Content uploaded by Murlidharan B
Author content
All content in this area was uploaded by Murlidharan B on Dec 01, 2016
Content may be subject to copyright.
Full Terms & Conditions of access and use can be found at
http://www.tandfonline.com/action/journalInformation?journalCode=lmmp20
Download by: [Muralidharan B] Date: 16 October 2016, At: 00:05
Materials and Manufacturing Processes
ISSN: 1042-6914 (Print) 1532-2475 (Online) Journal homepage: http://www.tandfonline.com/loi/lmmp20
Single-Spark Analysis of Electro-Discharge
Deposition Process
B. Muralidharan, H. Chelladurai, Praveen Singh & Mukesh Kumar Roy
To cite this article: B. Muralidharan, H. Chelladurai, Praveen Singh & Mukesh Kumar Roy (2016)
Single-Spark Analysis of Electro-Discharge Deposition Process, Materials and Manufacturing
Processes, 31:14, 1853-1864, DOI: 10.1080/10426914.2015.1127936
To link to this article: http://dx.doi.org/10.1080/10426914.2015.1127936
Accepted author version posted online: 29
Dec 2015.
Published online: 29 Dec 2015.
Submit your article to this journal
Article views: 109
View related articles
View Crossmark data
Single-Spark Analysis of Electro-Discharge Deposition Process
B. Muralidharan
1
, H. Chelladurai
1
, Praveen Singh
1
, and Mukesh Kumar roy
2
1
Discipline of Mechanical Engineering, PDPM IIITDM Jabalpur, Jabalpur, MP, India
2
Discipline of Natural Science, PDPM IIITDM Jabalpur, Jabalpur, MP, India
Electro-discharge deposition (EDD) process is one of the newly developing, nonlithographic additive manufacturing methods to fabricate
3D microproducts. The advantage of the EDD process is that 3D microcomponents can be manufactured from the materials which are
electrically conductive in nature. In this paper, a comprehensive attempt to develop a thermophysical model for single discharge deposition
process has been reported. Experiments are designed and conducted by central composite design (CCD) and the obtained results are
compared with simulation results. Parametric analysis is carried out to study the effects of EDD process parameters like current, pulse-on
time, and duty factor on height, width, and weight of deposition. The objective of this research work is to simulate and predict the material
deposition mechanism of EDD process in a single pulse. Simulation results are compared with experimental results and it is found that the
developed model can predict results which are closer to experimental results. It is concluded from the results that the thermophysical model
can be further modified to establish EDD process with optimum process parameters.
Keywords Analysis; Deposition; Height; Modeling; Process; Simulation.
INTRODUCTION
The market survey shows that the demand for micro-
products are increasingly high due to their applications
in wide variety of fields like automobile, biomedical,
microelectromechanical systems (MEMS), and aero-
space industries [1,2]. In order to integrate the geometri-
cal size and capability (i.e., microfeatures with more
integrated functionalities) of the product, various micro-
and nanomanufacturing technologies are rapidly devel-
oping. Cui [3] discussed some of the research works
developed in the area of micro- and nanofabrication
technologies with their advantages, applications, and
limitations. Both additive and subtractive processes
can be employed in the area of micro- and nanofabrica-
tion technologies. With the aim of green manufacturing,
additive processes are preferred to subtractive processes
in micro level. Also, some limitations faced by subtrac-
tive processes are difficulty in building machining setup,
microtool handling, quality control, and high invest-
ment cost, among others. The role of different micro-
and nanomanufacturing technologies (both lithographic
and nonlithographic technologies), their applications,
and process capabilities are discussed by Brousseau
et al. [4]. Dimov et al. [5] articulated the limitations of
lithographic technologies that they can be helpful only
up to 2D and 2.5D features with limited material usage.
It is suggested that processes like electrophysical and
chemical process have been evolved as one of the suitable
alternative methods for manufacturing 3D micropro-
ducts with improved process capability when compared
to lithographic techniques [6]. Electro-discharge depo-
sition (EDD) process is a newly developing additive
manufacturing process based on Electrical discharge
machining (EDM) principle, wherein reverse polarity is
connected to build 3D microfeatures. Earlier, some of
the researchers [7,8] observed the transfer of material
from tool to workpiece during reverse EDM. Chen
et al. [9] performed electrical discharge coating (EDC)
process on 6061-T6 aluminum (Al) alloy by using
titanium (Ti)-sintered electrodes in wet and dry mode,
where they found the feasibility of surface modification
process by EDC. Huang et al. [10] surface modified the
shape memory alloys (SMA) Ti
50
Ni
50
with nitrogen gas
as dielectric medium and pure titanium as tool electrode.
Gill and Kumar [11] observed the transfer of material
from tool to workpiece in a surface-alloying process on
hot die steel H11 by using copper–tungsten as tool elec-
trode. Simao et al. [12] reported a research work in which
WC=Co is used as the electrode, resulted in the depo-
sition of a uniform alloyed layer over the workpiece using
EDM in order to achieve enhanced surface alloying. Soni
and Chakraverti [13] reported that the migration
phenomenon in EDM has been carried out intentionally
to enhance the surface properties of the workpiece
material. This process of material transfer can be further
developed to manufacture 3D free form microfeatures
using conducting electrodes. Wang et al. [14] deposited
microbrass cylinders with 200 mm in diameter and Chi
et al. [15] fabricated 3D brass spiral structure in dry mode
EDM. Jain et al. [16] have reported the fabrication of
microparts by using EDM and parametric analysis has
been done in order to find the significant contributing
Received November 6, 2014; Accepted November 20, 2015
Address correspondence to B. Muralidharan, Discipline of
Mechanical Engineering, PDPM IIITDM Jabalpur, Jabalpur 482005,
MP, India; E-mail: b.murlidharan@gmail.com,b.murlidharan@
iiitdmj.ac.in
Color versions of one or more of the figures in the article can be
found online at www.tandfonline.com/lmmp.
Materials and Manufacturing Processes, 31: 1853–1864, 2016
Copyright #Taylor & Francis Group, LLC
ISSN: 1042-6914 print=1532-2475 online
DOI: 10.1080/10426914.2015.1127936
1853
factors in the EDD process. Some of the researchers [17,
18] discussed about various process parameters of EDD
process where they attempted to produce 3D microfea-
tures. Many researchers developed several thermophysi-
cal models for EDM by considering suitable
assumptions to develop EDM variants with optimized
process parameters and also applying different
approaches like analytical, mathematical, and numerical
methods. Some of the research works reported in litera-
ture, with assumptions like, spark plasma to be cylindri-
cal shape [19], uniform disc heat source [20], pointed heat
source [21], expanding circle heat source model [22], con-
stant heat flux radius [24,25], and heat flux radius
expanding in nature [20], which have given a basic idea
to develop a simulation model for an EDD process.
The objective of the present research work focuses the
development of thermophysical model for EDD process
to understand the basic phenomenon. In order to vali-
date the developed thermophysical model, experiments
are conducted and the experimental results are com-
pared with the simulation results. Instead of doing some
random confirming experiments, the experiments are
designed and conducted by the concepts of design of
experiments (DOE), i.e., adopting response surface
methodology (RSM). The experimental work serves
two purposes, first to validate the simulation model,
secondly to perform parametric analysis.
MATERIALS AND METHODS
In this paper, initially the model is developed for sin-
gle pulse discharge and the corresponding results are
analyzed.
In an EDM process when the tool (anode) and work-
piece (cathode) are brought in closer proximity, there
will be a spark discharge by the breakdown mechanism.
The electrons will move from the cathode surface to
anode with high velocity and the ions will drift toward
cathode surface due to the applied potential difference.
A plasma channel is established which liberates high
heat energy. Due to heat energy released, some of the
material from tool and workpiece is melted and vapor-
ized. In the EDD process, copper wire of 3–5 mmin
diameter is used as the tool and polished stainless steel
plate is used as the workpiece (substrate). In a single
pulse, it is observed that the tool material is melted
and deposited over the substrate. Because of complex
nature of discharge mechanism and deposition phenom-
enon, the thermophysical model is developed using
COMSOL (4.3) multi-physics software to understand
the deposition process.
Some of the assumptions considered for developing
the model are the following:
1. Analysis is done for single pulse discharge.
2. One spark is generated for one pulse.
3. Workpiece and tool are considered to be homo-
geneous and isotropic.
4. Gaussian distribution of heat flux is assumed as heat
source [23,26].
5. Radius of plasma depends on current and pulse-on
time. The spark channel generated is assumed as
cylindrical in nature [27].
6. Modeling has been done by considering axisym-
metric domain.
7. Heat transfer between the electrodes is considered to
be through conduction.
8. Heat source is supplied for known time interval.
The EDM process is considered to be a problem of
nonlinear transient thermal analysis. Thermal diffusion
heat equation is taken as governing equation for the
heating process due to single-spark discharge from
literatures [26] and [28] as shown in Eq. (1).
1
r
@
@rkr @T
@r
þ@
@zk@T
@z
¼qC@T
@tð1Þ
where T is the temperature (K), K is the thermal con-
ductivity of electrodes (W=m.K), qis the density of
electrode materials (kg=m
3
), C is the specific heat
capacity of solid material (J=kg.K), and r and z are
the coordinates axes. Figure 1shows the developed
thermal model with boundary conditions. It is shown
that for a given plasma radius, the applied heat flux is
Gaussian in nature. In this work, the EDD process is
carried in dry mode condition (air as dielectric
medium). The heat lost to air is modeled as convective
boundary conditions on the surface b
1
. Boundaries
b
2
and b
3
are assumed to be far away from the heat
source and heat transfer across these boundaries is con-
sidered to be zero. The boundary b
4
is the axis of sym-
metry and heat transfer in this zone (across this
boundary) is assumed to be zero.
FIGURE 1.—Axisymmetric thermal model for EDD.
1854 B. MURALIDHARAN ET AL.
where q(r) is the heat flux applied at anode surface
(W=m
2
), h
c
is the convective heat transfer coefficient
of dielectric medium (W=m
2
K), R is the spark
radius(m), and T
0
is the room temperature (K), r is
the coordinate axis; and n gives direction normal to
the boundary.
The growth and size of the plasma depends on
operating input parameters. Different methods are
proposed by researchers to calculate the plasma radius.
Due to the complexity in spark discharge process it is
difficult to measure the plasma radius by experimental
methods. Erden [29] proposed an empirical relation-
ship for spark radius, where the spark radius depends
on spark time. The anode erosion model developed
by Patel et al. [22] calculated spark radius as multipli-
cation of leading constant and time exponent. Spark
radius based on boiling point is suggested by Pandey
and Jilani [30]. In the present model, the equation
proposed by Ikai and Hashigushi [27] is used. They cal-
culated the spark radius (R) as a function of discharge
current(I)andpulse-ontime(T
on
). The equivalent
heat input radius used in the present model is given
by Eq. (2):
R¼2:04 e3
I0:43 Ton0:44
ð2Þ
In the present EDD model, Gaussian distribution of
heat flux proposed by Ikai and Hashigushi [27] is used,
which is shown in Eq. (3).
qrðÞ¼qoexp 4:5r
R
2
ð3Þ
q(r) is the heat flux at radius r (W=m
2
) and q
o
is the
maximum heat flux (W=m
2
) at radius r ¼0.
If it is assumed that the total power of every pulse is
consumed by only single spark then the Eq. (3) can be
written as follows:
qðrÞ¼4:45 FAVI
pR2exp 4:5r
R
2
ð4Þ
where F
A
is the fraction of heat input to the anode
surface, V is the discharge voltage (V), and I is the dis-
charge current (A). The percentage of heat supplied to
the electrodes is also one of the important parameters
to be taken into consideration during EDM process.
Here the steel workpiece gets less amount of heat than
copper tool which has higher thermal conductivity and
lower thermal diffusivity. Many researchers [21,22]
assumed that a constant fraction of total power is
transferred to the electrodes. DiBitonto et al. [21]
reported that fraction of heat supplied to the anode
was 0.08. It is observed by [31,32] that the heat distri-
bution to anode was 0.48 and 0.14, respectively. In the
present research paper, different proposed values of F
A
are analyzed. The F
A
value of 0.22 is found to be in a
good match between simulation, experimental, and
predicted observations. Therefore, F
A
value equivalent
to 0.22 is used in the present model development.
From the thermal model, the temperature range above
the melting point (1356 K) of copper is assumed to be
melted. From the temperature distribution, the volume
of melted material is calculated. This melted volume
consists of materials both in liquid and vapor state.
After melting process, transport phenomenon and drift
diffusion plays a major role in deposition mechanism.
From the literature [33], it is found that the temperature
of EDM plasma ranges from 5500 to 10,000 K and the
plasma pressure reaches to about 4.5–10 bar. Therefore,
it is realized that at this elevated temperature and press-
ure, the removed material from the anode will be in the
form of mixture state, i.e., both in liquid and gaseous
states. At this state, the mixture consists of species like
gaseous species, surface species, and bulk species. Hence,
in the present work deposition model is developed by
assuming the theory of heavy species transport [34].
The equation that describes the transport process is
given by Maxwell–Stefan equations as stated in [35].
The species Cu and O
2
are observed as major species
from the EDAX compositional analysis reported in
[17], and the same species are used for the modeling of
species transport phenomenon.
The modeling flowchart for EDD process is illustrated
in Fig. 2. This process flow diagram gives a layout of
interaction between melting and deposition process
models during single spark. Multicomponent diffusion
equation for a species reaction, which has k ¼1 to Q spe-
cies and j ¼1 to N reactions is given by Eq. (5):
q@
@twk
ðÞþqu:rðÞwk
ðÞ¼r:jkþRkð5Þ
where j
k
is the diffusive flux vector, R
k
is the rate
expression for species k (kg=(m
3
s)), u is the mass aver-
aged fluid velocity vector (m=s), qis the density of the
mixture (kg=m
3
), and w
k
is the mass fraction of the k
th
species.
The diffusive flux vector is defined as in Eq. (6):
jk¼qwkVkð6Þ
where V
k
is the multicomponent diffusion velocity for
species k.
k@T
@z
¼
qrðÞ if r R
hcTT0
ðÞif r >R
0 pulse off time
8
<
:
on boundary b1and @T
@n¼0 on boundaries b2;b3;and b4
SINGLE-SPARK ANALYSIS OF EDD PROCESS 1855
Multicomponent diffusion velocity for a species is
defined by Eq.(7):
Vk¼XQ
j¼1DkjdkDT
k
qwk
rln T ð7Þ
where Dk
j
is the multicomponent Maxwell–Stefan
diffusivity (m
2
=s), T is the gas temperature (K), DT
kis
the thermal diffusion coefficient (kg=(ms)), and d
k
is
the diffusion driving force (1=m). Equation (8)deter-
mines the reaction rate of the surface species. The
amount of melted and vaporized copper material
deposited over the substrate is determined by surface
reaction rate. The surface reaction rate for reaction q
i
is
given by
qi¼kf;iYK
k¼1cvk;i
kð8Þ
The rate constant k
f,i
is given by Eq. (9):
kf;i¼ci
1ci
2
1
Ctot
ðÞ
m
1
4ffiffiffiffiffiffiffiffiffiffi
8RT
pMn
sð9Þ
where c
k
is the molar concentration of species k which
may be a volumetric or surface species, m is the reaction
order minus 1, T is the surface temperature (K), R is the
gas constant, M
n
is the mean molecular weight of the gas
mixture (kg=mol), and c
i
is the dimensionless sticking
coefficient.
A species occupying a surface phase is called surface
species. A surface phase can be defined as number of
atoms stick to build species thickness. Each surface spe-
cies occupies surface sites, which is considered to be a
position on the surface where a species can stay. The
number of moles of surface sites per unit area is called
surface site concentration, which is considered as a
property of the material surface and it is assumed to
remain constant. Equation (10) is solved for surface site
concentration for a surface species:
dC
dt ¼XN
i¼1qiDrið10Þ
where Cis the surface site concentration (mol=m
2
), q
i
is
the reaction rate for reaction i (mol=m
2
), and Dr
i
is the
change in site occupancy number for reaction i (dimen-
sionless). The site occupancy number specifies that a big
molecule may cover more than one site on a surface.
Equation (11) specifies the site fraction number. The
composition of surface phases can be specified in terms
of site fractions.
dZk
dt ¼R
Ctot
ð11Þ
where C
tot
is the total surface site concentration
(mol=m
2
), Z
k
is the site fraction (dimensionless), and R
is the surface rate expression (mol=m
2
). Equation (12)
determines the height of deposited material over the sur-
face. For the bulk surface species, the following equation
is solved for the deposition height:
dh
dt ¼
RMw
qð12Þ
where H is the total growth height (m), M
w
is the
molecular weight (kg=mol), and qis the density of the
bulk species (kg=m
3
).
EXPERIMENTAL SETUP
The schematic of the experimental setup for depo-
sition process is shown in Fig. 3. The experiments are
carried out EDM-smart ZNC (Electronica Machine
FIGURE 2.—EDD process modeling flowchart.
1856 B. MURALIDHARAN ET AL.
Tools Limited, Pune, Maharashtra, India). Since copper
has good physical, electrical, thermal, and chemical
properties, their ranges of applications are significantly
wide, increasing from micro to macro level. The contri-
bution of stainless steel is also increasing in all machine
parts and mechanical applications. Hence in this
research work, copper and stainless steel are selected
as tool and workpiece materials, respectively. Proper
tool and workpiece configurations are identified by con-
ducting trial experiments.
Copper wire of diameter 3.3 mm is taken as tool and
steel plate of dimension (5 mm 5mm1.5 mm) is
taken as workpiece. The experiments are conducted
using air as dielectric medium. Care has been taken to
highly polish the steel substrate (200–400 nm, surface
roughness) before conducting experiments. Digital stor-
age oscilloscope (DSO) is connected to the experimental
setup in order to track the pulses from which a single
pulse discharge can be found.
Machining parameters, like current, duty factor,
pulse-on time and voltage, influence the deposition
efficiency. Because the deposition process is not fully
established, the significant process parameters are ident-
ified by designing the experiments according to the con-
cepts of DOE. Here three independent input parameters
are selected, i.e., current (I) in amp, duty factor (t), and
pulse-on time (T
on
) in sec, for conducting experiments.
Variation in voltage changes the interelectrode gap
(IEG). In order to maintain a constant IEG, voltage is
kept constant throughout the experiments. The input
parameters with their three levels are shown in Table 1.
In the present research, central composite design (CCD)
has been used to design the experiments by using
Design-Expert statistical software (7.0, Stat-Ease, Inc.,
Minneapolis, MN, USA). Obeng et al. [36] suggested
that CCD requires less experiments than full factorial
design to get information about a process. CCD can
be a useful alternative method to factorial design which
is reported by Box and Hunter [37]. In CCD, experi-
ments are designed on the basis of 2
k
Factorial, where
k is the number of variables. In the present design, the
number of experiments for small fraction with three fac-
tors is 15. The designed experiments are shown in the
Table 2.
The responses are measured in terms of height (H) in
m, weight (W) in kg, and width (WD) in m. Since the
experiments are conducted for single discharge, the
deposited height, width, and weight are in microns level
and hence the images are taken through Atomic force
microscope (AFM) and Zeiss Axio Scope microscope
in order to measure the output responses. From the typi-
cal AFM image as shown in Fig. 4, the height of
deposited material over the substrate can be measured.
Width of the deposited spot is measured by using optical
microscope. The typical microscopic image of deposited
spot for width measurement is shown in Fig. 5. The
weight of deposition can be estimated from the grain
analysis module present in AFM.
TABLE 1.—Factors and their levels.
S.No. Factors
Levels
101
1. Current (I), A 0.5 1 1.5
2. Pulse-on time (T
on
), s 0.25E-6 0.50E-6 0.75E-6
3. Duty factor (t) 1 2 3
FIGURE 3.—Schematic diagram of experimental setup.
TABLE 2.—Experimental design and responses.
Ex. No
Input parameters Responses
IT
on
t H WD W
1 1 0.00000050 2 1.8E-06 1.07E-05 6.68E-13
2 0.5 0.00000075 3 1.3E-06 7.5E-06 2.41E-13
3 0.5 0.00000050 2 1.5E-06 6.67E-06 4.60E-13
4 1 0.00000050 1 3.0E-06 1.07E-05 1.10E-12
5 1 0.00000050 3 2.1E-06 1.03E-05 9.03E-13
6 0.5 0.00000025 1 3.5E-06 7.19E-06 1.08E-12
7 1 0.00000050 2 2.4E-06 1.03E-05 7.82E-13
8 1 0.00000050 2 2.2E-06 1.07E-05 7.92E-13
9 1 0.00000050 2 2.5E-06 1.03E-05 7.46E-13
10 1.5 0.00000075 1 6.5E-07 1.19E-05 1.83E-13
11 1 0.00000050 2 2.3E-06 1.03E-05 8.50E-13
12 1.5 0.00000050 2 1.1E-06 1.08E-05 4.22E-13
13 1 0.00000075 2 8.0E-07 9.8E-06 1.53E-13
14 1.5 0.00000025 3 2.8E-06 1.2E-05 1.15E-12
15 1 0.00000025 2 3.01E-06 1.15E-05 1.09E-12
FIGURE 4.—Typical AFM image showing deposited height.
SINGLE-SPARK ANALYSIS OF EDD PROCESS 1857
SIMULATION STUDIES
An axisymmetric model of the tool is developed by
considering copper as tool electrode and steel as work-
piece material. The isotherm above the melting point
of copper is assumed to be consisting of melted and
vaporized copper material which is ready to deposit
over the substrate material and the volume of melted
material can be estimated from this isotherm. The
component species involved in the transport phenom-
enon are selected based on EDAX compositional
analysis reported by [17]. The physical, thermal, and
chemical properties of the reaction species are given
as input for domain reactions. When the material is
melted, the mixture of melted and gaseous species with
molar concentration and the molar volume undergoes
surface reaction. The reaction rate will be determined
by the constants like sticking coefficient of the corre-
sponding species, plasma temperature, and plasma
pressure. The height of the deposited material can be
estimated from surface rate expressions, molecular
weight, and density of the species from the domain
surface after reaction. Width of the deposited material
can be estimated from the plasma radius. The weight
of the deposited material is evaluated from the total
volume of all species after reactions. Figure 6shows
the height of deposited material in isometric view.
Figure 7shows the total mass of the species deposited
over the surface.
Response surface methodology (RSM) is a math-
ematical and statistical technique useful in developing
and optimizing a new process. Raymond Mayers et al.
[38] reported that RSM is applied to approximate a
model for an unknown physical mechanism and this
helps to idenfy the most significant input parameters
influencing the output response. The relationship
between single dependent variable (response) and the
number of independent variables (regressors) are
related by a mathematical model called regression
model. The result of any experiment can be expessed
quantitatively in terms of an empirical regression
model reported by [39]. In general, response
variable Y may be related to k regressor variables
and is given by Eq. (13):
Y¼b0þXk
i¼1biXiþXk
i¼1biiXi2
þXXi<jbij XiXjð13Þ
where Yis the corresponding response, e.g., weight,
height, etc.; X
i
X
j
are the levels of k quantitative
process variables; and b
0
,b
i
,b
ii
,b
ij
are the regression
coefficients.
The height, width, and weight models are developed
by analyzing the data listed in Table 2and it is given
FIGURE 5.—Typical optical microscope image showing deposited width.
FIGURE 6.—Accumulated growth height.
FIGURE 7.—Weight of deposited material.
1858 B. MURALIDHARAN ET AL.
as follows:
Height ¼3:90E 06 þ6:85E 06 I2:022 Ton
2:97E 06 t3:5ITon 3:6E 08
Itþ0:55 Ton t2:71E 06 I2
þ5:70E 07 t2
ð14Þ
Width ¼1:13E 06 þ1:97E 05 Iþ0:926 Ton
þ4:24E 07 t2:42 ITon 1:802E
06 It0:95 Ton t5:40E 06
I2þ4:13E 07 t2
ð15Þ
Weight ¼6:93E 13 þ3:79E 12 Iþ9:22E 08
Ton1:18E 12 t1:71E 06 ITon
þ9:67E 15 It1:41E 07 Ton t
1:49E 12 I2þ2:74E 13 t2
ð16Þ
Analysis of variance (ANOVA) has been used to check
the adequacy of the developed model. The p-values of the
predictive model are calculated. The p-value which is less
than 0.0500 indicates that model terms are significant as
reported by [38]. The ANOVA for the second-order
model for Eq. (14) has been presented in Table 3, for
Eq. (15) it is presented in Table 4, and for Eq. (16)itis
presented in Table 5. It is observed that, p-value from
Table 3is 0.0101, from Table 4is 0.0002, and from
Table 5is <0.0001. This shows the significance of the
developed model. Since there are many insignificant
model terms found in Tables 3–5, the model has to be
improved using backward regression elimination
method, which eliminates the terms, having insignificant
effect on output responses. The backward elimination is
TABLE 3.— ANOVA for developed response surface from Eq. (14).
Source Sum of squares df Mean square F-value p-value
Model 9.71E-12 9 1.0798E-12 10.0936295 0.0101 Significant
Current (I) 8.00E-14 1 8E-14 0.74781758 0.4267
Pulse on (T
on
) 2.44E-12 1 2.44426E-12 22.8482622 0.0050
Duty factor (t) 4.05E-13 1 4.05E-13 3.78582651 0.1093
IT
on
2.55E-13 1 2.55208E-13 2.38561599 0.1831
It 4.32E-16 1 4.32E-16 0.00403821 0.9518
T
on
t 2.52E-14 1 2.52083E-14 0.23564044 0.6479
I
2
1.14E-12 1 1.14699E-12 10.7217554 0.0221
T2
on 8.41E-15 1 8.41864E-15 0.07869513 0.7903
t
2
9.03E-13 1 9.0353E-13 8.44594517 0.0335
Residual 5.34E-13 5 1.06978E-13
Lack of fit 2.42E-13 1 2.4289E-13 3.32725759 0.1422 Not significant
Pure error 2.92E-13 4 7.3E-14
Cor total 1.02E-11 14
TABLE 4.— ANOVA for developed response surface from Eq. (15).
Source
Sum of
Squares df
Mean
square F-value p-value
Model 3.72E-11 9 4.13E-12 55.27 0.0002 Significant
Current (I) 8.52E-12 1 8.52E-12 114.05 0.0001
Pulse on (T
on
) 1.43E-12 1 1.43E-12 19.25 0.0071
Duty factor (t) 8.00E-14 1 8.00E-14 1.06 0.3484
IT
on
1.22E-13 1 1.22E-13 1.63 0.2576
It 1.08E-12 1 1.08E-12 14.47 0.0126
T
on
t 7.52E-14 1 7.52E-14 1.00 0.3619
I
2
5.76E-12 1 5.76E-12 77.09 0.0003
T2
on 4.87E-13 1 4.87E-13 6.52 0.0510
t
2
2.05E-13 1 2.05E-13 2.74 0.1582
Residual 3.73E-13 5 7.47E-14
Lack of fit 1.81E-13 1 1.81E-13 3.78 0.1235 Not
significant
Pure error 1.92E-13 4 4.8E-14
Cor total 3.75E-11 14
TABLE 5.— ANOVA for developed response surface from Eq. (16).
Source
Sum of
squares df
Mean
square F-value p-value
Model 1.93E-24 9 2.14E-25 18.37 <0.0001 Significant
Current (I) 3.85E-28 1 3.85E-28 0.03 0.86
Pulse on (T
on
) 4.55E-25 1 4.55E-25 38.95 0.00
Duty factor (t) 4.66E-26 1 4.66E-26 3.99 0.10
IT
on
6.14E-26 1 6.14E-26 5.25 0.07
It 3.12E-29 1 3.12E-29 0.00 0.96
T
on
t 1.67E-27 1 1.67E-27 0.14 0.72
I
2
2.94E-25 1 2.94E-25 25.23 0.00
T2
on 3.89E-26 1 3.89E-26 3.33 0.12
t
2
2.54E-25 1 2.54E-25 21.74 0.00
Residual 5.84E-26 5 1.16E-26
Lack of fit 3.60E-26 1 3.60E-26 6.42 0.06 Not
significant
Pure error 2.24E-26 4 5.60E-27
Cor total 1.99E-24 14
SINGLE-SPARK ANALYSIS OF EDD PROCESS 1859
a stepwise method which begins by developing a full
regression model using all independent variables. Then
at a specified alpha level, a F-test or t-test for significance
is performed on each regression coefficients. The back-
ward elimination will continue until all independent
variables remaining in the model will have t-values above
a predefined set level. The advantage of backward elim-
ination method is that there is a possibility to look at
all the independent variables in the model before remov-
ing the variables that are not significant. The new models
for height, width, and weight of deposition are obtained
after applying backward elimination reduction method:
Height ¼3:59E 06þ6:60E 06 I0:87Ton
2:73E 06 t3:5ITon2:71E
06I2þ5:70E 07 t2ð17Þ
Width ¼5:18E 07 þ1:73E 05 I3:39
Ton þ1:80E 06 t1:802E 06
It4:67E 06 I2ð18Þ
Weight ¼7:74E 13 þ3:86E 12 I2:04E
07 Ton1:24E 12 t1:71E
06 ITon 1:49E 12 I2
þ2:74E 13 t2ð19Þ
The ANOVA for improved model from Eq. (17)has
been presented in Table 6. It is observed from Table 6
that the p-value is <0.0001, which shows the signifi-
cance of the model. Also, from Table 6, the factors such
as, current, pulse-on time, and duty factor are found to
be significant factors for deposition height. The
ANOVA for the improved model described by Eqs.
(18)and(19) has been presented in Table 7and
Table 8.FromTable7,itisobservedthat,p-value
is <0.0001 and from Table 8,p-value is <0.0001,
showing that both the modified models for width and
weight are significant. Current and pulse-on time are
found to be significant factors for width which is
observed from Table 7.FromTable8, pulse-on time
and duty factor are found to be significant factors for
weight of deposition. Lack of fit is not significant for
all the developed models which are desired. So, the
developed model can represent responses, when the
corresponding input parameters are used in the model.
The above developed empirical model can be used to
predict the values of height, weight, and width of
deposition significantly. The variations between the
experimental, simulation, and predicted responses are
showninFig.8. Between experimental and predicted,
the calculated range of error percentage for height is
from 21.12% to 18.60%, for width is from 6.62% to
4.54% and for weight is from 9.99% to 20.94%,
respectively. Between experimental and simulation,
the calculated range of error percentage for height is
from 11.69% to 21.92%, for width is from 13.99% to
18.13%, and for weight is from 17.26% to 12.24%,
respectively. These variations in error percentage are
due to experimental setup difficulties and assumptions
made while developing the model. Also, the average
TABLE 6.— ANOVA for developed response surface from Eq. (17).
Source
Sum of
squares df
Mean
Square F-value p-value
Model 9.68E-12 6 1.61E-12 22.69 <0.0001 Significant
Current (I) 5.104E-13 1 5.10E-13 7.176 0.0280
Pulse on (T
on
) 7.174E-12 1 7.17E-12 100.88 <0.0001
Duty
factor (t)
4.05E-13 1 4.05E-13 5.694 0.0441
IT
on
2.552E-13 1 2.55E-13 3.588 0.0948
I
2
1.334E-12 1 1.33E-12 18.766 0.0025
t
2
9.396E-13 1 9.39E-13 13.211 0.0066
Residual 5.689E-13 8 7.116E-14
Lack of fit 2.769E-13 4 6.92E-14 0.948 0.5198 Not
significant
Pure error 2.92E-13 4 7.3E-14
Cor total 1.025E-11 14
TABLE 7.— ANOVA for developed response surface from Eq. (18).
Source
Sum of
squares df
Mean
square F-value p-value
Model 3.67E-11 6 6.13E-12 63.17 <0.0001 Significant
Current (I) 2.96E-11 1 2.96E-11 305.51 <0.0001
Pulse on (T
on
) 1.43E-12 1 1.43E-12 14.83 0.00
Duty factor (t) 1.66E-17 1 1.66E-17 0.00 0.98
It 1.08E-12 1 1.08E-12 11.14 0.01
I
2
5.64E-12 1 5.64E-12 58.19 <0.0001
T2
on 7.75E-13 1 7.75E-13 7.98 0.02
Residual 7.76E-13 8 9.70E-14
Lack of fit 5.84E-13 4 1.46E-13 3.04 0.15 Not
significant
Pure error 1.92E-13 4 4.8E-14
Cor total 3.75E-11 14
TABLE 8.— ANOVA for developed response surface from Eq. (19).
Source
Sum of
squares df Mean are F-value p-value
Model 1.93E-24 7 2.75E-25 32.10 <0.0001 Significant
Current (I) 5.68E-28 1 5.68E-28 0.06 0.80
Pulse on (T
on
) 1.38E-24 1 1.38E-24 161.13 <0.0001
Duty factor (t) 4.66E-26 1 4.66E-26 5.43 0.05
IT
on
6.14E-26 1 6.14E-26 7.15 0.03
I
2
2.94E-25 1 2.94E-25 34.33 0.00
T2
on 3.89E-26 1 3.89E-26 4.53 0.07
t
2
2.54E-25 1 2.54E-25 29.58 0.00
Residual 6.01E-26 7 8.59E-27
Lack of fit 3.77E-26 3 1.25E-26 2.24 0.22 Not
significant
Pure error 2.24E-26 4 5.60E-27
Cor total 1.99E-24 14
1860 B. MURALIDHARAN ET AL.
predicted error calculated between experimental and
predicted for height, width, and weight are 0.29%,
0.016%, and 0.53%, respectively. The average
predicted error calculated between experimental and
simulation results for height, weight, and width are
0.58%, 3.66%, and 1.92%. Since the average predicted
error percentages were less, the developed empirical
model has better fit.
RESULTS AND DISCUSSION
In the present work, model based on deposition
phenomenon, experimental work, and RSM model are
developed. In order to justify the methodology and
assumptions made for developing phenomenological
model, the simulation results obtained by using phenom-
enological model are compared with experimental
work. To conduct parametric analysis, RSM model is
developed from systematically designed experiments
with input factors and the corresponding recorded out-
put responses. Now, from the RSM model, main effects
and interaction effects are plotted to understand the
interaction of input factors with the output responses.
The inferences from the plots are discussed below.
The effects of current on height for various duty
factors are shown in Fig. 9(a). It is observed that height
increases initially until 1A and beyond that it starts
decreasing. Current is a function of heat energy gener-
ated. Therefore, beyond 1A amount of heat generated
is comparatively high and more amount of material is
melted and vaporized. This increases the possibility of
more amount of vaporized matter escaping from the
plasma column, which reduces the height with increasing
width. Effect of pulse-on time on height for various duty
factors are shown in the Fig. 9(b). It is seen that when
the duty factor increases, height decreases. Tool wear
will be higher at shorter pulse than at longer pulses
[40]. Because of higher tool wear more amount of
material is transferred from tool to workpiece at shorter
pulse time. Effects of duty factor on height for different
current ranges are shown in the Fig. 9(c). For a constant
pulse-on time, when duty factor increases, pulse width
will be raised. It is understood that at higher pulse-off
time heat dissipation will be more between two success-
ive pulses which causes less amount of material to be
removed from tool. Therefore height decreases with
increase in duty factor.
Effects of duty factor on height for different current
ranges are shown in the Fig. 8c. For a constant pulse-on
time, the effect of current on width for varying pulse-on
time is shown in Fig. 10(a). It is noted that width
increases with increase in current. As said above,
amount of heat generated will be more with increase in
current. Therefore, more amount of material is melted
and vaporized at higher current values. So, the diameter
of the plasma column will increase; this increases the
width with reduced height of deposition over the
substrate material. Width and weight decreases with
FIGURE 8.—Variation of experimental, predicted, and simulation values for all runs.
FIGURE 9.—Effect of significant factors on height of deposition.
SINGLE-SPARK ANALYSIS OF EDD PROCESS 1861
increase in pulse-on time for various current values and
different duty factor, respectively, are shown in the
Figs. 10(b) and 11(a). This is because shorter pulse leads
to higher tool wear rate as mentioned earlier. Therefore,
there will be increased width and more deposition at
shorter pulse-on time which can be observed from the
corresponding figures. Effects of duty factor on weight
for different ranges of pulse-on time are shown in the
Fig. 11(b). When duty factor increases, pulse-off time
increases for a constant pulse-on time. So, at higher
pulse-off time, heat loss will be more between two suc-
cessive pulses which leads to less amount of material
to be removed from tool. Therefore, weight decreases
with increase in duty factor.
The maximum height measured is 3.5E-06 m for the
machining parameters of I ¼0.5A, T
on
¼0.25 ms and
t¼1. The height of deposition process is directly pro-
portional to surface rate expressions. Tool wear will be
more at lesser pulse-on time which increases the molar
volume of the melted mixture domain. This increase in
molar volume increases the height of deposition at mini-
mum pulse-on time. The minimum width measured is
6.7E-06 m for the machining parameters of I ¼0.5A,
T
on
¼0.5 ms, and t ¼2. The width of deposition depends
on plasma radius. Plasma radius is directly proportional
to current and pulse-on time. So the deposition width is
minimum at the lowest range of current values. Mini-
mum width is desired for dimensional integrity and
accuracy. The maximum weight measured is
1.15E-12 kg for the machining parameters of I ¼1.5A,
T
on
¼0.25 ms and t ¼3. Because of higher current values
and lesser pulse-on time, there will be a combination of
more tool wear with maximum temperature in the melt-
ing zone. This increases surface reaction rate by raising
the molar concentration and mean molecular weight of
gas mixture. So the deposition weight increases signifi-
cantly. As observed from previous results [16,17], from
the present research work it is noted that lesser pulse-on
time assists in deposition of maximum height and
maximum weight. Also, less current creates deposition
with minimum width [18] desired.
The main barrier for widespread usage of this tech-
nology is the machining setup to eliminate oxide and
carbon formation in the composition of final product,
maintaining size of tool used, controlling deposition
rate, and etching of the final product from the base
material without altering any properties of the depo-
sition material. The EDD process overcomes the manu-
facturing cost and limitations faced by lithographic
technologies like geometry, application range, and
material usage to produce 3D microproducts.
CONCLUSIONS
A methodical development and investigation of single
discharge deposition process is reported in this paper for
various conditions. Microproducts with high tempera-
ture application, automobile, aerospace, MEMS, and
customized medical implants are some of the application
areas of the products manufactured by this process. The
FIGURE 10.—Effect of significant factors on width of deposition.
FIGURE 11.—Effect of significant factors on weight of deposition.
1862 B. MURALIDHARAN ET AL.
range of application is limited to the conducting materi-
als with the melting point less than 7000 K for the
specific parameters suggested. The following conclusions
can be drawn from the present research work:
1. A thermophysical model for single discharge EDD
process was developed. Gaussian heat distribution
has been used to conduct thermal analysis and the
concept of heavy species transport phenomenon
was applied to understand deposition process. Here
height, width, and weight of depositions have been
estimated.
2. Parametric analysis has been done in order to find
the influence of the significant process parameters
on the performance of the deposition process.
3. It has been observed that pulse-on time plays a sig-
nificant role in height and weight of deposition.
The maximum height measured is 3.5E-06 m and
maximum weight measured is 1.15E-12 kg, which
corresponds to shorter pulse-on time of 0.25 ms.
4. It has been observed that current plays a significant
role in width of deposition. Minimum width of depo-
sition measured is 6.7E-06 m, which corresponds to
lowest current value of 0.5A.
It is also found that the developed model can predict
the shape of deposited material over the substrate as
well. This model can be further enhanced to establish
an EDD process.
ACKNOWLEDGMENT
The authors acknowledge the Department of Mechan-
ical Engineering, IIT Kanpur, Kanpur, India, for
providing COMSOL software support.
REFERENCES
1. Ehmann, K.F.; Bourell, D.; Culpepper, M.L.; Hodgson, T.J.;
Kurfess, R.T.; Madou, M.; Rajurkar, K.P.; DeVor, R.E.
International assessment of research and development in
micromanufacturing. WTEC Panel Report 2005, pp. 1–7.
2. Alting, L.; Kimura, F.; Hansen, H.N.; Bissacco, G. Micro
engineering. CIRP Annals Manufacturing Technology 2003,
52 (2), 635–658.
3. Cui, Z. Micro-Nano Fabrication, Higher Education Press:
Springer, China, 2005; pp. 1–11.
4. Brousseau, E.B.; Dimov, S.S.; Pham, D.T. Some recent
advances in multi-material micro and nano manufacturing.
International Journal of Advanced Manufacturing Technology
2010,47 (1–4), 161–180.
5. Dimov, S.S.; Matthews, C.W.; Glanfield, A.; Dorrington,
P.A. Roadmapping study in multi-material micro manufac-
ture. Proceedings of the Second International Conference on
Multi-Material Micro Manufacture 4M2006, Grenoble,
France, September 20–22, 2006, xi–xxv.
6. Rajurkar, K.P.; Levy, G.; Malshe, A.; Sundaram, M.M.
J. McGeough, Hu, X.; Resnick. DeSilva, R. Micro and Nano
machining by electro-physical and chemical processes. CIRP
Annals Manufacturing Technology 2006,55 (2), 643–666.
7. Gangadhar, A.; Shanmugam, M.S.; Philip, P.K. Surface
modification in electro-discharge processing with powder
compact tool electrode. Wear 1991,143, 45–55.
8. Goto, A.; Magara, T.; Imai, Y.; Miyake, H.; Saito, N.;
Mohri, N. Formation of hard layer on metallic material by
EDM. Journal of the Japan Society of Electrical Machining
Engineers 1997,31 (68), 26–31.
9. Chen, H.J.; Wu, K.L.; Yan, B.H. Characteristics of Al alloy
surface after EDC with sintered Ti electrode and TiN powder
additive. The International Journal of Advanced Manufactur-
ing Technology 2014,72 (1–4), 319–332.
10. Huang, T.S.; Hsieh, S.F.; Chen, S.L.; Lin, M.H.; Ou, S.F.;
Chang, W.T. Surface modification of TiNi-based shape mem-
ory alloys by dry electrical discharge machining. Journal of
Materials Processing Technology 2015,221, 279–284.
11. Gill, A.S.; Kumar, S. Surface alloying of H11 die steel by
tungsten using EDM process. International Journal of
Advanced Manufacturing Technology 2015,78 (9–12),
1585–1593.
12. Simao, J.; Lee, H.G.; Aspinwall, D.K.; Dewes, R.C.;
Aspinwall, E.M. Workpiece surface modification using
electric discharge machining. International Journal of Machine
Tool and Manufacture 2003,43, 121–128.
13. Soni, J.S.; Chakraverti, G. Experimental investigation on
migration of material during EDM of die steel (T215 Cr12).
Journal of Materials Processing Technology 1996,56,
439–451.
14. Wang, Y.K.; Xie, B.C.; Wang, Z.L.; Peng, Z.L. Micro EDM
deposition in air by single discharge thermo simulation.
Transactions of Nonferrous Metals Society of China 2011,
21, 450–455.
15. Chi, G.; Wang, Z.; Xiao, K.; Cui, J.; Jin, B. The fabrication of
a micro-spiral structure using EDM deposition in the air.
Journal of Micromechanics and Microengineering 2008,18,
1–9.
16. Jain, V.K.; Shashank.; Ajay Sidpara; Himanshu Jain. Some
aspects of micro-fabrication using electro-discharge depo-
sition process. Proceedings of International Symposium on
Flexible Automation ISFA 2012, 2012,108, 419–424.
17. Muralidharan, B.; Chelladurai, H. Experimental analysis of
electro-discharge deposition process. International Journal of
Advanced Manufacturing Technology 2015,76 (1–4), 69–82.
18. Muralidharan, B.; Chelladurai, H.; Ramkumar, J. Experi-
mental investigation on electro discharge deposition (EDD)
process. International Conference - ASME 2013 International
Mechanical Engineering Congress & Exposition, SanDiego,
CA, USA. November 15–21, 2013.
19. Beck, J.V. Transient temperatures in a semi-infinite cylinder
heated by a disk heat source. International Journal of Heat
and Mass Transfer 1981,24 (10), 1631–1640.
20. Jilani, S.T.; Pandey, P.C. Analysis of surface erosion in
electrical discharge machining. Wear 1983,84 (3), 275–284.
21. DiBitonto, D.D.; Eubank, P.T.; Patel, M.R.; Barrufet, M.A.
Theoretical models of the electrical discharge machining pro-
cess I. A simple cathode erosion model. Journal of Applied
Physics 1989,66 (9), 4095–5103.
22. Patel, M.R.; Barrufet, M.A.; Eubank, P.T.; DiBitonto, D.D.
Theoretical models of the electrical discharge machining
process.II. The anode erosion model. Journal of Applied
Physics 1989,66 (9), 4104–4111.
SINGLE-SPARK ANALYSIS OF EDD PROCESS 1863
23. Kuriachen, B.; Varghese, A.; Somashekhar, K.P.; Panda, S.;
Mathew, J. Three-dimensional numerical simulation of
microelectric discharge machining of Ti-6Al-4V. International
Journal of Advanced Manufacturing Technology 2015,79 (1–4),
147–160.
24. Snoeys, R.; VanDijck, F.S. Investigation of electro discharge
machining operations by means of thermo-mathematical
model. CIRP Annals 1971,20 (1), 35–37.
25. Van Dijck, F.S.; Dutre, W.L. Heat conduction model for the
calculation of the volume of molten metal in electric
discharges. Journal of Physics D: Applied Physics 1974,7(6),
899–910.
26. Yadav, V.; Jain, V.K.; Dixit, P.M. Thermal stresses due to
electrical discharge machining. International Journal of
Machine Tools & Manufacture 2002,42, 877–888.
27. Ikai, T.; Hashigushi, K. Heat input for crater formation in
EDM. Proceedings of international symposium for
electro-machining, ISEM XI, EPFL, 1995, 163–170.
28. Joshi, S.N.; Pande, S.S. Thermo-physical modeling of
die-sinking EDM process. Journal of Manufacturing Processes
2010,12, 45–56.
29. Erden, A. Effect of materials on the mechanism of
electric discharge machining (EDM). Transaction of ASME
Journal of Engineering Materials and Technology 1983,108,
247–251.
30. Pandey, P.C.; Jilani, S.T. Plasma channel growth and the
resolidified layer in EDM. Precision Engineering 1986,8(2),
104–110.
31. Hashimoto, X.H.; Kunieda, M.; Nishiwaki, N. Measurement
of energy distribution in continuous EDM process. Journal of
the Japan Society for Precision Engineering, 1996,62 (8),
1141–1145.
32. Yeo, S.H.; Kurnia, W.; Tan, P.C. Electro thermal modeling
of anode and cathode in micro-EDM. Journal of physics D:
Applied Physics 2007,40 (8), 2513–2521.
33. Adineh, V.R. Radiative heat loss of electrical discharge
machining process through hydrocarbon oil and deionized
water dielectric liquids. Plasma Chemistry and Plasma Proces-
sing 2012,32, 369–392.
34. Kee, R.J.; Coltrin, M.E.; Glarborg, P. Chemically Reacting
Flow Theory and Practice, Wiley, 2003; pp. 445–530.
35. Hagelaar, G.J.M. Modeling of microdischarges for display
technology. Ph.D thesis, 2000; pp. 13–27.
36. Obeng, D.P.; Morrell, S.; Napier, T.J.N. Application of cen-
tral composite rotatable design to modelling the effect of
some operating variables on the performance of the
three-product cyclone. International Journal of Mineral Pro-
cessing 2005,769,181–192.
37. Box, G.E.P.; Hunter, J.S. Multi-factor experimental design
for exploring response surfaces. Annals of Mathematical Stat-
istics 1957,28 (1), 195–241.
38. Raymond Mayers, H.; Douglas Montgomery, C.; Christene
Anderson-Cook, M. Response Surface Methodology Process
and Product Optimization Using Designed Experiments, John
Wiley &Sons, 2009; pp. 1–82.
39. Douglas Montgomery, C. Design and Analysis of Experi-
ments, John Wiley & Sons, 2009; pp. 388–443.
40. Mohri, N.; Suzuki, M.; Furuya, M.; Saito, N. Electrode wear
process in electric discharge machining. CIRP Annals
Manufacturing Technology 1995,44, 165–168.
1864 B. MURALIDHARAN ET AL.
