Investigation of parameters and morphology of coated WC tool while machining X-750 using NSGA-II

In the present work, Tungsten carbide insert PVD coated with TiAlN is used as tool as shown in figure 1(a). The tool insert has the signature as the angle of approach: 91°; clearance angle: 0°; inclination angle: −6°; nose radius: 1.2mm, and rake angle: −6°. Tool holder Model number LI FENG JING MI WTJNR 25252M16 is used for holding the tool bit while experimentation on turning operation. Scanning electron microscopy (SEM) of the PVD coated tool insert before turning operation as shown in figure 1(b) to compare with SEM of tool insert after turning operation to investigate the morphology of insert, done at Chandigarh University, Chandigarh, India. Nickel based super alloy X-750 (crystallite size: 144°A according to Scherrer formulations) is used as workpiece, size 30 mm in diameter and 110 mm in length. The elements present in the work piece are provided in table 2 and it is observed that after Ni, Cr has the maximum contribution, followed by Fe, Ti and other elements that have a small content percentage.

Zoom In Zoom Out Reset image size

The experiments were planned according to response surface methodology-based Box-Bhenken Design (BBD), is based on 12 mid edge nodes and three central nodes to satisfy the 2nd order equation and its prime benefit is to generate higher order response with fewer required experiments. The other benefit of the BBD is the reduction in the number of experiments (figure 2(a)). Additionally, CCD select one value lower and higher than the given range, due to which the search space is extrapolated, but BBD works in the given range. Thus, BBD is selected in the present work. The experiments are conducted within the polynomial region, and the design is rotatable (or near rotatable) [37]. A total of seventeen experiments were planned and performed as per the run order. NSGA-II is used to predict the solutions of MRR and TW at different parametric settings of tool rotational speed (TRS), depth of cut (DoC), and feed (F).

Zoom In Zoom Out Reset image size

In the present work, Tungsten carbide inserts coated with TiAlN are used to machine, i.e. turning operation on Nickel based super alloy X-750 by a semi-automatic lathe machine named Bhilkhu, situated at Ambala, India. Machining parameters with their levels are shown in table 3. The 17 experiments are performed with different parameters selected in the present work are tool rotational speed (TRS), depth of cut (DoC) and feed (F) as suggested by (DoE) RSM. The experimental setup is shown in figure 2(b). The MRR (in mm³ min⁻¹) is measured using weighting balance ((ViBRA/ESSAC with a tuning-fork sensor)), as shown in figure 2(c), and TW (Conation- SuXma) (in mm) is calculated using an optical microscope, as shown in figure 2(d). Three readings are recorded for both responses. The average of three is used for analysis. After machining, the morphology of tool insert and chips formed at optimised value is studied using a scanning electron microscope (SEM) as shown in figure 2(e) (SEM and EDS setup at NITTTR Chandigarh, India) (Jeol make JSM-IT100, Japan). Energy Dispersive Spectroscopy (EDS) is used for the elemental mapping of the rake face of the tool insert for the existence of different elements (F). The process implemented in the current work is represented in figure 2(f).

The ANOVA clarifies the significant input parameters in the process and their percentage contribution to the evaluation of the responses. The table 5 depicts the ANOVA for MRR and TW, and it is evident from the data that all the parameters have influential contributions to the computation of MRR. The two-factor interaction model was observed to be significant in the case of MRR. The P-values of all the input parameters are less than 0.05, which verifies the influential contribution of all the parameters. A high F-value of the input parameter corresponds to the highest contribution for the evaluation of the input parameter. From this, it is interpreted that DoC has the maximum contribution for MRR, followed by F and TRS because F-value has highest value 11985.78, 5327.01 and 3749.64 respectively. The value of R2, Pred-R2, and Adj-R2 also verifies the presence of a suitable ANOVA. In the present study the value of Adeq-Prec. is 201.068 i.e. greater than value 4 with condition if Adeq-Prec. is more than value 4 reflects good model. The present model is good one.

The statistical summary of TW shows that the linear model fits the computation of the TW. The P-value, less than 0.05 for the parameters, signifies the influential role of all parameters in the TW. The F-values show that F has the maximum role in the TW, trailed by DoC and TRS. The lack of fit is non-significant, with a value of 0.7735. The importance of R², Pred-R², Adj-R² and Adeq-Prec. suggests a good model and good ANOVA. The difference between Pred-R² and Adj-R² is observed within the permissible limit.

Figure 3(a) presents the predicted versus actual plot for MRR, and it has been observed that all the residuals fall along a straight line. For a suitable ANOVA, all the residuals should lie in a straight line, and it is true for the present work, which means that the ANOVA for MRR represents a good model for future outcomes. The value of MRR with all three varying input parameters is depicted in the cube plot for MRR in figure 3(b). The plot shows that MRR varies from 612.65 to 810.65 mm³ min⁻¹ at F: 0.06 mm rev⁻¹ and DoC: 0.1 mm, while TRS varies from 900 to 1600 rpm. In other events, when TRS varies from 900 to 1600 rpm, the MRR varies from 1434.65 to 2820.65 at F: 0.06 mm rev⁻¹ and DoC: 0.3 mm. Similarly, when TRS varies from 900 to 1600 rpm, the MRR varies from 3162.65 to 5340.65 mm³ min⁻¹ at F, and DoC is 0.12 mm rev⁻¹ and 0.3 mm, respectively.

Zoom In Zoom Out Reset image size

All the residuals falling along the straight line, representing uniform distribution, reveals suitable ANOVA for the present work, as shown in figure 4(a) actual versus predicted plot for TW. The TW at a different set of three input parameters may be interpreted from figure 4(b). The cube plot shows the variation of TW corresponding to all three input parameters. When TRS varies from 900 to 1600 rpm, then tool wear varies from 0.24 to 0.32 mm at F: 0.06 mm rev⁻¹ and DoC: 0.10 mm. In another case, Tool wear increases from 0.58 to 0.67 mm when TRS increases from 900 to 1600 rpm at DoC: 0.30 mm and F: 0.12 mm rev⁻¹. Similarly, TRS and DoC are kept constant at 1600 rpm and 0.1 mm, respectively, and changing the F value from 0.06 mm rev⁻¹ to 0.12 mm rev⁻¹; the TW changes from 0.42 mm to 0.5 mm

Zoom In Zoom Out Reset image size

The models (empirical) developed for the MRR and TW are given in equations (1) and (2), respectively.

$$\begin{eqnarray}&&\begin{array}{c}MRR\,=2125.07563-1.69714\times TRS-10607.14286\times DoC-23571.42857\\ \,\times \,F+8.48571\times TRS\,\times \,DoC+18.85714\times \,TRS\times \,F+1.18000E+005\times \,DoC\,\times \,F\end{array}\end{eqnarray} \tag{ 1 }$$

$$\begin{eqnarray}&&TW=-0.17339+0.000117857* \,TRS+0.862* \,DoC+2.916* F\end{eqnarray} \tag{ 2 }$$

NSGA-II, Non-dominated sorting genetic algorithm, is suitable to assess the all points by mathematical calculations for optimization, and opted to generate the set of optimal solutions, known as Pareto set and also called nondominated solutions which makes compromise between all objectives without any biasing [30]. It provides the set of good quality solution, which gives more flexibility to users for the investigation of preferences in a particular situation. It is also found that NSGA-II provides better solution as compared to other optimization technique especially in the situation, where the responses are more than one and are of contradictory in nature. Although, there are other available methods, which are simple to implement, but still NSGA-II works better in multi-objective problems [38, 39]. In this the Pareto front is calculated according to the predicted optimal set of solution. A non-dominated solution is one in which a compromise has been set between the performance characteristics without degrading the other solutions. In this method, a solution (say f) is compared with all the predicted solution on that particular setting, if any solution does not dominate solution 'f'; then solution 'f' is termed as non-dominated solution and falls in the set of Pareto front. As compared to other available technique this technique is easy to apply and the results are comparable to other technique. In the current research work, two machining characteristics MRR and TW were considered with larger the better type quality and lower the better type quality respectively. Thus, MRR and TW are opposite, which means if MRR increases, then TW increase. Therefore, on one side production rate increases (with the rise in MRR) and another side, the operating cost also increases (with the rise in TWR). Hence, NSGA II is utilized to optimize two responses i.e. MRR and TW, simultaneously by setting the compromise between them [40, 41]. The machining response is provided in equations (3) and (4).

$$\begin{eqnarray}&&Objective\,\left(1\right)=\,-\,\left(MRR\right)\end{eqnarray} \tag{ 3 }$$

$$\begin{eqnarray}&&Objective\,\left(2\right)=\,\left(TWR\right)\end{eqnarray} \tag{ 4 }$$

The flow diagram of the optimization process is depicted in figure 5. The below-mentioned working parameters are selected to investigate the solutions with minimum computational effect, (i) mutation probability: 0.2; crossover probability: 0.8; the maximum number of generations: 50; population size: 100. The empirical model developed as equations (3), and (4) are solved in the working range of machining parameters. The machining parameter range is given from equations (5)–(7).

$$\begin{eqnarray}&&900\,\leqslant \,Rotation\,speed\leqslant \,1600\end{eqnarray} \tag{ 5 }$$

$$\begin{eqnarray}&&0.1\,\leqslant \,Depth\,of\,cut\,\leqslant \,0.3\end{eqnarray} \tag{ 6 }$$

$$\begin{eqnarray}&&0.06\,\leqslant \,Feed\,\leqslant \,0.12\end{eqnarray} \tag{ 7 }$$

The Pareto optimal front of both objectives is investigated using NSGA-II, as shown in figure 6. A total of thirty solutions were predicted and are mentioned in table 6. Table 6 shows that the whole range of solutions was predicted with no bias with the MRR or TW side. Both the machining responses are given equal importance, and all the Pareto fronts co-existed in the population. It is clear from the solutions that one solution is no better than another, and any solution can be considered acceptable. The choice of solution can be considered according to the solution obtained. If higher productivity is desired, then a solution with a higher MRR is selected. If a solution with the minimum operating cost is desired, then a low value of TW is chosen.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The optimization of machining parameters using NSGA-II shows the significant improvement in the response variable i.e. MRR and TW. The confirmation experiments are conducted on the randomly selected machining parameter setting from table 6.

The confirmation experiments (table 7) are repeated three times and the average of three is compared with the predicted solutions. The comparison between the predicted and experimental solution is made and the percentage variation is less than 7%, which shows the good result reproduction. In previous published results [42, 43], the percentage error is less than 7%. The error may occur because of the human error, procedural error or environmental error. Micro-structural images are captured at TRS₉₀₀DoC_0.06F_0.1, and TRS₁₆₀₀DoC_0.12F_0.15 suggested by NSGA-II is shown in figure 7. There is remarkable abrasive mark, means visible crater wear is observed on the coated tungsten carbide tool by using Scanning Electron Microscopy (SEM) after machining as shown in figure 7. Figures 7(a) and (b) shows the minimum crater wear is 148.054 μm and 178.045 μm on TRS₉₀₀DoC_0.06F_0.1, and TRS₁₆₀₀DoC_0.12F_0.15 respectively. Furthermore, maximum crater wear is 198.494 μm and 298.101 μm on TRS₉₀₀DoC_0.06F_0.1, and TRS₁₆₀₀DoC_0.12F_0.15 respectively. Moreover, average crater wear is 196.24 μm and 254.95 μm on TRS₉₀₀DoC_0.06F_0.1, and TRS₁₆₀₀DoC_0.12F_0.15 respectively.

Zoom In Zoom Out Reset image size

Tool wear is measured 0.392 and 0.7316 at TRS₉₀₀DoC_0.06F_0.1, and TRS₁₆₀₀DoC_0.12F_0.15 respectively by Optical microscope as shown in figures 8(a) and (b), which signifies a significant tool wear. However, the EDS shows the presence of WC, Ti and N due to coating of TiAlN on the WC tool insert. Figure 8(c) depicts the optimal image of the tool insert, which signifies significant tool wear.

Zoom In Zoom Out Reset image size

It is observed that, at TRS₁₆₀₀DoC_0.12F_0.15 parametric setting material removal rate (2851.93) and tool wear (0.7316) is more than at TRS₉₀₀DoC_0.06F_0.1 parametric setting with material removal rate (482.53) and tool wear (0.392) due to high friction produced because of high tool rotational speed, more depth of cut and more feed rate.

The chip morphology or chip removal in the machining process indicates the nature of the tool-work interface. This morphology may be extruded, elongated, blocky or segmental/lamellar, which depends upon the status of the process. During the machining process, no lubricant is used, and the coated carbide tools are used for the machining process. The chips are collected after a particular gap with regular intervals of time. At the initial stage, chips are not uniform, and the materials sometimes come in contact with a tool, or sometimes a gap is formed due to deflection. Thus, once the machining process starts, the chips are collected after an interval of time. So that uniform chips are obtained and can be easily analyzed. Usually, the segmental chips are obtained at low speed. As the speed increases, the chips are serrated, and continuous status is obtained. Figure 9(a) shows the transition of the chip from serrated to continuous serrated. The serrated chips have just started forming at TRS: 900 RPM; DoC: 0.1 mm; F: 0.09 mm rev⁻¹. The two arrows in figure 9(a) depict the two regions, representing the chips' inhomogeneity. These types of chips are formed in dry machining due to the high temperature. Also, the red path is drawn to check the serration status. This indicates the transition stage in which one triangular portion is complete, along with some irregularities. Once the speed is increased from this limit, continuous serration is observed, as in figure 9(b). The inhomogeneity in the chip is visible between the arrows. The formation of triangles is uniform due to high-temperature machining conditions, which are TRS: 1600 RPM; DoC: 0.15 mm; F: 0.12 mm rev⁻¹.

Zoom In Zoom Out Reset image size

It is clear from the analysis that the input parameters plays major role in the deformation of the material. In the primary shear-zone, the shear localized instability is observed in the adjacent segments. With the increase in the cutting speed, a rapid decrement in the contact area is observed.