2003 - Multivariate Statistical Analysis of The Wearing On Diamond Beads in The Cutting of Andesitic Rocks

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 14

Key Engineering Materials Vol 250 (2003) pp 118-130 Online: 2003-09-15

© (2003) Trans Tech Publications, Switzerland


doi:10.4028/www.scientific.net/KEM.250.118
Citation & Key Engineering Materials Vols. (2003) pp.
© 2003 Trans Tech Publications, Switzerland

Multivariate Statistical Analysis of the Wearing on Diamond Beads in the


Cutting of Andesitic Rocks

Y. Özçelik
Hacettepe University, Department of Mining Engineering, 06532 Beytepe, Ankara, Turkey

Keywords: Marble, Diamond wire cutting, Diamond bead, Wearing, Multicollinearity, Factor
analysis, Regression analysis, Statistical modelling

Abstract. The wearing rate on the beads of diamond wire is the major parameter that effects the cost
of a hard rock marble quarry in the diamond wire cutting method. Therefore, it is very important to
develop the predictive equations that can be used to assess the wearing on beads during the rock
cutting. This study is primarily concerned with this problem. Diamond wire cutting studies in
industrial scale were performed at andesitic rocks of different mechanical and physical
characteristics. The andesitic rocks are located at Çubuk District of Ankara Province (TURKEY). The
wearings on beads in two modes of vertical and horizontal cuttings are evaluated separetely and
modelled as function of rock properties, machine and cutting characteristics. Multivariate statistical
methods include factor and regression analysis in reduced and unreduced forms. The unreduced
model covers all the independent variables. In the reduced model one of the variables that is strongly
interrelated (X2, X11, X13, X16, X17) is eliminated and the rest is used in the analysis. Reduced and
unreduced model equations predicting the wearing of diamond beads in the wire cutting of andesitic
rocks are developed.

Introduction
The increasing demand for stone products has inevitably led to the opening of numerous new stone
quarries all over the world and the use of diamond wire cutting machines to produce well-shaped
blocks of marble of carbonate origin with high recovery is common in modern quarries. Diamond
wire cutting not only speeds up the extraction process but, once started, can operate without constant
supervision. A number of cuttings can work on the quarry face at the same time, and each of which
requires only enough space for the power unit and its track. Nowadays, the diamond wire cutting has
been extended to some marble quarries of volcanic origin [1].
Direction of Drive Wheel
System Movement for Vertical Cutting

Vertical Hole

Diamond Wire

ROCK Diamond Wire Cutting


Machine
Drive
Horizontal Hole Wheel

Cutting Area

Drive
Wheel
ROCK
Diamond Wire

Rail System
Direction of Drive Wheel
System Movement for Horizontal Cutting

Fig.1 Diagram of diamond wire cutting layout

In diamond wire cutting, the cutting action primarily involves the pulling of continous loops of

All rights reserved. No part of contents of this paper may be reproduced or transmitted in any form or by any means without the written permission of Trans
Tech Publications, www.ttp.net. (ID: 130.216.129.208, University of Auckland, Auckland, New Zealand-16/04/15,01:59:36)
Key Engineering Materials Vol. 250 119

Key Engineering Materials Vols. 119

spinning wire mounted with diamond bonded steel beads through the stone. Thus, a path is cut by the
combination of spinning wire and the pulling force action on it. The initial step for making a vertical
cut is to drill two holes, one vertical and one horizontal, which intersect at a 90° angle. Then, the
diamond wire threaded through these holes, and over the drive wheel, clames at the two ends to form
a continuous loop. The drive wheel may be set at any angle, from vertical to horizontal, required to
facilitate cutting. The diamond wire cutting machine is mounted on temporary track, along which it
reverses to maintain tension in the wire as it cuts through the stone. When the limit of the track is
reached, the machine is returned to its starting point and the length of the wire is reduced accordingly
to permit cutting to continue. Water is applied with the spin direction of the wire as a coolant and as a
means of removing the particles of stone from the cut. [1-6]. Diagram of diamond wire cutting layout
is shown in Fig.1 [1].
The diamond wire is simply a steel cable on which small beads bonded with diamond abrasive are
mounted at a regular interval with spacing material placed between the beads (Fig.2).
Steel Ring

Matrix+Diamond Particals
1.5 mm
5 mm

Wire Rope
Diamond Bead
11-12 mm

Wire Rope

Fig.2 Typical illustration of diamond wire and cross-section of diamond bead

The beads provide the actual cutting action in this operation. They are bonded with diamond by
one of two methods: electroplating or impregnated metal powder bonding.
Since wire replacement is undoubtedly the major cost item especially in the case of hard rocks, the
problem of assessing bead wear is critical to the optimization of rock cutting operations with diamond
wire [5,7].
The aim of this study is to investigate the effects of the rock and machine parameters on the cutting
performance of diamond wire cutting machines on andesitic rock. For this purpose, the rocks of
different mechanical properties in the field are cut horizontally and vertically by using the diamond
wire cutting machines. Some laboratory experiments were performed with the specimens taken from
the rocks.
Using the information obtained from these works, the statistical models describing the wearing on
the beads which is the prime factor controlling the economy of diamond wire cutting have been
developed and tested.
Consequently, the numerical determination of the wearing in terms of rock and machine
parameters is realized. The statistical evaluation is done using the Excel 7.0, Statistica 5.0 and SPSS
7.0 packet programs.
120
Table 1 Results of field and laboratory studies
120

Laboratory Studies Vertical Cutting Horizontal Cutting

X1 X2 X3 X4 X5 X6 X7 X8 X9 X10 X11 X12 X13 X14 X15 X16 X17 Y X14 X15 X16 X17 Y

Çubuk1 1.95 2.98 5.79 18.7 3.72 2.82 33 4.53 41 6.8 7.2 19.7 3.19 19.45 64.24 158 29700 0.07 18.1 32.76 145 27200 0.06

Çubuk2 2.1 2.43 4.97 31.59 5.05 3.05 39.93 8 46 7.6 6.8 18.9 2.79 19.15 63 154 28900 0.145 5 20 40 7500 0.04

Çubuk3 2.42 1.25 3.1 51.5 9.55 5.15 65 8.4 48.3 20.8 5.5 16.1 1.33 58.1 78.61 465 77000 0.283 19.3 24.2 155 29300 0.099

Çubuk4 2.39 1.5 3.59 56.01 8.86 3.77 63.62 8.19 45.7 23.5 5.1 14.4 1.41 28 50.37 224 42000 0.172 13 27.86 104 19500 0.082

Çubuk5 1.81 3.68 6.61 17.7 3.56 1.7 49.93 5.02 36.8 7.9 8.7 22.7 2.21 24.3 43.91 196 36750 0.093 7.3 22.5 60 11200 0.025

Çubuk6 2.44 1.03 2.52 44.94 9.83 5.53 67 9.78 49.2 15.5 5.4 15.6 1.22 34.3 54.17 276 51300 0.302 7.3 17.4 60 11200 0.048

Çubuk7 2.15 2.5 4.43 38.5 6.15 3.55 43.7 7.9 47.5 8 6.8 18 2.57 26 57.94 208 39000 0.093 4.3 13.5 36 6750 0.016

Çubuk8 2.4 1.48 3.53 58.35 8.82 4.84 61.38 9.76 47.1 30.56 5.6 17 1.54 79.58 26.13 640 120000 0.411 56 19 456 85500 0.315

Çubuk9 2.45 1.35 3.16 50.5 9.25 5 63.7 9.81 47.9 24.6 5.3 15.5 1.41 41.83 26 256 63000 0.437 34.25 20 275 51250 0.282

Çubuk10 2.41 1.42 3.35 54.23 9.2 4.92 62.48 9.8 47.5 25.4 5.4 16 1.48 44.08 28 352 66000 0.445 35 20.25 280 52500 0.32

Çubuk11 2.4 1.5 3.55 52.5 8.95 4.75 60.3 9.63 47 26.4 5.3 15.4 1.61 42.2 36 337.6 63300 0.305 66 52 528 99000 0.35

Çubuk12 2.37 1.56 3.6 55 9.1 4.9 61.2 9.76 47.2 28.3 5.5 16.2 1.55 109 81 872 163500 0.645 72 51 576 108000 0.35
Machining of Natural Stone Materials

Çubuk13 1.89 3.25 6.15 18.15 3.68 2.58 42.5 4.72 38.4 7.3 8.3 21.4 2.64 40 48.1 320 60000 0.172 58.15 67.9 465.2 87225 0.263

Çubuk14 2.18 2.18 4.7 34.61 5.53 3.29 41.2 7.93 46.7 7.76 6.8 18.5 2.71 9.3 19.5 74.4 13950 0.042 7.6 14.5 60.8 11400 0.04
Machining of Natural Stone Materials

Çubuk15 2 2.75 4.8 37.5 6.05 3.25 43.5 6.62 45 8.01 7 19 2.61 36.4 54.6 291.2 54600 0.194 33.8 50.41 280.6 51460 0.18

Minimum 1.81 1.03 3.1 18.15 3.56 1.7 33 4.53 36.8 7.03 5.1 14.4 1.33 9.3 19.5 74.4 13950 0.042 4.3 13.5 36 6750 0.016

Maximum 2.45 3.68 6.61 58.35 9.55 5.53 67 9.81 49.2 30.56 8.7 22.7 3.19 109 81 872 163500 0.645 72 67.9 576 108000 0.35

Mean 2.24 2.01 4.21 41.59 7.23 3.99 53.92 8.09 45.45 17.17 6.26 17.53 1.98 40.77 48.77 321.6 60600 0.254 29.14 30.22 234.77 43932 0.165

Variance 0.05 0.7 1.57 226.05 6.07 1.38 134.78 3.87 14.69 86.34 1.39 6 0.45 650.17 359.08 41967 1433432 0.029 566.98 283.3 36467 1200000 0.018
Key Engineering Materials Vol. 250 121

Key Engineering Materials Vols. 121

The Ankara andesitic rock under study has the porfiric texture with approximately %68.5-74.8
Matrix, %0.8-2.5 Biotite, %0.1-1.8 Hornblende, %15.8-26.8 Plagioclase, %0.3-1.2 Opaque Minerals
and %1.0-7.0 Amorphous Quartz. Mineralogic constitute of the matrix consists of %60 volcanic
glass, 35% plagioclase’s microlite and 5% hematite-limonite. Volcanic glass and plagioclase’s
microlite fairly form a compact texture. Ankara andesite marble quarry shows massive structure, but
sometimes lava flow structure. This lava flow shows bedding structure in decimetrical thickness. The
position of bedding is between N30°-35°E/26°-45°SE in north and, E-W/37°-55°N or
N63°W/37°-55°NE in south [1]. Location map of the investigation area is shown in Fig.3.

Fig.3 Location map of the investigation area (Turkey)

Field and Laboratory Studies


Parallel to the purpose of this study, the only and first diamond wire cutting operations (vertical and
horizontal) have been performed at an andesite quarry in Ankara Province in Turkey. Cutting time,
cutting area of the blocks and energy and water consumption for andesites from 15 different localities
have been recorded. The physical and mechanical properties of the rock are determined in laboratory
under ISRM [8] and T.S.E. (Turkish Standarts) [9] standards. The results obtained are presented in
Table 1.

Statistical Analysis of the Test Results


In order to facilitate the analysis and interpretation of the results, it was decided to categorize the data
into dependent (Y) and independent (X1,……,X18) variables. Table 2 presents the dependent and
independent variables and their symbols used in this statistical analysis.
The statistical analysis consists of determination of the interrelations among the variables, the
factor analysis and regression analysis.
For the statistical evaluation, each parameter included in the research has been taken as a variable.
The data related to these variables, have been obtained in two categories as vertical and horizontal
depending upon direction of wire cutting. In the wire cuttings done in the rocks with identical
properties, the measurements in horizontal and vertical wire cuttings differ in cutting time, cutting
area, wearing on beads and water and energy consumption.
Determination of the Correlation Coefficient Matrix and Simple Linear Regression between
Dependent and Independent Variables. Pearson Correlation Coefficient was selected as correlation
coefficients due to the fact that the variables were obtained by measurements and denoted the
continual structure.
Cov(X, Y) (X i - X )(Yi - Y )
rP = = -1£r<1. (1)
SxSy (X i - X ) 2 (Yi - Y ) 2

where,
X is mean and Sx is standard deviation.
122 Machining of Natural Stone Materials

122 Machining of Natural Stone Materials

Table 2 Independent and dependent variables and their symbols

Parameters Symbols
Specific Gravity X1
Water Adsorption X2
Apparent Porosity X3
Uniaxial Compression Stren. X4
Tensile Strength X5
Point Load Strength X6
Rock Parameters Shore Scleroscope Hardness X7
NCB Cone-Indenter X8
Schmidt Hardness X9
Modulus of Elasticity X10
Los Angelas Abr. Stren. (100) X11
Los Angelas Abr. Stren. (500) X12
Böhme Surface Abrasion Stren. X13
Cutting Time X14
Cutting Area X15
Machine And Cutting Parameters
Energy Consumption X16
Water Consumption X17
Bead Parameter Wear Rate Y

The interrelations among the variables with respect to Pearson coefficients are presented in Table 3
As seen from the Table, the relations between the variables are well correlated either positively or
negatively. But, cutting time, cutting area and energy and water consumption are not well correlated
with other variables. Remarkably, the measurements taken in vertical cutting indicate the wearing is
related to the other variables whereas no relation has been obtained in measurements of horizontal
cutting. Furthermore, the important result is that the modulus of elasticity is in good relation with the
variables except the cutting area.
In this study, suitable statistical models have been used for the investigation of the effects of
variables convenient to the purpose of the study, on the wearing. Regressive models for regression
analysis have been established. Because, these models determine the individual effects of variables by
evaluating them individually.
The statistical models established in this study are linear due to the reason that the results obtained
exhibit linear functions in SPSS 7.0 statistical packet program.
The effects of the other variables on the dependent variable are searched in two ways. In the first
way, the individual effects of independent variables are modelled by simple linear regression. In the
second way, their overall effects on wearing are modelled by multiple linear regression.
The effect of the variables on diamond beads has been analysed by simple linear regression
analysis. The results obtained are given in Table 4. According to Table 4, the rock parameters namely
the modulus of elasticity and specific gravity and the machine parameters namely the cutting time and
energy and water consumption are the most important parameters affecting the bead wearing.
However, the variables namely the water adsorption, the appearent porosity, the Los Angelas abrasion
strength and Böhme surface abrasion strength have inverse effect on wearing, because, when these
variables get higher values the cuttability of rock gets easy.
Table 3. Correlation coefficient matrix for vertical and horizontal cutting measurements

Vertical Cutting Horizontal Cutting


Variables X1 X2 X3 X4 X5 X6 X7 X8 X9 X10 X11 X12 X13 X14 X15 X16 X17 Yver. X14 X15 X16 X17 Yhor.

X1 1 -0.99* -1 -0.95* 0.98* 0.95* 0.83* 0.93* 0.90* 0.83* -0.98* -0.96* -0.83* 0.45 -0.2 0.42 0.44 0.64* -0.02 -0.4 -0 -0 0.25

X2 -0.99* 1 0.99* -0.92* -0.97* -0.96* -0.81* -0.91* -0.90* -0.78* 0.97* 0.95* 0.81* -0.4 0.13 -0.4 -0.4 -0.62* 0.03 0.38 0.03 0 -0.22

X3 -0.98* 0.99* 1 -0.91* -0.97* -0.96* -0.82* -0.92* -0.91* -0.75* 0.96* 0.95* 0.82* -0.4 0.13 -0.4 -0.4 -0.59* 0.08 0.41 0.08 0.1 -0.17

X4 0.95* -0.92* -0.91* 1 0.95* 0.87* 0.83* 0.91* 0.82* 0.91* -0.94* -0.92* -0.83* 0.57 -0.2 0.55 0.56 0.66* 0.05 -0.37 0.05 0.1 0.31

X5 0.98* -0.97* -0.97* 0.95* 1 0.95* 0.92* 0.90* 0.82* 0.86* -0.95* -0.93* -0.92* 0.5 -0.2 0.48 0.49 0.67* 0.02 -0.35 0.02 0 0.27

X6 0.95* -0.96* -0.96* 0.87* 0.95* 1 0.81* 0.89* 0.86* 0.78* -0.91* -0.87* -0.81* 0.51 -0.1 0.49 0.5 0.69* 0.09 -0.29 0.09 0.1 0.32

X7 0.83* -0.81* -0.82* 0.83* 0.92* 0.81* 1 0.75* 0.56* 0.84* -0.77* -0.75* -1.0* 0.46 -0.3 0.43 0.44 0.62* 0.1 -0.23 0.1 0.1 0.31

X8 0.93* -0.91* -0.92* 0.91* 0.90* 0.89* 0.75* 1 0.90* 0.77* -0.88* -0.85* -0.75* 0.51 -0.3 0.48 0.51 0.70* -0.05 -0.47 -0 -0 0.24

X9 0.90* -0.90* -0.91* 0.82* 0.82* 0.86* 0.56* 0.90* 1 0.55* -0.87* -0.86* -0.56* 34 0 0.32 0.33 0.5 -0.23 -0.53 -0.2 -0 0.02

X10 0.83* -0.78* -0.75* 0.91* 0.86* 0.78* 0.84* 0.77* 0.55* 1 -0.82* -0.76* -0.84* 0.68* 0.36 0.66* 0.68* 0.76* 0.32 -0.17 0.32 0.3 0.57*

X11 -0.98* 0.97* 0.96* -0.94* -0.95* -0.91* -0.77* -0.88* -0.87* -0.82* 1 0.99* 0.77* -0.4 0.15 -0.4 -0.4 -0.60* 0.06 0.4 0.06 0.1 -0.21

X12 -0.96* 0.95* 0.95* -0.92* -0.93* -0.87* -0.75* -0.85* -0.86* -0.76* 0.99* 1 0.75* -0.3 0.12 -0.3 -0.3 -0.52* 0.09 0.37 0.08 0.1 -0.17

X13 -0.83* 0.81* 0.82* -0.83* -0.92* -0.8 -1.00* -0.75* -0.56* -0.84* 0.77* 0.75* 1 -0.5 0.27 -0.4 -0.4 -0.62* -0.1 0.23 -0.1 -0 -0.31

X14 0.45 -0.4 -0.4 0.57 0.5 0.51 0.46 0.51 0.34 0.68* -0.4 -0.3 -0.5 1 0.22 1.00* 1.00* 0.89* ---- ---- ---- ---- ----
Key Engineering Materials Vols.

X15 -0.2 0.13 0.13 -0.2 -0.2 -0.1 -0.3 -0.3 0 -0.4 0.15 0.12 0.27 0.22 1 0.25 0.19 0.04 ---- ---- ---- ---- ----
Key Engineering Materials Vol. 250

X16 0.42 -0.4 -0.4 0.55 0.48 0.49 0.43 0.48 0.32 0.68* -0.4 -0.3 -0.4 1 0.25 1 0.99* 0.85* ---- ---- ---- ---- ----

X17 0.44 -0.4 -0.4 0.56 0.49 0.5 0.44 0.51 0.32 0.66* -0.4 -0.3 -0.4 1 0.29 0.99* 1 0.89* ---- ---- ---- ---- ----

Yver. 0.64* -0.62* -0.6 0.66* 0.67* 0.69* 0.62* 0.70* 0.5 0.76* -0.60* -0.5 -0.62* 0.89* 0.04 0.85* 0.89* 1 ---- ---- ---- ---- ----

X14 -0 0.03 0.08 0.05 0.02 0.09 0.1 -0.1 -0.2 0.32 0.06 0.09 -0.1 ---- ---- ---- ---- ---- 1 0.82* 1.00* 1.00* 0.89*

X15 -0.4 0.38 0.41 -0.4 -0.4 -0.3 -0.2 -0.5 -0.5 -0.2 0.4 0.37 0.23 ---- ---- ---- ---- ---- 0.82* 1 0.81* 0.82* 0.56*

X16 -0 0.03 0.08 0.05 0.02 0.09 0.1 -0 -0.2 0.32 0.06 0.08 -0.1 ---- ---- ---- ---- ---- 1.00* 0.81* 1 1.00* 0.89*

X17 -0 0.03 0.08 0.05 0.02 0.09 0.1 -0 -0.2 0.32 0.06 0.09 -0.1 ---- ---- ---- ---- ---- 1.00* 0.82* 1.00* 1 0.89*

Yhor. 0.25 -0.2 -0.2 0.31 0.27 0.32 0.31 0.24 0.02 0.57* -0.2 -0.2 -0.3 ---- ---- ---- ---- ---- 0.89* 0.56* 0.89* 0.89* 1

N: 15
* signed varibles have significant relation in %95 confidence level
123

123
124 Machining of Natural Stone Materials

124 Machining of Natural Stone Materials

Table 4 Simple Linear Regression Results

Vertical Cutting Horizontal Cutting


Dep. Ind. Effect. Mean. Coef. St. Effect. Mean. Coef. St.
Var. Var. (b) (p) Deter. (R2) Errors (b) (p) Deter. (R2) Errors
X1 0.5132 0.024** 0.45 0.138 0.2413 0.419 0.17 0.129
X2 -0.1361 0.031** 0.42 0.141 -0.0595 0.476 0.14 0.130
X3 -0.091 0.041** 0.41 0.144 -0.0367 0.571 0.11 0.131
X4 0.0085 0.019** 0.51 0.135 0.0047 0.298 0.26 0.126
X5 0.0529 0.017** 0.53 0.134 0.0251 0.375 0.20 0.128
X6 0.1128 0.014** 0.56 0.131 0.0555 0.290 0.23 0.126
Wear Rate (Y)

X7 0.0107 0.032** 0.51 0.141 0.0051 0.304 0.19 0.127


X8 0.0632 0.012** 0.51 0.129 0.0305 0.432 0.20 0.129
X9 0.0223 0.095* 0.23 0.155 0.0066 0.954 0.04 0.133
X10 0.0155 0.004** 0.69 0.117 0.0103 0.042** 0.51 0.109
X11 -0.0963 0.041** 0.42 0.144 -0.0462 0.484 0.16 0.130
X12 -0.0415 0.086* 0.33 0.154 -0.0191 0.581 0.12 0.131
X13 -0.1841 0.032** 0.51 0.141 -0.0869 0.305 0.19 0.127
X14 0.0058 0.000** 0.75 0.083 0.0052 0.000** 0.79 0.062
X15 0.0004 0.912 0.00 0.179 0.0041 0.046** 0.27 0.110
X16 0.0007 0.000** 0.70 0.094 0.0006 0.000** 0.79 0.061
X17 0.0000 0.000** 0.75 0.082 0.0000 0.000** 0.79 0.061
*p<a=0.05 and **p<a=0.10 are significant

Factor and Multiple Regression Analysis. A multiple regression analysis can not be used due to
the fact that the independent variables are strongly interrelated (multicollinearity) to each other. In
other words, the preassumption that the covariance between variables is zero [cov(Xi,Xj)=0] has been
invalid. This difficulty has led to the use of the Factorial Analysis aiming at reducing the number of
variables by grouping the strongly interrelated variables as a new variable.
The essential purpose of factor analysis is to describe, if possible, the covariance relationships
among many variables in terms of a few underlying, but unobservable, random quantities called
factors. Basically, the factor model is motivated by the following argument. Suppose variables can be
grouped by their correlations. That is, all variables within a particular group are highly correlated
among themselves but have relatively small correlations with variables in a different group. It is
conceivable that each group of variables represents a single underlying construct, or factor, that is
responsible for the observed correlations.
Factor analysis can be viewed as an attempt to approximate the covariance matrix S. However, the
approximation based on the factor analysis model is more elaborate. The primary question in factor
analysis is whether the data are consistent with a prescribed structure.
The observable random vector X, with p components, has mean m and covariance matrix S. The
factor model postulates that X is linearly dependent upon a few unobservable random variables F1,
F2… Fm, called common factors, and p additional sources of variation e1, e2… ep, called errors. In
particular, the factor analysis model is
X 1 - m 1 = l11 F1 + l12 F2 + ....... + l1m Fm + e 1
X 2 - m 2 = l 21 F1 + l 22 F2 + ....... + l 2 m Fm + e 2
. (2)
M M M M
X p - m p = l p1 F1 + l p 2 F2 + ....... + l pm Fm + e p

or, in matrix notation,


Key Engineering Materials Vol. 250 125

Key Engineering Materials Vols. 125

X -m = L F + e . (3)
( p´1) ( p´m ) ( m´1) ( p´1)

The coefficient l ij is called the loading of the ith variable on the jth factor, so the matrix L is the
matrix of factor loadings [10].
The complexity of most scientific mechanism is such that in order to be able to predict an
important response, a well-known multiple regression model is needed. The quality of the estimated
regression model is often tested through the very well-known method of ANOVA
(Analysis-Of-Variance). ANOVA approach is used to partition the total sum of squares into a portion
due to regression and a portion due to errors. However, it describes a technique whereby the total
variation is being analyzed or divided into meaningfull components [11].
This study contains a large amount of variables and models. Figure 4 shows the flowchart
considered in this study.

Collection of field and


laboratory data

Definition of dependent and


independent variables

Determination of descriptive statistics


of dependent (wearing rate) and
independent variables

Analysis of the relationship


between dependent and
independent variables

Investigation of the effects of the


dependent variables among
the independent variables

Simple Linear Regression Multiple Linear Regresion


Analysis Analysis

Yes Is the No Performing the


INTERPRETATION of the Results FACTOR ANALYSIS for
and developed models
R e peating the o perations fo r red uced va riables

adequate? removing the multicollinearty


FINISH

Testing the developed models and


investigation of relationships
between measured and predicted values

INTERPRETATION of the Results Yes Are the Results


and Relationships No Removing of
and unnecessary variables
FINISH true and reliable?

Fig.4 Flowchart considered in this study


126 Machining of Natural Stone Materials

126 Machining of Natural Stone Materials

In this study, factorial analysis of the experimental data for vertical and horizontal cutting is
performed using SPSS 7.0 and STATISTICA 5.0 softwares and the results are given in Table 5-a and
b. The factor loads of variables relating to vertical and horizontal measurements are shown in Table
5-a and b.
Table 5-a Factor loads of unreduced variables for vertical and horizontal measurements
Vertical Horizontal
Variables Factor 1 (F1) Factor 2 (F2) Factor 1 (F1) Factor 2 (F2)
X1 0.9739 * 0.0343 0.9712 * -0.0446
X2 -0.9613 * -0.0199 -0.9599 * 0.0526
X3 -0.9572 * 0.0133 -0.9586 * 0.0.953
X4 0.9213 * 0.1885 0.9433 * 0.0293
X5 0.9651 * 0.0987 0.9710 * 0.0129
X6 0.9186 * 0.1244 0.9299 * 0.0635
X7 0.8453 * 0.1080 0.8524 * 0.1275
X8 0.9023 * 0.1126 0.9104 * -0.0879
X9 0.8459 * -0.0258 0.8423 * -0.2817
X10 0.8046 * 0.3501 0.8432 * 0.3074
X11 -0.9402 * -0.0171 -0.9453 * -0.0766
X12 -0.9201 * 0.0649 -0.9247 * 0.0958
X13 -0.8453 * -0.1103 -0.8529 * -0.1270
X14 0.4102 0.8892 * 0.0152 0.9752 *
X15 -0.2554 0.6263 * -0.3678 0.8337 *
X16 0.3840 0.9016 * 0.0184 0.9747 *
X17 0.3997 0.8886 * 0.0182 0.9748 *
Amount of Explanation 11.29 2.80 11.07 3.79
Rate of Explanation 0.66 0.16 0.65 0.22
* p < a=0.05 are significant
Table 5-b Factor loads of reduced variables for vertical and horizontal measurements
Vertical Horizontal
Variables Factor 1 (F1) Factor 2 (F2) Factor 3 (F3) Factor 1 (F1) Factor 2 (F2)
X1 0.9751 * 0.0234 -0.0200 0.9917 * -0.0652
X3 -0.9643 * 0.0674 0.1175 -0.9774 * 0.1141
X4 0.9253 * 0.0983 0.1850 0.9679 * 0.0089
X5 0.9645 * 0.0300 0.0676 0.9904 * 0.0056
X6 0.9355 * -0.0877 0.0114 0.9524 * 0.0367
X7 0.8196 * 0.2199 0.2064 0.8655 * 0.1468
X8 0.9160 * 0.0835 0.0406 0.9364 * -0.1331
X9 0.8812 * -0.2445 -0.2460 0.8636 * -0.3290
X10 0.7942 * 0.2898 0.4637 0.8675 * 0.2965
X12 -0.9189 * 0.0429 0.1383 -0.9412 * 0.1005
X14 0.4693 -0.2526 0.8146 * 0.0377 0.9739 *
X15 -0.1762 -0.9516 * 0.1249 -0.3639 0.8742 *
Amount of 8.56 1.19 1.07 8.91 1.98
Explanation
Rate of Explanation 0.71 0.10 0.09 0.74 0.16
* p < a=0.05 are significant
Key Engineering Materials Vol. 250 127

Key Engineering Materials Vols. 127

According to Table 5-a and b, the variables were assembled on two factors while the loads
approach to 1 at any factor, it approached to 0 at the other factor. Thus, it is clasiffied that which
variable would be represented by which factor. The factor to which the variables belong with 95 %
confidence interval is illustrated in Table 5-a while these two factors explain together the 82% of total
variance, 18 % of information loss for the vertical cutting , which is in acceptable limit, occurs. As
seen in Table 5-a and b, the amount of wearing accepted as bead variable is obtained as dependent
variable in the models and the remaining 17 independent variables are assembled on two different
factors.
The relationships between dependent and independent variables are investigated by using new
variables which are produced with factorial analysis and thus a new statistical model is derived. The
results obtained from the regression analysis are given in Table 6. These regression models have been
obtained in two separate forms for vertical and horizontal cuttings. The first form (unreduced) covers
all the independent variables. In the second form (reduced), one of the variables which are strongly
interrelated (X2, X11, X13, X16, X17) is eliminated and the rest is used in the regression analysis.

Table 6 Regression models derived for the results of vertical and horizontal cutting

Meaningfull Standart Explanation


Models
(P) Error (R2)
Unreduced Y1= 0.262+0.109F1+0.107F2 0.0003 0.077 0.84
Vertical Y2=0.262+0.111F1-
Reduced 0.0026 0.086 0.82
0.021F2+0.103F3
Unreduced Y1= 0.137+0.035F1+0.109F2 0.00013 0.057 0.83
Horizontal
Reduced Y2= 0.137+0.039F1+0.107F2 0.00037 0.063 0.79

The factor coefficients show the effect of parameters on the amount of wearing in the linear model
for the vertical cutting Y1Vertical=0.262+ 0.109F1+ 0.107F2. For example, when machine and cutting
factor (F2) is held constant, the effect of rock factor (F1) on the amount of wearing is calculated as
0.109. Similarly, the effect of machine and cutting factor is 0.107. It can be seen from these results
that the effect of two factors for vertical cutting are close to each other.

Table 7-a Results of partial correlation for unreduced model for vertical and horizontal measurement

Vertical Horizontal
(F1) (F2) (F1) (F2)
Wearing Rate (Y) 0.840 0.830 0.550 0.930
Meaningful (P) 0.001 0.002 0.064 0.000

Table 7-b. Results of partial correlation for reduced model for vertical and horizontal measurement

Vertical Horizontal
(F1) (F2) (F3) (F1) (F2)
Wearing Rate (Y) 0.671 -0.270 0.611 0.480 0.890
Meaningful (P) 0.034 0.203 0.043 0.071 0.000

The partial effect of the each factor F1 and F2 on wearing is determined by its partial correlation
coefficient which is investigated by varying the factor while the other factor is kept constant. This
methodology gives the comparative effects of the factors on wearing. The statistical results obtained
by partial correlation method are presented in Tables 7-a and b. In these tables, the partial correlation
128 Machining of Natural Stone Materials

128 Machining of Natural Stone Materials

between the factors (F1 and F2) and wearing are given for both horizontal and vertical cuttings. In
Table 7-a, for unreduced model of horizontal cutting, the relationships which exist between the
wearing and rock properties (F1) and wearing and machine and cutting parameters (F2) are 55% and
93% respectively. This result confirms that the effect of machine and cutting parameters on wearing is
to be more pronounced in horizontal cutting.

Table 8-a Factor score coefficient of unreduced variables for vertical and horizontal measurements

Vertical Horizontal
Variables Factor 1 (F1) Factor 2 (F2) Factor 1 (F1) Factor 2 (F2)
X1 0.0821 -0.0700 -0.0877 -0.0003
X2 -0.0808 0.0748 0.0867 -0.0019
X3 -0.0799 0.0880 0.0868 -0.0132
X4 0.0804 -0.0027 -0.0849 -0.0195
X5 0.0825 -0.0430 -0.0874 -0.0155
X6 0.0791 -0.0286 -0.0835 -0.0284
X7 0.0727 -0.0289 -0.0763 -0.0443
X8 0.0775 -0.0320 -0.0824 0.0119
X9 0.0703 -0.0835 -0.0771 0.0640
X10 0.0735 0.0732 -0.0747 -0.0918
X11 -0.0790 0.0741 0.0855 -0.0084
X12 -0.0759 0.1058 0.0837 -0.0138
X13 -0.0727 0.0280 0.0763 0.0442
X14 0.0501 0.3266 0.0028 -0.2580
X15 -0.0138 0.1956 0.0367 -0.2158
X16 0.0481 0.3340 0.0025 -0.2579
X17 0.0492 0.3273 0.0025 -0.2579

Table 8-b Factor score coefficients of reduced variables for vertical and horizontal measurements

Vertical Horizontal
Variables Factor 1 (F1) Factor 2 (F2) Factor 3 (F3) Factor 1 (F1) Factor 2 (F2)
X1 0.1227 -0.0129 -0.1000 0.1111 -0.0033
X3 -0.1324 0.0954 0.2012 -0.1080 0.0289
X4 0.0968 0.0645 0.1118 0.1107 0.0341
X5 0.1138 -0.0018 -0.0118 0.1131 0.0330
X6 0.1187 -0.1039 -0.0722 0.1098 0.0479
X7 0.0779 0.1722 0.1492 0.1032 0.1018
X8 0.1082 0.0436 -0.0314 0.1028 -0.0399
X9 0.1385 -0.2505 -0.3322 0.0885 -0.1428
X10 0.0510 0.2478 0.4106 0.1080 0.1788
X12 -0.1277 0.0744 0.2166 -0.1043 0.0230
X14 -0.0014 -0.1806 0.7531 0.0339 0.5017
X15 -0.0017 -0.7935 0.0816 -0.0149 0.4382

Both factors used in the regression model given above are originated from the linear combination
of variables. When the factor analysis is performed, the factor load of any variable approaches to 1,
Key Engineering Materials Vol. 250 129

Key Engineering Materials Vols. 129

while the others approach to 0. This shows the factor and the ratio for the representation of a variable.
Meanwhile, each factor consists of variables. Therefore, effects of these variables on the factor are
given in terms of factor scores.
The score coefficients of each variable obtained from the factorial results in Tables 5-a and b, are
presented in Table 8-a and b. The results given in the Table 8-a and b show that the effect of variables
on Factor 1 is approximately similar.
The statistical validity of any regression model can be tested by variance analysis method.
However, different approaches can also be used for this purpose. One of these methods can be used to
demonstrate the differences between measured and predicted values with scattered graphs.
Consequently in this study, Figures for the correlation between the results, those obtained from field
measurements and those calculated (predicted) from the linear regression model are given. The
calculated results for unreduced and reduced models that belong to the vertical and horizontal
measurements are given in Fig 5 and Fig 6.
Predicted = 0.04307 + 0.8356 * Measured Predicted = 0.04793 + 0.81705 * Measured
r = 0.91 r = 0.90
0.7 0.7

0.6 0.6

0.5 0.5
Predicted (mm)
Predicted (mm)

0.4 0.4

0.3 0.3

0.2 0.2

0.1 0.1

0 Regression 0 Regression
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 95% confid. 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 95% confid.

Measured (mm) Measured (mm)

Unreduced Reduced
Fig.5 The validity of the unreduced and reduced model for vertical cutting
Predicted = 0.02279 + 0.83309 * Measured Predicted = 0.02817 + 0.79366 * Measured
r = 0.91 r = 0.89
0.4 0.4

0.35 0.35

0.3 0.3

0.25 0.25
Predicted (mm)

Predicted (mm)

0.2 0.2

0.15 0.15

0.1 0.1

0.05 0.05

0 0

-0.05 Regression -0.05 Regression


-0.05 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 95% confid. -0.05 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 95% confid.

Measured (mm) Measured (mm)

Unreduced Reduced
Fig.6 The validity of the unreduced and reduced model for horizontal cutting

Conclusion
This study constitutes an initial step for the analysis of vertical and horizontal diamond wire cutting
operations in andesitic rocks of different physical and mechanical properties. The results obtained
from the vertical and horizontal cuttings are evaluated separately and each in two separate forms of
reduced or unreduced by using the multivariate statistical analysis methods. From the results of
statistical analysis, four wearing rate models are obtained. Thus, it looks logical to use one of these
models instead of all. Among these models, the reduced model either vertical or horizontal is
130 Machining of Natural Stone Materials

130 Machining of Natural Stone Materials

recommended to predict the wearing on beads during the rock cutting with diamond wire, because this
model involves a small quantity of variable in comparison to the others.

Acknowledgements
The author would like to thank Professor Seyfi KULAKSIZ for his many helpful comments and
suggestions during the study.

References
[1] Y. Özçelik: Investigation of the Working Conditions of Diamond Wire Cutting Machines in
Marble Industry (Ph. D. Thesis, Hacettepe University, Ankara, Turkish, 1999) (unpublished).
[2] Y. Özçelik, S. Kulaksız, M.Z. Aydın and H. Yurdugul: A Statistical Method For Practical
Assessment Of Sawability With Diamond Wire Cutting Machine of Ankara-Çubuk Andesites (9th
International Congress on Rock Mechanics, Paris, France, 1999).
[3] Cai and R. Mancini: Dimensional Stone (March 1989), p. 24
[4] Q. Cappuzi: Diamond Wire Cutting Technology in Granite Quarry (ACIMM for Marble, 1989,
November-December, 24).
[5] A.C. Hawkins, A.P. Antenen and G. Johnson: Dimensional Stone (September 1990), p. 44
[6] T. Marles: Industrial Diamond Review No. 2 (1990), p. 62
[7] A. Bortolussi, R. Ciccu, P.P. Manca and G. Massacci: Simulation and Optimization of Rock
Cutting with Diamond Wire (XXII. International Symposium APCOM, Berlin, 1990).
[8] ISRM, Rock Characterization Testing and Monitoring; Suggested Methods (Oxford, 1981).
[9] T.S.E.: Testing and Examination Methods of Natural Building Stones (T.S. 669, T.S.E. Press,
Ankara, Turkish, 1987).
[10] R.A. Johnson and D.W. Wichern: Applied Multivariate Statistical Analysis, (Prentice-Hall, Inc.,
New Jersey, 1982).
[11] R.E. Walpole and R.H. Mayers: Probability and Statistics for Engineers and Scientists
(Macmillan Publishing Company, New York, 1989).
Machining of Natural Stone Materials
10.4028/www.scientific.net/KEM.250

Multivariate Statistical Analysis of the Wear on Diamond Beads in the Cutting of Andesitic Rocks
10.4028/www.scientific.net/KEM.250.118

You might also like