Journal of Mechanical Science and Technology 00 (2010) 0000~0000

www.springerlink.com/content/1738-494x

submitted manuscript under review

Experimental investigation and 3D finite element prediction of the white layer thickness, heat affected zone, and surface roughness in EDM process
Mohammadreza Shabgard1, Samad Nadimi Bavil Oliaei2, Mirsadegh Seyedzavvar1 and Ahmad Najadebrahimi1
2 1 Department of Mechanical Engineering, University of Tabriz, Tabriz, Iran Department of Mechanical Engineering, Middle East Technical University, Ankara, Turkey

(Manuscript Received 000 0, 2009; Revised 000 0, 2009; Accepted 000 0, 2009) -please leave blank ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

Abstract

In this work, an axisymmetric three-dimensional model for temperature distribution in the Electrical Discharge Machining process has been developed using the finite element method to estimate the surface integrity characteristics of AISI H13 tool steel as workpiece. White layer thickness, depth of heat affected zone, and arithmetical mean roughness consisted of the studied surface integrity features on which the effect of process parameters, including pulse on-time and pulse current were investigated. Additionally, the experiments carried out under the designed full factorial procedure to validate the numerical results. Both numerical and experimental results show that increasing the pulse on-time leads to a higher white layer thickness, depth of heat affected zone, and the surface roughness. On the other hand, an increase in the pulse current results in a slight decrease of the white layer thickness and depth of heat affected zone, but a coarser surface roughness. Generally, there is a good agreement between the experimental and the numerical results.
Keywords: Electrical discharge machining, White layer thickness, Depth of heat affected zone, Surface roughness ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

1. Introduction
Considering the challenges brought on by advanced technologies, the electrical discharge machining (EDM) process is one of the best alternatives for machining an ever increasing number of high-strength, non-corrosion, and wear resistant materials [1, 2]. AISI H13 tool steel is considered as a significant one of these materials that has a widespread application in mold industries [3]. EDM utilizes rapid, repetitive spark discharges from a pulsating direct-current power supply between the workpiece and the tool submerged into a dielectric liquid [4]. During each discharge, intense heat is generated, causing local melting or even evaporating of the work material. Depending on the plasma flushing efficiency (%PFE), the collapse of plasma channel causes very violent suction and sever bulk boiling of some of the molten material and removal from the molten crater [5]. The metal remaining in the crater re-solidifies,
This paper was recommended for publication in revised form by Associate Editor 000 000-please leave blank. Corresponding author. Tel.: +90 531 322 7654, Fax.: +90 312 210 2536 E-mail address: e170538@metu.edu.tr. KSME & Springer 2010
*

which is called the white layer or recast layer, and develops a residual stress that often causes micro cracks. An annealed heat affected zone (HAZ) lay directly below the recast layer. The micro cracks created in the white layer could penetrate into the heat affected zone. Additionally, this layer is softer than the underlying base material. This annealed zone could weaken prematurely and cause the material to develop stress fractures that could lead to anything from a minor malfunction to a catastrophic failure. The quality of an ED machined surface is becoming more and more important to satisfy the increasing demands of sophisticated component performance, longevity and reliability. Optimum utilization of the EDM process requires the selection of an appropriate set of machining parameters that would result in the minimum thickness of the recast layer and the depth of heat affected zone [6]. Several studies have been carried out to determine appropriate ED machining parameter combinations from the aspect of surface integrity [7, 8]. However, these studies were based on the use of experimental approaches and statistical analyses. In few studies, pure theoretical approaches have been proposed to estimate the outputs of EDM process, using FE or analytical methods [9, 12]. For instance, Ben Salah et al. [9] presented a numerical model to study the temperature distribu-

0000

M. Shabgard et al. / Journal of Mechanical Science and Technology 00 (2010) 0000~0000

tion in EDM process. They used the thermal results to predict the material removal rate and the total surface roughness. They reported that taking into account the temperature dependence of the conductivity is of crucial importance to the accuracy of the numerical results and gives the better correlation with experimental observations. Marafona et al. [10] employed a FE model to estimate the surface roughness and the removed material from both anode and cathode. They reported that the anode material removal efficiency is smaller than that of the cathode because there is a high amount of energy going to the anode and also a fast cooling of this material. They explained that this phenomenon can be explained by the differences of thermal conductivity of the cathode and anode. Kansal et al. [11] developed a model to calculate the temperature distribution in the workpiece material by employing ANSYS software in the powder mixed electrical discharge machining. They utilized the results of finite element simulation to estimate the material removal rate. The usage of temperature dependent material properties was one of the major features of their model which led to a better accuracy in prediction of MRR. Joshi and Pande [12] introduced an intelligent process modeling and optimization of EDM process. In their model, they employed the FEM to estimate the output parameters of EDM process including MRR and %TWR. The dependency of material properties to the temperature and spark radius to the discharge duration have been emphasized in their research. Considering the existing tendency for improving the quality of EDMed product, it is essential to develop numerical models to estimate relationship between the predominant EDM machining parameters and the resulting machined surface integrity, i.e., surface roughness, white layer thickness and depth of HAZ. Although honored institutions and researchers all over the world have already initiated some researches in this area, a method for predicting the aforementioned parameters have never been presented. In the present study, the white layer thickness and depth of heat affected zone were predicted and a new approach for exploring the effect of %PFE (plasma flushing efficiency) has been introduced. Counting the number of normal pulse, through the implication of the utilized circuit and well programmed user subroutines provides useful tools for the study of EDM process. The developed axisymmetric three-dimensional finite element models are used to simulate the temperature distribution in the electrical discharge machining process. The results of the finite element simulation were employed to estimate the white layer thickness (WT), depth of heat affected zone (HD), and arithmetical mean roughness (Ra) of AISI H13 tool steel as workpiece. To verify the accuracy of the numerical results, EDM experiments have been carried out followed by metallurgical analyses. The influence of pulse on-time (Ti) and pulse current (I) were detailed based on numerical results and experimental observations.

2. Model details
2.1 Heat input In the literature, mainly two different heat input models are used, the point source model and the Gaussian heat input [9]. In the present work, the Gaussian heat distribution has been employed to approximate the heat from the plasma channel [13]. This model has two factors, the fraction of heat distributed to the workpiece (F) and the radius of the area heated by the plasma. Eq. (1) represents the heat flux qw(r) as a function of radius r [14]: qw(r)=4.5(FUbI/R2(t))exp{-4.5(r/R(t))2} (1)

where Ub is the electric potential, I the current density and R(t) the radius of plasma channel in micro-meter. The value of F is an empirically determined constant. Dibitoto et al. [5, 15] reported that the fraction of heat going to the cathode and anode were18.3% and 8%, respectively, with remainder lost to the dielectric. In other study, Xia et al. [16] reported a distribution of 34% and 48% discharge energy into the cathode and the anode, respectively. Modeling of heat distribution in anode and cathode in micro-EDM, Yeo et al. [17] proposed a fraction of 14% and 39% of discharge energy to the anode and the cathode, respectively. In the present study, the different proposed values for heat distribution into the workpiece (cathode) have been examined. Meanwhile, the best value for F which resulted in the better proximity of numerical results to the experimental observations was equal to 18.3%. Ikai and Hashiguchi [18] showed that the radius of plasma channel is related to the current intensity and pulse duration. In this research, it was decided to determine the radius of the plasma channel based on the work of Ikai and Hashiguchi. In particular, this radius is called the equivalent heat input radius, and it can be expressed as a function of the pulse current and the pulse on-time as shown in Eq. (2): R(t)=2.04I0.43Ti0.44 (2)

Fig. 1. Thermal model of the EDM process.

M. Shabgard et al. / Journal of Mechanical Science and Technology 23 (2009) 1261~1269

1263

2.4 Assumptions The following assumptions have been considered in the thermal modeling procedure: 1) The workpiece is considered as a semi-infinite body since the volume of removed material is much smaller than the volume of the workpiece. 2) The bilateral thermal effects of successive sparks are neglected. 3) The effects of sparking gap on discharge characteristics are supposed to be negligible. 4) The phase changes during the analysis are neglected. 5) The crater formed on the workpiece due to each discharge is assumed to have circular parabolic geometry. 6) The redeposit of recast layer in the crater after each spark is considered uniform. 7) The recast layer located outside the crater which is deposited on the workpiece after each spark is neglected. 2.5 Application of FEM in thermal modeling of EDM The implemented numerical analysis carried out in this study was based on the Gaussian distributed heat flux during each spark of the EDM process. The standard finite element software ABAQUS/CAE was employed for simulating the temperature distribution in the workpiece during the discharge process. One of the special features of this software is its ability in considering the temperature dependent material properties for workpiece. Consequently, the temperature dependency of workpiece material properties was taken into account in the simulation stage. This consideration increases the compatibility of predicted values with the experimental observation. The latent heat of melting, which was considered as one of the major features of workpiece and had a crucial effect on the result of simulation, was considered in the numerical analysis. Based on ABAQUS code, a transient thermal analysis was conducted to calculate the temperature field in the workpiece and to study the heat distribution with respect to different machining parameter settings. The size of domain for the thermal analysis was dependent on the input parameters of the EDM process, since the radius of discharge channel was determined based on the machining settings (Eq. (2)). The numerical analyses were performed using a 10-node quadratic heat transfer tetrahedron (DC3D10) mesh with the global size of 13m. Approximately 45000 elements have been generated on the instance.

Fig. 2. A view of three-dimensional temperature distribution obtained at the end of pulse on-time.

2.2 Heat transfer The evolution of the temperature within the workpiece during each discharge is governed by the differential equation for heat transfer without internal heat generation term [19]. It is assumed that the heat transferred to the workpiece as a result of a single spark is axisymmetric and thus, this equation in cylindrical coordinates takes the following form: (1/r)(/r)(r.(T/r))+2T/z2=(.C/k)(T/t) (3)

where r and z are the coordinate axes, T is the temperature, , k, and c are the density, thermal conductivity, and specific heat capacity of the workpiece material, respectively. 2.3 Boundary conditions The considered workpiece domain is shown in Fig. 1. In the domain, heat flux for a single spark is applied on the surface B1 up to sparking radius R using Gaussian distribution. On the remaining region on B1 surface, the convection heat transfer takes place due to the cooling effect of the dielectric fluid. As the boundaries B2 and B3 are very far from the position of spark collision, so no heat transfer conditions have been assumed for these boundaries. For B4, which is the axis of symmetry, the heat flux has been taken as zero as there is no net heat gain or loss from this region. The symmetric boundary conditions are given as follows: when t>0 BCS: k(T/z)= hc(T-T0) r>R qw(r) rR on B1 0 for off-time (4)

and (T/n)=0 on B2, B3, B4

(5)

The initial temperature, Ts can be taken as the temperature T0 of the dielectric fluid in which the workpiece is submerged. Thus, TS=T0 at t=0 (6)

To apply the heat flux varying with time on the work domain, a discontinuous method for discretization in time has been utilized by employing an arbitrary polynomial order. Let t0<t1<t2<..<tn=Ton be a sequence of times and a partition of time interval [0, Ton]. A dflux subroutine was used to calculate the heat flux exposed on the discharge location of instance.

0000

M. Shabgard et al. / Journal of Mechanical Science and Technology 00 (2010) 0000~0000

Radius of crater (m)

The theoretical crater volume defined by the parabolic geometry is described by the following equation: VC(FEM)= S rc2 (8)

Depth of crater (m)

where S and rc are the depth and radius of the crater, respectively. 2.5.2 Plasma flushing efficiency The 3D points comprised in VC(FEM) represent those that are over the liquid temperature of the workpiece material. The values of rC and S were obtained by computing the temperature distribution profiles along the radius and depth of the workpiece. The plasma flushing efficiency (%PFE), that was introduced in previous studies as the metal removal efficiency [10], is defined as the ratio of the actual volume of removed material per pulse versus the theoretical volume of melted material per pulse. Its value is described by Eq. (9). %PFE=100VC(EXP)/VC(FEM) (9)

Fig. 3. Iso-thermal counter at the melting temperature.

The temperature profiles obtained from the FE analysis were used to calculate the amount of material removed from the specimen. Fig. 2 shows a typical problem solved using the finite element simulation.
2.5 Determination of surface integrity characteristics 2.5.1 The morphology of cater cavity The temperature contour-plots are shown in Fig. 2 for a typical simulation. To calculate the cavity volume at the end of each discharge, the isothermal contour at the melting temperature was drawn through the (xi,zi) points obtained by the FEM. Fig. 3 represents this contour as well the parabolic crater cavity form suggested by Salonitis et al [20]. Using the (xi,zi) nodes resulted from the FEM and employing the equation (Eq. (7)) introduced by Joshi and Pande [12] the cavity volume is calculated. VC(FEM)= ((xi+xi+1)/2)2 (zi+1-zi) (7)

where VC(EXP) is the volume of removed material per discharge obtained through empirical observation. VC(EXP) and VC(FEM) are calculated according to the Eqs. (8) and (10), respectively. VC(EXP)=(M1-M2)/(Nnp) (10)

where M1 and M2 are the workpiece weight before and after machining (g), respectively, Nnp the number of normal pulses, and finally the density of the workpiece. 2.5.3 White layer thickness, depth of heat affected zone, and surface roughness Based on the numerical results, Eqs. (11) and (12) were used to calculate the white layer thickness and the depth of heat affected zone, respectively. WT=S-(S%PFE) HD=DRB-(S%PFE) (11) (12)

where VC(FEM) is the volume of the crater and xi and zi the coordinates of the node of the isothermal contour. It has been proved that the crater volume obtained by using Eq. (7) is equal to the volume by the assumption of a circular parabolic geometry for the crater [20] since the parabolic contour fits the isothermal contour passing through the (xi,zi) coordinates. This geometry is shown in Fig. 4.

where DRB is the depth of re-crystallization temperature border at the end of pulse on-time which is obtained from FE model (see Fig. 16). To calculate Ra, the equation proposed by Salonitis et al. [20] has been modified to take into account the %PFE:
Ra=%PFES((rc+rS) /rc)2 (13)

where rS is the radius of plasma channel at the end of pulse on-time.

Fig. 4. Crater geometry performed by each discharge.

Journal of Mechanical Science and Technology 00 (2010) 0000~0000

www.springerlink.com/content/1738-494x

submitted manuscript under review

Table 1. Mechanical and physical properties of AISI H13 [21]. Temperature C Density kg/dm3 Specific heat J/(kg .K) 460 550 590 1454 0C Electrical resistivity Ohm.mm2/m Modulus of elasticity N/mm2 Thermal conductivity (W/m.K) 24.30 27.70 27.50 1315 0C

20C 7.80 500C 7.64 600C 7.60 Liquidus temperature

0.52 215103 0.86 176103 0.96 165103 Solidus temperature

3. Experimental procedure
3.1 Experimental setup and model validation methodology The workpiece material used in this study was AISI H13 tool steel. The main mechanical and physical properties of such a workpiece material at different temperatures are given in Table (1). The tool material was forged commercial pure copper. The experiments were performed on a die sinking EDM machine (CHARMILLES ROBO-FORM200) which operates with an iso-pulse generator. Machining tests were carried out at five pulse current settings as well as four pulse on-time settings. As a result, 20 experiments could have been designed. Each machining test was performed for 15 minutes. Table (2) represents the experimental test conditions. For on-line registering the number of different kind of pulses (Normal, Arc, Open, and short-circuit pulses) during the EDM process, an oscilloscope (Hitachi VC-6524) of storage type and an electronic circuit were employed to capture the gap voltage and current variations versus time, which were then transferred and stored on a PC hard disk through a serial cable and port connection. Additionally, in order to count the number of each type of pulses, a program in FORTRAN language was written and linked to the pulse monitoring software (ITM). A digital balance (CP2245-Surtorius) with a resolution of 0.1mg was used for weighing the workpieces before and after the machining process.
Table 2. Experimental test conditions. Generator type Dielectric fluid Flushing type Supply voltage (V) Reference voltage (V) Pulse current (A) Polarity Pulse on-time (s) Pulse interval (s) Tool material Tool shape Iso-pulse (ROBOFORM 200) Oil Flux ELF2 Normal submerged 200 70 8,12,16,20, and 24 Positive 12.8, 25, 50, and 100 6.4 Commercial pure copper Cylindrical 18.3mm & L=20mm Table 3. The average depth of heat affected zone (HD) at some machining settings. Settings Average HD(m) I Ti 12.8 12.0 25 15.7 8 50 24 100 34.4 12.8 12.5 12 25 50 100 12.8 25 16.5 23 34.8 13 17.8 Settings I 16 Ti 50 100 12.8 25 50 100 12.8 25 50 100 Average HD(m) 23.5 32.7 12 16.2 21.5 30.2 11 15 21 29.6

20

24

16

The surface roughness parameter (Ra) was measured using a surface roughness measuring instrument (MahrPerthomether M2) with a cut-off length of 0.85 mm. The white layer is so infiltrated with carbon that has a separate, distinct structure, totally distinguishable from the parent material in the Scanning Electron Microscopy (SEM) images [22]. Consequently, the amount of white layer thickness has been measured by measuring this layers thickness at 30 different points by SEM and accounting for their average (Figs. 5-7). So the machined specimens were sectioned transversely

WT

Fig. 5. SEM micrograph showing cross-section of EDMed piece (I=8A & Ti=25s).

0000

M. Shabgard et al. / Journal of Mechanical Science and Technology 00 (2010) 0000~0000

WT

Fig. 6. SEM micrograph showing cross-section of EDMed piece (I=24A & Ti=50s).

Fig. 8. Optical micrograph representing the effect of penetrating point of micro-hardness tester (I=24A & Ti=100s).

by a wire electrical discharge machine and prepared under a standard procedure for metallographic observation. Etching was performed by immersing the specimens in 5% Nital reagent. VEGA_TESCAN scanning electron microscopy was employed at this stage. On the other hand, since there are not much significant differences between HAZ and the parent material in the microscopic images that could be identified by, measuring of microhardness is a reasonable way to obtain the depth of heat affected zone [23]. With this in mind, micro-hardness from cross-section of machined specimens was measured to determine the depth of heat affected zone (Table 3). The microhardness of specimens was measured by the OLUMPUS LM700 micro-hardness tester. 3.2Micro-hardness and micro-analyze results Figure 8 represents a typical effect of the penetrating point (indenter) of the employed OLUMPUS LM700 microhardness tester. The profiles of micro-hardness versus depth

from top surface of machined workpiece at different pulse settings are depicted in Figs. 9 and 10. It was found that the micro-hardness of machined specimens increases gradually form top surface to the point that is deemed to be the margin of heat affected zone. This phenomenon could be explained as follow: Figures 11 and 12 show the results of micro analysis carried out by the SEM analysis probe. According to these results, an increase of carbon content in the surface layers after machining can be observed. Carbon, resulting from pyrolysis of the hydrocarbon dielectric, is transferred towards the molten workpiece material as well as penetrates into the heat affected zone that its temperature is higher than the re-crystallization temperature. By increasing the carbon amount in the surface layer, regarding to the occurrence of super quenching, after collapsing the plasma channel and during the cooling stage, the amount of retained austenite increases [6]. This increase in the amount of retained austenite decreases the hardness significantly [24].

WT

Micro-hardness (HRC)

Depth from top surface (m) Fig. 7. SEM micrograph showing cross-section of EDMed piece (I=24A & Ti=100s). surface (Ti=100s). Fig. 9. Micro-hardness distribution below the top machined

M. Shabgard et al. / Journal of Mechanical Science and Technology 23 (2009) 1261~1269

1263

Micro-hardness (HRC)

Depth from top surface (m) Fig. 10. Micro-hardness distribution below the top machined surface (Ii=24A).

Fig. 12. SEM micro analysis result of workpiece cross-section (after machining) (I=24A; Ti=100s)

4. Results and discussion

4.1 Numerical results Figs. 13-16 show the variation of temperature with distance (along radius and depth of discharge position) for different values of pulse current at pulse on-time of 100s. Figs. 14 and 15 represent the key points required for calculating the white layer thickness, depth of heat affected zone and surface roughness, as rC1, rC2, and rC3 are the radius, and S1, S2, and S3 the depth of the molten crater at different pulse currents. Using these key points and Eqs. (8)-(13), the amounts of WT, HD, and Ra are calculated. 4.2 Experimental verification of the predicted values for the white layer thickness, depth of heat affected zone, and surface roughness Figs. 17-22 represent the variation of white layer thickness, depth of heat affected zone, and surface roughness with pulse on-time and pulse current for both theoretical and empirical results. The average deviations between the predicted WT,

HD, Ra and the experimental results are found 9.65%, 12.65%, and 18.3%, respectively. Generally, there is an acceptable correlation between the numerical results and experimental observations. This proves the validity of the presented model. 4.3 The effect of pulse on-time and pulse current on the white layer thickness, depth of heat affected zone, and surface roughness It was found that the effects of pulse on-time and pulse current on the white layer thickness, depth of heat affected zone, and surface roughness of the ED machined workpiece are quite different. The differences of these influences are detailed as follow: 4.3.1 The effect of pulse on-time and pulse current on the

white layer thickness and depth of heat affected zone

The increase in the white layer thickness and depth of heat affected zone by the increase in pulse on-time can be obviously seen for the both FEM and experimental results (Figs. 1718). This is justified by the fact that with an increase in pulse

Temperature (C)

Radial distance (mm) Fig. 11. SEM micro analysis result of workpiece cross-section (before machining) Fig. 13. The temperature distribution along the radial direction from the centerline of discharge position (Ti=100s).

0000

M. Shabgard et al. / Journal of Mechanical Science and Technology 00 (2010) 0000~0000

Temperature (C)

Radial distance (mm) Fig. 14. The temperature distribution along the radial direction at the A zone shown in figure (13) (Ti=100s); MT: Melting Temperature.

Temperature (C)

Depth from top surface at center (mm) Fig. 16. The temperature distribution along the depth of workpiece at the B zone shown in figure (15) (Ti=100s); MT: melting temperature; RT: re-crystallization temperature

on-time, plasma flushing efficiency decreases (Fig. 23). As a result, the ability of plasma channel for ejecting the molten material from the molten crater decreases. Subsequently, this remained molten material in the molten crater re-solidifies and forms the white layer upon the machined surface. Furthermore, any increase in the discharge duration results in more heat distributed into the workpiece per pulse, and consequently, more underlying material is affected by the high temperature. Overly, this phenomenon causes the increase in the white layer thickness and heat affected zone. Furthermore, better explanation is that the amount of molten material which can be flushed away at the end of each discharge is dependent on the plasma flushing efficiency [10]. Clearly, the %PFE is dependent on the discharge energy (W), energy changing rate (dW/dt), geometrical dimensions of the molten crater, pressure of the plasma channel (P), and changing rate of plasma channel pressure (dP/dt). According to the

Fig. 23, depending on the amount of mentioned parameters, %PFE has an optimum amount. As, with an increase in the pulse on-time, at first the %PFE increases and then decreases. The cause of this phenomenon could be justified by this reason that the increase in pulse on-time causes the decrease in the energy changing rate, as this causes a significant increase in diameter while not much increase in average temperature of the plasma channel, which leads to decrease in the pressure of the discharge channel and its changing rate. So, with regard to the mechanism of bulk boiling phenomenon, the amount of molten material, which is ejected from the molten material crater at the end of discharged, decreases and as a result, the %PFE decreases. Slight decrease in the white layer thickness and depth of heat affected zone by the increase in the pulse current is obvious from Figs. 20 and 21. Although the increase in pulse current leads to an increase in the dimensions of the molten

White layer thickness (m)

Temperature (C)

Depth from top surface at center (mm) Fig. 15. The temperature distribution along the depth of workpiece at the centerline of discharge position (Ti=100s).

Pulse on-time (s) Fig. 17. White layer thickness (WT) vs. pulse on-time (Ti) (I=24A).

M. Shabgard et al. / Journal of Mechanical Science and Technology 23 (2009) 1261~1269

1265

Depth of heat affected zone (m)

White layer thickness (m)

Pulse on-time (s) Fig. 18. Depth of heat affected zone (HD) vs. pulse on-time (Ti) (I=12A).

Pulse current (A) Fig. 20. White layer thickness (WT) vs. pulse current (I) (Ti=25s).

crater and the heat penetrating depth, the plasma flushing efficiency increases as pulse current increases (Fig. 24). The increase in plasma flushing efficiency causes more molten material to be swept away from the molten crater, therefore thinner layer of re-deposited material appears on the surface of workpiece. Since an increase in the penetrating depth of heat into the workpiece and plasma flushing efficiency counterbalance each other's effect, an increase in the pulse current has no significant effect on the depth of the heat affected zone. More explanations are that, with an increase in the pulse current and with a constant amount of pulse on-time, the amount of the energy changing rate increases, as this causes an increase in diameter and a sharp rise in average temperature of the plasma channel spontaneously, which leads to increase in the pressure of gap and its changing rate. So, regarding about the mechanism of bulk boiling phenomenon, the amount of molten material, which is ejected from the molten puddle at the end of each discharge, increases and as a result, the %PFE

increases (Fig. 24) as the reports of Marafona et al. [10] prove this matter. 4.3.2 The effect of pulse on-time and pulse current on the surface roughness The surface roughness is obviously affected by the amount of discharge energy. According to the Figs. 19 and 22, the increase in pulse on-time and pulse current cause the increase of surface roughness (Ra). This could be explained by the fact that, as the pulse on-time increases, the amount of transferred heat into the specimen increases, so the dimensions of molten crater increases. Furthermore, as mentioned previously, the %PFE decreases as the pulse on-time increases. Increase in the molten crater dimensions and decrease in the %PFE cause more molten material could not be swept away from the molten material crater by the collapse of plasma channel. During the cooling process, the remaining molten material resolidifies and forms the white layer on the machined surface.

Pulse on-time (s) Fig. 19. Arithmetical mean roughness (Ra) vs. pulse on-time (Ti) (I=16A).

Depth of heat affected zone (m)

Surface roughness (m)

Pulse current (A) Fig. 21. Depth of heat affected zone (HD) vs. pulse current (I) (Ti=25s).

0000

M. Shabgard et al. / Journal of Mechanical Science and Technology 00 (2010) 0000~0000

Pulse current (A) Fig. 22. Arithmetical mean roughness (Ra) vs. pulse current (A) (Ti=50s).

Plasma flushing efficiency

Surface roughness (m)

Pulse current (A) Fig. 24. Plasma flushing efficiency (%PFE) vs. pulse current (I).

2) The effect of this white layer also increases the surface roughness [22]. On the other hand, as the pulse current increases, discharge strikes the surface of the sample more intensely, and creates an impact force on the molten material in the crater and causes more molten material to be ejected out of the crater, and the surface roughness of machined surface increases.

3) 4)

5. Conclusions
Results from a numerical model for estimating the surface integrity characteristics, including the white layer thickness, depth of heat affected zone, and arithmetical mean roughness of the AISI H13 tool steel as workpiece machined by electrical discharge process have been presented. The leading conclusions are as follows: 1) An increase in pulse on-time leads to the increase in the white layer thickness and depth of heat affected zone.

5)

Slight decrease could be observed in the white layer thickness and depth of heat affected zone by an increase in the pulse current. The increase in both pulse on-time and pulse current, results in a coarser surface roughness By constant level of discharge energy, high pulse current and low pulse on-time leads to a reduction in the white layer thickness and depth of heat affected zone on the surface of EDMed workpiece. The presented FEM for EDM process has the capability of predicting WT, HD, and Ra of machined parts with average deviations of 9.65%, 12.65%, and 18.3%, respectively.

Acknowledgment
The authors of this study are indebted to the Razi Metallurgical Laboratory, Metallurgical Laboratory of Sahand University of Technology, universal workshop of Training Center of Iran Tractor Manufacturing Company, and advance machining workshop of Manufacturing Engineering Department of University of Tabriz. Also, we would like to appreciate the help of authors Professors J. Khalil Allafy and T.B. Navid Chakharlu for their invaluable technical support.

Plasma flushing efficiency

Nomenclature
c : Specific heat capacity of the workpiece material DRB : Depth of re-crystallization temperature border at the end of pulse on-time HD : Depth of heat affected zone I : Current density k : Thermal conductivity of the workpiece material M1 : Workpiece weight before machining M2 : Workpiece weight after machining Nnp : Number of normal pulses : Density of the workpiece material rc : Radius of the crater

Pulse on-time (s) Fig. 23. Plasma flushing efficiency (PFE%) vs. pulse on-time (Ti).

M. Shabgard et al. / Journal of Mechanical Science and Technology 23 (2009) 1261~1269

1267

rS time R(t) S T0 Ti Ts Ub VC(EXP) charge VC(FEM) charge WT

: Radius of plasma channel at the end of pulse on: Radius of plasma channel : Depth of the crater : Temperature of the dielectric fluid : Pulse on-time : Initial temperature : Electric potential : Empirical volume of removed material per dis: Numerical volume of removed material per dis: White layer thickness

602.
H.K. Kansal, S. Singh and P. Kumar, Numerical simulation of powder mixed electric discharge machining (PMEDM) using finite element method, Mathematical and Computer Modeling, 47 (11-12) (2008) 1217-1237. S.N. Joshi and S.S. Pande, Intelligent process modeling and optimization of die-sinking electric discharge machining, Applied Soft Computing, 11 (2) (2011) 27432755. V. Yadav, V. K. Jain and P. M. Dixit, Thermal stress due to electrical discharge machining, The International Journal of Machine Tools & Manufacture, 42 (8) (2002) 877-888.

References
O. A. Abu Zeid, On the effect of electro discharge machining parameters on the fatigue life of AISI D6 tool steel, Journal of Materials Processing Technology, 68 (1) (1997) 27-32. A. Hascalyk and U. Cayda, Experimental study of wire electrical discharge machining of AISI D5 tool steel, Journal of Materials Processing Technology, 148 (3) (2004) 362-367. G. Castro, A. Fernandez-Vicente, and J. Cid, Influence of the nitriding time on the wear behavior of an AISI H13 steel during a crankshaft forging process, Wear, 263 (7-12) (2007) 1375-1385. A. Abdullah and M. R. Shabgard, Effect of ultrasonic vibration of tool on electrical discharge machining of cemented tungsten carbide (WC-Co), The International Journal of Advanced Manufacturing Technology, 38 (11-12) (2008) 11371147. D. D. Dibitoto, P. T. Eubank, M. R. Patel and M. A. Barrufet, Theoretical models of the electrical discharge machining process. I. A simple cathode erosion model, Journal of applied physics, 66 (9) (1989) 4095 4103.

S. K. Hargrove and D. Ding, Determining cutting parameters in wire EDM based on workpiece surface temperature distribution, The International Journal of Advanced Manufacturing Technology, 34, (2007) 296299.
M. R. Patel, M. A. Barrufet and P. T. Eubank, Theoretical models of the electrical discharge machining process. II. The anode erosion model, Journal of applied physics, 66 (9) (1989) 4104 4111. Xia, H. Hashimoto, M. Kunieda and N. Nishiwaki, Measurement of energy distribution in continuous EDM process, Journal of Japanese Society of Precision Engineering, 62 (8) (1996) 11411145. S.H. Yeo, W. Kurnia and P.C. Tan, Electro-thermal modelling of anode and cathode in micro-EDM, Journal of Physics D: Applied Physics, 40 (8) (2007) 25132521. T. Ikai and K. Hashigushi, Heat input for crater formation in EDM, Proceedings of International Symposium for ElectroMachining- ISEM XI, EPFL, Lausanne, Switzerland April, (1995) 163-170. B. Izquierdo, J. Antonio Snchez, N. Ortega, S. Plaza and I. Pombo, Insight into fundamental aspects of the EDM process using multi-discharge numerical simulation, International Journal of Advanced Manufacturing Technology,52 (1-4) (2010) 195-206. K. Salonitis, A. Stournaras, P. Stavropoulos and G. Chryssolouris, Thermal modeling of the material removal rate and surface roughness for die-sinking EDM, The International Journal of Advanced Manufacturing Technology, 40 (3-4) (2009) 316-323. BHLER EDELSTAHL, http://www.bohler-edelstahl.at H. T. Lee and T. Y. Tai, Relationship between EDM parameters and surface crack formation, Journal of Materials Processing Technology, 142 (3) (2003) 676-683. A. Hascalyk and U. Caydas Experimental study of wire electrical discharge machining of AISI D5 tool steel, Journal of Materials Processing Technology, 148 (3) (2004) 362367. J. Simao, H.G. Lee, D.K. Aspinwall, R.C. Dewes, E.M. Aspinwall, Workpiece surface modification using electrical discharge Machining, International Journal of Machine Tools & Manufacture, 43 (2) (2003) 121128.

J. C. Rebelo, A. D. Morao, D. Kremer and J. L. Lebrun, Influence of EDM pulse energy on the surface integrity of martensitic steel, Journal of Materials Processing Technology, 84 (1-3) (1998) 90-96.
A. G. Mamalis, G. C. Vosniakos and N. M. Vaxevanidis, Macroscopic phenomena of electro-discharge machined steel surface: an experimental investigation, Journal of Mechanical Working Technology, 15 (8) (1987) 335-356. M. Boujelbene, E. Bayraktar, W. Tebni and S. Ben Salem, Influence of machining parameters on the surface integrity in electrical discharge machining, Archives of Materials Science and Engineering, 37 (2) (2009) 110-116. N. Ben Salah, F. Ghanem, and K. Ben Atig, Numerical study of thermal aspects of electric discharge machining process, International Journal of Machine Tools and Manufacture, 46 (7-8) (2006) 908 -911.

J. Marafona and J. A. G. Chousal, A finite element model of EDM based on Joule effect, International Journal of Machine Tools and Manufacture, 46 (6) (2006) 595

0000

M. Shabgard et al. / Journal of Mechanical Science and Technology 00 (2010) 0000~0000

Mohammadreza Shabgard received his PhD in manufacturing and production engineering at University of Tabriz, and currently is an associate professor at this university. His research interests mainly include the advanced manufacturing methods.

Samad Nadimi Bavil Oliaei received his MSc in manufacturing and production engineering at University of Tabriz, and currently is a PhD student at the Middle East Technical University. He is doing research in the field of micro-mechanical machining.

Mirsadegh Seyedzavvar received his MSc in manufacturing and production engineering at University of Tabriz under supervision of Professor M. Shabgard. His research interests mainly include the advanced manufacturing methods.

Ahmad Najadebrahimi received his MSc in manufacturing and production engineering at University of Tabriz, and currently is a researcher at the research center of East Azerbaijan branch of Iran Khodro Co. His research interests mainly include the utilization of FEM in welding and machining methods.