Application of Central Composite and Orthogonal Array Design ...

Application of Central Composite and

Orthogonal Array Designs for Predicting
the Cutting Force
Srinivasa Rao G.*1 and Neelakanteswara Rao A.2
Mechanical Engineering Department, RVR&JC College of Engineering,
Guntur-522019, A.P., India, [email protected]
1

2
Mechanical Engineering Department, National Institute of Technology,
Warangal-506004, A.P.,India, [email protected]

Abstract: Development of empirical models for cutting forces

in turning operation has been a major activity in metal cutting
research. This paper presents an application of central composite
face centered (CCF) and L 18(21 37) orthogonal array designs
for developing cutting force model in case of turning AISI 1015
steel with HSS tool. Three process parameters namely cutting
speed, feed, depth of cut and three tool parameters namely side
cutting edge angle, inclination angle, normal rake angle were
considered in developing the cutting force model. Each
parameter was set at three levels. Based on the experimental
data, a mathematical model in terms of process and tool
parameters was developed for main cutting force using multiple
linear regression. Confirmation tests were performed to verify
the predictability of the developed model.
Keywords: Turning, cutting force, central composite, orthogonal
array, multiple regression.

1. INTRODUCTION
In a machining process, turning operation plays an important role
in reducing a particular work piece from the original stock to the
desired shape and size. In order to achieve economic objective of
*Corresponding Author: [email protected]
Journal of Metallurgical Engineering, 1(1-2) January-December 2011

17

Srinivasa Rao G. and Neelakanteswara Rao A.

F

this process, optimal cutting conditions have to be determined.

Although one can determine the desirable cutting conditions based
on experience or hand book data, it does not ensure that the
conditions obtained will be optimal or near optimal for that
particular work-tool combination. Thus, in order to determine the
optimal cutting conditions, reliable mathematical models need to
be established. To ensure the effectiveness of the models, the design
of experimental techniques should be used to plan the machining
experiments efficiently and multiple regression methods can then
be used for the particular work-tool combination based on the
machining data collected on a specific machine.
Force modeling in turning is important for a multitude of
purposes including tool life estimation, chatter prediction, tool wear
monitoring, thermal analysis, etc. Empirical approach has been a
popular approach in developing cutting force model. Cutting force
in turning has been found to be influenced in varying amounts by a
number of factors such as speed, feed, depth of cut, work material
characteristics, tool geometry, use of cutting fluids, etc [1, 2]. In the
early phases of empirical work, researchers used one-factor-at-atime strategy, i.e., varying one factor while keeping all other factors
constant. This strategy cannot provide generalized conclusions about
factor effects. Further, interactions between factors cannot be studied.
To overcome these limitations, researchers shifted their strategy to
full factorial designs and orthogonal array designs. Full factorial
designs provide more complete information, but require large
number of experiments. Orthogonal array designs, which are in some
way fractional factorial designs, require less number of experiments,
but provide lesser information compared to full factorial designs.
Experiments based on 33 full factorial designs were conducted
[3] for prediction of tool life, force and power in terms of speed,
feed and depth of cut. Optimization of the cutting speed, feed rate,
and depth of cut with considerations of multiple performance
characteristics including tool life, cutting force and surface roughness
was done [4] using L9 orthogonal array (OA) design experiments.
18

Journal of Metallurgical Engineering, 1(1-2) January-December 2011

Application of Central Composite and Orthogonal Array Design ...

F

The effects of shape of cutting edge, work piece hardness, feed rate
and cutting speed on surface roughness and resultant forces were
experimentally investigated [5] by using a four-factor two-level
factorial design. An experimental investigation was conducted to
determine the effects of cutting speed, feed, effective rake angle and
nose radius on the surface roughness in the finish hard turning of
the bearing steel based on a 3 4 full factorial deign [6]. An L 9
orthogonal array has been used [7] to determine the optimum levels
for the parameters insert radius, feed rate, and depth of cut on surface
roughness while turning AISI 1030 steel bars.
An abductive network technique was adopted [8] to construct a
prediction model for surface roughness and cutting force and once
the process parameters (cutting speed, feed rate and depth of cut)
are given, the surface roughness and cutting force can be predicted
by this network. Artificial neural network approach has been
proposed [9] for modeling cutting forces. By seeing the literature, it
can be observed that the study of factorial effects and the empirical
model building was performed based on OA or factorial designs,
particularly in case of force modeling. Further, tool related factors
are excluded from most of the studies.
Central composite designs, which are a combination of 2-level
factorials and one-factor-at-a-time experiments, are known to be best
fit for fitting a second-order model [10]. In recent past, researchers
have been using central composite designs for developing surface
roughness models [11-13]. One of the key issues in empirical based
research is selection of experimental designs. It has been observed
that researchers adopted different experimental designs for similar
works. It might be of curiosity to verify the performance of
orthogonal array design and Central composite designs with respect
to cutting force. In this work, an attempt has been made to study
the effectiveness of the two designs for the same work-tool
combination. Further, in the present work, both process parameters,
namely, cutting speed, feed and depth of cut, and tool parameters,
namely, side cutting edge angle, inclination angle and normal rake
Journal of Metallurgical Engineering, 1(1-2) January-December 2011

19

Srinivasa Rao G. and Neelakanteswara Rao A.

F

angle are considered for the development of the cutting force model
while turning AISI 1015 steel with HSS tool. Experiments were
conducted based on central composite face centered design and L18
orthogonal array design. Separate models were developed using
the data obtained from each design. Confirmation experiments were
conducted to verify the adequacy of the models.
2. PROCEDURES AND METHODS
2.1 Cutting Force Model
The proposed relationship between the cutting force and machining
independent variables can be represented by the following:
Force = kAm Bn Cp Dq Er Fs

(1)

Where F is the main cutting force, k, m, n, p, q, r, s are the

constants, A (side cutting edge angle), B (inclination angle),
C (normal rake angle) are tool parameters, and D (cutting speed),
E (feed rate), F (depth of cut) are process parameters. To facilitate
the determination of constants and parameters, the mathematical
model is linearized by performing a logarithmic transformation. The
logarithmic transformed mathematical model is given by:
ln (Force) = ln k + m ln A + n ln B + p ln C + q ln D + r ln E + s ln F
(2)
Though a multiplicative model like equation (1) implicitly
incorporates interaction effects, upon its logarithmic transformation,
equation (2) becomes a simple linear form without any interaction
terms.
The second order model also is useful when the second order
effects and the two way interactions amongst the process parameters
and tool parameters were significant. The general second order
model for six parameters is of the form shown below:
Y = c0 + c1x1 + c2x2 + c3x3 + c4x4 + c5x5 + c6x6 + c11x12 + c22x22 + c33x32 +
c44x42 + c55x52 + c66x62 + c12x1x2 + c13x1x3 + c14x1x4 + c15x1x5 + c16x1x6 +
c23x2x3 + c24x2x4 + c25x2x5 + c26x2x6 + c34x3x4 + c35x3x5 + c36x3x6 +c45x4x5 +
c46x4x6 + c56x5x6
(3)
20

Journal of Metallurgical Engineering, 1(1-2) January-December 2011

Application of Central Composite and Orthogonal Array Design ...

F

Where x1, x2, x3, x4, x5, x6 are the independent variables, c0, c1, c2,..
c56 are the constants and y is the response. Equation (3) is useful
when the second order effects of variables and the two way
interactions among the variables are significant. In the present study,
the parameters of equations (2) and (3) have been estimated by the
multiple linear regression using a SPSS software package.
2.2 Materials and Processes
The experiments were carried out on TMX-2030 engine lathe. The
HSS tools (Co- 10%, W-9.3%, Cr-4.0%, Mo-3.6%, C-1.26%) with
required cutting angles were ground on a tool and cutter grinding
machine using the standard procedure [1]. Tool geometry used was
as follows: end cutting edge angle 10, normal side clearance angle
8, and normal end clearance angle 8, were fixed and remaining
side cutting edge angle, inclination angle and normal rake angles
were changed during experimentation.
Table 1
The Chemical Composition of AISI 1015 Steel
Element

Si

Mn

Cr

Ni

Mo

Fe

0.157

0.155

0.598

0.037

0.021

0.012

0.006

0.001 99.0

Table 2
Level of Control Factors
Factor
symbol

Factor

Side cutting edge angle(degrees)

Inclination angle(degrees)

Normal rake angle (degrees)

Cutting speed(m/min)

Feed (mm/rev)

Depth of cut(mm)

Level1(-1)

Level2 (0) Level3 (+1)

10

20

30

10

20

15

25

40

60

80

0.052

0.104

0.156

0.2

0.4

0.6

Journal of Metallurgical Engineering, 1(1-2) January-December 2011

21

Srinivasa Rao G. and Neelakanteswara Rao A.

F

Table 3
The Experimental Layout: L 18 Orthogonal Array
Exp.No. A
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18

1
1
1
0
0
0
+1
+1
+1
1
1
1
0
0
0
+1
+1
+1

Force(kgf)

1
0
+1
1
0
+1
0
+1
1
+1
1
0
0
+1
1
+1
1
0

1
0
+1
0
+1
1
1
0
+1
+1
1
0
+1
1
0
0
+1
1

1
0
+1
1
0
+1
1
0
+1
1
0
+1
1
0
+1
1
0
+1

1
0
+1
+1
1
0
0
+1
1
0
+1
1
+1
1
0
1
0
+1

1
0
+1
+1
1
0
+1
1
0
1
0
+1
0
+1
1
0
+1
1

3.00
8.50
15.50
18.50
2.50
11.00
16.00
7.00
5.00
5.00
15.00
7.00
12.00
9.00
5.00
4.00
12.00
8.00

The material used in the tests for controlled machining was AISI
1015 steel. The chemical composition of the AISI 1015 steel with a
137 HB hardness, is given in Table 1. The steel bar stock was 65 mm
diameter, 300 mm length and these bars were trued, centered and
cleaned by removing a 2 mm depth of cut from the outside surface,
prior to the actual machining tests. The main cutting force was
measured with a multi component digital force indicator (IEICOS
made and model 652). The force signals were amplified by a
3-channel charge amplifier. The range of force measurement is
0-100 kg.
2.3 Experimental Design
Design of experimental techniques was used for execution of the
plan of experiments, for six variables at three levels, whereby the
levels are the values taken by the factors. The factors to be studied
22

Journal of Metallurgical Engineering, 1(1-2) January-December 2011

Application of Central Composite and Orthogonal Array Design ...

F

and the level of each factor are given in Table 2. The levels of each
factor were selected based on machining data hand book [14].
Experiments were conducted as per central composite face centered
(CCF) design [10] and orthogonal array based designs [15]. This
research assumes that the three-four- and five-factor interactions
are negligible, because high order interactions are normally assumed
highly impossible in practice. For six factors, the CCF design consists
of 45 runs, which includes a 26-1 fractional factorial portion
(32 experiments), 12 axial points and a central point. For six 3-level
factors, the smallest (in terms of number of experiments) orthogonal
array design is L18 (21 37), which consists of eight columns, and
one 2-level factor and seven 3-level factors can be accommodated.
For the present work, the factors side cutting edge angle (A),
inclination angle (B), normal rake angle (C), cutting speed (D), feed
rate (E), and depth of cut (F) were assigned to columns 2, 3, 4, 5, 7
and 8, and columns 1 and 6 were kept empty. The empty columns
provide the necessary degrees of freedom for error estimation. The
experimental layouts along with the response (main cutting force)
obtained are shown in Tables 3 and 4. In the tables, 1 indicates
level 1 of the factor, 0 indicates level 2 of the factor and +1 indicates
level 3 of the factor. Experiments were conducted in random order
to avoid any bias.
Table 4
The Experimental Layout : CCF (6) Design
Exp.No. A
1
1
2
1
3
1
4
1
5
1
6
1
7
1
8
1
9
1
10
1

B
1
1
1
1
1
1
1
1
+1
+1

C
1
1
1
1
+1
+1
+1
+1
1
1

D
1
1
+1
+1
1
1
+1
+1
1
1

E
1
+1
1
+1
1
+1
1
+1
1
+1

F
1
+1
+1
1
+1
1
1
+1
+1
1

Force(kgf)
3.00
20.00
8.50
7.00
5.00
6.00
1.50
15.00
8.00
7.00
Table Contd

Journal of Metallurgical Engineering, 1(1-2) January-December 2011

23

Srinivasa Rao G. and Neelakanteswara Rao A.

F

Table 4 Contd
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46

1
1
1
1
1
1
+1
+1
+1
+1
+1
+1
+1
+1
+1
+1
+1
+1
+1
+1
+1
+1
1
+1
0
0
0
0
0
0
0
0
0
0
0
0

+1
+1
+1
+1
+1
+1
1
1
1
1
1
1
1
1
+1
+1
+1
+1
+1
+1
+1
+1
0
0
1
+1
0
0
0
0
0
0
0
0
0
0

1
1
+1
+1
+1
+1
1
1
1
1
+1
+1
+1
+1
1
1
1
1
+1
+1
+1
+1
0
0
0
0
1
+1
0
0
0
0
0
0
0
0

+1
+1
1
1
+1
+1
1
1
+1
+1
1
1
+1
+1
1
1
+1
+1
1
1
+1
+1
0
0
0
0
0
0
1
+1
0
0
0
0
0
0

1
+1
1
+1
1
+1
1
+1
1
+1
1
+1
1
+1
1
+1
1
+1
1
+1
1
+1
0
0
0
0
0
0
0
0
1
+1
0
0
0
0

1
+1
1
+1
+1
1
+1
1
1
+1
1
+1
+1
1
1
+1
+1
1
+1
1
1
+1
0
0
0
0
0
0
0
0
0
0
1
+1
0
0

3.00
19.00
3.00
16.00
6.00
5.00
7.00
7.50
2.50
19.00
2.50
16.50
6.00
5.50
2.50
22.00
7.00
6.50
6.00
5.00
2.00
15.50
8.50
8.00
8.00
9.00
9.00
8.00
9.00
8.00
4.50
12.00
4.00
12.00
8.00
8.00

3. DATA ANALYSIS AND DISCUSSION OF RESULTS

The plan of tests was developed aiming at determining the relation
between the process parameters and tool parameters with the cutting
force. The analysis of experiments was made into two phases. The
24

Journal of Metallurgical Engineering, 1(1-2) January-December 2011

Application of Central Composite and Orthogonal Array Design ...

F

first one concerned the analysis of the effects of factors and of the
Interactions. Models for cutting force in terms of process parameters
and tool parameters were developed in second phase. Finally, the
comparison between the models has been made.
3.1 Analysis of the Factors and Interactions
From L18 orthogonal array design, factor affects at three levels can
be obtained and interaction affects cannot be studied. Since the
experimental design is orthogonal, it is then possible to separate
out the effect of each parameter at different levels. The influence of
each control factor on the response considered i.e. cutting force, has
been performed with level mean analysis. A level mean of a factor
is the average of the response value of experiments in which the
factor is at the particular level. For example, the mean value of the
response for the side cutting edge angle at level 1, 2 and 3 can be
calculated by averaging the response for experiments 1-3 & 10-12,
4-6 & 13-15 and 7-9 & 16-18 respectively. The mean of the response
for each level of the other cutting parameters can be computed in a
similar manner. The control factor with the strongest influence is
determined by the difference between mean values of the factor at
high and low levels.
From the factorial portion of CCF design, both factor and
interaction effects (at two levels) can be obtained. It can be observed
from axial and central portion of CCF design, considering
experiments from 33 to 45, factor effects (at three levels) of each
factor can be obtained when all other factors are at 0 levels. Using
the experimental data, level means have been calculated. The level
means obtained from L18 design, factorial portion of CCF design,
and axial portion of CCF design are given in Table 5 to 7. The
influence of each control factor can be more clearly presented with
response graphs. A response graph shows the change of the response
when the settings of the control factor are changed from one level to
the other. The slope of the line determines the power of influence of
a control factor. Corresponding response plots are presented in Figs.
1 to 3.
Journal of Metallurgical Engineering, 1(1-2) January-December 2011

25

Srinivasa Rao G. and Neelakanteswara Rao A.

F

Table 5
Average Response for L 18 Design
Level 1
Level 0
Level +1

A
9.00
9.67
8.67

B
9.75
9.00
8.58

C
10.33
8.33
8.67

D
9.75
9.00
8.58

E
5.08
9.58
12.67

F
5.08
9.25
13.00

Table 6
Average Response for Factorial Portion of CCF Design
Level 1
Level +1

8.31
8.31

8.28
8.34

9.34
7.28

8.56
8.06

4.59
12.03

4.34
12.28

Table 7
Average Response for One Factor at a Time Analysis
Level 1
Level 0
Level +1

8.50
8.00
8.00

8.00
8.00
9.00

9.00
8.00
8.00

9.00
8.00
8.00

4.50
8.00
12.00

4.00
8.00
12.00

Fig. 1: Response Plot for L18 Design

Fig. 2: Response Plot for Factorial of CCF Design

26

Journal of Metallurgical Engineering, 1(1-2) January-December 2011

Application of Central Composite and Orthogonal Array Design ...

F

Fig. 3: Response Plot for One Factor at a Time Analysis

From the tables of level mean analysis, it can be observed that

the feed rate and depth of cut have been showing consistent
behavior. As the feed and depth of cut increases from 1 level to
+1 level, the force increases, whereas other factors do not show the
same consistency. For example, as the side cutting edge angle
changes from 1 level to +1 level in the L18 data, the cutting force
increases from 1 level to 0 level then decreases from 0 level to
+1 level. The same is not the case with one-factor-at-a-time analysis
portion of CCF design (Table 7). Perhaps, the reason might have
been the presence of interaction effects. The analysis of interactions
gives additional information about the process. Interaction effects
can be obtained by calculating all combinations of two control
factors. For example, the interaction Ax D has four possible
combinations of control factor settings: A1D1, A1D2, A2D1 and
A2D2. The interaction matrix enables the construction of interaction
graphs, which indicate the existence or non-existence of interaction
between two control factors. If the lines in the interaction graph are
parallel, it indicates non-existence of interaction. The interaction
matrix and interaction graphs are shown in Table 8 and Fig. 4. From
the interaction graphs, it can be observed that the interactions
between the feed and depth of cut, feed and normal rake angle,
depth of cut and normal rake angle are quite prominent. The
interaction graph of control factors speed and feed shows a lesser
influence, and remaining interactions are not prominent. In essence,
it can be concluded that feed and depth of cut are having strong
impact on cutting force followed by rake angle and speed.
Journal of Metallurgical Engineering, 1(1-2) January-December 2011

27

Srinivasa Rao G. and Neelakanteswara Rao A.

F

Table 8
Interaction Matrices for the Cutting Force
AXB B1
B2
A1
8.25 8.38
A2
8.31 8.32

AXC C1 C2
A1
9.43 7.19
A2
9.25 7.38

AXD D1 D2
A1
8.50 8.12
A2
8.63 8.00

AXE
A1
A2

E1
4.75
4.43

E2
11.89
12.18

AXF
A1
A2

F1
F2
4.44 12.18
4.25 12.38

BXC C1 C2
B1
9.31 7.25
B2
9.38 7.31

BXD D1 D2
B1
8.43 8.13
B2
8.69 8.00

BXE
B1
B2

E1
4.50
4.69

E2
12.06
12.00

BXF
B1
B2

F1
F2
4.44 12.12
4.25 12.44

CXD D1 D2
C1
9.63 9.06
C2
7.50 7.06

CXE E1 E2
C1
5.19 13.50
C2
4.00 10.56

CXF
C1
C2

F1
4.88
3.81

F2
13.81
10.75

DXE
D1
D2

E1
E2
4.63 12.50
4.56 11.56

DXF F1
F2
D1
4.56 12.56
D2
4.13 12.00

EXF F1
F2
E1
2.50 6.69
E2
6.19 17.87

Fig. 4: Interaction Graphs for Parameters

28

Journal of Metallurgical Engineering, 1(1-2) January-December 2011

Application of Central Composite and Orthogonal Array Design ...

F

Fig. 4(a): Interaction Graphs (Continued)

3.2 Model for L18 Design

As stated earlier L18 design does permit estimation of interaction
effects, and accordingly L18 design data has been used to fit the first
order model. The input data to SPSS software is provided in the
Journal of Metallurgical Engineering, 1(1-2) January-December 2011

29

Srinivasa Rao G. and Neelakanteswara Rao A.

F

logarithmic transformation of actual values of factors. Backwards

linear regression, which eliminates the insignificant factors one at a
time, option of SPSS is used to estimate the parameters and the final
results are shown in Table 9.
Table 9
Parameter Estimates
Variable

Parameter estimate

Intercept
ln(C)
ln(E)
ln(F)

Standard error

0.126
0.027
0.040
0.040

42.901
-5.644
21.632
21.635

5.392
-0.154
0.873
0.875

Sig.
0.000
0.000
0.000
0.000

Analysis of variance for the model

Source

df

Sum of squares Mean Square

F-value

Sig.

Model
Error
Total

3
14
17

5.864
8.482E-02
5.949

322.609

0.000

1.955
6.059E-02

R-square = 0.986

Based on the above results, the cutting force model developed

is given by:
ln (Force) = 5.392 0.154 ln (C) + 0.873 ln (E) + 0.875 ln (F)
or
Force = 219.64 C0.154 E0.873 F0.875
The R-square value of 0.986 indicates that 98.60% of the
variability in cutting force was explained by the model. It can be
observed that normal rake angle, feed and depth of cut are only
coming into model. Based on the mathematical model, it can be
observed that feed, depth of cut and normal rake angle are showing
a prominent effect on the cutting force. The other factors have no
significant effect on the cutting force. Perhaps it is consistent with
the level means analysis of L18 data.
30

Journal of Metallurgical Engineering, 1(1-2) January-December 2011

Application of Central Composite and Orthogonal Array Design ...

F

3.3 Model for CCF Design

As stated earlier, central composite designs are best for fitting a
second order model, and accordingly CCF data is used to fit a second
order model. Procedure stated above is applied for developing
cutting force model using CCF design. The input data to SPSS
software is provided in coded form of factors i.e. 1 to +1. To be
precise, value of the factor in coded scale is = (actual value of the
factor central value in the range)/ (difference between maximum
value and central value in the range). For example the cutting speed
value of 40 m/min is coded as 40-60/20 = 1. The parameter
estimates obtained from SPSS software are shown in Table 11.
Table 10
Parameter Estimates
Variable

Parameter Estimate

Constant
C
D
E
F
CxE
Cx F
Dx E
Ex F

Standard Error

8.304
-1.000
-0.265
3.721
3.971
-0.437
-0.500
-0.219
1.875

0.082
0.096
0.096
0.096
0.096
0.099
0.099
0.099
0.099

Sig.

100.953
-10.451
-2.767
38.886
41.498
-4.436
-5.070
-2.218
19.011

0.000
0.000
0.009
0.000
0.000
0.000
0.000
0.033
0.000

Analysis of variance for the model

Source

df

Model
Error
Total

8
37
45

Sum of Squares

Mean Square

1171.222
11.517
1182.739

146.403
0.311

F-value

Sig.

470.352

.000

R-square=0.990
Table 11
Cutting conditions used in confirmation tests
Test

2
3
4

20
30
20

10
20
10

25
15
5

40
60
40

0.156
0.104
0.156

Journal of Metallurgical Engineering, 1(1-2) January-December 2011

F
0.6
0.4
0.6
31

Srinivasa Rao G. and Neelakanteswara Rao A.

F

by:

The cutting force model developed using CCF design was given

Force = 8.304 1.0C 0.265D + 3.721 E + 3.971F 0.437CxE

0.5CxF 0.219DxE + 1.875ExF
as:

Cutting force model in terms of actual factors can be expressed

Force = 8.304 1.0 normal rake angle 0.265 cutting speed +

3.721 feed + 3.971 depth of cut 0.437 normal rake angle x feed
0.50 normal rake angle x depth of cut 0.219 cutting speed x feed +
1.875 feed x depth of cut.
The R-square value of 0.990 indicated that 99.00% of the
variability in cutting force was explained by the model. In
comparison to the previous model, it can be observed that the cutting
speed apart from normal rake angle, feed and depth of cut is also
coming into the model. Further, four interaction terms are also into
the model. This is consistent with the level means and interaction
effects analysis of CCF design.
3.4 Confirmation Tests
Conducting confirmation experiments has been the final step of the
design of Experimental (DOE) process. The confirmation is
performed by conducting tests using combinations of the factors
and levels that are not previously evaluated. Table 12 shows the
conditions used in the confirmation tests. Table 13 shows the results
obtained where a comparison was done between the foreseen values
from the models developed in the present work with the values
obtained experimentally.
Table 12
Confirmation Tests and their Comparison with the Results
Test
2
3
4
32

CCF design
Experiment Model
Error (%)
16.5
8.00
21.00

16.54
8.28
20.30

0.24
3.50
3.33

L18 design
Experiment Model
16.50
8.00
21.00

16.95
7.74
21.68

Error (%)
2.73
3.37
3.24

Journal of Metallurgical Engineering, 1(1-2) January-December 2011

Application of Central Composite and Orthogonal Array Design ...

F

It can be observed that the error percentages associated with

both the models have been within the limits. Therefore, we can
consider the empirical models, which correlate the cutting force with
the process and tool parameters, with a reasonable degree of
approximation within the given working conditions.
4. CONCLUSIONS
In this work, the cutting force models have been developed by
considering both process and tool parameters. Central composite
and orthogonal array designs are used to develop the models. Based
on the work, the following conclusions may be drawn:
1.

It can be observed from L18 design that normal rake angle,

feed and depth of cut are only coming into model whereas
in CCF design the cutting speed apart from normal rake
angle, feed and depth of cut is also coming into the model.
Further, four interaction terms are also into the model.
Interactions between feed and depth of cut, feed and normal
rake angle, depth of cut and normal rake angle, and
cutting speed and feed are having significant impact on the
cutting force.

2.

Both L18 and CCF (6) designs have been showing same
prediction accuracy for cutting force. From cutting force
equation, it can be observed the absence of quadratic effects
of factors, and the linear effect of factors has been only
identified. In such cases, the L18 design is sufficient to reveal
the information regarding the process.

3.

Significance of interaction terms of parameters has been

clearly predicted in CCF design, whereas none of them are
considered in L18 design. This is owing to the fact that in
Taguchis design interactions between control factors are
aliased with their main effects.

4.

By explicit incorporation of the interaction terms, the CCF

design model gives better insight into the process. So, CCF
design model developed for cutting force serves as a good
alternative to the popular multiplicative model.

5.

The significance of normal rake angle stresses the need for

inclusion of tool parameters in empirical model building
studies.

Journal of Metallurgical Engineering, 1(1-2) January-December 2011

33

Srinivasa Rao G. and Neelakanteswara Rao A.

F

REFERENCES
[1] Armarego E.J.A., and Brown R.H., The Machining of Metals, (1969) PrenticeHall, New Jersey.
[2] Arshinov V., and Alekseev G., Metal Cutting Theory and Cutting Tool
Design, (1976) MIR Publishers, Moscow.
[3] Chua M.S., and Rahman M., Determination of Optimal Cutting Conditions
using DOE and Optimization Techniques, International Journal of Machine
Tools and Manufacturer, (1993) 33(2), pp. 297-305.
[4] Nian C.Y., Yang W.H., and Tarang Y.S., Optimization of Turning Operations
with Multiple Performance Characteristics, Journal of Materials Processing
Technology, (2000), Vol. 100, pp. 417-423.
[5] Ozel T., Hsu T.K., and Zeren E., Effects of Cutting Edge Geometry, Work
Piece Hardness, Feed Rate and Cutting Speed on Surface Roughness and
Forces in Finish Turning of Hardened AISI H13 Steel, Int.Jr.of Advanced
Manufacturing Technology, (2005), Vol. 25, pp. 262-269.
[6] Singh D., and Rao P.V., A Surface Roughness Prediction Model for Hard
Turning Process, International Journal of Advanced Manufacturing Technology,
( 2007), Vol. 32, pp. 1115-1124.
[7] Nalbant M., Gokkaya H., and Sur G., Application of Taghcui Method in the
Optimization of Cutting Parameters for Surface Roughness in Turning,
Materials and Design, (2007), Vol. 28, pp. 1379-1385.
[8] Lin W.S., Lee B.Y., and Wu C.L., Modeling the Surface Roughness and
Cutting Force in Turning, Journal of Materials Processing Technology, (2001),
Vol. 108, pp. 286-293.
[9] Szecsi T., Cutting Force Modeling Using Artificial Neural Networks, Journal
of Materials Processing Technology, (1999), Vol. 92-93, pp. 344-349.
[10] Montgomery D.C., Design and Analysis of Experiments, (1984), Wiley,
New York.
[11] Choudary I.A., and Baradie M.A., Surface Roughness Prediction in the
Turning of High-strength Steel by Factorial Design of Experiments, Journal
of Materials Processing Technology, (1997), Vol. 67, pp. 55-61.
[12] Puerts Arbizu I., and Luis Perez C.J., Surface Roughness Prediction by
Factorial Design of Experiments, Journal of Materials Processing Technology,
(2003), Vol. 143-144, pp. 390-396.
[13] Noordin M.Y., Venkatesh V.C., Sharif S., Elting S., and Abdulla A.,
Application of Response Surface Methodology in Describing the Performance
of Coated Carbide Tools when Turning AISI 1045 Steel, Journal of Materials
Processing Technology, (2004), Vol. 145, pp. 46-58.
[14] Central Machine Tool Institute, Bangalore. Machine Tool Design Hand
Book, (1989), Tata McGraw-Hill Publishing Co., Ltd. New Delhi.
[15] Phadke M.S., Quality Engineering using Robust Design, (1989), PrenticeHall, New Jersey.
34

Journal of Metallurgical Engineering, 1(1-2) January-December 2011