* Corresponding author. Tel.: +9831568294 (M), +33-2548-2655 (R) E-mail address: [email protected] (S. Chakraborty) © 2019 by the authors; licensee Growing Science, Canada doi: 10.5267/j.msl.2018.12.004
Management Science Letters 9 (2019) 467–494
Contents lists available at GrowingScience
Management Science Letters
homepage: www.GrowingScience.com/msl
Applications of optimization techniques for parametric analysis of non-traditional machining processes: A Review
Shankar Chakrabortya*, Bijoy Bhattacharyyaa and Sunny Diyaleyb
aDepartment of Production Engineering, Jadavpur University, Kolkata, West Bengal, India bDepartment of Mechanical Engineering, Sikkim Manipal Institute of Technology, Majitar, Sikkim, India C H R O N I C L E A B S T R A C T
Article history: Received: October 27, 2018 Received in revised format: No-vember 28, 2018 Accepted: December 7, 2018 Available online: December 7, 2018
The constrained applications of conventional machining processes in generating complex shape geometries with the desired degree of tolerance and surface finish in various advanced engineering materials are being gradually compensated by the non-traditional machining (NTM) processes. These NTM processes usually have higher procurement, maintenance, operating and tooling cost. Hence, in order to attain their maximum machining performance, they are usually operated at their optimal or near optimal parametric settings which can easily be determined by the application of different optimization techniques. In this paper, 133 international research papers published during 2012-16 on parametric optimization of NTM processes are extensively reviewed to have an idea on the selected process parameters, observed responses, work materials machined and optimization techniques employed in those processes while generating varying part geometries for their indus-trial use. It is observed that electro discharge machining is the mostly employed NTM process, applied voltage is the identified process parameter with maximum importance, surface roughness and material removal rate are the two maximally preferred responses, different steel grades are the mostly machined work materials and grey relational analysis is the most popular tool utilized for parametric optimization of NTM processes. These observations would help the process engineers to attain the machining performance of the NTM processes at their fullest extents for different work material and shape feature combinations.
© 2019 by the authors; licensee Growing Science, Canada
Keywords: Non-traditional machining pro-cess Response Process parameter Work material Optimization
1. Introduction Difficulties in machining complicated and intricate shape features in varied hard-to-machine, high-strength-temperature-resistant materials, superalloys, metal matrix composites (MMCs) and other ad-vanced engineering materials for aviation, nuclear power, wafer fabrication and automobile applica-tions using conventional machining processes have caused to the evolution of an array of non-tradi-tional machining (NTM) processes. In conventional material removal processes, like turning, milling, shaping, drilling etc., forces are applied on the workpiece with the help of a cutting tool to remove excess material in the form of chips. It induces plastic deformation within the workpiece leading to material removal due to shear action. On the other hand, in NTM processes, instead of employing sharp
468
cutting tools, materials are removed using mechanical, thermal, electrical or chemical energy or com-binations of them. In some of the NTM processes, the tool does not even make any contact with the workpiece, and it is also not mandatory that the tool material should have higher hardness than the work material. Basically, in NTM processes, material removal takes place from the workpiece surface in the form of minute particles while attaining superior surface smoothness and dimensional accuracy. These processes are generally classified based on the form of energy deployed for removal of material, i.e.: a) Mechanical processes: Water jet machining (WJM), ultrasonic machining (USM), abrasive jet ma-
chining (AJM) etc. b) Electrochemical processes: Electrochemical machining (ECM), electro jet drilling (EJD), electro-
chemical deburring (ECD) etc. c) Electrical thermal processes: Electrical discharge machining (EDM), wire electrical discharge ma-
chining (WEDM), laser beam machining (LBM), electron beam machining (EBM) etc. d) Chemical processes: Chemical machining (CHM), photochemical machining (PCM) etc. e) Hybrid machining processes: Electrochemical grinding (ECG), electrochemical discharge machin-
ing (ECDM), electrochemical honing (ECH), abrasive water jet machining (AWJM), travelling wire electrochemical spark machining (TW-ECSM) etc.
These NTM processes have exceptionally low material removal rate (MRR) and consume excessive specific energy. Most of them have comparatively high procurement cost, tooling and fixture cost, power consumption and operating cost, and maintenance cost. Therefore, for productive and economic exploration of the capacities of NTM processes, their different machining/process parameters need to be optimally selected. Identification of the most relevant process parameters and their settings for any NTM process mainly depend on the expert knowledge and skill of the concerned operator. Sometimes, manufacturer’s catalogues may also help in setting these parametric combinations to achieve the best machining performance. But, in most of the cases, these parametric combinations are conservative and depart from their optimal settings which cause hindrance in full exploitation of the capabilities of NTM processes. Selection of the optimal or near optimal parametric mix for different NTM processes thus becomes a vital decision making task. A variety of analytical techniques in the form of mathematical algorithms has been fruitfully employed for parametric optimization of NTM processes in order to fully explore their machining potentials and capabilities. At this very time, it now becomes essential to have an exhaustive investigation on the application and efficacy of various optimization techniques in deter-mining the best parametric settings for different NTM processes. In this paper, altogether 133 research papers published in different international journals during 2012-16 dealing with the applications of ECM, EDM, WEDM, LBM, USM and hybrid machining (HM) processes in real time manufacturing environment are extensively reviewed. This review mainly fo-cuses on the identification of the type of work material used for the machining operation, machining parameters and responses chosen for the considered NTM processes, nature of the optimization prob-lem developed and optimization technique(s) employed for achieving the best machining performance. The details of this analysis for the six above-mentioned NTM processes are presented here-in-under. 1.1 ECM process The first machining process similar to ECM was patented by Gusseff in 1929, and its successful com-mercial application was started in late 1950s and early 1960s in aerospace and other manufacturing industries to perform shaping and finishing operations. It is a carefully regulated anodic dissolution process to shape the workpiece (anode) using a tool (cathode), approaching towards the workpiece with a constant feed rate. The electrolyte (NaCl, NaNO3) flowing at high speed through the inter-electrode gap usually removes the dissolved metal from the machining zone. Between the pre-shaped cathode tool and the anode workpiece, a DC voltage (10-25 V) is applied, causing atomic level reactions to take
S. Chakraborty et al. / Management Science Letters 9 (2019) 469
place within the electrolytic medium which is mainly responsible for removal of metal from the work-piece to achieve its desired shape. The final workpiece shape thus becomes an approximate negative mirror copy of the tool electrode (Rajurkar et al., 1999). In ECM process, as the mechanical or physical properties of the work material do not influence the material removal mechanism, it can machine different electrically conductive materials regardless of their hardness, toughness or thermal characteristics. Almost all kinds of metal, including high-alloyed nickel, titanium-based alloys, superalloys and MMCs, can be efficiently machined using this process. It has several advantages, like no tool wear, high MRR, good surface finish, no need for deburring operation, and capability of generating complex shape geometries (contours, ring ducts, grooves etc.) for subsequent use in aerospace, automotive, defense and medical industries. Nowadays, it is also being successfully utilized in micro-machining and fabrication (dimensions within 1-999 μm) of engineering components (Bhattacharyya et al., 2002). For achieving maximum benefits from this machining sys-tem, its different process parameters can be selected as applied voltage, current, electrolyte temperature, electrolyte flow rate, electrolyte concentration, inter-electrode gap and tool material (copper, brass or bronze) (Senthilkumar et al., 2013). In view of difficulties encountered from the conventional machin-ing processes, like high tool wear and high tooling cost, this process now proffers an effective alterna-tive to fulfill the requirements of the machining personnel. Table 1 exhibits the details of the research works executed during 2012-16 on parametric optimization of ECM processes. It mainly includes various control parameters and responses selected for ECM op-eration, type of the work material considered for this process and optimization technique(s) adopted for its parametric optimization. The analysis of this information is graphically represented in Fig. 1. From this figure, it can be revealed that among various machining parameters of ECM process, applied volt-age is the most important one, followed by tool feed rate, electrolyte flow rate, electrolyte concentration and inter-electrode gap. With respect to the responses, MRR is provided with the maximum importance, followed by surface roughness and radial overcut. Among the chosen work materials, various grades of steel are maximally machined using ECM processes, followed by different MMCs, and titanium and its alloys. From the parametric optimization point of view, various advanced optimization techniques, like biogeography-based optimization, cuckoo search optimization, artificial bee colony optimization etc. are identified as the most popular methods, followed by grey relational analysis (GRA) and genetic algorithm (GA). In single objective optimization, among all the considered responses, each of them is separately optimized and different parametric combinations are derived for each of the responses which are quite difficult to maintain from the machining point of view. Rather, multi-objective optimization is more practical because in this approach, a unique parametric setting is obtained while simultaneously optimizing all the conflicting responses. Among the 14 research papers identified dealing with para-metric optimization of ECM processes, ten papers considered only multi-objective optimization of the responses, while the remaining papers provided emphasis on both single and multi-objective optimiza-tion of the responses. 1.2 EDM process Although, Joseph Preistly first observed the principles of the EDM process in 1770, two Soviet re-searchers, the Lazarenkos’, succeeded in the development of a machining process in the 1940’s that formed the foundation for the present EDM system. In this process, electrical energy is utilized to originate electrical spark between an electrode and a workpiece, and material removal principally takes place due to electro-discharge erosion. An intense heat with temperature between 8000o-12000o C is generated by this electric spark, which when carefully controlled and localized, can only affect the workpiece surface. The metal removal mechanism is based on the application of a pulsating (on/off) electrical charge carrying high frequency current through the electrode to the workpiece, which causes controlled erosion of minute particles of metal from the workpiece. Instant vapourization and melting of the material are thus responsible for material removal. The tool and the work material are submerged
470
in a dielectric medium (kerosene or deionized water which also acts as a coolant and washes out the eroded metal particles), and a gap is steadily maintained between the tool and the workpiece. In EDM process, both the tool and the workpiece material must be good conductor of electricity. Thus, any material that is electrically conductive (steel, titanium, superalloys, brass etc.) can easily be machined using this process (Ho and Newman, 2003). Its major advantages include limited heat affected zone (HAZ), surface hardening, no burr formation, generation of complex part geometries and faster ma-chining operation. It has huge applications in die and mold making industries. Its various control pa-rameters are open circuit voltage, spark gap, pulse-on time, pulse-off time, maximum (peak) current, polarity, dielectric medium etc.
(a)
(b)
(c) (d)
Fig. 1 Analysis on the parametric optimization of ECM processes A list of the reviewed research papers on parametric optimization of EDM processes is provided in Table 2. On the other hand, the pertinent information regarding the process parameters and responses selected, work materials machined and optimization techniques adopted are provided in Fig. 2. It can be revealed from this figure that supply current, pulse-on time and pulse-off time are the three most predominant parameters for EDM process. They are subsequently followed by other parameters, like gap voltage, duty factor/duty cycle, applied voltage, tool rotational speed, pulse duration, tool electrode lift time, flushing pressure etc. according to their preference. Some least important process parameters, like feed rate, work time, capacitance, machining polarity, dielectric level, tool material, dielectric flow rate, aspect ratio of the tool, no load voltage, inter-electrode gap etc. are merged together into a single group (treated as others). These parameters are not so common in EDM processes, and their availability and settings mainly count on the type of EDM machine employed for material removal and part gener-ation. Among the responses, MRR, surface roughness and electrode wear rate (EWR) have the maxi-mum importance, followed by overcut, taper ratio, white layer thickness, surface crack density etc. Circularity, process energy, residual stress, process time, dielectric consumption etc. are identified as the least important responses. It is interestingly noted that different grades of steel (mainly AISI varie-ties) and alloys (Inconel 718, Invar, aluminum alloys and Rene 80) are the widely machined work
02468
101214
0
2
4
6
8
10
12
14
MRR Surfaceroughness
Overcut Cylindricity
0
1
2
3
4
5
Steels MMCs Ti and itsalloys
Tungstencarbide
Otheralloys
0
1
2
3
4
S. Chakraborty et al. / Management Science Letters 9 (2019) 471
materials using EDM process. Different MMCs, titanium and its alloys, and ceramics (aluminum oxide) are also machined by this process. For parametric optimization of EDM process, GRA is the most popular technique, followed by the other advanced optimization methods (like bio-geography based optimization, teaching learning-based optimization, particle swarm optimization etc.), principal com-ponent analysis (PCA), non-sorting genetic algorithm (NSGA-II), desirability function, neural net-works (NN), simulated annealing (SA) etc. For this process, only a single paper considered single ob-jective optimization, 31 papers attempted multi-objective optimization and the remaining four papers took into account both single and multi-objective optimization of the responses. 1.3 WEDM process The principle of material removal in WEDM process is quite identical to that of EDM process. In this process, material is gradually worn out from the workpiece using a series of sparks generating between the workpiece and the wire isolated by a flow of dielectric fluid. This fluid is continuously supplied into the machining zone, enabling complex shapes being generated with high accuracy. The wire is made of thin copper, tungsten or brass having diameter 0.05-0.3 mm. Because of its versatility, it is used in several areas, like aviation, medical, electronics and semiconductor applications, tool and die making industries, manufacturing of fixtures, gauges, cams, gears, strippers, punches, electrodes etc. As this process employs no force and does not form burrs, it can be effectively applied for machining of delicate parts (Ho et al., 2004). Its different process parameters include peak current, supply voltage, pulse-on time, pulse-off time, polarity, work material, size and speed of wire, feed rate, gain, rate of flushing, type of the dielectric medium etc. Table 1 Parameters, responses, work materials and optimization techniques considered in ECM processes
Sl. No.
Name of the authors Work material machined
Single/ Multi-
objective
Optimization tool(s) adopted
Process parameters Responses
1. Abuzied et al. (2012) - Multiple Artificial neural network
Electrolyte flow rate, applied volt-age, tool feed rate
MRR, surface roughness
2. Dhobe et al. (2014)
Titanium Both Quality loss function Electrolyte concentration, electrolyte flow rate, tool feed rate, inter-elec-trode gap, applied voltage
MRR, surface roughness
3. Gopal and Chakrachar (2012)
EN31 steel Multiple Grey relational analysis
Electrolyte concentration, applied voltage, tool feed rate
MRR, overcut, cylindricity
4. Jegan et al. (2013) Metal matrix composites
Multiple Weighted sum genetic algorithm
Current, applied voltage, tool feed rate, electrolyte concentration
MRR, surface roughness
5. Kalaimathi et al. (2014)
Monel 400 alloy Multiple Desirability function Inter-electrode gap, applied voltage, electrolyte concentration
MRR, surface roughness
6. Manikandan et al. (2015)
Ti-6Al-4V titanium alloy
Both Grey relational analysis
Tool feed rate, electrolyte concentration, electrolyte flow rate
MRR, overcut
7. Mukherjee and Chakraborty (2013)
EN8 steel
Both
Biogeography-based optimization
Electrolyte concentration, applied voltage, electrolyte flow rate, inter-electrode gap
MRR, overcut
8. Rao and Pad-manabham (2015)
Aluminium matrix and boron carbide metal matrix com-
posite
Both Utility concept Applied voltage, electrolyte flow rate, tool feed rate, reinforcement content
MRR, surface roughness, radial overcut
9. Sathiyamoorthy et al. (2015)
High carbon high chromium die tool
steel
Multiple Genetic algorithm Applied voltage, inter-electrode gap, tool feed rate, electrolyte discharge rate
MRR, surface roughness
10. Sathiyamoorthy and Sekar (2016)
AISI 202 stainless steel
Multiple Genetic algorithm Applied voltage, tool feed rate, elec-trolyte discharge rate
MRR, surface roughness
11. Sohrabpoor et al. (2016)
Cemented tungsten carbide
Multiple Cuckoo optimization algorithm
Electrolyte concentration, applied voltage, electrolyte flow rate, tool feed rate
MRR, surface roughness, radial overcut
12. Solaiyappan et al. (2014)
Silicon carbide composite
Multiple Hybrid fuzzy-artificial bee colony algorithm
Applied voltage, electrolyte flow rate, current, inter-electrode gap, tool feed rate, electrolyte concentration
MRR, surface roughness, overcut
13. Tang and Yang (2013) Special stainless steel
00Cr12Ni9Mo4Cu2
Multiple Grey relational analy-sis
Applied voltage, electrolyte pressure, electrolyte concentration, tool feed rate
MRR, side gap, surface roughness
14. Teimouri and Shor-abpoor (2013)
Cemented tungsten carbide
Multiple Adaptive neuro-fuzzy inference system,
cuckoo optimization algorithm
Electrolytic concentration, applied voltage, electrolyte flow rate, tool feed rate
MRR, surface roughness
47
2
Tab
le 2
Pa
ram
eter
s, re
spon
ses,
wor
k m
ater
ials
and
opt
imiz
atio
n te
chni
ques
con
side
red
in E
DM
pro
cess
es
Sl.N
o.
Nam
e of
aut
hor(
s)
Wor
k m
ater
ial
mac
hine
d Si
ngle
/ M
ulti-
obje
ctiv
e O
ptim
izat
ion
tool
(s) a
dopt
ed
Proc
ess p
aram
eter
s R
espo
nse(
s)
1.
Aic
h an
d B
aner
jee
(201
6)
Hig
h sp
eed
stee
l M
ultip
le
Teac
hing
lear
ning
-bas
ed o
ptim
iza-
tion
algo
rithm
Su
pply
cur
rent
, pul
se-o
n tim
e, p
ulse
-off
tim
e M
RR
, sur
face
roug
hnes
s
2.
Ani
tha
et a
l. (2
016)
A
ISI D
2 to
ol st
eel
M
ultip
le
Arti
ficia
l neu
ral n
etw
ork
Puls
e cu
rren
t, pu
lse-
on t
ime,
dut
y cy
cle,
ap-
plie
d vo
ltage
M
RR
, sur
face
roug
hnes
s
3.
Ay
et a
l. (2
013)
In
cone
l 718
M
ultip
le
Gre
y re
latio
nal a
naly
sis
Dis
char
ge c
urre
nt, p
ulse
dur
atio
n
Tape
r rat
io, h
ole
dila
tion
4.
Bar
aska
r et a
l. (2
013)
EN
8 ca
rbon
stee
l M
ultip
le
Non
-dom
inat
ed so
rting
gen
etic
al
gorit
hm
Dis
char
ge c
urre
nt,
pul
se-o
n tim
e, p
ulse
-off
tim
e M
RR
, sur
face
roug
hnes
s
5.
Beh
era
et a
l. (2
015)
ZA
27/S
iC
met
al m
atrix
co
mpo
site
Mul
tiple
G
rey
rela
tiona
l ana
lysi
s Po
wde
r con
cent
ratio
n in
die
lect
ric, S
iC%
in d
i-el
ectri
c, p
ulse
d cu
rren
t, du
ty c
ycle
M
RR
, su
rfac
e ro
ughn
ess,
tool
wea
r rat
e
6.
Bha
rti e
t al.
(201
2)
Inco
nel 7
18
Mul
tiple
N
on-d
omin
ated
sorti
ng g
enet
ic a
l-go
rithm
Sh
ape
fact
or, d
isch
arge
cur
rent
, pul
se-o
n tim
e,
gap
volta
ge, d
uty
cycl
e, fl
ushi
ng p
ress
ure,
tool
el
ectro
de li
ft tim
e
MR
R, s
urfa
ce ro
ughn
ess
7.
Chak
ravo
rty e
t al.
(201
2)
Cer
amic
s (A
l 2O3 +
30
vol%
Ti
C),
stai
nles
s ste
el 3
04
Mul
tiple
PC
A-b
ased
gre
y re
latio
nal a
naly
-si
s, PC
A-b
ased
pro
porti
on o
f qua
l-ity
loss
redu
ctio
n, P
CA
-bas
ed
TOPS
IS, w
eigh
ted
prin
cipa
l com
-po
nent
ana
lysi
s
Mac
hini
ng p
olar
ity, p
eak
curr
ent,
appl
ied
volt-
age,
no
load
vol
tage
, pu
lse
dura
tion,
se
rvo
vo
ltage
, cap
acita
nce,
resi
stan
ce, f
eed
rate
, too
l ro
tatio
nal s
peed
MR
R,
surf
ace
roug
hnes
s, el
ectro
de w
ear r
ate,
ent
ranc
e cl
eara
nce,
exi
t cle
aran
ce
8.
Dav
e et
al.
(201
2)
Inco
nel 7
18
Mul
tiple
Ta
guch
i los
s fun
ctio
n O
rbita
l rad
ius,
orbi
tal s
peed
, gap
vol
tage
, cur
-re
nt, p
ulse
-on
time,
dut
y fa
ctor
M
RR
, too
l wea
r rat
e, su
rfac
e ro
ughn
ess
9.
Dew
anga
n an
d B
isw
as (2
013)
A
ISI P
20 to
ol st
eel
Mul
tiple
G
rey
rela
tiona
l ana
lysi
s D
isch
arge
cur
rent
, wor
k tim
e, li
ft tim
e, p
ulse
-on
tim
e, in
ter-
ele
ctro
de g
ap
MR
R, t
ool w
ear r
ate
10.
Dew
anga
n et
al.
(201
5)
AIS
I P20
tool
stee
l M
ultip
le
Gre
y-fu
zzy
logi
c D
isch
arge
cur
rent
, t
ool
wor
k tim
e, t
ool
lift
time,
pul
se-o
n tim
e
Whi
te l
ayer
thi
ckne
ss,
sur-
face
cr
ack
dens
ity, s
urfa
ce ro
ugh-
ness
11
. G
olsh
an e
t al.
(201
2)
Al/S
iC c
ompo
site
M
ultip
le
Non
-dom
inat
ed so
rting
gen
etic
al-
gorit
hm
Puls
e-on
tim
e, a
vera
ge g
ap v
olta
ge, p
ulse
pea
k cu
rren
t, pe
rcen
t vol
ume
frac
tion
of S
iC
MR
R, s
urfa
ce ro
ughn
ess
12.
Jaga
dish
and
Ray
(201
5)
AIS
I D2
tool
stee
l M
ultip
le
Gre
y re
latio
nal a
naly
sis
Puls
e du
ratio
n, p
eak
curr
ent,
diel
ectri
c le
vel,
flush
ing
pres
sure
Pr
oces
s tim
e,
proc
ess
en-
ergy
, too
l wea
r rat
io, c
once
n-tra
tion
of a
eros
ol,
diel
ectri
c co
nsum
ptio
n
13.
Maj
umda
r (20
12)
Mild
stee
l Si
ngle
G
enet
ic a
lgor
ithm
C
urre
nt, p
ulse
-on
time,
pul
se-o
ff ti
me
Elec
trode
wea
r rat
e 14
. M
ajum
der (
2013
) A
ISI 3
16 L
N st
ainl
ess s
teel
M
ultip
le
Fuzz
y-ba
sed
parti
cle
swar
m o
pti-
miz
atio
n Su
pply
cur
rent
, pul
se-o
n tim
e, p
ulse
-off
tim
e M
RR
, ele
ctro
de w
ear r
ate
15.
Maj
umde
r et a
l. (2
014)
A
ISI 3
16 L
N st
ainl
ess s
teel
M
ultip
le
Des
irabi
lity-
base
d pa
rticl
e sw
arm
i
ii
Supp
ly c
urre
nt, p
ulse
-on
time,
pul
se-o
ff ti
me
MR
R, e
lect
rode
wea
r rat
e 16
. M
ajum
dar
(201
5)
Mild
stee
l M
ultip
le
Gen
etic
alg
orith
m, s
imul
ated
an-
neal
ing,
par
ticle
swar
m o
ptim
iza-
tion,
arti
ficia
l neu
ral n
etw
ork
Dis
char
ge c
urre
nt,
puls
e-on
tim
e,
puls
e-of
f tim
e M
RR
, to
ol w
ear r
atio
S. C
hakr
abor
ty e
t al.
/ M
anag
emen
t Sci
ence
Let
ters
9 (2
019)
47
3
Tab
le 2
Pa
ram
eter
s, re
spon
ses,
wor
k m
ater
ials
and
opt
imiz
atio
n te
chni
ques
con
side
red
in E
DM
pro
cess
es (C
ontin
ued)
Sl
.No.
N
ame
of a
utho
r(s)
W
ork
mat
eria
l m
achi
ned
Sing
le/
Mul
ti-ob
ject
ive
Opt
imiz
atio
n to
ol(s
) ado
pted
Pr
oces
s par
amet
ers
Res
pons
e(s)
17.
Min
g et
al.
(201
6)
SiC
/Al
com
posi
te
Mul
tiple
G
enet
ic a
lgor
ithm
with
de
sira
bilit
y fu
nctio
n Pe
ak c
urre
nt, p
ulse
-on
time,
pul
se-o
ff ti
me,
ap-
plie
d vo
ltage
M
RR
, sur
face
roug
hnes
s
18.
Mog
hadd
am a
nd K
olah
an
(201
5)
AIS
I 231
2 ho
t wor
ked
stee
l M
ultip
le
Sim
ulat
ed a
nnea
ling
Peak
cur
rent
, pul
se-o
n tim
e, p
ulse
-off
tim
e, a
p-pl
ied
volta
ge,
duty
fact
or
MR
R, t
ool w
ear r
ate,
surf
ace
roug
hnes
s 19
. M
ohan
ty e
t al.
(201
6a)
AIS
I D2
stee
l M
ultip
le
Non
-dom
inat
ed so
rting
gen
etic
al-
gorit
hm
Dis
char
ge c
urre
nt,
puls
e-on
tim
e, d
uty
fact
or
MR
R, t
ool
wea
r ra
te, r
esid
-ua
l stre
ss
20.
Moh
anty
et a
l. (2
016b
) In
cone
l 718
M
ultip
le
Parti
cle
swar
m o
ptim
izat
ion
Dis
char
ge c
urre
nt,
open
circ
uit
volta
ge,
duty
fa
ctor
, pu
lse-
on t
ime,
flu
shin
g pr
essu
re,
tool
m
ater
ial
MR
R,
surf
ace
roug
hnes
s, el
ectro
de w
ear r
ate,
ove
rcut
21.
Muk
herje
e an
d C
hakr
abor
ty
(201
2)
Die
stee
l
Bot
h B
ioge
ogra
phy-
base
d op
timiz
atio
n al
gorit
hm
Peak
cur
rent
, av
erag
e ga
p vo
ltage
, pu
lse-
on
time,
per
cent
vol
ume
frac
tion
of S
iC p
rese
nt in
al
umin
um m
atrix
Surf
ace
roug
hnes
s, w
hite
la
yer t
hick
ness
, sur
face
crac
k de
nsity
, M
RR
, to
ol
wea
r ra
te, g
ap si
ze
22.
Padh
ee e
t al.
(201
2)
EN31
stee
l M
ultip
le
Non
-dom
inat
ed so
rting
gen
etic
al-
gorit
hm
Puls
e-on
tim
e, p
eak
curr
ent,
duty
fact
or,
con-
cent
ratio
n of
the
abra
sive
M
RR
, sur
face
roug
hnes
s
23.
Pand
a et
al.
(201
5)
Stai
nles
s ste
el (S
304)
M
ultip
le
Gre
y re
latio
nal a
naly
sis,
parti
cle
swar
m o
ptim
izat
ion
Dis
char
ge cu
rren
t, di
elec
tric f
low
rate
, pul
se-o
n tim
e, p
ulse
-off
tim
e
MR
R, t
ool w
ear r
ate,
surf
ace
roug
hnes
s 24
. Po
rwal
et a
l. (2
012)
In
var
Mul
tiple
A
rtific
ial n
eura
l net
wor
k G
ap v
olta
ge, c
apac
itanc
e, sp
indl
e sp
eed
MR
R, t
ool w
ear r
ate,
hol
e ta
-pe
r 25
. Pr
adha
n (2
012)
A
ISI D
2 st
eel
Mul
tiple
Pr
inci
pal c
ompo
nent
ana
lysi
s, gr
ey
rela
tiona
l ana
lysi
s D
isch
arge
cur
rent
, pul
se-o
n tim
e, a
pplie
d vo
lt-ag
e, d
uty
cycl
e
MR
R, s
urfa
ce ro
ughn
ess
26.
Priy
adar
shin
i and
Pal
(201
6)
Tita
nium
allo
y (T
i-6A
l-4V
) B
oth
Prin
cipa
l com
pone
nt a
naly
sis,
grey
re
latio
nal a
naly
sis
Puls
e du
ratio
n, d
uty
fact
or, d
isch
arge
cur
rent
, ga
p vo
ltage
M
RR
, too
l wea
r rat
e, su
rfac
e ro
ughn
ess
27.
Rad
hika
et a
l. (2
015)
A
lum
iniu
m a
lloy
(Al-S
i10M
g)
Mul
tiple
G
rey
rela
tiona
l ana
lysi
s Pe
ak c
urre
nt, f
lush
ing
pres
sure
, pul
se-o
n tim
e M
RR
, su
rfac
e ro
ughn
ess,
tool
wea
r rat
e 28
. R
aja
et a
l. (2
015)
D
ie st
eel
Mul
tiple
Fi
refly
alg
orith
m
Supp
ly c
urre
nt, p
ulse
-on
time
Surf
ace
roug
hnes
s, m
achi
n-in
g tim
e 29
. Sa
hu a
nd N
ayak
(201
5)
AIS
I P20
tool
stee
l M
ultip
le
Gen
etic
alg
orith
m
Puls
e-on
tim
e, d
isch
arge
cur
rent
M
RR
, ove
rcut
, too
l wea
r rat
e 30
. Si
ngh
(201
2)
Alu
min
ium
met
al m
atrix
com
po-
site
s M
ultip
le
Gre
y re
latio
nal a
naly
sis
Asp
ect r
atio
, dut
y cy
cle,
gap
vol
tage
, pul
se cu
r-re
nt, p
ulse
-on
time,
tool
ele
ctro
de li
ft tim
e M
RR
, su
rfac
e ro
ughn
ess,
tool
wea
r rat
e 31
. Ta
ng a
nd D
u (2
014)
Sp
ecia
l sta
inle
ss st
eel
00C
r12N
i9M
o4C
u2
Mul
tiple
G
rey
rela
tiona
l ana
lysi
s To
ol f
eed
rat
e, a
pplie
d vo
ltage
, el
ectro
lyte
pr
essu
re, e
lect
roly
te c
once
ntra
tion
MR
R,
side
ga
p,
surf
ace
roug
hnes
s 32
. Te
imou
ri an
d B
aser
i (20
12)
EN 3
2 m
ild st
eel
Bot
h A
rtific
ial b
ee c
olon
y a
lgor
ithm
Pu
lse
curr
ent,
gap
volta
ge, p
ulse
-on
time,
dut
y fa
ctor
, air
inta
ke p
ress
ure,
spin
dle
spee
d M
RR
, sur
face
roug
hnes
s
33.
Teim
ouri
and
Bas
eri
(201
4)
SPK
(X21
0Cr1
2) c
old
wor
k st
eel
Mul
tiple
A
dapt
ive
neur
o-fu
zzy
infe
renc
e sy
stem
, con
tinuo
us a
nt c
olon
y op
-tim
izat
ion
algo
rithm
Mag
netic
fie
ld
inte
nsity
, ro
tatio
nal
spee
d,
prod
uct o
f cur
rent
and
pul
se-o
n tim
e M
RR
, sur
face
roug
hnes
s
34.
Than
gadu
rai a
nd A
sha
(201
4)
Alu
min
ium
bor
on c
arbi
de c
om-
posi
te
Mul
tiple
D
esira
bilit
y fu
nctio
n, g
enet
ic a
lgo-
rithm
Pu
lse-
on ti
me,
pul
se-o
ff ti
me,
cur
rent
M
RR
, su
rfac
e ro
ughn
ess,
to
ol w
ear r
ate
35.
Uyy
ala
and
Kum
ar (2
014)
R
ENE
80 n
icke
l sup
er a
lloy
Bot
h G
rey
rela
tiona
l ana
lysi
s Pe
ak c
urre
nt, p
ulse
-on
time,
pul
se-o
ff ti
me
MR
R, s
urfa
ce ro
ughn
ess,
ra-
dial
ove
rcut
36
. Y
adav
and
Yad
ava
(201
5)
Tita
nium
allo
y (T
i-6A
l-4V
) M
ultip
le
Gre
y re
latio
nal a
naly
sis,
prin
cipa
l co
mpo
nent
ana
lysi
s To
ol e
lect
rode
spe
ed, p
ulse
-on
time,
dut
y fa
c-to
r, ga
p cu
rren
t M
RR
, su
rfac
e ro
ughn
ess,
circ
ular
ity
474
(a)
(b)
(c) (d)
Fig. 2 Analysis on the parametric optimization of EDM processes
In Table 3, the parameters and responses considered, work materials machined and optimization tech-niques adopted for parametric optimization of WEDM processes are listed based on the reviewed re-search papers. Figure 3 presents the related detailed analyses. It can be noticed that for WEDM process, wire feed rate, pulse-off time and pulse-on time are the most frequently set process parameters, followed by wire tension, current, gap voltage, flushing pressure, pulse duration, servo feed and workpiece thick-ness. There are also some insignificant process parameters, like spark gap, capacitance, duty factor, pulse frequency, taper angle, power, dielectric flow rate, rotational speed, corner servo etc. as set by different researchers depending on their availability and settings in the WEDM set-ups. Among the responses, surface roughness is provided with the maximum priority, followed by MRR, cutting rate, overcut, wire wear rate (WWR), kerf width and angular error. Spark gap, machining time and white layer thickness are also the other responses which are less frequently considered by the researchers. In WEDM processes, different grades of steel and alloys (specially Inconel) are the most commonly machined work materials, followed by titanium and its alloys, MMCs, aluminum and its alloys, and tungsten carbide. For paramet-ric optimization of WEDM processes, different advanced optimization methods along with NSGA are the most popular approaches among the researchers, followed by GA, GRA, utility concept, desirability function, SA, NN and PCA. Out of 28 research papers surveyed, 17 of them dealt with multi-objective optimization, and the remaining papers considered both single and multi-objective optimization of WEDM responses. 1.4 LBM process
In LBM process, a laser beam with high energy density is made to focus on the workpiece surface. The work volume is heated by the absorbed thermal energy and transformed into a molten, vaporized or modified into another chemical state, and is subsequently removed from the machining zone using a high pressure assist gas jet (Meijer, 2004). Thus, melting, vaporization and chemical degradation (chemical bonds are disintegrated causing the material to dissipate) are the three stages of the material removal mechanism in LBM process (Dubey & Yadava, 2008). Ruby (chromium alumina alloy), Nd-glass laser,
048
12162024283236
048121620242832
0
4
8
12
16
20
24
Steels Alloys MMCs Ti and itsalloys
Ceramics
02468
1012
S. Chakraborty et al. / Management Science Letters 9 (2019) 475
Nd-YAG laser etc. are the examples of solid state laser, whereas, Helium-Neon, Argon, CO2 etc. are the gas lasers. The LBM process has several advantages, such as no built-up edge formation, low operating cost, no tool wear, rapid machining ability, capability of generating very tiny holes at difficult entrance angles etc. It is most suitably deployed for welding of non-conductive and refractory materials, and also for drilling, cutting, grooving, scribing, trimming and patterning operations. Its main process parameters include pulse shape, pulse frequency, wave length, duration, laser energy, assist gas type and pressure, focal length and position etc. A list of the reviewed research papers published during 2012-16 on para-metric optimization of LBM processes is provided in Table 4, and Fig. 4 exhibits the parameters and responses selected, work materials machined and optimization techniques deployed in those processes. Pulse frequency, assist gas pressure, cutting speed and pulse width are identified as the most significant process parameters, followed by laser power, lamp current and focus position. There are also some other insignificant parameters, e.g. Y feed rate, workpiece thickness, arc radius etc., which are less frequently chosen by the past researchers. Surface roughness, HAZ, hole taper and kerf taper are the maximally preferred responses for their optimization; although, other responses, like top kerf width, upper deviation, channel width, burr height, depth deviation, MRR, lower deviation are also given due importance. In these processes, some unimportant responses, like burr width, depth of separation line, drag line separa-tion and channel width are also noticed to exist. Among the work materials, aluminium and its different alloys are primarily machined using LBM process, followed by various ceramics (alumina and zirconia) and thermoplastic polymers. Other materials, like different grades of steel, Inconel 718, and titanium and its alloys are also machined using LBM process, but their occurrences are observed to be quite less as compared to other work materials. For parametric optimization of LBM processes, GRA is identified as the most effective method, followed by the application of different advanced optimization techniques. Other optimization methods, like GA, PCA, NSGA etc. are also occasionally adopted by the past re-searchers for the said purpose. Among 18 research papers reviewed, 16 papers considered multi-objective optimization, while one paper dealt with single objective optimization of the responses. There is only a single paper where both single and multi-objective optimization of the responses were considered.
(a)
(b)
(c)
(d)
Fig. 3 Analysis on the parametric optimization of WEDM processes
04812162024
048
1216202428
0
2
4
6
8
10
12
Steels Alloys Ti and itsalloys
MMCs Al and itsalloys
Tungstencarbide
02468
1012
47
6 Tab
le 3
Pa
ram
eter
s, re
spon
ses,
wor
k m
ater
ials
and
opt
imiz
atio
n te
chni
ques
con
side
red
in W
EDM
pro
cess
es
Sl.N o.
Nam
e of
aut
hors
W
ork
mat
eria
l ma-
chin
ed
Sing
le/
Mul
ti-
obje
ctiv
e
Opt
imiz
atio
n to
ol(s
) ad
opte
d Pr
oces
s par
amet
ers
Res
pons
es
1.
Agg
arw
al e
t al.
(201
5)
Inco
nel 7
18
Bot
h D
esira
bilit
y fu
nctio
n W
ire f
eed
rate
, pu
lse-
on t
ime,
pul
se-o
ff t
ime,
w
ire te
nsio
n, g
ap v
olta
ge
Cut
ting
rate
, su
rfac
e ro
ughn
ess
2.
Azh
iri e
t al.
(201
4)
Alu
min
ium
silic
on c
ar-
bide
com
posi
te
Mul
tiple
A
dapt
ive
neur
o-fu
zzy
infe
r-en
ce s
yste
m, g
rey
rela
tiona
l an
alys
is
Puls
e-on
tim
e, p
ulse
-off
tim
e, g
ap v
olta
ge, d
is-
char
ge c
urre
nt, w
ire fe
ed, w
ire te
nsio
n C
uttin
g ve
loci
ty, s
urfa
ce
roug
hnes
s
3.
Boo
path
i and
Siv
akum
ar
(201
3)
Hig
h sp
eed
stee
l M
ultip
le
Epsi
lon
dom
inan
ce
ap-
proa
ch o
f gen
etic
alg
orith
m
Dis
char
ge cu
rren
t, pu
lse-
on ti
me,
pul
se-o
ff ti
me,
ga
p vo
ltage
, air-
mis
t pre
ssur
e M
RR
, sur
face
roug
hnes
s
4.
Bho
opat
hi a
nd S
ivak
umar
(2
016)
H
SS-M
42 to
ol st
eel
Mul
tiple
A
rtific
ial
bee
colo
ny a
lgo-
rithm
Sp
ark
curr
ent,
oxyg
en-m
ist i
nlet
pre
ssur
e, m
ix-
ing
flow
rate
, pul
se-o
n tim
e
MR
R, s
urfa
ce ro
ughn
ess
5.
Cha
lisga
onka
r and
Kum
ar
(201
3)
Tita
nium
B
oth
Util
ity c
once
pt
Wire
feed
, wire
tens
ion,
pul
se-o
n tim
e, p
ulse
-off
tim
e, g
ap v
olta
ge, p
eak
curr
ent
Cut
ting
spee
d,
surf
ace
roug
hnes
s 6.
G
arg
et a
l. (2
012)
Ti
tani
um 6
-2-4
-2
Mul
tiple
N
on-d
omin
ated
sor
ting
ge-
netic
alg
orith
m
Puls
e-on
tim
e, p
ulse
-off
tim
e, g
ap v
olta
ge, w
ire
feed
, wire
tens
ion,
pea
k cu
rren
t C
uttin
g sp
eed,
su
rfac
e ro
ughn
ess
7.
Jaya
dith
ya e
t al.
(201
4)
Inco
nel 8
25
Mul
tiple
U
tility
con
cept
W
ire f
eed,
wor
kpie
ce th
ickn
ess,
puls
e-on
tim
e,
puls
e-of
f tim
e
Cut
ting
spee
d,
surf
ace
roug
hnes
s, di
men
sion
al
devi
atio
n 8.
K
ovač
ević
et a
l. (2
014)
N
itrid
ed st
eel,
AIS
I D
5/D
IN 1
.260
1 st
eel,
alum
iniu
m, t
itani
um a
l-lo
y
Mul
tiple
Ex
haus
tive
itera
tive
sear
ch
proc
edur
e Pu
lse-
on ti
me,
pul
se-o
ff ti
me,
ser
vo f
eed,
pea
k cu
rren
t M
RR
, cut
ting
spee
d, su
r-fa
ce ro
ughn
ess,
over
cut
9.
Kris
han
and
Sam
uel (
2013
) A
ISI D
3 to
ol st
eel
Mul
tiple
N
on-d
omin
ated
sor
ting
ge-
netic
alg
orith
m
Puls
e-of
f tim
e, s
park
gap
, ser
vo fe
ed, r
otat
iona
l sp
eed,
flus
hing
pre
ssur
e M
RR
, sur
face
roug
hnes
s
10.
Kum
ar a
nd A
garw
al (2
012)
H
igh
spee
d st
eel
Bot
h N
on-d
omin
ated
sor
ting
ge-
netic
alg
orith
m
Puls
e pe
ak c
urre
nt,
wire
ten
sion,
wire
fee
d,
flush
ing
pres
sure
, pul
se d
urat
ion,
pul
se-o
ff ti
me
MR
R, s
urfa
ce ro
ughn
ess
11.
Kur
iach
en e
t al.
(201
5)
Tita
nium
allo
y
(Ti-6
Al-4
V)
Mul
tiple
Pa
rticl
e sw
arm
opt
imiz
atio
n G
ap v
olta
ge, c
apac
itanc
e, fe
ed ra
te, w
ire te
nsio
n M
RR
, sur
face
roug
hnes
s
12.
Min
g et
al.
(201
4)
Tung
sten
stee
l (Y
G15
) M
ultip
le
Non
-dom
inat
ed s
ortin
g ge
-ne
tic a
lgor
ithm
W
ire te
nsio
n, p
ulse
-on
time,
pul
se-o
ff ti
me,
wire
sp
eed,
wat
er p
ress
ure,
cut
ting
feed
rate
M
RR
, sur
face
roug
hnes
s
13.
Muk
herje
e et
al.
(201
2)
Inco
nel 6
01
Bot
h G
enet
ic a
lgor
ithm
, pa
rticl
e sw
arm
opt
imiz
atio
n, s
heep
flo
ck a
lgor
ithm
, an
t co
lony
op
timiz
atio
n,
artif
icia
l be
e co
lony
, bio
geog
raph
y-ba
sed
optim
izat
ion
Peak
cur
rent
, pu
lse
dura
tion,
pul
se f
requ
ency
, w
ire sp
eed,
die
lect
ric fl
ow ra
te, d
uty
fact
or, w
ire
tens
ion,
wat
er p
ress
ure,
MR
R, w
ear r
atio
, sur
face
ro
ughn
ess,
kerf
wid
th
14.
Nay
ak a
nd M
ahap
atra
(201
4)
Stai
nles
s ste
el
Mul
tiple
U
tility
con
cept
D
isch
arge
cur
rent
, pa
rt th
ickn
ess,
wire
spe
ed,
wire
tens
ion,
tape
r ang
le, p
ulse
dur
atio
n
Ang
ular
er
ror,
surf
ace
roug
hnes
s, cu
tting
spee
d
S. C
hakr
abor
ty e
t al.
/ M
anag
emen
t Sci
ence
Let
ters
9 (2
019)
47
7
Tab
le 3
Pa
ram
eter
s, re
spon
ses,
wor
k m
ater
ials
and
opt
imiz
atio
n te
chni
ques
con
side
red
in W
EDM
pro
cess
es (C
ontin
ued)
Sl
.N o.
Nam
e of
aut
hors
W
ork
mat
eria
l ma-
chin
ed
Sing
le/
Mul
ti-ob
-je
ctiv
e
Opt
imiz
atio
n to
ol(s
) ad
opte
d Pr
oces
s par
amet
ers
Res
pons
es
15.
Nay
ak a
nd M
ahap
atra
(201
6)
In
cone
l 718
B
oth
Bat
alg
orith
m
Part
thic
knes
s, di
scha
rge
curr
ent,
wire
spe
ed,
wire
tens
ion,
tape
r ang
le, p
ulse
dur
atio
n
Ang
ular
er
ror,
surf
ace
roug
hnes
s, cu
tting
spee
d
16.
Pras
ad a
nd K
rishn
a (2
015)
A
ISI D
3 di
e st
eel
Mul
tiple
H
arm
ony
sear
ch a
lgor
ithm
Puls
e-on
tim
e, p
ulse
-off
tim
e, d
iele
ctric
flo
w
rate
, wire
feed
, wire
tens
ion
K
erf,
wire
wea
r rat
io
17.
Raj
yala
kshm
i and
Ram
aiah
(2
013)
In
cone
l 825
B
oth
Gre
y re
latio
nal a
naly
sis
Puls
e-on
tim
e, p
ulse
-off
tim
e, c
orne
r ser
vo, w
ire
tens
ion,
gap
vol
tage
, ser
vo f
eed,
flu
shin
g pr
es-
sure
of d
iele
ctric
flui
d, w
ire fe
ed ra
te
MR
R,
surf
ace
roug
h-ne
ss, s
park
gap
18.
Rao
et a
l. (2
014)
A
l707
5/Si
Cp
met
al m
a-tri
x co
mpo
site
s B
oth
Hyb
rid g
enet
ic a
lgor
ithm
Pu
lse-
on t
ime,
pul
se-o
ff t
ime,
pea
k cu
rren
t, sp
ark
gap
volta
ge, s
ervo
feed
rate
, flu
shin
g pr
es-
sure
, wire
feed
rate
, wire
tens
ion
MR
R,
surf
ace
roug
h-ne
ss, w
ire w
ear r
ate
19.
Rao
and
Kris
hna
(201
4)
Al7
075/
SiC
p m
etal
ma-
trix
com
posi
tes
Mul
tiple
N
on-d
omin
ated
sor
ting
ge-
netic
Alg
orith
m
Parti
cula
te s
ize,
vol
ume
of S
iCp,
pul
se-o
n tim
e,
puls
e-of
f tim
e, w
ire te
nsio
n M
RR
, su
rfac
e ro
ugh-
ness
, wire
wea
r rat
io
20.
Saha
and
Mon
dal (
2016
) N
ano
stru
ctur
ed h
ard
faci
ng m
ater
ial
Mul
tiple
G
rey
rela
tiona
l an
alys
is,
prin
cipa
l co
mpo
nent
ana
ly-
sis
Serv
o vo
ltage
, w
ire t
ensi
on,
wire
fee
d ra
te,
puls
e-on
tim
e, p
ulse
-off
tim
e
MR
R,
surf
ace
roug
h-ne
ss, m
achi
ning
tim
e
21.
Saha
et a
l. (2
013)
Ti
tani
um c
arbi
de
Bot
h N
euro
-gen
etic
tech
niqu
e Pu
lse o
n-tim
e, p
ulse
off
-tim
e, w
ire fe
ed ra
te, g
ap
volta
ge
Cut
ting
spee
d,
kerf
w
idth
22
. Sh
ahal
i et a
l. (2
012)
D
IN 1
.454
2 st
ainl
ess
stee
l B
oth
Mic
ro-g
enet
ic a
lgor
ithm
Po
wer
, tim
e-of
f tim
e, g
ap v
olta
ge, s
ervo
vol
tage
, in
vers
, num
ber o
f fin
ish
pass
es
Surf
ace
roug
hnes
s, w
hite
laye
r thi
ckne
ss
23.
Shay
an e
t al.
(201
3)
Cem
ente
d tu
ngst
en c
ar-
bide
B
oth
Des
irabi
lity
appr
oach
, par
ti-cl
e sw
arm
opt
imiz
atio
n Pu
lse-
on ti
me,
pul
se-o
ff ti
me,
dis
char
ge c
urre
nt,
wire
te
nsio
n,
gap
volta
ge
Cut
ting
velo
city
, sur
face
ro
ughn
ess,
over
size
24.
Som
ashe
kar e
t al.
(201
2)
Alu
min
ium
M
ultip
le
Sim
ulat
ed a
nnea
ling
Gap
vol
tage
, cap
acita
nce,
feed
rate
M
RR
, su
rfac
e ro
ugh-
ness
, ove
rcut
25
. Su
brah
man
yam
and
Sar
car
(201
3)
Die
stee
l M
ultip
le
Gre
y re
latio
nal a
naly
sis
Puls
e-on
tim
e, p
ulse
-off
tim
e, p
eak
curr
ent,
gap
volta
ge, w
ire te
nsio
n, w
ire fe
ed ra
te, s
ervo
feed
, flu
shin
g pr
essu
re
MR
R, s
urfa
ce ro
ughn
ess
26.
Var
un e
t al.
(201
6)
M
onel
400
B
oth
Des
irabi
lity
func
tion,
par
ti-cl
e sw
arm
opt
imiz
atio
n Pu
lse-
on ti
me,
pul
se-o
ff ti
me,
pea
k cu
rren
t, w
ire
feed
M
RR
, sur
face
roug
hnes
s
27.
Yan
g et
al.
(201
2)
Tung
sten
M
ultip
le
Sim
ulat
ed a
nnea
ling
Pu
lse-
on ti
me,
pul
se-o
ff ti
me,
arc
-off
tim
e, w
ire
tens
ion,
wat
er p
ress
ure,
serv
o vo
ltage
, wire
feed
ra
te
MR
R,
surf
ace
roug
h-ne
ss, c
orne
r dev
iatio
n
28.
Zhan
g et
al.
(201
4)
Hig
h ca
rbon
, hig
h
chro
miu
m a
lloy
tool
st
eel
Mul
tiple
N
on-d
omin
ated
sor
ting
ge-
netic
alg
orith
m
Puls
e-on
tim
e, p
ulse
-off
tim
e, p
ulse
curr
ent,
wire
tra
velli
ng sp
eed,
trac
king
coe
ffic
ient
M
RR
, sur
face
roug
hnes
s
478
1.5. USM process
In USM process, a tool having the appropriate shape geometry oscillates over the workpiece at an ultra-sonic frequency of 19~25 kHz and amplitude of 15-50 μm. Between the tool and the workpiece, the machining zone is deluged with abrasive particles (Al2O3, SiC, B4C, diamond etc.) mixed with water to form of a water-based slurry. When the tool oscillates over the workpiece, the abrasive particles make indentations to remove material from the workpiece. Crack initiation, propagation and brittle fracture are the three phases causing removal of material in USM process. It can be effectively employed for gener-ating square, round, irregular shaped holes and surface impressions on hard and brittle materials, like glass, ceramics, stones, carbides, silicon nitride, nickel/titanium alloys etc. (Thoe et al., 1998). Amplitude and frequency of vibration, feed force and pressure, abrasive size and material, contact area of the tool, volume concentration of abrasive in slurry etc. are the different parameters influencing the machining performance of USM process.
(a)
(b)
(c)
(d)
Fig. 4 Analysis on the parametric optimization of LBM processes
Table 5 presents a list of the reviewed research papers published during 2012-16 on parametric optimi-zation of USM processes, and Fig. 5 provides an idea on the process parameters and responses, work materials and optimization techniques selected in those considered processes. It can be easily revealed that grit size and power rating are the two important control parameters mostly preferred by the past researchers, followed by the type of tool material, slurry concentration and tool feed rate. Tool profile, and type and thickness of the work material are some of the least preferred machining parameters for USM process. In this process, maximum importance is allocated to the optimization of surface roughness and MRR, followed by tool wear rate (TWR) and overcut. Titanium is identified as the most widely machined work material in USM process. It is also employed for machining of different ceramics (mainly zirconia) and MMCs. Among the optimization tools, GRA, utility concept and different advanced opti-mization techniques are mainly utilized for parametric optimization of USM processes. Other tools, such as GA, NN and Taguchi loss function are also applied for the same purpose. There are applications of weighted signal-to-noise ratio and multi-response signal-to-noise ratio for simultaneous optimization of USM responses. From Table 5, it can be noticed that four research papers dealt with multi-objective optimization, while three papers considered both single as well as multi-objective optimization of the responses for USM process.
02468
10121416
012345678
0
1
2
3
4
5
6
Al and itsalloys
CeramicsPolymers Steels Inconel718
Ti and itsalloys
0
2
4
6
8
10
GRA Adv. opt.methods
GA PCA Fuzzylogic
NSGA
S. C
hakr
abor
ty e
t al.
/ M
anag
emen
t Sci
ence
Let
ters
9 (2
019)
47
9
Tabl
e 4
Pa
ram
eter
s, re
spon
ses,
wor
k m
ater
ials
and
opt
imiz
atio
n te
chni
ques
con
side
red
in L
BM
pro
cess
es
Sl.
No.
N
ame
of a
utho
rs
Wor
k m
ater
ial m
a-ch
ined
Si
ngle
/Mul
ti-ob
-je
ctiv
e O
ptim
izat
ion
tool
(s) a
dopt
ed
Proc
ess p
aram
eter
s R
espo
nses
1.
Ach
erje
e et
al.
(201
4)
Poly
met
hyl-
met
h-ac
ryla
te
Mul
tiple
G
rey
rela
tiona
l ana
lysi
s La
mp
curr
ent,
scan
ning
spe
ed,
puls
e fr
eque
ncy,
pul
se w
idth
C
hann
el d
epth
, bur
r hei
ght,
burr
wid
th
2.
Gan
guly
et a
l. (2
012)
Zi
rcon
ium
oxi
de
Mul
tiple
G
rey
rela
tiona
l ana
lysi
s La
mp
curr
ent,
puls
e fr
eque
ncy,
ai
r pr
essu
re, p
ulse
wid
th
Hol
e ta
per,
heat
aff
ecte
d zo
ne
3.
Kib
ria e
t al.
(201
3)
Alu
min
a ce
ram
ics
Mul
tiple
G
rey
rela
tiona
l ana
lysi
s A
vera
ge p
ower
, pul
se fr
eque
ncy,
rota
-tio
nal s
peed
, air
pres
sure
, Y fe
ed ra
te
Surf
ace
roug
hnes
s, de
pth
devi
atio
n
4.
Kua
r et a
l. (2
012)
A
lum
iniu
m o
xide
M
ultip
le
Gre
y re
latio
nal a
naly
sis
Lam
p cu
rren
t, pu
lse
freq
uenc
y,
air
pres
sure
, pul
se w
idth
H
ole
tape
r, he
at a
ffec
ted
zone
5.
Mad
ic e
t al.
(201
4)
Stai
nles
s ste
el
Mul
tiple
G
rey
rela
tiona
l ana
lysi
s La
ser
pow
er, c
uttin
g sp
eed,
ass
ist g
as
pres
sure
, foc
us p
ositi
on
Bur
r hei
ght,
drag
line
sepa
ratio
n, d
epth
of
sepa
ratio
n lin
e 6.
M
adic
et a
l. (2
015)
St
ainl
ess s
teel
M
ultip
le
Cuc
koo
sear
ch a
lgor
ithm
La
ser
pow
er, c
uttin
g sp
eed,
ass
ist g
as
pres
sure
, foc
us p
ositi
on
Surf
ace
roug
hnes
s, he
at a
ffec
ted
zone
, to
p ke
rf w
idth
7.
M
ishr
a an
d Y
adav
a (2
013a
) In
cone
l 718
M
ultip
le
Gre
y re
latio
nal
anal
ysis
, pr
inci
pal
com
pone
nt a
naly
sis
Puls
e w
idth
, pu
lse
freq
uenc
y, p
eak
pow
er, w
orkp
iece
thic
knes
s M
RR
, hol
e ta
per,
heat
eff
ecte
d zo
ne
8.
Mis
hra
and
Yad
ava
(201
3b)
Alu
min
ium
M
ultip
le
Gre
y re
latio
nal
anal
ysis
, pr
inci
pal
com
pone
nt a
naly
sis
Ave
rage
pow
er, p
ulse
wid
th, p
ulse
fre-
quen
cy, n
ozzl
e st
and-
off d
ista
nce
MR
R, h
ole
tape
r, he
at a
ffec
ted
zone
9.
Muk
herje
e et
al.
(201
3)
Zirc
oniu
m o
xide
, al-
umin
ium
oxi
de
Bot
h A
rtific
ial b
ee c
olon
y al
gorit
hm
Lam
p cu
rren
t, pu
lse
freq
uenc
y,
air
pres
sure
, pul
se w
idth
, cut
ting
spee
d H
eat a
ffec
ted
zone
, tap
er, u
pper
dev
ia-
tion,
low
er d
evia
tion,
dep
th d
evia
tion
10.
Pand
ey a
nd D
ubey
(201
2a)
Tita
nium
allo
y M
ultip
le
Gen
etic
alg
orith
m
Gas
pre
ssur
e, p
ulse
wid
th,
puls
e fr
e-qu
ency
, cut
ting
spee
d Su
rfac
e ro
ughn
ess,
kerf
tape
r
11.
Pand
ey a
nd D
ubey
(201
2b)
Dur
alum
in
Mul
tiple
Ta
guch
i-bas
ed fu
zzy
logi
c G
as p
ress
ure,
pul
se w
idth
, pu
lse
fre-
quen
cy, c
uttin
g sp
eed
Ker
f wid
th, t
op k
erf d
evia
tion,
bot
tom
ke
rf d
evia
tion
12
. Pa
ndey
and
Dub
ey (2
013)
D
ural
umin
M
ultip
le
Gre
y-fu
zzy
logi
c
Gas
pre
ssur
e, p
ulse
wid
th,
puls
e fr
e-qu
ency
, cut
ting
spee
d Su
rfac
e ro
ughn
ess,
kerf
ta
per,
kerf
w
idth
13
. Pa
war
and
Ray
ete
(201
4)
Stai
nles
s ste
el
Mul
tiple
N
on-d
omin
ated
sorti
ng g
enet
ic a
l-go
rithm
G
as
pres
sure
, cu
tting
sp
eed,
la
ser
pow
er, p
ulse
freq
uenc
y K
erf w
idth
, tap
er a
ngle
, sur
face
roug
h-ne
ss
14.
Phip
on a
nd P
radh
an (2
012a
) A
lum
iniu
m a
lloy
Sing
le
Gen
etic
alg
orith
m
Oxy
gen
pres
sure
, pu
lse
wid
th,
puls
e fr
eque
ncy,
cut
ting
spee
d K
erf t
aper
, sur
face
roug
hnes
s
15.
Shar
ma
and
Yad
ava
(201
2)
Alu
min
ium
allo
y M
ultip
le
Gre
y re
latio
nal a
naly
sis
Oxy
gen
pres
sure
, pu
lse
wid
th,
puls
e fr
eque
ncy,
cut
ting
spee
d Su
rfac
e ro
ughn
ess,
kerf
tape
r
16.
Shar
ma
and
Yad
ava
(201
3)
Alu
min
ium
allo
y M
ultip
le
Gre
y re
latio
nal a
naly
sis
Arc
ra
dius
, ox
ygen
pr
essu
re,
puls
e w
idth
, pul
se fr
eque
ncy,
cut
ting
spee
d K
erf d
evia
tion,
ker
f tap
er
17.
Tam
rin e
t al.
(201
5)
Poly
met
hyl m
etha
c-ry
late
, pol
ycar
-bo
nate
, p
olyp
ropy
lene
Mul
tiple
G
rey
rela
tiona
l ana
lysi
s Po
wer
, cut
ting
spee
d, c
ompr
esse
d ai
r pr
essu
re
Hea
t aff
ecte
d zo
ne, d
iam
eter
of t
he c
ut
18.
Texi
dor e
t al.
(201
3)
AIS
I H13
tool
stee
l M
ultip
le
Parti
cle
swar
m o
ptim
izat
ion
Scan
ning
spe
ed, p
ulse
inte
nsity
, pul
se
freq
uenc
y C
hann
el d
epth
, cha
nnel
wid
th, s
urfa
ce
roug
hnes
s
480
(a)
(b)
(c) (d)
Fig. 5 Analysis on the parametric optimization of USM processes
1.6 HM processes
Technological improvement of NTM processes can be efficiently attained when different machining ac-tions or phases are combined together for material removal. This combination of individual processes leads to the subsequent development of a HM process where the combined advantages of the constituent processes can be achieved while avoiding or reducing some of their adverse effects when they are applied individually (El-Hofy, 2005). The performance characteristics of a HM process are expected to be better than those of the single-phase processes with respect to productivity, accuracy and surface quality (Sundaram, 2014). In AJWM process, the mechanical energy of water and abrasive particles is utilized to remove material from the workpiece. In this process, abrasive particles, like sand (SiO2), glass beads etc. are mixed with the water jet to enhance its machining capability by many folds (Janković et al., 2012). The ECDM is also a hybrid NTM process combining the features of ECM and EDM processes. It consists of a tool (cathode), an auxiliary electrode (anode) and a workpiece, which are isolated by a gap filled with electrolyte, and a pulsed DC is applied between them. This causes generation of electrical discharges between the tool and electrolyte where the workpiece is placed, thus attaining both electro-chemical dissolution and electro discharge erosion of the workpiece (Mediliyegedara et al., 2005). The TW-ECSM process combines the material removal mechanisms of ECM and WEDM processes (Malik and Manna, 2017). It can be efficaciously employed for machining of hard-to-machine non-conductive materials that cannot be machined by the other NTM processes, like EDM, ECM, WEDM etc. Electrical discharge diamond grinding (EDDG) is a hybridized NTM process, consisting of EDM with rotary disc electrode and grinding using diamond abrasives (Yadav et al., 2012). During EDDG, the electrically non-conducting reinforcements that hinder the generation of sparks can be removed by the abrasion action of diamond abrasives. On the other hand, the wheel loading and clogging problems can be avoided due to electrical sparks, and dressing of diamond wheel can also take place. In magnetic abrasive finishing (MAF) process, the workpiece is kept between the two poles of a magnet, and the machining gap between the workpiece and the magnet is flooded with magnetic abrasive particles. A magnetic abrasive flexible brush is thus formed, which acts as a multipoint cutting tool, due to the effect of magnetic field in the
01234567
0123456
0
1
2
3
4
Titanium Ceramics MMCs
0
1
2
3
Others GRA Adv.opt.
methods
Utilityconcept
GA Lossfunction
NN
S. Chakraborty et al. / Management Science Letters 9 (2019) 481
machining gap (Kumar et al., 2013). The ECG combines electrochemical dissolution and mechanical grinding processes (Goswami et al., 2009). The workpiece is connected to the positive electrode and an electrically conductive grinding wheel is made as the negative electrode. As the electric current flows between the workpiece and wheel through the electrolyte, electrochemical dissolution takes places caus-ing material to be removed from the workpiece. Mechanical abrasion is also responsible for material removal to some extent.
Table 5 Parameters, responses, work materials and optimization techniques considered in USM processes
Sl.No. Name of author(s)
Work material
machined
Single/ Multi-
objective
Optimization tool(s) adopted
Process parameters Responses
1. Chakravorty et al. (2013)
Cobalt, tungsten carbide
Multiple
Weighted signal-to-noise ratio, utility theory, grey relational analysis, multi-response signal-to-noise ratio
Tool material, abrasive slurry material, slurry concentration, grit size of slurry material, power rating
MRR, tool wear rate, surface roughness
2. Cheema et al. (2013)
Titanium (ASTM Gr 2),
Titanium (ASTM Gr 5)
Both Utility concept Workpiece material, abrasive grit size, abrasive slurry concen-tration, power rating, tool mate-rial
Surface roughness, tool wear ratio, hole oversize
3. Das et al. (2013)
Zirconia Multiple Genetic algorithm Abrasive grit size, abrasive slurry concentration, power rating, tool feed rate
MRR, surface roughness
4. Goswami and Chakraborty
(2015)
Zirconia Both Gravitational search algo-rithm, fireworks algorithm
Grit size, slurry concentration, power rating, tool feed rate
MRR, surface roughness
5. Kataria et al. (2016)
WC-Co composite
Multiple Grey relational analysis Cobalt content, workpiece thickness, tool profile, grit size, power rating
MRR, tool wear rate
6. Kumar (2014) Titanium (ASTM Gr 1)
Both Taguchi loss function Tool material, abrasive type, grit size, power rating
Surface roughness, micro-hardness
7.
Teimouri et al. (2015)
Titanium (ASTM Gr 1)
Multiple Adaptive neuro-fuzzy in-ference system, imperialist competitive algorithm
Tool material, grit size, power rating
MRR, tool wear rate, surface roughness
In Table 6, a list of research papers published during 2012-16 on parametric optimization of HM pro-cesses (i.e. AJWM, ECDM, TW-ECSM, EDDG, MAF and ECG) is provided. The responses set in these six HM processes, work materials machined and optimization techniques employed for deriving the best parametric mix are shown in Fig. 6. As the control parameters in the considered HM processes are widely varying and entirely depend on the type of the machining set-up employed, it is not at all a wise decision to make an exhaustive list of those parameters. It can be perceived from Fig. 6 that among the responses, MRR and surface roughness are allotted with the maximum importance, followed by kerf width, HAZ, overcut, TWR, current density and depth of cut. In these processes, glass, MMCs, different grades of steel, and aluminium and its alloys are mostly machined for generation of different shape features for industrial use, followed by ceramics, brass, Inconel, tungsten carbide and silicon. With respect to the techniques adopted for parametric optimization of HM processes, GRA and advanced optimization meth-ods become quite popular among the researchers. Other techniques, like NSGA. GA, NN, loss function, PCA, utility concept and desirability function are also applied for the same purpose. Among 30 papers surveyed on HM processes, 18 of them considered multi-objective optimization, while only four dealt with both single as well as multi-objective optimization of the responses. There are also eight research papers where the authors optimized only a single response.
2. Discussions When all the reviewed papers are classified based on their year of publication, as depicted in Fig. 7, it can be clearly revealed that the earlier researchers maximally considered EDM process for its parametric optimization, followed by HM and WEDM processes. It is also quite interesting to notice that in this direction, maximum research works were mainly carried out in 2012 and 2014 for almost all the consid-ered NTM processes. When all the responses, as considered by the past researchers for deriving their
482
optimal values, are analyzed in Fig. 8, it can be observed that surface roughness has the top priority, followed by MRR. Other responses, like TWR, overcut, cutting rate, HAZ, taper, kerf width etc. have also moderate importance. There are several less significant responses, like cylindricity, white layer thickness, surface crack density, channel depth, burr height, depth deviation, lower deviation, current density etc. which are occasionally considered by the past researchers in order to satisfy varying end product requirements. It is a well known fact that NTM processes are mainly employed for generation of complex shape geometries on varying hard-to-machine materials which cannot be machined by the con-ventional material removal methods. From the reviewed papers, it can be clearly visualized that these processes were mainly utilized for machining of different grades of steel, followed by MMCs and various metal alloys to meet their high demands for diverse industrial applications. A list of different work ma-terials machined by the considered NTM processes is graphically presented in Fig. 9. The application of various tools and techniques for parametric optimization of the considered NTM processes is exhibited in Fig. 10. It is quite interesting to observe that among all these applied methods, GRA supersedes the others due to its mathematical simplicity, comprehensiveness and capability for performing multi-objec-tive optimization of the responses quite easily. But when GA, NSGA and SA are coupled together with the other advanced optimization methods, they become the most popular techniques due to their ability to solve both single and multi-objective optimization problems. These techniques are also capable of finding out the global optimal solutions for parametric optimization problems. Among the employed advanced optimization methods, particle swarm optimization algorithm is mostly preferred by the re-searchers, followed by artificial bee colony optimization and cuckoo optimization techniques, as shown in Fig. 11. Biogeography-based optimization, teaching learning-based optimization, ant colony optimi-zation, firefly algorithm, bat algorithm, sheep flock algorithm, harmony search algorithm, gravitational search algorithm and fireworks algorithm are the other techniques applied for parametric optimization of the considered NTM processes according to their preference. It is observed that amongst 133 research papers surveyed, in only 10 papers (7.52%), a single response was optimized; in 96 papers (72.18%), multiple responses were optimized simultaneously; and 27 papers (20.30%) dealt with both single as well as multi-objective optimization of the responses.
(a)
(b)
(c)
Fig. 6 Analysis on the parametric optimization of HM processes
05
10152025
0123456
012345678
S. C
hakr
abor
ty e
t al.
/ M