You are on page 1of 6

International Journal of Rock Mechanics & Mining Sciences 138 (2021) 104595

Contents lists available at ScienceDirect

International Journal of Rock Mechanics and Mining Sciences


journal homepage: http://www.elsevier.com/locate/ijrmms

A new modification to the Kuz-Ram model using the fragment size


predicted by image analysis
Abiodun Ismail Lawal
Department of Mining Engineering, Federal University of Technology, PMB 704, Akure, Ondo State, Nigeria

A R T I C L E I N F O A B S T R A C T

Keywords: Fragmentation of rocks or mineral deposits plays an imperative role in the determination of economic viability of
Rock fragmentation mining operations. The determination of the fragment size is usually done using either the Kuz-Ram model or
Image analysis image analysis. However, the Kuz-Ram model often overestimates the fragment size though it is simple to
Kuz-Ram model
implement, while the prediction of the image analysis is accurate but difficult to implement and costly. Hence,
Least square method
Rock factor
this study proposed a new modified Kuz-Ram model based on the data obtained in the literature. To achieve the
aim of this study, an analytically adjusted original Kuz-Ram model and the image analysis datasets were used to
formulate an objective function which was solved using the least square method implemented in the MATLAB
environment. The proposed model was validated using datasets outside those used in developing the model and
the performance of the proposed model was assessed using the error analysis. The overall error between the
proposed model and the image analysis is about 3.5%, while that of the original Kuz-Ram model and the image
analysis is above 60%. Hence, the proposed model can accurately approximate the fragment size predicted using
image analysis with no cost and easy implementation.

1. Introduction coarse fragments accurately, however, the model can give an unrealistic
prediction of the number of fines generated during blasting.6 In addi­
The effective fragmentation of rocks is a key factor in the determi­ tion, its performance when applied to non-homogeneous rock is not
nation of the economic viability of the rock/mineral excavation as the satisfactory.9 Hence, the Kuz-Ram model usually overestimates the
other stages of mining activities viz.: loading, hauling, and crushing fragment size.
depend on how good the rock is fragmented. A well-fragmented rock To improve the predictions of the Kuz-Ram model, the crush zone
means less energy is required during loading and crushing with little or model (CZM) which applies two different Rosin-Rammler functions to
no cost in secondary blasting. Owing to the importance of the rock estimate the fragment size distribution curve is proposed by Kanchibotla
fragmentation, various researchers have proposed methods for deter­ et al.10 Ouchterlony6 also developed the
mining the size of rock fragment prior to drilling and blasting oper­ Kuznetsov-Cunningham-Ouchterlony (KCO) model to predict the frag­
ations.1–7 The pioneering model was proposed by Kuznetsov2 by ment size distribution by substituting Swebrec function in place of the
developing Rosin and Rammler1 distribution function. Cunningham4 original Rosine-Rammler equation. Gheibie et al.11 proposed a modifi­
later integrated both the Kuznetsov2 and Rosin and Rammler1 distri­ cation to the Kuz-Ram model using the data obtained from Sungun
bution function to come up with a new model generally known as the Copper Mine. However, the Kuz-Ram model and its extensions still
Kuz-Ram model. depend on the use of rock factor which has been identified as the cause
Kuz-Ram model is generally used for the quantification of the rock of the deficiency in Kuz-Ram model.9
fragmentation size prior to drilling and blasting because the model In recent decades, the computer-aided photographic techniques have
combines the blast design parameters and the rock fragment size dis­ been developed for the determination of size distribution of the actual
tribution together.5,8 In the Kuz-Ram model, a rock factor and unifor­ blasting rather than empirical prediction. Wipfrag, FragScan, Split­
mity index are computed which enable quantification of characteristic Desktop, and WIEP are some of the commercially available software for
size and size distribution according to the Rosin-Rammler procedure. the analysis of size distribution of fragments.12,13 The mode of operation
The strength of the Kuz-Ram model is that the model can predict the of the aforementioned software involves the analysis of the digital

E-mail address: ailawal@futa.edu.ng.

https://doi.org/10.1016/j.ijrmms.2020.104595
Received 19 May 2019; Received in revised form 3 October 2020; Accepted 22 December 2020
Available online 4 January 2021
1365-1609/© 2020 Elsevier Ltd. All rights reserved.
A.I. Lawal International Journal of Rock Mechanics and Mining Sciences 138 (2021) 104595

images obtained through the digital cameras. Hence, this process can be Table 1
automated and continuous without obstructing the production cycle.14 Description and rating of the parameters for the rock factor estimation.23
Though, the predicted size distribution of rock fragments through this Symbol Description Rating
approach is more accurate than the empirical models but it has some
RMD Rock Mass Description
inherent limitations15 which include the errors that are inherent in Powdery/friable 10
image processing software. In addition, the method cannot be used Vertically jointed 20
during the design stage and hence, the blasting parameters cannot be Massive 50
pre-controlled using this approach. Also, the approach is difficult to JF Joint factor (JPS + JPA)
JPS Vertical Joint Spacing
employ as compared to the Kuz-Ram model as the cost of acquiring the <0.1 10
software and the expert that will operate it maybe be very high. 0.1 to oversize 20
Therefore, the field engineers rely on the empirical models for the Oversize to Drilling pattern size 50
determination of fragments size and distributions. JPA Joint plane angle
Dip out of the face 20
Machine learning (ML) approaches such as artificial neural networks
Strike perpendicular to the face 30
(ANN), ANN optimized with the artificial bee colony (ANN-BC), ANN Dip into face 40
optimized with the fire fly algorithm (ANN-FA), classification and
RDI density influence 25 × ρr − 50
regression tree (CART), and multiple linear regression16–19 have also
been used to predict the rock fragmentation sizes. However, the ML HF Hardness factor
if Ea < 50 GPa HF = 0.3333E
methods still rely on the image analysis results which has been described
If E > 50 GPa HF = 0.2UCS**
to be time consuming and costly to perform.9 In addition, the ANN
a
method which is found to be more promising among the ML methods Young’s modulus, **uniaxial compressive strength.
that have been used to predict the rock fragmentation sizes is a
black-box approach which makes its practical implementation the run-off mine in the iron ore mine. However, the mode of applications
difficult.20,21 of the image analysis software is similar. They are designed for the
The application of the empirical model is still very common in the computation of the fragment size by analyzing digital images through
routine determination of rock fragment size and distribution. The Kuz­ the use of a digital camera.
netsov2 and its modifications consist of two primary parameters viz.: the
characteristic size and a uniformity index, based on geometric param­ 2.1. Kuz-Ram model for the fragment size determination
eters of the drilling and blast design.22 The characteristic size determines
the fragment size distribution and the uniformity index defines the The most commonly used empirical/analytical model for the deter­
spread of the fragment size.22 In the Kuz-Ram model proposed by mination of the fragment size prior to blasting is the Kuz-Ram model. In
Cunningham,4 it is required to compute the rock factor before the spite of its deficiencies, it is the most commonly used method for the
characteristic size can be determined. The rock factor is computed by determination of fragment size. The proposed study is intended to
multiplying the rock mass characteristics with an assumed coefficient improve the performance of the Kuz-Ram model by modifying its rock
known as a land coefficient.9 Lilly23 suggested value for the land coef­ factor component. Kuznetsov2 proposed the formula relating rock mass
ficient. Lilly23 suggested value can give a realistic prediction in homo­ between the explosive amount (specific charge) and the average size:
geneous rock mass without the discontinuities, while its prediction fails
Xm = AK − 4/5
Q1/6 (1)
when applied to non-homogeneous rock.9 Gheibie et al.11 also proposed
a land coefficient similar to that of Lilly23 to improve the predicted
where Xm is the mean particle size in cm, K is the powder factor in kg/
fragmentation size. Recently, Tosun et al.9 proposed a new modification
m3, A is the rock factor, and Q is mass in kg of the nitroglycerine-based
to the land coefficient for the limestone deposit in Turkey by linear
explosive used per hole.
correlation to predict the Kuz-Ram model with that of the image anal­
The mean fragment size presented in Eq. (1) is for nitroglycerine
ysis. However, in their analysis, the way the constant of the equation of
based explosives which have high detonation velocity. Hence, Eq. (1)
the linear correlation is handled is not stated.
can be corrected with respect to the commonly used explosive (ANFO)
Hence, the main aim of this research is to harness the advantage of
which is less powerful than the nitroglycerine-based explosive, as fol­
the image analysis to correct the anomaly in the rock factor using the
lows (Eq. (2)):
original Kuz-Ram model as the base model coupled with the least square
approach implemented in the MATLAB environment. The proposed Xm = AK − 4/5
Q1/6 (115/SANFO )19/30 (2)
model is expected to, therefore provide solutions to the problem of rock
factor required in the Kuz-Ram model and its modifications. The details where SANFO is the relative weight strength of the explosive to ANFO
of the procedures used in achieving the aim of this study are discussed in (ANFO ≅ 100), the rock factor A is assumed by Kuznetsov to vary be­
the following sections. tween 7 and 13.
Cunningham3 corrected the rock factor by relating it to the blast­
2. Materials and methods ability index proposed by Lilly23 as presented in Eq. (3). It was suggested
that the rock factor should at least take into consideration the physical
The drilling and blasting parameters from the limestone quarry and mechanical properties of the rock and the structural geology.3,9 The
located in Western Anatolia Cement Plant (Batı Anadolu Cement Plant) definition and rating of different parameters in Eq. (3) can be found in
located in Izmir, Turkey and that of Chadormalu iron ore mine located in Table 1.
180 km northeastern of Yazd in Central Iran available in literature were
A = 0.06(RMD + JF + RDI + HF) (3)
used for this study. The limestone data were obtained from the research
of Tosun et al.,9 while that of iron ore were gotten from the research of It is important to note that the coefficient 0.06 in Eq. (3) proposed by
Shad et al.24 The results of the image analysis are available for both the Lilly23 was determined as a land coefficient9 which has been identified
limestone and the iron ore respectively. Fourteen (14) blasts data were as one of the major causes of the deficiencies in the Kuz-Ram model. The
used in all. Eight (8) of which were from the Arkavadi 1st Region, while land coefficient was changed to 0.071 in the modified Kuz-Ram model
the remaining six (6) were from the Chadormalu iron ore mine. Wipfrag proposed by Gheibie et al.11 However, their proposed modification has
software was used to analyse the images of the blasted rock in the not totally rectified the deficiency in the Kuz-Ram model. This is the
limestone quarry, while the Gold size program was used for the image of main reason why there is no large disparity between the original

2
A.I. Lawal International Journal of Rock Mechanics and Mining Sciences 138 (2021) 104595

Table 2 where k is an unknown constant which is to be determined, KR1 and KR2


The image analyses and Kuz-Ram model results.
can be expressed as follows:
Location Type of deposit Source IA (mm) Kuz-Ram
(cm) KR1 = (RMD + JF + RDI + HF) (6)
9
Arkavadi 1st region Limestone Tosun et al. 167.3 25.95
182.3 29.63 KR2 = K − 4/5
Q1/6 (115/SANFO )19/30 (7)
181.9 27.17
188.0 29.02 The unknown constant in Eq. (5) is determined using the least square
163.4 26.39 method of error minimization as presented in Eq. (8).
151.5 23.58 ∑( )2
157.3 23.93 Fobj = IA − Xmnew (8)
164.0 26.65
Chadormalu Iron ore Shad et al.24 175.3 29.22 where IA is the image analysis results (Table 2) and the new mean
168.6 28.10 fragment size Xnewm as presented in Eq. (5)
161.3 26.88
Data obtained from the limestone quarry and iron ore mine are as
120.0 20.00
123.5 20.58 presented in Table 2. Equation (8) is then formulated with the intention
125.5 20.92 to get the value for the unknown constant which gives the lowest
objective function using Newton’s method. The initial required step in
using this method is to compute IA, Eqs. (6) and (7), and then create the
Kuz-Ram model and the modified Kuz-Ram model proposed by Gheibie objective function using Eq. (8). Then the partial derivative of the
et al.11 It is evident in the work of Gadikor25 who optimized the drilling objective function with respect to the unknown for example k1 to kn is
and blasting operations in an open-pit copper mine in USA using both computed and expressed in the form of the Jacobian matrix (e.
original and modified Kuz-Ram model together with the Split-Desktop g. ∂fi /∂xj = ∂(f1 , f2 …fn )/∂(k1 , k2 …kn )). The Newton’s method (xn+1 =
image analysis software. Hence, this study modifies the Kuz-Ram
xn − f(xn )/f (xn ) is then used to iteratively solve the equation for the

model by modifying the land coefficient associated with the rock mass
unknown with an initial guess value. If the guess is good, the convergent
characteristics in Lilly’s23 rock factor (Eq. (3)) using the least square
solution is obtained with the error less than or equal to the predefined
method.
error tolerance and if otherwise, the process will be repeated until the
convergent solution is achieved. More details on Newton’s method can
2.2. Development of the proposed modification be found in Lawal.26 To minimize the mathematical rigor, a MATLAB
code is then developed to solve Eq. (3) using an inbuilt Newton’s
The Kuz-Ram model presented in Eq. (2) can basically be divided function.
into two viz; rock mass characteristics and the blasting parameters as The convergent solution obtained for the unknown variable k in Eq.
presented as follows: (5) is 0.03739. Hence, Eq. (2) can be re-written as in Eq. (9). The major
advantage of Eq. (9) over the existing model is that it combines the el­
Xm = 0.06(RMD + JF + RDI + HF)K − 4/5
Q1/6 (115/SANFO )19/30 (4)
ements from the image analysis and that of the empirical model, hence a
Xmnew = kKR1 KR2 (5) more accurate prediction of the fragment size can be obtained.

Fig. 1. Comparison of the proposed modification with the image analysis and Kuz-Ram model results.

3
A.I. Lawal International Journal of Rock Mechanics and Mining Sciences 138 (2021) 104595

Table 3 Xm = 0.03739KR1 KR2 (9)


Error analysis of the proposed model and Kuz-Ram model.
IA Kuz-Ram % Proposed % 3. Results and discussion
(mm) (cm) Error (cm) Error

Arkavadi 1st 167.3 25.95 55.09 16.17 3.34 3.1. Comparison with the image analysis and Kuz-Ram model
region 182.3 29.63 62.53 18.47 1.29
181.9 27.17 49.39 16.94 6.89 The results of the modified Kuz-Ram model proposed in this study is
188.0 29.02 54.34 18.08 3.81
compared with the results of the image analysis and Kuz-Ram model for
163.4 26.39 61.48 16.44 0.64
151.5 23.589 55.65 14.70 3.00 both the limestone quarry in Arkavadi 1st region of as reported in Ref. 9
157.3 23.939 52.14 14.91 5.18 and Chadormalu iron mine documented in Ref. 24 as shown in Fig. 1. It
164.0 26.65 62.49 16.61 1.27 can be seen that the results of the proposed modification to the Kuz-Ram
Chadormalu 175.3 29.22 66.67 18.21 3.87 model are very close to the accurate results obtained from the image
168.6 28.10 66.67 17.51 3.87
161.3 26.88 66.67 16.75 3.87
analysis. The Kuz-Ram model vary widely from the results of the image
120.0 20.00 66.67 12.46 3.87 analysis which is generally agreed to be the most accurate method of
123.5 20.58 66.67 12.83 3.87 determining the fragmentation size aside the sieve analysis.13 The error
125.5 20.92 66.67 13.04 3.87 analysis was also carried out to further establish the accuracy of the
proposed model (Eqs. (10) and (11)) as presented in Table 3. It can be
seen that the overall average error between the proposed modified
Table 4 Kuz-Ram model and the image analysis is about 3.5% which is much
Error analysis the proposed model and Kuz-Ram model for the Arkavadi 2nd lower than the 60.9% which is the average error between the original
region. Kuz-Ram model and the image analysis results.
IA Kuz-Ram *% Proposed % To further establish the veracity of the proposed model, results of the
(cm) (cm) Error (cm) Error image analysis and that of the original Kuz-Ram model from Arkavadi
Arkavadi 2nd 160.8 25.10 60.64 15.62 2.83 2nd region (Table 4) are compared with the proposed modified Kuz-Ram
Region 169.5 25.15 60.40 15.68 7.50 model as presented in Fig. 2. It can be seen from Fig. 2 that the proposed
145.3 23.84 60.43 14.86 2.27 modification to the Kuz-Ram model is very close to the value obtained
171.3 27.06 60.29 16.88 1.45
from the image analysis. The overall average error (Eqs. (10) and (11))
|IA − Kuz − Ram| between the proposed modified Kuz-Ram model and the image analysis
*.%Error = × 100%
IA result is about 3.5%, while the overall error between the original Kuz-
Ram model and the image analysis result is about 60.4% as shown in
Table 4.
It can be inferred from the results presented in Figs. (1) and (2), and
Table 3 and 4 that the proposed model can give accurate predictions of
the fragment size that are very close to the actual values from the image
analysis. In addition, the proposed modification is much simpler to

Fig. 2. Comparison of the proposed modification with the image analysis and Kuz-Ram model results of the Arkavadi 2nd region.

4
A.I. Lawal International Journal of Rock Mechanics and Mining Sciences 138 (2021) 104595

Table 5 implement with no cost as compared with the image analysis which is
The sources of the data adopted for the application.27 more time consuming and very costly.
Quarry Location Rock Type No of References
|IA − Kuz − Ram|
blasts %Error = × 100% (10)
IA
El Alto quarry, Spain Limestone 1 Segarra and Sanchidrián,28
Sanchidrián et al.29 |IA − Proposed|
Mt. Coot-tha quarry, Hornfels 1 Kojovic et al.,30 LeJuge and %Error = × 100% (11)
Australia Cox31
IA
Bårarp, Sweden Granitic gneiss 7 Olsson and Bergqvist,32
Moser et al.33 3.2. Application of the proposed model
Kållered, Sweden Gneissic granite 6 Gynnemo34
Billingsryd, Sweden 6 Gynnemo34
The veracity of the proposed model is further established by
Rolla quarry, MO, Dolerite 29 Otterness et al.35
USA comparing the results from the modified Kuz-Ram model with the
High Forest quarry, Dolomitic 20 Ash,36 Dick et al.37 original Kuz-Ram model determined using 112 data out of the 169 blasts
MN, USA limestone data reported in the literature27 (Table 5). All the 112 blasts were done
Guan Shan copper Rhyoporphyry 8 Ma et al.38 in the field on an in situ rock mass in both small-scale and large-scale
mine, Jiang Shu,
China
production blasts.27 It is also important to note that where the drilling
Waterloo quarry, IA, Limestone 7 Stagg et al.39 and blasting parameters and rock mass parameters have lower and
USA upper values, their lower and upper limits were used for the computa­
St Paul Park quarry, Dolomite 6 Stagg and Rholl40 tion of the Kuz-Ram model. Similarly, where there was more than one
MN, USA
blasts, the parameters required for the implementation of the Kuz-Ram
Granite Falls quarry, Granitic gneiss 3 Stagg and Otterness41
MN, USA model are the same. In addition, the image analysis results of the blasts
Manitowoc quarry, Dolomite 6 Stagg and Otterness41 data used were not reported in the literature.
WI, USA Fig. 3 shows the result obtained from the original Kuz-Ram model
Vändle quarry, Granite 6 Ouchterlony et al.,42,43 and the modified Kuz-Ram model proposed in this study. It can be seen
Sweden Ouchterlony and Moser,44
Liu et al.45
from the predictions of the fragment size obtained from the original Kuz-
Långåsen quarry, Granodiorite 6 Ouchterlony et al.46,47 Ram that they are higher than those proposed in this study. This is
Sweden consistent with the observations from the previous sections. However,
the error analysis (Eqs. (10) and (11)) performed on the two models
shows that the overall error is 60.47% which is consistent with the
earlier levels of error observed between the original Kuz-Ram model and
the actual results from the image analysis in the previous sections.
Hence, the proposed model can give reasonable prediction of the frag­
ment sizes very close to the predictions of the image analysis with
overall error margin of less than 5%.

Fig. 3. Comparison of the proposed model with the original Kuz-Ram model for the data in Table 4.

5
A.I. Lawal International Journal of Rock Mechanics and Mining Sciences 138 (2021) 104595

4. Conclusions 21 Onifade M, Lawal AI, Aladejare EA, Bada S, Idris MA. Prediction of gross calorific
value of solid fuels from their proximate analysis using soft computing and regression
analysis. Int. J. Coal Prep. Util.. 2019 https://doi.org/10.1080/
This research modified the prediction of the fragment size by the 19392699.2019.1695605, 2019.
Kuz-Ram model using the results of the image analysis. The modification 22 Spathis AT. A correction relating to the analysis of the original Kuz-Ram model.
was achieved through the data obtained in the literature coupled with Fragblast - Int. J. Blast. Frag. 2004;8(4):201–205.
23 Lilly PA. An empirical method of assessing rock mass blastability. Proceedings of Large
the least square approach implemented in the MATLAB environment. Open Pit Mine Conference. Newman: October; 1986:89–92, 1986.
The outcome of the analyses indicated that the land coefficient of the 24 Shad HIA, Sereshki F, Ataei M, Karamoozian M. Investigation of rock blast
rock factor is 0.03739 contrary to the 0.06 proposed by Lilly.23 The fragmentation based on specific explosive energy and in-situ block size. Int J Min Geol
Eng. 2018;52–1:1–6.
overall percentage error between the proposed model and the actual 25 Gadikor J. Optimization of drilling and blasting practices at a western us open pit copper
value is about 3.5%, while the overall error between the original mine. Graduate Theses & Non-Theses. Montana Tech; 2018.
Kuz-Ram model and the actual value is about 65%. This study shows that 26 Lawal AI. A Generalized Lateral Earth Pressures Assessment Model for Earth-Retaining
Structures Supporting Sloping Frictional-Cohesive Backfill. PhD Thesis. City Univ. Hong
the proposed modification has considerably reduced the large disparity Kong; 2018.
between the image analysis results and that of the Kuz-Ram model. 27 Sanchidrián JA, Ouchterlony F. A distribution-free description of fragmentation by
Hence, the rock fragment sizes can be more accurately predicted to a blasting based on dimensional analysis. Rock Mech Rock Eng. 2017;50(4):781–806.
28 Segarra P, Sanchidrián JA. The effect of blast design parameters on fragmentation in El
high degree of accuracy using the proposed model. Alto quarry (Madrid, Spain). Report 92, Less Fines Project, EC Contract No. GlRD-CT-
2000- 00438. Universidad Politécnica de Madrid; 2003.
29 Sanchidrián JA, Segarra P, López LM. A practical Procedure for the measurement of
Declaration of competing interest fragmentation by blasting by image analysis. Rock Mech Rock Eng. 2006;39(4):
359–382.
30 Kojovic T, Michaux S, McKenzie C. Impact of blast fragmentation on crushing and
The authors declare that they have no known competing financial screening operations in quarrying. Proceedings of Explo’95-A Conference Exploring the
interests or personal relationships that could have appeared to influence Role of Breakage in Mining and Quarrying. 1995:427–436. Brisbane; 4-7 September.
the work reported in this paper. 31 LeJuge GE, Cox N. The Impact of Explosive Performance on Quarry Fragmentation. in:
Proceedings of Explo’95-A Conference Exploring the Role of Breakage in Mining and
Quarrying, Brisbane; 4-7. 1995:445–452. September 1995.
References 32 Olsson M, Bergqvist I. Fragmentation in quarries. Proceedings of Discussion Meeting BK
(In Swedish). Stockholm: Swedish Rock Construction Committee; 2002:33–38.
1 Rosin P, Rammler E. The laws governing the fineness of powdered coal. J Inst Fuel. 33 Moser P, Grasedieck A, Olsson M, Ouchterlony F. Comparison of the blast
1933;7:29–36. fragmentation from lab-scale and full-scale test at Bararp. In: Holmberg R, ed.
2 Kuznetsov VM. The mean diameter of the fragments formed by blasting rock. Sov Min Proceedings of EFEE 2nd World Conference on Explosives and Blasting Technique. 2003:
Sci. 1973;9:144–148. 449–458. Prague; 10-12 September 2003.
3 Cunningham CVB. The Kuz-Ram Model for Prediction of Fragmentation from Blasting. in: 34 Gynnemo M. Investigation of Governing Factors in Bench Blasting. Full-Scale Tests at
Proceedings of 1st International Symposium on Rock Fragmentation by Blasting. Sweden: Killlered and Billingsryd. Gothenburg, Sweden: Publ A84, Chalmers University, Dept.
Lulea University of Technology; August 1983:23–27, 439-453. Geo; 1997 (In Swedish).
4 Cunningham CVB Fragmentation Estimations and the Kuz-Ram Model-Four Years on. in: 35 Ottemess RE, Stagg MS, Rholl SA, Smith NS. Correlation of shot design parameters to
Proceedings of the 2nd International Symposium on Rock Fragmentation by Blasting. USA. fragmentation. Proceedings of 7th Symposium on Explosives and Blasting Research, Las
August 1987:23–26, 475-87. Vegas, Nevada. 6-7 February 1991:179–190.
5 Cunningham CVB. The Kuz-Ram Fragmentation Model-20 Years on. in: Proceedings of 36 Ash RL. The Influence of Geological Discontinuities on Rock Blasting. Minneapolis: PhD
3rd World Conference on Explosives and Blasting. Brighton, UK. 2005:201–210, 13–16 Thesis. Univ. Minnesota; 1973.
September. 37 Dick RA, Fletcher LR, D’Andrea DV. A study of fragmentation from bench blasting in
6 Ouchterlony F. The Swebrec© function: linking fragmentation by blasting and limestone at a reduced scale. USBM Report of Investigation RI 7704. US Bureau of Mines;
crushing. Min. Tech.-Trans. Inst. Min Met.: Sect. A. 2005;114(1):29–44. 1973, 1973.
7 Ouchterlony F. The case for the median fragment size as a better fragment size 38 Ma B, Zeng S, Zou D, Guo C. A Study of Bench Blasting in Rhyoporphyry at a Reduced
descriptor than the mean. Rock Mech Rock Eng. 2016;49(1):143–164. Scale and the Statistical Analysis of the Regularity for Fragmentation Distribution. in:
8 Afum BO, Temeng VA. Reducing drill and blast cost through blast optimization-A Proceedings of Lst Intemational Symposium on Rock Fragmentation by Blasting.
case study. Ghana Min. J. 2015;15(2):50–57. Luleå, Sweden; 23-26 August 1983. pp. 857–872.
9 Tosun A, Konak G, Toprak T, Karakus D, Onur AH. Development of the Kuz-Ram 39 Stagg MS, Rholl SA, Otterness RE. The Effect of Explosive Type and Delay between Rows
model to blasting in a limestone quarry. Arch Min Sci. 2014;59(2):477–488. on Fragmentation. in: Proceedings of 15th Conference on Explosives and Blasting
10 Kanchibotla SS, Valery W, Morrell S. Modelling fines in blast fragmentation and its Technique. New Orlean; January. 1989:353–366.
impact on crushing and grinding. Proceedings of Explo’99-A Conference on Rock Breaking. 40 Stagg MS, Rholl SA. Effects of accurate delays on fragmentation for single-row blasting in
Kalgoorlie; 7-11 November 1999:137–144. a 6.7 m (22-ft) bench. Proceeding of 2nd International Symposium on Rock Fragmentation
11 Gheibie S, Aghababaei H, Hoseinie SH, Pourrahimian Y. Modified kuz-ram by Blasting. Keystone, CO; 23-26 August 1987:210–223.
fragmentation model and its use at the Sungun copper mine. Int J Rock Mech Min Sci. 41 Stagg MS, Otterness RE. Screen analysis of full-scale production blasts. Proceedings of
2009;46(6):967–973. 11th Symposium on Explosives and Blasting Research. Nashville; 5-9 February 1995:
12 Ozkahraman HT. Fragmentation assessment and design of blast pattern at Goltas 298–313.
Limestone Quarry, Turkey. Int J Rock Mech Min Sci. 2006;43(4):628–633. 42 Ouchterlony F, Olsson M, Nyberg U, Potts G, Andersson P, Gustavsson L. Optimal
13 Kabwe E. Velocity of detonation measurement and fragmentation analysis to Fragmentering I Krosstäker-Fältförsök I Vändletäkten. Report 1:11. Stockholm: MinBas
evaluate blasting efficacy. J. Rock Mech. Geotech. Eng. 2018;10:523–533. Project; 2005 (in Swedish).
14 Sanchidrián JA, Segarra P, Ouchterlony F, López LM. On the accuracy of fragment 43 Ouchterlony F, Olsson M, Nyberg U, Andersson P, Gustavsson L. Constructing the
size measurement by image analysis in combination with some distribution Fragment Size Distribution of a Bench Blasting Round, Using the New Swebrec Function.
functions. Rock Mech Rock Eng. 2009;42(1):95–116. in: Proceedings of 8th Intemational Symposium on Rock Fragmentation by Blasting
15 Maerz NH, Zhou W. Calibration of optical digital fragmentation measuring systems. (Fragblast 8). Santiago. 2006:332–344, 7-11 May.
Fragblast-Int. J. Blast. Frag. 2000;4(2):126–138. 44 Ouchterlony F, Moser P. Likenesses and Differences in the Fragmentation of Full-Scale
16 Enayatollahi I, Bazzazi AA, Asadi A. Comparison between neural networks and and Model-Scale Blasts”. Proceedings of 8th International Symposium on Rock
multiple regression analysis to predict rock fragmentation in open-pit mines. Rock Fragmentation by Blasting (Fragblast 8). Santiago. 2006:207–220, 7–11 May 2006.
Mech Rock Eng. 2014;47:799–807. 45 Liu H, Lindqvist PA, Akesson U, Kou SQ, Lindqvist JE. Characterisation of Rock
17 Ebrahimi E, Monjezi M, Khalesi MR, Armaghani DJ. Prediction and optimization of Properties Using Texture-Based Modelling- a Geometallurgical Approach. in: Proceedings
backbreak and rock fragmentation using an artificial neural network and a bee of Conference Mineral Engineering, Luleå. 2011:8–9. February 2011.
colony algorithm. Bull Eng Geol Environ. 2016;75:27–36. 46 Ouchterlony F, Nyberg U, Olsson M, Vikstriim K, Svedensten P. Optimal Fragmentering
18 Asl PF, Monjezi M, Hamidi JK, Armaghani DJ. Optimization of flyrock and rock I Krosstiikter, Fültfürsiik I Langasen. Report 2010:2, MinBas Project 1.2.1: Optima!
fragmentation in the Tajareh limestone mine using metaheuristics method of firefly Fragmentering Vid Spriingning 11. Stockholm: Swebrec-Swedish Blasting Research
algorithm. Eng Comput. 2018;34:241–251. Center; 2010 (in Swedish).
19 Mehrdanesh A, Monjezi M, Sayadi AR. Evaluation of effect of rock mass properties on 47 Ouchterlony F, Nyberg U, Olsson M, Widenberg K, Svedensten P. Effects of Specific
fragmentation using robust techniques. Eng Comput. 2018;34:253–260. Charge and Electronic Delay Detonators on Fragmentation in an Aggregate Quarry,
20 Lawal AI, Idris MA. An artificial neural network-based mathematical model for the Building KCO Design Curves. in: Proceedings of 11th Intemational Symposium on Rock
prediction of blast-induced ground vibrations. Int J Environ Stud. 2019. https://doi. Fragmentation by Blasting (Fragblast 11). Sydney. 2015:24–26. August 2015. pp.
org/10.1080/00207233.2019.1662186. 727–739.

You might also like