Available online at www.sciencedirect.

com

ScienceDirect
Materials Today: Proceedings 11 (2019) 1015–1023 www.materialstoday.com/proceedings

I2CN_2018

MULTI-OBJECTIVE OPTIMIZATION OF MILLING PROCESS

PARAMETERS IN GLASS FIBRE REINFORCED POLYMER VIA GREY
RELATIONAL ANALYSIS AND DESIRABILITY FUNCTION

Mohammed Yaser E.Ka, K Shunmugeshb *

a,b
Department of Mechanical Engineering, Viswajyothi College of Engineering and Technology, Vazhakulam
Ernakulam , 686670,India
*[email protected]

Abstract

Finding the greatest or least value of a function for some constrain, that is correct regardless of the solution, this process is
optimization. Inorder to lift up or improve the quality of machining products, to reduce the machining costs and for the raise of
production rate, there is an important issue, ie the selection of optimum cutting parameters. In this project there is a performance
of different factors of simultaneous optimization as the parameters in milling operation by combining the method of GRA and
DFA. The simultaneous optimization is also known as multi-objective optimization. The epoxy glass fibre is the material used for
milling operation and the parameters are delamination factor, machining force and surface roughness. On which every parameters
have to be in minimized magnitude. The Grey relational analysis (GRA) is based on the grey system theory. GRA is used to
study the relation among various attributes in a system and for solving the complicated interrelationships among the multiple
responses. DFA is one of the most used methods in the industry for optimizing multiple response features. The analysis of the
desirability function is used to convert multiple response characteristics into unique response characteristics.

© 2018 Elsevier Ltd. All rights reserved.

Selection and/or Peer-review under responsibility of International Multi- Conference on Computing, Communication, Electrical &
Nanotechnology: Materials Science.

Keywords:Grey Relational Analysis GRA, Desirability Function Analysis DFA, Multi criteria decision making (MCDM).

1. Introduction

Milling is the machining operation frequently used in the manufacture of pieces of reinforced fiberplastic,
because components made of compounds are commonly produced by network-shaped components that oftenrequires

2214-7853© 2018 Elsevier Ltd. All rights reserved.

Selection and/or Peer-review under responsibility of International Multi- Conference on Computing, Communication, Electrical &
Nanotechnology: Materials Science.
1016 Mohammed Yaser E.K and K Shunmugesh / Materials Today: Proceedings 11 (2019) 1015–1023

the removal of excess material to control tolerances and milling is used as a corrective operation forproduce well
defined and high quality surfaces[1].Fiberglass composite materials have become an economic alternative to exotic
materials[2]. Milling of GFRP composites is a rather complex task due to its heterogeneity and the number of
problems, such as the delamination of the surface, which appears during the working process, associated to the
characteristics of the material and to the cutting parameters [3]. Optimization techniques help us to select the
optimum combination of machining parameters. It is the process of finding the greatest or least value of a function
for some constraint that will be right regardless of solution. Inshort optimization discovers the very apt value for the
function within the given domain. Therefore, it seems that the optimization of the machining process parameters for
all types of FRP compounds be an emerging area in research. Also delamination, under-surface damage and fiber
removal will take place after processing. There are many research studies exposed to improve the workability of
FRP materials over the last five decades [4] -[5].

The grey systems theory proposed by Deng in 1982 proved to be useful in dealing with problems with
insufficient, poor and uncertain information. The grey relational analysis based on this theory can be adopted more
effectively to solve the complicated interrelation between the designated performance characteristics [6]. GRA
provides an efficient solution to uncertainty, problem of multiple and discrete data. The relationship between the
machining parameters and performance can be found with Gray's relational analysis. In addition, relational grey
grade will use the discrete measurement method. The degree of grey relational information for the decision-making
process has been defined to solve the difficult situation of the decision-makers [7].

Desirability function analysis (DFA) is one of the most commonly used methods in the sector for the
optimization of multiple response problems [8]. The advantage of this method is that many factors can be analyzed
using less data. It does not imply a complicated theory or mathematical calculation such as traditional approaches
and therefore can be used by engineers without a solid statistical basis. The focus on the analysis of the desirability
function and its applications [9]-[12].

Phadke et al. (1989), according to him it is difficult to optimize simultaneously responses in complex process by
single-response method and engineering judgment is primarily used to resolve such complicated problems. An
engineer’s judgment often increases the degree of uncertainty during decision making process, making it most
critical to the quality of finished product[13].

Lin et al. (2002) have explored the optimization of the parameters for electrical discharge machining process. The
findings are verified by GRA. The study also analyses the effect of data normalization and data integrity in GRA to
predict the rank of the parameter effect in the case of insufficient data derived from the Taguchi method[14].

Mohammed et al. (2007) build a multiple regression model for surface roughness to study the effects of spindle
speed, cutting feed rate, depth of cut and their two way interactions. The cutting parameters were selected as four
levels of cutting speed, seven levels of feed rate and three levels of depth of cut.The results showed the cutting feed
as the most dominant factor and interactions cutting feed-depth of cut ,and cutting feed-spindle speed the most
significant[15].

Kini et al. (2010) studied the process parameters effect for surface roughness and MRR in turning of GFRP
composites in dry cutting conditions. A second order model to predict the cutting speed, feed rate, tool nose radius
and depth of cut is developed with 95% confidence level for surface roughness and MRR[16].

Dubey et al. (2014) had conducted study for the multiple response optimization of end milling parameter using
grey based taguchi method. The feed rate was identified as the most influential process parameter on surface
roughness[17].

Hsu et al. (2004) presents an integrated optimization approach based on neural networks, exponential desirability
functions and Tabu search to optimize a fused biconic taper process for a Taiwanese fiber-optic passive component
 Mohammed Yaser E.K and K Shunmugesh / Materials Today: Proceedings 11 (2019) 1015–1023 1017

manufacturer. The confirmation results demonstrated the practicability and effectiveness of the proposed
approach[19].

Tosun et al. (2006) used GRA for optimising the drilling process parameters for the work piece surface roughness
and the burr height. Various drilling parameters, such as feed rate, cutting speed, drill and point angles of drill were
considered. An orthogonal array was used for the experimental design. Optimal machining parameters were
determined by the grey relational grade obtained from the grey relational analysis for multi-performance
characteristics (the surface roughness and the burr height). Experimental results have shown that the surface
roughness and the burr height in the drilling process improved effectively.

1. Experimental Details

The material used for experiment: - Fiber reinforced glass polymer.

Fiber glass is made by using individual glass fibers into different forms. According to geometry mainly there are
two types of glass fibers. Details about the fiber reinforced glass polymer are given in the Table 1.

Table 1 Mechanical properties of glass fiber

Properties Values

Density 1.8

Elongation 1.7%

Tensile strength 1500 MPa

Tensile modulus 140 a

In this study, the identified machining parameters and their range of values are mentioned in the Table 2.

Table 2 Machining parameter levels

Machine: SMG 300 CNC milling machine

Factors Level 1 Level 2 Level 3

Spindle speed(RPM) 2000 4000 6000

Feed rate(mm/rev) .1 .2 .3

Depth of cut(mm) .1 .15 .2

The below graph shows the data obtained during the total 27 experiments by interchanging depth of cut, feed and
speed with each other.
1018 Mohammed Yaser E.K and K Shunmugesh / Materials Today: Proceedings 11 (2019) 1015–1023

Figure 1 plots the main effects of surface roughness with spindle speed, feed rate and depth of cut.

Figure 1 Surface Roughness main effects

Figure 2 plots the main effects of machining force with spindle speed, feed rate and depth of cut.

Figure 2 Machining Force main effects

Figure 3 plots the main effects of delamination factorwith spindle speed, feed rate and depth of cut.

Figure 3 Delamination Factor main effects

4. Methodology

Main objective in design of experiment: - Determine the best RPM, depth of cut, cutting speed for milling the
glass FRPs.
 Mohammed Yaser E.K and K Shunmugesh / Materials Today: Proceedings 11 (2019) 1015–1023 1019

Design of experiment: - To minimize the number of experiments in order to attain optimum conditions. ie
reached by multi criteria decision making (MCDM).

4.1 Grey Relational Analysis

The multiple optimization of the performances was carried out using grey relational analysis, in which the
optimization processes are carried out step by step to obtain the grey relational degree (GRG).

Step 1

Preprocess raw data (Grey relational generation)

Preprocess raw data using the smaller-the-better criterion from Eq. (1). The smaller-the-better model aims to obtain
the minimum quality characteristics.

Pre- processed data= y i

max xi  xi
yi  (1)
max xi  min xi

Step 2

Deviation Sequence of pre- processed data

Deviationsequence  1  y i (2)

Step 3
Grey relational coefficient of deviation sequence

Calculate grey relational coefficient using Eq. (3) to express the relational degree between the desired response and
the experimental value.

(min xi  .5  max xi )
GRC  (3)
( xi  .5  max xi )

Step 6

Calculate grey relational grade with Eq. (4) to evaluate the multiple response as a single index.

Grey Relational Grade= Average of the GRC (4)

And rank them accordingly

4.2 Desirability Function Analysis

DFA is one of the most used methods in the industry for optimizing multiple response features.
1020 Mohammed Yaser E.K and K Shunmugesh / Materials Today: Proceedings 11 (2019) 1015–1023

Step 1
Calculate individual desirability function of each response by using Eq. (5).

w
 yi  max yi 
yi    (5)
 min yi  max yi 
Step 2
Calculate the composite desirability function using the Eq. (6) given below.

yi  w w w
x1  x 2  ........ x n
w (6)

Step 3
Rank the composite desirability function in descending order to find out the optimum parameters for the
machining of GFRP.

2. Results and Discussion

The workability in this work was assessed by the surface roughness (Ra) of the machined work surface of the
piece, machining force and delamination factor. The results obtained through the experiments are shown in graph 1,
graph 2 & graph 3.

5.1 Grey Relational Analysis (GRA)

Based on experimental data, GRA is used to convert the optimization of multiple responses in the optimization of
a single objectiveof the GRG.

5.1.1 Result for pre- processed row data and deviation sequence

After the calculation of pre- processed value of each response from the raw data the next step is to calculate the
deviation sequence of the result obtained for the pre- processed data.

5.1.2 Grey relational coefficient , weighted grey relational grade and rank

Result obtained for the grey relational coefficient is given in Table 3. It is calculated from the data which is
obtained after the calculation of deviation sequence. Weighted grey relational grade is the average of the grey
relational coefficient of experiment and it is ranked in descending order of the value obtained for the grey relational
grade and shown in the Table 3.
 Mohammed Yaser E.K and K Shunmugesh / Materials Today: Proceedings 11 (2019) 1015–1023 1021

Table 3 Grey relational coefficient, weighted grey relational grade and rank

Sl. grey relational coefficient weighted grey relational

rank
No. Surface Roughness Force Delamination Factor grade
1 0.451505017 0.45287906 0.492307692 0.465563923 15
2 0.460750853 0.435524047 0.444444444 0.446906448 16
3 0.436893204 0.419450083 0.432432432 0.429591906 21
4 0.373961219 0.41080788 0.372093023 0.385620707 23
5 0.354330709 0.390616944 0.390243902 0.378397185 24
6 0.448504983 0.399824099 0.450704225 0.433011102 20
7 0.333333333 0.406110416 0.333333333 0.357592361 26
8 0.358090186 0.365492845 0.372093023 0.365225351 25
9 0.347043702 0.333333333 0.380952381 0.353776472 27
10 0.645933014 0.628074054 0.542372881 0.605459983 7
11 0.627906977 0.468563183 0.615384615 0.570618258 9
12 0.605381166 0.433199924 0.524590164 0.521057085 12
13 0.467128028 0.556698506 0.47761194 0.500479491 14
14 0.445544554 0.445948597 0.432432432 0.441308528 18
15 0.517241379 0.467791727 0.551724138 0.512252415 13
16 0.410334347 0.45891379 0.385542169 0.418263435 22
17 0.480427046 0.467599259 0.376470588 0.441498965 17
18 0.467128028 0.426054358 0.41025641 0.434479599 19
19 0.918367347 1 0.888888889 0.935752079 1
20 1 0.673481481 0.761904762 0.811795414 3
21 0.870967742 0.601800371 1 0.824256038 2
22 0.72972973 0.740632128 0.615384615 0.695248824 5
23 0.671641791 0.640823231 0.711111111 0.674525378 6
24 0.906040268 0.538753259 0.666666667 0.703820065 4
25 0.477031802 0.673082618 0.542372881 0.564162434 10
26 0.589519651 0.514369767 0.62745098 0.577113466 8
27 0.6 0.468756445 0.615384615 0.561380353 11

From the optimization of machining parameters using Grey Relational Analysis it is found out that the optimum
parameters are 6000rpm spindle speed, 0.1mm/rev feed rate and 0.1mm depth of cut.

5.2 Desirability function analysis

To analyse the desirability function we have to find out the individual desirability and composite desirability
from that we have to rank according to descending order the ranks will help us to find out the optimum parameters.
Table 4 given below shows the obtained values for the individual desirability, composite desirability and rank.
1022 Mohammed Yaser E.K and K Shunmugesh / Materials Today: Proceedings 11 (2019) 1015–1023

Table 4 Desirability function analysis

Individual desirability
Sl. No. Composite desirability rank
Ra(µm) Fm(N) Fd
1 0.041629687 0.047013 0.091432306 0.749823983 15
2 0.050200088 0.031873 0.039291914 0.724517331 16
3 0.029722504 0.020514 0.029484986 0.694831073 20
4 0.002094606 0.0155 0.002185778 0.577304544 23
5 0.000266736 0.006758 0.006634812 0.543239408 24
6 0.039019038 0.010234 0.044958117 0.69498845 19
7 0 0.013097 0 0 26
8 0.000450506 0.001252 0.002185778 0.504263628 25
9 6.72003E-05 0 0.003989369 0 26
10 0.336537795 0.313919 0.163935034 0.873772369 7
11 0.302778787 0.063111 0.290402211 0.841474094 9
12 0.261580835 0.030069 0.136467939 0.796952521 13
13 0.056561428 0.1872 0.073370162 0.787288545 14
14 0.036526141 0.040619 0.029484986 0.715671173 17
15 0.117976545 0.062268 0.179016044 0.801673132 12
16 0.013431652 0.052946 0.005195402 0.658958804 22
17 0.070957275 0.062059 0.002993656 0.688352853 21
18 0.056561428 0.024863 0.015205663 0.69927045 18
19 0.85678174 1 0.808174628 0.987791441 1
20 1 0.400066 0.570828296 0.952434268 2
21 0.769766898 0.265612 1 0.948716397 3
22 0.498425573 0.529817 0.290402211 0.91810757 4
23 0.385572251 0.337837 0.472702247 0.911247421 5
24 0.8344092 0.15821 0.386992162 0.906349737 6
25 0.067140818 0.399299 0.163935034 0.833766751 11
26 0.233402913 0.121484 0.312757247 0.854371231 8
27 0.251935704 0.063323 0.290402211 0.836323643 10

From the optimization of machining parameters using Desirability Function Analysis it is found out that the
optimum parameters are 6000rpm spindle speed, 0.1mm/rev feed rate and 0.1mm depth of cut.

The given Table 5 shows the optimum parameters for machining of GFRP found out using GRA and GFA.

Table 5 Optimum parameters

control factors response
Spindle Speed Feed Rate Depth of Cut Surface Roughness Force Delamination Factor
6000 0.1 0.2 0.95 22.79 1.005
 Mohammed Yaser E.K and K Shunmugesh / Materials Today: Proceedings 11 (2019) 1015–1023 1023

3. Conclusion

i. Material has a higher surface roughness at lower cutting speed.

ii. Force and delamination factor was found comparatively to be minimum at the maximum speed(6000 rpm)
minimum feed rate(.1 rev/mm) and minimum depth of cut(.1 mm).

iii. In order to obtain a higher surface finish machining should be carried out at higher cutting speed, low feed
rate and by low depth of cut.

References

1. S. Jahanmir, M. Ramulu, and P. Koshy, Machining of Ceramics and Composites(Marcel Dekker, Inc., New York, 2000).
2. W.F. Smith, Principles of materials science and engineering (MacGraw-Hill, 1990)
3. Jamal Y. Sheikh-Ahmad, 2009. Machining of Polymer composites. Springer Science and Business Media LLC.
4. Palanikumar K, Modeling and analysis of delamination factor and surface roughness in drilling GFRP composites. Materials and
Manufacturing Processes, 25 (10), 2010, 1059–1067.
5. Palanikumar K, Prakash S, and Shanmugam K, Evaluation of delamination in drilling GFRP composites, Materials and Manufacturing
Processes, 23 (8), 2008, 858–864.
6. lu H S., Chang C K , HwangN C and Chung C T, " Grey relational analysis coupled with PCA for optimization design of the cutting
parameters in high sped end milling", Journal of Material processing Technology, 209, 3808-3817, [2009].
7. J. Song, Y. Dang, and Z. Hua, “Study on group decision-making method based on grey cluster model,” Control and Decision, vol. 25, no.
10, pp. 1593–1597, 2010.
8. Derringer G and Suich. 1980. Simultaneous optimization of several response variables. J. Qua. Technol. 12(4): 214.
9. Ku, W.L.; Hung, C.L.; Lee, S.M.; Chow, H.M. Optimization in thermal friction drilling for SUS 304 stainlesssteel. International Journal of
Advanced Manufacturing Technology 2011, 53, 935 – 944.7.
10. Naveen Sait, S. Aravindan, A. Noorul Haq, Optimisation of Machining Parameters of Glass-Fibre-ReinforcedPlastic (GFRP) Pipes by
desirability Function Analysis using Taguchi Technique,International Journal of Advanced Manufacturing Technology, Vol. 43, pp. 581 –
589, 2009.8.
11. J. R. Turner, J. F. Thayer,Introduction to Analysis of Variance: Design, Analysis, and Interpretation(Sage, Beverly Hills, CA, 2001).9.
12. R. S. Bogartz,Introduction to Analysis of Variance(Praeger, Westport, CT, 1994).
13. Phadke, M. S. (1989). Quality engineering using robust design, New Jersey: AT&T Bells Laboratory/Prentice-Hall.
14. Lin, J. L. and Lin (2002). International Journal of Tools & Manufacturing Technology.
15. Mohammed T. Hayajneh, Montasser S. Tahat, Joachim Bluhm (2007) ,‟Study of the Effects of Machining Parameters on the Surface
Roughness in the End-Milling Process‟, Jordan Journal of Mechanical and Industrial Engineering,(2005).
16. Kini, M.V., Chincholkar, A.M. (2010) “Effect of machining parameters on surface roughness and material removal rate in finish turning of
± 30˚ glass fibre reinforced polymer pipes”, Material and Design.
17. AbhishekDubey, DevendraPathak, Nilesh Chandra, AjendraNath Mishra , Rahul Davis, “A parametric design study of end milling
operation using grey based Taguchi method”.
18. Ching-ming Hsu (2004).“An integrated optimization approach based on neural networks, exponential desirability functions and Tabu
search”.
19. NihatTosun(2006). Determination of optimum parameters for multi-performance characteristics in drilling by using grey relational analysis,
Int J AdvManuf Technology, 28(1) 450– 455.