utility.pdf

Journal of Engineering Science and Technology
Vol. 10, No. 1 (2015) 81 - 96
© School of Engineering, Taylor’s University
81
PARAMETRIC OPTIMIZATION IN ELECTROCHEMICAL
MACHINING USING UTILITY BASED TAGUCHI METHOD
SADINENI RAMA RAO1,
*, G. PADMANABHAN2
1
Department of Mechanical Engineering, SVCET, RVS Nagar, Chittoor-517127, India
2
Department of Mechanical Engineering, S. V. University, Tirupathi-517502, India
*Corresponding Author: ramaraosadineni1@gmail.com
Abstract
The present work deals the application of Taguchi method with utility concept
to optimize the machining parameters with multiple characteristics in
electrochemical machining (ECM) of Al/B4C composites. L27 orthogonal array
was chosen for the experiments. METATECH ECM setup is used to conduct
the experiments. The ECM machining parameters namely applied voltage,
electrolyte concentration, electrode feed rate and percentage of reinforcement
are optimized based on multiple responses, i.e., material removal rate, surface
roughness and radial over cut. The optimum machining parameters are
calculated by using utility concept and results are compared with ANOVA. The
results show that the feed rate is the most influencing parameter which affects
the multiple machining characteristics simultaneously. The optimum parametric
combination to maximize the material removal rate and to minimize surface
roughness and radial over cut simultaneously are, applied voltage 16 V, feed
rate 1.0 mm/min, electrolyte concentration 30 g/L and reinforcement content
5 wt%. Experimental results show that the responses in electrochemical
machining process can be improved through this approach.
Keywords: Electrochemical machining, utility function, Taguchi method, ANOVA.
1. Introduction
Aluminium metal matrix composites (AMMCs) have high strength to weight
ratio, low wear rate and low thermal coefficient of expansion and are used in
aerospace, biomedical, defence, automotive industries, etc. [1]. Since abrasive
properties of the reinforced particles, high tool wear associated with the
conventional machining of AMMCs and hence the machining is difficult or costly
[2]. Electrochemical machining is the one of the unconventional machining process

82 S. R. Rao and G. Padmanabhan
Journal of Engineering Science and Technology January 2015, Vol. 10(1)
Nomenclatures
A Applied voltage
Al Aluminium
B Tool feed rate
B4C Boron carbide
C Electrolyte concentration
D Percentage of reinforcement
DF Degrees of freedom
k Number of responses
MS Mean sum of squares
n Number of observations
NaCl Sodium chloride
Pi Preference number of response ‘i’
p Probability
q Number of machining parameters
S/N ratio Signal-to-noise ratio
SS Sum of squares

Overall mean of the response
U Utility value
Wi Weight of a response ‘i’
xi Any value of response ‘i’

∗ Predicted optimum value of response ‘i’

/ Minimum or maximum acceptable value of response ‘i’
Z Constant
Greek Symbols
Mean utility value
Mean utility value at optimum conditions

Predicted utility value
µ Estimated mean of the response
Abbreviations
AMMC Aluminium metal matrix composite
ANOVA Analysis of variance
ECM Electrochemical machining
IEG Inter electrode gap
MRR Material removal rate
ROC Radial over cut
SR Surface roughness
used for machining extremely hard materials into complex shapes which is
difficult to machine by conventional methods [3]. In ECM process machining is
done by the controlled movement of the tool into the workpiece, which produces
the tool shape in the worpiece by mass migrational transport [4].
Effect of intervening variables like, feed rate, electrolyte, flow rate of the
electrolyte and voltage on the responses material removal rate, roughness and
over-cut were studied in electrochemical machining of SAE-XEV-F Valve-Steel
and find out feed rate is the main parameter which effects on the material removal

Parametric Optimization in Electrochemical Machining Using Utility Based . . . . 83
rate [5]. Senthil Kumar, et al., studied the effect of various process parameters
like applied voltage, electrolyte concentration, feed rate and percentage of
reinforcement on the response material removal rate in electrochemical machining
of A356/SiCp composites [6].
Multiple regression analysis and artificial neural networks (ANN) were used for
the multi-objective optimization of ECM process [7]. Later on, grey relational analysis
was used to optimize multi-responses viz. material removal rate, overcut, cylindricity
error and surface roughness in electrochemical machining of EN31 steel and observed
that feed rate is the most influencing parameter in multi-response optimization [8].
Moreover, ANN had also been used by Abuzeid et al., for the prediction of ECM
process parameters with applied voltage, feed rate and electrolyte flow rate as inputs
and material removal rate and surface roughness as outputs [9]. Surface roughness and
passivation strength of electrolyte in electrochemical polishing of 316L stainless steel
were optimized using grey relational analysis and observed that electrolyte
composition plays a vital role on the surface roughness [10].
The present work investigates the effect of applied voltage, electrode feed
rate, electrolyte concentration and percentage of reinforcement on multi
performance characteristics (responses) in electrochemical machining of Al/B4C
composites using utility based Taguchi method.
2. The Utility-Based Taguchi Method
Taguchi method works only for optimization of single response problems. The
utility concept is used to convert multi response problems into single response. In
the present work, utility-based Taguchi optimization method follows the
optimization method developed by Taguchi. The procedure of the utility-based
Taguchi optimization method is shown in Fig. 1. In Fig. 1 steps 1, 2 and 7 are
procedure of Taguchi method and steps 3 to 6 are procedure of utility concept.
Fig. 1. Procedure of utility-based Taguchi method.

2.1. Experimental design and execution
A large number of experiments are required when classical parameter design is
used and the complexity increases with increasing the number of process
parameters [11]. Taguchi method solves this problem with a special design of
orthogonal arrays (OA). Entire process parameters space is studied with only a
small number of experiments by using Taguchi’s OA [12]. Therefore, in the
optimization problems selection of OA is the first step and the experiments are
conducted according to the selected OA.
2.2. Signal-to-noise ratio calculation
In the Taguchi method, the term ‘signal’ represents the desirable value (mean) for
the output characteristic and the term ‘noise’ represents the undesirable value
(standard deviation, SD) for the output characteristic. Therefore, the S/N ratio is
the ratio of the mean to the SD. There are several S/N ratios available, depending
on the type of characteristic; lower the better (LB), nominal the best (NB), or
higher the better (HB) [13].
The S/N ratio for higher-the-better criterion is given by Taguchi as:
= −10

! (1)
The S/N ratio for lower-the-better criterion is given by Taguchi as:
= −10

! (2)
where ‘y’ is the observed data and ‘n’ is the number of observations.
2.3. Optimum value of each response
The optimum value of performance characteristic is predicted at the optimal
levels of significant process parameters. The estimated mean of the performance
characteristic ‘i’ can be determined as [14]:
= #

+ %

+

+ '

+ ⋯ − (* − 1)
(3)
where #

,,is the average value of the performance characteristic ‘i’ at the
optimum level of process parameter ‘A’, ‘q’ is the number of process parameters
and ‘
’ is the overall mean of the performance characteristic ‘i’.
2.4. Preference number calculations
After obtaining the optimum values, a preference scale is constructed for each
response. In multi objective problems the range and unit of one performance
characteristic may differ from others. To reduce the variability in performance
characteristics, preference numbers are used. According to Gupata and Murthy
[15] the preference scale should be a logarithmic one. The minimum acceptable
quality level for each response is set out at 0 preference number and the best

available quality is assigned a preference number of 9. If a log scale is chosen the
preference number (Pi) is given by Eq. (4).
,,- = . log
23
23
/ (4)
where ‘xi’ is any value of performance characteristic or attribute ‘i’,
/
is
minimum or maximum acceptable value of performance characteristic or attribute
‘i’ and ‘Z’ is a constant.
At optimum value (
∗
) of attribute i, Pi = 9.
So, . =
4
567
83
∗
8
3
/
(5)
2.5. Assign weight to each response
Weight is nothing but relative importance of each performance characteristic. It
depends on the customer’s requirement or on the end use of product. The assigned
weight should satisfy the following condition:
9 = 1
:
; ,, (6)
where, Wi is the weight assigned to the performance characteristic ‘i’ and the sum
of the weights for all performance characteristics is equal to 1.
2.6. Utility value calculation
The utility of a product on a particular characteristic measures the usefulness of
that particular characteristic of the product. The overall utility,() of a product is
the sum of utilities of each of the quality characteristics and is given by
= 9-
:
; (7)
2.7. Optimal factor levels for utility value
From S/N ratio analysis of utility values, optimal level of performance
characteristics is determined. Since utility is a higher-the-better (HB) type of
quality characteristic, S/N ratios are calculated using Eq. (1). The optimal level of
the utility value is the level with the greatest S/N value.
3. Experimentation
A 25 mm diameter and 20 mm length LM6 Al/B4C composite samples were
fabricated with the help of stir casting technique with different percentage of
reinforcement of B4C (2.5, 5.0 and 7.5 wt%). LM6 is an Aluminium-Silicon alloy
containing 11-13% Si and small amount of Mg, Mn, Fe, Cu, Ni, Zn, Ti, etc.
Twenty seven experiments were conducted on METATECH ECM setup using
Taguchi L27 orthogonal array. METATECH ECM setup is shown in Fig. 2. The
setup consists of a control panel for changing the voltage, current, feed rate etc.; a
machining chamber in which tool and workpiece are fixed; a pump for circulating

the electrolyte and tool feed system (servo system) to maintain constant gap
between the tool and workpiece. A circular cross section tool coated with epoxy
on the lateral surface was used. NaCl solution with different concentration was
used as an electrolyte since NaCl solution has no passivation effect on the job
[16]. Electrolyte was pumped to the machining zone through the central hole of
the tool. An electrolyte flow rate of 30 liters per minute, and an inter electrode
gap of 0.5 mm was maintained constant for all the experiments.
Fig. 2. Electrochemical machining setup.
The responses like, material removal rate (MRR), surface roughness (SR) and
radial over cut (ROC) were measured by varying the predominant machining
parameters namely applied voltage; electrode feed rate, electrolyte concentration
and reinforcement content (i.e. wt% of B4C). The selected factors and their levels
are given in Table 1. The MRR was measured by weight loss using Eq. (8). The
diameter of the machined hole was measured using digital vernier caliper and the
ROC was obtained from this data by using Eq. (9). The surface roughness was
measured using a Talysurf tester.
MRR= (weight before machining – weight after machining) / machining time (8)
ROC = (Hole diameter – Tool diameter) / 2 (9)
Table 1. Machining parameters and their levels.
Machining
parameters
Symbol Unit
Level
1 2 3
Applied
Voltage
A Volts 12 16 20
Feed rate B mm/min 0.2 0.6 1.0
Electrolyte
concentration
C g/lit 10 20 30
Reinforcement
content
D wt% 2.5 5.0 7.5

4.Taguchi and Utility Analysis
4.1.Experimental design and execution
Experiments were conducted by varying the predominant machining parameters
namely, applied voltage; electrode feed rate, electrolyte concentration and
reinforcement content. Taguchi L27 orthogonal array and the measured responses
are given in Table 2. Effect of machining parameters on the responses is
discussed in the following subsections.
Table 2. Taguchi L27 orthogonal array and
measured response values and their S/N ratios.
Exp.
No.
Process parameters Responses S/N ratios
A B C D MRR SR ROC MRR SR ROC
1 1 1 1 1 0.268 4.948 0.96 -11.437 -13.889 0.354
2 1 1 2 2 0.335 5.002 0.94 -9.4991 -13.983 0.537
3 1 1 3 3 0.227 4.591 0.79 -12.879 -13.238 2.047
4 1 2 1 1 0.353 4.920 0.75 -9.0445 -13.839 2.499
5 1 2 2 2 0.448 4.498 0.65 -6.9744 -13.060 3.742
6 1 2 3 3 0.420 4.725 0.80 -7.5350 -13.488 1.938
7 1 3 1 1 0.689 4.555 0.67 -3.2356 -13.170 3.478
8 1 3 2 2 0.545 4.356 0.64 -5.2721 -12.782 3.876
9 1 3 3 3 0.703 4.232 0.65 -3.0609 -12.531 3.742
10 2 1 1 2 0.321 4.882 0.91 -9.8699 -13.772 0.819
11 2 1 2 3 0.329 4.823 0.94 -9.6561 -13.666 0.537
12 2 1 3 1 0.488 4.254 1.05 -6.2316 -12.576 -0.424
13 2 2 1 2 0.379 4.540 0.76 -8.4272 -13.141 2.384
14 2 2 2 3 0.302 4.431 0.69 -10.400 -12.930 3.223
15 2 2 3 1 0.583 3.998 0.99 -4.6866 -12.037 0.087
16 2 3 1 2 0.615 4.274 0.75 -4.2225 -12.617 2.499
17 2 3 2 3 0.619 4.346 0.70 -4.1662 -12.762 3.098
18 2 3 3 1 0.812 3.598 0.93 -1.8089 -11.1212 0.630
19 3 1 1 3 0.282 5.472 0.91 -10.995 -14.7629 0.819
20 3 1 2 1 0.599 4.797 1.10 -4.4515 -13.6194 -0.823
21 3 1 3 2 0.603 4.640 1.16 -4.3937 -13.3304 -1.289
22 3 2 1 3 0.526 5.214 0.85 -5.5803 -14.3434 1.412
23 3 2 2 1 0.688 4.897 1.03 -3.2482 -13.7986 -0.257
24 3 2 3 2 0.732 4.531 1.08 -2.7098 -13.1239 -0.668
25 3 3 1 3 0.688 5.002 0.64 -3.2482 -13.9829 3.876
26 3 3 2 1 0.887 4.389 0.99 -1.0415 -12.8473 0.087
27 3 3 3 2 0.944 3.989 1.00 -0.5006 -12.0173 0.060
4.1.1. Effect of machining parameters on material removal rate
Effect of machining parameters on MRR is shown in Fig. 3. MRR increases with
increase in voltage (A), feed rate (B) and electrolyte concentration (C) and
decreases with increase in percentage of reinforcement (D). With increase in
applied voltage, the machining current in the inter electrode gap (IEG) increases,
which leads to the enhancement of MRR. Increase in feed rate reduces the IEG

that leads to increase in the current density in the gap. This effect causes rapid
anodic dissolution which increases the MRR. With increasing the electrolyte
concentration the electrical conductivity of the electrolyte increases and also that
releases large number of ions in IEG, which results in higher machining current in
IEG and causes higher MRR. With increasing the percentage of reinforcement,
the electrical conductivity of the work piece decreases, because the reinforced
particles are poor conductors than the base material. Thus increase in the
percentage of reinforcement leads to lower metal removal rate.
Fig. 3. Effect of machining parameters on material removal rate.
4.1.2. Effect of machining parameters on surface roughness
Figure 4 shows the effect of machining parameters on surface roughness. At low
voltage, the current density in IEG is low which results in etching pits, leads to
rough surface. Moreover, high voltage causes excessive heating of the workpiece
which deteriorates the surface finish. Therefore, lower surface roughness values
are obtained at middle level of voltage. With increasing the feed rate the metal
dissolution is steady and uniformity in anodic dissolution results good surface
finish. At low electrolyte concentration depletions of ions occurs which leads to
poor surface finish. Thus increasing the electrolyte concentration improves the
surface finish. With increase in percentage of reinforcement the electrical
conductivity of the workpiece decreases due to non-conductive nature of the
reinforced particles and these reinforced particles are not participated in
electrolytic action, but the surrounding matrix material is removed by electrolysis
process which gives poor surface finish.
Fig. 4. Effect of machining parameters on surface roughness.

4.1.3. Effect of machining parameters on radial over cut
Effect of machining parameters on ROC is shown in Fig. 5. ROC increases with
increasing applied voltage (A) and electrolyte concentration (C) and decreases
with increasing feed rate (B) and reinforcement content (D). A higher
electrolyte concentration leads to the formation of greater volume of reaction
products, e.g. sludge’s and precipitations and also inhibits the initiation of gas
bubbles, e.g. O2, H2, etc. These phenomenal effects lead to the possibility of the
passage of stray current to the machining periphery thereby increase the ROC.
At higher feed rates the operational stability occurs which reduces the metal
removal in the lateral of the hole, causing the reduction in ROC. The increase in
voltage causes greater electrolyzing current to be available in the IEG, as well
as causing a greater stray current intensity, which leads to higher ROC.
Electrical conductivity of the work piece decreases with the increase in
percentage of reinforcement. This causes the reduction in metal removal rate in
radial direction, which leads to lower ROC.
Fig. 5. Effect of machining parameters on radial over cut.
4.2. Signal-to-noise ratio calculation
In the present work, the S/N ratio for MRR is calculated using higher-the-better
criterion by using Eq. (1) and the S/N ratio for ROC and SR is calculated using
lower-the-better criterion by using Eq. (2). The calculated S/N ratios of the
responses MRR, SR and ROC are given in Table 2.
4.3. Predicted optimum value of each response
Mean of each response for each level of factor is shown in Table 3. The average
of the S/N ratio of the responses MRR, SR and ROC for each level of each factor
is shown in Table 4. Regardless of the category of the performance
characteristics, a greater S/N value corresponds to a better performance.
Therefore, the optimal level of the machining parameters is the level with the
greatest S/N value. From Table 3 and Table 4 based on the analysis of means and
S/N ratios, the optimal machining parameter levels are A3 B3 C3 D1, A2 B3 C3 D1
and A1 B3 C1 D3 for MRR, SR and ROC, respectively. The optimal level of each
machining parameter is bolded in Tables 3 and 4.
0.7
0.75
0.8
0.85
0.9
0.95
1
A1 A2 A3 B1 B2 B3 C1 C2 C3 D1 D2 D3
Radial
over
cut
Machining parameters

Table 3. Average response value by factor levels (mean).
Material removal rate Surface roughness
Level A B C D A B C D
1 0.443 0.389 0.462 0.601 4.647 4.823 4.867 4.484
2 0.508 0.496 0.532 0.551 4.350 4.639 4.615 4.524
3 0.661 0.727 0.617 0.460 4.770 4.305 4.284 4.760
Delta 0.218 0.339 0.155 0.142 0.421 0.519 0.583 0.276
Cell mean = 0.5373 Cell mean = 4.5890
Radial over cut
Level A B C D
1 0.761 0.967 0.796 0.934
2 0.842 0.842 0.849 0.872
3 0.973 0.768 0.932 0.770
Delta 0.212 0.199 0.137 0.164
Cell mean = 0.8588
Table 4. Average response value by factor levels (S/N ratio).
Material removal rate Surface roughness
Level A B C D A B C D
1 -7.660 -8.824 -7.340 -5.021 -13.33 -13.65 -13.72 -12.99
2 -6.608 -6.512 -6.079 -5.763 -12.74 -13.31 -13.27 -13.09
3 -4.019 -2.951 -4.867 -7.502 -13.54 -12.65 -12.61 -13.52
Delta 3.641 5.873 2.473 2.482 0.80 1.00 1.12 0.53
Radial over cut
Level A B C D
1 2.468 0.347 2.063 0.685
2 1.584 1.615 1.606 1.370
3 0.350 2.440 0.733 2.348
Delta 2.118 2.094 1.330 1.663
The predicted optimal value of each response characteristics can be determined
[14] as:
MRR = A3 + B3 + C3 + D1 – 3 × mean = 0.661+ 0.727 + 0.617
+ 0.601 – 3,×,0.5373 = 0.9941
SR = A2 + B3 + C3 + D1 − 3 × mean = 4.350 + 4.305
+ 4.284 + 4.484 – 3,×,4.5890 = 3.6560
ROC = A1 + B3 + C1 + D3 – 3,×,mean = 0.761 + 0.768
+ 0.796 + 0.770 – 3,×,0.8588 = 0.5186.

4.4. Preference number calculation
After obtaining the optimum value, a preference scale is constructed for each
response. The preference scale should be a logarithmic one [15]. Zero is the
preference number of the minimum acceptable quality level and 9 is the
preference number of the best available quality level for each response. For the
log scale the preference number (P?) is calculated by using Eqs. (4) and (5).
• Material removal rate
@AA
∗
= predicted optimum value of MRR = 0.9941 g/min
@AA
/
= minimum acceptable value of MRR = 0.2 g/min
(since all MRR values greater than 0.2).
Using @AA
∗
, @AA
/
values and Eqs. (4) and (5), the preference scale for MRR is
given by
-@AA = 12.9236 ,
2GHH
.I
(10)
• Surface roughness
JA
∗
= predicted optimum value of SR = 3.6560 µm
JA
/
= maximum acceptable value of SR = 6.0 µm
(since all SR values are less than 6.0).
Using ,JA
∗
, ,JA
/
values and Eqs. (4) and (5), the preference scale for SR is
-JA = −41.8322 ,
2MH
N.
(11)
• Radial over cut
AOP
∗
= predicted optimum value of ROC = 0.5186 mm
AOP
/
= maximum acceptable value of ROC = 1.5mm
(since all ROC values are less than 1.5).
Using ,AOP
∗
, AOP
/
values and Eqs. (4) and (5), the preference scale for ROC
is given by
-AOP = −19.5118 ,
2HRS
.T
(12)
The preference number of the responses MRR, SR and ROC is calculated by
using Eqs. (10), (11) and (12) respectively and are given in Table 5.
4.5. Assign weight to each response
In the present study it is given equal importance to all responses and hence equal
weight has been assigned to them. The weights assigned to various responses are
WMRR = WSR = WROC = 1/3.

Table 5. Preference number and utility value of the responses.
Exp.
No.
Preference number
Utility Order
MRR SR ROC
1 1.642652 3.502257 3.783134 2.976014 26
2 2.895078 3.305060 3.961602 3.387247 25
3 0.710745 4.862740 5.435286 3.669590 22
4 3.188830 3.605356 5.875744 4.223310 21
5 4.526465 5.234538 7.088797 5.616600 11
6 4.164233 4.340067 5.328657 4.610986 16
7 6.942419 5.005761 6.831901 6.260027 4
8 5.626503 5.817327 7.220224 6.221351 5
9 7.055321 6.341995 7.088797 6.828704 2
10 2.655477 3.746219 4.236553 3.546083 24
11 2.793642 3.967114 3.961602 3.574119 23
12 5.006471 6.247797 3.023500 4.759256 15
13 3.587710 5.065687 5.763466 4.805621 13
14 2.313026 5.507188 6.582562 4.800926 14
15 6.004803 7.375371 3.522286 5.634153 10
16 6.304715 6.162583 5.875744 6.114347 7
17 6.341102 5.859082 6.460590 6.220258 6
18 7.864347 9.290517 4.052265 7.069043 1
19 1.928448 1.673504 4.236553 2.612835 27
20 6.156762 4.065317 2.629155 4.283745 19
21 6.194118 4.669865 2.178947 4.347643 18
22 5.427340 2.550937 4.814748 4.264342 20
23 6.934267 3.690485 3.186523 4.603758 17
24 7.282204 5.101737 2.784698 5.056213 12
25 6.934267 3.305060 7.220224 5.819850 9
26 8.360194 5.680213 3.522286 5.854231 8
27 8.709757 7.416314 3.437090 6.521054 3
4.6. Utility value calculation
The utility value of experiments has been calculated [17] by using Eq. (13) and is
given in Table 5.
=, ,,-@AA × 9@AA + -JA × 9JA + -AOP × 9AOP (13)
Figure 6 shows the utility value for MRR, SR and ROC, and it can be clearly
observed that experiment 18 has the highest utility value. Therefore, the
parameters of experiment 18 have the near optimal parameters setting for the best
multi-performance characteristics such as MRR, SR and ROC.
Fig. 6. Utility value for MRR, SR and ROC.

4.7. Optimal factor levels for utility value
The effects of each machining parameter at different level can be separated by the
orthogonal experimental design. For example, the mean of utility value for the
applied voltage at level 1, 2 and 3 can be calculated by taking the average of the
utility values for the experiments 1–9, 10–18 and 19–27, respectively. The mean
of the utility value for each level of other machining parameters can be computed
in the similar manner and is given in Table 6. In Table 6, the data in bold type is
the optimal parameter level, i.e. A2 B3 C3 D2. The difference between the
maximum and minimum value (Range) of the utility is also indicated in Table 6.
The factor which has the maximum range is the most influencing factor on the
multi-performance characteristics. The maximum range value in Table 6 is 2.444,
and the corresponding control factor, i.e. the feed rate, has the strongest effect on
multi-performance characteristics. The order of importance of the controllable
factor to the multi-performance characteristics in the electrochemical machining
process, in sequence, can be listed as: Feed rate, electrolyte concentration, applied
voltage and reinforcement content. Fig. 7 shows the utility value for different
machining parameters at different levels. Basically, the larger the utility value is,
the closer will be the product quality to the ideal value. Thus the larger utility
value is desired. From Fig. 7 it is inferred that A2 B3 C3 D2 is the optimum
parameter level for multi-response optimization.
Table 6. Utility response table for each level of machining parameters.
Machining
parameter
Utility value Max−Min
(Range)
Level 1 Level 2 Level 3
A 5.109 5.379 4.943 0.436
B 3.967 5.052 6.411 2.444
C 4.747 5.143 5.541 0.794
D 5.231 5.250 4.950 0.300
Mean value of utility = 5.1436
Fig. 7. Effect of ECM parameters on multiple performances.
5. Analysis of Variance
Analysis of variance (ANOVA) is a statistical tool used to investigate the
significance of the all parameters and their interactions by comparing the mean
square with an estimate of the experimental error at a specific confidence level.
3
4
5
6
7
A1 A2 A3 B1 B2 B3 C1 C2 C3 D1 D2 D3
Utility
value
Machining parameters level

The relative influence of the parameters is measured by total sum of square value
(U) and is given by:
U = (V − V)I

; (14)
where ‘n’ is the number of experiments in the orthogonal array, V is the mean
S/N ratio for the ith
experiment and V is the total mean S/N ratio of all
experiments. Further, the Fisher’s F-ratio, the ratio between the regression mean
square and the mean square error, is used to identify the most significant factor on
the performance characteristic. Table 7 shows the results of ANOVA for utility
value. In Table 7, the applied voltage, feed rate and electrolyte concentration are
the significant factors which influence the utility (multi-performance) since their
p values are below 0.05.
Table 7. Results of ANOVA for utility value.
Source DF SS MS F P
A 2 0.8714 0.4357 3.81 0.042
B 2 26.9903 13.4951 117.98 0.000
C 2 2.8367 1.4184 12.40 0.000
D 2 0.5090 0.2545 2.23 0.137
Error 18 2.0589 0.1144
Total 26 33.2663
6. Confirmation Experiment
Improvement of performance characteristic at optimum level is verified by
conducting the confirmation experiment. The estimated utility value

,
using the optimum level of machining parameters can be calculated as:

utility.pdf

More Related Content

Similar to utility.pdf (20)

More from SudhansuPanda15 (6)

Recently uploaded (20)

utility.pdf