Materials Today: Proceedings xxx (xxxx) xxx

Contents lists available at ScienceDirect

Materials Today: Proceedings

journal homepage: [Link]/locate/matpr

Grey relational analysis for multi-response optimization of process

parameters in green electrical discharge machining of Ti-6Al-4V alloy
Bhiksha Gugulothu a,⇑, G. Krishna Mohana Rao b, Minyahil Bezabih a
a
Department of Mechanical Engineering, Post Box no 144, Bule Hora University, Ethiopia
b
Department of Mechanical Engineering, Jawaharlal Nehru Technologcal University, Kukatpally, Hyderabad, India

a r t i c l e i n f o a b s t r a c t

Article history: This paper presents optimization of process parameters in electrical discharge machining (EDM) of Ti-
Received 25 April 2020 6Al-4 V alloy with multiple responses using grey relational analysis. The responses considered in this
Accepted 6 June 2020 study are metal removal rate (MRR), surface roughness (Ra) and recast layer thickness (RLT) whereas pro-
Available online xxxx
cess parameters are discharge current, pulse on time and pulse off time and graphite powder concentra-
tion(g/l). The approach uses weighted grey relational grade as performance index to determine optimal
Keywords: levels of parameters. Based on weighted grey relational grade, optimal levels of parameters are identified
Electrical discharge machining
and their percentage contributions are determined by analysis of variance. It is found that pulse on time
Grey relational analysis
Taguchi method and Multi-response
is the significant parameter for multi-response optimization in this study. The results of confirmation test
optimization indicate percentage improvements in MRR, Ra and RLT as 5.60%, 2.33%and 5.97% respectively. Hence grey
relational analysis can be effectively used for continuous improvement in EDM quality in modern man-
ufacturing industries.
Ó 2020 Elsevier Ltd. All rights reserved.
Selection and peer-review under responsibility of the scientific committee of the 2nd International Con-
ference on Manufacturing Material Science and Engineering.

1. Introduction (GRA) is applied to optimize parameter levels for EDM. GRA has
been proved useful in dealing with complex and multivariate prob-
Electrical discharge machining (EDM) is extensively used for lems such as machining with poor, insufficient and uncertain infor-
machining very hard, high strength, high temperature resistant mation. Further it can be adopted for solving complicated
materials used in aerospace, tool and die industries. Complicated interrelationship among the selected responses. GRA gives grey
shapes and sizes with very high precision and accuracy can be relational grade that provides the knowledge of parameters that
easily produced. Hence EDM is most demanding in industrial appli- affect multiple responses.
cations [1]. EDM is applicable to electrically conductive materials Ti-6Al-4 V alloy has high strength to weight ratio, good temper-
and its supreme advantage is that there is no physical contact ature resistance and prominent corrosion resistance which makes
between the electrode and the work piece. But EDM machining it successfully used in various applications such as aerospace, auto-
performance heavily depends on the many process parameters. motive, chemical plant, pressure vessels and medical fields [2].
The desired levels of these parameters are usually selected based However, machining this alloy in conventional machining would
on experience or handbook values of manufacturers. But such be a problem due to their low toughness and low thermal conduc-
selection procedure of parameter levels does not ensure whether tivity and high chemical reactivity with all the cutting tool materi-
a near optimal machining performance is achievable or not. Thus als to be machined. Therefore, an unconventional machining
different engineers may produce opposing results when addressing method such as Electrical Discharge Machining (EDM) has been
the same problem. Hence it is acute essential for manufacturers to used to machine this alloy [2–3].
search for a simple method to optimize parameters quickly for
more than one response. In this paper, grey relational analysis 1.1. Grey relational analysis

Jeyapaul et al. through literature review concluded that

⇑ Corresponding author. multiple-responses can be optimized by using pure engineering
E-mail address: bhikshamg@[Link] (B. Gugulothu). judgment rather than through Taguchi methods such as regression

[Link]
2214-7853/Ó 2020 Elsevier Ltd. All rights reserved.
Selection and peer-review under responsibility of the scientific committee of the 2nd International Conference on Manufacturing Material Science and Engineering.

Please cite this article as: B. Gugulothu, G. K. M. Rao and M. Bezabih, Grey relational analysis for multi-response optimization of process parameters in
green electrical discharge machining of Ti-6Al-4V alloy, Materials Today: Proceedings, [Link]
2 B. Gugulothu et al. / Materials Today: Proceedings xxx (xxxx) xxx

analysis, analysis of data development based on ranking approach, high MRR and low SR & RLT, appropriate process parameters com-
grey relational analysis, principle component analysis etc. Uncer- binations were determined. Signal-to-noise ratio i.e. S/N ratio was
tainty resulted in decision-making process due to this. Though obtained by converting MRR, SR, RLT experimental results. For bet-
approach of assigning weight to solve the multi-response problem ter machining performance, MRR should be higher value i.e.
was used, defining a weight for each response in real situation was ‘‘higher-the-better” and SR & RLT should be lower value i.e.
difficult [4]. ‘‘lower-the-better”. Quality characteristic deviating from the
Ramakrishnan et al. Investigated Wire EDM multi response desired value is measured by S/N ratio in Taguchi method. Equa-
optimization for machining tool Steel by applying L9 orthogonal tion (1) defines the S/N ratio g as
array through several cutting situations such as wire tension, wire
g ¼ 10 log10 ðM:S:DÞ ð1Þ
feed speed, delay time, pulse on time and ignition current intensi-
ties. ANOVA showed that dissimilar optimal parameters resulted in Where, M.S.D. is the mean-square deviation for the output
dissimilar weight assignment. Experiments are done by taking SR, characteristic.
MRR and wire wear ratio into account and found that pulse on time Equation (2) represents the mean-square deviation (M.S.D.) as
and current had great effect on the MRR comparatively with
1 Xm
1
others. [5]. M:S:D: ¼ ð2Þ
Gouri et al. pronounced that weights of realistic and experimen- m i¼1 MRR2
tal result analysis responses can only be obtained through PCA position m: the number of experiments and MRRi : MRR value
technique compared to other techniques like GRA, weighted signal for the ith experiment.
to noise ratio, wicker and multiple response signals to noise ratio SR and RLT to be considered to get optimal machining perfor-
and wicker. [6]. mance. Equation (3) expresses the M.S.D. and equation (4) defines
El-Taweel et al. experimentally studied the influence of EDM the mean square deviation for RLT.
process parameters on CK45 Steel by employing Al-Cu-Si-TiC com-
posite as electrode material. TWR and MRR are observed to be 1 Xm
M:S:D: ¼ SR2 ð3Þ
related. Since this is contradictory, TWR and MRR are given equal m i¼1 i
weightage centered on composite desirability function methodol-
ogy for optimization [7]. Where SR2i : The value of SR for the i th
test
Lin and Lee et al. declared 30% enhancement in EDM with mag- 1 Xm
netic force machining performance by utilizing GRA. No-load volt- M:S:D: ¼ RLT 2i ð4Þ
m i¼1
age, peak current and pulse duration are observed to be the
enhancement factors from ANOVA studies [8]. Where RLT2i : The value of RLT for the i th test
Kao et al. obtained 12%, 19% and 15% enhancement respectively Calculations for average experimental results of material
for MRR, SR and electrode wear ratio through experimentation of removal rate, surface roughness, recast layer thickness and corre-
Titanium alloy (Ti-6Al-4 V) EDM process parameter optimization sponding signal to noise ratios are done by using equations (2–
by applying Taguchi technique with grey relation theory. Open 4). Separation of each machining parameter effect at different
voltage, duty factor, discharge current and pulse duration are the levels was done after S/N ratio calculation. Average of the experi-
EDM process parameters. [9]. ments S/N ratios at each level for each process parameter is used
Anil Kumar et al. obtained 10% increase in Silicon abrasive to calculate the mean S/N ratio. MINITAB 17 software is used to
mixed EDM machining performance through GRA process opti- prepare mean of response tables and mean of S/N graphs for
mization. Peak current, duty factor, Silicon abrasive powder con- MRR, SR, RLT. Taguchi method allows evaluation of individual
centration in dielectric fluid and pulse on time is chosen for parameters effects independent of the other parameters for pro-
examining SR and MRR of EN-24 tool Steel by employing Copper cess parameters optimization. The design parameters prominently
electrode. [10]. affecting the performance measures were determined by using
Meena et al. used Grey Relational analysis to improve target analysis of variance (ANOVA). Equation (5) is used to calculate
output parameters such as MRR, TWR and over cut (OC) by opti- the first total sum of squared deviations SST from total mean S/N
mizing micro-EDM process parameters while machining of Ti- ratio gm in ANOVA.
6Al-4 V with Tungsten Carbide electrode. Voltage found to be the
most influential parameter [11]. X
n
SST ¼ ðgi  gm Þ2 ð5Þ
i¼1
1.2. Grey relational analysis and principal component analysis
Where n: the number of experiments in the orthogonal array
Tong et al. exercised signal to noise ratio to enhance Copper thin andgi : the mean S/N ratio for ith experiment.
film mechanical and chemical polishing of Taiwan based integrated Significance of process parameters on performance is found by
circuits manufacturer [12]. using ANOVA and F-test. High percentage contribution of process
Wang optimized dynamic multiple response problem and parameters on EDM has prominent effect on the response. MRR,
related Eigen vectors by applying grey relation model multiple- SR, RLT results by ANOVA are prepared. Process parameter combi-
criteria evaluation and PCA. Outcome is observed to be better com- nation and optimum results are also indicated by analysis of mean.
pared to Eigen vectors methods [13].
2.1.1. The confirmation test
2. Methodology and analysis Performance measures improvement were predicted and con-
firmed by design parameters optimization. Equation (6) is used
2.1. Taguchi method-The optimal condition for response measures and to calculate the estimated S/N ratio ( g
b ) by using the optimal level
identifies the compelling factors of the design parameters.
X
o
Evaluation of performance measures by process parameters g ¼ gm þ ðgi  gm Þ ð6Þ
under optimal conditions was done by Taguchi method. To achieve i¼1

Please cite this article as: B. Gugulothu, G. K. M. Rao and M. Bezabih, Grey relational analysis for multi-response optimization of process parameters in
green electrical discharge machining of Ti-6Al-4V alloy, Materials Today: Proceedings, [Link]
 B. Gugulothu et al. / Materials Today: Proceedings xxx (xxxx) xxx 3

Wheregm : the total mean S/N ratio, g

b i : the mean S/N ratio at the Equation (7) gives response normalization for larger-the-better
optimal level, and OA. The number of main design parameters that type
affects the quality characteristic.  
MRR, SR, RLT optimum conditions are validated by conducting X ij  Minj Xij
X ij ¼    Þ ð7Þ
experiments. Maxj X ij  Minj X ij

Equation (8) gives response normalization for smaller-the-

2.1.2. Steps in Taguchi method better type
The following steps are involved in Taguchi method !
 
Maxj X ij  X ij
 Prominent EDM parameters and performance characteristics X ij ¼     ð8Þ
Maxj X ij  Minj X ij
Identification.
 Deciding experimental region of process parameter effecting X ij : The normalized S/N ratio value after grey relational
the machining performance and parameter levels selection by generation
trial experimentation During jth experiments for j = 1, 2 nq
 Orthogonal array (Experimental matrix design) selection. Step 2: Regardless of responses, maximum normalized value
 Execution of Experimentation. Equation (9) gives the maximize value ‘‘R”, known as referenced
 Process parameters optimization through Taguchi method. value
 Using ANOVA, calculation of percentage contribution of each  
process parameters on EDM R ¼ Max X ij ð9Þ
 Optimized parameters levels verification by conducting confir-
mation tests Step 3: Equation (10) gives (the absolute difference (Dij)
 EDM machined surfaces microstructure evaluation at optimized between each normalized value and the reference value R regard-
parameter levels and various parametric combinations less of the responses
 
 
Dij ¼ X ij  R ð10Þ
3. Grey relational analysis
Step 4: Equation (11) gives the GRA coefficient (fij ) for each nor-
In control theory, ‘’White’’ system is understood as completely
malized value
known system, while a completely unknown one is called as
‘‘Black” system. ‘‘Grey” system is known to be in between ‘‘White”    !
mini minj Dij þ fMaxi Maxj Dij
and ‘‘Black” limits that getting an approach with model the system fij ¼   ð11Þ
Dij þ fMaxi Maxj Dij
for which the date collection is limited, no accurate and character-
ized by random uncertainty. Deng (1993) initiated uncertainty of f : the deferent coefficient of GRA lie in the level of value from 0
small sample based Grey system theory. to 1. & 0.5 is the mostly used value [4–8]
Probability distribution is not required in Grey theory whereas Step 5: Equation (12) gives the weighted grey relational grade
statistical methods need it for the values in the evaluation matrix (cj ) for each jth experiment
and sometimes it cannot be determined because of limited avail- !
1 Xm X
ability of data. q

GRA is a data normalization evaluation technique, which cj ¼ wi Dij ð12Þ

m i¼1 j¼1
enables reliable modelling and estimation of system behaviour
through discrete sequential data. wi : The weighing value for ith performance and m is the num-
P
ber of performance, 0 < cj  1 and m i¼1 wi ¼ 1
3.1. Steps in grey relational analysis (GRA) Step 6: Larger-the-better response is the basis to find each
parameter optimal level.
The problem of complicated interrelationships among multi Step 7: Using experiments and weighted grey relational grades,
responses is solved by using GRA. Final weighted grey relational perform ANOVA and draw inferences.
grade is taken as the alone objective function to be optimized Step 8: Execution of experiments and verification of results.
while solving the multi response problem. Types of steps made
in Grey relational analysis were as follows [13].
4. Experimental setup EDM details

3.1.1. Phase I The die sink EDM (Make: Formatics, Bangalore, India) with
Step 1: Study process parameters and their levels identification power supply unit and controller PSR-20 (Make: Electronica, Pune,
Step 2: Orthogonal array selection by taking time and cost into India) was used for conducting the experiments. The details of the
account experimental set up and the closed view of working tank are
Step 3: Orthogonal array parameters allocation and their inter- shown in Fig. 4.1 and Fig. 4.2 respectively.
actions to columns, if required.
Step 4: Condition of at least two experiments for various
4.1. Modified experimental set up for PMEDM
orthogonal parameter combinations and obtaining the response
values The experimental set up was modified for conducting the
experiments in the case of powder mixed dielectric fluid. The sche-
3.1.2. Phase II matic diagram of experimental set up for PMEDM is shown in
Step1: As the one data sequence of responses S/N ratios normal- Fig. 4.1.1. The modified experimental set up is shown in Fig. 4.1.2
ized values is different from other data sets, linear normalization is and a close view of the modified experimental set up is shown in
normally required. Grey relational generation is known as the lin- Fig. 4.1.3. A close view of machining tank of a modified experimen-
ear normalization of S/N ratio in the range between zero and one. tal set up is shown in Fig. 4.1.4.

Please cite this article as: B. Gugulothu, G. K. M. Rao and M. Bezabih, Grey relational analysis for multi-response optimization of process parameters in
green electrical discharge machining of Ti-6Al-4V alloy, Materials Today: Proceedings, [Link]
4 B. Gugulothu et al. / Materials Today: Proceedings xxx (xxxx) xxx

Fig. 4.1. Experimental set up.

Fig. 4.2. Closed view of working tank (dielectric fluid without powder)

In this modified experimental set up, powder is mixed into the so that circulation of powder mixed dielectric fluid is better. Small
dielectric fluid in a separate tank. The experimental set up helps in pumps are used for appropriate circulation of powder mixed
constant reuse of powder mixed dielectric fluid with its small dielectric fluid during machining and into the discharge gap. Two
quantity. It consists of a small transparent acrylic sheet container pumps are kept in a machining tank, out of which one is used to
known as machining tank. It is kept in the working tank of EDM circulate dielectric fluid from machining tank to bucket reservoir
machine. The workpiece is firmly clamped in the vice which is and another is used for flushing. Third pump is kept in the bucket
placed in the same tank and immersed in the electrode EDM oil. and used to supply the dielectric fluid from the bucket to the
Stirrer is used to avoid settling of powder particles due to gravity machining tank.

Please cite this article as: B. Gugulothu, G. K. M. Rao and M. Bezabih, Grey relational analysis for multi-response optimization of process parameters in
green electrical discharge machining of Ti-6Al-4V alloy, Materials Today: Proceedings, [Link]
 B. Gugulothu et al. / Materials Today: Proceedings xxx (xxxx) xxx 5

Fig. 4.1.1. Schematic diagram of experimental set up for PMEDM

Fig. 4.1.2. Modifications experimental set up for PMEDM

5. Selection of work piece material 7. Suitable orthogonal array

The work piece material selected for the experiments is Tita- Four process parameters and each parameter having three
nium alloy i.e. Ti-6Al-4 V. The chemical composition and proper- levels are used in present work. The degrees of freedom of each fac-
ties of Ti-6Al-4 V are given in Table 5.1 and Table 5.2, respectively. tor is two (i.e. number of levels-1) and the interaction between A
and B, A and C and A and D are considered. Degree of freedom
6. Selection of the electrode material and dielectric fluid for individual factor = 4  2 = 8, degree of freedom for interac-
tion = 3 (3–1)  (3–1) = 12, overall mean 1 therefore total degree
The experiments are conducted by using Copper electrode and of freedom = 21. Minimum number of experiments to be per-
dielectric fluid was drinking [Link] physical characteristics of formed is 21. The nearest OA suitable for this is 27. Therefore, L27
drinking water are shown in Table 6.1. (313) OA is selected Table [Link] [Link] [Link] 7.

Please cite this article as: B. Gugulothu, G. K. M. Rao and M. Bezabih, Grey relational analysis for multi-response optimization of process parameters in
green electrical discharge machining of Ti-6Al-4V alloy, Materials Today: Proceedings, [Link]
6 B. Gugulothu et al. / Materials Today: Proceedings xxx (xxxx) xxx

Fig. 4.1.3. Closed view of machining tank of modified experimental set up for PMEDM

Fig. 4.1.4. Closed view of modified experimental set up for PMEDM

Table 5.1
Chemical composition of Ti-6Al-4 V.

Element C Al V N O Fe H Ti
% Max. 0.08 5.5–6.5 3.5–4.5 0.05 0.13 0.25 0.01 Balance

Please cite this article as: B. Gugulothu, G. K. M. Rao and M. Bezabih, Grey relational analysis for multi-response optimization of process parameters in
green electrical discharge machining of Ti-6Al-4V alloy, Materials Today: Proceedings, [Link]
 B. Gugulothu et al. / Materials Today: Proceedings xxx (xxxx) xxx 7

Table 5.2 Optimal conditions of SR (A1B1C1D2) from Fig. 7.4.1

Properties of Ti-6Al-4 V. A1 = Discharge current = 10Amp
Property Quantity B1 = Pulse on time = 25 ms
Hardness (HRC) 36 C1 = Pulse of Time = 24 ms
Melting point (°C) 1649 D2 = Graphite powder concentration = 9.0 g/l
Density (g/cm3) 4.5 Fig. 7.4.1 shows means of weighted grey relational grades. Basi-
Ultimate tensile strength (MPa) 897–1000 cally larger the value of grey relational grade, better are corre-
Thermal conductivity (W/mK) 7.2
Specific heat (J/kg K) 560
sponding multiple responses. Accordingly, the level that gives
Mean coefficient of thermal expansion 0–100 °C/°C 8.6x10-6 largest grade is selected for optimization. From Table 7.3 and
Volume electrical resistivity (ohm-cm) 170 Fig. 7.4.1, the optimal levels for multiple responses are identified
Elastic Modulus (GPa) 114 as A1 B1C1D2 i.e. discharge current at level 1 (10 Amps), pulse on
time at level 1 (25 ls) and pulse off time at level 1 (24 ls) and
powder concentration at level 2 (9 g/l). The steep slope of weighted
Table 6.1 grey relational grade graph indicates more influence of process
Characteristics of Drinking water.
parameters on multiple responses.
Characteristic Value
Appearance Clear and Colorless 8. Analysis of variance (ANOVA)
Specific gravity at 30 °C 1.0004
Flash point (°C) ——————————
Pour point (°C) 3 ANOVA is carried out for weighted grey relational grades to
Viscosity at 38 °C (cst) 0.78 determine which parameter significantly affects all the four
Copper corrosion Not worse than 1 responses and its percentage contribution. Results of ANOVA
Dielectric strength (KV/min) ————————
shown in Table 8.1 indicate that percentage contribution of pow-
der concentration is 41.88% which is highest among all four param-
eters; pulse on time is 10.36%, which is second largest. Hence in
Table 6.2
this investigation, powder concentration is relatively significant
Drinking water properties.
process parameter affecting the multiple responses, followed by
Total dissolved solids 98 mg/l pulse on time (10.36%), pulse off time (6.43%) and discharge cur-
Total Suspended solids 10 mg/l rent (1.37%). It is well-known fact that the process parameters
Total solids 120 mg/l could have significant effects on selected response individually.
Dissolved O2 7.0 mg/l
Dissolved CO2 9 mg/l
Dissolved N2 0.01 mg/l 8.1. Confirmation experiments of multi optimization of EDM of Ti-6Al-
Chlorides 20.7 mg/l 4 V using grey relational analysis
Sulphate 32.0 mg/l
Total alkalinity 34.0 mg/l
pH 6.55 The confirmation experiments are carried out by setting param-
Total hardness 55.1 mg/l eters at optimal levels. Discharge current, pulse on time, pulse off
Resistivity at 25 °C 4000 O /cm3 time and powder concentration are set at A1B1C1D2 i.e. 10 Amps,
25 ls, 24 ls and 9 g/lt respectively. Table 8.2 presents the results
of confirmation experiments obtained by the proposed method
and those of initial parameters A1 B1 C1 D1. From Table 8.2, it can
be observed that MRR is increased from 2.5 mm3/min to
Table 7
Working range of the process Parameters and their levels. 2.64 mm3/min, SR is decreased from 2.14 mm to 2.09 mm and RLT
is decreased from 8.04 mm to 7.56 mm. The corresponding %
Parameters Units Levels
enhancement in material removal rate, surface roughness and
Peak current (A) Amps 10 15 20
recast layer thickness are 5.6, 2.3 and 5.9% respectively.
Pulse on time (B) ms 25 45 65
It is understood clearly that MRR and RLT are combining
Pulse off time (C) ms 24 36 48
Graphite powder conc.(D) g/l 4.5 9 13.5 improved whereas SR is slightly improved. The predicted grey rela-
tional grade is calculated by the equation (13).

X
q
cpre ¼ cm þ ðci  cm Þ ð13Þ
The working range of the process parameters and their levels i¼1
shown in Table 7 and L27 experimental was shown in table 7.1
The response is better if the S/N ratio is larger. All values of S/N = A1 + B1 + C1 + D2 – 3cm
ratios in Table 7.1 are substituted in equations (8) and (9) to get = 0.571796 + 0.53862 + 0.612636 + 0.557989 – 3 (0.539232)
normalized values as shown in Table 7.2. Larger normalized values = 0.663345
indicate better performance and maximum normalized value equal Overall average grade is improved from 0.6633 to 0.8750.
to 1 indicate the best performance. The results given in Table 7.2
are substituted in equation (12) to calculate grey relational coeffi- 9. Conclusions
cients of material removal rate, surface roughness, and recast layer
as shown below the Table 7.3. This study utilizes, based GRA to improve multiple responses
Now, the response table of Taguchi’s grey relational analysis is MRR, SR and RLT in EDM of Titanium alloy and determines optimal
employed to calculate the average weighted grey relational grade levels of parameters based on weighted grey relational grade. The
for each parameter level. It is done by taking average of grey rela- proposed procedure is relatively simple and does not involve much
tional grades with same levels of parameter in each column of mathematical processing, thus making it quite easy for engineers
orthogonal array. Average grey relational grade of each level of cor- to use without much statistical background. The conclusions are
responding parameter is shown in Table 7.4. summarized as follows

Please cite this article as: B. Gugulothu, G. K. M. Rao and M. Bezabih, Grey relational analysis for multi-response optimization of process parameters in
green electrical discharge machining of Ti-6Al-4V alloy, Materials Today: Proceedings, [Link]
8 B. Gugulothu et al. / Materials Today: Proceedings xxx (xxxx) xxx

Table 7.1
S/N ratio of MRR, Ra and RLT of grey relational analysis.

Ex. A(Discharge BPulse on CPulse off DGraphite Powder *Avg S/N ratio *Avg Ra S/N ratio *Avg RLT S/N ratio
No current)Amp time (ms) time (ms) Concentration (g/l) MRR MRR (microns)
1 10 25 24 4.5 2.5 7.958 2.14 6.608275 8.04 18.10512
2 10 25 36 9 2.2121 6.896 2.126 6.551265 8.9 18.9878
3 10 25 48 13.5 2.2776 7.149 3.641 11.22441 10.109 20.09416
4 10 45 24 9 3.0578 9.708 2.001 6.024942 8.25 18.32908
5 10 45 36 13.5 3.6811 11.319 4.211 12.4877 8.9054 18.99307
6 10 45 48 4.5 2.4108 7.643224 4.67 13.38634 11.2533 21.0256
7 10 65 24 13.5 3.983 12.004 3.6505 11.24705 8.4 18.48559
8 10 65 36 4.5 4.0632 12.177 4.2885 12.64611 10.3795 20.32353
9 10 65 48 9 1.9492 5.797 5.27 14.43621 11.8675 21.48718
10 15 25 24 4.5 3.9447 11.920 2.308 7.264716 8.7 18.79039
11 15 25 36 9 3.3277 10.442 2.787 8.902739 9.03 19.11376
12 15 25 48 13.5 2.2178 6.918 3.502 10.88632 9.78 19.80678
13 15 45 24 9 3.5575 11.022 2.7035 8.638527 8.6945 18.78489
14 15 45 36 13.5 4.2765 12.62177 3.6575 11.26369 11.19 20.9766
15 15 45 48 4.5 4.0892 12.232 4.33 12.72976 11.44 21.16852
16 15 65 24 13.5 5.11178 14.171 4.25 12.56778 10.2953 20.25278
17 15 65 36 4.5 6.7889 16.635 4.495 13.05459 11.804 21.44058
18 15 65 48 9 3.2235 10.166 5.56 14.9015 11.5157 21.22581
19 20 25 24 4.5 5.0293 14.03015 3.505 10.89376 9.88 19.89514
20 20 25 36 9 4.1478 12.356 3.2215 10.16116 9.5774 19.62495
21 20 25 48 13.5 3.4018 10.634 3.6545 11.25656 11 20.82785
22 20 45 24 9 6.7215 16.549 4.27 12.60856 10.9 20.74853
23 20 45 36 13.5 7.746 17.781 4.01 12.06289 13.24 –22.43776
24 20 45 48 4.5 8.55 18.639 4.5895 13.23531 10.5366 20.45401
25 20 65 24 13.5 7.606 17.623 5.46 14.74385 13.4492 –22.57393
26 20 65 36 4.5 9.4142 19.475 5.533 14.85921 12.615 –22.01775
27 20 65 48 9 8.0598 18.126 6.41 16.13716 10.87605 20.72942

Table 7.2
Normalized values of the S/N ratios of material removal rate, surface roughness and recast layer thickness.

Exp no Normalization Dij

MRR SR RLT MRR SR RLT
1 0.158034 0.942314 1.00 0.841966 0.057686 0.000
2 0.080342 0.947952 0.80 0.919658 0.052048 0.198
3 0.098872 0.485823 0.55 0.901128 0.514177 0.445
4 0.285926 1.000000 0.95 0.714074 0.000000 0.050
5 0.403729 0.360896 0.80 0.596271 0.639104 0.199
6 0.134963 0.272029 0.35 0.865037 0.727971 0.654
7 0.453782 0.483584 0.91 0.546218 0.516416 0.085
8 0.466441 0.345231 0.50 0.533559 0.654769 0.496
9 0.000000 0.168207 0.24 1.000000 0.831793 0.757
10 0.447647 0.877398 0.85 0.552353 0.122602 0.153
11 0.339638 0.715414 0.77 0.660362 0.284586 0.226
12 0.081977 0.519257 0.62 0.918023 0.480743 0.381
13 0.382042 0.741542 0.85 0.617958 0.258458 0.152
14 0.498931 0.481939 0.36 0.501069 0.518061 0.643
15 0.470492 0.336959 0.31 0.529508 0.663041 0.686
16 0.612223 0.352977 0.52 0.387777 0.647023 0.481
17 0.792399 0.304836 0.25 0.207601 0.695164 0.746
18 0.319436 0.122195 0.30 0.680564 0.877805 0.698
19 0.601893 0.518521 0.60 0.398107 0.481479 0.401
20 0.479527 0.590968 0.66 0.520473 0.409032 0.340
21 0.353623 0.482644 0.39 0.646377 0.517356 0.609
22 0.786063 0.348944 0.41 0.213937 0.651056 0.592
23 0.876148 0.402906 0.03 0.123852 0.597094 0.970
24 0.938857 0.286965 0.47 0.061143 0.713035 0.526
25 0.864566 0.137785 0.00 0.135434 0.862215 1.000
26 1.000000 0.126377 0.12 0.000000 0.873623 0.876
27 0.901365 0.000000 0.41 0.098635 1.000000 0.587

1. Taguchi orthogonal designed experiments for EDM are success- 3. There is an improvement in weighted grey relational grade
fully conducted. from 0.6633 to 0.8750 that confirms the improvement in the
2. The parameters are optimized for MRR, SR and RLT. The optimal performance of EDM using optimal levels of parameters. The
levels of discharge current, pulse on time and pulse off time are percentage improvement in MRR, SR and RLT are 5.6%, 2.3%
A1B1C1D2 i.e. 10 Amps, 24 ls and 25 ls, 9 g/l respectively. and 5.9% respectively.

Please cite this article as: B. Gugulothu, G. K. M. Rao and M. Bezabih, Grey relational analysis for multi-response optimization of process parameters in
green electrical discharge machining of Ti-6Al-4V alloy, Materials Today: Proceedings, [Link]
 B. Gugulothu et al. / Materials Today: Proceedings xxx (xxxx) xxx 9

Table 7.3
Grey relational coefficients and grades for performance.

Grey relational coefficient (GRC)

[Link] MRR SR RLT Grade (GRG) Rank
1 0.372588 0.896562 1.000 0.756365 2
2 0.352198 0.905718 0.717 0.658241 4
3 0.356855 0.493011 0.529 0.459636 22
4 0.411837 1.000000 0.909 0.773565 1
5 0.456092 0.438942 0.716 0.536877 13
6 0.366290 0.407176 0.433 0.402307 25
7 0.477912 0.491925 0.854 0.608100 6
8 0.483765 0.432987 0.502 0.472848 21
9 0.333333 0.375434 0.398 0.368867 27
10 0.475126 0.803082 0.765 0.681159 3
11 0.430900 0.637279 0.689 0.585715 8
12 0.352603 0.509818 0.568 0.476696 19
13 0.447244 0.659232 0.767 0.624396 5
14 0.499466 0.491130 0.438 0.476070 20
15 0.485669 0.429907 0.422 0.445779 24
16 0.563204 0.435911 0.510 0.503003 17
17 0.706613 0.418353 0.401 0.508709 16
18 0.423526 0.362896 0.417 0.401224 26
19 0.556727 0.509435 0.555 0.540456 12
20 0.489969 0.550036 0.595 0.545055 10
21 0.436157 0.491470 0.451 0.459457 23
22 0.700342 0.434384 0.458 0.530933 15
23 0.801472 0.455749 0.340 0.532490 14
24 0.891038 0.412189 0.488 0.596912 7
25 0.786864 0.367049 0.333 0.495750 18
26 1.000000 0.364001 0.363 0.575832 9
27 0.835234 0.333333 0.460 0.542814 11

Table 7.4
Average grey relational grade for each level of corresponding factor.

Parameters Level 1 Level 2 Level 3 Max-Min (D) Rank

A(Discharge Current) 0.5596* 0.5225 0.5355 0.0371 4
B (Pulse on time) 0.5736* 0.5466 0.4975 0.0762 2
C (pulse off time) 0.6126* 0.5435 0.4615 0.1511 1
D(Graphite powder concentration) 0.5534 0.5590* 0.5053 0.0536 3

Fig. 7.4.1. Means of mean weighted grey relational grades

Please cite this article as: B. Gugulothu, G. K. M. Rao and M. Bezabih, Grey relational analysis for multi-response optimization of process parameters in
green electrical discharge machining of Ti-6Al-4V alloy, Materials Today: Proceedings, [Link]
10 B. Gugulothu et al. / Materials Today: Proceedings xxx (xxxx) xxx

Table 8.1
ANOVA of weighted grey relational grades.

Source Degreesof freedom Sum square Mean sumof square Percentage of Contribution (%)
A (Discharge current) 2 0.8903582 0.4451791 1.375202232
B (Pulse on time) 2 6.7121976 3.356098802 10.36732084
C (Pulse off time) 4 4.1659226 1.041480651 6.434473297
D (Graphite Powder concentrations) 2 27.120625 13.5603125 41.88914534
A*B (interaction) 4 9.8676563 2.466914087 15.24108279
A*C (interaction) 2 2.8302547 1.415127363 4.371468266
A*D (interaction) 4 0.9258313 0.231457821 1.429992161
Error 6 12.230955 2.038492581 18.89131507
Total 26 100 100

Table 8.2
Weighted grey relational grades of initial, predicted and confirmation experiment.

response Initial data Optimal parameters

A 1 B 1 C 1 D1 PredictionA1 B1 C1 D2 ExperimentA1 B1 C1 D2 %improvement
MRR 2.5 – 2.64 5.6
SR 2.14 – 2.09 2.3
RLT 8.04 – 7.56 5.9
Overall average grade (cm) 0.53922 0.663345 0.8750

4. Among four parameters, pulse of time has the significant effect [4] R. Jeyapaul, P. Shahabudeen, K. Krishnaiah, Quality management research by
considering multi-response problems in the Taguchi method – A review, Int. J.
on selected multiple responses, followed by pulse on time, pow-
Adv. Manuf. Technol. 26 (2005) 1331–1337.
der concentration and discharge current. [5] R. Ramakrishna, L. Krishnamurthy, Multi response optimization of wire EDM
5. The proposed approach simplifies the multi-response optimiza- operations using robust design of experiments’, International Journal of
tion of EDM process parameters to a great extent. Thus solu- Advanced Manufacturing Technology 29 (2006) 105–111.
[6] S.K. Gouri, S. Pal, Comparison of performances of five prospective approaches
tions obtained from this method can be useful to for the multi-response optimization, Int. J. Adv. Manuf. Technol. 48 (2008)
manufacturers to find optimal solutions of EDM parameters. 1205–1220.
This approach can be applied effectively to other manufacturing [7] El-Tawel T.A, ‘‘Multi optimization of EDM with Al-Cu-Si-TiC p/m electrode’’
international journal of advanced manufacturing technology, volume 44,
processes also. (2009) pp.100-113
[8] Y.C. Lin, H.S. Lee, Optimization of machining parameters using magnetic force
assisted EDM based on grey relational analysis, Int. J. Adv. Manuf. Technol. 42
(2009) 1052–1064.
[9] Y.C. Kao, S.S. Tsao, Y. Hsu, Optimization of the EDM parameters on machining
Declaration of Competing Interest Ti-6Al-4V with multiple quality characteristics, Int. J. Adv. Manuf. Technol.
(2010) 395–402.
The authors declare that they have no known competing finan- [10] Anil kumar, maheswari S, Sharma and Naveen beri ‘‘A study of multi objective
parametric optimization of Silicon Abrasive mixed Electric discharge
cial interests or personal relationships that could have appeared machining of tool Steel Journal of material and manufacturing process, vol.
to influence the work reported in this paper. 25 No.10, (2005) pp.1041-1047.
[11] Meena ‘‘Optimized micro-EDM process parameters using grey relational
analysis when machining of Ti-6Al-4V was performed with Tungsten
Carbide electrode” International Journal of Advanced Manufacturing
References
Technology, (2012), pp.401-409.
[12] Lee-Ing Tong, Chung Ho Wang, Hung Chen Cheng, Optimization of multiple
[1] Singh M. K. Chapter1 Introduction to Unconventional Manufacturing Process, responses using principal component analysis and technique for order
1st Edition, New Age International (P) Ltd., (2007) New Delhi, India: pp.1-2 preference by similarity to ideal solution, Int. J. Adv. Manuf. Technol. 27
[2] E.O. Ezugwu, Z.M. Wang, Titanium alloys and their machinability- a review, J. (2005) 407–414.
Mater. Process. Technol. 68 (1997) 262–274, in English. [13] Chung-Ho-Wang, Dynamic multi-response optimization using principal
[3] ASM Handbook Introduction to Titanium and Titanium alloys, Properties and component analysis multiple criteria evaluation of the grey relational model,
selection: Nonferrous alloys and Special Purpose Materials. ASM International, Int. J. Adv. Manuf. Technol. 32 (2007) 617–624.
Ohio, USA, Volume 2, (1993)

Please cite this article as: B. Gugulothu, G. K. M. Rao and M. Bezabih, Grey relational analysis for multi-response optimization of process parameters in
green electrical discharge machining of Ti-6Al-4V alloy, Materials Today: Proceedings, [Link]