Int J Adv Manuf Technol

DOI 10.1007/s00170-007-1068-y

ORIGINAL ARTICLE

A study of electrical discharge grinding using a rotary

disk electrode
Hung Rung Shih & Kuen Ming Shu

Received: 24 October 2006 / Accepted: 27 April 2007

# Springer-Verlag London Limited 2007

Abstract This study aims to investigate the electrical

discharge grinding (EDG) using a rotary disk electrode.
From a practical perspective, the electrode is designed to
mimic the machining process of a surface grinder with
horizontal spindles. First, the machining ability of cold
working tool steel AISI D2 by EDG is investigated. Then,
the optimal machining parameters are found through
ANOVA analysis. The experimental results show that both
the lower electrode wear rate and the higher materials
removal rate are obtained when a rotary disk electrode with
positive polarity is conducted on EDG. In addition, the
roughness of machined surfaces using an electrode with
negative polarity can reach about 2 m Rt, and no
incidence of concentrated discharge or short circuit are
found in any machining conditions.
Keywords Electrical discharge grinding . Rotary disk
electrode . ANOVA analysis . Optimal machining parameters

1 Introduction
Scientific and engineering advances in the last few decades
have placed remarkable demands on metalworking industry. One aspect of these demands is that metals with high
strength-to-weight ratios have been developed to serve
specific purposes. These metallurgical trends have led to
H. R. Shih (*) : K. M. Shu
Department of Mechanical and Computer-Aided Engineering,
National Formosa University,
Hu-wei, Yunlin, Taiwan, Republic of China
e-mail: [email protected]
K. M. Shu
e-mail: [email protected]

the developments of chemical, electrical, and other alternative means for material removal beginning in the 1940s.
Since it is easy to automate, operate, and machine any
material with electric conductivity, electrical discharge
machining (EDM) has been shown to be a versatile method
for machining difficult-to-work materials, including heattreated steels, tungsten carbides and various conductive
ceramics. However, low machining efficiency is one of its
main disadvantages. How to reduce machining time and
maintain reasonable accuracy at the same time has always
been the topic of research interest [1].
In the process of die-sinking electrical discharge machining, in general, either fixed electrodes are used to
produce die cavities or a rotary device works in conjunction
with a CNC to control the electrodes path in various EDM
profiling [24]. However, some limitations have to be
placed on this type of machine. For example, a large
surface with good flatness is hard to be machined by this
machining process [5]. On the other hand, wire-cut
electrical discharge machining (WEDM), which employs a
continuously moving thin copper wire as the electrode, can
easily fabricate a large surface. Nevertheless, the vibration
of the wire electrode arising from the explosions in
electrical discharge seems to be unavoidable and is the
dominant factor for the flatness of machined surfaces [6].
Electrochemical discharge grinding (ECDG) combines
the material removal actions of electrical discharge grinding
with electrolytic attack to remove materials from electrically conductive workpieces. There is a possibility that
chemicals of electrolytic action might interact locally with
cobalt components, thus producing residual carbide, which
may cause defects on a workpiece or machined surfaces
with poor quality.
In contrast to nontraditional processes, the traditional
contact-type grinder can achieve good quality on surface

Int J Adv Manuf Technol

roughness (less than 0.1 m Ra), but wear of a cutting

wheel has a negative effect on the precision of angular and
linear dimensions. It is particularly obvious in grinding if
the materials are hard but brittle. Occasionally, it even
causes a workpiece with chipping or cracks. The workpiece
may suffer damage from machining processes.
In the domain of micro electro-mechanics, a silicon
wafer is used as a substrate and coated with a metal film for
conductivity in the post process of LIGA. It is then used to
fabricate a microstructure, which has a large ratio of depth
to width. Surfaces of the final product from this electroforming process are required to be well trimmed. If a
contact-type trimming process is used, it leads to a
machining failure, which causes the microstructure to
deform, collapse, and the substrate to peel off completely.
In response to the needs as mentioned above, a study on
electrical discharge grinding (EDG) is performed. Therefore, the purpose of this paper is to investigate the effect of
the EDG using a rotary disk electrode on the surface
texture, by conducting a series of experiments.

2 Experimental method
The machine used in this study is P36+E50 model made by
MAX SEE Industry Co., Ltd. It is an EDM machine with
CNC control. A horizontal rotary device, which turns
electrodes to discharge, is designed to work with the
EDM machine. A sketch for both is shown in Fig. 1. The
dielectric fluid used in these experiments was Castrol SPE
oil having a hydrocarbon base with a kinematics viscosity
of 2.08 cSt at 40C.
In experiments, electrolytic copper rings were used for
electrodes. They were mounted on mandrels made of
medium carbon steel. The dimensions of an electrode are
shown in Fig. 2.
Because EDM is widely used in mold fabrication, the cold
working tool steel AISI D2 is selected as the material for

Fig. 1 Sketch of equipment

Fig. 2 Rotary disk electrode

specimens. The specimens are pre-made in size of 1010

4.6 mm. There are several process variables can be investigated during the EDM process. Parameters involved in this
study are polarity, pulse duration, discharge depth, rotating
speed, discharge direction (two type discharge direction are
shown in Fig. 3), dielectric fluid and flushing direction.
Major correlated parametric values or ranges adopted are
shown in Table 1. Taguchis robust design methodology was
used for designing the experiment. The level for each factor
was chosen and shown in Table 1. An L18(21 37) orthogonal
array was designed for the experiment.
Before the experimental data was measured, surface
cleaning of all electrodes and specimens had been
performed by immersing in acetone under ultrasonic
vibration for 5 minutes. The measurements of microhardness of specimens were performed on a micro
sclerometer (MHV-2000, made by Shimadzu, Japan).
Before a reading was taken, a sustained load at 25 g was
applied for 5 seconds. Hardness reading of specimens
before EDM grinding is HV=495. Six measurements were
taken on each specimen and the average data were adopted
for analysis.
We first drew a 1 mm line on the micrograph, and then
measure the crack density by counting the average
intersection number between this line and micro cracks.

Fig. 3 Discharge directions

Int J Adv Manuf Technol

Table 1 Design scheme of experimental parameters for EDG
Observed
value

Control
parameters

Level

Surface
roughness
(m Rt)

Electrode
polarity
Discharge
direction
Peak
current (A)
Discharge
depth
(mm)
Flushing
direction
Pulse
duration
(s)
Rotating
speed
(RPM)
Error

(+)

(-)

Upward

Downward

Upward*

0.03

0.05

0.07

Front

None

Side

50

100

300

600

900

Material
removal rate
(g/min)

Electrode wear
rate (g/min)

The flat area between the craters hump and bottom was
elected as the standard measurement position of the microhardness.

3 Results and discussion

Fig. 5 The relationship between roughness, pulse duration and peak

current

(MRR) and surface roughness increase with respect to peak

current and pulse duration in cases where polarity is either
positive or negative. MRR and surface roughness both have
a greater value while machining with a positive polarity
(Electrode +, Workpiece ).
A single discharge energy equation is shown as follow:

3.1 Effect of EDM parameters on EDG

W IP2
Rg
t P

Results of the experiments are shown in Figs. 4 and 5.

From the figures, it is shown that both material removal rate

where W: discharge energy, Rg: voltage between electrode

and workpiece, p: pulse duration.
From this equation, we know that electrical current is the
most dominant factor (in squared order) to discharge
energy, and pulse duration is the second dominant factor.
The discharge depth in a single shoot from copper electrode
to steel has the following form [7]:
0:6
h 1:3
t 0:2
P
IP

Fig. 4 The relationship between MRR, pulse duration and peak

current

where h:discharge depth (m).

From this equation, we observe that discharge depth is
proportional to electrical current in power of 0.6 and to
pulse duration in power of 0.2. This observation is
consistent with the experiments outcome.
From Tables 2 and 3, it shows that the dominant factors
for MRR in a descending order are peak current, pulse
duration and polarity. Note that electrical current of
discharge is the most dominant factor for MRR. In the
range of the experimental settings, rotating speed has no
significant effect on MRR and roughness. In EDM
machining, the magnitude of electrode wear rate is
correlative to the precision for linear and angular dimensions on workpieces. For this reason, in addition to

Int J Adv Manuf Technol

Table 2 ANOVA and F-test for MRR
Parameter

Degree

Square sum

Variance

FA0

F0.05,

Electrode polarity
Discharge direction
Peak current
Discharge depth
Flushing direction
Pulse duration
Rotating speed
Error

1
2
2
2
2
2
2
4

48.6324
1.7942
547.9678
3.073
5.2144
457.9121
2.5925
4.4748

48.6324
0.8971
273.9839
1.5365
2.6072
228.9561
1.2963
1.1187

*43.4722
0.8019
**244.9128
1.3735
2.3306
**204.6626
1.1587

7.71
6.94
6.94
6.94
6.94
6.94
6.94

n1, n2

Contribution (%)
4.54
0.17
51.13
0.29
0.49
42.73
0.24
0.42

** Significant parameter; * Sub-significant parameter

roughness, the electrode wear rate is considered as another

important controlled factor in machining. From Fig. 6, we
observe that electrode wear rate is proportional to peak
current. In the cases for positive polarity, the electrode wear
rate is higher than the one for negative polarity. Also, the
electrode wear rate is in reverse proportion to pulse
duration. From the figure, we also discover that, under
some circumstances, the electrode does not lose weight; it
gains weight instead. This phenomenon is particularly
obvious in the case where pulse duration is 75 s and no
flushing.
There are two possibilities for the electrode to gain
weight: (1) the dielectric fluid generally is made of
hydrocarbon compounds. In a machining process, the
electric arc will release great heat and cause hydrocarbon
compounds to decompose into carbons and other compounds. The melting point of carbon is up to 3500C.
When carbons adhere to an electrodes surface, they protect
the electrode from wear and this greatly reduces electrodes
wear rate. (2) When cutting debris vaporized from a workpiece
splashes onto electrodes and mixes with electrodes material, an
alloy clot with a different color is formed on electrodes surface.
Figure 7 shows a micrograph of an electrodes surface.
Instead of losing weight, this electrode gains weight after
experiencing an EDG process. The parameters setting for
this process is the following: peak current=6A, pulse
duration=100 s, no flushing, negative polarity. In the

figure, a white meshed coagulation is shown. We suppose it

was formed by the molten cutting debris. Debris is
vaporized while grinding and splashed from workpiece
onto electrodes surface. Based on this theory, another special
machining process may emerge if parameters setting can be
controlled appropriately to allow the electrodes material
uniformly sprays onto a workpiece.
3.2 Microstructure of EDM machined surfaces
A part of an EDM grinded surface is shown in the
micrograph in Fig. 8. The whole surface is finished with a
single discharge. The primary objective of EDM machining
is to remove surfaced materials from workpieces. In reality,
only a few materials are removed and most materials, being
cooled by dielectric fluid, solidify instantly on the surface.
Due to this instantaneous quenching, tensile stress is
developed on the hardened layer [6]. When the stress
energy is great, some cracks may emerge. In a study for
single shoot discharge [7], cracks fan out radially from the
central spot of discharge. In addition to cracks, sometimes,
globules and pockmarks also appear. Globules appear at
pits where melted metals re-clot. Pockmarks are cavities on
the machined surface and are formed by vaporized bubbles
of dielectric fluid. The two pictures in Fig. 8 showing two
surfaces machined with positive and negative polarities are
compared in the following: given the same setting for other

Table 3 ANOVA and F-test for roughness

Parameter

Degree

Square sum

Variance

FA0

F0.05,

Electrode polarity
Discharge direction
Peak current
Discharge depth
Flushing direction
Pulse duration
Rotating speed
Error

1
2
2
2
2
2
2
4

9.9760
0.6414
55.4582
4.5490
2.6704
70.9938
0.2090
0.5548

9.9760
0.3207
27.7291
2.2745
1.3352
35.4969
0.1045
0.1387

**71.9250
2.3122
**199.9214
*16.3987
*9.6265
**255.9257
0.7534

7.71
6.94
6.94
6.94
6.94
6.94
6.94

** Significant parameter; * Sub-significant parameter

n1, n2

Contribution (%)
6.84
0.44
38.04
3.12
1.83
48.70
0.14
0.38

Int J Adv Manuf Technol

Fig. 6 The relationship between electrode wear rate, pulse duration

and electrical current

parameters, many traces of molten metals are seen on

machined surface, but complete traces of discharges are
scarcely seen in a case where positive polarity is used. On
the contrary, in a case where negative polarity is used,
complete traces of discharges are clearly seen on machined
surfaces. From this phenomenon, we can learn that the
discharged energy distributes differently in two cases.
3.2.1 Microcracks
In quenching, carbon steel develops cracks because its
material is transformed into Martensite. In the process of
transformation, two things related to size have changed a
shrink in steel itself due to instant cooling and an increase
in volume due to the transformation of Austensite into

Fig. 7 Adhered compounds on electrodes surface. (x500)

Martensite. Due to the changes in size, residual stress

occurs and it causes cracks to develop in steel. Under room
temperature, there is about a 1.4% linear expansion in the
transformation into martensite. If a metal with high hardenability is cooled to room temperature, it is very easy to
develop martensite, which is very hard but brittle, and the
sign of fracture for steel is even obvious. Surfaced materials
of workpieces are vaporized and melted by the great heat
and released during an EDM machining. The melted
materials are either washed away by dielectric fluid or
remain stayed. The stayed melt clots again to form a white
layer. Finally, the white layer is cooled down promptly.
During this swift cooling period, variation of temperature is
great. This process is similar to a quenching process in heat
treatment. Figure 9 shows the effect of pulse duration on
craters crack density in cases where positive and negative
polarities are used, respectively. In the figure, it shows that
craters crack density is proportional to pulse duration. This
phenomenon is probably caused by the fact that a large
increase in pulse duration is actually equivalent to a
decrease in machining rate and, on the contrary, an increase
in carburization and quenching effects. This makes a
workpieces surface even harder and more brittle - making
it more likely to develop cracks. It is also shown in Fig. 10
that the craters crack density, in a case where negative
polarity is used, is much higher than that where positive
polarity is used. The reason for this is that with the use of
negative polarity, carbon decomposed from dielectric fluid
will become a negative ion and attract the workpieces,
which are positive. Therefore, the concentration of carbon
near the workpieces surface increases, leading to an
increase in the carbon content diffused into the white layer.
3.2.2 Micro hardness
In EDM machining, surfaces of workpieces are subject to
diffusion of certain chemical elements, such as copper and
carbon, and quenching and instant heating. Due to these
effects, there is a significant difference in material properties between surface and interior of workpieces, which
includes hardness propriety. Microscope hardness readings
of workpieces under different machining conditions are
shown in Fig. 10. From the figure, we can observe that
microscope hardness readings increase with respect to pulse
duration. This is because the greater the energy discharged,
the easier it is to transform material into martensite and
diffuse carbon into workpieces. The averaged surface
hardness of a workpiece has a higher value in cases where
negative polarity is used than that in cases where positive
polarity is used. The primary reason is such that the longer
the pulse duration, the easier it is for carbon to adhere
unto the workpieces surface (positive polarity) and diffuse
into the white layer.

Int J Adv Manuf Technol

Fig. 8 Micrographs for machined surfaces under different
EDM parameters. (x500)

3.2.3 Best setting for EDM in surface grinding

The best setting shown in Fig. 11, from ANOVA analysis,
for EDM in surface grinding is as follows: negative
polarity, upward discharge, 3A, 0.07 mm, no flushing,
2 s, and 300 rpm, and its corresponding roughness
reading of the machined surface is 2 m Rt. Because a
machined surface is created from a series of electrical
discharges; therefore, the surface can be considered as a
collection of continuously created craters. The roughness
of a machined surface depends on craters diameter and
depth, which is the result of a single electrical discharge.
To obtain good quality of surface roughness, it requires
large crater diameter and shallow flat depressed base. To

create a shallow flat depressed base, the parameters for

electrical discharge must be controlled and maintained
properly. Also the clearance between electrode and
workpiece must be maintained properly so that debris
separates freely without introducing any unwanted electrical discharges. Under such a good condition, there is no
chance to damage the machined surface due to excessive
energy exposure resulting from multiple electrical discharges at same spot. Emission of vaporized dielectric
fluid, which engages in the cooling function for great heat,
shall also be considered. If any vaporized dielectric fluid
exists in the clearance, it will degrade electrical insulation
provided by dielectric fluid. Cavities on machined surfaces, resulting from infiltration of vaporized dielectric fluid,

Int J Adv Manuf Technol

Fig. 11 Machined surface (best setting). (x500)

Fig. 9 The relationship between unit crack density and pulse duration

have a negative effect on surfaces roughness and

mechanical properties.
Figure 12 shows several enlarged pictures of a surface
taken from a workpieces side view. From the pictures, we
can see the effect of energy scale on corner radius. The
larger the energy scale, the larger the corner radius. A
similar statement also holds for surface roughness. Two
effects lead to the birth of a corner: (1) energy concentrates
more at corners and this effect enables more materials to be
removed than that of flat surfaces. (2) Materials, melted at
corners by heat and not washed away by dielectric fluid,

will solidify into corners with circular profiles due to

cohesion. The more the materials melt, the larger the
corners radius is; i.e., the more the materials melt, the
deeper the melting is.

4 Applications
A potential application would be to cut the honeycomb as
shown in Fig. 13, which is made of aluminums foil with
86 m average thickness and of 1.6081 gm weight before
machining. Lower strength and thickness of honeycomb
keep it from clamping and/or cutting by using normal
grinding method before deformation occurs. For fixing the
honeycomb, four conductive blocks were located around it
and the other two 20 mg-weight blocks were, respectively,
placed on the honeycomb at the side of the to-be-cut
groove. No flushing was applied in this process of EDG.
Also, lower electrode rotating speed must be set to prevent
the honeycomb from vibrating by disturbing fluid. The
machining parameters were set as follows: negative
polarity, downward discharge, peak current 3A, discharge
depth 2.0 mm, no flushing, pulse duration 2 s, and
100 rpm rotating speed. When EDG was performed, no
deformation was observed on the edge of honeycomb.

5 Conclusion
A theoretical and experimental study on electrical discharge
grinding (EDG) using a rotary disk electrode has been
carried out. From the experiments and the research
described above, the following conclusions are drawn:
Fig. 10 The relationship between hardness and pulse duration

1. In order to obtain higher MRR on EDG, parameters

such as positive electrode polarity, higher peak current

Int J Adv Manuf Technol

Fig. 12 Corner radii. (x500)

Fig. 13 Applications of cut of

honeycomb using EDG

Int J Adv Manuf Technol

and longer pulse duration are suggested. To obtain

lower roughness on EDG, parameters such as negative
electrode polarity, lower peak current and shorter
pulse duration are suggested. The surface roughness
of machined surfaces could reach about 2 m Rt in
this study when optimum machining parameters were
set.
2. The machining parameters, such as rotating speed,
discharge direction and flushing direction, have no
significant effect on MRR and roughness in this
study.
3. Lower electrode wear rate was obtained with positive
electrode polarity and with longer pulse duration
machining parameters.
4. When using negative electrode polarity on EDG, the
micro craters crack density and micro-hardness value
are increased with pulse duration increase.

References
1. Shu KM, Tu GC (2003) Study of electrical discharge grinding
using metal matrix composite electrodes. Int J Mach Tools Manuf
43:845854
2. Yan BH, Wang CC, Liu WD, Huang FY (2000) Machining
characteristics of Al2O3/6061A1 composite using rotary EDM
with a disk like electrode. Int J Adv Manuf Technol 16(5):322333
3. Soni JS, Chakraverti G (1994) Machining characteristics of
titanium with rotary electro-discharge machining. Wear 171:5158
4. Don KS (1994) Electro discharge machining. Fu-Wen publish Co,
p 40
5. Bayramoglu M, Duffill AW (2004) CNC EDM of linear and
circular contours using plate tools. J Mater Process Technol
148:196203
6. Puri AB, Bhattacharyya B (2003) Modeling and analysis of the
wire-tool vibration in wire-cut EDM. J Mater Process Technol
141:295301
7. Enache S (1993) Dynamic stability of the technological machining
system in EDM. Ann CIRP 42(1):209214