Strojniški vestnik - Journal of Mechanical Engineering 65(2019)10, 557-564 © 2019 Journal of Mechanical Engineering. All rights reserved. D0l:10.5545/sv-jme.2019.6194 Original Scientific Paper Received for review: 2019-06-13 Received revised form: 2019-08-26 Accepted for publication: 2019-09-10 Analysis of EDM Process Parameters on Inconel 718 Using the Grey-Taguchi and Topsis Methods Thangavel Yuvaraj* - Paramasivam Suresh Muthayammal Engineering College (Autonomous) Rasipuram, Department of Mechanical Engineering, India Nickel-based superalloys are gaining importance for their growing usage in aerospace industries. Amidst the various advanced machining processes, electro discharge machining (EDM) is considered to be an important one for its ability to machine materials irrespective of its intrinsic properties. In this study, Inconel 718 is considered as a work material, and an L18 orthogonal array (OA) experimental plan is utilized to machine the work material. The influential factors, which affect the EDM performance characteristics, are identified using analysis of variance (ANOVA). Not much work has been done regarding using grey-Taguchi technique for order of preference by similarity to ideal solution (TOPSIS) methods, although these methods can be easily applied for multi-objective optimization. These methods provide the best results with the available sparse data. The best combination of machining factors is determined using grey-Taguchi and TOPSIS methods. Based on the conducted experiments, voltage (V) and pulse off-time (toff) show a notable contribution on output performance. The optimal combination of input parameter through grey-Taguchi is found to be 10 A, 30 V, 200 ps, and 20 ps respectively, for the EDM parameters: current (I), V, pulse on-time (ton) and toff for improved response. Moreover, the best parameter setting (I = 10 A, V = 30 V, ton =100 ps and toff = 20 ps) is identified using the TOPSIS method for the performance measures machining rate (MR), tool wear rate (TWR), overcut (OC), and taper overcut (TOC). Further tool wear behaviour is also studied through scanning electron microscope (SEM) images by varying the voltage. Keywords: Inconel, ANOVA, grey-Taguchi, overcut, taper Highlights • EDM process parameters(voltage, current, pulse on- time and pulse off- time) were optimized through the L18 orthogonal experimental design method, and the grey relational analysis (GRA) method, considering multi-responses, such as machining rate (MR), tool wear rate (TWR), overcut (OC), and taper overcut (TOC). • The methods grey-Taguchi and TOPSIS were used to study the influential parameters that provide the best results with the available sparse data. • Based on GRA and TOPSIS, the optimum level parameters for EDM have been identified. Furthermore, the tool wear behaviour is also studied through SEM images by varying the voltage. • The optimal combination of the input parameter to acquire better responses based on grey-Taguchi has been found to be (10 A), V (30 V), ton (200 ps), and toff (20 ps). According to ANOVA, the voltage and toff plays a prominent role in machining Inconel 718. • The best combinations identified using the TOPSIS method for better performance measures is 10 A, 30 V, 100 ps and 20 ps. 0 INTRODUCTION Electro discharge machining (EDM) has a one-off significant process for the machining of hard materials and superalloys. Heat-resistant superalloys (HRSA), especially the Inconel alloy, find applications in aerospace and marine components, cryogenic storage tanks, and nuclear reactor components. Understanding the importance of the Inconel alloy, manufacturing scientists and engineers are attempting to understand the behaviour of it through EDM processes. In the machining of Inconel 718, Shen et al. [1] applied high-speed EDM with air as a dielectric medium and produced components with better surface quality. The importance of powder-mixed dielectric fluid in the EDM for machining Inconel alloy was analysed by Talla et al. [2]. The researchers revealed better results for surface quality and accuracy while using different powders, such as graphite, silicon, and aluminium [3] and [4]. Torres et al. [5] investigated the behaviour of an Inconel 600 alloy through EDM process. They studied the electrical parameters' influence and concluded that the change in polarity has a significant influence on surface quality.Tanjilul et al.[6] reported that a novel flushing method and a machining current significantly influences the debris removal. The size of the debris particles increases with increasing machining current. Rajesha et al. [7] studied the effect of process parameters such as current, duty cycle, sensitivity control, inter-electrode gap control, and flushing pressure on the material removal rate (MRR) and surface roughness (SR). They found that the pulse current and duty factor has the highest influence. Kuppan et al. [8] reported that the MRR and SR increase with the increase in peak current, duty factor, and electrode speed. *Corr. Author's Address: Department of Mechanical Engineering, Muthayammal Engineering College (Autonomous), Rasipuram, India,yuvarajt2019@gmail.com 557 Strojniski vestnik - Journal of Mechanical Engineering 65(2019)10, 557-564 Mohanty et al. [9] conducted an L9 orthogonal array (OA) and optimized the EDM process parameters of Inconel 825 using grey relational analysis (GRA). The parameters combination ( I (1A), ton (10 ^s), and duty cycle (75 %)) showed good results for the rate of material removed, surface quality, and radial overcut. Mohanty et al. [10] highlighted the importance of cryogenic treatment of tool in EDM on output performance using technique for order of preference by the similarity to ideal solution (TOPSIS) method and teaching-learning-based optimization algorithm. Dang [11] optimized the EDM parameters using a kriging model and particle swarm optimization method,finding that the model and method is suitable for the optimization of EDM process. Lin et al. [12] optimized and enhanced the EDM process parameters for Inconel 718 through multi-objective optimization technique using grey-Taguchi. Muthuramalingam and Mohan [13] machined SS 201 through EDM and established the importance of peak current. Lin et al. [14] conducted experiments on Ti-6Al-4V alloy through Micro-EDM and studied the influence of process parameters and gaps using the grey relational analysis (GRA)-Taguchi technique. The use of the GRA with Taguchi technique yields better results for tool electrode wear and overcut.Based on the above literature, the characteristics of EDM for machining holes on Inconel 718 are influenced by various input parameters. Each performance characteristic has different combinations of optimal process parameters and thus, in the case of multiple responses, the selections of optimal machining parameters are difficult. The grey system, proposed by Deng [15], handles the vague information and thus the GRA method is recommended as a principal method for multiple response optimisation. Therefore, researchers optimized the machining of Inconel 718 using electrical parameters and inter-electrode gap (IEG).In this paper, EDM process parameters: I, V, ton and toff were optimized through the L18 orthogonal experimental design method, and the GRA method considering multi-responses, such as machining rate (MR), tool wear rate (TWR), overcut (OC) and taper overcut (TOC). Not much work has been done using the grey-Taguchi and TOPSIS methods although these can be easily applied for multi-objective optimization. The significant contribution of this research is in using these methods to study the influential parameters that provide the best results with the available sparse data.Based on GRA and TOPSIS, the optimum level parameters for EDM have been identified. Furthermore, the tool wear behaviour is also studied through scanning electron microscope (SEM) images by varying the voltage. 1 EXPERIMENTAL The experimental set-up for EDM machining process is shown in Fig. 1. It consists of a maximum working voltage of 415 V, maximum current of 25 A, work table size of 600 mm x 400 mm, a maximum electrode length of 400 mm, and a servomotor for inter-electrode gap control. Inconel 718 has been selected as a workpiece material, whereas a brass electrode of 0 0.5 mm is used with EDM oil as the dielectric medium. Tables 1 and 2 provides the details of chemical and mechanical properties of Inconel 718 [16]. The thickness of workpiece is 3.1 mm. Since the discharge energy is primarily determined by current (I), voltage (V), pulse on-time (ton), and pulse off-time (tof), these factors are used as the input parameters. These parameters are selected based on the literature review, and levels are identified based the preliminary experiments;10 A, 12 A and 14 A have been considered as current variables; 30 V, 40 V and 50 V have been chosen as voltage variables;100 ^s, 150 ^s and 200 ^s have been chosen as ton values with tf values of 20 ^s, 30 ^s and 40 ^s. Table 3 shows the experimental layout using an L18 OA. MR is calculated by dividing the length of the through hole with the machining time required to complete the through hole. TWR is calculated using the relation mass is p x v, where p is brass density, v volume of brass tool (nr2h), h brass tool height. During the start and end of each Fig. 1. Experimental setup 558 Thangavel, Y. - Paramasivam, S. Strojniski vestnik - Journal of Mechanical Engineering 65(2019)10, 557-564 Table 1. Chemical composition of Inconel 718 (weight %) C Mn SI Cr NI Co Mo Nb+Ta Ti Al Fe 0.040 0.08 0.08 18.37 53.37 0.23 3.04 5.34 0.98 0.50 17.80 experiment, the length of the brass tool is measured, and the difference is noted as h; this value is used to calculate the volume of brass electrode. The ratio of volume of electrode wear to the time taken to complete the experiment is called TWR. OC is calculated using optical microscopic images. OC is defined by AR = Re - Rt, where Re is the entrance radius of the machined hole and Rt the tool electrode radius.The difference in Re and Rt results in OC. TOC is defined by TOC=D - d/ (2L), where, D is entry diameter of the machined hole, d exit diameter of the machined hole, and L thickness of the workpiece. Table 2. Mechanical properties of Inconel 718 Hardness [HB] Yield limits [MPa] Tensile stress [MPa] 388 1375 1170 2 RESULTS MR is the measure of machinability of the material. Hence, for a characteristic like MR, "larger-the-better" Table 3. L18 orthogonal array Range of factors Actual value of parameters Experiment No. A B C D I V ton IPS] toff ^S] MR [mm/min] TWR [g/min] OC [mm] TOC [mm] 1 1 1 1 1 0.019137 0.00261 0.2419 0.04839 2 1 2 2 2 0.042738 0.00466 0.1630 0.06966 3 1 3 3 3 0.065654 0.00205 0.1978 0.03706 4 2 1 1 2 0.056586 0.00351 0.1885 0.05388 5 2 2 2 3 0.049971 0.00410 0.1328 0.03871 6 2 3 3 1 0.062078 0.00317 0.1398 0.04179 7 3 1 2 1 0.090978 0.00905 0.1398 0.01508 8 3 2 3 2 0.051362 0.00366 0.1143 0.06070 9 3 3 1 3 0.072961 0.00675 0.0911 0.04752 10 1 1 3 3 0.044799 0.00371 0.1398 0.03871 11 1 2 1 1 0.039450 0.00436 0.1027 0.04284 12 1 3 2 2 0.032244 0.00559 0.1444 0.04806 13 2 1 2 3 0.081764 0.00526 0.1537 0.03786 14 2 2 3 1 0.050321 0.00456 0.1491 0.05987 15 2 3 1 2 0.043268 0.00399 0.1003 0.04493 16 3 1 3 2 0.070008 0.00488 0.1676 0.06497 17 3 2 1 3 0.062579 0.00699 0.1235 0.11196 18 3 3 2 1 0.081983 0.00456 0.1769 0.09516 is considered and the obtained MR data values are homogenized as shown below [14]: ./n xi (i)- min x- (i) x* (i ) = — n—iin' w minXj (i)-minxj (i) where x* (i) are the homogenized MR after the preprocessing, is the signal-to-noise ratio of the MR, where i = 1 for MR; i = 1, 2, 3, ..., 18 for experiments 1 to 18. EDM performance is also measured using TWR, OC and TOC. Hence, to achieve better machining quality, the "smaller-the-better" is considered in view of minimizing TWR, OC, and TOC. Therefore the actual sequence must be normalized as follows [14]: , , . max xf (i)- x, (i) x* (i) =-(J) • M' (2) max Xj (i)- mm Xj (i) where x0* (i), A0j (i), xj (i) are reference sequence, deviation sequence and comparability sequence, respectively. Ao j (i )=K (i)-(i)|. (3) Analysis of EDM Process Parameters on Inconel 718 Using the Grey-Taguchi and Topsis Methods 559 Strojniski vestnik - Journal of Mechanical Engineering 65(2019)10, 557-564 Similarly, other computation was carried out for 18 experiments and the values of all A0j for j = 1, 2, 3, ..., 18 are represented in Table 4. In continuation of data preprocessing, a coefficient for grey relational analysis is found using the relation given below [14]: S, (i ) = A . +Z -A mm ~ m (4) Ao, (k ) + Z-Amax' where Z is a unique coefficient and Z is considered as 0.5 because all parameters are given equal weight. The grey relational grade (GRG) presented in Table 5 is calculated by averaging the GRC and overall Table 4. Performance characteristics of the processed data Experiment Performance characteristics after data processing Deviation sequences run reference MR TWR OC TOC MR TWR OC TOC sequence [mm/min] [g/min] [mm] [mm] [mm/min] [g/min] [mm] [mm] 1 0.0000 0.9197 0.0000 0.6562 1.0000 0.0803 0.0000 0.3438 2 0.3285 0.6270 0.5231 0.4366 0.6715 0.3730 0.5231 0.5634 3 0.6475 1.0000 0.2924 0.7731 0.3525 0.0000 0.2924 0.2269 4 0.5213 0.7921 0.3539 0.5995 0.4787 0.2079 0.3539 0.4005 5 0.4292 0.7069 0.7231 0.7561 0.5708 0.2931 0.7231 0.2439 6 0.5977 0.8399 0.6769 0.7243 0.4023 0.1601 0.6769 0.2757 7 1.0000 0.0000 0.6769 1.0000 0.0000 1.0000 0.6769 0.0000 8 0.4486 0.7697 0.8462 0.5291 0.5514 0.2303 0.8462 0.4709 9 0.7492 0.3289 1.0000 0.6652 0.2508 0.6711 1.0000 0.3348 10 0.3572 0.7634 0.6769 0.7561 0.6428 0.2366 0.6769 0.2439 11 0.2827 0.6695 0.9231 0.7135 0.7173 0.3305 0.9231 0.2865 12 0.1824 0.4937 0.6462 0.6596 0.8176 0.5063 0.6462 0.3404 13 0.8718 0.5416 0.5846 0.7649 0.1282 0.4584 0.5846 0.2351 14 0.4341 0.6418 0.6154 0.5377 0.5659 0.3582 0.6154 0.4623 15 0.3359 0.7221 0.9385 0.6919 0.6641 0.2779 0.9385 0.3081 16 0.7081 0.5952 0.4923 0.4851 0.2919 0.4048 0.4923 0.5149 17 0.6047 0.2939 0.7846 0.0000 0.3953 0.7061 0.7846 1.0000 18 0.8748 0.6418 0.4308 0.1734 0.1252 0.3582 0.4308 0.8266 Table 5. Computed Grey relational grade GRC Grey relational grade (GRG) Experiment No. MR 4(i) TWR 4?) OC 4(3) MR à™ y=1/4 • (4i)+4-2)+43)+44)) 1 0.3333 0.8617 1.0000 0.5926 0.6969 2 0.4268 0.5727 0.4887 0.4702 0.4896 3 0.5865 1.0000 0.6310 0.6879 0.7263 4 0.5109 0.7063 0.5856 0.5552 0.5895 5 0.4669 0.6304 0.4088 0.6722 0.5446 6 0.5542 0.7574 0.4248 0.6446 0.5952 7 1.0000 0.3333 0.4248 1.0000 0.6895 8 0.4755 0.6847 0.3714 0.5150 0.5117 9 0.6660 0.4270 0.3333 0.5989 0.5063 10 0.4375 0.6788 0.4248 0.6722 0.5533 11 0.4108 0.6020 0.3514 0.6357 0.5000 12 0.3795 0.4969 0.4362 0.5949 0.4769 13 0.7959 0.5217 0.4610 0.6802 0.6147 14 0.4691 0.5826 0.4483 0.5196 0.5049 15 0.4295 0.6428 0.3476 0.6188 0.5097 16 0.6314 0.5526 0.5039 0.4927 0.5451 17 0.5585 0.4146 0.3892 0.3333 0.4239 18 0.7997 0.5826 0.5372 0.3769 0.5741 560 Thangavel, Y. - Paramasivam, S. Strojniški vestnik - Journal of Mechanical Engineering 65(2019)10, 557-564 assessment of the multiple objective optimization determined using Eq. (5). Yj = 11** (a)' (5) where jj is the GRG of jth experiment and b is the number of performance characteristics. 3 DISCUSSION The multi-response performance index presented in Table 6 presents the average value of the GRG for every level. The highest value of GRG indicates the best possible level of the process parameters. The calculated higher GRG value indicates the closeness to the optimal value. The total mean of the GRG for the eighteen runs was estimated and is presented in Table 6. The optimal parameter combination for better MR and lesser TWR, OC and TO is found to be (AjBjCsDj) as given in Table 6. Table 6. Multi response performance index Symbol Level 1 Level 2 Level 3 Main effect (max-min) A 0.5738 0.5598 0.5418 0.0321 B 0.6148 0.4958 0.5648 0.1191 C 0.5377 0.5649 0.5728 0.0351 D 0.5934 0.5204 0.5615 0.0730 y, = = 0.5585 Moreover, Fisher's test (F test) is also performed to establish machining parameters' influence on the performance characteristic [17]. ANOVA for GRG is presented in Table 7. Based on the ANOVA, table voltage and pulse off-time show a higher percentage contribution, hence voltage and pulse off-time are dominant parameters that affect the MR, TWR, OC and TOC. In EDM, increase voltage increase the current required for machining which improves the ionization effect between the tool and electrode. This ionization effect increases the temperature of the tool and electrode, resulting in melting of workpiece material. The molten material resulted from heating is evaporated during pulse off time of the EDM process. Hence, the voltage and pulse off-time are considered as significant factors in EDM. Table 7. ANOVA table Factors DoF SS MSj F ratio