Keywords

1 Introduction

Al–SiCp metal matrix composites (MMCs) have gained plentiful application in various industries, for instance, in automotive and aerospace sectors, owing to their one of a kind blend of mechanical properties, wear resistance, and retention of strength at elevated temperatures [1,2,3,4,5]. However, it has been witnessed that their full potential use is not escalating and hindered due to lack of machinability, with most of the conventional machining processes, often results into high tool wear, poor surface roughness, and high machining cost [6]. In this context, numerous researchers studied and investigated the suitability of advanced and nonconventional machining processes (such as electric discharge machining, abrasive jet machining, electron beam machining, and laser beam machining) to overcome the aforesaid bottlenecks [7, 8]. Among these, electric discharge machining (EDM) has been found as one of the best-suited machining technologies for machining Al–SiCp MMC [9]. EDM is a thermomechanical process widely used to machine the hard and tough material with the ease [10]. The heat energy librated from the electrical sparks melts the workpiece surface and removed the material in the form of micron-size debris [11]. Prakash et al. reported the mechanism of material removal during EDM process in details, as illustrated in Fig. 1 [12]. In EDM, countless electrical sparks have been generated within fraction of seconds, which induced heat energy causing removal of material from workpiece surface [13]. In the presence of dielectric fluid, this librated heat energy thermally affected the top surface layer of workpiece material due to sudden quenching. Thus, various surface deformities like high surface roughness, high surface cracks, and micron size pits/dimples were shaped on the machined surface which breaks down the surface quality [14]. Keeping in mind the end objective to reduce the generation of surface flaws like micro-cracks, high ridges of re-disposition of molten pool, and high roughness, numerous modifications and progressions in EDM process has been carried out by numerous analysts [15, 16]. In the as-adopted modifications and progressions, the machining mechanisms of one or more process have been superimposed to take advantage of one process over other and called as hybridization. These hybridizations are such as electro-discharge diamond grinding (EDDG), ultrasonic vibration-assisted EDM (UVA-EDM), rotary-assisted EDM (RAEDM), HyFlex EDM, electro-discharge coating/surface modification by composite or green tool electrode, near dry EDM and powder mixed-EDM (PM-EDM) [17,18,19,20,21,22,23,24,25]. In comparison to all, the PM-EDM has been accepted and used as the most encouraging hybridized process to deal with reduction of surface defects and to enhance the surface quality [26,27,28,29,30]. Prakash et al. [31] investigate the capability of PM-EDM not only to enhance the surface quality but also to improve the machining performance. It has been reported that with the dispersion of powder particles in the dielectric fluid, the uniformity and sparking area increased but the increased discharge gap which reduced the thermal energy resulted from the electrical sparks. As a result of this, small and tiny discharge craters were developed on the machined surface and lead to decrease in the SR value of the machined surface. On the other hand, with the dispersion of powder particles, the sparking area increased. As a consequence the top layer of workpiece is expelled in the form of micron debris from large locations; thus increased the MRR value.

Fig. 1
figure 1

Schematic illustrated the mechanism of material removal in EDM process [12]

Full size image

Singh et al. studied the capability of PM-EDM to enhance the SR, MRR, and TWR characteristics by using tungsten powder particles as dielectric solvent [32,33,34]. It has been reported that tungsten powder in dielectric fluid enhanced the machining gap and increased the machining contact area, as a result the material was removed in micron and sub-micron sized in large proportion with small pit size. Mohan et al. reported the application of PM-EDM for drilling the hole in Al–SiC MMCs and studied the effect of tool polarity along with other machining parameters on the machined hole surface quality [35]. Sidhu et al. investigated the efficacy of PM-EDM process to enhance the surface characteristics, machining performance, and surface properties of Al–SiCp MMC and it has been reported that microhardness of the substrate surface was enhanced by the PM-EDM process using material migration method [36, 37]. Pecas and Henriques [38] reported the potential of PM-EDM for the finishing of workpiece material. Recently, Prakash et al. [39] explored the capability of PM-EDM as finishing process and studied the effect of surface finish achieved by PM-EDM process on the fatigue performance of Ti-based implant material. A number of researchers used PM-EDM process to enhance the machining performance, surface characteristics, and surface properties using micron-sized powder particles in the dielectric fluid [40,41,42,43,44,45]. The application of nonmetallic powder particles as a dielectric solvent has been used for the machining the work material with the aim of producing nanofinished part [46,47,48]. Miao et al. explored the capability of MWCNT mixedEDM process for the finishing of workpiece using miniature tool electrode. It was reported that with the use of MWCNT in EDM dielectric fluid, the machined surface quality and machining efficiency was improved by 70 and 66%, respectively [49]. Izman et al. [50] used the potential of MWCNT mixed-EDM for the reducing the recast layer thickness, improving the MRR and surface finish. It was reported that MRR and surface was improved by 7 and 9%, respectively. Prabhu and Vinayagam [51] reported that nanolevel surface finish ~75 nm was achieved on AISI-D2 steel substrate by MWCNT mixed-EDM process. Further, the utility of MWCNT-mixed-EDM for the finishing of Inconel 825 has been reported [52]. It was reported that surface finish of the components was enhanced by 34% than EDMed substrate. Sari et al. reported the nanofinishing of AISI H-13 tool by EDM using MWCNT mixed additive in dielectric fluid. The effect of MWCNT concentration on the re-deposition layer was studied and it was reported that the thickness of re-deposited layer was significantly. MWCNTs minimize the thickness of recast layer due to larger heat absorption [53]. When MWCNTs is added in to dielectric of EDM, produces 20% improvement in SR of AISI D2 tool steel [54]. Recently, Shabgard and Khosrozadeh [55] investigated the effect of MWCNT in dielectric fluid on SR, MRR, and TWR and it has been reported that the TWR and SR were significantly reduced by the use of MWCNT in dielectric fluid.

Optimization is very imperative to determine the best possible setting of input-process parameters to maximize the response characteristics. A number of single/conventional techniques such as Taguchi, response surface methodology, Grey relation have been adopted in the past to optimize the process parameters of EDM and PM-EDM process [56,57,58,59,60,61]. These techniques are only applicable for optimizing single parameter and are usually not favorable for multiparametric objectives, since outcomes get clashed. Non-dominated sorted genetic algorithm (NSGA)-II was found suitable for the multiobjective optimization of EDM and PM-EDM process parameters to maximize the machine efficiency and to obtain quality surface [62, 63]. Padhee et al. [64] implemented the application of NSGA-II to determine the optimal setting of PM-EDM process parameters to maximize MRR and to minimize SR for EN-32 steel. Recently, Mohanty et al. [65] used the capability of particle swarm optimization (PSO) for the optimization of nano-Al2O3-mixed-EDM process parameters for maximizing MRR and minimizing SR and TWR. Kennedy and Eberhart presented PSO first time in 1995 as an effective transformative computational procedure to optimize the response characteristics [66]. The PSO is developed to solve the consistent nonlinear streamlining issues after identifying the behavior of natural swarm bird coming back to perch and observed in numerous types of winged creatures [67,68,69]. The PSO method can produce top notch arrangements inside short computation time and stable merging attributes [70,71,72]. Because of the great execution of the PSO system, numerous process engineers/scientists discover the PSO as a productive option over other inquiry calculations particularly when managing multi-target streamlining issues [73].

The surface quality and the machining performance of Al–30SiCp MMC is a challenging issue in the machining process. However, there is no study available on the machining of Al–30SiCp MMC using MWCNT-mixed-EDM. In this way, the current study investigated the effect of MWCNT concentration on SR and MRR, further more multiobjective particle swarm optimization has been carried out to determine the optimal best setting to maximize the MRR and minimize SR. Till date no research study is available, which considered implication of PSO for multiobjective optimization of NPM-EDM process for Al-30SiCp MMC by using CNTs as powder additive.

2 Experimental Planning and Optimization

2.1 Materials and Experimentation

Widely used Al–305SiCp MMC was used as workpiece for machining using powder mixed electric discharge machining process. The surface of Al–305SiCp MMC was well grounded and cleaned with ethanol followed by drying at room temperature. The machining of Al–30SiCp workpiece was carried out by using multi-walled carbon nanotube (MWCNT) mixed-EDM process. The MWCNTS powder particles were used as dielectric solvent and mixed in dielectric fluid. The SEM micrograph of MWCNT powder particles are shown in Fig. 2. The experimental setup for the surface modification was in-house developed and called as nanopowder-mixed electric discharge machining (NPM-EDM), as shown in Fig. 3. In order to perform a NPM-EDM process, a separate tank of the dielectric capacity of 5 L was designed and MWCNT powder was mixed in the dielectric fluid at various concentrations 0, 2, 4, 6, 8 g/l. The die-sinking EDM machine (ELECTRONICA model 5535) has been utilized to conduct the experiments. A circular shape copper alloy rod of size ϕ 10 × 50 mm was used as an electrode for the machining process. Table 1 shows the detailed experimental conditions for the NPM-EDM process. From the initial trials, four input-process parameters were chosen for the examination. Table 1 presents the process parameters and their level.

Fig. 2
figure 2

SEM micrograph showing the morphology and size of MWCNT particles

Full size image
Fig. 3
figure 3

Experimental setup of HA mixed electric discharge machining process

Full size image
Table 1 Process parameters and their levels
Full size table

2.2 Materials Characterization

The surface roughness (SR) and material removal rate (MRR) are considered as output response characteristics. The surface roughness was measured by Mitutoyo surface roughness tester. The MRR was computed by dividing the material removal per unit time, as per the procedure adopted previously [32]. In experimentation, central composite rotatable design (CCRD) has been used as a module of response surface methodology. Table 2 shows the design of experiment and obtained value of MRR and SR.

Table 2 Experimental design with response characteristics values of MRR and SR
Full size table

2.3 Optimization Using MO-PSO

Because of clashing nature of output characteristics as MRR and SR, the single optimal settings of process parameter is not fulfilling the goals. In such circumstances, MO-PSO gives better execution when contrasted with the customary improvement strategy because of their heartiness, independency of slope data, and utilization of inborn parallelism in looking through the plan space. The algorithm flowchart of MO-PSO algorithm is shown in Fig. 4.

Fig. 4
figure 4

Algorithm flowchart of MO-PSO

Full size image

3 Results and Discussions

3.1 Effect of Process Parameter on MRR and SR

Figure 5 shows the 3D response surface plot for the material removal rate (MRR) with respect to process parameters. Figure 5a shows the effect of interaction of peak current (Ip) and pulse duration (Ton) on MRR. The MRR increased with peak current (Ip), this is because when Ip increased, a large amount of heat is liberated and sunk into the workpiece. As a consequence, the size and shape of pits/craters on the modified surface increases, which further increases the MRR. The MRR value increased from 1.05 to 1.28 mm3/min when peak current increased from 5 to 25 A. The MRR value first decreased with the pulse duration but, after certain value of pulse duration (150 μs), it starts increasing and goes on. This is due to the fact that discharge energy is proportional to pulse duration, thus an increase in the later enlarged the depth and width of craters [21, 74]. The MRR value increased from 1.03 to 1.55 mm3/min when pulse duration increased from 150 to 250 μs at peak current and duty cycle of 5 A and 8%, respectively. The highest MRR (1.55 mm3/min) is achieved at high level of peak current and high level of pulse duration. Figure 5b shows the effect of interaction of pulse duration (Ton) and powder concentration (Pc). The MRR increases with the MWCNT concentration in the dielectric fluid at any value of Ip and Ton. This is because, with the dispersion of MWCNT powder particles in the dielectric fluid, the spark locations increases and removed the material in larger proportion form the workpiece surface As a consequence, the top layer of workpiece is expelled in the form of micron-debris from large locations; thus, increased the MRR value. The MRR value increased from 1.15 to 1.38 mm3/min, when MWCNT concentration increased from 0 to 6 g/l at pulse duration and duty cycle of 50 μs and 8%, respectively. The highest MRR (1.66 mm3/min) is achieved at high level of peak current and high level of MWCNT concentration.

Fig. 5
figure 5

3D response surface plot of MRR with respect to input-process parameters and their interaction

Full size image

The MRR value is high in all cases of MWCNT mixed-EDM as compared to EDM. Figure 5c shows the variation of MRR value with respect to MWCNT concentration and duty factor. The MRR value increased with the duty factor at any value of peak current and pulse duration. This is due to the fact that as the duty cycle increases the pulse interval decreases and pulse duration increases. As a result, large amount of discharge energy sank into the workpiece material and causing removal of material in the form of deep and large craters. The MRR value increased from 0.90 to 1.35 mm3/min, when duty cycle increased from 8 to 80% at peak current and pulse duration of 5 A and 50 μs, respectively. The highest MRR (1.35 mm3/min) is achieved at high level of duty factor and high level of MWCNT concentration. The MRR increased very rapidly in combination with MWCNT concentration and pulse duration. The maximum MRR value has been obtained at high value of peak current, pulse duration, duty cycle, and MWCNT concentration. The best optimal condition where high MRR was obtained is A3, B3, C3, and D3. Table 3 presents the analysis of variance (ANOVA) for the MRR and showing all of the input-process parameters have significant contribution toward increasing the MRR. The mathematical model for the perdition of MRR was computed and represented in Eq. (3.1).

Table 3 ANOVA table for MRR
Full size table
$$\begin{aligned} {\text{MRR}} & = 1.16243 - 5.27625{\text{E}}{-}003*A - 4.62664{\text{E}}{-}003*B \\ & \quad + 7.01198{\text{E}}{-}003*C - 0.013202*D + 1.11325{\text{E}}{-}004*A*B \\ & \quad + 1.72313{\text{E}}{-}004*B*D - 1.00059{\text{E}}{-}003*C*D \\ & \quad + 9.71031{\text{E}}{-}006*B2 + 7.85020{\text{E}}{-}003*D2 \\ \end{aligned}$$
(3.1)

Figure 6 shows the 3D response surface plot for the surface roughness (SR) with respect to process parameters and their interactions. Figure 6a shows the effect of interaction of peak current (Ip) and pulse duration (Ton) on SR. As the peak current increases the SR increases, this is because when peak current increases, a large amount of heat is liberated and sunk into the workpiece. As a consequence, the increase in the size and shape of pits/craters increases the SR value from 1.22 to 1.80 μm. Similar trend was observed for the case of pulse duration parameter, as evidently. The SR value increases with the increase in pulse duration, because the discharge energy increased with the increase in pulse duration; thus deep and wide craters were developed on the machined surface. The SR value increased from 1.22 to 1.50 μm when pulse duration increased from 50 to 250 μs. The lowest value of SR (1.22 μm) is obtained at low level of peak current and pulse duration.

Fig. 6
figure 6

3D response surface plot of SR with respect to input-process parameters and their interaction

Full size image

Figure 6b shows the effect of interaction of MWCNT concentration (Pc) and pulse duration (Ton) on SR at any value of Ip and Ton. The SR value first decreases with MWCNT powder concentration but it starts increasing when the MWCNT powder concentration increases beyond 4 g/l. By the dispersion of MWCNT powder particles in the deionised water, the uniformity and sparking area increased but the increased discharge gap reduced the thermal energy resulted from the electrical sparks. As a result of this, small and tiny discharge craters were developed on the machined surface and lead to a decrease in the SR value of the machined surface [21]. The MWCNT powder concentration increased beyond 4 g/l, the powder particle get stuck with the plasma and get deposited on the machined surface. The SR value decreased from 1.79 to 1.35 μm, when MWCNT powder concentration increased from 0 to 4 g/l and it start increases from 1.35 to 2.14 μm when powder concentration increased from 6 to 8 g/l. The SR value increased with respect with the increase in duty cycle, due to the fact that as the duty cycle increased the pulse interval decreased and pulse duration increased. As a result, large amount of discharge energy sinked into the workpiece material and causing removal of material in the form of deep and large craters. The SR value increased from 1.79 to 2.12 μm, when duty cycle increased from 8 to 72%. The SR decreased very rapidly in combination with low peak current and 4 g/l MWCNT powder concentration. The minimum SR value was obtained at low value of peak current, pulse duration, duty cycle, and high level of MWCNT concentration. The best optimal condition where low SR was obtained is A1, B1, C1, and D3. Table 4 presents the analysis of variance (ANOVA) for the SR and showing all input-process parameters has significant contribution toward the SR. The mathematical model for the perdition of SR is computed, as represented in Eq. (3.2).

Table 4 ANOVA table for SR
Full size table
$$\begin{aligned} {\text{SR}} & = 1.30396 - 7.79792{\text{E}}{-}003*A + 1.34333{\text{E}}{-}003*B \\ & \quad + 7.71354{\text{E}}{-}003*C - 0.16473*D + 3.02500{\text{E}}{-}003*A*D \\ & \quad - 1.51953{\text{E}}{-}003*C*D + 7.79375{\text{E}}{-}004*A2 + 0.024402*D2 \\ \end{aligned}$$
(3.2)

3.2 Multiobjective Optimization Using MO-PSO

It is noticed that when SR reduced, the MRR reduces. The MRR requires to be maximized while regulating the SR. Because of clashing nature of MRR and SR, it is important to locate the optimal set of conditions of input-process parameters that yield high MRR and least SR. Accordingly, the function MRR is changed over into minimization type and the objective functions are altered as below:

  • Objective 1 = Minimize (1/MRR);

  • Objective 2 = Minimize (SR).

With a specific end goal to perform such multiobjective optimization, a source code of the proposed algorithm MO-PSO was executed utilizing MATLAB. Figure 7 shows the spread of Pareto front which contains the 100 optimal set of conditions where both functions were optimized. In this way, the choice of a result against different relies on item necessity as well as decision of process designer/engineer. It has been observed that MO-PSO predicted the best results within boundaries. Out of 100 global best solutions, the first 5 solutions are presented in Table 5.

Fig. 7
figure 7

Pareto optimal front of the MRR and SR using MO-PSO

Full size image
Table 5 Optimal solutions obtained from MO-PSO with confirmation test
Full size table

3.2.1 Comparison of Experimental and Optimal MRR

In the investigation and comparing of experiential results of MRR with Pareto front, it is observed that the maximum MRR is 1.45 mm3/min at peak current = 20 A, pulse-on = 200 μs, duty cycle = 24%, and MWCNT powder concentration = 6 g/l, corresponding to experiment no. 24 in Table 2. On the other hand, the maximum MRR obtained from the Pareto optimal front is 1.23 mm3/min at peak current = 25 A, pulse-on = 250 μs, duty cycle = 80%, and MWCNT powder concentration = 0 g/l as shown in Table 2 (serial no. 1). The MRR predicted obtained from the optimal solution is slightly lower than the experimental value. Moreover, from the 3D response of MRR (Fig. 5), it can be clearly seen that the MRR decreases if the pulse duration and duty cycle increases. So, it indicates that proper selection of process parameter level gives the maximum MRR. From Table 2 (experimental results no. 1), the minimum MRR is 0.987 mm3/min at peak current = 20 A, pulse-on = 200 μs, duty cycle = 24%, and MWCNT powder concentration = 6 g/l. On the other hand, the minimum MRR obtained from the Pareto optimal front is 1.134 mm3/min at peak current = 15.59 A, pulse-on = 169.61 μs, duty cycle = 65.17%, and MWCNT powder concentration = 4.08 g/l. The results show that the lower MRR obtained from optimal solution is much higher (about more than 1.7 times) as compared to the experimental value. This is because the MRR increases with increase of MWCNT powder concentration as more stable spark is generated in the machining zone.

3.2.2 Comparison of Experimental and Optimal SR

In the investigation and comparing of experiential results of SR with Pareto front, it is observed that the maximum SR is 1.987 μm at peak current = 20 A, pulse-on = 200 μs, duty cycle = 24%, and MWCNT powder concentration = 6 g/l corresponding to experiment no. 12 in Table 2. On the other hand, the maximum SR obtained from the Pareto optimal front is 1.4285 μm at peak current = 25 A, pulse-on = 250 μs, duty cycle = 80%, and MWCNT powder concentration = 0 g/l corresponding as shown in Table 2 (serial no. 1). The predicted SR value obtained from the optimal solution is very much lower than the experimental. Moreover, from the 3D response of SR (Fig. 6), it can be clearly seen that the SR decreases if the pulse duration decreases. So, it indicates that proper selection of process parameter level gives the maximum SR. From the Table 2 (experimental results no. 17), the minimum SR is 1.305 μm at peak current = 5 A, pulse-on = 150 μs, duty cycle = 40%, and MWCNT powder concentration = 4 g/l corresponding. On the other hand, the minimum SR obtained from the Pareto optimal front is 1.097 μm at peak current = 15.59 A, pulse-on = 169.61 μs, duty cycle = 65.17%, and MWCNT powder concentration = 4.08 g/l corresponding. The results show that the lower SR obtained from optimal solution is much higher (about more than 2.66 times) as compared to the experimental value. This is because the SR decreases with increase of MWCNT powder concentration as more stable spark is generated in the machining zone.

4 Conclusions

In this chapter, particle swarm optimization (PSO) is used for the multi-objective optimization of process parameters of MWCNT mixed-EDM process. The following conclusions are drawn after the critical observation of the results obtained:

  1. 1.

    The MRR increases with peak current, pulse duration, duty factor, and MWCNT powder concentration. The addition of MWCNT in dielectric fluid increases the uniformity and spark locations, as a result, more materials is removed from workpiece surface in comparison to EDM only.

  2. 2.

    The SR increases with peak current, pulse duration, and duty factor. It decreases with increase of MWCNT powder concentration, but, after 4 g/l, it starts increasing. The addition of MWCNT in dielectric fluid increases the discharge gap; as a result the reduced level of thermal energy caused the formation of micron and sub-micron debris, thereby improved the surface finish.

  3. 3.

    The MO-PSO provides a number of combinations of process parameters in comparison to experimental values, which indicates that proper selection of process parameters enhances machining efficiency.

  4. 4.

    The combinations of high peak current, pulse duration, duty factor, powder concentration are more suitable for getting higher value of MRR.

  5. 5.

    The combination of low peak current, moderate pulse duration, high duty factor, and 2–4 g/l MWCNT powder concentration are more suitable for the better surface finish.

  6. 6.

    The MO-PSO provides set of optimal solutions as presented in Pareto front. The optimum solutions facilitate for process engineer to select the optimal value of control parameters depending upon product requirement.