P., Poole, A.B., 1992b. Producing armourstone within aggregate quarries. In: Magoon, O.T., Baird, W.F.
(Eds.), Durability of Stone for Rubble Mound Breakwaters.
ASCE, pp. 200210.

Вам также может понравиться