Performance evaluation of SiC powder mixed electrical discharge machining: a case of wire cut mode with re-circulating molybdenum wire

A review of the available literature on powder mixed electrical discharge machining (PMEDM) indicates that most of the research has been done for “die sinking machining mode” whereas the “wire cut machining mode” has not received due attention despite being an important process variant. This work employs silicon carbide (SiC) powder mixed dielectric fluid for machining of AISI D2 in “wire cut” mode with re-circulating molybdenum wire (an economic and chemically stable proposition as a tool). The effect of five process parameters (powder concentration, peak current, pulse on time, nozzle flushing pressure and stand-off distance) has been evaluated on surface roughness, kerf width, material removal rate, and wire wear ratio using Taguchi’s approach followed by analysis of variance (ANOVA) for determining statistically significant factors at 95% confidence level. It is found that for surface roughness, higher current and low to moderate concentration levels (2 to 4g/l) deteriorate surface quality; higher values of pressure and stand-off distance are also seen to adversely affect it. For material removal rate, pulse on time as well as its interactions with powder concentration and current, are statistically significant. A higher pulse on with smaller and moderate powder concentrations (2g/l and 4g/l) reduces MRR. For the wire wear ratio, the current is the sole significant factor (PCR of ~65%). SEM analysis of the machined workpiece for the maximum MRR condition quantifies the recast layer as ~19microns. An indirect comparison with the reported values for non-powdered EDM process indicates that for the similar wire (molybdenum), the use of SiC powder maintains the surface roughness and kerf values, for a much harder D2 material used in this work.


Introduction
Powder mixed electric discharge machining (PMEDM) has offered itself as an effective variant of the conventional EDM process. It involves adding certain conductive powders to the dielectric which influence spark attributes during machining and result in high surface quality and increased material removal rate [1,2]. The improvement in surface quality is attributed to an increase in discharge gap which reduces the generated heat flux, which then helps form smaller craters (thus better surface quality) [2][3][4][5]. Material removal rate (MRR), on the other hand, gets improved because of the increased spark frequency, which is the result of reduced dielectric resistivity [2]. Ease of removal of debris which is made possible by wider discharge gap [5] and smaller craters further contributes to an increase in MRR [2].
Review of the available PMEDM literature indicates that various powder relevant variables, namely the "material" itself, "concentration," and "size," influence the process outcomes differently [5][6][7][8][9]. Wong et al. [5] found the specific workpiece/powder-mixed dielectric combinations to be resulting into beneficial outcomes in their work wherein the effectiveness of Si, Al, SiC, Gr, MoS 2 and crushed glass were compared. Graphite and Si were found to be producing mirror like surfaces for SKH-54 tool steel workpiece material (even up-to a pulse current of 2 Amp) whereas Al was reported to be more effective in producing mirror like surfaces for the workpiece material SKH-51 [5]. For the other response measures as well, the favorable outcomes are not found to be specific to the use of any one powder material. Tzeng and Lee [8] reported Cr addition (in Kerosene) to be contributing most, to the increase in MRR than Cr, Cu, Al and SiC powders. They however found effectiveness of the powders towards tool wear rate to be different than pattern observed for MRR. Maximum tool deterioration was reported for the case of SiC, followed by Al and Cr. On another note, Cu was not found to be much influential for either response measure in their work. The ineffectiveness of Cu as an additive was attributed to its high density that kept it from remaining suspended in dielectric [8,9]. Chow et al. [10], in their work on micro-slit EDM reported SiC to be having higher material removal depth in comparison to Al at their respective optimum concentrations. In yet another work [11], addition of SiC powder in the de-ionized water was seen to increase the material removal depth. More recently, SiC powder mixed dielectric has been reported to have positive effect in reducing the surface roughness for hardened steel [12]. In yet another work employing a different powder material, i.e., Gr, its use was not only reported to improve MRR and tool wear rate [2], but was also found to alloy with the machined surface [11]. In a recent study, Sahu and Mandal [13] reported Gr based PMEDM to be generating the surface with maximum flatness for Nimonic 263 in the work that also employed alumina powder in the EDM oil. Other than the material (of powder) itself, the "powder concentration" has also been seen to play its part. The particles beyond an optimum level are mentioned to have similar effect to that of overcrowded debris in the gap thus causing unstable machining [2]. The ensuing bridging effect, short circuiting, arcing and settling issues have been attributed for adversely affecting the surface characteristics [2,4,8]. The use of powder concentrations upto 20 g/l is mostly reported; only limited work could be observed which employed excess concentrations [2]. Another important powder relevant variable seen to be investigated is the "particle size". It is seen to relate directly with the discharge gap produced. In general, the larger particles have been reported to result in wider discharge gaps whereas the s m a l l e r p a r t i c l e s p r o d u c e s m a l l e r g a p s [ 2 , 8 ] . Correspondingly, the improved material removal rate and tool wear ratio have been associated with the smaller particles [8]. In this regard, work can be found in both, micro and nano regimes [2,14,15]. The micro-regime and nano-regime have been seen to be employed mostly in the size ranges of 1-55μm and 20-60nm respectively [2].
While the preceding section discussed how various powder relevant variables generally affect the process outcomes, another important context of discussion pertains to their influence when compared with other process variables in a certain parametric combination. Kansal et al. [6] optimized the process parameters of powder mixed dielectric machining for MRR and SF using the response surface methodology (RSM). Silicon powder in the range of 20-30 microns was added to kerosene oil (base dielectric) while machining EN-31 tool steel with Cu electrode. They found the quadratic model to be statistically significant for explaining the relationship between the independent and the output variables. Powder addition not only increased the surface finish but also high concentration of the powder tended to increase MRR. They found maximum MRR at Si concentration of 2g/l and the peak current of 12 Amp (the highest employed in their work). Later, Kumar et al. [16] optimized the process for another tool steel grade EN-24 while employing the same process parameters but with different electrode dia. They employed grey relation analysis for process optimization. Batish et al. [17], in another work, evaluated parametric combination that not only involved various powder specific but also other conventional EDM variables while they machined various die steel grades. They found the powder concentration to be more dominant than certain electrical parameters (such as pulse on time and current) in terms of significance for the tool wear rate. For the case of overcut, they noted powder concentration to be the second highest contributing factor behind pulse on time. Singh and Yeh [18] used L18 orthogonal array with gray relational analysis to optimize the process when SiC abrasive powder mixed dielectric was used to machine 6061Al/Al2O3p/ 20p aluminum matrix composite. Eight input variables were considered in their work with performance measures being the material removal rate, tool wear rate and surface roughness. As per their confirmatory runs, an increase was noted for MRR (from 0.4267 to 0.530 g/min) whereas reduction was recorded for the tool wear rate (from 0.0112 to 0.0096 g/min) and the surface roughness (3.64 to 2.82 microns). They also concluded pulse on time to be relatively more important than other investigated parameters for the multiple performance characteristics studied in their work. The aspect ratio, tool electrode lift time, duty cycle, abrasive particle size, abrasive powder concentration, gap voltage, and pulse current were reported to follow the order of importance. In another work on composite machining via PMEDM, Singh et al. [19] used tungsten powder mixed dielectric to machine aluminum alloy 6061/10% SiC composite. They used RSM to model the effect of four electrical parameters (current, pulse on, pulse off and gap voltage) on MRR with and without the powder presence. They reported that although all four electrical parameters had significant role in MRR for both i.e. powder mixed and conventional EDM, however change in MRR with respect to pulse off time was more with PMEDM than that obtained for simple EDM. Additionally, the surface analysis of the machined workpiece indicated 42.85% reduction in thickness of the recast layer for the case of PMEDM in comparison to when only base dielectric was used. Long et al. [7] while machining different grades of steel in presence of titanium powder mixed dielectric concluded that interactions of titanium powder concentration with the workpiece material and with the electrode material were among the main factors that influenced MRR. In a more recent work, Kumar et al. [20] while employing different powder materials to machine Inconel-800 at different machining conditions, found powder particles along with current, pulse on time and tool material to be influential factors for the tool wear rate.
As evident from above literature, among the parametric sets investigated in specific studies, the powder specific variables have played different (but important role) in differing scenarios. Interestingly, it is pertinent to note that all of the aforementioned studies were undertaken in the "die sinking machining mode." The "wire cut machining mode" which is an important process variant and otherwise enjoys a respectable place in research and industry due to its extended capabilities, has not been given much attention by researchers in PMEDM domain. Very limited research could be found for wirecut PMEDM and that too only considered special machining scenarios [21,22]. Jarin et al. [21] while employing the wire cut mode, mixed carbon nanopowder (at three concentration levels of 0.2, 1, and 2 g/L) in commercially available dielectric to machine gold coated Si wafer at different voltage and capacitance values to compare the performance with an unmixed dielectric. In general, they noted higher MRR when machined with powder added dielectric, wherein the maximum increase was reported to be~33% at certain conditions in comparison to un-mixed dielectric. It was also observed that specifically, at lower discharge energies, MRR increased with increasing concentration of the powder. Incidentally the pattern of MRR increase was inconsistent at higher discharge energies which they attributed to more abrupt and uncontrolled erosion. They also noted higher spark gaps at higher discharge energies. In another study, Yeh et al. [22] mixed sodium pyrophosphate powder in pure water to machine polycrystalline silicon ingot in wire cut mode. The effects on cutting speed, kerf loss, and surface roughness were evaluated. They reported 2.4 times higher cutting speed and an improvement of 16% in surface roughness with powder mixed dielectric in comparison to an unmixed dielectric. An EDS analysis however, revealed infiltration of phosphorus into the machined surface for the case of sodium pyrophosphate mixed dielectric.
Ironically, other than the aforementioned special scenarios, no study could be located that thoroughly evaluates the role of powder mixed EDM in wirecut mode. It is important to realize that the dynamics of certain variables in wire cut mode are different than diesinker mode and therefore, the role of process specific variables in conjunction with the powder addition is found to lack a thorough investigation. It has also been cited by others [9] that there exists a need to further explore the powder material's potential especially when so many variables get combined [9]. Looking at the effectiveness of powders for "die sinking machining mode" and prevalence of significant research gap in the "wire cut machining mode of EDM," a thorough investigation in this regard is called for. It would be worthwhile to evaluate the role of variables and quantify the benefits for the powder assisted wire cut machining mode.
This work quantifies the role of different process variables while employing SiC powder during machining of AISI D2 steel in EDM wire cut mode with re-circulating molybdenum wire which not only provides an economic proposition to be used as a tool but is also reported to have less tendency to chemically react and contaminate the machined surface [23,24]. An average size of~15 micron SiC powder was taken with three levels of concentration, stand-off distance, flushing pressure, pulse on time and current. The surface roughness, material removal rate, kerf width and wire wear ratio were the output criteria. Taguchi L27 array was used for experimentation followed by main effect plots and analysis of variance (ANOVA) for determination of significant parameters at 95% confidence level. SEM analysis was used to evaluate the re-cast layer thickness of the optimized samples.

Workpiece specimen and setup details
A plate (20 × 10 × 10mm) of hardened AISI D2 steel with a hardness value of 56 HRC was used as the workpiece material. The typical composition and properties are given in Tables 1 and 2 respectively. Re-circulating molybdenum wire (dia 0.18 mm) was used as cutting tool to produce a cut of  10 mm length via EDM wirecut machine 1 . Figure 1 provides the schematic of the workpiece and the cut produced. Commercially available 99.9% pure silicon carbide powder with an average size of~15 micron was mixed in commercially available medium duty soluble hydrocarbon oil dielectric at three concentration levels in a tank. A rotary mixer at 2850 rpm was used for the purpose and it was operated continually during experimentation to prevent settling. The solution was pumped to the machining zone for jet flushing (from both, upper and lower nozzle) through piping network and pressure regulating valves at a flow rate of 40 l/min. Two calibrated pressure gauges attached at 1.5" from the outlets were employed to measure supply pressure variations at different experiment conditions. The workpiece was not submerged in dielectric so that role of parameters (stand-off distance and pressure) which could possibly influence the dielectric flow through the gap could be properly studied. Figure 2 presents the actual machining setup and schematic of the main components and working principle.

Design of experiments
Taguchi L27 orthogonal array was used as design of experiment (DOE) for the experimentation phase. Five factors namely current, pulse on, powder concentration, stand-off distance and pressure were varied at three different levels whereas the powder type & material were constants. The levels of variables were selected while taking guide from the available literature on PMEDM. Table 3 gives the details of the variables' levels employed during experimentation. Table 4 contains constants and key parameters of machine.

Post-experimental measurements and analysis
The surface roughness was quantified by measuring surface roughness parameter "Ra" using surface texture meter 2 with 4mm evaluation length. Ra was measured at four random locations over the cut surface and the average value is reported herein. Coordinate measuring machine 3 was used to measure Kerf width of the workpiece after making a cut of 9mm length (before parting it off completely). Images were taken at the magnification of 70× and three measurements were done (graphically) for each upper and lower side of the workpiece. Average values are reported herein.
Material removal rate (MRR) and Wire wear ratio (WWR) was calculated using equations (1) and (2) where WWL is wire weight loss and IWW is initial wire weight.
The main effect plots and analysis of variance (ANOVA) were used for determination of significant parameters at 95% confidence level. The recast layer thickness of the sample with maximum MRR was measured using scanning electron microscope 4 . Etching of the specimens was done in a mixture composed of 2% Nital for 1 min.
3 Results, analysis, and discussion Table 5 gives the results for the experiments performed using powder mixed wire electric discharge machining.

Surface roughness
The main effect plots of average surface roughness values are shown in Figure 3. Two interactions were studied in the analysis. The effect of interactions is given in Figure 4 whereas Table 6 gives the ANOVA calculations done to identify significant factors. The current had the maximum percentage contribution (PCR) of 29.02% followed by the powder concentration (PCR of 21.7%). The surface finish is seen to be deteriorated with an increase in current. At higher current levels, high intensity sparks would be generated thus producing more deep craters. This translates into higher roughness values. Looking at the general trend for the powder concentration, it gets indicated that surface quality deteriorates with it 1 EDM Machine: Model DK 7725A Fig. 1 Schematic of the workpiece specimen and cut 2 Surface Texture meter: Taylor Hobson Surtronic S128 3 Coordinate measuring machine: CE450 DV 4 SEM: JEOL JSM 6100 in general. Specifically in the range of 2 to 4g/l, the surface deteriorates and then afterwards at higher concentration (6g/l), some improvement may be expected. It can be explained on the basis that as powder gets added, it inhibits the insulating properties of the dielectric thus reducing resistance to current flow. In the lower ranges of the concentration, this reduced resistance would be confined to limited areas/paths (where bridging would occur) so energy could be more focused and intense as it impinges on workpiece surface thereby resulting into increase in crater depths. As the concentration increases to 6g/l, the saturation would help in achieving widespread bridging thus distributing the energy to multiple locations at workpiece; this would improve the surface finish. The standoff distance and pressure are also two factors that are found to be statistically significant. A possible explanation for their role could be put forth as well. The two factors have got to do with the flow characteristics of the incoming fluid stream and their higher values are seen to result in the deteriorated surface finish. A higher value of stand-off distance may have contributed to uniform entry of the powdered dielectric (moving along the wire) into the gap (as opposed to splashing etc that may have been the case otherwise). This phenomenon as well as higher pressure may have influenced the number of particles and the effective time each individual particle may be available with for contributing to the whole process. The availability of larger number of particles (by virtue of smooth entry to the gap due to appropriate stand-off distance) with higher pressure head (as opposed to the velocity head) is expected to have given the powders enough time to have culminated into producing deeper craters on the surface. The results     also show that although pulse on time is not statistically significant in independent capacity however its interactions with current and powder concentration are found to be significant. A moderate pulse on time of 130 μsec with minimum powder concentration (2g/l) and current (2A) is found to be resulting into minimum surface finish. This combination seems to spread and dilute discharge energy thereby producing minimum craters. Although no work could be found for the role of powder in wire EDM mode for D2 steel, however some indirect comparisons may be made. A previous work that used molybdenum wire for the electrical discharge machining of HSLA steel without any powder additive, reported the surface roughness of the scale of~6microns [27]. It is pertinent to mention that HSLA steel employed in the previous work had hardness value of 38 HRC with experimentation conducted at pulse on values of 32 & 96 microseconds. Although the average value of the surface roughness obtained in the work presented herein (with SiC powdered EDM) is also of the comparable scale (~6microns) however the work presented herein has employed D2 steel that has a much higher hardness of 56 HRC; additionally high pulse on values (100-150μs) have been employed. In a more recent work, Ishfaq et al. [28] reported an improved surface roughness value of 4.47microns with moly wire (without powder) but their work employed much softer SS304 in comparison to the workpiece employed herein. In this context, the current work demonstrates the capability of SiC powdered EDM for generating good quality surfaces for harder workpieces. Figure 5 gives the comparison of the surfaces generated of two extreme qualities along with their measurements. As already presented in Table 5, experiment run 2 gave the least surface roughness with an average Ra of 5.0 microns. On the other hand experiment run 7 resulted in the highest Ra value of 7.8 microns. As can be seen from the topographic images of the surfaces produced, the parametric combination (60mm, 2g/l, 30kPa, 2A, 130μs ) of experiment run 2 is found to be much effective as it results in much smoother surface (with 35% less average Ra) in comparison to surface generated with highest average Ra value.
Additionally, the surfaces produced at experiment conditions yielding some important results for the other output measures considered in this work are presented in Figure 6 with their surface roughness values mentioned against each. Not only that the surface topography is visible for these but it may also be deduced that the primary difference in the machined surfaces is exhibited by Ra value.   Figure 7 presents the main effects plots for material removal rate (MRR) whereas the effects of significant interactions are shown in Figure 8. The corresponding ANOVA calculations are presented in Table 7. The analysis yields that not only the pulse on time itself is a significant factor but also its interactions with current and powder concentration are significant. Additionally, pressure is also seen to have statistical significance. As can be seen from the results, pulse on time has a negative effect on MRR; an increase in pulse on time reduces MRR. It is believed that with an increase in pulse duration, the pulse energy is shared by many electrons/ions which may have contributed to the decrease in MRR. The argument takes strength considering that the pulse on time has got a strong interaction effect with powder concentration (incidentally this interaction had a maximum PCR of 28.29%). The interaction plots show that at two lesser powder concentration levels of 2g/l and 4g/l, there is a continual decrease in MRR with increase in pulse on. The situation changes when the higher concentration level of 6g/l is used which increases the MRR. The distribution of pulse energy at lower concentration levels of powder is expected to dilute the effect of heat intensity which could have eroded the material at much faster rate thus resulting in lesser MRR. At higher concentration of 6g/l however, the increased MRR may be attributed to the bridging effect that becomes more dominant; stronger multiple spark locations at the work piece thus enable the higher MRR in this case. The interaction between pulse on time and current is also significant with PCR of 12.06%. For longer pulse durations of 130μs and 150μs, 3A current is found to be resulting in higher MRR with certain other set of variables however for shorter pulse duration of 100μs, 4A current generates maximum MRR. This is likely due to the fact that shorter pulse duration (100μs) may have insufficient time for the chips to be removed hence maximum current is required for effective deionization of the die-electric while longer pulse durations (130μs and 150μs) requires moderate current (3A) for similar effect. For the effect of pressure which comes out to be another significant parameter, it may be hypothesized that the extreme values of pressure facilitate more favorable fluid flow characteristics that enable a good mix of powder particles availability as well as time for them to contribute positively to MRR. At intermediate pressure value, may be the flow characteristics are such that a transitory phase exists that is less than the optimum mix of powder quantity & their time in the gap.

Kerf width
The images of kerf width measurements for experiment runs 22, 9, 7, 2, 12, 1 and 5 (conditions where maximum and least parameters were measured) are given in Figure 9. The main effect plots for kerf width are given in Figure 10 whereas Table 8 shows the ANOVA calculations. No factor is found to be statistically significant at the pre-defined confidence level of 95%. Though, it is seen that "stand-off" is the factor that has PCR in excess of 15%. The trend of main effect indicates that increase in stand-off distance increases the kerf width. The possible reason for this could be that since increase in   stand-off increases the distance from where fluid gets thrown, it is very likely that this increase reduced splashing of the fluid near the gap. Too close a supply with fluid splashing would prevent its entry to the gap in an effective manner. On the other hand, an increase in the stand-off distance could make fluid to move along the wire making its entry more uniform thereby facilitating more particles to be available in the gap at a certain time. The effect of the larger number of powder particles once available between electrode and the workpiece, would be to increase the discharge frequencies. thus resulting in larger over cut. The average kerf value of~0.28mm obtained for the current work with re-circulating molybdenum wire is of the same order as the one (~0.29mm) reported by Ikram et al. [29], when they employed single pass brass wire to machine D2 steel of comparable thickness at optimized conditions but without any powder addition. This shows that the currently reported values are pretty much maintained at previously reported values even as a re-circulating wire is employed for this hard material. Another comparison can be made with a more recent work by Ishfaq et al. [28] wherein they employed molybdenum wire for machining of much softer SS 304 and reported the post improvement kerf value of~0.322mm. Thus, in current work, good control over the process in terms of kerf width is indicated by employing SiC powder.

Wire wear ratio
The results of main effects plots for wire wear ratio are shown in Figure 11 while ANOVA analysis is shown in Table 9. Among all evaluated factors, current is the only factor which was statistically significant with maximum PCR of~65%. The enhanced current intensity of the spark seems to be taking toll on the tool (wire). It is important to mention that wire wear ratio has not been found to be quantified in the PMEDM domain so a direct comparison may not be possible. Nevertheless, Batish et al. [17] while working in die sinking mode of powder mixed EDM for die steels, observed similar pattern for the tool wear rate. In general, the minute amount of wire wear observed herein is in line with the previous findings [24]. Azam et al. [24] reported molybdenum wire to be much stable in their work. Their evaluation of the wire post-machining, yielded it to be un-corroded with no contaminated surface; additionally, there was negligible change in its diameter. .

Recast layer thickness
The machined work surface was analyzed for assessment of the thickness of recast layer generated during WEDM process where highest material removal rate was achieved, i.e., experiment no. 22 (80mm, 4g/l, 30kpa, 4A, 100μs). Measurements were taken at multiple locations and an average value is reported herein. A representative SEM image is shown in Figure 12.
The measurements indicated the re-cast layer thickness of around~19microns. The values ranging from~14 to~21microns of the re-cast layer alone have been presented in previously reported literature with moly wires and HSLA steel without powder [23]. Here too, the value obtained is of the similar order.

Conclusions
In this study, the effect of five process parameters on the machining outputs (surface roughness, material removal rate, kerf width and wire wear ratio) of AISI D2 steel has been investigated in powder mixed wire electric discharge machining using Taguchi L27 orthogonal array. Following conclusions can be drawn from the presented work.
& For surface roughness (Ra) of the machined workpiece, four of the five investigated parameters namely current, powder concentration, pressure and stand-off distance were found to be statistically significant at 95% confidence level. Furthermore, interactions of pulse on time with current as well as with powder concentration, were statistically significant. Current was the most contributing factor for Ra with PCR of 29.02% followed by powder concentration (PCR 21.70%). As a general trend, the surface quality deteriorated at higher currents, specifically at low to moderate concentration levels (2 to 4g/l). This was most probably due to their role in facilitating high intensity sparks which would result in deeper craters on machined surface (and thus high Ra). A higher powder concentration (6g/l), tended to show improvement in surface quality which could be attributed to greater distribution of the discharge energy as a result of greater number of powder particles. On another note, the higher values of pressure and stand-off distance were seen to be associated with deteriorated surface quality. The effect of pressure and stand-off distance has been explained on the basis of their possible role in influencing flow characteristics of the dielectric stream which would in turn influence the number of particles available within the gap between workpiece and tool at a certain instant. & For material removal rate, pulse on time as well as its interactions with powder concentration and current, has been found as statistically significant. Additionally, dielectric pressure also came out to be significant. Here, higher pulse on time was associated with lower MRR. The results are better explained when due consideration is given to interactions. A higher pulse on with smaller & moderate powder concentrations (2g/l and 4g/l) is expected to distribute/dilute the effect of heat intensity thereby reducing MRR. A higher powder concentration (6g/l) resulted in improved MRR which could be explained on the basis of enhanced bridging phenomena. & For wire wear ratio, current was the sole statistically significant factor with PCR of~65%. A higher value of current resulted in more wire wear ratio. & An SEM analysis of the machined workpiece done for the conditions where maximum MRR was achieved indicated the recast layer of~19 microns. & An indirect comparison with the reported values indicate that the use of SiC powder maintains the surface roughness and kerf values for a much harder D2 material when machined with re-circulating molybdenum wire.