Journal of Mechanical Science and Technology 24 (5) (2010) 1083~1090

www.springerlink.com/content/1738-494x
DOI 10.1007/s12206-010-0305-8

Optimization of EDM process for multiple performance characteristics

using Taguchi method and Grey relational analysis†
Jong Hyuk Jung and Won Tae Kwon*
Department of Mechanical and Information Engineering, University of Seoul, Seoul, 130-743, Korea

(Manuscript Received September 24, 2009; Revised November 2, 2009; Accepted November 30, 2009)

----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

Abstract

Electrical discharge machining (EDM) is one of the most extensively used non-conventional material removal processes. The Taguchi
method has been utilized to determine the optimal EDM conditions in several industrial fields. The method, however, was designed to
optimize only a single performance characteristic. To remove that limitation, the Grey relational analysis theory has been used to resolve
the complicated interrelationships among the multiple performance characteristics. In the present study, we attempted to find the optimal
machining conditions under which the micro-hole can be formed to a minimum diameter and a maximum aspect ratio. The Taguchi
method was used to determine the relations between machining parameters and process characteristics. It was found that electrode wear
and the entrance and exit clearances had a significant effect on the diameter of the micro-hole when the diameter of the electrode was
identical. Grey relational analysis was used to determine the optimal machining parameters, among which the input voltage and the ca-
pacitance were found to be the most significant. The obtained optimal machining conditions were an input voltage of 60V, a capacitance
of 680pF, a resistance of 500Ω, the feed rate of 1.5µm/s and a spindle speed of 1500rpm. Under these conditions, a micro-hole of 40µm
average diameter and 10 aspect ratio could be machined.
Keywords: Electrical discharge machining (EDM); Taguchi method; Grey relational analysis
----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

parameters on the performance of micro-EDM with an elec-

1. Introduction
trode of WC in producing quality micro-holes in both transis-
Electrical discharge machining (EDM) is one of the most tor- and RC-type generators [9]. Research into micro holes
extensively used non-conventional material removal processes. with aspect ratios above 10 [9-11] also has been conducted
Its unique feature, that is, the use of thermal energy to ma- over the years. However, the average diameter of the ma-
chine electrically conductive parts regardless of hardness has chined micro holes is above 80µm, and therefore to obtain
been its distinctive advantage in the manufacture of moulds, specimens of smaller diameter and higher aspect ratio, the
dies, automotive, as well as aerospace and surgical compo- machining parameters must be optimized. The Taguchi
nents. The growing popularity of micro-EDM can be attrib- method has been widely employed in several industrial fields
uted to advantages including low set-up cost, high-aspect-ratio and relevant research work. Liao et al. used the Taguchi
parts, enhanced precision, and significant design freedom [1]. method to determine the optimal parameter settings in Wire-
Another characteristic of EDM is the lack of direct contact EDM [12]. Lin et al. adopted the method to obtain the optimal
between the electrode and the workpiece, thus eliminating machining parameters of a hybrid process of EDM incorporat-
mechanical stress, chatter and vibration during machining [2]. ing ball-burnish machining [13]. The original Taguchi method
Masuzawa fabricated micro pins, micro nozzles and micro was designed to optimize only a single performance character-
pipes using EDM [3-6] and Allen used micro-EDM technol- istic. The Grey relational analysis theory posited by Deng [14]
ogy to manufacture ink jet nozzles [7]. Liu presented a process has been proved to effectively resolve the complicated interre-
using micro-EDM combined with high-frequency dither lationships among multiple performance characteristics of the
grinding (HFDG) to improve the surface roughness of micro- EDM process [15-17]. Specifically, Deng, using EDM, opti-
holes [8]. Jahan investigated the influence of major operating mized the material removal rate (MRR), the surface roughness
†
(SR) and the tool wear ratio (TWR) [15], and later, for the
This paper was recommended for publication in revised form by Associate Editor
Dae-Eun Kim machining of Al-10%SiCp by EDM, the MRR, TWR, SR,
*
Corresponding author. Tel.: +82 2 2210 2403, Fax.: +82 2 2210 5575 taper (T) and radial overcut (ROC) were likewise optimized
E-mail address: [email protected] [16].
© KSME & Springer 2010
1084 J. H. Jung and W. T. Kwon / Journal of Mechanical Science and Technology 24 (5) (2010) 1083~1090

The goal of the present study was to determine the optimal

machining parameters for formation of micro-holes of mini-
mum diameter and maximum aspect ratio under given ma-
chining conditions. The Taguchi method was employed to
elucidate the effect of the machining parameters on the char-
acteristics of the EDM process. Additionally, Grey relational
analysis was used to find the optimal machining parameters
satisfying the multiple characteristics of the EDM process.

2. Machining of micro-hole using EDM

2.1 Experimental set-up
Fig. 1. Measurement of electrode diameter using image processing
In a home-made EDM machine used in this study, three lin-
technology (σ = 2.53µm).
ear DC motor (Dover, Trilogy LM-210) and a controller
(Computer, CM233AX) were installed to control displace-
ment in each of the X, Y and Z directions, respectively. An
encoder (Heidenhain LIP401A) with the resolution of
0.1m/count was mounted on each axis to relay high-precision
displacement information to the controller. A smart motor
(Antimatics Inc. 2315D DC brushless type) along with a static
pneumatic bearing were installed in the spindle system in or-
der to achieve 0.05µm rotation precision in the radial direction
and up to 3000rpm rotational spindle speed. Removal of de-
bris generated during the machining process being indispensi-
ble to maintaining smooth processing during EDM machining,
the electrode rotation technique was selected over the ultra-
sonic vibration and internal/external flushing debris-removal-
technology alternatives. We chose an RC circuit to provide
discharge energy, owing to the attendant ease of control of
discharge energy using voltage and capacitance [9]. Positive- Fig. 2. Measured and calculated diameter along electrode.
polarity machining by means of a negative electrical source
connected to the electrode and a positive electrical source Taking this phenomenon and the depth of the hole (400µm)
connected to the work material, was utilized. This type of into consideration, the length of the electrode was determined
machining allowed us a higher machining rate along with to be 1300µm. Whereas the manufacturing tolerance of the
lesser electrode wear [11]. The electrode was produced by electrode is not a significant issue in general EDM processing,
WEDM (wire-EDM) machining. An induction motor (SPG in micro-EDM processing it certainly is. An automatic image-
cop. S6106GB-V12) was installed to transfer the brass wire at processing method was developed to provide the necessarily
a constant speed of 40mm/min.. Also a constant tension of precise electrode diameter measurement. In this method, first,
3.5N was maintained during electrode machining using a hys- the image of the electrode is captured by microscope, after
teresis clutch break PHT-1 (Ogura clutch cop.) to minimize which image- processing technology is utilized to detect the
the deviation of the wire from the trajectory. electrode edge. With the least square method, two straight
lines off the edge in the axial direction are drawn, and then the
2.2 Machining of electrode center line is obtained by taking the average of the two straight
The electrode, as already stated, was produced by WEDM lines. By means of a line drawn perpendicular to the center
machining [3]. Reduction of the electrode diameter due to line, the diameter of the electrode can be determined. Fig. 1
wire wear was minimized by means of the constant speed of shows the measured diameter at several points of the electrode
wire transfer and consistent wire tension just noted. The WC along the rotational axis, and Fig. 2 shows both the measured
electrode, of 50mm length and 300µm diameter was machined diameter and the calculated least square diameter based on
into a smaller diameter electrode using WEDM processing. A same. We then proceeded to find the optimal machining con-
relatively large discharge energy (>18µJ) was used in rough ditions under which the electrode, using the Taguchi method,
formation of the electrode, followed by the finishing process- can be manufactured as quickly as possible. Several tens of
ing utilizing a much smaller energy (<4µJ). The electrode experiments showed that the accuracy of the electrode is in-
wear in the axial direction with kerosene as the dielectric fluid fluenced more by the experience of the worker than by the
was twice that with de-ionized water as dielectric fluid [10]. processing parameters.
 J. H. Jung and W. T. Kwon / Journal of Mechanical Science and Technology 24 (5) (2010) 1083~1090 1085

Table 1. Machining condition for EDM processing. electrode of 30-36µm diameter for use in our experiments.
Material WC
Electrode 2.3 Machining of micro hole
Diameter (µm) 30 ~ 36
In the course of the EDM process, discharging occurs when
Material SS 304
Work material the electrode and the work material are closed enough to in-
Thickness (µm) 400 duce a spark. If the electrode touches the work material during
Dielectric fluid Kerosene machining, meaning that sparks have not been induced, the z
axis of the machine retreats backward to make a gap. The
Voltage (V) 60, 110
voltage difference between the electrode and the work mate-
680, 1300,
Capacitance (pF) rial is then measured to determine whether the electrode and
2000
the work material have been short-circuited or not. After the
Experimental 500, 1500,
condition
Resistance (Ω) machining is completed, the wear length of the electrode is
3000
measured by touching it to the un-machined surface of the
Feed rate (µm/s) 1, 1.5, 2
work material and calculating the difference in the z axis be-
1500, fore and after machining. When the diameter of the micro-
RPM
2200, 2900 hole to be machined is over 70-80µm, the machining time and
electrode wear are shorter when the dielectric fluid is water.
Table 2. Experimenter’s log and experiment results. However, for a smaller micro-hole, kerosene is more effective
Tool En- [10], which is confirmed in our experiment.
Resis Elec- Exit
Volt Capaci- Feed rota- trance
No tance Time trode clear- No. of
age tance rate tional clear-
. (s) Wear ance Shorts 3. Experiments and results
(V) (pF) (µm/s) speed ance
(Ω) (µm) (µm)
(RPM) (µm)
3.1 Experimental condition
1 60 680 500 1 1500 3480 431 6.57 1.26 321
2 60 680 1500 1.5 2200 1800 468 5.35 1.65 53 The purpose of the experiments was to form a micro hole of
3 60 680 3000 2 2900 1440 457 7.105 3.415 19 maximum aspect ratio and minimum diameter. From the result
4 60 1300 500 1 2200 1860 307 7.29 3.09 64 of our preliminary experiments, we settled on the use of a
5 60 1300 1500 1.5 2900 1500 378 6.625 2.575 22 400µm thickness SS304 work material to machine a maxi-
mum aspect ratio micro-hole. WC of 30-36µm diameter and
6 60 1300 3000 2 1500 1800 546 7.99 1.98 70
1300µm length was manufactured for use as an electrode. To
7 60 2000 500 1.5 1500 2640 362 6.335 0.685 275
determine the machining conditions suitable for a hole of
8 60 2000 1500 2 2200 2040 522 6.78 0.63 130
maximum aspect ratio and minimum diameter machined with
9 60 2000 3000 1 2900 1920 474 6.335 1.145 21 a home-made electrode, the Taguchi method was used. The
10 110 680 500 2 1500 1500 479 11.465 3.155 17 selected machining parameters, listed in Table 1, were voltage,
11 110 680 1500 1 1500 2100 542 10.285 0.495 30 capacitance, resistance, feed rate and spindle speed. An L18
12 110 680 3000 1.5 2200 1560 445 10.105 3.055 20 (21x37) orthogonal array was selected to determine the 18 trial
13 110 1300 500 1.5 2900 1980 605 9.815 1.045 46 conditions and their results, tabulated in Table 2. Two levels
14 110 1300 1500 2 1500 1620 554 11.8 1.68 27 of voltage were selected along with three levels of capacitance,
15 110 1300 3000 1 2200 2040 515 12.085 0.495 20
resistance, feed rate and spindle speed.
16 110 2000 500 2 2200 2220 767 11.205 1.145 81
17 110 2000 1500 1 2900 2220 639 11.08 5.4 9
3.2 Result of experiment
18 110 2000 3000 1.5 1500 1800 518 12.58 3.51 42 The results of the experiment are illustrated in Fig. 3, and
the quantitative results for electrode wear, entrance clearance,
It has also been revealed that the most important factor in exit clearance, machin ing time and number of shorts are listed
machining the smallest hole with the maximum aspect ratio is in Table 3. Entrance clearance indicated the difference be-
the diameter of the electrode. It is generally accepted that tween the radius of the electrode and the machined micro-hole
minimal-diameter micro-hole requires an even smaller- at the micro-hole entrance. Exit clearance, correspondingly, is
diameter electrode. However, there is a limitation of the di- the difference between the radius of the electrode and the ma-
ameter on how small the diameter of an electrode can be. Ac- chined hole at the micro-hole exit. Accordingly, we wanted
cording to our preliminary experiments, an electrode with the these two clearances to be as tight as possible so as to machine
diameter below 25-30µm could not stand the harsh environ- the smallest micro-hole. In the ideal case, this value would be
ment of EDM processing, and accordingly would break. 0, meaning that the diameters of the electrode and the micro
Breakage may be attributed specially to the collision of an hole are identical.
electrode with the work material. Hence, we manufactured an
1086 J. H. Jung and W. T. Kwon / Journal of Mechanical Science and Technology 24 (5) (2010) 1083~1090

Table 3. Experimental results by Taguchi method.

Electrode Entrance Exit

Machining No. of
wear clearance clearance
time (s) Shorts
(µm) (µm) (µm)
Voltage 60 438 6.7 1.8 2053 108
(V) 110 563 10.7 2.2 1893 32
680 470 8.5 2.2 1980 77
Capaci-
tance 1300 484 9.3 1.8 1800 42
(pF)
2000 547 9.1 2.1 2140 93
500 492 8.8 1.7 2280 134
Resistance
1500 517 8.7 2.1 1880 45
(Ω)
3000 493 9.4 2.3 1760 32
1 485 8.9 2.0 2270 78
Feedrate
1.5 463 8.5 2.1 1880 76
(µm/s)
2 554 9.4 2.0 1770 57
1500 492 9.3 1.6 2240 128
RPM 2200 504 8.8 1.7 1920 61
2900 505 8.7 2.8 1760 22

Table 4. Electrical characteristics according to variation of voltage,

capacitance and resistance.

Peak Discharge Discharge

current power energy
(I) (W) (µJ)
Voltage 60 3.78 0.59 2.39
(V) 110 5.77 1.98 8.03
680 3.50 1.28 2.67
Capaci
tance 1300 4.83 1.28 5.10
(pF)
2000 5.99 1.29 7.85
500 4.77 2.55 5.21
Resis
tance 1500 4.77 0.87 5.21
(Ω)
3000 4.77 0.43 5.21

3.2.1 Analysis of the results according to the input voltage

variation
When the voltage increased from 60V to 110V, the elec-
trode wear increased from 438µm to 563µm. For the sacrifice
of electrode wear, the machining time was reduced from
2053sec. to 1893sec. Both the entrance and exit clearance
increased at the same time. The number of shorts changed
from 108 to 32. The discharge energy in each case increased
from 2.39µJ to 8.03µJ, as indicated in Table 4. With the same
input voltage increase (from 60V to 110V), the electrode wear,
the entrance and exit clearances and the discharge energy all Fig. 3. Plots of factor effects.
increased, while the machining time and number of shorts
decreased. Those phenomena could be attributed to the in- clearances, while the successful removal of the debris effected
creased input voltage that led to the increased discharge en- decreases in the machining time and the number of shorts.
ergy. The increased discharge energy resulted in a larger
sparks, which removed the debris generated during the ma- 3.2.2 Analysis of the results according to the capacitance
chining quite effectively. The higher discharge energy resulted variation
in accelerated electrode wear and increased entrance and exit As the capacitance increased from 680pF to 2000pF, the
 J. H. Jung and W. T. Kwon / Journal of Mechanical Science and Technology 24 (5) (2010) 1083~1090 1087

electrode wear and the machining time rose from 470µm to

547µm and from 1980sec. to 2140sec., respectively. The en- 3.2.5 Analysis of results according to the spindle speed varia-
trance and exit clearances and the number of shorts remained tion
almost the same. The increased electrode wear could be at- It is known that the spindle speed has one of the most sig-
tributed to the increased discharge energy. nificant effects on the removal of debris generated in the EDM
process [18]. In the present experimentation, as the spindle
3.2.3 Analysis of the results according to the resistance speed increased from 1500, via 2200, to 2900rpm, the number
variation of shorts decreased from 128, via 61, to 22. The machining
The machining parameters most influenced by the variation time also decreased from 2240, via 1920, to 1760sec.. As the
of the resistance were the machining time and the number of spindle speed increased, the debris was removed more quickly,
shorts. As the resistance increased from 500Ω to 3000Ω, the which led to a reduced number of shorts and a shorter machin-
machining time and the number of shorts decreased from ing time. The electrode wear and the entrance clearance re-
2280sec. to 1760sec. and from 134 to 32, respectively. As can mained almost identical. The exit clearance increased from
be seen in Table 4’s results, as the resistance increased, so too 1.6µm to 2.8µm.
did the discharge power decrease. This decrement of the dis-
charge energy caused a reduction in the debris size, which led 3.2.6 Relationship between electrical characteristics and
to a reduction in the number of shorts and in the machining electrical parameter
time. The variation of the electrode wear in the axial direction Eqs. (1), (2), (3) show the relationships between electrical
as was that of the entrance and exit clearances, was negligible. characteristics and input voltage, capacitance and resistance,
respectively [17]. The calculated results are given in Table 4.
3.2.4 Analysis of the results according to feed rate variation Vgap and L2 (inductance) are 26.5V and 0.2µH, respectively.
The feed rate variation greatly affected the electrode wear V − Vgap
and the machining time. When the feed rate changed from Ip ≈ (1)
L2 / C
1µm/sec. to 1.5µm/sec., the machining time was reduced from
1 1
2270sec. to 1880sec. and the electrode wear decreased from W= CV 2 (2)
485µm to 463µm. The 1.5µm/sec. feed rate showed a better 2 − R1 C ln (1 − Vd / V ) + π L2 C
result in both machining time and electrode wear. This meant 1
Eunit , RC = CV 2 (3)
that the interval between discharges was long enough to re- 2
charge the circuit when the feed rate was either 1µm/sec. or
3.3 Determination of the machining condition
1.5µm/sec. As a result, the increment of the feed rate did not
change the discharge energy but only reduced the machining In the previous section, it was shown that the electrode wear
time. The electrode wear did not change significantly in this and entrance and exit clearances are directly related to the size
case. By contrast, as the feed rate increased from 1.5µm/sec. of the micro-hole in 5 of machining process characteristics. The
to 2µm/sec., the electrode wear increased and the machining machining time and the number of shorts does not show a sig-
time decreased. An increment of the feed rate to 2µm/sec. nificant relation to the size of the micro-hole, which, rather, is
resulted in their being insufficient time for recharge of the related to productivity. As a result, the electrode wear along
circuit. Because of the insufficient amount of energy resulting with the entrance and exit clearances were determined to be the
from the insufficient recharging time, there was an insufficient parameters to optimize in order to minimize the hole size. We
amount of discharge energy compared with the case for the had to determine the machining parameters such the input volt-
feed rates of 1µm/sec. and 1.5µm/sec. Consequently, we age, capacitance, resistance, the feed rate and the spindle speed
needed to increase the number of discharges to finish the ma- to optimize the electrode wear and entrance and exit clearance
chining. And due to the enhanced number of discharges, the characteristics. It is a complicated procedure to determine 5
electrode wear also increased. machining parameters for optimizing 3 characteristics. To over-
The discharge energy decrement and the feed rate incre- come the difficulty, Grey relational analysis was employed [14].
ment had opposite effect on the machining time. When the
discharge energy decreased, the machining time increased 3.3.1 Grey relational analysis
accordingly. But when the feed rate increased, the machining The original reference sequence and the sequence for com-
time decreased. In this experiment, the effect of the increased parison can be represented as xo (k) and xi (k), i=1, 2, ... , m;
feed rate was larger than that of the decreased discharge en- k=1, 2, ..., n respectively. Here, m is the total number of ex-
ergy. Resultantly, the machining time and the number of periments, while n is the total number of observation data. If
shorts decreased as the feed rate increased to 2.0µm/sec. and the target value of the original sequence is 'the-larger-the-
the discharge energy decreased. This meant that even though better', then the original sequence is normalized according to
the discharge energy was reduced because of the increased
xio (k ) − min xio (k )
feed rate, it was yet sufficient speed of the electrode to induce xi* (k ) = (4)
max xio (k ) − min xio (k )
a spark between the electrode and the work material.
1088 J. H. Jung and W. T. Kwon / Journal of Mechanical Science and Technology 24 (5) (2010) 1083~1090

If it is 'the-smaller-the-better', then the original sequence is Table 5. Experimental results after normalization process (Reference
normalized as sequence for all parameters is 1).

Electrode Entrance Exit

max xio (k ) − xio (k ) wear clearance clearance
xi* (k ) = (5)
max xio (k ) − min xio (k ) 1 0.7304 0.8313 0.8440
2 0.6500 1.0000 0.7645
In the case where there is 'a specific value', the original se- 3 0.6739 0.7573 0.4047
quence is normalized as 4 1.0000 0.7317 0.4709
5 0.8457 0.8237 0.5759
| xio ( k ) − OB |
xi* (k ) = 1 − (6) 6 0.4804 0.6349 0.6972
max{max xi (k ) − OB; OB − min xio (k )}
o

7 0.8804 0.8638 0.9613

Here, OB is the target value. 8 0.5326 0.8022 0.9725

9 0.6370 0.8638 0.8675
3.3.2 Grey relational coefficient and Grey relational grades 10 0.6261 0.1542 0.4577
11 0.4891 0.3174 1.0000
The Grey relational coefficient is defined as follows [14]
12 0.7000 0.3423 0.4781
∆ + ζ∆ max 13 0.3522 0.3824 0.8879
γ ( xo* ( k ), xi* (k )) = min
∆ oi (k ) + ζ∆ max (7) 14 0.4630 0.1079 0.7584

0 < γ ( x ( k ), x (k )) ≤ 1
* * 15 0.5478 0.0685 1.0000
o i
16 0.0000 0.1902 0.8675
Δ oi (k) is the deviation sequence of the reference sequence x*0 17 0.2783 0.2075 0.0000
(k) and the comparability sequence x*i (k), that is: 18 0.5413 0.0000 0.3853

∆ oi (k ) =| xo* (k ) − xi* (k ) | Table 6. Calculated Grey relational coefficient and grade.

∆ max = max max | x (k ) − x ( k ) |

*
o
*
j (8) Grey relational coefficient
∀j∈i ∀k
Electrode Entrance Exit Grey rela-
∆ max = min min | x (k ) − x (k ) |
* *
tional grade
∀j∈i ∀k
o j wear clearance clearance
1 0.6497 0.7477 0.7622 0.7196
where ζ is the distinguishing coefficient, ζ ∈ [0,1] . 2 0.5882 1.0000 0.6798 0.7556
A Grey relational grade is a weighted sum of the Grey coef- 3 0.6053 0.6732 0.4565 0.5781
ficient, and is defined as 4 1.0000 0.6508 0.4859 0.7119

n
5 0.7641 0.7393 0.5411 0.6812
γ ( xo* , xi* ) = ∑ β k γ ( xo* ( k ), xi* (k )) 6 0.4904 0.5779 0.6229 0.5635
k =1
n
(9) 7 0.8070 0.7859 0.9281 0.8400
∑β
k =1
k =1 8 0.5169 0.7166 0.9478 0.7268
9 0.5793 0.7859 0.7905 0.7183
10 0.5721 0.3715 0.4797 0.4743
Here the Grey relational grade γ ( x , x ) represents the level
*
o
*
i
11 0.4946 0.4228 1.0000 0.6390
of correlation between the reference and the comparability
sequence. If a particular comparability sequence is more im- 12 0.6250 0.4319 0.4893 0.5152
portant to the reference sequence than the other comparability 13 0.4356 0.4474 0.8168 0.5664
sequence, the Grey relational grade for that comparability 14 0.4822 0.3592 0.6742 0.5050
sequence and the reference sequence will exceed that for the 15 0.5251 0.3493 1.0000 0.6246
other Grey relational grades. 16 0.3333 0.3817 0.7905 0.5017
17 0.4093 0.3868 0.3333 0.3763
3.3.3 Evaluated Grey relational coefficient and grade 18 0.5215 0.3333 0.4486 0.4343
The experimental results for the electrode wear, the en-
trance clearance and the exit clearance are listed in Table 2. tional grade. The results are tabulated in Table 6. In this case,
Since smaller values for those parameters were desirable, the the reference sequences were set to 1, that is, xi* (k)=1, and the
data sequence had ‘the-smaller-the-better’ characteristic. distinguishing coefficient ζ was set to 0.5.
Hence, Eq. (5) is employed for the data processing, the results
of which are listed in Table 5. Eq. (7) was utilized to deter- 3.3.4 ANOVA results for Grey grade
mine the Grey relational coefficient, and Eq. (9) the Grey rela- Table 7 lists the results of the analysis of variance
 J. H. Jung and W. T. Kwon / Journal of Mechanical Science and Technology 24 (5) (2010) 1083~1090 1089

Table 7. Results of analysis of variance (ANOVA).

Level Level Level Sum of Mean Contri-

Factor DOF F value
1 2 3 square square bution
Voltage 0.700 0.515 1 1.158 1.158 97.00 66.29
Capaci-
1.238 0.609 0.600 2 1.154 0.577 48.32 33.02
tance
Resis-
0.636 0.614 0.573 2 0.006 + 0.003
tance
Feed
0.632 0.632 0.558 2 0.011+ 0.005
rate (a) 45µm hole entrance (b) 35µm hole exit
Cutting
0.578 0.60 0.644 2 0.007+ 0.003 Fig. 4. Micro-hole machined under selected optimal condition (kero-
speed
sene, 60V, 680pF, 500Ω, 1.5µm/s, 1500RPM).
Error (6) (0.024) 0.012

Total 9 2.336 1.747 99.32

+
Indicates the sum of squares added together to form the pooled error
sum of squares shown in parentheses

(ANOVA) for the electrode wear, entrance clearance and exit

clearance using the calculated values for the Grey relational
coefficients and Grey relational grades of Table 6. The Table
7 figures show that the contributions of the voltage and ca-
pacitance were 66.29 and 33.02, respectively. These two were
found to be the most significant controlling parameters, mean-
ing that they controlled the electrode wear, entrance clearance
and exit clearance simultaneously and very effectively.

3.3.5 Determination of process parameters

According to the ANOVA analysis of the Grey relational
grades discussed in 3.3.4, the most significant factors were the Fig. 5. Number of shorts and position of electrode in z direction for
voltage and capacitance, the values for which were 60V and length of time.
680pF, respectively. Since the remaining parameters were
insignificant, their values were taken from Table 3: resistance conditions for drilling of a micro-hole of minimum diameter
500Ω, feed rate 1.5µm/s, spindle speed 1500RPM. To prove and maximum aspect ratio. The Taguchi method was em-
the effectiveness of the selected optimal machining conditions, ployed to determine the relations between the machining pa-
a confirmation experiment was executed using an electrode rameters and the process characteristics. The machining pa-
with a diameter of 30.6µm. It was shown that the machined rameters affecting the hole diameter were revealed by execut-
micro-hole had an average diameter of 40µm and an aspect ing 18 experiments. It was found that the electrode wear and
ratio of 10. The entrance and exit diameter were 45µm and the entrance and exit clearances have a significant effect on
35µm respectively, as shown in Fig. 4. The electrode wear the diameter of the micro-hole when the diameter of the elec-
was 273µm, while the entrance and exit clearances were trode is identical. To determine the machining parameters
7.49µm and 2.06µm, respectively. Compared with the Table 3 affecting the electrode wear and the entrance and exit clear-
results, the electrode wear was the smallest, while the entrance ances, Grey relational analysis was used. The input voltage
clearance was close to the smallest. The exit clearance did not and the capacitance were found to be the most significant
show any significant improvement. This was natural, consid- controlling parameters. The obtained optimal machining con-
ering that the average and standard deviations of exit clear- dition were an input voltage of 60V, a capacitance of 680pF, a
ance are 2.03µm and 0.31µm, respectively. The number of resistance of 500Ω, a feed rate of 1.5µm/s and a spindle speed
shorts in the z direction of the electrode are indicated in Fig. 5. of 1500rpm. Under these conditions, a micro-hole of 40µm
As can be seen, the number linearly increased as the depth of average diameter and an aspect ratio of 10 could be machined.
the hole increased. This speaks to the consistency of the EDM
process in forming the smallest microhole under the given Acknowledgement
conditions.
This work was supported by Nuclear Research & Devel-
opment Program of the Korea Science and Engineering Foun-
4. Conclusions dation (KOSEF) grant funded by the Korean government
In this study, we attempted to find the optimal machining (MEST) (grant code: M2009-0078442).
1090 J. H. Jung and W. T. Kwon / Journal of Mechanical Science and Technology 24 (5) (2010) 1083~1090

[13] Y. C. Lin, B. H. Yan and F. Y. Huang, Surface modifica-

References
tion of Al-Zn-Mg aluminum alloy using combined processes
[1] H. S. Lim, Y. S. Wong, M. Rahman and M. K. Edwin Lee, of EDM with USM, J. Mater. Process. Technol., 115 (2001)
A study on the machining of high-aspect ratio micro- 359-366.
structure using micro-EDM, J. Mater. Process. Technol., [14] J. Deng, Introduction to Grey system, J. of Grey System, 1
140 (2003) 318-325. (1) (1989) 1-24.
[2] K. H. Ho and S. T. Newman, State of the art electrical dis- [15] J. L. Lin and C. L. Lin, The use of the orthogonal array
charge machining (EDM), Int. J. Mach. Tools Manuf., 43 with Grey relational analysis to optimize the electrical dis-
(2003) 1287-1300. charge machining process with multiple performance charac-
[3] T. Masuzawa, M. Fujino and K. Kobayashi, Wire Electro- teristics, Int. J. Mach. Tools Manuf., 42 (2002) 237-244.
discharge grinding for Micro-machining, Ann CIRP., 34 (1) [16] P. N. Singh, K. Raghukandan and B. C. Pai, Optimization
(1985) 431-434. by Grey relational analysis of EDM parameters on machin-
[4] T. Masuzawa, C. L. Kuo and M. Fujino, A combined Elec- ing Al-10% SiCp composites, J. Mater. Process. Technol.,
trical Machining Process for Micro Nozzle Fabrication, Ann 155-156 (2004) 1658-1661.
CIRP., 43 (1) (1994) 189-192. [17] P. J. Cho, Characteristics of RC circuit with Transistor in
[5] Z. Y. Yu, T. Masuzawa and M. Fujino, Micro-EDM for Micro-EDM, Seoul National University (2003) Ph. D. Dis-
Three-dimensional Cavities-development of Uniform wear sertation.
method, Ann CIRP., 47 (1) (1998) 169-172. [18] J. S. Soni and G Chakraverti, Machining characteristics of
[6] C. L. Kuo and T. Masuzawa, a Micro-pipe fabrication proc- titanium with rotary electro-discharge machining, Wear, 171
ess, Proc. IEEE MEMS ’91, 80-85. (1994) 51-58.
[7] D. M. Allen and A. Lecheheb, Micro Electo-discharge Ma-
chining of Ink Jet Nozzle: Optimum Selection of Material Jong Hyuk Jung received his B.S. and
and Machining Parameter, J. Mat. Proc. Tech., 58 (1996) M.S in Mechanical Information Engi-
53-63. neering from University of Seoul. He is
[8] H. S. Liu, B. H. Yan, C. L. Chen and F. Y. Huang, Applica- currently an associate director in In-
tion of micro-EDM combined with high-frequency dither dustrial Application R&D Institute in
grinding to micro-hole machining, Int. J. Mach. Tools DAEWOO SHIPBUILDING & MA-
Manuf., 46 (2006) 80-87. RINE ENGINEERING. He is involved
[9] M. P. Jahan, Y. S. Wong and M. Rahman, A study on the in developing robots for productivity
quality micro-hole machining of tungsten carbide micro- improvement and systems for welding and painting automa-
EDM process using transistor and RC-type pulse generator, tion.
Int. Mater. Proc. Tech., 209 (2009) 1706-1716.
[10] D. W. Seo, M. S. Park, S. M. Yi and C. N. Chu, Machining Won Tae Kwon received his B.S., M.S.
Characteristics of Micro-EDMed holes According to Dielec- degree from Seoul National University,
tric Fluid, Capacitance and Ultrasonic Vibration, J. Korean and Ph.D. degree from Northwestern
Soc. Precision Eng., 24 (2007) 42-29. University in 1982, 1984 and 1992,
[11] G. M. Kim, B. H. Kim and C. N. Chu, Machining Rate and respectively. Prof. Kwon is currently a
Electrode Wear Characteristics in Micro-EDM of Micro- Professor of the Department of Me-
hole, J. Korean Soc. Precision Eng., 16 (1999) 94-100. chanical and Information Engineering of
[12] Y. S. Liao, J. T. Huang and H. C. Su, A study on the ma- University of Seoul, Korea. His research
chining-parameters optimization of wire electrical discharge fields include micro machining and machining process opti-
machining, J. Mater. Process. Technol., 71 (1997) 487-493. mization.