1. International Journal of Science Engineering and Technology Vol. 1, No. 3, 2008
ISSN: 1985-3785
Available online at: www.ijset.org
© 2008 ILRAM Publisher
Application of Taguchi Method in the Optimization of Turning
Parameters for Surface Roughness
1
G. Akhyar, 1C.H. Che Haron, 1J.A. Ghani
1
Department of Mechanics and Materials, Universiti Kebangsaan Malaysia
Abstract
The quality of design can be improved by improving quality and productivity in company-wide activities.
Taguchi’s parameter design is an important tool for robust design, which offers a simple and systematic
approach to optimize a design for performance, quality and cost. Taguchi optimization methodology is applied to
optimize cutting parameters in turning Ti-6%Al-4%V extra low interstitial with coated and uncoated cemented
carbide tools under dry cutting condition and high cutting speed. The turning parameters evaluated are cutting
speed of 55, 75, and 95 m/min, feed rate of 0.15, 0.25 and 0.35 mm/rev, depth of cut of 0.10, 0.15 and 0.20 mm
and tool grades of K313, KC9225 and KC5010, each at three levels. The analysis of results show that the
optimal combination of parameters are at cutting speed of 75 m/min, feed rate of 0.15 mm/min, depth of cut of
0.10 mm and tool grade of KC9225. The cutting speed and tool grade have a significant effect on surface
roughness are 0.000 and have a contribution are 47.146% and 38.881%, respectively. At optimal condition,
contribution of each cutting parameter on surface roughness is reached at 20.47 from tool grade, 21.01 from feed
rate, 11.54 from depth of cut and 11.17 from cutting speed.
KEYWORDS: Taguchi Method, Turning, Ti-6Al-4V ELI and Coated Carbide Tool.
1. Introduction cutting speed of 355 m/min, low feed rate of 0.1mm
per tooth and low depth of cut of 0.5 mm.
The quality of design can be improved by Application of Taguchi’s method for parametric
improving the quality and productivity in company- design was carried out to determine an ideal feed
wide activities. Those activities concerned with rate and desired force combination Although small
quality, include in quality of product planning, interactions exist between a horizontal feed rate and
product design and process design [1, 3]. Robust desired force, the experimental results showed that
design is an engineering methodology for obtaining surface roughness decreases with a slower feed rate
product and process condition, which are minimally and larger grinding force, respectively [7].
sensitive to the various causes of variation, and Conceptual S/N ratio approach of Taguchi method
which produce high-quality products with low provides a simple, systematic and efficient
development and manufacturing costs [1]. methodology for optimizing of process parameters
Taguchi’s parameter design is an important tool for and this approach can be adopted rather than using
robust design. It offers a simple and systematic engineering judgment. Furthermore, the multiple
approach to optimize design for performance, performance characteristics such as tool life, cutting
quality and cost. Signal to noise ratio and force, surface roughness and the over all
orthogonal array are two major tools used in robust productivity can be improved by useful tool of
design. Signal to noise ratio, which measures Taguchi method [8].
quality with emphasis on variation, and orthogonal This paper describes the turning of Ti-6Al-4V
arrays, which accommodates many design factors ELI with parameters of turning at three levels and
simultaneously [1, 2]. four factors each. The main objective is to find a
Taguchi method offers the quality of product is specific range and combination of turning
measured by quality characteristics such as: parameters and interaction to achieve the lowest
nominal is the best, smaller is better and larger is surface roughness value.
better [1, 3]. Optimization using Taguchi method in
end milling using conceptual S/N ratio approach
and Pareto ANOVA proceed, the Taguchi’s robust 2. Taguchi method, design of experiment, and
design method is suitable to analyze the metal experimental details
cutting problem. Ghani et.al [6] found that the
2.1. Taguchi method
conceptual S/N ratio and Pareto ANOVA
approaches for data analysis draw similar
conclusion in process end milling use at high
Corresponding Author: G. Akhyar, Department of Mechanics and Material Engineering, University
Kebangsaan Malaysia, Kuala Lumpur, Malaysia, E-mail: sati1771@gmail.com

2. G. Akhyar et al. / Int. J. Sci. Eng. Tech. Vol. 1, No. 3, 2008, 60-66
Taguchi defines as the quality of a product, in 3. Factors, which do not affect the S/N ratio or
terms of the loss imparted by the product to the process mean.
society from the time the product is shipped to the In practice, the target mean value may change
customer [2]. Some of these losses are due to during the process development applications in
deviation of the product’s functional characteristic which the concept of S/N ratio is useful are the
from its desired target value, and these are called improvement of quality through variability
losses due to functional variation. The uncontrollable reduction and the improvement of measurement.
factors, which cause the functional characteristics of The S/N ratio characteristics can be divided into
a product to deviate from their target values, are three categories when the characteristic is
called noise factors, which can be classified as continuous: nominal is the best, smaller the better
external factors (e.g. temperatures and human errors), and larger is better characteristics.
manufacturing imperfections (e.g. unit to unit
variation in product parameters) and product 2.2. Experimental details
deterioration. The overall aim of quality engineering
is to make products that are robust with respect to all The experiments were carried out with four
noise factors. factors at three levels each, as shown in Table 1.
Taguchi has empirically found that the two stage The fractional factorial design used is a standard
optimization procedure involving S/N ratios, indeed L27 (313) orthogonal array with 20 degree of
gives the parameter level combination, where the freedom [1]. This orthogonal array is chosen due
standard deviation is minimum while keeping the to its capability to check the interactions among
mean on target [2, 3, 4]. This implies that factors.
engineering systems behave in such a way that the The machining trials were carried out on the
manipulated production factors that can be divided lathe machine (Colchester T4 with maximum 6000
into three categories: rpm) in dry condition, as recommended by the tool
1.Control factors, which affect process variability as supplier for the specific work material. The three
measured by the S/N ratio. inserts used were uncoated carbide tool K313
2. Signal factors, which do not influence the S/N (WC-Co), coated carbide tool KC9225 (TiN-
ratio or process mean. Al2O3-TiCN-TiN) CVD and KC5010 (TiAlN)
PVD, respectively.
Table 1. Factors and levels used in the experiment
Levels
Factors
0 1 2
A- Cutting speed (m/min) 55 75 95
B- Feed rate (mm/rev) 0.15 0.25 0.35
C- Depth of cut (mm) 0.10 0.15 0.20
D- Tool type K313 KC9225 KC5010
The maximum flank wear land (VB) was depth of cut and tool grade) to achieve low value of
measured at regular interval of one pass machining the surface roughness. The experimental data for
using Mitutoyo Tool Maker Microscope with 20x the surface roughness values and the calculated
magnification. The surface roughness of machined signal-to-noise ratio are shown in Table 2. The S/N
surface was then measured accordingly surface ratio values of the surface roughness are calculated,
roughness tester model Mpi Mahr Perthometer. The using the smaller the better characteristics [3, 4].
turning process was stop when VB reached 0.2 mm.
3. Results and Discussions 1
S
N
 10 log
n
 y 
2
(1)
3.1 Signal to Noise Ratio (S/N)
The main objective of the experiment is to optimize
the turning parameters (cutting speed, feed rate,
61

3. G. Akhyar et al. / Int. J. Sci. Eng. Tech. Vol. 1, No. 3, 2008, 60-66
Table 2. Experimental results for surface roughness and its calculated S/N ratios.
Factor Surface S/N ratio for
Exp.
Designation roughness, surface
run. A B C D
Ra (m) roughness
1 0 0 0 0 A0B0C0D0 1.71 -4.66
2 0 0 1 1 A0B0C1D1 1.17 -1.364
3 0 0 2 2 A0B0C2D2 2.50 -7.959
4 0 1 0 1 A0B1C0D1 1.09 -0.749
5 0 1 1 2 A0B1C1D2 4.94 -13.875
6 0 1 2 0 A0B1C2D0 3.48 -10.832
7 0 2 0 2 A0B2C0D2 6.01 -15.578
8 0 2 1 0 A0B2C1D0 6.49 -16.245
9 0 2 2 1 A0B2C2D1 2.82 -9.067
10 1 0 0 0 A1B0C0D0 0.53 5.514
11 1 0 1 1 A1B0C1D1 1.56 -3.863
12 1 0 2 2 A1B0C2D2 1.44 -3.168
13 1 1 0 1 A1B1C0D1 4.67 -13.387
14 1 1 1 2 A1B1C1D2 3.02 -9.601
15 1 1 2 0 A1B1C2D0 0.97 0.264
16 1 2 0 2 A1B2C0D2 3.94 -11.91
17 1 2 1 0 A1B2C1D0 2.56 -8.165
18 1 2 2 1 A1B2C2D1 6.19 -15.834
19 2 0 0 0 A2B0C0D0 2.02 -6.108
20 2 0 1 1 A2B0C1D1 1.73 -4.761
21 2 0 2 2 A2B0C2D2 1.10 -0.828
22 2 1 0 1 A2B1C0D1 3.56 -11.029
23 2 1 1 2 A2B1C1D2 2.37 -7.495
24 2 1 2 0 A2B1C2D0 4.31 -12.69
25 2 2 0 2 A2B2C0D2 1.29 -2.212
26 2 2 1 0 A2B2C1D0 4.26 -12.589
27 2 2 2 1 A2B2C2D1 5.20 -14.321
Table 2 shows the actual data of surface Average S/N ratio for each level of experiment
roughness along with its computed S/N ratio is calculated based on the value of Table 1, and
value. Whereas the S/N ratio for each levels of is shown in Table 2. The different values of the
the surface roughness as shown in Table 3. In S/N ratio between maximum and minimum
the standard L27 (313) orthogonal array, factor (main effect) are also shown in Table 2. The
A, B, C and D are arranged in columns 1,2, 5 feed rate and the tool grade are two factors with
and 9, respectively. Whereas interaction factors the highest different in values of 9.553 and
between the cutting speed and feed rate (AxB), 8.445, respectively. Based on Taguchi
the cutting speed and depth of cut (AxC) and prediction that the bigger different in value of
the feed rate and depth of cut (BxC) are S/N ratio shows a more effect on surface
arranged in columns 3, 6 and 8, respectively. roughness or more significant. Therefore, it can
be concluded that, increase changes the feed
3.2 Analysis of variance for S/N ratio rate reduces the surface roughness significantly.
Furthermore, the tool geometry changes, mainly
Taguchi recommends to analyze data
tool nose radius, increase or decrease the
using the S/N ratio that will offer two
surface roughness significantly.
advantages; it provides a guidance for selection
The result of S/N ratio analysis for the
the optimum level based on least variation
surface roughness values, which was calculated
around on the average value, which closest to
using Taguchi Method, is shown in Table 3.
target, and also it offers objective comparison
Then, analysis of variance is shown in Table 4,
of two sets of experimental data with respect to
which consists of DF (degree of freedom), S
deviation of the average from the target [3]. The
(sum of square), V (variance), F (variance ratio)
experimental results are analyzed to investigate
and P (significant factor) [3, 4]. In most
the main effects and differences between the
engineering cases, the significant value selected
main effects of level 0, 1 and 2 on the variables.
was 5% ( = 0.05).
62

4. G. Akhyar et al. / Int. J. Sci. Eng. Tech. Vol. 1, No. 3, 2008, 60-66
Table 3. Average for S/N ratio and main effect of surface roughness
Level average
No. columns Variable Factors Designation Max. - Min. Rank
0 1 2
1 Cutting speed A -8.918 -7.221 -8.213 1.697 4
2 Feed rate B -3.022 -8.757 -12.574 9.553 1
5 Depth of cut C -7.063 -8.961 -8.329 1.897 3
9 Tool grade D -9.403 -3.252 -11.697 8.445 2
Table 4. ANOVA analysis of S/N ratio for surface roughness
Sum of Variance Percent Contribution
No Varible Factors Designation DF F
Square (S) (V) (P) (%)
1 Cutting speed A 2 13.083 6.542 1.36 0.327 1.482
2 Feed rate B 2 416.179 208.089 43.13 0.000 47.146
3 Cutting speed x feed rate AxB 4 44.858 11.214 2.32 0.170 5.082
5 Depth of cut C 2 16.800 8.400 1.74 0.253 1.903
6 Cutting speed x depth of cut AxC 4 5.958 1.490 0.31 0.862 0.675
8 Feed rate x depth of cut BxC 4 13.697 3.424 0.71 0.614 1.552
10 Tool grade D 2 243.223 171.611 35.57 0.000 38.881
Error 6 28.949 4.825 3.279
Total 26 100
Table 4 shows that the significant value 0.170 from AxB, 0.862 from AxC and 0.614%
of the feed rate and tool grade (P) is 0.000. It from BxC. While, a contribution for each
means that the feed rate and tool grade interaction is small.
influences significantly on the surface The most significant factor, which
roughness value at significant value of 0.05. affects the surface roughness measured in
In addition to P value for the cutting speed and turning Ti-6Al-4V, is the feed rate therefore
depth of cut are insignificant. The feed rate and the quality of surface roughness can be
the tool grade have a contribution for the controlled by a suitable feed rate value.
surface roughnesses are 47.146 % and Previous researchers suggest similar results.
38.881%, respectively. From this result, it can They claimed that the surface roughness well
be concluded that the feed rate is more strongly depends on the feed rate followed by
significant factor and give most contribution the cutting speed. Jaharah et al. [6]
on the surface roughness. Bhattacharyya found recommended to obtain better surface finish
that the surface roughness was primarily for specific test range in end milling was use
dependent on the feed rate and the nose radius of high cutting speed (355 m/min), low feed
of tool [9]. The nose radius related to tool rate (0.1 mm/tooth) and low depth of cut (0.5
grade and tool geometry. Since three types of mm).
tool were applied in this experiment have Table 5 shows level of contribution
different tool nose radius, so effect of tool nose each parameter and interaction on surface
geometry changes on surface roughness was roughness values for estimating optimum
significant. condition. The biggest contribution is from
The interaction between the cutting feed rate of 0.15 mm/rev (21.01%), and then
speed and feed rate (AxB), the cutting speed followed by tool grade of KC9225 (20.47%).
and depth of cut (AxC) and the feed rate and The contribution from depth of cut and cutting
depth of cut (BxC) are also insignificant. speed were 11.54% and 11.17%, respectively.
These significant values of interaction are
63

5. G. Akhyar et al. / Int. J. Sci. Eng. Tech. Vol. 1, No. 3, 2008, 60-66
The optimum condition in turning of factors is the cutting speed and feed rate of
aerospace material Ti-6Al-4V ELI which level 1, the cutting speed and depth of cut of
produces a low surface roughness is at cutting level 0, and the feed rate and depth of cut of
speed of level 1 (0.75m/min), feed rate of level level 0. The total contribution for optimum
0 (0.15 mm/rev), depth of cut of level 0 (0.10 condition is 383,898 and its expected result is
mm) and tool grade of level 1 (KC9225). 376.028.
Meanwhile, optimum condition for interaction
Table 5. Estimate of optimum condition for the smaller the better characteristics.
Level of
Factors description Levels Contribution Contribution (%)
description
Cutting speed (A) 0.75 1 42.881 11.17
Feed rate (B) 0.15 0 80.672 21.01
Cutting speed x federate (AxB) - 1 51.780 13.49
Depth of cut (C) 0.1 0 44.303 11.54
Cutting speed x depth of cut (AxC) - 0 43.300 11.28
Feed rate x depth of cut (BxC) - 0 42.360 11.03
Tool grade (D) KC9225 1 78.602 20.47
Contribution for all factors (total) 383.898 100.00
Current grand average of performance -7.87
Expected result at optimum condition 376.028
Main Effects Plot (data means) for SN ratios
Cutting speed Feedrate
-4
-6
-8
Mean of SN ratios
-10
-12
0 1 2 0 1 2
Depth of cut T ype of tool
-4
-6
-8
-10
-12
0 1 2 0 1 2
Signal-to-noise: Smaller is better
Figure 1. Main effects for factors machining verse S/N ratio of surface roughness.
The main effects for each level of parameter It can be seen from Figure 1 that B0 is the
on surface roughness are shown in Figure 1. maximum value with –3.022 of S/N ratio, and
The best choice for machining titanium alloy is decrease dramatically to B1 (-8.757), and then
based on S/N ratio as followed; at cutting speed follow to B2 (–12.574). For the graph of feed
of -8.918 (A1), feed rate of -3.022 (B0), depth of rate, the slope between the horizontal and feed
cut of -7.063 (C0) and tool grade of -3.252 (D1). rate line is bigger. It means that the feed rate
The best combination is A1B0C0D1 that means changes effected significantly on surface
at low high cutting speed, low feed rate, low roughness, and the same trend can also be
depth of cut and CVD coated carbide tool.
64

6. G. Akhyar et al. / Int. J. Sci. Eng. Tech. Vol. 1, No. 3, 2008, 60-66
observed on the graph of tool grade factor for (A1) and the feed rate at level 0 (B0) have a
each level. maximum value.
Figure 2 shows the interaction between the It can be also seen from Table 5 that the
cutting speed and feed rate (AxB), the cutting optimum predicted result for each interaction
speed and depth of cut (AxC) and the feed rate gives contribution is 13.49% from AxB, 11.28%
and depth of cut (BxC). The S/N ratio value at from AxC and 11.03% from BxC. The
(AxB)1 is a best interaction because of it gives contribution for all factors is 383.898, while the
the biggest delta value, and then followed by expected result at optimum condition is 376.028.
interaction (AxC)0. The cutting speed at level 1
Interaction Plot (data means) for SN ratios
0 1 2
0
C utting
speed
-5
0
C utting spee d 1
-10 2
0
Feedrate
0
-5
1
F eedr ate 2
-10
0
Depth
of cut
-5
0
Depth of cut 1
-10 2
0 1 2 0 1 2
Signal-to-noise: Smaller is better
Figure 2. Interaction S/N ratio for surface roughness with smaller is better
4. Conclusions References
From the findings of the following can be [1] Park, S.H. Robust Design and Analysis
concluded: for Quality Engineering; Chapman &
1. Taguchi’s robust design method is suitable Hall, London, 1996.
to optimize the surface roughness in turning [2] Phadke, M.S. Quality Engineering
Ti-6Al-4V ELI. Using Design of Experiment, Quality
2. The significant factors for the surface Control, Robust Design and The
roughness in turning Ti-6Al-4V ELI were Taguchi Method; Wadsworth & Books,
the feed rate and the tool grade, with California, 1988.
contribution of 47.146% and 38.881%, [3] [Ranjit, R. Design of experiment Using
respectively. The Taguchi Approach; John Wiley &
3. The optimal condition for surface roughness Sons Inc., New York, 2001.
in turning Ti-6Al-4V ELI was resulted at [4] Ranjit, R. A Primer on The Taguchi
cutting speed of 75 m/min, feed rate of 0.15 Method, Society of Manufacturing
mm/rev, depth of cut of 0.10 mm and CVD Engineers; Dearborn, Michigan, 1999.
coated carbide with KC9225. [5] Bendell, T. Taguchi Method,
4. The optimal interaction parameter was Proceedings of the European
between the cutting speed and feed rate at Conference of Taguchi Method,
level 1 (75 m/min). Elsevier, Amsterdam, July 13-24, 1988.
[6] Ghani, J.A., Choudhury, I.A., Hasan,
H.H. Application of Taguchi Method in
Optimization of End Milling
Parameters, Journal of Materials
65

7. G. Akhyar et al. / Int. J. Sci. Eng. Tech. Vol. 1, No. 3, 2008, 60-66
Processing Technology (2004) 145: 84– Parameters on Cutting Force and
92. Torque During Drilling of Glass-Fiber
[7] Liu, C.H., Andrian, C., Chen, C.A., Polyester reinforce Composite, Journal
Wang, Y.T. Grinding Force Control in of Composite Structures (2005) 71:
Automatic Surface Finish System, 407–413.
Journal of Materials Processing [9] Bhattacharyya. Metal cutting theory
Technology (2005) 170: 367–373. and practice, New Central Book
[8] Mohan, N.S., Ramachndra, A., Agency, Calcutta 1998.
Kulkarni, S.M. Influence of Process
66