1

MEASUREMENTS OF SIZE DISTRIBUTION OF BLASTED

ROCK USING DIGITAL IMAGE PROCESSING


By:
NAME I.D.
Zubair Ahmed Nizamani (Group Leader) 09 MN 28
Shahzad Ali Rajput (Assistant Group Leader) 09 MN 85
Sanaullah Bhoot 09 MN 58
Manzoor Ali Rahimoon 09-08 MN 44
Nasir Ali Magsi 09 MN 60

SUPERVISOR:
MR. AHSAN ALI MEMON
Assistant Professor


DEPARTMENT OF MINING ENGINEERING
MEHRAN UNIVERSITY OF ENGINEERING AND TECHNOLOGY
JAMSHORO


Submitted in partial fulfillment of the requirements
for the degree of Bachelor in Mining Engineering
2013
2






"Read... Read in the name of thy
Lord who created; [He] created the
human being from blood clot. Read in
the name of thy Lord who taught by
the pen: [He] taught the human being
what he did not know" (AL-QURAN)

3

DEDICATION
This thesis is dedicated to
MY BELOVED PARENTS
who have supported me all the way
since the beginning of my studies.
&
SPECIAL GIRL
who was the source of motivation and inspiration for me.

4


MEHRAN UNIVERSITY OF ENGINEERING & TECHONOGY
JAMSHORO


CERTIFICATE

This is to certify that the thesis entitled Measurements of Size Distribution of Blasted
Rock Using Digital Image Processing in partial fulfillment of the requirements for the
award of Bachelor of Technology degree in Mining Engineering at Mehran University of
Engineering & Technology, Jamshoro is an authentic work carried out following students

NAME I.D.
Zubair Ahmed Nizamani (Group Leader) 09 MN 28
Shahzad Ali Rajput (Assistant Group Leader) 09 MN 85
Sanaullah Bhoot 09 MN 58
Manzoor Ali Rahimoon 09-08 MN 44
Nasir Ali Magsi 09 MN 60

under my supervision and guidance.


_____________ _______________
Ahsan Ali Memon External Examiner
Assistant Professor
(Thesis/Project Supervisor)


________________
CHAIRMAN
Department of Mining Engineering

Date _________
5

ACKNOWLEDGEMENT


First of all we would like to thank Almighty Allah, The most merciful, compassionate,
gracious and beneficial Who helped to complete our thesis/project.

We wish to express our profound gratitude and indebtedness to Ahsan Ali Memon, Assistant
Professor, Department of Mining Engineering for introducing the present topic and for his
inspiring guidance, constructive criticism and valuable suggestion throughout the project
work. His able knowledge and supervision with unswerving patience guided my work at
every stage, for without his warm affection and encouragement the fulfillment of the task
would have been difficult, especially Engr. Fahad Siddiqui, Lecturer, Department of Mining
Engineering who guided us and gave effective suggestions at every point of completing this
thesis.

We are also thankful to Engr. Mushtaq Ali Abro, Quarry Manager, Dewaan Cement Factory
Karachi who co-operated with us and helped us in collection of required information for the
completion of our thesis work.

Last but not least, my sincere thanks to Prof. Dr. Mohammad Ali Shah, Chairman,
Department of Mining Engineering who provided us better study environment and motivated
us to complete our thesis work.

6

ABSTRACT

The basis of this analysis was to measure the size distribution of blasted rock using the digital
image processing software Split-Desktop system. Quick and accurate measurements of size
distribution are essential for managing fragmented rock and other materials. Various
fragmentation measurement techniques are available and used by industry/researchers but
most of the methods are time consuming and not precise.

The size distribution analysis of the rock fragmentation by sieving is a direct and accurate
method but it is very time consuming and costly. Fragmentation analysis by digital image
processing is a low cost and quick method. Split system is one of the digital Image processing
software developed to compute the size distribution of fragmented rock from digital images.
Fragmentation is the ultimate measure of efficiency of any production blasting operations.
The degree of fragmentation plays an important role in order to control and minimize the
loading, hauling, and crushing costs.

In this study, size distributions were analyzed by using Split Desktop system. In the
analysis, the mean fragment size obtained is 250.75 mm and top-size 941.27mm. 7.45% of the
fragments are below 25.40mm. A thorough appraisal of blasting operation is suggested to
enhance the efficiency of all the post-blast operations such as Loading, Hauling, crushing and
Grinding and also reduces the cost of secondary breakage.

Keywords: Rock blasting, Fragmentation, Digital image processing, Split-Desktop
7

CONTENTS

CHAPTER # 01 INTRODUCTION Page No.
1.1 Background 1
1.2 General Description 2
1.3 Optimum Fragmentation 3
1.4 Significance of optimum rock
fragmentation
3
1.5 Achievement of optimum rock
fragmentation
3
1.6 Motivation 4
1.7 Objectives of the work 4

CHAPTER # 02 LITERATURE REVIEW
2.1 Mechanism of Rock Fragmentation by
Blasting
5
2.2 Different Parameters of Rock Breakage 6
2.2.1 Explosive properties 6
2.2.2 Rock properties 7
2.2.3 Charge loading and blasting parameters
and blast geometry
7
2.3 Fragmentation Measurement
Techniques
7
8

2.3.1 Sieving or screening 8
2.3.2 Oversize boulder count method 9
2.3.3 Explosive consumption in secondary
blasting method
9
2.3.4 Shovel loading rate method 9
2.3.5 Bridging delays at the crusher method 9
2.3.6 Visual analysis method 10
2.3.7 Photographic or manual analysis
method
10
2.3.8 Conventional and high speed
photogrammetric method
11
2.3.9 High speed photography or image
analysis method
11
2.3.9(a) IPACS 13
2.3.9(b) TUCIPS 13
2.3.9(c) FRAGSCAN 13
2.3.9(d) SPLIT DESKTOP 14
2.3.9(e) FRAGALYST 15
2.3.9(f) WIPFRAG 16

CHAPTER # 03 THE SPLIT SYSTEM AND
EXPERIMENTAL WORK

3.1 Introduction 17
9

3.1.1 Software and Hardware Requirements 19
3.1.2 Difference in Version of Split-Desktop 20
3.2 Description of Site 21
3.3 Methodology 23
3.3.1 Image Acquisition at Quarry 24
3.3.2 Image scaling 25
3.3.3 Fragment Delineation 26
3.3.3.1 Noise Size 27
3.3.3.2 Watershed ratio 27
3.3.3.3 Gradient ratio 27
3.4 Computation of Size Distribution
Curves
27
3.5 Sources of error 29
3.5.1 Sampling Errors 29
3.5.2 Poor Delineation of Fragments 29
3.5.3 Missing Fines 29

CHAPTER # 04 RESULTS AND DISCUSSIONS
4.1 Results 31
4.2 Combined Size Distribution 41
4.3 Discussion and Conclusion 42
Bibliography 43


10

List of Figures

Figures No. Figure Description Page No.
Figure 1.1 Typical image of rock fragmentation by blasting 2
Figure 2.1 Clear view of Blasting 6
Figure 3.1 Simple Image inserting in Software 18
Figure 3.2 Front view of quarry face F5, horizontal and vertical
bedding planes are clearly visible
21
Figure 3.3 An image taken at Dewan cement quarry for size
distribution measurement.
24
Figure 3.4 Delineation of the particles 26
Figure 3.5 Size Distribution Curves 28
Figure 4.1 Size Distribution Curve of Image Pic1 32
Figure 4.2 Size Distribution Curve of Image Pic2 33
Figure 4.3 Size Distribution Curve of Image Pic3 34
Figure 4.4 Size Distribution Curve of Image Pic4 35
Figure 4.5 Size Distribution Curve of Image Pic5 36
Figure 4.6 Size Distribution Curve of Image Pic6 37
Figure 4.7 Size Distribution Curve of Image Pic7 38
Figure 4.8 Size Distribution Curve of Image Pic8 39
Figure 4.9 Size Distribution Curve of Image Pic9 40
Figure 4.10 Size Distribution Curve of Combined Images 41

11

List of Tables

Table No. Table Description Page No.
Table 3.1 System requirements for Split-Desktop 19
Table 3.2 Blasting Parameters 2



12

INTRODUCTION

1.1 Background
Rock blasting is one of the most dominating operations in open pit mining efficiency or
quarrying. As many downstream processes depend on the blast-induced fragmentation, an
optimized blasting strategy can influence size distributions and make safe economical
environment.

A successful, complete breakage takes place when the amount of explosive and the geometry
of the blast e.g. burden, spacing, height are balanced in a way that the cracks expand all the
way to the free face and gases push the rock forward to form a well- swollen pile.

The effect of blasting on fragmentation is assessed in two different aspects: Seen and Unseen.
The size distribution of blasted fragments is the seen part of blasting results, which can be
measured quantitatively by sieving or image analysis techniques. The unseen effect of
blasting is the fracture generation within the fragments, these fracture can be classified as
either macro-fractures or micro-fractures. Macro-fractures are comparatively large and can be
seen on the surface of fragments; but micro-fractures are only seen through a microscope.
The results of a production blast are mainly presented by fragmentation of the broken rock.
The fragmentation is described in terms of geometrical characteristics of the particles i.e. size,
angularity or roundness. The cumulative size distribution function, CDF, provides a complete
description of the former. It is either obtained from physical sieving of the material, which is
very costly in large-scale blasts, or by non-physical sieving methods such as image analysis.
13

1.2 General Description
Fragmentation is the process of breaking the solid in situ rock mass into several smaller pieces
capable of being excavated or moved by material handling equipment. Breakage of rock mass
is assisted by conventional drilling blasting operation which is the most important method of
fragmentation in almost every quarry.

Figure 1.1 Typical image of rock fragmentation by blasting

There are a number of controllable as well as uncontrollable parameters that govern the
fragmentation of rock. The controllable parameters can be controlled by effective blast
designing and use of appropriate explosive for blasting. While the uncontrollable parameters
as the name suggests cannot be controlled. But certain measures have to be taken to minimize
the effects of these parameters in rock blasting in order to attain an optimum rock
fragmentation.
14

1.3 Optimum Rock Fragmentation
The rock fragmentation obtained as an outcome of blasting operations is said to be optimum,
when it contains maximum percentage of fragments in the desired range of size. The desired
size usually means the size that is demanded and can be effectively utilized by the consumers
for further operations devoid of any processing. The desired size for different consumers is
different. For example, the size of dolomite fragments required for railway tracks is
comparatively smaller than the coarser ones those used by a cement industry.

1.4 Significance of Optimum Rock Fragmentation
The significance of optimum rock fragmentation is, to fulfill the varying demands of different
consumers for assorted sizes of rock fragments, to reduce the cost of crushing and grinding or
palletization operations, and finally uphold the economics of mining. For this the rock must
be 2 fragmented in such a way that further processing (usually termed as Milling) is not
required. In other words, if the cost per ton of broken ore is greater than the price it
commands when sold as the final product, then the production is not considered to be
economic. Hence the cost of milling should be minimized and,

1.5 Achievement of Optimum Rock Fragmentation
To achieve an optimum rock fragmentation a blast with optimized controllable parameters
should be designed so that the effects of the uncontrollable parameters could be minimized.
The controllable parameters for it should be ensured that the primary blast results in optimum
fragmentation. Optimum fragmentation can be fixed after conduction of trial blasts in a mine
15

and quantification of fragmentation. Quantification of fragmentation refers to the
measurement of fragmentation in order to predict the necessary corrections in the blast design.
These corrections when applied to the blast design results in almost acceptable fragmentation.

1.6 Motivation
It is well known that rock is generally treated as a heterogeneous material and the
heterogeneity of rock causes sizes distribution of fragmented rocks in blasting. Rock
fragmentation has been used an index to estimate the effect of bench blasting for the mining
industry. The measurement of rock fragmentation using image analysis techniques has
become an active research field because of its usefulness. This trend involves an effort to
eliminate the need for traditional and costly sieve analysis. Sieving analysis is still used for
examining results of image analysis because of its limitations. Among these limitations, small
particles that are seldom represented in images of blasted rocks have been a big obstacle in
determining fragment size distribution by image analysis, especially, in large-scale blasting.

1.7 Objectives of the Work
The objectives of the project are as follows:
To analyze the fragmentation characteristics of the blasted rock using Digital Image
Processing.
To Determine the overall size distribution of blasted muck pile.

16

LITERATURE REVIEW

2.1 Mechanism of Rock Fragmentation by Blasting
Various parameters like explosive parameters, blast geometry, strength of rock, geo
technical conditions affect the degree of fragmentation of rock. The blasting operation causes
the rock fail due to crushing, tensile fracture, release of load, strain energy generation,
shearing action, flexural rupture etc.

After an explosive is initiated, the site around the drill hole will crush and will deform
plastically. The effects of an explosion can be divided into:
- The charge explodes and it is divided into high-pressure, high-temperature gases.
- The gases are applied to the borehole, which contains them .Then it creates a strain
field in the rock.
- This strain field, due to its impulse nature, generates a strain wave that is propagated
in the rock and damages it.
- This damage is the centre of the cracks in the rock.
- The gas pressure is reduced via the cracks and separates the rock fragments.
- The pressure of these gases applied to the face of the fragments, produces forces that
propel the fragments.
- The fragments adopt a ballistic trajectory.
- In areas if the damage to the rock was insufficient to generate fragments, the strain
wave continues its trajectory until it runs out of energy that dissipates by making the
rock vibrate.
17


Figure 2.1 Clear view of Blasting

2.2 Different Parameters of Rock Breakage
The parameters are divided mainly into the following: Properties of explosive, Blast geometry
and charge loading parameters.

2.2.1 Explosive properties
Different properties of explosive like V.O.D, density of explosive, shock wave energy and gas
pressure, volume of gas, composition of explosives, powder factor, and type of detonation,
primers, nature and strength of explosives affect the rock fragmentation.
18

2.2.2 Rock properties
The properties of rock that affect the rock breakage or fragmentation are dip, strike,
compressive strength, tensile strength, shear strength, density, elastic property, bedding plane
structure, presence of geological disturbances like faults, folds, fractured ground.

2.2.3 Charge loading and blasting parameters and blast geometry
The parameters which are included in this category are diameter and the length of shot holes
and charges, stemming material and height of stemming, degree of decoupling, method and
sequence of initiation, blast hole diameter, spacing and burden, distribution of explosive along
the hole, loading density, angle of blast hole, number of holes in a row, number of rows, sub
grade drilling, climate condition, amount of strata to be broken, requirement of shape of the
excavation, factors of loading, transporting and requirement of crushing and screening etc.

2.3 Fragmentation Measurement Techniques
Blast optimization requires a degree of compromise between the competing objectives of
maximum fragmentation, minimum dilution and minimum costs for drilling and explosives.
Also, mining companies and quarry operations have to examine and reduce production costs
to remain competitive. But no single factor, such as cost of explosives, can be properly
evaluated without measurements of fragmentation and rock quality. Hence the need to
manage production costs necessitates the need to measure the post-blast fragmentation.
Quantification of fragmentation on a larger scale is an extremely complicated task. Because it
needs a substantial amount of time to find out manually the grain size distribution in a muck
19

pile. Research has been carried out worldwide with different methods and tools for
measurement of fragmentation. These methods are listed below.
Sieving or Screening.
Oversize boulder count method.
Explosive consumption in secondary blasting method.
Shovel loading rate method.
Bridging delays at the crusher method.
Visual analysis method.
Photographic or manual analysis method.
Conventional and high speed photogrammetric method.
High speed photography or image analysis method.

2.3.1 Sieving or screening
Sieving or screening is a direct and accurate method of evaluation of size distribution of
particles or fragmentation. However, for production blasting, this method is costly, time-
consuming and inconvenient. This method is feasible in case of small scale blasts. In this
method the rock fragments are screened through sieves of different mesh numbers for
different fragment sizes. Then the screened out fragments are grouped according to their size
and the number of fragments in each size range is counted to predict the nature of the blast.

20

2.3.2 Oversize boulder count method
In Oversize boulder count method, manual counting of the oversize boulders in the muck pile
which cannot be handled by the shovel is done. This directly gives an over-size index with
respect to the total in-situ rock mass blasted. It is a very popular method of determining the
post-blast fragmentation.

2.3.3 Explosive consumption in secondary blasting method
In Explosive consumption in secondary blasting method, an index regarding the consumption
of explosives in secondary blasting by either pop shooting or plaster shooting is determined.
This index is then used for comparing the degree of fragmentation of a group of blasts.

2.3.4 Shovel loading rate method
The shovel loading rate method assumes that the faster the mucking the better the
fragmentation. In this method the loading rate of shovel for a particular muck pile is taken in
to account. This technique may be used more accurately for a comparative account of the
nature of fragmentation of a group of blasts.

2.3.5 Bridging delays at the crusher method
In the Bridging delays at the crusher method, the delay in bridging at the crusher mainly due
to oversize boulders is observed. This attributes in determining the number of oversize
boulders in the muck pile. This method is usually preferable in a small production site rather
than in large scale blasting situations.
21

2.3.6 Visual analysis method
The Visual analysis method is a subjective assessment method. In this method the post-blast
muck is viewed immediately after blasting and a subjective assessment is made. This
technique is not dependable as the superficial view of the muck cannot enlighten anything
about the hidden portion.

2.3.7 Photographic or manual analysis method
In photographic method delineating of fragments on the photographs of muck pile is carried
out manually to determine the number of fragments using a graph paper. For this, 0.15m x
0.10m size photographs of the muck pile are printed. Each photograph is then placed under a
transparent paper by fixing it firmly with the help of pins. All the fragments are delineated on
the transparent paper. Delineation is started with large fragments because they have more
effect on the results. It is tried to detect and delineate fragments as small as possible. The
scale placed in the middle of the muck pile is used to convert the measured distance on the
photograph to actual distance. Then, a Xerox copy of the traced paper is placed on a graph
paper. The area of the reference scale on graph paper is noted down and then a scale factor
(actual area of scale/graph area of scale) is determined. For every identifiable fragment, the
area covered by the fragment is measured by counting the number of small blocks on the
graph paper covered by that fragment. The area is then multiplied with the scale factor. For
converting the area into volume, the third dimension is determined using the method of
equivalent circle of area.
The parameters are calculated as follows:

22

|
.
|

\
|

=

Area 4
diameter Equivalent


Spherical volume (m
3
) = Area x Equivalent diameter

Weight of the fragment (kg) = Spherical volume x density of the rock

The manual analysis of each photograph takes about one to two hours.

2.3.8 Conventional and high speed photogrammetric method
This method is more reliable and accurate than the photographic method. It can provide three
dimensional measurements and thereby helps in the calculation of fragmentation volume.

2.3.9 High speed photography or image analysis method
Nowadays High speed photography or Digital images processing and analysis systems
emerged with the advance in technology are becoming increasingly popular in fragmentation
measurement. This is due to their advantages over photographic methods. Consequently
several countries and organizations have developed their own image analysis systems.

There are several methods of size distribution measurement and fall under two broad
categories; direct method and indirect methods. The sieve analysis is the direct and accurate
method of measuring size distribution. Although it is the most accurate technique among
23

others, it is not practical for such a large scale due to being both expensive and time
consuming. For this reason, indirect methods which are observational, empirical and digital
methods have been developed. Observational methods include the visual observations of
muck-piles immediately following the blasting. It is widely used by blasting engineers to
arrive at an approximation. In some empirical models such as Larssons equation, SveDe Fo
formula, KUZ-RAM model, etc, blasting parameters are considered to determine the size
distribution of blasted rock.

Recent fragmentation assessment techniques using digital image processing program allow
rapid and accurate blast fragmentation size distribution assessments. Digital image software
was developed through the 1990s and at present it is a worldwide accepted tool in the mining
and mineral processing industries. Its main advantage is that it can be used on a continuous
basis without affecting the production cycle, which makes it the only practical tool for
evaluating fragmentation of the run of mine. However, some errors are also associated with
the digital image analysis. It is extremely hard to obtain accurate estimates of rock
fragmentation after blasting. Following are the main reasons for error in using image analysis
programs.

Image analysis can only process what can be seen with the eye. Image analysis programs
cannot take into account the internal rock, so the sampling strategies should be carefully
considered.
Analyzed particle size can be over-divided or combined; which means larger particles
can be divided into smaller particles and smaller particles can be grouped into larger
24

particles. This is a common problem in all image-processing programs. Therefore,
manual editing is required.
Very fine particles can be underestimated, especially from a muckpile after blasting.
There is no good answer to avoid these problems.

In this investigation, the SPLIT-DESKTOP system was used for size distribution
computation.
Some of these systems include:

IPACS
TUCIPS
FRAGSCAN
SPLIT
FRAGALYST
WIPFRAG
IPACS
The IPACS consists of grabbing, scaling, image enhancing, grey level image segmentation,
shape analysis (merging and splitting) and processing parameters as the software functions.
The host computer required for this image analysis system is an industrial PC. Therefore this
system is well suited for industrial purposes. The Processing speed and accuracy of IPACS
are good, and the system is conducted automatically with a video input picture.

25

TUCIPS
The TUCIPS has been developed to measure blast fragmentation at Technical University
Clausthal (Germany). This system involves general algorithms specially created algorithm for
muck pile image analysis. This system is suitable for practical use because there is just five
percent (5%) deviation in the practical test with this program.

FRAGSCAN
The FRAGSCAN uses the method of measurement of the size distribution of blasted rock
from dumper or conveyer belt with the help of a camera and mathematic morphology
technique. The FRAGSCAN equipment is composed of a camera, an Image acquisition card,
a control data card, computer type PC and a light. Conversion from surface to volume
distribution is made possible by using a spherical model. This operating system is fully
automatic tool and provides reliable as well as consistent results because extensive
experimentation has provided satisfying results. This system is better for industrial usage.
SPLIT DESKTOP
The SPLIT Desktop is image analysis software developed by the University of Arizona to
figure out size distribution of rock fragment. It is operated with eight bit grayscale images of
rock fragments. There are two kinds of SPLIT programs; one is an automatic and continuous
program that is used on the conveyor belt and the other is a manual program which uses the
saved images. However, the same algorithm is used in both programs. A digital camera is
used to get the image of the bench face, which is to be used in SPLIT. The maximum size of
image that can be processed using SPLIT is 1680x1400 pixels, so the maximum size of image
26

needs to be considered during sampling images because image editing may be required in
SPLIT, and a larger image may not be opened in SPLIT without such editing.

Image samples are obtained during charging the blast holes. Approximately five to seven (5-
7) pictures are taken at each blasting, and three to five (3-5) appropriate pictures for analyzing
in SPLIT are chosen. The digital camera should be held such that the long axis of the
photograph is vertical. The image should be taken with the camera lens perpendicular to the
muck pile surface. An article of known dimensions must be in the picture in order to provide
scale. A white fig may be used as a scale material on the bench face. The same scale material
must be used from image to image for analyzing all pictures in SPLIT regarding each
blasting. Also, the number of scale materials should be the same from image to image for
analysis. Fragmentation assessment is achieved by analyzing the scaled photographs of the
muck pile.

FRAGALYST
The Fragalyst is an image analysis system developed by CMRI Regional Centre, Nagpur
(India) and Wavelet Group of Pune (India). This system consists of capturing video
photographs of the muck pile, down loading the photographs to the computer, or capturing the
photos of muck pile from field by digital camera/ordinary camera then converting the images
to grey scale, image enhancement, calibration and blob (grain) analysis. With the aid of menu-
driven software, it is possible to determine the area, size and shape of the fragments in a muck
pile/grain aggregates on the basis of grey scale difference. The 2-D information available
from software can further be processed for stereological analysis for 3-D information.
27


WIPFRAG
The WipFrag image analysis software uses the technique of analysis of digital image of the
blasted rock with granulometry system to predict the grain size distribution in the muck pile.
Typically, camcorder images of the muck pile are acquired in the field. A scale device is used
in each view to reference the sizing. The muck pile is photographed or videotaped and this
image is transferred to the WipFrag system. The broken rock image is transformed into a
particle map or network. Network areas are converted into volumes and weights and the
resulting data is displayed as a graph. The fidelity and speed of fragment edge detection allow
fully automatic remote monitoring at a rate of one image per 3 to 5 seconds. More fragments
are resolved, over a greater size range. WipFrag allows comparing the automatically
generated net against the rock image. The fragment boundaries are analyzed efficiently using
Edge Detection Variables (EDV). Any inaccuracies can be corrected by manual editing with a
mouse to improve edge detection. Manual editing, however, is needed only if image quality is
poor and is simplified by a "smart edit" function that erases and draws lines, linking them
automatically to the existing fragment net.

28

THE SPLIT SYSTEM AND EXPERIMENTAL WORK

3.1 Introduction
The Split software was originally developed at the University of Arizona, and in 1997 the
technology was transferred to a newly formed company, Split Engineering. The Split software
allows post-blast fragmentation to be determined on a regular basis throughout a mine, by
capturing images of fragmented rock in muck-piles, on haul trucks, or from primary crusher
feed or product. The resulting size distribution data can then be used to accurately assess the
fragmentation associated with different parts of a shot. And in particular, this data can be used
to assess and improve the accuracy of fragmentation models (Higgins et al, 1999).
Fragmentation models are also being improved by utilizing drill-monitoring data. Drill-
monitoring data includes raw drilling data such as rotary torque, penetration rate, and pull
down pressure, as well as calculated quantities such as drilling specific energy or the Aquila
Blast ability Index (Peck and Gray, 1995). Because drill-monitoring data is available from
every blast hole, it provides data throughout the rock mass to be blasted. As part of this
project fragmentation studies are being conducted at several large open pit mines in Arizona.
At these mines Split-Online systems are installed at the primary crushers. On these systems,
cameras installed at the truck dumps monitor primary crusher feed and cameras installed at
the discharge belts monitor primary crusher product. The primary crusher feed information is
then traced back to the original position of this rock on the shot using mining dispatch
systems. This information is used to assess post-blast fragmentation and can be correlated
with rock mass and blasting information on a hole by hole basis.

29


Figure 3.1 Simple Image inserting in Software

SPLIT is an image processing program for determining the size distribution of rock fragments
at various stages of rock breaking in the mining and processing of mineral resources. The
desktop version of SPLIT refers to the user-assisted version of the program that can be run by
mine engineers or technicians at on-site locations. The desktop SPLIT system consists of the
SPLIT software, computer, keyboard and monitor. There must be a mechanism (software
and/or hardware) for downloading digital or video camera images onto the computer. For
digital cameras the software that is supplied with the camera is required and for video camera
images a frame grabber board is necessary. For higher resolution images and for ease of
image selection, than is available by most frame grabbers, a digital camera is recommended.
Resolution of the images should be at least 512x512. The first step is for the user to acquire
images in the field and download these images onto the computer. The source of these images
can be a muck pile, haul truck, leach-pile, draw point, waste dump, stockpile, conveyor belt,
30

or any other situation where clear images of rock fragments can be obtained. The SPLIT
program first assists the user in properly scaling the images. SPLIT can then automatically
delineate the fragments in each of the images and determine the size distribution of the rock
fragments. SPLIT allows the resulting size distributions to be plotted in various forms (linear-
linear, log-linear, log-log, and Rosin- Rammler). The size distribution results can also be
stored in a tab-delineated file for access in separate spreadsheet and plotting programs.

3.1.1 Software and Hardware requirements
The hardware and software for required the split-Desktop Version 3.0 easily are mentioned in
table 3.1

Table3.1 System requirements for Split-Desktop
Computer/ processor PC compatible with 100 MHz processor of higher
Operating System
- Windows 7, 32 and 64 bit
- Windows Vista, 32 and 64 bit
- Windows XP
- Windows Server 2003
- Windows Server 2008, 32 and 64 bit
RAM 64 MB or Higher
Hard-Disk
At least 100 MB free to load manipulate and process sets of
multiple images
Monitor Higher Resolution (16-bit) or higher

31

3.1.2 Difference in version of Split-Desktop
If you have used previous versions of Split-Desktop, you may not even recognize this release
as the same software. The user interface is totally new, and the process of calculating size
distribution results has been streamlined.
Previous versions of Split-Desktop created a lot of files and then left file management up to you.
Split-Desktop 3.0 and later now use a self contained project file that includes all of your images,
settings and output options. Binary files are no longer part of Split-Desktop. Delineations are simpler
and usually better than in previous versions. The sometimes confusing array of delineation
parameters has been reduced to one simple slider bar that will increase or decrease the amount
of delineation.

Scaling has been simplified and the scales are now visible in the image. You can insert one to
three scales anywhere in the image, and modify or delete them later.
The calculations have been improved too. Not only are they faster, but the combining formula
used for merging multiple images into one result has been updated and brought more in line
with the typical field practice.
32

3.2 Description of Site
The limestone quarry belongs to Dewan cement (formerly Pak-land cement) located near
Karachi, Pakistan. The limestone deposit is of Miocene age and belongs to Gaj formation. The
geology is simple and essentially uniform. In the upper 1-2 m, there is an overburden of
weathered clay shale of sandy nature and low cohesion. The limestone formation below this
has a thickness of 6-25 m; the bedding planes are horizontal or sub-horizontal and crossed by
some nearly vertical joints as shown in Fig. 1. The upper part of limestone deposit is highly
fractured causing hole-collaring problems during drilling. The quarry is mined in one bench.
Limestone rock is medium-hard and has compressive strength of 87 MPa and density is 2.66
tons/m
3
.

Figure 3.2: Front view of quarry face F5, horizontal and vertical bedding planes are
clearly visible.
33

Drilling is done with heavy duty down-the-hole hammer drill to a preset blasting pattern. The
blasting parameters are designed to suit the rock conditions and gradation requirements. The
holes are charged with primed cartridge at the bottom with Shock-tube for detonation. ANFO
is filled as column charge. Two types of high explosives are used; Gelatinous dynamite and
Emulite. Each hole contains 15 kg of high explosive and 60 kg of ANFO. Other blasting
parameters are given in Table No. 3.2

Table No. 3.2 Blasting Parameters
Parameters Description
Hole diameter 105mm
Bench Height 9-10m
Sub Drilling 0.5m
Burden 4 feet
Spacing 3.5 feet
Stemming 0.5-1m
Blasting pattern Rectangular
Initiation System Shock Tube
Powder Factor 0.4kg/m
3

No. of Holes 32



34

3.3 Methodology
FLOW CHART OF SIZE DISTRIBUTION PROCESS
Acquire Image

Add image to project

Set Resolution

Crop Image (if needed)

Delineation

Scaling

Edit Delineation

Fines Estimation

Show Results

Export Results
35

3.3.1 Image Acquisition at Quarry
Image acquisition of blasted rock for size distribution analysis is the most critical phase of the
analysis. Important issues in image sampling are: The location of the image, the image angle
from the surface of the muck-pile, and the scale of the image. In order to obtain good images,
which are both capable of being analyzed and representative of the entire rock assemblage,
sampling strategies must be carefully considered.
The location of image taking is important, and there are two sampling methods, random and
systematic. Both methods have been used for this investigation. Another consideration is the
angle of the surface being photographed. Ideally, the surface should be perpendicular to the
camera lens.

Figure 3.3: An image taken at Dewan cement quarry for size distribution
measurement.
36

A digital camera was used to get the images of the blasted muck, which were used in SPLIT.
Images were taken randomly in the field and balls of 21.9 and 15.9 cm in diameters were used
to provide scale in the images. Single and dual scaling object were used in this investigation.
Total 15 images were taken for analysis.

3.3.2 Image scaling
For material piles, you may need to take images of different scale in order to obtain a decent
sample of the material:
1) Large scale including boulders and areas of fines. The horizontal length of the image
should be about 20 ft (7 m). These images will contain the top size material and will
adequately sample the coarse material as well as provide indications of the large areas of
fines.
2) Medium scale of typical regions of 2 to 10 inch (5 to 25 cm) material. The horizontal
length of the image should be about 8 ft (3 m). These images will provide a closer look at the
medium size material (material in size between the top size and the fines) and will lower the
fines cutoff value (the value at which the software stops measuring and begins to estimate).
3) Small scale which is zoomed in images of representative samples of the finer material. The
horizontal length of the image should be about 1.5 ft (0.5 m). These images will try to
measure the fine material to give an indication of the size distribution within the large areas of
37

fines that may be present on the surface of the large scale images. Many zoomed-in fines
images would need to be acquired to change the distribution of the entire sample, but these
images can help with measuring the fines and lowering the fines cutoff value as opposed to
using the fines estimation equation in the software.
Take approximately equal numbers of images at each scale although if you are not interested
in the size distribution of the smallest scale of material and are happy to accept a Schumann or
Rosin-Rammler curve in this range, you may omit taking the zoomed-in images.
3.3.3 Fragment Delineation
In this step Split-Desktop performs the automatic delineation of the particles. The three most
important delineation parameters are Noise size, Watershed ratio and Gradient ratio.

Figure 3.4 Delineation of the particles
38

3.3.3.1 Noise Size:
The noise size parameter is used to determine the size, in pixels, of the smallest pixel
grouping that is used in the split algorithm. Noise size value may range from 3 to 90 and
default value is 7. If the image contains larger rock fragments and boulders, noise value may
set to as higher as 80 to 90 and if it contains finer fragments, the value may reduce to 3.
The Noise size value for this investigation was found empirically by using various values and
finally 22 were found best-fit for the images.

3.3.3.2 Watershed ratio:
The watershed size ratio controls the number of divisions made during the watershed
algorithm which is used to make additional divisions based on the shape of the particles. The
default value is 1.5, which usually gives satisfactory result for most images. Increasing this
number makes fewer divisions and decreasing it makes more. This value can be changed
typically between 0.33 and 3. In this investigation, watershed ratio was set at 1.85.
3.3.3.3 Gradient ratio:
The gradient is a numerical measure of grayscale change from light to dark. The typical
average Gradient Ratio is 0.14. A higher value will create fewer dividing lines and a lower
value will create more. The gradient ratio for this analysis was set at 0.18.

3.4 Computation of Size Distribution Curves
Once the delineation of images has been completely done, computation of size distribution
can be carried out. In this step, the distribution of fines in each image can be calculated using
two approaches Rosin-Rammler or Schumann distribution. In the present study, a
39

combination of these two approaches was used to best-fit the fines distribution. The final step
and the most critical influence on the size calculation is the Fines Estimation. Split-Desktop
can measure particles automatically, but in every image there is a point below the resolution
of the image where particles can no longer be seen" and delineated. At this point, Split-
Desktop will estimate the remaining finer material. The "fines" cutoff chiefly depends on the
resolution in pixels/unit of the image. Since the black pixels in the image represent both fines
and outlines of particles, a percent of these pixels is included in the fines calculation. This
percentage of black to be counted as fines can vary for each muck-pile and can be adjusted by
the user. For the images that contain too much fines, the High option can be selected and also
other options such as None, Low and Medium can be selected accordingly depending upon
the fines percentage in each image. As shown in figure

Figure 3.5 Size Distribution Curves
40

3.5 Sources of error
There are potentially three sources of significant error while processing in Split System;
sampling errors, poor edge net fidelity, and missing fines.
3.5.1 Sampling Errors
Sampling errors, the process of taking an image of the fragmentation have the potential to be
the most serious of all the errors. Such errors result if the camera is pointed at a place in the
muck pile where the coarse blocks or zones of fines dominate.
3.5.2 Poor Delineation of Fragments
Poor delineation of individual fragments results in erroneous results. Poor delineation arises
from a combination of two sources:
- Poor images, e.g. contrast too low or high, too grainy, lighting inadequate or uneven,
or the size of the fragments in the image is too small.
- Highly textured rock, where shadows and/or colorings on the surface of the rocks are
as prominent as the shadows between rock fragments.

3.5.3 Missing Fines
Where the smallest fragments in a distribution are not delineated on the image, either because
they are too small relative to the image to be resolved, or they have fallen in and behind larger
fragments, there is clearly a bias towards over representing the size of the distribution. Where
the distribution has a relatively narrow size range (well sorted, or poorly graded) this is
normally not a problem. However, where the distribution has a relatively wider size range
(poorly sorted, or well graded), typically with size differences of more than 1 order of
41

magnitude, missing fines start affecting the measurement results. Split Desktop has the ability
to deal with the missing fines problem using either an empirically based calibrations or by
using multiple images taken at different scales of observation.

42

RESULTS AND DISCUSSIONS

4.1 Results
Total 70 images were taken during the field visit to Pak Land Cement limestone quarry
immediately after the blasting. Nine most representative images of blasted muck-pile were
analyzed using Split-Desktop Software and mean values were obtained. Following are the
obtained size distribution curves of each and finally combined image.

43

CUMMULATIVE SIZE DISTRIBUTION

Picture taken at Site Picture Delineation

Figure 4.1: Size Distribution Curve of Image Pic1

44

CUMMULATIVE SIZE DISTRIBUTION

Picture taken at Site Picture Delineation

Figure 4.2: Size Distribution Curve of Image Pic2
45

CUMMULATIVE SIZE DISTRIBUTIO

Picture taken at Site Picture Delineation

Figure 4.3: Size Distribution Curve of Image Pic3

46

CUMMULATIVE SIZE DISTRIBUTION

Picture taken at Site Picture Delineation

Figure 4.4: Size Distribution Curve of Image Pic4


47

CUMMULATIVE SIZE DISTRIBUTION

Picture taken at Site Picture Delineation

Figure 4.5: Size Distribution Curve of Image Pic5

48

CUMMULATIVE SIZE DISTRIBUTION

Picture taken at Site Picture Delineation

Figure 4.6: Size Distribution Curve of Image Pic6

49

CUMMULATIVE SIZE DISTRIBUTION

Picture taken at Site Picture Delineation

Figure 4.7: Size Distribution Curve of Image Pic7

50

CUMMULATIVE SIZE DISTRIBUTION

Picture taken at Site Picture Delineation

Figure 4.8: Size Distribution Curve of Image Pic8

51

CUMMULATIVE SIZE DISTRIBUTION

Picture taken at Site Picture Delineation

Figure 4.9: Size Distribution Curve of Image Pic9
52

4.2 COMBINED SIZE DISTRIBUTION

Figure 4.10: Size Distribution Curve of Combined Images

53

4.3 Discussion and Conclusion
The results obtained from the analysis of muck-pile images using Split-Desktop shows that
the mean fragment size is 250.75 mm and F20, F80, and Top-size are 108.03 mm, 417.19 mm
and 941.27mm respectively.

The primary crusher installed at the quarry accepts the feed size as large as 1000 mm and
crush down to the 25 mm. Results indicate that approximately 7.45% of the fragments are
below 25.45 mm.

Results also indicate that only 0% of the material is above 1000 mm therefore it doesnt
require secondary breakage. The Rosin-Rammler uniformity index of the entire muck-pile is
0.81. This index is generally used to approximate the size distribution of rock in blasted
muck-piles. The value ranges between 0.5 (very non-uniform) and 2 (very uniform). So the
obtained index value confirms non-uniform size distribution. Non uniform size distribution
affects the loading and hauling operations and crushers efficiency.

As the results indicate that 7.45% fragments are below 25.45 mm, which is product size of
primary crusher, this percentage can be enhanced by optimizing the overall blasting operation.

The Burden and spacing are two most important factors in the blasting because these factors
can be adjusted to obtain required fragmentation. Proper explosive in an appropriate quantity
can also results in good fragmentation and reduce the overall cost of production.

54

Bibliography

Robert S.lewis, E.M, Element of Mining, Reprinted by KHAYABAN pass Lahore,
1983
Dahlhielm.S(1996) industrial application of image analysis-the IPACAS system
proceeding measurement of blast fragmentation
Girdner, k.k, kemeny, J.M, Srikant.A & Mcgill.R (1996) The split system for
analyzing the size distribution of fragmented rock proceeding measurement of blast
fragmentation
Jimeno C.L, Jimeno E.L, Carcedo, F.J.A (1995) drilling and blasting of rocks
Norton B (2005) Private communication with as expert in Split Engineering
Split Enginnering LLC (2001) Split-Desktop Software manual.

Web References
http://www.mine-engineer.com/mining/open_pit.htm
www.spliteng.com
http://www.miningequipmentforsale.net/mining-equipment-for-sale/resize-blasted-
rock-to-stone-crusher-size.html