wDry

ARTICLE IN PRESS

International Journal of Machine Tools & Manufacture 50 (2010) 431–443

Contents lists available at ScienceDirect

International Journal of Machine Tools & Manufacture

0890-69

doi:10.1

n Corr

E-m

journal homepage: www.elsevier.com/locate/ijmactool

Experimental characterization of material removal in dry electricaldischarge drilling

P. Govindan, Suhas S. Joshi n

Department of Mechanical Engineering, Indian Institute of Technology Bombay, Mumbai – 400 076, India

a r t i c l e i n f o

Article history:

Received 3 November 2009

Received in revised form

10 February 2010

Accepted 11 February 2010Available online 21 February 2010

Keywords:

Dry EDM

Material removal rate

Tool wear rate

Dimensional accuracy

Single discharge

Taguchi methods

55/$ - see front matter & 2010 Elsevier Ltd. A

016/j.ijmachtools.2010.02.004

esponding author. Tel.: +91 22 25767527.

ail address: [email protected] (S.S. Joshi).

a b s t r a c t

Dry electrical discharge machining is one of the novel EDM variants, which uses gas as dielectric fluid.

Experimental characterization of material removal in dry electrical discharge drilling technique is

presented in this paper. It is based on six-factor, three-level experiment using L27 orthogonal array. All

the experiments were performed in a ‘quasi-explosion’ mode by controlling pulse ‘off-time’ so as to

maximize the material removal rate (MRR). Furthermore, an enclosure was provided around the

electrodes with the aim to create a back pressure thereby restricting expansion of the plasma in the dry

EDM process. The main response variables analyzed in this work were MRR, tool wear rate (TWR),

oversize and compositional variation across the machined cross-sections. Statistical analysis of the

results show that discharge current (I), gap voltage (V) and rotational speed (N) significantly influence

MRR. TWR was found close to zero in most of the experiments. A predominant deposition of melted and

eroded work material on the electrode surface instead of tool wear was evident. Compositional

variation in the machined surface has been analyzed using EDAX; it showed migration of tool and

shielding material into the work material. The study also analyzed erosion characteristics of a single-

discharge in the dry EDM process vis-a-vis the conventional liquid dielectric EDM. It was observed that

at low discharge energies, single-discharge in dry EDM could give larger MRR and crater radius as

compared to that of the conventional liquid dielectric EDM.

& 2010 Elsevier Ltd. All rights reserved.

1. Introduction

Dry EDM is one of the novel EDM techniques, which employs gasas a dielectric medium instead of liquid. It mainly involves supply ofa gas through a rotating thin-walled pipe electrode, which alsoflushes out the debris from inter-electrode gap [1]. The process holdsthe potential to be a viable alternative to conventional liquiddielectric EDM for precision oriented machining applications. Themajor advantage of this method is its simplicity.

In conventional EDM, in addition to the problems associated witha complex dielectric supply system, a kerosene or oil-baseddielectric causes carbon deposition over the machined surfaces. Onthe other hand, use of water-based dielectric leads to formation ofcracks, electrolysis and corrosion of the EDMed surface [2,3]. Theforce generated after the dielectric breakdown in EDM due toexpansion and contraction of a bubble formed by evaporation,dissociation and ionization of dielectric liquid and electrodematerials is higher in liquid as compared to gas, since expansionof the bubble is prevented by greater inertia and viscosity of thedielectric liquid. Since the electrical permittivity of the liquid

ll rights reserved.

medium is higher than the gas medium, the electrostatic force ishigher as well [4]. It is known that the liquid dielectric hasdifficulties in accessing the sparking region, besides its fumes arehazardous to the environment.

Mist EDM, a new fine finishing EDM process [5], also causes areduction in efficiency, high tool wear, pollution, difficulty incleaning and piping required for mist circulation. Another emergingtechnology, viz. powder mixed EDM, increases the cost of machiningand is environment-unfriendly [6]. Dry EDM technology overcomessome of these demerits and has a few additional advantages such asformation of thinner white layer [7,8], low dielectric constantresulting in easy breaking of dielectric and formation of plasma,lesser viscosity of gas and lower heat concentration causing betterdebris removal and flushing.

At present, the investigations related to dry EDM process aremainly aimed at demonstrating the feasibility of the process andimproving the basic process outputs. In this regard, Kunieda et al. [1]has demonstrated dry EDM process in machining of steel, whereoxygen gas was used as a dielectric medium. Furthermore, apotential method to enhance MRR by controlling pulse-off-time,using ‘quasi-explosion mode’ was proposed in Ref. [7]. Yu et al. [9]belonging to the same research group applied dry EDM milling tomachine ‘difficult-to-cut’ cemented carbides. They found that theprocess reduces machining time and is cost-effective. Zhang et al.

www.elsevier.com/locate/ijmactool

dx.doi.org/10.1016/j.ijmachtools.2010.02.004

mailto:[email protected]

ARTICLE IN PRESS

P. Govindan, S.S. Joshi / International Journal of Machine Tools & Manufacture 50 (2010) 431–443432

[10–12] employed ultrasonic vibrations to improve MRR of theprocess and introduced surface roughness as a response variable. Inthese investigations, wall thickness of the electrode and amplitudeof ultrasonic vibration were identified as two additional inputparameters. In addition, the material removal mechanisms inultrasonic assisted dry EDM of cemented carbides were studied inRef. [13]. An investigation involving control of gap length in dry EDMusing piezoelectric actuator was conducted by Kunieda et al. [14],which improved MRR as well as the process stability. Kao et al. [5]developed near dry EDM using liquid–gas mixture as a dielectricfluid for machining aluminium. Moreover, Tao et al. [3] explored thecapability of dry EDM for better MRR and near-dry EDM forfine surface finish, and found that among the few electrodematerial-dielectric medium combinations, copper–oxygen was thebest for dry EDM and graphite with water–nitrogen mixture for thenear-dry EDM. In a recent work by Saha and Choudhury [15], a set ofcentral composite design (CCD) experiments were conducted. Theyfound that discharge current, duty factor, air pressure and spindlespeed significantly influenced MRR, and all the parameters exceptgap voltage influenced the average surface roughness (Ra).

Nevertheless, one of the major unresolved issues related to dryEDM is the low MRR. Furthermore, low stability, arcing andmicro-crack formation over the surface, poor surface finish andadherence of debris to the electrode are some other problems thatare yet to be resolved adequately [2,3]. Some researchers haveshown that gas pressure is a significant parameter that influencesthe process performance [1–3,6–15]. However, the gas is releasedinto the atmosphere during the process. In spite of variousresearch works, it appears that no attempts have been made tosolve the problem of uncontrollable plasma expansion in a dryEDM process. It is felt that if in some way, the release of gas can beconstrained to create a back pressure, and hence the plasmaexpansion could be controlled during the process.

Therefore, the objective of this paper is to investigate dry EDM indrilling by employing a shield around the plasma. It is proposed torun the process in the ‘quasi-explosion mode’. The process analysisinvolves a study of effect of processing conditions on MRR, TWR, andaccuracy. Pulse ‘off-time’ and clearance between the shield and theelectrode at the sparking region were used in conjunction with theother regular dry EDM parameters. Also, analysis of the mechanismof material removal has also been carried out by performing single-discharge experiments in dry EDM. The details are elaborated in thefollowing sections of the paper.

2. Experimental work

In order to characterize the dry EDM process qualitatively andquantitatively through an assessment of the influence of input

Oxygen gas su

Shield

Debrisparticle

Dry EDM p

Gap voltage

Discharge current

Pulse off-time

Oxygen pressure

Electrode speed

Shielding clearance

Input variables

Fig. 1. Schematic of the process and theme of

parameters on the machining performance, the following threeapproaches have been used in this work:

(i)

pply

roces

expe

To perform the experiments under well-known ‘quasi-explosion’ conditions [7], by controlling pulse ‘off-time’, soas to maximize the MRR.

(ii)
To provide an enclosure around the electrode, aiming atcreating a back pressure thereby restricting expansion of theplasma. This action is expected to stabilize the process andhelp prevent spark column contamination.
(iii)
To analyze the radius of crater and MRR obtained in a single-discharge in dry EDM and compare it with that of the liquiddielectric EDM.
It is felt that the analysis of the various characteristics of dry EDMprocess shown in Fig. 1 could reveal the performance of the process.Since MRR and TWR are the most commonly analyzed performanceindicators in this work, the dimensional oversize and machinedsurface topography have also been included as response variables.

2.1. Experimental design

In this experimentation, various independent parameters areselected (see Fig. 1) based on the authors’ preliminary experi-mentation and the past literature, as mentioned in Table 1. It wasevident that MRR is higher with reverse polarity (work ispositive). Therefore, all the experiments were performed usingthe reverse polarity. From the physics of the process, three two-factor interactions (i.e. voltage� current, voltage�pulse off-timeand current�pulse off-time) which could influence energy ofspark have been chosen for the experimentation. Accordingly, theDOF of this experiment is 26. Therefore, the L27 OA (orthogonalarray) has been selected for the experimentation. The levels ofparameters and allotment of various factors to various columns ofL27 array based on the linear graph [16] are depicted in the righthand bottom corner of Table 2.

2.2. Experimental set-up and procedure

2.2.1. Experimental set-up

A ELECTRA PS ZNC EDM machine having a programmableZ-axis control, with NC multi-step programming facility with arotary attachment, has been used for conducting the dry EDMdrilling experiments on SS304 (see Table 3 for materialcomposition). A schematic diagram and a photograph of the dryEDM set-up are shown in Fig. 2a,b. Dry EDM set-up consisting ofan oxygen cylinder with a regulator for flow control wasconnected to the copper electrode.

Workpiece

Rotatingelectrode

s

MRR

TWR

Oversize

Machined surface

Approaches 1 and 2

MRR in single discharge

Crater radius in singledischarge

Approach 3

Output variables

rimental work on dry EDM drilling.

ARTICLE IN PRESS

Table 1Experimental parameters, levels and reason for selection.

No. Parameter, units Levels Reason for selection

1 Discharge current (I), A 12, 15, 18 Higher levels of pulse current increases thermal loading and causes damage to the work, therefore lower current

levels selected.

2 Gap voltage (V), V 50, 65, 80 Based on the range available with the machine and within a range in one of the recent investigations [3].

3 Pulse off-time (Toff), ms 22, 33, 67 To satisfy ‘quasi-explosion’ condition. The three levels of pulse ‘off-time’ chosen are: one-third, one-sixth and one-

ninth of pulse ‘on-time’. Approximately one-sixth has been proposed as the range for quasi-explosion mode earlier

[7].

4 Gas pressure (P), MPa 0.15, 0.20,

0.25

Range available with the existing set-up. Similar machining conditions used in few investigations [7,9].

5 Electrode speed (N), rpm 100, 200,

300

Rotation of the tool enhances dielectric flow and stability of the process [1]. However, very high speed also reduces

accuracy [17]. Therefore, the range selected is 100–300 rpm.

6 Radial clearance of shield at

bottom (Cb), mm

4.0, 4.5,

5.0

Due to high mobility of gas molecules, which causes uncontrollable plasma expansion, a shield is provided. An

aluminium shield was provided, having lowest ‘‘rCp’’ factor (2.42) compared to copper (3.44) and iron (3.46).

Table 2L27 array with levels of input parameters, their allocation (including selection of interactions), response variables (MRR, TWR and oversize) and linear graph for

independent variables assignment.

Trial

No.

Input parameters and their levels Response variables (for original trials and replicated trials)

V

(Volts)

I

(A)

Toff

(ms)

P

(MPa)

N

(rpm)

Cb

(mm)

MRR

(mm3/min)

MRRR

(mm3/min)

TWR

(mm3/min)

TWRR

(mm3/min)

Oversize at locations along depth of hole (%)

Entry 50% 90%

OS OSR OS OSR OS OSR

1 50 12 22 0.15 100 4 0.376 0.416 �0.0056 0.0242 �29.29 �41.16 �16.74 �21.54 �2.67 �7.36

2 50 12 33 0.20 200 4.5 0.414 �0.0093 �32.33 �22.66 �6.64

3 50 12 67 0.25 300 5 0.441 0.458 �0.0074 �0.0112 �29.94 �36.83 �17.70 �20.13 �8.15 �2.81

4 50 15 22 0.20 200 5 0.552 �0.0018 �36.23 �23.83 �8.75

5 50 15 33 0.25 300 4 0.620 �0.0056 �38.69 �23.70 �8.53

6 50 15 67 0.15 100 4.5 0.515 0.0000 �34.60 �20.82 �2.76

7 50 18 22 0.25 300 4.5 0.794 0.811 �0.0018 �0.0204 �44.01 �43.33 �23.89 �24.34 �5.02 �3.78

8 50 18 33 0.15 100 5 0.679 0.634 �0.0005 �0.0037 �43.96 �40.53 �27.00 �25.22 �6.31 �7.05

9 50 18 67 0.20 200 4 0.691 �0.0167 �39.86 �24.85 �7.17

10 65 12 22 0.20 300 4.5 0.426 �0.0074 �32.79 �21.08 �5.26

11 65 12 33 0.25 100 5 0.380 0.0000 �31.06 �16.54 �5.91

12 65 12 67 0.15 200 4 0.449 0.0130 �37.92 �26.15 �3.82

13 65 15 22 0.25 100 4 0.517 0.523 0.0167 0.0018 �36.57 �40.39 �14.66 �21.35 �8.88 �4.27

14 65 15 33 0.15 200 4.5 0.513 0.552 �0.0204 �0.0056 �37.11 �39.85 �19.59 �24.67 �3.72 �4.78

15 65 15 67 0.20 300 5 0.578 �0.0093 �37.81 �22.41 �3.63

16 65 18 22 0.15 200 5 0.699 0.729 �0.0037 �0.0093 �40.2 �45.89 �28.42 �27.16 �2.39 �7.9

17 65 18 33 0.20 300 4 0.755 �0.0093 �44.74 �24.19 �2.80

18 65 18 67 0.25 100 4.5 0.679 0.649 0.0093 0.0018 �41.44 �44.74 �21.61 �22.62 �1.79 �3.45

19 80 12 22 0.25 200 5 0.303 0.357 �0.0112 �0.0112 �29.54 �32.37 �16.39 �20.49 �2.81 �5.45

20 80 12 33 0.15 300 4 0.401 0.0056 �35.58 �21.68 �1.48

21 80 12 67 0.20 100 4.5 0.315 0.347 �0.0074 �0.0056 �32.16 �37.13 �20.51 �18.62 0.81 �7.27

22 80 15 22 0.15 300 4.5 0.441 0.454 �0.0074 0.0093 �38.84 �37.03 �21.68 �21.81 �4.84 �4.99

23 80 15 33 0.20 100 5 0.437 �0.0112 �35.62 �20.40 �6.59

24 80 15 67 0.25 200 4 0.586 0.464 0.0093 �0.0018 �43.38 �41.78 �24.13 �23.37 �4.94 �6.07

25 80 18 22 0.20 100 4 0.580 �0.0018 �40.88 �26.25 0.31

26 80 18 33 0.25 200 4.5 0.624 �0.0130 �40.71 �25.52 �7.74

27 80 18 67 0.15 300 5 0.605 �0.0018 �38.32 �23.96 �5.14

Assignment of factors to columns and selection of interactions based on the linear graph

ASSIGNMENT OF FACTORS TO COLUMNS AND SELECTION OF INTERACTIONS BASED ON THE LINEAR GRAPH Col No. 1 2 5 9 10 12

Linear graph and arrangement of independent variables to columns of L27 OA [16]

EMPTY COLUMNS

3, 4, 6, 7, 8, 11, 13

MRR - MRR for original trials , MRRR- MRR for replicated trials, TWR- TWR for original trials, TWRR- TWR for replicated trials, OS- Oversize for original trials,

OSR

-Oversize for replicated trials

I × Toff

1- V

2-I 12- Cb

V × I V × Toff

5-Toff9-P 10-N

P. Govindan, S.S. Joshi / International Journal of Machine Tools & Manufacture 50 (2010) 431–443 433

ARTICLE IN PRESS

Table 3Thermal and physical properties of work piece, tool, dielectric and shield.

Element Specific

heat (J/g K)

Melting

point (K)

Thermal

conductivity (W/m K)

Density Chemical composition (wt%)

SS 304 0.5 1455 16.2 8.03 g/cm3 Cr Ni Mn Si C P S Fe

18.0 8.0 2.0 0.75 0.08 0.04 0.03 Balance

Copper 0.4 1375 401 8.9 g/cm3 100% Cu

Aluminium 0.9 933.2 237 2.7 g/cm3 100% Al

Oxygen 0.92 500.5 0.0267 1.429 g/L 99.9% pure

Exit90% L

10% L

50% L

Entry

ClampingforceDb

Dt Shield

H

Fig. 2. (a)–(e) Experimental set-up for dry EDM: (a) schematic diagram, (b) photograph, (c) SS 304 split work piece in assembly and one-half of the split specimen, (d)

specifications of a section of shield used in experiments (Dt=diameter of the shield at top=9 mm, Db=diameter of the shield at the bottom=13, 14 and 15 mm), (e) typical

hole SEM photograph showing locations of the oversize measurement.


In this study, the work material was designed in the form of a splitspecimen [17,18], and dimensions of each part were: 27 mm�14mm�10 mm (see Fig. 2c). The mating surfaces (interfaces) on thespecimen were ground and polished using waterproof SiC papers ofgrit size varying from 600 to 1200 to achieve parallelism.

Knowing that in dry EDM, oxygen gas and copper electrodeprovides the best option [13,15], thereby they were used inexperiments. Electrodes were prepared by using round coppertubes (OD: 4.75 mm, ID: 3.25 mm) of 60 mm long with polishedflat end faces. A hollow cylindrical aluminium shield having aheight of 15 mm with three different levels of radial clearances (2,2.5 and 3 mm) around the electrode, near the sparking zone, wasused (Fig. 2d). Height H as shown in Fig. 2d is taken as 15 mm inall experiments, based on initial experiments carried out. Heightwas varied between 10 and 20 mm. It was observed that theoptimum value of MRR was at a height of 15 mm. The propertiesand specifications of work, tool, dielectric and shield are alsolisted in Table 3.

2.2.2. Experimental procedure

To make the holes exactly at the intersection of two parts ofthe split work piece, accurate alignment of the electrode withreference to the split workpiece intersection was done. Dryelectrical discharge drilling experiments based on L27 orthogonalarray were performed in a random order. Among the total 27trials, 12 experiments were replicated. Each experiment wascontinued for an hour. After machining, specimens were cleanedusing an ultrasonic cleaner to remove loose debris depositedaround the work surfaces. Fig. 2e shows a scheme of typicalmeasurements of oversize on drilled holes.

The weight loss of workpiece and tool, during the dry EDMprocess, was measured using Sartorius CP 4235 precision scale.Knowing the density of stainless steel 304 (workpiece) and copper(electrode), MRR and TWR are estimated:

MRR ðin mm3=minÞ ¼ðWiÞw�ðWf Þw

rss

�1

Tm� 1000 ð1Þ

ARTICLE IN PRESS


where (Wi)w is the weight of the workpiece before machining, (Wf)w

is the final weight of the workpiece after machining, rss is thedensity of SS304 in g/cm3 and Tm is the machining time in minutes,

TWRðin mm3=minÞ ¼ðWiÞe�ðWf Þe

rcu

�1

Tm� 1000 ð2Þ

where (Wi)e is the weight of the electrode before machining, (Wf)e isthe weight of the electrode after machining, rcu is the density ofcopper in g/cm3 and Tm is the machining time in minutes. The samemethod has been followed for all the experiments.

Depth achieved during the process and outside diameter of thedrilled holes was measured using a Nikon MM-400 microscope.Surface morphology of the cross section of machined surfaces andcompositional variations using EDAX were observed on a FEI Quanta-200 SEM.

3. Results and discussion

The results of the experiments including the values of responsevariables (MRR and TWR) are presented in Table 2.

3.1. Analysis of material removal rate

The MRR values (in Table 2) show that the highest MRR(0.811 mm3/s) was achieved for 7th trial-replication (50 V, 18 A,22 ms, 0.25 MPa, 300 rpm and 4.5 mm) and the lowest MRR(0.303 mm3/s) was for 19th trial (80 V, 12 A, 22 ms, 0.25 MPa,200 rpm, 5 mm). In order to investigate the effect of machiningparameters on MRR, statistical analysis using analysis of variance(ANOVA) has been performed (see Table 4).

Table 4ANOVA for MRR (mm3/min).

ANOVA for material removal rate (MRR)

Parameter DOF Seq SS Adj SS Adj MS F P Statistical

significance

V 2 0.1173 0.0547 0.02738 29.09 0.00 OI 2 0.5145 0.4679 0.23395 248.5 0.00 OToff 2 0.0003 0.0001 0.00007 0.08 0.92 X

P 2 0.0054 0.0041 0.00205 1.64 0.13 X

N 2 0.0278 0.0299 0.01496 11.94 0.00 OCb 2 0.0046 0.0050 0.00250 2.66 0.09 X

VnI 4 0.0037 0.0025 0.00129 1.38 0.27 X

VnToff 4 0.0044 0.0044 0.00224 2.38 0.11 X

InToff 4 0.0056 0.0056 0.00280 2.98 0.07 X

Error 14 0.0188 0.0188 0.00094

Total 38 0.7028

V=Gap voltage; I=discharge current; Toff=pulse-off time; P=pressure; N=spindle

speed; Cb=clearance of shield at bottom.

MR

R (m

m3 /m

in)

0.4

0.5

0.6

0.7

Voltage(V)

Current(A)

Off-time(

35 50 65 80 12 15 18 22 33 44 55 66

Fig. 3. (a)–(f) Main effects plots of inpu

The ANOVA results show that the gap voltage (V), dischargecurrent (I) and rotational speed of the electrode (N) are thesignificant factors (at 95% confidence level) that influence theMRR (see Table 4). It is observed that none of the interactions aresignificant in influencing MRR. The trends of each factor in maineffects plots are determined using analysis of means (AOM) plotsin Fig. 3a–f.

3.1.1. Effect of discharge current on MRR

Statistical analysis using ANOVA (see Table 4) reveals thatdischarge current (I) is the most significant parameter due to thehighest F value. The main effects plot (see Fig. 3b) indicates thatMRR increases linearly with current (I) at all levels. With a variationin current from 12 to15 A, and a further increase up to 18 A, a linearincrease in average MRR has been observed. From ANOVA table forMRR, a very higher F value (248.5) indicates that discharge current I

is more significant than gap voltage V, which is contrary to the usualfindings. It is found that due to an increase in current, there is anincrease in MRR at all levels of other parameters, which is attributedto an increase in pulse energy. From all previous investigations inEDM as well as dry EDM, it has been known that material removalincreases with an increase in spark energy, a function directlyproportional to discharge current (I) [23], due to greater transfer ofthermal energy to the machining zone.

3.1.2. Effect of gap voltage on MRR

Statistical analysis using ANOVA (see Table 4) shows that thegap voltage (V) is also a significant parameter at 95% confidencelevel. As can be observed from the main effects plot (see Fig. 3a),an increase in voltage appears to cause a decrease in MRR. Anincrease in gap voltage from 50 to 65 V causes a decrease inaverage MRR by 1.69%. As the voltage changes from 65 to 80 V,further reduction in MRR by 18.26% has been observed. Generally,it is anticipated that an increase in gap voltage increasesdischarge energy given by

E¼1

2CV2 ð3Þ

where C is the capacitance of the medium and V is the voltageacross the gap. However, the relation does not appear to holdcompletely for dry EDM though the voltage has a significantinfluence (from statistical results).

In conventional EDM, a reduction in MRR with an increase in gapvoltage (V) as high as 43%, was evident [19]. This is due to anincrease in electric field, which helps discharge to occur even at highgap widths and insufficient cooling of the work due to localizedconcentration of discharge [20]. A similar trend should be prevalentsuch that MRR also decreases beyond a certain voltage in dry EDM. Ahigh voltage (above 80 V) increases gap of sparking, which reducesvelocity of gas at the work surfaces consequently affecting flushing

Pressure(MPa)

Speed(rpm)

Clearance(mm)

0.15 0.2 0.25 100 200 300 4 4.5 5

t parameters associated with MRR.

ARTICLE IN PRESS


and promoting arcing [15]. However, in this investigation, though asimilar phenomena has been observed, even after a gap voltage of65 V, increased MRR values were obtained under better flushingconditions, viz. increase in oxygen pressure and rotary speed, even athigher gap voltages.

3.1.3. Effect of spindle speed on MRR

The spindle speed (N) is also statistically significant (at 95%confidence level) for MRR (see Table 4). It appears from the maineffects plot depicted in Fig. 3e that there is a slight overallincrease in MRR when the spindle speed (N) increases. As therotation speed changes from 100 to 200 rpm, an increase in MRRby 2.46% is observed. Similar observations were made by otherresearchers [21,22]. They have shown that MRR increases withspeed because of centrifugal force and whirl which help flushdebris out, and consequently improves MRR. Similarly, in dryEDM, a rotation of tool reduces short circuit [1].

In EDM process, under normal circumstances, spindle speed(N) is the third significant parameter after gap voltage (V) andcurrent (I). When the effect of tool rotation on MRR in dry EDM iscompared with that of the EDM, it is observed that spindle speedhas a lower F value (11.94), compared to gap voltage and current.However, spindle speed appears to be more influential in thepresent investigation, with the same P value as that of mentionedtwo parameters. This could be due to the fact that a densermedium may not allow effective debris expulsion. However, in

Table 5ANOVA for TWR (mm3/min).

ANOVA for tool wear rate (TWR)

Parameter DOF Seq SS Adj SS Adj MS F P Statistical

significance

V 2 0.00005 0.00004 0.000023 0.36 0.704 X

I 2 0.00009 0.00008 0.000040 0.62 0.548 X

Toff 2 0.00013 0.00010 0.000053 0.82 0.454 X

P 2 0.00037 0.00041 0.000205 3.15 0.065 X

N 2 0.00034 0.00030 0.000153 2.35 0.121 X

Cb 2 0.00031 0.00031 0.000156 2.41 0.116 X

VnI 4 0.00017 0.00009 0.000046 0.71 0.502 X

VnToff 4 0.00033 0.00016 0.000081 1.25 0.308 X

InToff 4 0.00028 0.00016 0.000083 1.29 0.298 X

Error 14 0.00130 0.00130 0.000065

Total 38 0.00341

V=Gap voltage; I=Discharge current; Toff=Pulse-off time; P=Pressure; N=Spindle

speed; Cb=Clearance of shield at bottom.

TW

R (m

m3 /m

in) 1

0-3

-8

-6

-4

-2

0

Voltage(V)

Current(A)

Off time(µs)

35 50 65 80 12 15 18 22 33 44 55 66

Fig. 4. (a)–(f) Main effects plots of inpu

conventional EDM, under similar machining conditions, theincrease in MRR with an increase in speed of rotation (by70 rpm) is relatively higher (up to 6%) than in dry EDM [21],due to better heat transfer from electrode surface and lesserreattachment of debris.

3.1.4. Other factor effects

It is clear that oxygen pressure (P), pulse ‘off-time’ (Toff) andshielding clearance (Cb) are not significant parameters (seeANOVA results in Table 4). The past investigations related to dryEDM show that most of the researchers have chosen gas pressureas an important parameter, but its statistical significance in theprocess was not evident. The main effects plot of the threevariables (in Fig. 3c, d, f) also showed similar variation.

3.2. Analysis of tool wear

TWR values were estimated for the original and replicatedtrials as given in Table 5. The absolute values of tool wear showthat there are two main phenomena governing tool wearcharacteristics: erosion of electrode material and deposition ofmaterial on the electrode. The results of ANOVA and AOM on thedata are shown in Table 5 and Fig. 4a–f, respectively.

In spite that none of the parameters appear to be significant at95% confidence level, some of them appear to be significant at about90% confidence level: P—oxygen pressure, followed by Cb—radialclearance and N—spindle speed (P at 94.5%, Cb at 88.5%, N at 88%). Ascan be seen in main effects plots (Fig. 4a–f), central level of therelatively significant parameters (P, N and Cb) gives minimum toolwear (i.e. maximum material deposition on the tool).

Considering the effect of oxygen pressure, the highest materialdeposition is observed at the central level (0.2 MPa), and it isfound to be reduced further at the highest level (0.25 MPa). It isunderstood that in dry EDM, debris deposition is very high at lowinput pressures of oxygen due to inadequate cooling of electrodesurface [3,15]. Therefore, an increase in deposition of material atthe surfaces of inner walls of tool was observed, due to anincrease in oxygen pressure from 0.15 to 0.20 MPa. It appears thatpressure close to 0.20 MPa is sufficient to cool the electrodesurface, but is not sufficient to blow away the debris particles.

Furthermore, an increase in radial clearance of shield is foundto increase the deposition effect on the tool. At larger clearances,cooling of electrode will be more effective, whereas for smallclearances, it may not be so. Therefore, the deposition could beless at smaller clearances.

-8

-5

-2

1

Pressure(MPa)

-6

-4

-2

0

2

Speed(rpm)

Clearance (mm)

0.1 0.15 0.2 0.25 0 100 200 300 4 4.5 5

t parameters associated with TWR.

ARTICLE IN PRESS


It is known that in conventional EDM, with an increase inspindle speed, formation of a deposited layer is observed aroundthe electrode periphery [21,24]. A further increase in speed leadsto an increase in heat transfer rate from electrode surface andsubsequently causes separation of a deposited layer. This effectappears to be similar to that in present investigation of dryEDM. It may be noted that other parameters especially, theenergy related parameters, viz. voltage (V), current (I) and pulseoff time (Toff), are not significant in influencing the tool wear ofthe process. In conclusion, the parametric conditions thatminimize and maximize the response variables are summarizedin Table 6.

3.3. Analysis of oversize

Analysis of oversize shows that the holes machined using dryEDM are of U-shape specifically after 50% of length. Oversize in

-42

-40

-38

-36

-34

Ove

rsiz

e-en

try

(%)

-42

-39

-36

-33

-30

-42

-40

-38

-36

-34

Bold parameter labels indicates stati

V I Toff P

V I Toff P N

V I Toff P N Cb

Entry

50%

90%

Holesurface

Voltagge(V)

Current(A)

Off-time(µs)

1135 12 2250 65 80 15 18 33 44 59

Fig. 5. (a) Statistical significance of parameters influencing oversize at locations (ent

oversize at entry.

Table 6Parametric conditions corresponding to different phenomena related to tool wear.

Governing phenomena Tool wear

Condition Maximum Minimum

Trial no. 1 (replicated) 13 (replicated) 18 (replicated)

Parametric conditions

V (V) 50 65 65

I (A) 12 15 18

Toff (ms) 22 22 67

P (MPa) 0.15 0.25 0.25

N (rpm) 100 100 100rpm

Cb (mm) 4 4 4.5

the hole dimension in comparison with the original electrodedimension has been estimated as the outside diameter error (%). Itwas done at three locations, viz., entry, 50% depth and 90% depthof each hole machined. Fig. 5a shows the factors that arestatistically significant in influencing the hole oversize atvarious depths. The analysis is based on the results of ANOVA,which is not presented here. The main effect plots indicatingtrends in factor effects at the entry are displayed (see Fig. 5b–g).Similar plots are available for other depths too, as shown inFig. 6a–f (at 50% depth) and in Fig. 7a–f (at 90% depth).

The ANOVA results in Fig. 5a show that at entry of the hole,current (I) and clearance of the shield at bottom (Cb) are thesignificant parameters (at 95% confidence level) that govern theoversize. However, as can be observed in main effects plots (seeFig. 5b–g), the trend in variation of OD error (%) due to these twoparameters, appears to be opposite to each other. For example, alinear increase in I causes a corresponding increase in oversize,whereas a linear increase in clearance of the shield decreases the

stical significance at 95% confidence level

N Cb V*I V*Toff

Cb V*I V*Toff I*Toff

I*Toff

V*I V*Toff I*Toff

Pressure(Mpa)

Speed(rpm)

Clearance(mm)

0.15 100 45 66 0.2 0.25 200 300 4.5 5

ry, 50% and 90%). (b)–(g) Main effects plots of input parameters associated with

Debris deposition on electrode Zero tool wear/deposition

Maximum Minimum Zero variation

14 7 (replicated) 8 6 11

65 50 50 50 65

15 18 18 15 12

33 22 33 67 33

0.15 0.25 0.15 0.15 0.25

200 300 100 100 100

4.5 4.5 5 4.5 5

ARTICLE IN PRESS

-26

-24

-22

-20

-18

Ove

rsiz

e-50

% d

epth

(%)

Voltage(V)

Current(A)

Off-time(

Pressure(MPa)

Speed(rpm)

Clearance(mm)

35 12 22 0.15 100 450 65 80 15 18 33 44 55 66 0.2 0.25 200 300 4.5 5

Fig. 6. (a)–(f) Main effects plots of input parameters associated with oversize at 50% depth.

Ove

rsiz

e-90

% d

epth

(%)

-6.5

-6

-5.5

-5

-4.5

-4

Voltage(V)

Current(A)

Off time(µs)

Pressure(Mpa)

Speed(rpm)

Clearance(mm)

6535 50 80 12 15 18 22 33 44 55 66 0.15 0.2 0.25 100 200 300 4 4.5 5

Fig. 7. (a)–(f) Main effects plots of input parameters associated with oversize at 90% depth.


oversize. In conventional EDM, there is a small increase inoversize (4%), with an increase in current, even at low values(below 12 A) [25]. A similar pattern is observed from the maineffects plots (see Fig. 5b–g). It has also been observed that neitherof the other chosen factors (V, Toff, P, N) nor the correspondinginteractions (V� I, V� Toff, I� Toff) statistically influence theoversize.

At the centre of the hole, i.e. at 50% depth, it is evident thatcurrent (I), oxygen pressure (P) and spindle speed (N) aresignificant at 95% confidence level (Fig. 5a). It is observed frommain effects plots (Fig. 6a–f) that an increase in current causes alinear increase in OD error (%). However, an increase in pressure(P) results in a decrease in OD error (%). The spindle speedcorresponding to the intermediate level (200 rpm) yields max-imum OD error (%). At 50% depth of hole, neither of otherparameters (V, Toff and Cb) nor the chosen interactions are found tobe statistically significant.

At 90% depth of hole, spindle speed (N) is the only statisticallysignificant parameter (Fig. 5a). Further, the main effect plots(Fig.7a–f) show that the oversize appears to be maximum at themiddle level of speed (N). However, at the other two levels ofspindle speed (N), OD error (%) is found to be minimum. Insummary, the extreme values of oversize at entry, 50% depth and90% depth are presented in Table 7.

3.4. EDAX analysis of dry ED machined surfaces of hole

Energy dispersive X-ray (EDAX) method of analysis has beenused to identify the elemental composition in the machinedsurfaces generated after dry EDM. EDAX analysis of the samplescorresponding to the parametric conditions that give the best andthe worst MRR and TWR has been presented in Table 8. An area atthe middle of the cross section of specimen was selected (seeFig. 8a, b), and average percentage composition of elements atvarious points covering the region has been estimated.

EDAX analysis showed that the prominent elements detectedin the samples other than iron (Fe) were nickel (Ni), followed byoxygen (O) and chromium (Cr). The results also indicated thatoxygen (1.08–11.36%), copper (0.85–4.7%), and aluminium(0.227–0.395%) were the additional elements observed. A typicalcentral region on a dry EDMed surface corresponding to trial #19(80 V, 12 A, 22 ms, 0.25 MPa, 200 rpm and 5 mm) has been chosenfor EDAX analysis, and the corresponding EDS pattern indicatingintensities of various elements is shown (see Fig. 8b). A study ofliterature shows that during a rotary EDM process, migration ofmaterial from tool to work piece occurs, which increases with anincrease in spindle speed [26]. EDAX analysis indicated that somequantity of copper (tool material) and aluminium (shieldmaterial) occurred on the work surface. In addition, an increase

ARTICLE IN PRESS

Table 7Extreme values of oversize at various locations of hole and their parametric conditions.

Location Entry 50% depth 90% depth

Condition Maximum Minimum Maximum Minimum Maximum Minimum

OD error (%) �45.89% �29.29% �28.42% �14.66% �8.88% 0.31%

Trial no. 16th (replicated) 1st 16th 13th 13th 25th

Parametric conditions

V (V) 65 50 65 65 65 80

I (A) 18 12 18 15 15 18

Toff (ms) 22 22 22 22 22 22

P (MPa) 0.15 0.15 0.15 0.25 0.25 0.20

N (rpm) 200 100 200 100 100 100

Cb (mm) 5 4 5 4 4 4

Table 8Extreme parametric conditions for MRR and TWR and corresponding EDAX results.

Element Original composition,

wt%

MRR TWR

Highest (Trial no.7

(replicated))

Lowest (Trial no. 19) Highest (Trial no.1

(replicated))

Deposition

Average composition,

wt%


wt%


wt%

Highest (Trial no.14),

wt%

Lowest (Trial no.8),

wt%

Iron (Fe) 71.17% 69.83 59.89 68.76 69.14 60.61

Nickel (Ni) 8.0 22.36 18.14 18.26 25.63 32.097

Chromium

(Cr)

18.0 3.822 8.447 3.027 1.79 2.1

Oxygen (O) – 2.127 11.36 2.985 1.087 2.88

Manganese

(Mn)

2.0 0.247 0.647 0.845 0.082 0.175

Aluminium

(Al)

– 0.395 0.232 0.387 0.395 0.227

Silicon (Si) 0.75 0.142 0.43 0.217 0.185 0.185

Copper (Cu) – 0.937 0.85 4.707 1.22 3.097

Carbon (C) 0.08 0.125 – 0.745 0.467 0.547

5000

4000

3000

2000

1000

0

Full scale counts: 2234

O

CrMnNiCu

Fe

AlSi

keV

Cr MnCr

Fe

Fe

20 counts

Ni NiCu

Cursor: 10.192 keV

0 2 4 6 8 10

Cu

Fig. 8. (a), (b) EDAX analysis of dry EDMed surfaces: (a) typical central location for analysis: experiment #19 (80 V, 12 A, 22 ms, 0.25 MPa, 200 rpm, 5 mm) and (b) EDS

spectrum (at 1000� magnification).


in carbon content on machined surface was found, which issimilar to conventional EDM [29] and micro-EDM [18] processes.

3.5. Machined surface topography analysis

The dry EDM machined samples that correspond to theseverest and the mildest conditions of MRR, TWR and oversizewere investigated using scanning electron microscopy (SEM)at various regions of the surface and at various magnifications(80� , 300� , 600� and 1200� ). It is envisaged that aninteraction between spark generated and work surface causes a

change in the surface. A close examination of the images showsthat in most of the cases, micro-cracks have been developed. Ananalysis of morphology of surfaces has been presented in Table 9.

There have been a few attempts to characterize morphologicalfeatures in an EDM process in the past [27–30]. In general, in anEDM process, crack formation is expected, where its densityincreases with an increase in discharge energy and carbon content[27]. Furthermore, in an EDM process, orientation of the crackscan be horizontal or vertical. Phenomena such as overlapping ofcraters, formation of features such as global appendages [28],spherical particles due to solidification of expelled molten metal,and pock-marks formed due to entrapped gases [30] are also

ARTICLE IN PRESS

Table 9Morphological features of dry EDMed surfaces, corresponding to extreme conditions.

Condition/Trial no SEM micrograph Comments on features

Maximum MRR (Trial no.7 (replicated))

Few micro-cracks and large blow-holes found.

Black patches indicating some carbon deposition. Large blow-holes observed.

Minimum MRR (Trial no.19)

Very few micro-cracks Solidified features (globular solidification)

Maximum TWR (Trial no.1 (replicated))

Larger and deeper micro-cracks Blow holes and porous structures

observed in some region

Zero TWR (Trial no.6)

Dimples, river lines and few micro-cracks observed

Few blowholes and folding indicating localized melting and solidification

Maximum Debris deposition (Trial no.14)

Relatively smooth surface Only very few continuous micro-

cracks

Minimum Debris deposition (Trial no.8)

Few dimples observed Few micro-cracks due to void coalescence

Projections over work surface due to tool material deposition

Maximum oversize (Trial no.16 (replicated))

Discontinuous micro-cracks Very few continuous micro-cracks

Minimum oversize (Trial no.25)

River lines in few region Continuous micro-cracks


found. Similar morphological features were observed in thisstudy. However, the orientation of the cracks was highly random,and only few deep, continuous cracks were seen. Low discharge

current and high pulse duration were identified as reasons forcrack formation [31], which are also evident in the case of dryEDM (see Table 9).

ARTICLE IN PRESS

Table 10Optimum parametric conditions for MRR, TWR and oversize in dry EDM process.

Responses Voltage (V) Current (I) Pulse-off-time (Toff) P (pressure) N (speed) Clearance at bottom (Cb) Values Trial no.

Maximum depth 50 18 22 0.25 300 4.5 2.2 mm 7Maximum MRR 50 18 22 0.25 300 4.5 0.811 mm3/min 7(R)Zero TWR 50 15 67 0.15 100 4.5 0 6

65 12 33 0.25 100 5 11Minimum OD error

Entry 50 12 22 0.15 100 4 �29.29% 150% 65 12 22 0.25 100 4 �14.66% 1390% 80 18 22 0.20 100 4 0.31% 25


3.6. Optimum parametric conditions

This experimental work has enabled to obtain a number offactors significantly affecting dry EDM process. The optimumparametric conditions to maximize MRR, and minimize TWR andhole oversize are presented in Table 10.

Fig. 9. Scheme of a single discharge EDM method.

Crater radius, Rc

Conical shaped crater (approximated)

Depth ofCrater, z

ANODE (+)

Anode melt cavity

3.7. Single-spark analysis of dry EDM

Single-spark dry EDM experiments were conducted at theparametric conditions similar to that in Ref. [32] knowing thatthey are important to understand the material removal mechan-ism [32]. Fig. 9 illustrates schematic of the set-up.

The experiments were conducted on a S50CNC EDM machineusing a single-pulse generator. The workpiece material was aSS304 block of dimensions 27 mm�14 mm�10 mm. Copperrods of Ø5 mm with pointed tip were used as electrodes. Oxygengas is supplied at a low pressure of 0.05 MPa. A minimum gapdistance is ensured so as to apply only a single-spark and generatea single crater on the work surface. The programming for thesingle-spark operation was done in MDI mode. Six single-sparkdry EDM experiments were performed at various I, Ton, Toff

combinations as depicted in Fig. 12c.Voltage has been keptconstant as 25 V so that results are comparable with that ofPatel et al. [32].

Fig. 10. Simplified shape of the anode erosion cavity for MRR and crater radius

evaluation.
3.7.1. Crater radius estimation and volume estimation using
theoretical relations

The values of crater depths obtained after simulation (zsim)corresponding to each parametric condition for a single-sparkliquid EDM process were used in this work [32]. Approximatingthe shape of the anode crater as conical (Fig. 10), the volumeremoval rate (Vc) and the radius of the crater formed in single-discharge (Rc) have been evaluated for dry as well as liquiddielectric EDM. These are compared with the correspondingexperimental values obtained in this work. Considering thegeometry of a crater (in Fig. 10), volume of the crater (in mm3)is given by

VC ¼1

3pR2

c z ð4Þ

Knowing the depth of erosion from the simulation of a singledischarge in liquid dielectric EDM [32], the volume of crater canbe calculated theoretically and experimentally in liquid dielectricEDM as

VCðtheoretical�liquidÞ ¼1

3pRcðthÞ

2zsim ð5Þ

VCðexp�liquidÞ ¼1

3pRcðexpÞ

2zsim ð6Þ

where Rc(exp) is the experimental value of crater radius and Rc(th) isthe theoretical value of crater radius. zsim is obtained from theprocess simulation.

Similarly, knowing the pulse on-time (Ton), the number ofsparks per minute and the volume of a crater formed in liquiddielectric EDM can also be calculated in two ways:

No: of sparks in 1min¼N¼60� 106

ðTonþToff Þsparks=min ð7Þ

Experimental volume of the crater in liquid EDM per spark¼VCðexp-liquidÞ

Nmm3

ð8Þ

Theoretical volume of the crater in liquid EDM per spark¼VCðtheo-liquidÞ

Nmm3

ð9Þ

A similar method was followed for finding erosion rate in dryEDM. Depth measurement and diameter measurement were doneat 20 locations and their average values have been taken. Afterconducting depth and diameter measurements on the experi-mental cavities shown in Fig. 11a,b and approximating it asconical, the experimental volume of the crater per spark in dry

ARTICLE IN PRESS

Fig. 11. (a)–(b) Craters generated in single spark dry EDM, parametric conditions: (a) 5th (I=25 A, Ton=100 ms and Toff=4.2 ms), (b) 6th (I=36 A, Ton=180 ms and Toff=4.2 ms).

Expt no. 1 I (A) 5

Ton (µs)Toff (µs)

0

5

10

15

20

25

30

35

40

Cra

ter

radi

us (µ

m)

Discharge current (A)

Rc-liquid EDM-experimentalRc-liquid EDM-simulationRc-dry EDM

-5

0

5

10

15

20

25

30

MR

R (m

m3 /s

park

) X 1

0-6

Discharge current (A)

MRR-liquid EDM-experimental

MRR-liquidEDM-simulation

MRR-dry EDM

0 5 10 15 20 25 30 35 40 0 10 20 30 40

2 3 4 5 610 13 20 25 36

18 32 42 56 100 1802.4 2.4 3.2 3.2 4.2 4.2

Fig. 12. (a)–(c) A comparison of (a) crater radius and (b) erosion rates for single spark liquid EDM and single spark dry EDM. (c) parametric conditions.


EDM is obtained as

VCðdry-expÞ ¼1

3pRc dry-exp

2zdry-exp mm3=spark ð10Þ

3.7.2. Radius of single-spark dry EDM

A comparison between crater radii obtained after a single-sparkoperation in a liquid EDM and a dry EDM is presented in Fig. 12a. Itis observed that for the experiments conducted at relatively lowinput energies (experiments from 1 to 4), crater radius for a single-spark in dry EDM is higher than that of in the liquid EDM. However,for the liquid EDM, a higher radius of crater is found at the higherdischarge energies (see experiments 5 and 6).

3.7.3. MRR for single-spark dry and liquid EDM

A comparison between the erosion rates for a single-spark indry and liquid dielectric EDM is shown in Fig. 12b. It is observedthat at lower discharge energies (in experiments 1, 2 and 3), theerosion rate in a single-spark in dry EDM is slightly higher thanthat of in a single-spark liquid dielectric EDM. Further, at the highdischarge energies, it is observed that the liquid dielectric EDM

yields a very high MRR than the dry EDM (in experiments 4, 5 and6). It appears that a greater material melting and solidificationoccurred at the machined surface in experiment 6 (see Fig. 12b)than that in experiment 5 (see Fig. 12a).

4. Conclusions

In this work, experimental evaluation of dry electricaldischarge drilling was carried out. It has been shown that dryEDM can be performed by providing an enclosure (shielding) tothe sparking region. This study has revealed various character-istics of dry EDM process by measurement of oversize, examina-tion of machined surface morphology and EDAX analysis, inaddition to MRR and TWR study. The following remarks have beendetermined:

�
It was evident that gap voltage, discharge current andelectrode rotational speed significantly (with 95% confidence)influence MRR. However, none of the two-factor interactionswere statistically significant.

ARTICLE IN PRESS


�
TWR was found close to zero in most of the experiments. Apredominant deposition of melted and eroded work material onthe electrode surface instead of tool wear was evident. None ofthe input parameters and their interactions were statisticallysignificant (at 95% confidence level) in influencing TWR. � Dimensional measurement of diameter on holes was performed
at five different locations: entry, 10% depth of hole, 50% depth ofhole, 90% depth of hole and at exit. The maximum oversize value(�45.89%) was observed at the entry that corresponds to the16th trial (replication), and the minimum value (0.31%) at 90%depth of the hole corresponding to the 25th trial.
� EDAX analysis of the central region of machined cross-section
surfaces showed that there was a migration of tool (copper)and shield (aluminium) material as well as enrichment ofcarbon composition at work surface.
� Observation of morphology of the dry EDMed surfaces
revealed micro-crack formation due to thermal stresses,deposition of spherical particles and marks due to entrappedgases on the machined surface. Though, most of the crackswere discontinuous, a few shallow and continuous cracks werealso evident. In addition, dimples and river lines were observedin most of the cases.
� Optimum processing conditions for maximizing the MRR and
depth were achieved; these corresponds to the trial #7 (50 V,18 A, 22 ms, 0.25 MPa, 300 rpm and 4.5 mm). In experiments #6and #11, zero TWR was observed. Similarly, correspondingexperimental conditions for minimum oversize were in experi-ment #1 at entry, #13 at 50% depth and #25 at 90% depth.
� Single-spark analysis of dry EDM showed that at low input
energies, there is an increase of crater radius (Rc) as well asMRR in dry EDM as compared to the liquid dielectric EDM.However, a larger crater radius (Rc) and MRR were observed athigher discharge energies.

Acknowledgement

The authors wish to acknowledge financial support for thiswork from Department of Science and Technology, Government ofIndia. The authors also wish to acknowledge the technical supportfrom Electronica Machine Tools Limited, Pune (India).

References

[1] M. Kunieda, M. Yoshida, N. Taniguchi, Electrical discharge machining in gas,CIRP Annals—Manufacturing Technology 46 (1) (1997) 143–146.

[2] N.M. Abbas, D.G. Solomon, M.F. Bahari, A review on current research trends inelectrical discharge machining (EDM), International Journal of Machine Toolsand Manufacture 47 (7–8) (2007) 1214–1228.

[3] J. Tao, A.J. Shih, J. Ni, Experimental study of the dry and near dry electricaldischarge milling processes, Journal of Manufacturing Science and Engineer-ing 130 (1) (2008) 1–8.

[4] T. Wang, M. Kunieda, Dry WEDM for finish cut, Key Engineering Materials259–260 (1) (2003) 562–566.

[5] C.C. Kao, J. Tao, A.J. Shih, Near dry electrical discharge machining,International Journal of Machine Tools and Manufacture 47 (15) (2007)2273–2281.

[6] J. Tao, A.J. Shih, J. Ni, Near-dry EDM milling of mirror-like surface finish,International Journal of Electrical Machining 13 (1) (2008) 29–33.

[7] M. Kunieda, Y. Miyoshi, T. Takaya, N. Nakajima, Y. ZhanBo, M. Yoshida, Highspeed 3D milling by dry EDM, CIRP Annals—Manufacturing Technology 52(1) (2003) 147–150.

[8] M. Kunieda, B. Lauwers, K.P. Rajurkar, B.M. Schumacher, Advancing EDMthrough fundamental insight into the process, CIRP Annals—ManufacturingTechnology 54 (2) (2005) 64–87.

[9] Z. Yu, T. Jun, M. Kunieda, Dry electrical discharge machining of cementedcarbide, Journal of Materials Processing Technology 149 (1–3) (2004)353–357.

[10] Q.H. Zhang, J.H. Zhang, J.X. Deng, Y. Qin, Z.W. Niu, Ultrasonic vibrationelectrical discharge machining in gas, Journal of Materials ProcessingTechnology 129 (1–3) (2002) 135–138.

[11] Q.H. Zhang, J.H. Zhang, S.F. Ren, J.X. Deng, X. Ai, Study on technology ofultrasonic vibration aided electrical discharge machining in gas, Journal ofMaterials Processing Technology 149 (1–3) (2004) 640–644.

[12] Q.H. Zhang, R. Du, J.H. Zhang, Q.B. Zhang, An Investigation of ultrasonicassisted electrical discharge machining in gas, International Journal ofMachine Tools and Manufacture 46 (12–13) (2006) 1582–1588.

[13] M.G. Xu, J.H. Zhang, Y. Li, J.H. Zhang, S.F. Ren, Material removal mechanismsof Cemented carbides machined by ultrasonic Vibration assisted EDM in gasmedium, Journal of Materials Processing Technology 209 (4) (2008)1742–1746.

[14] M. Kunieda, T. Takaya, S. Nakano, Improvement of dry EDM characteristicsusing piezoelectric actuator, CIRP Annals—Manufacturing Technology 53 (1)(2004) 183–186.

[15] S.K. Saha, S.K. Choudhury, Experimental investigation and empirical model-ing of the dry electric discharge machining process, International Journal ofMachine Tools and Manufacture 49 (3–4) (2009) 297–308.

[16] M.S. Phadke, in: Quality Engineering using Robust Design, Prentice-HallPublication, New Jersey, 1989.

[17] M.S. Puranik, S.S. Joshi, Analysis of accuracy of high-aspect ratio holesgenerated using micro-EDM Drilling, Proceedings of the Institution ofMechanical Engineers, Part B, Journal of Engineering Manufacture 222 (11)(2008) 1453–1464.

[18] O. Thao, S.S. Joshi, Analysis of heat affected zone in the micro-electricdischarge machining, International Journal of Manufacturing Technology andManagement 13 (2–4) (2008) 201–213.

[19] S.H. Lee, X.P. Li, Study of the effect of machining parameters on the machiningcharacteristics in electrical discharge machining of tungsten carbide, Journalof Materials Processing Technology 115 (3) (2001) 344–358.

[20] J. Simao, H.G. Lee, D.K. Aspinwall, R.C. Dewes, E.M. Aspinwall, Workpiecesurface modification using electrical discharge machining, InternationalJournal of Machine Tools and Manufacture 43 (2) (2003) 121–128.

[21] B. Mohan, A. Rajadurai, K.G. Satyanarayana, Effect of SiC and rotation ofelectrode on electrical discharge machining of Al–SiC composite, Journal ofMaterials Processing Technology 124 (3) (2002) 297–304.

[22] J.S. Soni, G. Chakravarti, Machining characteristics of titanium with rotaryelectro-discharge machining, Wear 171 (1–2) (1994) 51–58.

[23] K.D. Chattopadhyay, S. Verma, P.S. Satsangi, P.C. Sharma, Development ofempirical model for different process parameters during rotary electricaldischarge machining of copper–steel (EN-8) system, Journal of MaterialsProcessing Technology 209 (3) (2009) 1454–1465.

[24] P. Koshy, V.K. Jain, G.K. Lal, Experimental investigations into electricaldischarge machining with a rotary disk electrode, Precision Engineering 15(1) (1993) 6–15.

[25] S. Singh, S. Maheshwari, P.C. Pandey, Some investigations into the electricaldischarge machining of hardened tool steel using different electrodematerials, Journal of Materials Processing Technology 149 (1–3) (2004)272–277.

[26] J.S. Soni, G. Chakraverti, Experimental investigation on migration of materialduring EDM of die steel (T215Cr12), Journal of Materials ProcessingTechnology 56 (1–4) (1996) 439–451.

[27] L.C. Lee, L.C. Lim, Y.S. Wong, Towards crack minimization of EDMed surfaces,Journal of Materials Processing Technology 32 (1–2) (1992) 45–54.

[28] L.C. Lee, L.C. Lim, V. Narayanan, V.C. Venkatesh, Quantification of surfacedamage of tool steels after EDM, International Journal of Machine Tools andManufacture 28 (4) (1988) 359–372.

[29] F. Ghanem, C. Braham, H. Sidhom, Influence of steel type on electricaldischarge machined surface integrity, Journal of Materials ProcessingTechnology 142 (1) (2003) 163–173.

[30] A.G. Mamalis, G.C. Vosniakos, N.M. Vaxevanidis, Macroscopic and micro-scopic phenomena of electro-discharge machined steel surfaces: an experi-mental investigation, Journal of Mechanical working Technology 15 (1)(1987) 335–356.

[31] H. Lee, F. Hsu, T. Tai, Study of surface integrity using the small area EDMprocess with a copper–tungsten electrode, Material Science and EngineeringA364 (1) (2004) 346–356.

[32] M.R. Patel, M.A. Barrufet, P.T. Eubank, Theoretical models of the electricaldischarge machining process. II. The anode erosion model, Journal of AppliedPhysics 66 (9) (1989) 4104–4111.

wDry

Documents