See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/342753635

Evaluation of empirical approaches in estimating mean particle size after

blasting by using nondestructive methods

Article  in  Arabian Journal of Geosciences · July 2020

DOI: 10.1007/s12517-020-05636-9

CITATIONS READS

5 89

2 authors:

Esma Kahraman Ahmet Mahmut Kilic

Cukurova University Cukurova University
50 PUBLICATIONS   59 CITATIONS    72 PUBLICATIONS   465 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Acil Durum Eylem Planı View project

P and S velocity View project

All content following this page was uploaded by Esma Kahraman on 27 December 2021.

The user has requested enhancement of the downloaded file.

Arabian Journal of Geosciences (2020) 13: 613
https://doi.org/10.1007/s12517-020-05636-9

ORIGINAL PAPER

Evaluation of empirical approaches in estimating mean particle

size after blasting by using nondestructive methods
Esma Kahraman 1 & Ahmet Mahmut Kilic 1

Received: 4 September 2019 / Accepted: 22 June 2020 / Published online: 7 July 2020
# Saudi Society for Geosciences 2020

Abstract
It is well-known that rocks have heterogeneous properties, and these properties played a curial role in blasting operations to
obtain fragmented rocks with different particle size distributions. Therefore, the estimation of the particle size distribution of the
rock is one of the most important parameters in terms of the economic operation of open-pit mines as it is directly related to
mining operations such as loading, transporting, crushing, and grinding. Many parameters divided into controllable and non-
controllable affect the efficiency of blasting operations. This study aimed to suggest new empirical models based on nondestruc-
tive test methods including P wave velocity and Schmidt hardness to predict the mean particle size of the rock after blasting.
WipFrag software was used to evaluate the pictures taken after blasting and to determine the average particle size distribution of
the rock mass. Experimental results obtained from simple and multiple regression models show that there is a good agreement
observed between the mean particle size of rock mass and its Schmidt hammer hardness and P wave velocity properties.

Keywords Blasting fragmentation . Mean particle size . P wave velocity . Schmidt hardness

Introduction parameters such as spacing, powder factor, specific charge,

burden, and hole depth as input parameters at rock fragmen-
Blasting is one of the most common used methods in mining tation prediction (Monjezi et al. 2009; Kulatilake et al. 2010;
and civil engineering projects, and the particle size distribu- Bahrami et al. 2011; Singh and Abdul 2012; Mehrdanesh et al.
tion range of the product obtained after blasting varies in line 2018; Akbari et al. 2015; Shams et al. 2015).
with the area to be used. The efficiency of blasting affects Other researchers associated study field geology and rock
other operations such as conveying, crushing, and grinding. mass characteristics with fragmentation. The influence of rock
Fragmentation after blasting should generally be desired par- mass properties on blasting operations was studied by long
ticle size, uniform, and highly economic. years, and these properties are more important for blasting
The affordability of the blasting and the estimation of the fragmentation. Jimeno and Jimeno (1995) showed that the
particle size distribution after blasting were evaluated by many strength of rock mass depends upon the geological features
researchers. Studies and discussions still continue on this and the strength of the intact rock. In addition, researchers
subject. indicated that densities and strengths of rocks are generally
Some researchers associated the blasting design parameters correlated. Widzyk-Capehart and Lilly (2002) tried to deter-
with fragmentation. Researchers utilized blasting design mine the rock properties that affect the explosibility. In their
study, the most important rock properties affecting the deto-
nation were revealed as modulus of elasticity, compressive
Responsible Editor: Murat Karakus
strength, tensile strength, Poisson ratio, bulk modulus, pres-
* Esma Kahraman
sure waves, and shear waves of rock material. Kılıc et al.
[email protected] (2009) determined the unit volume weight, water absorption,
uniaxial compressive strength, tensile strength, cohesion, and
Ahmet Mahmut Kilic internal friction angle of the rocks in Kırka borax mine and
[email protected] established a statistical relationship with part size distribution
1
after blasting. As a result of their analysis, the relationship
Department of Mining Engineering, Cukurova University, 01330
Balcali, Saricam, Adana, Turkey
between fragmentation and rock characteristics was found.
613 Page 2 of 8 Arab J Geosci (2020) 13: 613

Adana Cement Inc.

Turkey

Adana

Study Field

Fig. 1 Location of the study field

Saliu et al. (2013) indicated observed good correlation be- however, it is possible to determine most properties, especial-
tween blast efficiency and uniaxial compressive strength. In ly the strength properties, only under laboratory conditions. In
their study, it was observed that blasting efficiency increases addition to laboratory studies is time-consuming, labor-
with decrease in uniaxial compressive strength. intensive and costly. In this study, mean particle size after
Rock mass properties inside mine quarries are quite hetero- blasting was estimated by using Schmidt hammer hardness
geneous. This case prevents the operation at the same standard and P wave velocity values, which is known as nondestructive
in some operations such as drilling and blasting. method and gives information about rock properties in the
Fragmentation after blasting affects the formation transitions study area. In this article, regression analysis model was de-
and structural changes within the same area. The part size veloped to estimate mean particle size resulting from blasting
distribution of the batch obtained after blasting has a signifi- operations in the Caldag raw material field of Adana Cement
cant effect on the operating economy. Knowing the rock prop- Inc. At these models, nondestructive methods as input param-
erties will make it easier to estimate the fragmentation; eters to increase applicability in the study field were used.

a b
Fig. 2 Photo sampling at Caldag raw material field. a Surface of muckpile on-sight after blasting and b colored state of a surface of muckpile on-sight by
WipFrag
Arab J Geosci (2020) 13: 613 Page 3 of 8 613

Fig. 3 Size distribution example

of the fragmentation at Caldag
raw material field through the
image analysis

Methods to determination of mean particle (Kou and Rustan 1993) Saroblast Model, and Kuz-Ram
size after blasting Model (Cunningham 1983, 1987) are some of the fragmenta-
tion models that appeared in the literature (Hudaverdi et al.
Several methods have been developed for prediction of post- 2012). Kuz-Ram model is the most frequently applied and
blast rock fragmentation. These methods generally can be cat- referred model. This theory, first proposed by V. M.
egorized into empirical and image analysis processing. Kuznetsov (1973), gives a reasonable description of the blast-
ed rock fragmentation. The later studies of other researchers
especially Lilly (1986) and Cunningham (1983, 1987) were
Empirical methods important contribution for improving the efficiency of this
approach. The basis of Kuz-Ram fragmentation prediction
In the literature, prediction models of mean particle size after
model is the Kuznetsov’s Eq. (1) given below:
blasting are determined together with the blast design param-
 
eters and rock mass parameters. Each model is using different X 50 ¼ AðK Þ−0;8 Qe 6 115=S An−Fo 19=30
1
ð1Þ
blast design and rock mass parameters. Researchers devel-
oped most model using different blasting and rock mass pa- X50 = Mean particle size
rameters, but Swedish Detonic Research Foundation’s
(SveDeFo) Model (Hjelmberg 1983), Kouand Rustan’s

Fig. 4 a The Proceq Pundit PL-

200 equipment and b the Schmidt
hammer (N-type) testing
equipment
613 Page 4 of 8 Arab J Geosci (2020) 13: 613

30
method for checking the fragmentation after blasting and cal-
culating fragment size distribution. Some of image processing
25
softwares are IPACS, TUCIPS, FRAGSCAN, CIAS,
Schmidt Hardness Value

GoldSize, WipFrag, SPLIT, PowerSieve, and Fragalyst

20
(Sudhakar et al. 2006; Sereshki et al. 2016).
15
Case study on Adana Cement Inc. Caldag raw material
field
10

Caldag raw material field belonging to Adana Cement Industry is

5
located in the 12th km Incirlik area of Adana-Ceyhan road, in the
same region with cement production unit (Fig. 1).
0
Generally geological layer of this mine has four units as
ACS 1
ACS 2
ACS 3
ACS 4
ACS 5
ACS 6
ACS 7
ACS 8
ACS 9
ACS 10
ACS 11
ACS 12
ACS 13
ACS 14
ACS 15
ACS 16
ACS 17
ACS 18
ACS 19
ACS 20
ACS 21
ACS 22
ACS 23
ACS 24
ACS 25
ACS 26
ACS 27
ACS 28
ACS 29
ACS 30
follows:
Blasting Area Code
1. Limestone and clayey limestone (the rock contains fossils
Fig. 5 Change in Schmidt hammer hardness values for in Caldag raw
material field in general and yellow to beige in color)
2. Marl (yellow, blue to gray in color)
A = Rock factor (It is a value ranging between 7 to 13 3. Claystone (yellow to brown in color)
according to the strength of the rocks) 4. Sandstone (yellow to brown and gray to black in color)
K = Powder factor = Qe/V (Qe is the amount of explosive
per blast hole, kg) In blasting operations, Anfo (blasting agent), emulsion explo-
V = The rock volume broken per blast hole, m3 sive (priming), and millisecond electrical detonators were used.
S ANFO = The relative weight strength of explosive The blast hole was vertical, and diameter of blast holes is 89 mm
(ANFO = 100) in the quarry. In blasts, between 27 and 120 holes which hole
length varies between 6.5 and 9.5 m are detonated. Within the
scope of the study, a total of 30 blasts were observed.
Image analysis
Determined of mean particle size by using WipFrag
Methods for evaluating fragmentation efficiency such as sieve
analysis crusher performance are generally low in applicabil- The WipFrag software developed by WipWare Corporation in
ity and nonpractical methods. Nowadays, the most effective Canada, which was established by Maerz 1996a, 1996b, can work
method for the evaluation of fragmentation is to obtain digital on digital files, videos, and photographs (Maerz 1996a, 1996b).
images from rock masses and to process these images using This software uses images from a variety of sources such as
digital image processing techniques. This is a practical camcorders, fixed cameras, photographs, or digital files and files
saved in different formats such as jpeg, tiff, etc. After the photo-
7
graphs taken in the software are uploaded to program, the bound-
6
aries of the parts are determined by means of an automated algo-
rithm and form a piece of dimension pattern. Firstly, WipFrag
measures particle as the 2-D net, and then this measured particle
P-Wave Velocity km/s

5
converts a 3-D distribution using principles of geometric proba-
4 bility. The system allows various types of output according to
individual requirements, including cumulative size distribution
3
graphs and percentage passing at different sieve sizes (Sudhakar
et al. 2006).
2
In order to determine the particle size distribution after
1
blasting, the image analysis was carried out by taking several
photographs from each fragmented muckpile to represent the
0 pile. These photographs were taken by using two defined
ACS 10
ACS 11
ACS 12
ACS 13
ACS 14
ACS 15
ACS 16
ACS 17
ACS 18
ACS 19
ACS 20
ACS 21
ACS 22
ACS 23
ACS 24
ACS 25
ACS 26
ACS 27
ACS 28
ACS 29
ACS 30
ACS 1
ACS 2
ACS 3
ACS 4
ACS 5
ACS 6
ACS 7
ACS 8
ACS 9

scales (22 cm). In the first step, photographs were selected

as the input file for WipFrag software; scales were introduced
Blasting Area Code to the program and ignored from the fragmentation analysis.
Fig. 6 Change in P wave values for in Caldag raw material field In the next step, the boundaries of each particle were
Arab J Geosci (2020) 13: 613 Page 5 of 8 613

Table 1 Validation of the derived

simple regression models (F test Equation no./ Independent B (coeff.) Standart R2 ItI F value Sig
and t test) of samples regression model variables error value

1 SCH 1.779 0.114 0.89 15.540 241.479 0.000

Power Constant 2.879 0.915 3.,147 0.004
2 Vp 203.786 20.502 0.77 9.937 98.753 0.000
Lineer Constant − 410.282 87.781 4.674 0.000

determined, and these particles were automatically colored in was evaluated for predicted uniaxial compressive strength of
the program (Fig. 2). In the last step, the mean particle size and the rocks. This test technique is quicker, cheaper, and
size distribution of fragmented rocks were obtained separately nondestructive.
as the WipFrag output data (Fig. 3). A total of 30 blasting were observed in the study area, and
samples were taken from these blasting points to determine
Determined of P wave velocity and Schmidt hardness Schmidt hardness and P wave velocity of the rock mass.
Schmidt hammer hardness values are given in Fig. 5. The P
The P wave velocity was determined on cylindrical or cubic wave velocity values are given in Fig. 6.
specimens according to the ISRM (1981) by using the Proceq
Pundit PL-200 (Fig. 4a). The Proceq Pundit PL-200 includes
Statistical analysis
two transducers (each 54 kHz), pundit touchscreen, and 2
cables (each 1.5 m). Firstly; test equipment is calibrated by
A database was prepared using mean particle size after
using calibration rod. Secondly, measurement surfaces of the
blasting, P wave velocity, and Schmidt hardness values on
specimens should be smooth, and the distance (L) between the
the benches at the Caldag raw material field. Statistical anal-
two surfaces of the sample is measured. This surface is cov-
ysis of the test results obtained from the main rock samples
ered with ultrasonic testing couplants for good interaction
and WipFrag analysis was done by SPSS and Minitab
with transducer. After that, time elapsed (T) is measure, and
softwares.
P wave velocity is calculated by using Eq. (2).
L
Vp ¼ ðm=sÞ ð2Þ Simple regression analysis
T
Simple regression analyses of the test results were performed
The Schmidt hammer (Fig. 4b) was developed for measur- using least squares regression technique. Linear (y = ax + b),
ing hardness of concrete by Schmidt 1951. But nowadays, it is logarithmic (y = a + lnx), exponential (y = aex), and power
used for rock hardness measuring, and this measured result (y = axb) models were used to obtain equations representing

Fig. 7 The relationship between

Schmidt hardness and mean 1000
particle size in Caldag raw
material field 900

800
Mean Particle Size (mm)

700

600

500

400

300

200

100
10 12 14 16 18 20 22 24 26
Schmidt Hardness Value
613 Page 6 of 8 Arab J Geosci (2020) 13: 613

Fig. 8 The relationship between

P wave velocity and mean particle 1000
size in Caldag raw material field

800

Mean Particle Size (mm)

600

400

200

0
2 3 4 5 6
P-Wave Velocity (km/s)

the most appropriate relations and having the highest correla- velocity and mean particle size on samples of Adana
tion coefficients. Cement Caldag raw material area, and statistical analyses
To verify simple regression equations, F and t tests showed that it can be contributed to the model in two param-
were used (95% level of confidence). The summaries of eters (p < 0.05). To observe the reliability of multiple regres-
test results are given Table 1. Nondestructive test values sion equations, 95% level of confidence f and t tests were used
of the rock samples, all of which were produced from 30 and examined variance inflation factor (VIF) which must be
different blasting area, were significantly correlated with less than 10 (Table 2). The estimation model equation obtain-
X50 values of these rocks. As can be seen from Figs. 7 ed from multiple regression analysis is given in Eq. 5 (Fig. 9).
and 8, the X50 showed a power relationship with the SCH
and a lineer relationship with Vp. The equations of re-
X 50 ¼ −419:1 þ 64:2 V p þ 36:09 SCH ð5Þ
gression are:
Conclusions
X 50 ¼ 2:63079  SCH 1:81247
ð3Þ
In this research, the mean particle size after
blasting estimated with simple and multiple regression analy-
X 50 ¼ −410:30 þ 203:7 V p ð4Þ sis. In this context, 30 shots were investigated in different parts
of Adana Cement Inc. Caldag raw material field to estimate
where X50 is the mean particle size (mm), SCH is the Schmidt
mean particle size models by using nondestructive methods.
hardness, and Vp is the P wave velocity.
According to results of the study, following findings were
obtained.

Multiple regression analysis & P wave velocity and Schmidt hardness parameters were
considered in models as independent variables.
Multiple regression analysis was performed to investigate the Advantages of these parameters are easy to apply, eco-
relationship between Schmidt hammer hardness to P wave nomical, and time-saving.

Table 2 Validation of derived multiple regression models (F test and t test) of samples

Eq. no Indep. variables B (coeff.) Std. error R2 ItI F- Sig VIF

value value

1 Constant − 419.143 40.67 0.954 10.306 281.861 0.000

SCH 36.092 3.548 10.173 0.000 3.09
Vp 64.225 16.681 3.850 0.001 3.09
Arab J Geosci (2020) 13: 613 Page 7 of 8 613

Fig. 9 The relationship between

the estimated mean particle size 1000
and measured mean particle size
rocks in Caldag raw material field

Measured Mean Particle Size (mm)

after multiple regression analysis 800

600

400

200

0
0 100 200 300 400 500 600 700 800 900
Estimated Mean Particle Size (mm)

& In the simple regression analysis, a significant rela- estimation of the mean particle size is very complex.
tionship was established between X50 and SCH. Using These regression models can be used only for the study
the Eq. (3), the X50 value of the batch obtained as a field. But each study field can be modeled, and the con-
result of explosions in the field will be calculated in a trollable blasting parameters can be revised according to
simple way. the change of these properties (P wave velocity and
& In the same way, a significant relationship was established Schmidt hardness).
between X50 and Vp. Using the Eq. (4), the X50 value of
the batch obtained as a result of explosions in the field will Acknowledgments The authors are grateful to the Research Fund of
Cukurova University for their financial support and Adana Cement Inc.
be calculated in a simple way. Considering the study field,
for research support.
it can be seen that predictive ability of the equations de-
rived by simple regression is considerably higher. Funding information This work was supported by the Research Fund of
& The SCH and the Vp test given in Eqs. (3) and (4) are Cukurova University (project no: FBA-2019-11637).
dependable equations that can be used in the prediction of
the X50 with a correlation coefficient of 0.89 and 0.77.
& If there are data about the both Vp and SCH of rock sam- References
ples, the equation generated using the multiple regression
techniques can be used to predict X50. Akbari M, Lashkaripour G, Bafghi AY et al (2015) Blastability evalua-
& Multible regresssion tests show that the most significant tion for rock mass fragmentation in Iran central iron ore mines. Int J
Min Sci Technol 25(1):59–66
correlations with X50 are SCH and Vp experiments. Bahrami A, Monjezi M, Goshtasbi K, Ghazvinian A (2011) Prediction of
Avaliable model obtained from multiple regression stud- rock fragmentation due to blasting using artificial neural network.
ies to predict X50 value of study field rock mass is given in Eng Comput 27(2):177–181
Eq. (5). The X50 values predicted a correlation coefficient Cunningham C (1983) The Kuz-Ram model for production of fragmen-
tation from blasting. In Proc. 1st Symp. on Rock Fragmentation by
of 0.95 by using multiple regression equations. In this
Blasting, Lulea
case, SCH and Vp variables can be used for a more Cunningham C (1987) Fragmentation estimation and the Kuz–Ram mod-
precise estimation. el: four years on. In: Proc. 2nd Int. Symp. Rock Fragmentation by
& In this study, it is emphasized that a unique model can be Blasting. Colorado, U.S.A., 475–487
prepared by using nondestructive methods for the study Hjelmberg H (1983) Some ideas on how to improve calculations of the
fragment size distribution in bench blasting. In 1st International
field. Thus, blasting design parameters can be modified to
Symposium on Rock Fragmentation by Blasting Lulea University
adjust desired grain size, and more efficient fragmentation Technology Lulea Sweden 469-494
can be achieved. Hudaverdi T, Kuzu C, Fisne A (2012) Investigation of the blast fragmen-
& There are some geological and geomechanical uncer- tation using the mean fragment size and fragmentation index. Int J
tainties in relation to the problem of fragmentation by Rock Mech Min Sci 56:136–145
ISRM (1981) International society for rock mechanics, 1981. Rock char-
blasting in mining aplications. In addition, each field ap-
acterization, testing and monitoring. In: Brown ET (ed) ISRM sug-
plies different blasting design parameters. Therefore, gested methods. Pergamon, Oxford
613 Page 8 of 8 Arab J Geosci (2020) 13: 613

Jimeno CL, Jimeno EL (1995) Drilling and blasting of rocks. Balkema, Monjezi M, Rezaei M, Varjani AY (2009) Prediction of rock fragmenta-
Rotterdam tion due to blasting in Gol-E-Gohar iron mine using fuzzy logic. Int
Kılıc AM, Yaşar E, Erdoğan Y et al (2009) Influence of rock mass prop- J Rock Mech Min Sci 46(8):1273–1280
erties on blasting efficiency. Sci Res Essays 4(11):1213–1224 Saliu MA, Akindoyeni AF, Okewale IA (2013) Correlation between blast
Kou S, Rustan A, (1993) Computerized design and result prediction of efficiency and uniaxial compressive strength. Int J Eng Tech 3(8):
bench blasting. In Proc. lh Int. Symp. on Rock Fragmentation by 799–805
Blasting, HP Rossmanith ed. 263-271 Schmidt E (1951) A non-destructive concrete tester. Concrete 59:34–35
Kulatilake PHSW, Qiong W, Hudaverdi T, Kuzu C (2010) Mean particle Sereshki F, Hoseini SM, Ataeia M (2016) Blast fragmentation analysis
size prediction in rock blast fragmentation using neural networks. using image processing. Int J Min Geo-Eng 50(2):211–218
Eng Geol 114(3–4):298–311 Shams S, Monjezi M, Majd VJ, Armaghani DJ (2015) Application of
Maerz NH, (1996a) Image sampling techniques and requirements for fuzzy inference system for prediction of rock fragmentation induced
automated image analysis of rock fragmentation. Proceedings of by blasting. Arab J Geosci 8(12):10819–10832
the FRAGBLAST, 5, 115-120 Singh SP, Abdul H (2012) Investigation of blast design parameters to
optimize fragmentation. In proceedings of the 10th international
Maerz NH, (1996b) Reconstructing 3-D block size distributions from 2-d
symposium on rock fragmentation by blasting, 181-185
measurements on sections. Fragblast 5, Workshop on Measurement
Sudhakar J, Adhikari GR, Gupta RN (2006) Comparison of fragmenta-
of Blast Fragmentation, Canada, Aug, 25-29:39–43
tion measurements by photographic and image analysis techniques.
Mehrdanesh A, Monjezi M, Sayadi AR (2018) Evaluation of effect of Rock Mech Rock Eng 39(2):159–168
rock mass properties on fragmentation using robust techniques. Eng Widzyk-Capehart E, Lilly P (2002) A review of general considerations
Comput 34(2):53–260 for assessing rock mass blastability and fragmentation. Fragblast
6(2):151–168

View publication stats