CHAPTER 4 RESULTS AND DISCUSSION

Transcription

1 66 CHAPTER 4 RESULTS AND DISCUSSION 4.1 PERFORMANCE ANALYSIS In recent years, several researches have been carried out on EDM and die-sinking micro-edm using various electrodes. In die-sinking micro- EDM, the electrode plays a vital role as it is employed to produce corresponding mirror images. Hence, it is desirable to have stability in the shape of the electrode during the process. Therefore, a detailed experimental investigation was carried out with different machining conditions to evaluate the performance of electrodes such as tungsten, copper, copper tungsten and silver tungsten on EN24 die steel. During the study, MRR, TWR, overcut, circularity error, SR and HAZ characteristics were monitored to evaluate the response of the materials in die-sinking micro-edm and the results obtained are discussed in this chapter. In micro-edm, fabrication of micro-parts requires minimization of the pulse energy supplied into the gap, since the performance of the micro- EDM is directly related to the discharge energy. However, in most cases, transistor-type pulse generator is used to supply the discharge energy which is good for conventional EDM (Han et al. 2004). But transistor-type pulse generator sometimes may not fulfill all the requirements. On the contrary, RC-type pulse generator can produce very small discharge energy compared

2 67 to transistor-type and it is used to produce better quality micro-holes. (Jahan et al. 2009b). Therefore, RC-type pulse generator was used to study the performance of the die-sinking micro-edm in the present work. A series of experiments was conducted using different electrodes to study the effect of major operating parameters such as gap voltage, capacitance, feed rate, threshold and fixed polarity on machining performance. 4.2 EFFECT OF OPERATING PARAMETERS IN RC-TYPE PULSE GENERATOR In die-sinking micro-edm, it is an important step to select proper processing parameters. Gap voltage, capacitance, feed rate and threshold can greatly affect processing status to achieve desired results and efficiency Effect of Gap Voltage Gap voltage is one of the important operating parameters in the RCtype die-sinking micro-edm. It is related to effective electrode gap/distance between the workpiece and electrode during the spark. As sparks occur across a narrow gap between tool and workpiece, the gap condition has a strong influence on machining stability and thus has to be monitored throughout the machining process (Yeo et al. 2009a). When the current is flowing, the voltage drops and stabilizes at the working gap level. Therefore, with the increase of gap voltage, the MRR increases. This is because with the increase of gap voltage, the discharge energy increases along with the spark gap. Also, more materials that are removed from the workpiece can be flushed easily leaving space for the newly machined material. Furthermore, the tool wear also increases with the increase of gap voltage due to increased discharge energy.

3 Effect of Capacitance In RC-type, the capacitor controls the charging and discharging process. Therefore, the performance of the micro-edm in RC-type is more influenced by the capacitance. Actually, at higher value of capacitance, the MRR increases as the larger capacitance results in deeper craters. To achieve better surface quality there is a need to minimize the discharge energy and this can be obtained by using a very low value of capacitance in the RC-type Effect of Feed Rate A feed rate is the most significant parameter which influences MRR and surface roughness. When the feed rate is low, the tool moves slowly and this may cause more electrode wear as more sparks occur at the same position on the workpiece. With the increase of feed rate, both MRR and SR increase Effect of Threshold The threshold value has a significant influence on the die-sinking micro-edm because it determines the reliability of spark pulses. 4.3 ANALYSIS OF OUTPUT PERFORMANCE The performance of die-sinking micro-edm process was evaluated in terms of MRR, TWR, overcut, circularity error, SR and HAZ are discussed in this section Effect of Different Electrodes on MRR Experiments were carried out to evaluate MRR. It was found that MRR strongly depended on the properties of electrodes and discharge energy.

4 69 Figure 4.1 Parametric influences on MRR with different electrodes From Figure 4.1, it is found that among the electrode materials, Cu facilitates highest MRR while with W electrode only least MRR is realized. Next to copper, AgW showed better MRR than other two electrodes and also MRR registered a rise up to the range of gap voltage V, above which it drops down. A typically observed electrode specific influence of effective discharge energy on material removal rate for different electrodes is illustrated in Figure W Cu CuW AgW Effective discharge energy in µj Figure 4.2 Effective discharge energy on MRR with different electrodes

5 70 The experimental data on MRR for different operating conditions are listed in Table 4.1. E e = ½ CV 2 Th Effective discharge energy is calculated based on the equation Where, C Capacitance (nf), V Gap Voltage (V), Th Threshold (%) Table 4.1 Experimental results for MRR Sl. No. Parameters Material Removal Rate (MRR) mm 3 /min W Cu CuW AgW Effective discharge energy (µj) 1 80V, 0.1nF, 2µm, 20% V, 1nF, 4µm, 40% V, 10nF, 6µm, 60% V, 100nF, 8µm, 80% V, 0.1nF, 4µm, 60% V, 1nF, 2µm, 80% V, 10nF, 8µm, 20% V, 100nF, 6µm, 40% V, 0.1nF, 6µm, 80% V, 1nF, 8µm, 60% V, 10nF, 2µm, 40% V, 100nF, 4µm, 20% V, 0.1nF, 8µm, 40% V, 1nF, 6µm, 20% V, 10nF, 4µm, 80% E V, 100nF, 2µm, 60%

6 71 The different operating / parametric combinations with different effective discharge energy yielded varying amount of MRR as in Table 4.1. The significance of four levels of discharge energy on MRR is shown in Figure 4.2. Various levels and their ranges are given below: Level I to 0.576µJ Level II 1.28 to 4.32µJ Level III 10 to 78.4µJ Level IV 144 to 588µJ With level I of discharge energy MRR tends to rise up to mm 3 /min followed by a drop. The least value of MRR is seen with 0.576µJ discharge energy which can be attributed to relating high threshold (80%) used. With higher threshold there is a time delay in every release (pulse off duration) causing a reduction in MRR. With level II of discharge energy, it is seen that copper electrode exhibited progressing increase in MRR, while other electrodes tends to exhibit drop in MRR with higher discharge energy. With level III of discharge energy, it is seen that up to certain energy level MRR increases, followed by drop in higher energy. With level IV of discharge energy, MRR tends to drop down mostly with increasing discharge. From the illustration, it is seen that among the electrodes copper facilitated higher order MRR. Also considering four levels of discharge energy, it is seen that with higher discharge energy MRR drops down. Figure 4.3 represents the effective discharge energy with a constant feed rate of 2µm/s, 4µm/s, 6µm/s and 8µm/s, respectively.

7 W Cu CuW AgW W Cu CuW AgW Effective Discharge Energy (µj) Effective Discharge Energy (µj) (a) 2µm/s (b) 4µm/s W Cu CuW AgW W Cu CuW AgW Effective Discharge Energy (µj) Effective Discharge Energy (µj) (c) 6µm/s (d) 8µm/s Figure 4.3 Effective discharge energy on MRR with different feed rate It is seen that with fine feed rate of 2µm/s, a marginal variation can be seen and Cu electrode exhibits an appreciable higher order MRR compared to other electrodes, owing to its high order thermal conductivity. Also it is seen that beyond certain specific energy (4.32µJ) MRR either sets-in or drops down. With increased energy level possible wear of electrode and also excessive heating of the work material (reduction in hardness) can contribute to this. With 4µm/s feed rate, MRR tends to drop down progressively with increasing energy up to 78.4µJ, above which a rise can be seen. Influence of feed rate pertains to low energy loss (up to 4.32µJ)

8 73 low feed rate of 2µm/s shows a rise in MRR. It is appreciable with Cu and relatively smaller and only marginal difference between other electrodes (possibly due to the influence of W). With 6µm/s mostly reduced order compared to 2µm/s while a reduction in MRR is seen with Cu, other electrodes exhibit a rise in MRR. With 8µm/s while Cu exhibits a rise in MRR, others exhibit a drop possibly with higher feed rate of 8µm/s with associated tendency of shorting / higher order temperature. Higher conducting Cu shows a reduction with 4µm/s which could be attributed to desire clogging. Relative comparison of MRR with level of specific energy < 1µJ with low feed rate of 2µm/s, as low energy level of 0.064µJ the MRR was about mm 3 /min and with a marginal variation between different electrodes 0.3 to 0.4µJ. With feed rate of 4µm/s, MRR vary in the range of to mm 3 /min with Cu exhibiting a higher MRR. With higher feed rate of 8µm/s, a reduction in MRR ( to mm 3 /min) was observed. This can be attributed to possible tendency for short/reduction in electrode wear. With 4µm/s feed rate and energy level of 1.28µJ, a relatively low order MRR varying over mm 3 /min was observed. With 6µm/s feed rate and energy level of 1.96µJ, a rise in MRR ( to mm 3 /min) was observed. With higher feed rate of 8µm/s and energy level of 4.32µJ, a relatively higher order MRR (0.0027mm 3 /min) was observed with Cu electrode. Only a marginal order ( to mm 3 /min) was observed with other electrodes. It is seen that gap voltage and capacitance exert more influence on MRR than any other parameters. Further, it is seen that gap voltage up to medium range, lower capacitance, higher feed rate and medium threshold facilitate better MRR.

9 Effect of Different Electrodes on TWR Each electrical discharge generates a shock wave in the dielectric fluid which removes the surface layer of the workpiece to leave craters. Similar craters are formed on the electrode surface. The VMS image of tool wear for various electrodes with different parameters is shown in Figure 4.4. With the increase of discharge energy the tool wear increases. Moreover, copper electrode exhibits more wear than other electrodes. W - electrode Cu - electrode CuW - electrode AgW - electrode (a) Lower discharge energy (80V, 0.1nF, 2 µm/s and 20%) W - electrode Figure 4.4 (Continued) Cu - electrode

10 75 CuW - electrode AgW - electrode (b) Medium discharge energy (100V, 10nF, 8 µm/s and 20%) W - electrode Cu - electrode CuW - electrode AgW - electrode (c) Higher discharge energy (140V, 100nF, 2µm/s and 60%) Figure 4.4 VMS image of tool wear for different electrodes The experimental data on TWR for different operating conditions are listed in Table 4.2.

11 76 Table 4.2 Experimental results for TWR Sl. No. Parameters Tool wear ratio (TWR) in % W Cu CuW AgW Effective discharge energy (µj) 1 80V, 0.1nF, 2µm, 20% V, 1nF, 4µm, 40% V, 10nF, 6µm, 60% V, 100nF, 8µm, 80% V, 0.1nF, 4µm, 60% V, 1nF, 2µm, 80% V, 10nF, 8µm, 20% V, 100nF, 6µm, 40% V, 0.1nF, 6µm, 80% V, 1nF, 8µm, 60% V, 10nF, 2µm, 40% V, 100nF, 4µm, 20% V, 0.1nF, 8µm, 40% V, 1nF, 6µm, 20% V, 10nF, 4µm, 80% V, 100nF, 2µm, 60% W Cu CuW AgW Parametric combination (Gap Voltage V) Figure 4.5 Parametric influences on TWR in % with gap voltage

12 77 Figure 4.5 shows the parametric combination of various gap voltages on tool wear ratio. Cu shows higher tool wear ratio than other electrodes used. Due to high voltage, the gap increases and helps to improve the flushing conditions and stabilize the cut. The gradual increase in discharge energy also shows increase in wear. The observed variations of tool wear ratio (specific tool wear) with discharge energy are shown in Figure W Cu CuW AgW Effective discharge energy in µj Figure 4.6 Effective Discharge Energy on TWR in % with different electrodes It is seen that copper electrode exhibits the higher order TWR, while other electrodes exhibit relatively smaller order TWR. Among these, tungsten electrode exhibits least order. Figure 4.2 and 4.6 reveal that higher order TWR is associated with a reduction in MRR. With level II of discharge energy, the tool wear ratio tends to rise and drop down. It is seen that higher order threshold (80%) could have caused a reduction in TWR. With level III of discharge energy, it is seen that both TWR and MRR tend to rise up to certain energy level followed by a drop. With higher discharge energy, a reduction in TWR is associated with a

13 78 reduction in MRR. With level IV of discharge energy, TWR tends to drop and rise as in the case of MRR. From Figure 4.2 and 4.6 it is seen that with relatively smaller order energy level (level II) better erosion can be achieved with copper electrodes. Moreover, in the lower discharge energy minor variations in the tool wear ratio are observed. In the medium and high discharge energy, copper shows higher electrode wear than that of other electrodes as it exhibits identical machining conditions due to its highest thermal conductivity. Similarly, wear of copper is greater than the other electrodes used due to its lower density and hardness. Figure 4.7 represents the parametric combination on TWR of effective discharge energy with constant feed rate of 2µm/s, 4µm/s, 6µm/s and 8µm/s, respectively. From Figure 4.7 it is seen that with 2µm/s feed rate, an increase in TWR is observed up to 72µJ. Moreover, tungsten-based electrodes exhibit lower TWR than other electrodes W Cu CuW AgW W Cu CuW AgW Effective Discharge Energy (µj) Effective Discharge Energy (µj) (a) 2µm/s (b) 4µm/s Figure 4.7 (Continued)

14 W Cu CuW AgW W Cu CuW AgW Effective Discharge Energy (µj) Effective Discharge Energy (µj) (c) 6µm/s (d) 8µm/s Figure 4.7 Effective discharge energy on TWR with different feed rate With 4µm/s feed rate, TWR tends to show a rise with the increasing energy. With 6µm/s only marginal variations in TWR are seen with other electrodes whereas with Cu electrode a rise and fall can be observed. With 8µm/s also Cu exhibits a rise and fall in TWR compared to other electrodes. With lower feed rate when discharge energy increases, rise in TWR is observed with almost all electrodes. Due to the increase in feed rate, an increase in TWR is observed to a certain level and in the higher feed rate TWR drops down. This is because when feed rate increases the tool moves rapidly and reduces the stability of sparks and it may cause more electrode wear as more sparks occur at the same position on the workpiece. It is thus evident from the experimental illustrations, copper exhibits higher TWR followed by copper tungsten than other electrodes. Tungsten shows lower TWR due to its high melting temperature and lower thermal conductivity compared to other electrodes.

15 Observation on Overcut Apart from MRR and TWR, the performance of various electrodes is also assessed from the overcut point of view. In micro-edm process, the size of the eroded hole with different electrodes varies due to end erosion, side erosion and stiffness/tension of the various electrodes. During machining process, overcut occurs due to side erosion and removal of debris. Overcut is also one of the major parameters to be considered to evaluate the machining performance of die-sinking micro- EDM. Pradhan et al. (2009) observed from the experimental investigations that the peak current and pulse-on time used in the machining process influence the overcut of the machined micro-holes. In the present analysis, the difference between the radius of the micro-hole and the radius of the electrodes were measured. The accuracy of the micro holes was also evaluated by overcut of the fabricated micro holes. The experimental data on overcut for different operating conditions are listed in Table 4.3. Sl. No. Parameters Table 4.3 Experimental results for overcut Overcut (%) W Cu CuW AgW Effective discharge energy (µj) 1 80V, 0.1nF, 2µm, 20% V, 1nF, 4µm, 40% V, 10nF, 6µm, 60% V, 100nF, 8µm, 80% V, 0.1nF, 4µm, 60% V, 1nF, 2µm, 80% V, 10nF, 8µm, 20% V, 100nF, 6µm, 40% V, 0.1nF, 6µm, 80% V, 1nF, 8µm, 60% V, 10nF, 2µm, 40% V, 100nF, 4µm, 20% V, 0.1nF, 8µm, 40% V, 1nF, 6µm, 20% V, 10nF, 4µm, 80% V, 100nF, 2µm, 60%

16 81 The overcut was assessed for all the erosion trials. The typical variation of overcut (of the eroded hole) influenced by parametric combinations of four levels of gap voltage (80, 100, 120 and 140V) is shown in Figure W Cu CuW AgW Parametric combination (Gap Voltage V) Figure 4.8 Parametric combinations of overcut with gap voltage It is seen that with 80V and 100V centred parametric combination, minimum variations of overcut was observed. The overcut tends to increase with higher energy centred with the parametric combination of 120V and 140V. A rise in overcut with wide variation can also be seen for all the electrodes. Experimental investigations also revealed that Cu and CuW show the maximum overcut. From Figure 4.9, it is seen that with lower energy level I, overcut tends to rise with discharge energy. This is reflected in the observed rise in electrode wear. With level II of discharge energy, overcut tends to drop down. The observed rise in MRR (Figure 4.2) and drop in TWR (Figure 4.6) indicate a reduction in tool wear and associated drop in overcut. With energy level III and IV, a rising trend in overcut (Figure 4.9) is seen. This is supplemented by the rising TWR (Figure 4.6)

17 W Cu CuW AgW Effective discharge energy in µj Figure 4.9 Effective discharge energy on overcut variation in % for different electrodes The observation on overcut also indicated that good performance in erosion of die steel can be attained with relatively smaller discharge energy up to (level II). Also unlike the case of MRR and TWR, not much of the distinction in overcut with four different electrodes can be observed. It is also seen that the four electrodes could have exhibited varying order of end erosion (causing a difference in MRR). Marginal variations in side erosion could have resulted and reflected in the observation of overcut. With medium discharge energy range, relatively constrained overcut occurs and uniformly good quality erosion can thereby be obtained. With lower and higher discharge energy, high percentage of overcut is observed. Overcut tends to increase with higher voltage combination and higher discharge energy. It is found that with the increase of discharge energy, the overcut increases. At lower discharge energy, smaller amount of material is removed per discharge producing smaller craters, which in turn results in low overcut.

18 83 One of the main reasons for this result is that with the longer duration of discharge, the electrons released from the negative poles collide with the neutral particles in the dielectric fluid, resulting in greater ionisation effect. The greater the number of electrons and ions colliding with the workpiece, bigger the micro-hole expansion would be (Mathew et al. 2009) Parametric combination of gap voltage (V) W Cu CuW AgW Figure 4.10 Average and range of overcut with different gap voltages Figure 4.10 illustrates the average of the overcut and the range which is highlighted using band width. Among the electrodes, W exhibits only a mild/marginal variation in overcut with gap voltage parametric combination. Cu exhibits wide variations in overcut. It also tends to rise up to 100V, followed by a marginal drop. CuW shows increasing variation (overcut, with gap voltage parametric combination). AgW exhibits increased overcut up to 120V followed by a drop. According to the experimental results, the hole expansion (overcut) increases as the voltage increases. It is found that with the increase of discharge energy, the overcut increases. At higher levels of capacitance and voltage, spark energy per pulse is greater. This high spark energy produces a

19 84 larger amount of debris. The debris sticks on the workpiece trap and may cause unwanted spark. The unwanted sparks, cause tool material erosion, which results in less overcut. Debris can also cause shorting and reduced performance. Furthermore, it is found that tungsten followed by silver tungsten shows lower overcut than other electrodes Observation on Circularity Error Circularity is the measure of concentricity, associated with form/geometric accuracy. The circularity can also provide a quantitative evaluation of surface irregularities. From the plotted points, the maximum peak-to-valley values of the machined surface is then measured and reflected as the circularity of the hole. If the micro-tool has poor circularity, it may affect the geometric and dimensional accuracies of the machined micro-hole. The circularity of spark machined holes is influenced by two types of electric source, RC-type generator and transistor-type generator. It is proved from the experimental results that good surface finish and circularity is achieved by RC type generator (Jahan et al. 2009a). In micro-edm drilling, due to the vibration of the electrode high roundness error occur (Ali et al. 2009). The surface finish and circularity is also influenced by the rotation of the electrode (Egashira et al. 2010). Circularity of the spark eroded hole depends largely on the inter electrode gap condition, which depends on the stability of tool electrode, formation of HAZ on the workpiece and the side erosion. They are mostly energy specific. The experimental data of circularity error for different operating conditions are listed in Table 4.4.

20 85 Table 4.4 Experimental results for circularity error Sl. No. Parameters Circularity error in µm W Cu CuW AgW Effective discharge energy (µj) 1 80V, 0.1nF, 2µm, 20% V, 1nF, 4µm, 40% V, 10nF, 6µm, 60% V, 100nF, 8µm, 80% V, 0.1nF, 4µm, 60% V, 1nF, 2µm, 80% V, 10nF, 8µm, 20% V, 100nF, 6µm, 40% V, 0.1nF, 6µm, 80% V, 1nF, 8µm, 60% V, 10nF, 2µm, 40% V, 100nF, 4µm, 20% V, 0.1nF, 8µm, 40% V, 1nF, 6µm, 20% V, 10nF, 4µm, 80% V, 100nF, 2µm, 60% Typically observed variations of circularity error influenced by parametric combination centred around gap voltages 80,100,120 and 140V are illustrated in Figure 4.11.

21 W Cu CuW AgW Parametric combination (Gap Voltage V) Figure 4.11 Parametric combinations of circularity error with gap voltage It is also seen that, with lower and higher gap voltage, relatively high order deviation occurs, while error constrained machining can be achieved with medium range of gap voltage ( V). With relatively poor machining with lower voltage and arc prone erosion machining with high voltage, only marginal influence of gap voltage on circularity error can be seen with W and AgW electrodes. It is seen that Cu electrode exhibits reducing (drops) circularity up to 100V combination, above which a rise in circularity error occurs. Typically observed variations of circularity error with effective discharge energy are shown in Figure It is seen that with low level of discharge energy (level I) the circularity tends to rise and drop. With level II of discharge energy, circularity error tends to drop down with increasing energy. With III and IV level of energy it tends to rise and drop down with increasing energy. The illustrations of overcut and circularity error indicate widely different trends except for energy level II.

22 W Cu CuW AgW Effective discharge energy in µj Figure 4.12 Effective discharge energy of circularity error with different electrodes With energy level II both circularity error and overcut tend to drop down with increasing energy. This again confirms that good micro-edm process can be attained with relating smaller energy level. As in the case of overcut, both Cu and tungsten-based electrodes indicate mostly similar trends of variation. Moreover, it is also found that with the increase of discharge energy, the circularity of the micro-holes increases. This is due to the occurrence of secondary sparking more frequently during machining of deep holes. Furthermore, the experimental results are investigated in detail as shown in Figure It is seen that machining with W electrode results in almost the least order circularity error. And with higher voltage combination, a rise in circularity error can be seen. In the lower gap voltage of about 80V, a rise in circularity error is observed with Cu and CuW, and then it drops down.

23 88 Around 100V, only marginal variation can be identified with all the electrodes. Increasing voltage for about 140V and above, Cu shows a drop in circularity error. This is due to its high thermal conductivity and low melting temperature. In the range of the gap voltage around 100V, CuW shows higher circularity error than other electrodes. This is due to occurrence of secondary discharge caused by poor flushing. The relatively higher order circularity error and reduced MRR at higher voltage are due to high tool wear and sporadic machining. Further to identify the performance of electrodes, analyses were made with the average range of circularity error using different gap voltages. The average of the circularity error is illustrated in Figure 4.13 and the range is highlighted using band width Parametric combination of gap voltage (V) W Cu CuW AgW Figure 4.13 Average and range of circularity error with different gap voltages Cu experiences least circularity error with 100V conditions. Among the tungsten-based electrodes, AgW exhibits increasing circularity error up to 120V, followed by a drop. CuW and W exhibit a reduced order of circularity

24 89 error up to 120V, followed by a rise. It is seen that mostly Cu and CuW exhibit wider variation in circularity error, which is attributed to higher order tool wear and MRR Observation on Surface Roughness The machined surface is influenced by various parameters such as discharge energy, gap voltage, capacitance, feed rate, threshold and the properties of the electrode materials. Accurate surface of the machined workpiece was identified using the topographical representation of Talysurf CCI 3000A. Figure 4.14 shows the three-dimensional (3D) image of the measured surface texture using Talysurf CCI 3000A. In this figure Rt values (peak-to-valley surface roughness) for different electrodes are shown. (a) W-electrode (b) Cu-electrode Figure 4.14 (Continued)

25 90 (c) CuW-electrode (d) AgW-electrode Figure D image of surface texture of the micro-hole with different electrodes The experimental data on surface roughness for different operating conditions are listed in Table 4.5. From Figure 4.15, it is observed that the roughness values are highly influenced by the gap voltage. With the increase of gap voltage the surface roughness gradually increases. This increase in surface roughness is due to the increase of discharge energy per pulse. AgW electrode provides lower surface roughness when compared to other electrodes due to its electrical and thermal properties. On the other hand, copper exhibits higher surface roughness because of its higher thermal conductivity which leads to the formation of larger craters.

26 91 Table 4.5 Experimental results for surface roughness Sl. No. Parameters Surface Roughness (R a ) in µm W Cu CuW AgW Effective discharge energy (µj) 1 80V, 0.1nF, 2µm, 20% V, 1nF, 4µm, 40% V, 10nF, 6µm, 60% V, 100nF, 8µm, 80% V, 0.1nF, 4µm, 60% V, 1nF, 2µm, 80% V, 10nF, 8µm, 20% V, 100nF, 6µm, 40% V, 0.1nF, 6µm, 80% V, 1nF, 8µm, 60% V, 10nF, 2µm, 40% V, 100nF, 4µm, 20% V, 0.1nF, 8µm, 40% V, 1nF, 6µm, 20% V, 10nF, 4µm, 80% V, 100nF, 2µm, 60% W Cu CuW AgW Parametric combination (Gap Voltage V) Figure 4.15 Parametric combinations on SR for different electrodes

27 92 In die-sinking Micro-EDM, the amount of energy released in every spark determines the surface roughness. The experimental results from Figure 4.16 demonstrate that high discharge energy requires large machining areas and produces greater roughness. When the discharge energy is increased, the crater depth increases and consequently, the surface roughness also increased. Thus, it is found that low discharge energy is optimal for achieving minimum surface roughness W Cu CuW AgW Effective discharge energy (µj) Figure 4.16 Effective discharge energy on SR with different electrodes Figure 4.17 represents the parametric combinations of surface roughness of discharge energy and threshold with constant feed rate of 2µm/s, 4µm/s, 6µm/s and 8µm/s, respectively.

28 W Cu CuW AgW W Cu CuW AgW Effective Discharge Energy (µj) Effective Discharge Energy (µj) (a) 2µm (b) 4µm/s (c) W Cu CuW AgW (d) W Cu CuW AgW Effective Discharge Energy in µj Effective Discharge Energy in µj (c) 6µm/s (d) 8µm/s Figure 4.17 Effective discharge energy on SR with different feed rate The result shows that the surface roughness increases with the increase of discharge energy whereas feed rate exerts minimum influence. This may be due to the dominant influence of discharge energy. From Figure 4.2, it is seen that barring the case of copper, other tungsten-based electrodes exhibit mostly a reduction in MRR with higher energy level. However, it is seen that all the electrodes exhibit surface roughness with increasing discharge energy. The reduction in MRR with higher discharge energy is attributed to presence of machined debris in the

29 94 inter electrode gap which can cause electrode shorting and surface roughness. Also with higher discharge energy the workpiece (die steel) experiences higher order heating, formation of HAZ and associated surface texture. The amount of the heating of electrode depends on the discharge energy and electrode. In order to evaluate the HAZ characteristics, the zone of the material around the periphery of the spark eroded hole is observed through scanning electron microscope Observation on HAZ Normally, all thermal dominant processes of material are associated with HAZ, depending on the heat flux and the thermal characteristics of the work material. However, in the case of spark erosion, process specific HAZ is found. Further, apart from the heated substrate layer, a recast/ resolidified layer is formed over the surface due to rapid cooling of the molten material. In the micro-edm process, the HAZ are found with the three layers on the surface of the machined component. The top surface contains a thin layer of spattered material which can easily be removed by flushing. Underneath the spattered material a thin layer called re-cast layer is formed due to the rapid cooling effect of dielectric and gets adhered to the machined surface. Recast layer is a hard, brittle and porous and may contain even micro cracks. The next layer is the heat affected substrate zone which is affected due to the amount of heat conducted within the material. As in the case of welding, when adjoining the molten / solidified deposition, formation of HAZ occurs. In EDM, the heat of machining and subsequent depth of HAZ (based on thermal diffusibility) facilitates formation of HAZ. HAZ was studied by analyzing the surface region of the workpiece. The micrograph of recast/resolidified layer for the four electrodes is shown in Figure It is seen that the variations of recast/ resolidified layer is largely

30 95 influenced by the discharge energy and tool electrode. It is seen that among the electrodes, erosion with tungsten electrode results in fairly thick and uniform recast layer. The recast layer is dense, with decreased solidified spherical globules. With copper electrode, non-uniform recast layer is formed. Also the layer is discontinuous with a rougher outer surface. The relatively lower heating of the work surface (in the case of copper electrode) contributes to this. In the case of copper tungsten electrode, the work material experiences more heating results in recast layer and non-uniform continuous resolidified/recast spherical globules. Erosion of die steel with AgW electrode results in thicker, non-uniform recast layer with coarse solidified globules. (a1) (a) (a1) (a) W- electrode (b1) (b) (b1) (b) Cu-electrode Figure 4.18 (Continued)

31 96 (c) (c1) (c1) (c) CuW-electrode (d1) (d) (d1) (d) AgW-electrode Figure 4.18 SEM micrographs of HAZ for different electrodes using 100V of gap voltage, 0.1nF of capacitance, 4µm/s of feed rate and 60% of threshold The observation of recast layer supplements the observation of relating higher order MRR with Cu electrode followed by AgW, CuW, and W electrodes Observation on Micro-Hardness Apart from formation of recast/resolidified layer, the substrate beneath the recast layer experiences rapid heating and cooling, experiencing quench hardening. The amount of hardening depends on the heating of the material which is electrode specific.the EDMed surface has a relatively high

32 97 micro-hardness, which is due to the emigration of carbon from the oil dielectrics to the workpiece surface forming iron carbides in the white layer (Kruth et al. 1995). Typical monitored variations of micro hardness in this substrate region of spark machined die steel are shown in Figure It is seen that the material undergoes more hardening with tungsten compared to copper attributable to the difference in the substrate heating. For the different energy level, the variations of hardness were observed. It is seen that substrate undergoes relatively higher hardness with tungsten electrode, while the least is seen with copper Edge Distance from the hole surface in µm W Cu CuW AgW (a) 80V, 0.1nF, 2µm/s, 20%, Edge Distance from the hole surface in µm W Cu CuW AgW (b) 100V, 10nF, 8µm/s, 20% Figure 4.19 (Continued)

33 Edge Distance from the hole surface in µm W Cu CuW AgW (c) 140V, 100nF, 2µm/s, 60% Figure 4.19 Variation of micro-hardness on cross section of the machined surface using different electrodes with parameters Also all the tungsten-based electrodes exhibit mostly closer values, especially near the edge region. Further, with micro-edm, a HAZ of around 75µm thickness is seen. The observation on micro hardness also supports least order heating with copper consequent higher order MRR. Figure 4.20 shows the sample image of micro-hardness measured using micro Vickers hardness tester. Figure 4.20 Image of micro hardness measured using Vickers hardness tester

34 99 W and AgW exhibit highest micro-hardness whereas CuW and Cu show less hardness. From the EDX analysis, it is identified that the significant amount of carbon, migrated to the workpiece due to decomposition of dielectric, determines the hardness of the machined surface Observation on Micro-Cracks and Voids In die-sinking micro-edm, a recast layer is formed on the machined surface that contains craters and micro-cracks, which cause poor surface quality and reduce size accuracy. Micro-cracks are the result of the thermal stresses (quenching). This is due to the drastic heating and cooling rate and the non-uniform temperature distribution. Moreover, the surface immediately reaches the solidification temperature being cooled by the surrounding working fluid. The micro-voids can thus be attributed to the gas bubbles expelled from the molten material during solidification. In addition, the morphology of the EDM surface is dependent on the applied discharge energy. When the discharge energy increases, the machined surface exhibits a deeper crack or void and more pronounced defects. Micro-cracks and micro-voids formed on the surface exhibit poor surface quality as it affects the diameter size of the micro-hole and undermining the precision of the desired geometric shape. They also produce notch effects leading to stress concentration and reduction of fatigue strength. Figure 4.21 shows cracks and voids of micro EDMed specimens with severe propagation on the discharge surface. The observation reveals splat boundary. In addition, pin holes and several cracks are also evident in most of the cases.

35 100 (a) W-Electrode (b) Cu-Electrode (c) CuW-Electrode (d) AgW-Electrode Figure 4.21 SEM micrographs of micro-cracks and voids of discharge surface using 100V of gap voltage, 0.1nF of capacitance, 4µm/s of feed rate and 60% of threshold Comparing the four figures, it can be seen that for Cu and CuW more debris is spattered on the machined surface. It is observed that W and AgW specimen exhibits minimum debris and a few pin holes (micro-pore) and cracks can be identified. The overall performance analysis based on the effect of input parameters shows that gap voltage and capacitance are found to exert more influence than feed rate and threshold. Tungsten-based electrodes prove to be better as it exhibits good surface finish with lower TWR but copper exhibits the highest MRR.

36 Energy Dispersive X-ray Analysis of Machined Surfaces of the Hole Apart from the formation of recast/ resolidified layer, the machined surface beneath the recast/resolidified layer exhibits an intrinsic response in terms of hardness and chemistry. EDAX profile on the machined/ heated substrate material around the hole periphery is illustrated in Figure 4.22 and It can be seen that EDAX of surfaces machined with W and CuW shows dominant carbon particles. This can be attributed to relatively higher order heating of the work material, which could have caused localized/surface decomposition of the metal carbides (in die steel) associated with dominant free carbon. With copper and silver tungsten this effect is relatively absent. This supplements the observation of higher MRR with Cu and AgW electrodes. For all the four electrodes, the roughness increases with the increase of discharge energy. However, AgW exhibits better performance with regard to surface finish followed by W while CuW and Cu show the poorest. Even though copper has high electrical conductivity, the surface roughness increases due to its higher MRR and TWR. Moreover, it is also observed that the highly accelerated electron produced while using copper electrode creates larger size crater on the workpiece. This results in higher surface roughness.

37 102 Element Wt% At% CK OK FeK WL Matrix Correction ZAF (a) W-electrode Element Wt% At% CrK FeK CuK Matrix Correction ZAF (b) Cu-electrode Figure 4.21 (Continued)

38 103 Element Wt% At% CK OK MgK SiK CdL CrK FeK CuK WL Matrix Correction ZAF (c) CuW-electrode Element Wt% At% CK AgL FeK WL Matrix Correction ZAF (d) AgW-electrode Figure 4.22 EDAX analyses showing the percentage of migrated materials on the workpiece surface after machining at 120 V, 1nF, 8µm/s and 60% with different electrodes The elements present in the surface are clearly indicated by the peaks corresponding to their energy level. It is evident from Figure 4.22 and 4.23 an appreciable amount of tungsten migrates when compared with copper.

39 104 It is also noticed that the significant changes in the chemical composition of machined surface due to migration of material from the tool electrode. (a) W-electrode (b) Cu-electrode Figure 4.22 (Continued)

40 105 (c) CuW-electrode (d) AgW-electrode Figure 4.23 EDAX analyses showing the percentage of migrated materials on the workpiece surface after machining at 100 V, 0.1nF, 4µm/s and 60% with different electrodes

41 OPTIMIZATION Optimization technique was adopted to assess the important machining parameters for performance measures such as MRR, TWR and SR in the die-sinking micro-edm of EN24 die steel. In the present analysis, the larger the better characteristic was applied in the case of S/N ratio for MRR and the smaller the better characteristic for TWR and SR, since both tool wear as well as surface roughness had to be minimized. The analysis was made based on the effect of different electrodes used Taguchi-based Grey Relational Analysis using W Electrode Based on Taguchi method and GRA, S/N ratio was calculated and normalized in the range of 0 to 1. Based on this data, grey relational coefficients are calculated to represent the correlation between the ideal (best) and the actual normalized experimental data. Overall grey relational grade is then determined by averaging the grey relational coefficients corresponding to selected responses. The overall quality characteristics of the multi-response process depend on the calculated GRG. By following these steps the optimal parameters identified for W electrode is discussed in this section Calculation and analysis of GRG for W The optimization of the observed values was determined through comparison with the Taguchi S/N ratio. In the Taguchi method, the higher the level, the better the overall performance, meaning that the factor levels with the highest value should always be selected. The highest values for higher MRR, smaller TWR and smaller SR are given in Table 4.6.

42 107 Table 4.6 L 16 S/N ratio for tungsten (W) electrode Exp. No Parameters Experimental Values S/N Ratio A B C D MRR (mm 3 /min) TWR (%) SR (µm) MRR TWR SR In the grey relational analysis, a normalization of the S/N ratio is performed to prepare raw data for the analysis in which the original sequence is transferred to a comparable sequence. The values are highlighted in Table 4.7.

43 108 Table 4.7 Experimental results by GRA for W Exp. No. Normalized S/N ratio Derivation sequences oi MRR TWR SR MRR TWR SR Grey relational co-efficient for all the sequences expresses the relationship between the ideal (best) and actual normalized S/N ratio. The grey relational co-efficient was thus calculated based on Equation 3.1. The value is considered 1 throughout the analysis. The overall evaluation of the multiple performance characteristics is based on the grey relational grade. The grey relational grade is the average sum of the grey relational coefficients and

44 109 it was calculated using Equation 3.2. The grey relational co-efficient and the calculated grey relational grade are given in Table 4.8. Table 4.8 Grey relational coefficients and grey relational grade for W Exp. No. Grey relational co-efficient GC ij MRR TWR SR Grey relational Grade i The grey relational grade calculated for each sequence was taken as a response for the further analysis. Accordingly, the average for each experimental level was calculated using the highest value at the level for each parameter and it is given in Table 4.9.

45 110 Table 4.9 Average GRG for W Symbol Grey relational grade Rank Parameters Level1 Level2 Level3 Level4 (Max-Min) A Gap Voltage B Capacitance C Feed rate D Threshold Total mean value of GRG ( m ) = The grey relational grade graph for the overall grey relational grade for tungsten is shown in Figure The steep slope of grey relational grade graph indicates the more influencing machining parameters in the performance characteristics. It can be seen that the gap voltage of V influences significantly and 1-10nF capacitance, 4-6µm/s feed rate and 20% threshold are significant parameters. It is also seen that capacitance is found to be the most influencing parameters for tungsten electrode. 0.9 Main Effects Plot for Means 0.8 MEAN A B C D Figure 4.24 Main effects of the factors on the GRG for W

46 ANOVA for W electrode ANOVA was used to investigate the design parameters which significantly affected the quality characteristics. Therefore, ANOVA was done by analyzing the influence of gap voltage, capacitance, feed rate and threshold. The result of ANOVA for MRR, TWR and SR is calculated using the values of grey relational grades of Table 4.9. According to Table 4.10, the factor B, the capacitance with 82.89% of contribution, is the most significant controlled parameter for the die-sinking micro-edm. The gap voltage is with 10.89% contribution, the feed rate with 3.33% and the threshold with 2.89% of contribution if the maximum MRR and minimum TWR and SR are simultaneously and very effectively considered. Table 4.10 ANOVA of GRA for W Source of Variance Sum of Square DOF Mean Square/ Variance Contribution (%) Gap Voltage (V) Capacitance (nf) Feed rate (µm/s) Threshold (%) Error Total Confirmation Test Confirmation test was carried out to predict and verify the enhancement of the quality characteristics using the optimal parametric combination. The estimated GRG using optimal level of machining

47 112 parameters is calculated using Equation 3.3. The initial and predicted MRR, TWR and SR for optimal machining parameters are obtained from Tables 4.8 and 4.9, respectively. Table 4.11 shows the comparison of the experimental results using the orthogonal array (A1B1C1D1) and grey relational analysis (A2B1C2D1) of die-sinking micro-edm on EN24 using tungsten electrode. The response values obtained from the confirmation experiment are MRR = mm 3 /min, TWR = % and SR = 0.108µm. The MRR shows an increased value of mm 3 /min to mm 3 /min, the TWR shows a reduced value of % to % and the SR shows a reduced value of 0.11µm to 0.108µm, respectively. The corresponding improvements of MRR, TWR and SR are 30.38%, 1.41% and 1.81%, respectively. Table 4.11 Micro-EDM results of L 16 using the initial and optimal process factors for W Initial condition Optimal factors Prediction Experiment Level A1B1C1D1 A2B1C2D1 A2B1C2D1 MRR TWR SR GRG The optimal conditions are A2B1C2D1. Thus, it can be seen that combination of 100V gap voltage, 0.1nF capacitance, 6µm/s feed rate and 20% threshold facilitates mostly smaller order of discharge energy for tungsten electrodes.

48 Taguchi-based GRA using Cu Electrodes By following the steps involved in optimization techniques of Taguchi-based grey relational analysis, the optimal parameters identified for Cu electrode is discussed in this section Calculation and analysis of GRG for Cu Based on the Taguchi method, the calculated S/N ratio for copper electrode is given in Table Table 4.12 L 16 S/N Ratio for copper (Cu) electrode Exp. No Parameters Experimental Values S/N Ratio MRR TWR SR A B C D (mm 3 MRR TWR SR /min) (%) (µm)

49 114 Table 4.13 Experimental results of GRA for Cu Exp. No. Normalized S/N ratio Derivation Sequences oi MRR TWR SR MRR TWR SR In the grey relational analysis, a normalization of the S/N ratio is performed to prepare raw data for the analysis in which original sequence is transferred to a comparable sequence. The values are highlighted in Table 4.13.

50 115 Table 4.14 Grey relational coefficients and GRG for Cu Exp. No. Grey relational co-efficient GC ij MRR TWR SR Grey relational Grade i The grey relational co-efficient and the calculated GRG are given in Table The average for each experimental level was calculated using the highest value at the level for each parameter and it is given in Table Table 4.15 Average GRG for Cu Symbol Grey relational grade Parameters Level1 Level2 Level3 Level4 Rank (Max-Min) A Gap Voltage B Capacitance C Feed rate D Threshold Total mean value of GRG ( m ) =

51 116 Comparison of Table 4.9 and 4.15 (for W and Cu) shows a distinct change in ranking and mostly gap voltage and capacitance play a significant role. The overall GRG for copper is shown in Figure The steep slope of GRG graph indicates that capacitance is found to be the most influencing parameters followed by gap voltage. It can be seen that the gap voltage of V influences significantly and 1-10nF capacitance, 2-4µm/s feed rate and 60-80% threshold are significant parameters Main Effects Plot for Means 0.75 MEAN A B C D Figure 4.25 Main effects of the factors on the GRG for Cu ANOVA for Cu electrode ANOVA was used to investigate the design parameters which significantly affected the quality characteristics. Therefore, ANOVA was done by analyzing the influence of gap voltage, capacitance, feed rate and threshold. The results of ANOVA for MRR, TWR and SR were calculated using the values of grey relational grades of Table According to Table 4.16 the factor B, the gap voltage with 45.46% of contribution, is the most

52 117 significant controlled parameter for the die-sinking micro-edm. The capacitance is with 38.13% contribution, the feed rate with 10.10% and the threshold with 6.31% of contribution, if the maximum MRR and minimum TWR and SR are simultaneously and very effectively considered. Table 4.16 ANOVA of GRA for Cu Source of Variance Sum of Square DOF Mean Square/ Variance Contribution (%) Gap Voltage (V) Capacitance (nf) Feed rate (µm/s) Threshold (%) Error Total Confirmation Test The initial and predicted MRR, TWR and SR for optimal machining parameters are obtained from Table 4.14 and 4.15, respectively. Table 4.17 shows the comparison of the experimental results using the orthogonal array (A2B1C2D3) and grey relational analysis (A2B1C1D2) of die-sinking micro- EDM on EN24 using copper electrode. The response value obtained from the confirmation experiment are MRR = mm 3 /min, TWR = % and SR = 0.149µm. The MRR shows an increased value of mm 3 /min to mm 3 /min, the TWR shows a reduced value of % to % and the SR shows a reduced value of 0.152µm to 0.149µm, respectively. The corresponding improvements of MRR, TWR and SR are 2.49%, 3.28% and 1.97%, respectively.

53 118 Table 4.17 Micro-EDM results of L 16 using the initial and optimal process factors for Cu Initial condition Prediction Optimal factors Experiment Level A2B1C2D3 A2B1C1D2 A2B1C1D2 MRR TWR SR GRG It is also observed that relatively smaller feed rate for copper is appreciable to avoid possible shorting of electrodes due to the presence of debris in the electrode gap. Also higher threshold facilitates enhanced discharge energy permissible with copper electrodes Taguchi-based GRA using CuW Electrodes By following the steps involved in optimization techniques of Taguchi-based grey relational analysis, the optimal parameters identified for CuW electrode is discussed in this section Calculation and analysis of GRG for CuW Based on the Taguchi method, the calculated S/N ratio for CuW electrode is given in Table 4.18.

54 119 Table 4.18 L 16 S/N ratio for copper tungsten (CuW) electrode Exp. No Parameters Experimental Values S/N Ratio MRR TWR SR A B C D (mm 3 MRR TWR SR /min) (%) (µm) E In the grey relational analysis, a normalization of the S/N ratio is performed to prepare raw data for the analysis in which the original sequence is transferred to a comparable sequence. The values are highlighted in Table 4.19.

55 120 Table 4.19 Experimental results by GRA for CuW Exp. No. Normalized S/N ratio Derivation sequences oi MRR TWR SR MRR TWR SR Table The grey relational co-efficient and the calculated GRG are given in

56 121 Table 4.20 Grey relational coefficients and GRG for CuW Exp. No. Grey relational co-efficient GCij MRR TWR SR Grey relational Grade i The average for each experimental level was calculated using the highest value at the level for each parameter and it is given in Table 4.21.

57 122 Table 4.21 Average GRG for CuW Symbol Grey relational grade Parameters Level1 Level2 Level3 Level4 Rank (Max-Min) A Gap Voltage B Capacitance C Feed rate D Threshold Total mean value of GRG ( m ) = Copper tungsten needs lower order energy in terms of low gap voltage, capacitance, feed rate, and increased pulse off (higher threshold). The overall grey relational grade for copper tungsten is shown in Figure The steep slope of grey relational grade graph indicates that capacitance is found to be the most influencing parameters, followed by gap voltage. It can be seen that the gap voltage of V influences significantly and 1-10nF capacitance, 6-8µm/s feed rate and 60-80% threshold are significant parameters. 0.9 Main Effects Plot for Means 0.8 MEAN A B C D Figure 4.26 Main effects of the factors on the GRG for CuW

58 ANOVA for CuW electrode ANOVA was used to investigate the design parameters which significantly affected the quality characteristics. Therefore, ANOVA was done by analyzing the influence of gap voltage, capacitance, feed rate and threshold. The results of ANOVA for MRR, TWR and SR were calculated using the values of grey relational grades of Table According to Table 4.21, the factor B, the capacitance with 42.45% of contribution, is the most significant controlled parameter for the die-sinking micro-edm. The gap voltage is with 37.30% contribution, the threshold with 18.94% and the feed rate with 1.58% of contribution, if the maximum MRR and minimum TWR and SR are simultaneously and very effectively considered. Table 4.21 ANOVA of GRA for CuW Source of Variance Sum of Square DOF Mean Square/ Variance Contribution (%) Gap Voltage (V) Capacitance (nf) Feed rate (µm/s) Threshold (%) Error Total Confirmation Test The initial and predicted MRR, TWR and SR for optimal machining parameters are obtained from Tables 4.20 and 4.21, respectively. Table 4.23 shows the comparison of the experimental results using the orthogonal array (A1B2C2D2) and grey relational analysis (A1B1C1D3) of die-sinking micro- EDM on EN24 using copper tungsten electrode. The response value

59 124 obtained from the confirmation experiment are MRR = mm 3 /min, TWR = % and SR = 0.129µm. The MRR shows an increased value of mm 3 /min to mm 3 /min, the TWR shows a reduced value of % to % and the SR shows a reduced value of 0.132µm to 0.129µm. The corresponding improvement of MRR, TWR and SR are 4.31%, 0.55% and 2.27%, respectively. Table 4.23 Micro-EDM results of L 16 using the initial and optimal process factors for CuW Initial condition Optimal factors Prediction Experiment Level A1B2C2D2 A1B1C1D3 A1B1C1D3 MRR TWR SR Grey Relational Grade Taguchi-based GRA using AgW Electrodes By following the steps involved in optimization techniques of Taguchi-based grey relational analysis, the optimal parameters identified for AgW electrode is discussed in this section Calculation and analysis of GRG for AgW Based on the Taguchi method, the calculated S/N ratio for silver tungsten electrode is given in Table 4.24.

60 125 Table 4.24 L 16 - S/N ratio for silver tungsten (AgW) electrode Exp. No Parameters Experimental Values S/N Ratio MRR TWR SR A B C D (mm 3 MRR TWR SR /min) (%) (µm) E E E In the grey relational analysis, a normalization of the S/N ratio is performed to prepare raw data for the analysis in which the original sequence is transferred to a comparable sequence. The values are highlighted in Table 4.25.

61 126 Table 4.25 Experimental results by GRA for AgW Exp. No Normalized S/N ratio Derivation sequences oi MRR TWR SR MRR TWR SR Table The grey relational co-efficient and the calculated GRG are given in

62 127 Table 4.26 Grey relational coefficients and GRG for AgW Exp. No. Grey relational co-efficient GC ij MRR TWR SR Grey relational Grade i The average for each experimental level was calculated using the highest value at the level for each parameter and it is given in Table 4.27.

63 128 Table 4.27 Average GRG for AgW Symbol Grey relational grade Rank Parameters Level1 Level2 Level3 Level4 (Max-Min) A Gap Voltage B Capacitance C Feed rate D Threshold Total mean value of GRG ( m ) = Main Effects Plot for Means MEAN A B C D Figure 4.27 Main effects of the factors on the GRG for AgW The overall grey relational grade for copper is shown in Figure The steep slope of grey relational grade graph indicates that capacitance is found to be the most influencing parameter, followed by threshold and gap voltage ANOVA for AgW electrode ANOVA was used to investigate the design parameters which significantly affected the quality characteristics. Therefore, ANOVA was done by analyzing the influence of gap voltage, capacitance, feed rate and

64 129 threshold. The results of ANOVA for MRR, TWR and SR were calculated using the values of the grey relational coefficients and grey relational grades of Table According to Table 4.28, the factor B, the capacitance with 67.4% of contribution, is the most significant controlled parameter for the diesinking micro-edm. The threshold is with 12.44% contribution, the feed rate with 10.36% and the gap voltage with 9.8% of contribution, if the maximum MRR and minimum TWR and SR are simultaneously and very effectively considered. Table 4.28 ANOVA of GRA for AgW Source of Variance Sum of Square DOF Mean Square/ Variance Contribution (%) Gap Voltage (V) Capacitance (nf) Feed rate (µm/s) Threshold (%) Error Total Confirmation Test The initial and predicted MRR, TWR and SR for optimal machining parameters are obtained from Tables 4.26 and 4.27, respectively. Table 4.29 shows the comparison of the experimental results using the orthogonal array (A2B1C2D3) and grey relational analysis (A1B1C2D1) of die-sinking micro- EDM on EN24 using silver tungsten electrode. The response values obtained from the confirmation experiment are MRR = mm 3 /min, TWR = % and SR = µm. The MRR shows an increased value of mm 3 /min to mm 3 /min, the TWR shows a reduced value of % to % and the SR shows a reduced value of 0.094µm to

65 µm, respectively. The corresponding improvement of MRR, TWR and SR are 3.86%, 1.36% and 2.66%, respectively. Table 4.29 Micro-EDM results of L 16 using the initial and optimal process factors for AgW Initial condition Prediction Optimal factors Experiment Level A2B1C2D3 A1B1C2D1 A1B1C2D1 MRR TWR SR GRG AgW needs lower order energy in terms of low gap voltage, capacitance, feed rate and lower threshold Observation on Optimization Results The overall performance of various electrodes was analyzed using Taguchi-based GRA. From the observation, it can be concluded that gap voltage and capacitance are the main influencing parameters rather than feed rate and threshold. Moreover, the significant machining parameter values for whole machining performance are gap voltage in the range of V, capacitance 0.1nF, feed rate 2-6µm and threshold 20% -60%. Furthermore, it is also inferred that copper-based electrodes show better performance in the higher range of threshold (60%) whereas tungsten-based electrodes show better performance in the lower range of threshold. Hence, it is identified that the GRA based on the Taguchi method for the optimization of multiperformance characteristic is a very useful tool for predicting MRR, TWR and SR of EN24 die steel using different electrodes die-sinking micro-edm.

66 MATHEMATICAL MODELING The analysis and modeling were done for all the used electrodes such as tungsten, copper, copper tungsten and silver tungsten separately with reference to most significant parameters like gap voltage, capacitance, feed rate and threshold Data Analysis and Modeling - MRR From the data collected, a statistical model for material removal rate was developed. The analytical model for MRR was developed by modeling the relationship between MRR and machining conditions based on the relationship shown in Equation 4.1. The non-linear relationship is owing to the MRR which is normally non-linear and stochastic in nature MRR = A 1 V x 1 C y 1 F z 1 T w 1 (4.1) where MRR is the material removal rate in mm 3 /min V, C, F and T are the gap voltage (V), capacitance (nf), feed rate (µm/s) and threshold (%), respectively, and x 1, y 1, z 1 and w 1 are exponential constants and A 1 is constant of proportionality. The non-linear regression model was developed using SPSS software with the results of experiments Observation on tungsten electrode The model thus developed with regression coefficient of is represented in Equation 4.2 and the predicted estimates are indicated in Table 4.30 with the standard error and confidence interval. The percentage deviation of the experimental MRR from that of predicted for tungsten electrode is represented in Table 4.31.

67 132 Table 4.30 Parameter estimates for tungsten 95% Confidence Interval Parameter Estimate Std. Error Lower Bound Upper Bound A 6.564E E E-6 X Y Z W Table 4.31 Experimental and predicted MRR for tungsten Exp no Gap Voltage (V) Capacitance (nf) Feed rate (µm/s) Threshold (%) Experiment MRR (mm3/min) Predicted MRR (mm3/min) % Deviation E Material Removal Rate = 6.564E-7 Gap Voltage Capacitance Feed rate Threshold (4.2) Normally, in die-sinking micro-edm, machining performance is largely influenced by gap voltage, i.e., due to increase in the gap between the

68 133 electrodes a clean / debris free dielectric, restricted voltage to avoid arcing. Apart from gap condition, specific characteristics of electrode such as thermal/electrical conductivity / thermal potential also contribute to erosion. Further, lower order capacitance and threshold facilitate good machining. Accordingly, it is seen in die-sinking micro-edm, the machining parameters exert a direct influence on MRR. Among the parameters, it is seen that MRR is directly influenced by gap voltage and feed rate. With higher gap voltage, it is possible to get maximum MRR and thereby facilitating better / sustained machining. The plot of the predicted and experimental results is shown in Figure Figure 4.28 Experimental and predicted results of MRR for W It is seen that for tungsten electrode material, it is preferable to use higher gap voltage and feed rate, while other parameters such as capacitance and threshold play relatively lower order significant role.

69 Observation on copper electrode copper electrode is The non-linear relationship due to the material removal rate of MRR = A 2 V x 2 C y 2 F z 2 T w 2 where MRR is the material removal rate in mm 3 /min V, C, F and T are the gap voltage (V), capacitance (nf), feed rate (µm/s) and threshold (%), respectively, and x 2, y 2, z 2 and w 2 are exponential constants and A 2 is constant of proportionality. Table 4.32 Parameter estimates for copper Parameter Estimate Std. Error 95% Confidence Interval Lower Bound Upper Bound A 1.00E X Y Z W The model thus developed with regression coefficient of is represented in Equation 4.3 and the predicted estimates are indicated in Table 4.32 with the standard error and confidence interval. The percentage deviation of the experimental MRR from that of predicted values for copper electrode is represented in Table 4.33.

70 135 Table 4.33 Experimental and predicted MRR for copper Exp. Gap Voltage No (V) Capacitance (nf) Feed rate (µm/s) Threshold (%) Experiment MRR (mm3/min) Predicted MRR (mm3/min) % Deviation Material Removal Rate = 1.00E-5 Gap Voltage Capacitance Feed rate Threshold (4.3) Unlike the case of tungsten, with copper electrodes, higher gap voltage, and capacitance (higher discharge energy) facilitate enhanced MRR. It is also identified that it is preferable to use higher gap voltage and capacitance while other parameters such as feed rate and threshold play relatively lower order significant role. The plot of the predicted and experimental results is shown in Figure 4.29.

71 136 Figure 4.29 Experimental and predicted results of MRR for Cu Observation on copper tungsten electrode copper electrode is The non-linear relationship due to the material removal rate of MRR = A 3 V x 3 C y 3 F z 3 T w 3 where MRR is the material removal rate in mm 3 /min V, C, F and T are the gap voltage (V), capacitance (nf), feed rate (µm/s) and threshold (%), respectively, and x 3, y 3, z 3 and w 3 are exponential constants and A 3 is constant of proportionality. The model thus developed with regression coefficient of is represented in Equation 4.4 and the predicted estimates are indicated in Table 4.30 with the standard error and confidence interval. The percentage deviation of the experimental MRR from that of predicted values for copper tungsten is represented in Table 4.35.

72 137 Table 4.34 Parameter estimates for copper tungsten Parameter Estimate Std. Error 95% Confidence Interval Lower Bound Upper Bound A X Y Z W Table 4.35 Experimental and predicted MRR for copper tungsten Exp. Experiment Predicted % No Gap Voltage Capacitance Feed rate Threshold MRR MRR Deviation (V) (nf) (µm/s) (%) (mm3/min) (mm3/min) E Material Removal Rate = Gap Voltage Capacitance Feed rate Threshold (4.4)

73 138 For copper tungsten electrode, smaller gap voltage and capacitance and higher feed rate and threshold shows better MRR. It is also identified that, it is preferable to use higher feed rate and threshold while other parameters such as gap voltage and capacitance play relatively lower order significant role. Figure The plot of the predicted and experimental results is shown in Figure 4.30 Experimental and predicted results of MRR for CuW Observation on silver tungsten electrode copper electrode is The non-linear relationship due to the material removal rate of MRR = A 4 V x 4 C y 4 F z 4 T w 4 where MRR is the material removal rate in mm 3 /min V, C, F and T are the gap voltage (V), capacitance (nf), feed rate (µm/s) and threshold (%), respectively, and x 4, y 4, z 4 and w 4 are exponential constants and A 4 is constant of proportionality.

74 139 The model thus developed with regression coefficient of is represented in Equation 4.5 and the predicted estimates are indicated in Table 4.36 with the standard error and confidence interval. The percentage deviation of the experimental MRR from that of predicted values for silver tungsten is represented in Table Table 4.36 Parameter estimates for silver tungsten Parameter Estimate Std. Error 95% Confidence Interval Lower Bound Upper Bound A 2.520E E E-5 X Y Z W Table 4.37 Experimental and predicted MRR for silver tungsten Exp. No Gap Voltage (V) Capacitance (nf) Feed rate (µm/s) Threshold (%) Experiment MRR (mm3/min) Predicted MRR (mm3/min) % Deviation E E E

75 140 Material Removal Rate = 2.520E-6 Gap Voltage Capacitance Feed rate Threshold (4.5) It is seen that both tungsten and silver tungsten require relating higher gap voltage for sustained MRR while copper tungsten requires lower gap voltage. It is also identified that it is preferable to use higher gap voltage and capacitance while other parameters such as capacitance and threshold play relatively lower order significant role. The plot of the predicted and experimental results is shown in Figure Figure 4.31 Experimental and predicted results of MRR for AgW Data Analysis and Modeling TWR The analytical model for tool wear ratio was developed by modeling the relationship between TWR and machining conditions based on the relationship shown in Equation 4.6. The non-linear relationship due to TWR is TWR = A 1 V x 1 C y 1 F z 1 T w 1 (4.6)