Evaluating the performance of electrical discharge face grinding on super alloy monel 400

Conventional machining of super alloy Monel 400 is difficult due to its high strength, low heat conductivity, and gummy nature. In this work, the performance of the electrical discharge diamond face grinding (EDDFG) while processing supper alloy Monel 400 is assessed using the super abrasive diamond-coated grinding wheels of three different grit numbers. Preliminary experiments highlighted the superior performance of the EDDFG as compared to that of electrical discharge machining (EDM) and electrical discharge face grinding (EDFG); in terms of material removal rate (MRR), average surface roughness (Ra), and microscopic images of surfaces using scanning electron microscope (SEM). This work aimed to assess the influence of the diamond grit number (DG n ) of the grinding wheel along with other process parameters, viz. grinding wheel speed (GWS), peak current (I p ), and pulse-on-time (T on ) on Ra and MRR. The analysis of variance (ANOVA) confirmed that DG n and GWS significantly affect the response variables. Finally, the multi-objective genetic algorithm (GA) optimization approach was used to determine the optimal parametric settings of the EDDFG process.


Introduction
Monel 400, a nickel-copper-based superalloy, has superior properties, including high strength and high resistance to corrosive environments.Thus making it suitable for applications in gas turbines, surface materials for spacecraft, marine environments, chemical processing, and oil and gas industries [1,2].However, issues like material pull-out, smearing, complexity in breaking a chip, rapid tool wear, etc, are commonly observed during the conventional machining of Monel 400.Thus, it is categorized as a difficult-to-cut material [3,4].Superalloys like Monel 400 can be easily processed using unconventional processes like electrical discharge machining (EDM).
EDM is extensively used for machining difficult-to-cut materials and superalloys, capable of fabricating precise and intricate shapes irrespective of their hardness [5].EDM is a thermal erosion process with no direct contact between the tool and workpiece during machining [6].The workpiece erosion occurs due to controlled electric spark discharges.Sparks occur between the workpiece and the tool electrodes during the EDM process when the dielectric fluid breakdown [7].Although only a small amount of material is removed by each spark, this process is repeated hundreds of times per second, eventually resulting in the formation of the desired cavity within the workpiece.But in the EDM process, the high energy discharges result in the formation recast layer over the machined workpiece surface [8].Also, poor dielectric flushing contributes to defects like the clogging of debris over the machined surfaces [9].Nevertheless, the EDM also has low MRR and poor surface finish.The restrictions of EDM can be overcome by using hybrid techniques, such as electrical discharge grinding (EDG) and electrical chemical discharge machining, etc.
The EDG process combines the grinding wheel rotary motion with EDM spark erosion.Unlike the stationary tool used in EDM, the EDG uses a rotating electrically conductive metal disc-shaped grinding wheel.The material removal rate (MRR) and the surface finish of the machined workpiece are improved by efficient flushing within the inter-electrode gap (IEG), which can be made feasible by the rotation of the grinding wheel.Various researchers have suggested the application of metal-bonded diamond grinding wheels to improve the EDG process.This variation of the EDG process is known as electrical discharge diamond grinding (EDDG).The metal-bonded abrasive (diamond) particles are at the periphery of the grinding wheel.When the material melts and vaporizes during EDM, these diamond particles remove molten material from the IEG.Utilizing the abrasive effects of a diamond grinding wheel, the MRR and surface quality of the machined workpiece improves [10].
According to Kozak [11], using a metal-bonded diamond grinding wheel leads to a two-fold increase in the MRR compared to a metallic rotating electrode.Additionally, it is claimed that the grinding wheel continually self-dresses itself without the need of interrupting the process.Yadav et al [12] performed EDDG and investigated the influence of process parameters, i.e., GWS, I p , duty factor (DF), and T on , on response parameters.They found GWS and I p as most important factors affecting the EDDG process.Koshy et al [13] found that the MRR during EDDG was improved by almost 98% as compared to the grinding process alone.
EDDG can be performed in different configurations, as shown in figure 1 and named as (i) electrical discharge face diamond grinding (EDDFG), (ii) electrical discharge diamond surface grinding (EDDSG), and (iii) electrical discharge diamond cutoff grinding (EDDCG).In EDDFG, the feed direction is perpendicular to the work material, and the diamond grinding wheel rotates vertically while grinding through the face side of the grinding wheel.Singh et al [14] integrated EDM with conventional grinding and named as EDDFG.In order to identify the optimal machining process parameters, they used Taguchi and Grey rational analysis.Their investigation showed that the MRR improved by 86.49%, wheel wear rate (WWR) and R a noted a decrease of 21.70% and 14.86%, respectively.
Based on existing literature and the current state of knowledge, a prominent gap remains concerning the machining of nickel copper-based alloys through the EDDFG process.Limited research addresses the statistical importance of the EDDFG process, and the influence of different diamond grit numbers grinding wheel during the EDDFG process remains unexplored.This study focuses on investigating the MRR and R a of machined Monel 400 using EDDFG.Primarily to validate the preeminence of EDDFG, a comparative analysis encompassing EDM, EDFG, and EDDFG processes was conducted to evaluate their performance in terms of MRR, R a , and surface topography.Subsequently, EDDFG was performed with diamond-coated grinding wheels of different grit numbers.Employing the Central composite design (CCD) technique, a total of 30 experiments were planned and carried out.Through Analysis of variance (ANOVA), the relationships between input process parameters and their interactions with the response parameters have been studied.A predictive mathematical model was then formulated to estimate MRR and average surface roughness (R a ) based on process parameters.The accuracy and suitability of these models were verified.Additionally, the multi-objective optimization technique, genetic algorithm (GA), was successfully employed.The GA optimization process provided optimized results, which were compared against validation experiments, including the calculation of percentage absolute error.

Comparison between EDM, EDFG, and EDDFG processes
In this section, the efficiency and productivity of the EDDFG process are analyzed and compared with the competencies of other well-known processes like EDM and EDFG.The comparison is based on crucial factors such as surface integrity, MRR, and R a .The experiments were performed on the flat surface of Monel 400 (15 mm × 15 mm × 5 mm).The diameters of the tool electrodes were fixed at 12 mm.The pure copper rods were used as tool electrodes for EDM and EDFG process experiments, while the diamond-coated copper rod was used for the EDDFG process experiment.The non-diamond-coated copper rods are rotated for the EDFG process while remaining stationary to be used as a tool electrode for the EDM process.While for the EDDFG process, a diamond-coated rod rotates at a speed of 400 rpm.The comparative experimentations were done at continual input process parameters as I p (10 A), T on (100 μs), and pulse off time (T off ) at (5 μs) for all the processes.At the same time, a GWS of 400 rpm is used for EDFG and EDDFG processes, and the GWS for EDM is at 0 rpm.For the EDDFG process, diamond coated grinding wheel of 270 grit number was used for the experiment, and each experiment was performed for 15 min.After each experiment, the MRR and R a values were measured.The MRR was determined using equation (1).
Various configurations of the EDDG process [15].Reproduced with permission.Copyright Springer Nature.
Table 1.The R a values and SEM images of machined Monel 400 using different processes.

Process
Machined surface images and R a value graphs SEM images EDM SEM micrograph under the machining conditions: I p (10 A), T on (100 μs), GWS (0 rpm), and T off (5μs).Here, MRR was measured in mm 3 /min.Wi = workpiece mass before machining in gm, Wf = workpiece mass after machining in gm, ρ was the density for the workpiece, and t denotes the machining time in minutes.An micro weight balance (Make: CITIZEN, Model:CX 165, with a least count 0.1 mg) was used to measure the difference between the workpiece weight before and after machining to determine the loss in workpiece weight.The R a value for the machined surface was determined by the surface roughness tester (Make: MITUTOYO, Model: SJ 200).The R a was measured three times at random loacations, and the mean value was taken for analysis purpose.Roughness profiles were measured according to the ISO 1997 standard; with cut-off wavelength (λc) = 2.5 μm with a multiplier (N) = 5, stylus travel length (Lst) = 12.5 mm and evaluation length (Lev) = 8 mm.
To evaluate the surface topography the microscopic images of machines samples were captured at 500x using a SEM equipment (Make: FEI, Model; Nova NanoSEM 450).Nova NanoSEM 450 offers a range of magnifications from 5x to 1,000,000x and a resolutions of 1.6 nm at 1 kv and 1 nm at 15 kv (TLD-SE).For SEM analysis, the samples (machined Monel 400) were cut in size (12 mm × 12 mm × 5 mm) using wire-EDM.Then the sample were cleaned in an ultrasonic bath using acetone for 5 min and air dryed for 15 min.The prepared samples were mounted on stubs using conductive adhesive tabs, ensuring that the samples securely firmely attached on the stub.The stubs were placed inside the specimen chamber of SEM and then chamber was clasing for maintaining high vacuum.The electron beam accelerating voltage set at 15 kV and secondary electron (SE) imaging was chosen for surface topography.Finally, the images were taken at 500x magnifications level of different area.After completing the imaging session, carefully remove the sample from the SEM chamber.
The SEM and photographic images were also taken along with R a measured value graphs of the machined Monel 400 sample by the application of these processes to differentiate these machining processes in terms of surface integrity, as shown in table 1.
Based on the SEM images obtained when employing the EDM, EDFG, and EDDFG processes to machine Monel 400, it can be seen that in EDDFG, the machined surface had minimal variation in surface topography like recast layer, craters, and globules as compared to other machining processes, and a comparatively smoother surface is obtained.The measured R a values support the same observation.The reasons for the surface defects mentioned above are discussed in the next paragraph.The variation of MRR concerning machining processes is shown in figure 2(a).When compared to the EDM and EDFG processes, the MRR achieved through the EDDFG process is notably higher, about 2-fold and 1.5-fold, respectively.During EDDFG, the workpiece material becomes thermally softened due to the EDM process discharge action.Then the diamond-coated abrasive particles cause the abrasive action for further material removal from the workpiece surface.The variation of R a concerning the machining processes is shown in figure 2(b).The R a value attained in the EDDFG process is noticeably lower (approximately 2.2 and 1.4 times) than in the EDM and EDFG processes.Section 4.4.2.1 provides explanations or an in-depth discussion of the R a value decreased during the EDDFG process while processing Monel 400.
Due to the high discharge energy during spark erosion, the material on the workpiece surface melts and evaporates.A dielectric liquid flushes away some portion of molten metal, and the remaining molten metal solidifies and forms a recast layer.This recast layer is very hard, brittle, and difficult to etch, and apparent as a white layer under microscopic view.The existence of the recast layer results in undesirable effects, such as reducing the fatigue strength and deterioration of the surface integrity of the machined surface [16].The machined surface presented lots of non-directional craters.The dimensions of these craters greatly depend on the amount of discharge energy that interrupts sparks and impulsive forces transferred from the electrode to the workpiece in a stochastic manner via electric discharges in the EDM process [17].The structure reveals an accumulation of debris across the machined surface in the form of globules that could cause a higher R a value, which can be due to the quenching process during the machining and inappropriate flushing across the machining zone.However, during the EDFG and EDDFG processes, the dimension of the craters on the machined surface gradually decreased.This was mainly due to the rotating grinding wheels circumferential speed and the effective flushing of the dielectric fluid along with the local expansion of the temperature distributions.

Experimental procedure 3.1. Workpiece
All experiments were performed on Monel 400 alloy.Monel 400 plays an essential role in manufacturing marine frames, nozzles, valves, and other pipe systems for marine applications.A commercially available Monel 400 (97% pure nickel-copper alloy) was taken as a workpiece.The chemical elements in Monel 400 are outlined in table 2.

Experimental setup and procedure
The experimental investigation of the EDDFG process was conducted to assess its performance in terms of MRR and R a .Figure 3 shows the photographs of the EDM machine (Make: Electronica, Model: ELTECH D300) and the EDDFG attachment used for experimentation.The Monel 400 plate as workpiece was securely mounted on the EDDFG setup.For EDDFG operation, the diamond-coated copper grinding wheels, as shown in figure 4, (diameter 12 mm) of three grit numbers (140, 205, 270) were chosen as tool electrodes.During the experiment, the tool electrode was positioned facing the workpiece surface.The grinding wheel description is presented in  table 3. The input process parameters were diamond grit number (DG n ), grinding wheel speed (GWS), peak current (I p ), and pulse on time (T on ), while material removal rate (MRR) and average surface roughness (R a ) were the response parameters.The three levels for each of the four process parameters were decided based on previous trial runs and are listed in table 4. The duration of each experiment was fixed at 30 min and repeated twice to ensure the took results were.A series of experiments were conducted by systematically varying the levels of above four process parameters to observe their influence on response parameters.In this work, RSM is utilized, and the experiments were prearranged and conducted based on CCD, a well-known design methodology within RSM.The background of RSM and CCD is presented below in section 3.3.Each experimental run was conducted for a 30 min duration, and the resultant MRR and R a were determined.To ensure the reliability of the results, each experimental run was repeated three times.After each experiment, the MRR and R a values were measured with the similar procedure as described in the section 2.

Response surface methodology (RSM) and central composite design (CCD)
In statistics, RSM explores the relationships between explanatory process parameters and their responses to specific criteria [18].RSM central idea is to conduct a sequence of planned experiments to find the best responses [19].Second-order polynomial response analysis was used to investigate the effects of the process parameters on the response, as shown by equation (2): Where Y stands for the corresponding response, a 0 , a i , a ii , and a ij , stand as coefficients for free terms, linear terms, quadratic terms, and interaction terms.The CCD technique decreases the overall number of experimental trials crucial to estimate the effects of individual process parameters and their interactions on the responses [20].The CCD is a six-face center trial full-  For improving the model accuracy, non-significant terms were removed using the backward elimination procedure.Table 6 provides a summary of the ANOVA results for MRR. Figure 5 illustrates an excellent The percentage contribution was calculated by dividing the 'individual (model) term' sum of squares by the 'model' sum of squares.The determination coefficient (R 2 ) was computed to determine if the fitted model accurately fits the experimental data.The R 2 values close to unity indicate that the model fits the data more accurately.The MRR R 2 value, as indicated in table 6, is 0.97, representing that the developed model accurately represents the process and can explain up to 97% of the variation in MRR.
Divide the standard deviation by the mean to determine the model coefficient of variation (CV).The lower value of CV (5.74%) indicates that the performed experiments were more accurate and consistent [22].The signal-to-noise ratio is another characteristic of the Adeq Precision.A ratio better than four is usually recommended.The Adeq Precision of 44.85 specifies an adequate signal for the model.Which is significantly greater than four.

ANOVA for R a
The ANOVA results for R a are outlined in table 5.The fit summary generated by Design-Expert software indicated that the quadratic model was adequate.After backward elimination, table 7 summarizes the ANOVA results highlighting significant parameters only.Figure 6 shows a resemblance between the actual data and the model predicted data, indicating the reliability of the ANOVA results (table 7).
The model F value of 60 and Prob > F value lesser than 0.0001 (shown in table 7) reveal that the developed model is significant for R a .Model terms relevance is shown by the values of Prob > F less than 0.05, which denote significant effects on the response.Significant model terms for R a are A, B, C, D, AC, BC, CD, A 2 , B 2 , C 2 , and D 2 with percentage contributions of 7.01, 19.86, 44.86, 6.82, 7.53, 1.46, 1.09, 2.34, 1.10, and 3.45%, respectively.
In this case, R 2 is 0.96 (see table 7), indicating that the developed model accurately represents the process and can account for up to 96% of the variation in R a .The lower value of CV (2.24%) reveals that the performed experiments were more accurate and reliable.A sufficient signal for the model is proven by the Adeq Precision of 31.72, which is significantly greater than four.As a result, the model may be used to predict R a .

Regression equation for MRR and R a
The following second-order polynomial equation (3) provides the final empirical relationship between MRR and input process parameters, while equation (4) is for R a and input process parameters in terms of actual process parameters.Design-Expert software has computed the coefficients of the process parameters in the below shown second-order polynomial equations after analyzing data in table 5 To check absolute error between experimental and predicted values was defined by the mathematical formula as shown in equation (5).Table 8 provides the results of validation experiments.From the results, it can be stated that the developed regression models for

Influence of process parameters
In this section, the impact of specific process parameters on performance benchmarks has been discussed on MRR and R a using perturbation plots and 3D surface plots.

Influence of process parameters on MRR
The perturbation plot depicted the influence of significant process parameters on MRR for Monel 400 by the EDDFG process (see figure 7).While plotting the effect of individual process parameters, the other parameters were kept at the constant reference point in the middle (coded as 0).A steep incline line for DG n (A), T on (D), and the sharp incline line for I p (C) shows that the MRR increases with an increase of these process parameters, and it also demonstrates high affectability of MRR towards them as compared to GWS (B).However, it was found that the initial increase in MRR with an increase of GWS (B) then diminishes at the higher end of the GWS; this can be due to wheel glazing [23].The following section discusses the interaction effects and briefly explains the results of the interaction plots.

Interactive effect of process parameters on MRR
In table 6, there has been a substantial contribution to the model from the interactions between the I p and GWS ('CB'), T on and I p ('DC'), and I p and DG n ('CA').All these interaction plots are shown in figures 8(a)-(c).Figure 8(a) depicts the interaction effect of I p and GWS, and the MRR increases with an increase of I p and GWS.This increase is because as the I p increases, the discharge energy increases, resulting in a greater volume of material being melted and evaporated.Thus, for a higher I p , a high MRR could be achieved.A similar finding was also reported by Sanjeev et al [12] and Kumar et al [24].More importantly, GWS found to positively impact the EDDFG process by increasing the MRR.When the tool is stationary in the EDM process (die-sink EDM), the removal of debris particles is challenging and could negatively affect the MRR.In the case of EDDG, the rotating abrasive grinding wheel enhances debris removal by improving the flushing mechanism and thereby expelling the debris and molten metal away from the IEG [25].
Consequently, MRR was found to be enhanced.At higher values of T on , the effect of I p on MRR is more prominently noticeable.Figure 8(b) shows the interaction effect of T on and I p .Here MRR increases as the T on increases.This is because the amount of material removed from the workpiece mostly depends on the discharge energy, which is the product of the I p and T on [21].Furthermore, figure 8(c) clearly shows that MRR increases with increasing I p and DG n ; this is due to the material removal through mechanical abrasion action of diamond coated grinding wheel and thermo-physical MRR, which leads to more material removal from the workpiece [10].

Influence of process parameters on R a
Figure 9 illustrates the perturbation plot and the comparative effects of process parameters on R a for Monel 400 by the EDDFG process.From the plot (see figure 9), it can be stated that R a is highly sensitive to T on (D), followed by I p (C).But moderately sensitive to both GWS (B) and DG n (A) because of the comparatively less sharp lines.

Interactive effect of process parameters on R a
In table 7, a substantial contribution to the model from the interactions between the I p and GWS ('CB') and I p and DG n ('CA') is presented.Figures 10(a) and (b) illustrate these interaction plots, respectively.The  perturbation graph (figure 9) illustrates that R a increases with an increase for all values of T on .Discharge energy increases with an increase in T on during the process duration, and at higher T on , the same discharge energy will act on the workpiece material for a longer length of time.Therefore, at larger T on , larger craters will form, resulting in more R a [26].Also, Renjie Ji et al [27] enlightened the correlation between R a and the machining process parameters (I p and T on ), as shown in equation (6).
Where SR a the surface roughness (μm), T on is the pulse on time (μs), I p is the peak current (A), K R the constant, and a, b are the proportionate constants.It can be seen from equation (6) that under specific machining conditions, the SR a increases with the increase of T on and I p , respectively; consequently, SR a increases.Figure 10(a) depicts that as GWS increases, the machined workpiece R a value decreases.Higher GWS improves flushing efficiency and effectively minimizes debris buildup in IEG [28].Additionally, each diamond abrasive particle will remove a little amount of material when the GWS rises.As a result, an even surface can be achieved.A similar finding was also reported by Agrawal and Yadava [29].It is also evident from figure 10(a) that a good R a value cannot be attained with a higher value of I p .At higher values of I p, the more discharge energy impacting the workpiece material and, consequently, the MRR will increase, resulting in larger craters and uneven surfaces on the machined workpiece.Due to the inefficiency of abrasive 'protrusion height,' larger craters cannot be machined entirely during the grinding process.The term 'protrusion height' refers to the height at which diamond grain projects from the wheel bonding material.So, higher I p values lead to lower surface quality and higher R a values.Figure 10(b) depicts the interaction between R a versus I p and DG n , showing that surface finish improves at lower values of I p and DG n .The coarseness of grain decreases as the DG n increase.A fine grain decreases the amount of material removed and the pressures required for grinding during machining.Additionally, it may be inferred that both the EDM and grinding mechanisms are performing at lower values of I p .A similar finding was also reported by Shrivastava et al [30].

Optimization
In this research, it is found that the combination of input process parameters has a reversal impact on MRR and R a response parameters.It is crucial to optimize both the response parameters, which can be achieved using a multi-objective optimization approach.

Multi-objective meta-heuristic algorithms
Meta-heuristic algorithms are used for multi-objective optimization problems with objectives having conflicting natures.The multi-objective problem is framed when the behaviour of the solution space containing all the objective values for the feasible input is challenging to determine in selecting a desirable solution.The meta-heuristic algorithms are commonly used for constraint and unconstrained problems.The primary objective of multi-objective meta-heuristic algorithms is to identify a set of solutions, known as the Pareto front The GA is one of the famous evolutionary algorithms (EAs) to solve multi-objective optimization problems.GA is a robust, pervasive optimization technique that can be used without domain-specific heuristics.GA uses natural selection, survival, and interbreeding population recombination to design a novel and unique optimal combination of parameters [31].GA is commonly employed to optimize complex machining processes instead of conventional approaches.The algorithm relies on natural genetics and natural selection principles.
The GA algorithm initiates with a group of solutions referred as the solution population, similar to the chromosome set in human genetics.The best solution from the population of solutions is selected and then used to form a new population solution utilizing genetic operators (reproduction, crossover, mutation), mainly hoping that the new population may outperform the earlier one.The solutions were based purely on their fitness to construct a modern solution (offspring), and also GA is not restrained by non-linear functions that are hard to be differentiable.One key feature of GA is that it explores several paths at the same instant, starting with the initial population.Ganesan and Mohankumar proposed [32] a multi-objective optimization technique that utilizes the GA approach to optimize the cutting parameters in turning processes.Specifically, the technique seeks to optimize cutting depth, feed, and speed to achieve the best possible results.The flow chart of the GA execution is shown in figure 11.

Optimization by GA
A MATLAB-based simulation is widely used to execute GA optimization.MATLAB is popularly used for carrying out optimization processes.Here MATLAB R2020b is used to achieve the GA optimization.The fitness function primarily relates to the objective function, which must be optimized.The fitness function, commonly regarded as the objective function, is perhaps a central component of the GA whose value should be minimized or maximized.
The study found that the interaction of input process parameters has an opposite consequence on MRR and R a .Therefore, optimizing both output parameters requires a multi-objective optimization approach.A multiobjective GA optimization was designed to identify the optimal process parameters settings that maximize MRR and minimize R a simultaneously.For the EDDFG process, identify the level of input process parameters that will suit both of the conditions.The mathematical problem was formulated as a composite problem considering it an objective function of MRR and R a .The objective function F (Y) is given in equation (7) and is minimized by executing the GA.The range of the input process parameters: DG n, GWS, I p , and T on , was defined as lower and upper boundaries in the MATLAB program.
< GWS < 600.0 5.0 < I p < 15.0 50.0 < T on < 100.0 Where, − MRR (− sign for minimization of maximizing MRR), R a (minimization of R a ).MRR and R a were normalized, respectively.By normalization, all the parameters have values in the same range.The considerations used in GA and their respective values have been summarized in table 9.

Pareto front of the optimization technique
Multi-objective problems are more challenging to solve than single-objective problems because there is no one optimal solution; relatively, there is a set of acceptable optimal solutions.This set is called Pareto optimal front [33].The Pareto front is a set of optimal solutions acquired by considering all input data and plotting or generating a curve, known as the Pareto front curve.All of the points on this curve represent the locus of optimal solutions, and all points on the curve are optimal.The most appropriate optimal solution can be found at any position along the Pareto curve, which lies between the initial and final points of the Pareto frontier [34].The MRR and R a are represented by the Y and X axes in the Pareto front curve shown in figure 12.The Pareto-optimal solution front is   12. Showing the formation of the front and the subsequent assortment of optimal solutions.As none of the solutions in the Pareto-optimal front outperform the others, any solution within this front (between initial and final points) can be considered acceptable.

Optimization results
In this paper, the multi-objective optimization GA was designed and employed.As the GA optimization technique depends on the parameters specified, a systematic, in-depth investigation was done to identify the setting of appropriate input process parameters.The GA had chosen the following optimal levels of process parameters for its exhaustive search.The recommended optimal process parameters settings from GA are shown in table 10.After completing the optimization, the next step is verifying the results.The confirmation experiments have been performed at optimal process parameters settings obtained from GA. Confirmatory experiments results and absolute error are also shown in table 10.
From the absolute error, as shown in table 9, it can be said that the GA optimization gives better prediction and confirmation results.Hence, it was determined that the GA optimization is appropriate to attain the optimized EDDFG process parameters.i.e., DG n of 262, GWS of 589 rpm, I p of 14 A, T on of 77 μs, respectively.

Conclusion
This study explored the influence of the DGn of the grinding wheel, followed by other associated process parameters, on the EDDFG process.Initial investigation revealed that the EDDFG process is considerably more efficient than the EDM and EDFG processes.Subsequently, the experiments were planned and carried out using the CCD approach.The machining responses MRR and R a were assessed using an ANOVA analysis.The process optimization was carried out using GA, a multiobjective optimization technique.The resulting findings allow for the following inferences to be established: 1.The EDDFG process exhibited improved MRR and offered minimal surface topographical variations, offering a smoother surface finish when compared with EDM and EDFG processes.
2. Based on the findings of the ANOVA analysis, DG n significantly influences the performance metrics of the EDDFG process, contributing 13% to MRR and 7% to R a .GWS is another parameter contributing to 18% of MRR and 19% of R a .Additionally, it can be noted that R a reduced by 54%, but MRR increased substantially by 73% when DG n increased from 140 to 270.When GWS was increased from 200 to 600 rpm, MRR was also improved, and R a was reduced.
3. The investigation identified the best machining settings for producing a multi-performance output: DG n = 262, GWS = 589 rpm, I p = 14 A, T on = 77 μs.
4. The results of the optimization highlighted the effectiveness of GA in addressing multi-objective optimization problems.GA validated its ability to locate near-optimal solutions, thus confirming its reliability in multiobjective optimization.

Figure 2 .
Figure 2. The variation due to machining processes in (a) MRR and (b) R a .

Figure 3 .
Figure 3. Photograph of EDM machine and EDDFG attachment.

Figure 5 .
Figure 5.The plot of predicted versus actual response for MRR.

Figure 6 .
Figure 6.The plot of predicted versus actual response for R a .

Figure 8 .
Figure 8.The interaction graphs for MRR (a) Interaction of I p versus GWS (b) Interaction of T on versus I p (c) Interaction of I p versus DG n .

Figure 9 .
Figure 9. Perturbation plots for R a .

Figure 10 .
Figure 10.The interaction graphs for R a (a) Interaction of I p versus GWS (b) Interaction of I p versus DG n .

Table 3 .
Description of the grinding wheel.
[21]orial design that comprises all grouping of the process parameters.The design is described as a 'face-centered CCD.' The face-centered CCD has 30 runs or experiments[21].4.1.ANOVA for MRRDesign-expert software was used to analyze the experimental results for MRR shown in table 5.The tests for lack of fit, model summary statistics, and the sequential model sum of squares were executed.The quadratic model was employed to analyze MRR since it is statistically significant, and the fit summary implies that it is adequate.

Table 4 .
Process parameters machining levels.

Table 5 .
Experiments with process machining parameters and their measured response.betweentheactualvaluesandthe values anticipated by the model.As a result, it can be stated that the ANOVA results shown in table 6 are reliable.The model designed is significant for MRR, as revealed by the model F value of 116.25 and Prob > F value lesser than 0.0001 (which is provided in table6).The possibility of noise being the cause of the considerable model F value is 0.01% at the very minimum.Model terms relevance is shown by the values of Prob > F less than 0.05, which denote significant effects on the response.Significant model terms for MRR are A, B, C, D, AC, BC, CD, and B 2 , with percentage contributions of13.65,18.68,46.21, 5.16, 10.85, 1.52, and 2.67%, respectively. correlation

Table 7 .
ANOVA analysis for R a .
a are trustworthy representations of the empirical results.

Table 8 .
Results after validation experiments.

Table 9 .
GA parameters and their respective values.

Table 10 .
Results from GA optimization.
a Used levels of parameters due to resources limitations and machining equipment restrictions 20 Mater.Res.Express 10 (2023) 096516 A S Kulshrestha et al