1 Introduction

Among different steel grades, AISI 304 grade holds excellent corrosion resistance and hence employed in wide variety of applications such as vaporizer coils, kitchen sinks, food handling units, chemical processing units, automotive exhausts, cooking pans, pressure vessels and mining materials handling equipment. Machining of AISI 304 stainless steel is quite tedious process, due to its high strength and higher work hardening rate. This might lead to shorter tool life, high tool wear, higher machining time and machining cost. The tool life phenomenon and unpredictable cutting forces have direct impact on overall machining cost and dimensional accuracy of machined work piece. Several attempts were carried out in reducing the cutting forces with the view of achieving better machinability (Ref 1, 2).

Super hard thin-film surface coating on cutting tool inserts makes them more resistant to abrasion and oxidation. These coatings also act as solid lubricant at the tool–workpiece interface, which led to a decrease in temperature at cutting zone. Various materials were utilized as coating material for cutting tool inserts, among them titanium aluminum nitride-coated tool inserts exhibit better machinability than titanium nitride-coated tools (Ref 3,4,5). These surface coatings indirectly improve the working life of tool and reduce tool wear especially at increased machining speed (Ref 4, 6). As compared to single- and double-layer surface films, multilayer films experience enhanced hardness and reduce the possibility of tool failure (Ref 5, 7,8,9,10).

The machining parameters such as feed rate, cutting speed and depth of cut also influence the degree of machining in addition to surface coatings of tool insert. However, further optimizing the machining parameters will surely increase the tool life and quality of machining (Ref 11,12,13,14,15,16,17,18). It was indicated that cutting speed has greatest influence on cutting force, cutting temperature and tool life but feed rate has a significant influence on surface roughness. Furthermore, titanium oxide was present on the worn surface, which helps to increase the hardness of cutting zone, lower the friction coefficient and act as an insulator. As a result, the wear rate is reduced (Ref 10, 19). The machining parameters were optimized in turning of AISI 316L steel for attaining maximum material removal rate and surface roughness (Ref 20). For optimizing tool wear, cutting pressure, surface roughness and material removal rate, Taguchi technique and grey relational analysis were found to be more effective (Ref 21). Adoptive modeling was implemented for the prediction and validation of input to output relation with experimental outputs during turning of AISI 202L steel (Ref 22). The multi-response optimization of parameters was performed with the combined effect of Taguchi, regression analysis, response surface methodology and grey relational analysis (Ref 23,24,25). The optimal machining parameters on wet machining of F55 steel were achieved through Taguchi-grey relational analysis (Ref 26).

During machining of AISI 304 steel, the influence of cutting velocity on various output parameters such as tool wear and surface roughness was premeditated (Ref 27). The effect of various coolants during machining of AISI 304 steel on cutting tool performance was examined by Xavior (Ref 28). A technique of sound processing was incorporated in evaluating the optimized machining parameters such as surface roughness, formation of chip, built-up-edge formation and flank wear (Ref 29). Through design of experiments, the most reliable technique for optimizing various engineering problems is found to be Taguchi method (Ref 18, 30, 31). Design of experiments was most helpful in recognizing and optimizing machining parameters and predicting optimal combinations (Ref 32,33,34,35). Multi-output response optimization can be easily carried out with various techniques such as grey relational analysis, data evaluation and ranking, genetic algorithm, response surface methodology and artificial neural network (Ref 1, 9, 36,37,38).

Optimizing multiple output parameters with the view of attaining maximum performance was evaluated through various advance optimization tools such as artificial intelligence, grey relational analysis and fuzzy logic. The fuzzy logic might be helpful in addressing the indefinite and uncertain experimental values (Ref 39). For achieving effective weldability, fuzzy logic approach was applied for obtaining optimal welding parameters (Ref 40). Application of fuzzy logic approach with the grey relational analysis shall enhance degree of machining parameter optimization with the view of surpassing multi-response issues (Ref 41). Many research works were carried out in elevating the cutting tool performance through surface engineering of tool inserts. From various studies, it is evident that turning parameter optimization and input parameter influence over output machining responses were extensively explored. In addition to this, certain optimization techniques are found to be more effective in determining the optimal solution of both single and multi-response problems such as Taguchi technique, fuzzy logic and grey relational analysis (Ref 1, 32, 33, 37, 41,42,43).

Although numerous optimization tools are used to optimize manufacturing process parameters, there is still a high expectation of a simplified and more efficient multi-criteria decision making methodology (MCDM). The grey fuzzy technique in connection with the same method has shown itself to be efficient for the simplest methodology and similar results in accordance with the output values derived from other optimization tools. Several researchers used this approach to optimize various machining operations. Since the use of grey fuzzy-based Taguchi technique to optimize machining parameters during dry machining of AISI 304 steel was unexplored, this research study addresses the research gap by studying the same by employing the coated TiAlN, AlTiN and TiAlSiN tungsten carbide (WC) inserts. Moreover, the research study also extended in employing a simplified optimizing tool, namely grey fuzzy coupled Taguchi technique, for achieving appropriate machining process parameters in an efficient manner. The aim and objective of the research work is to deposit hard TiAlN, AlTiN and TiAlSiN coating on cutting tool insert and to perform multi-criteria decision making for identifying best optimal parameters in dry turning AISI 304 steel through grey fuzzy coupled Taguchi technique.

2 Experimental Detail and Methodology

2.1 Hard Coating on Cutting Tool

In this research study, uncoated tungsten carbide CNC cutting tool inserts were procured from Usman Tools, Coimbatore, Tamil Nadu. The insert is in the form of 4 cm diamond bar and width of the cutting-edge is 12 mm. The negative rake angle inserts have 8 mm nose radius (80° diamond sized insert) and conform to ISO CNMG 120408 standard geometrically. The hardness of uncoated insert was found to be 19.1 GPa.

Thin-film TiAlN, AlTiN, and TiAlSiN coatings were adhered to the surfaces of uncoated tungsten carbide turning tool inserts using commercially available high-power impulse magnetron sputtering system (HiPIMS) at Famex Coating India Pvt. Ltd., Coimbatore, Tamil Nadu, as shown in Fig. 1. The sputtering targets such as TiAl, AlTi, and TiAlSi were procured from Sigma-Aldrich (99.99% purity). The operating conditions for the coating deposition are listed in Table 1. The deposition was carried out in a hybrid deposition setup at a base pressure of 2 x 10−3 Pa (Ref 44). The operational parameters for deposition such as bias voltage, pressure and supply voltage were set to 400 V, 0.8 Pa and 950 V, respectively.

Fig. 1
figure 1

Schematic diagram of HiPIMS coating process

Full size image
Table 1 HiPIMS deposition parameter
Full size table

To ensure high level of cleanliness, all uncoated inserts were manually polished and ultrasonically cleaned prior to deposition. The substrates were cleansed for another 30 minutes using glow discharge at 1000 V substrate bias voltage and 1.5 Pa argon pressure. The high purity argon and nitrogen gas (99.99 percent) were utilized, and the temperature for deposition was regulated at 510 °C. The deposition chamber comprises 99.99% pure TiAl (Ti: 0.60 Al: 0.40), AlTi (Ti: 0.40, Al: 0.60) and TiAlSi targets (Ti: 0.64, Al: 0.3, Si: 0.06), which are mounted on the internal walls of deposition chamber in rectangular form. In HiPIMS coating technique, nitrogen was used as reactive gas, which has the flow rate of approximately 94 sccm. Nitrogen ion implantation was exploited to strengthen adhesion between coating and substrate before deposition. The substrate holder was continuously rotated through the stepper motor between the sputtering targets. The substrates were rotated three cycles to aid in the uniform distribution of depositing flux throughout the surface of substrates. Thin-film TiAlN, AlTiN, and TiAlSiN layers were sputter-coated on cutting tool inserts by employing the process parameters as listed in Table 1 (Ref 45). The photographic images of coated tools are shown in Fig. 2. The coated inserts were firmly fixed to ISO PCLNR 2525 M12 assigned tool holder for Jobber XL CNC turning center as shown in Fig. 3.

Fig. 2
figure 2

Images of coated tools

Full size image
Fig. 3
figure 3

Machining in CNC lathe

Full size image

2.2 Characterization of Coated Inserts

The coated tools were sectionally divided into two halves by wire cutting EDM machine to measure the thickness of coating. In addition to sample cleaning and assembly, coated tools require little special preparation. When scanned through the electron beam, specimens get charged and in turn create scanning error and other picture artifacts, especially in the secondary electron imaging area. Before the analysis of FE-SEM, a thin layer of conductive gold was sputter coated to the external surface in order to evade the effects of scanning errors. By exposing the cross section through Carl Zeiss’ MERLIN Field Emission Scan Electron Microscope (FE-SEM, Model: ZEISS EVO60), the thickness of coated tools was determined. X-ray diffraction (XRD, D8 Advance) with Cu K radiation (0.15406 nm) was performed at 40 kV and 40 mA with a fixed incidence angle of 1° and scanning from 30° to 90° with a step size of 0.02°. A scratch tester was used to determine the adhesion strength between the coating and the substrate (CSM, RST). Scratch tests were performed at a maximum force of 100 N and a scratch length of 3 mm. The microhardness of TiAlN, AlTiN, and TiAlSiN coatings was measured by Vickers micro hardness tester (Wilson 400 series), which was guided by ASTM E384 Standard. The microhardness tests were carried out with an applied load 1 kg, and an indentation time of 10 s. Final microhardness result was calculated by taking average of 5 measurements on each specimen.

2.3 Workpiece

In this research work, AISI 304 SS rods of 40 mm diameter were procured from Usman Tools, Coimbatore, Tamil Nadu. The chemical composition of AISI 304 austenitic stainless steel is depicted in Table 2.

Table 2 AISI 304 steel—chemical composition
Full size table

The long SS rod was cut into 9 small pieces of length 100 mm using cutting machine, and its end was faced using manual lathe as shown in Fig. 4.

Fig. 4
figure 4

AISI 304 steel work piece

Full size image

2.4 Sustainable Hard Machining of AISI 304 SS

The turning experiments were carried out in dry conditions on a circular rod of AISI 304 austenitic stainless steel. As shown in Fig. 3, the experiments were carried out in computer numerical control (CNC) Lathe machine (Model: JOBBER XL). TiAlN-, AlTiN-, and TiAlSiN-coated CNC turning inserts specified in Table 3 were used. The tests were conducted three times for each cutting speed.

Table 3 Machining data
Full size table

The experiments were designed based on the L9 orthogonal array to perform optimization study. Surface roughness (Ra and Rz), wear rate and MRR were considered as the output responses. Each experiment is performed three times, and the mean value of the output results was reported. To determine surface roughness, Mitutoyo surface tester with a travel length of 2.54 mm in the X-axis and a velocity of 0.05 cm/s was used. Wear rate was measured based on the mass difference of coated CNC tool insert before and after machining. Material removal rate was measured based on the mass difference of work piece before and after machining.

2.5 Multi- Criteria Decision Making-GFCT

In order to solve the simultaneous optimization of four quality characteristics such as Ra, Rz, wear rate and MRR, MCDM-based Grey fuzzy coupled Taguchi (GFCT) technique was employed. The multiple responses were first optimized by utilizing GRA (grey relational analysis) and then followed by fuzzy logic approach. Results of the fuzzy relationship grade have been optimized using S/N analysis in Taguchi single response optimization approach. The level, which has highest value for average grey fuzzy grade, is the optimum ideal condition for the output responses. ANOVA has been carried out to determine the influence and contribution by each parameter on multiple responses. Confirmation experiments were carried out to validate the optimal environment. The representation of proposed MCDM-based GFCT method is depicted in Fig. 5.

Fig. 5
figure 5

copyright 2020 (Ref 2)

Grey fuzzy coupled Taguchi method. Reprinted by permission from Springer Nature Customer Service Centre GmbH: Springer, Structural and Multidisciplinary Optimization, Achieving machining effectiveness for AISI 1015 structural steel through coated inserts and grey-fuzzy coupled Taguchi optimization approach, C. Moganapriya et al.,

Full size image

2.5.1 Grey Relational Analysis

GRA was selected to multiple such as Ra, Rz, wear rate and MRR. The optimization process in GRA constitutes three phases.

  • Normalization

  • Computation of grey relational coefficient

  • Grey relational grade

2.5.1.1 Normalization

First step in GRA is to standardize the measured data in order to reduce difference, which is referred to data pre-processing. It is normally articulated as a range, and the unit of response varies from one individual to the next. Pre-processing of data is a better strategy for converting all original series into a relative one. The data pre-processing methods used for grey relational evaluation are as follows:

  • To normalize the initial data sequence of the measured data for “higher-the-better” features (MRR), the following equation was adopted.

    $$ x*j = (x^\circ j\left( k \right) - \min x^\circ j\left( k \right))/(\max x^\circ j\left( k \right) - \min x^\circ j\left( k \right)) $$
    (1)

  • The “lower-the-better” features of measured data (Ra and Rz, wear rate) were normalized through the formula.

    $$ x*j = (\max x^\circ j\left( k \right) - x^\circ j\left( k \right))/(\max x^\circ j\left( k \right) - \min x^\circ j\left( k \right)) $$
    (2)

  • If a specific target value (preferred) is to be met, the original data set will be standardized using the following formula:

    $$ x*j = 1 - (\left( {x^\circ j\left( k \right) - x^\circ j} \right)/(\max x^\circ j\left( k \right) - x^\circ j\left( k \right))) $$
    (3)

x°j(k)—original set of data, x*j(k)—next set to data pre-processing, max x*j(k)—maximum of x°j(k), min x*j(k)—minimum of x°j(k), x°j—expected value.

2.5.1.2 Calculation of Grey Relational Coefficient

The grey contextual evaluation is used to determine the relevance of two structures. A grey relational coefficient is used to describe the sequences. It could be calculated as (k):

$$ \xi \left( k \right) = \left( {\Delta \min + \xi \Delta \max } \right)/\left( {\Delta {\text{oj}}\left( k \right) + \xi \Delta \max } \right)) $$
(4)

Δoj(k)—absolute quantity of deviation from x°j(k) and x*j(k) (series of deviation);

ξ—coefficient of distinguishing=0.5.

The distinguished potential would be higher if the magnitude is smaller.

$$ \Delta oj\left( k \right) = \left| {\left| {x*o\left( k \right) + x^\circ j\left( k \right)} \right|} \right| $$
(5)
2.5.1.3 Grey Relational Grade (GRG)

The grey relational grade was calculated by the average of coefficients.

2.5.2 Grey Fuzzy Relational Analysis

The following four steps as part of the fuzzy rule system are shown in Fig. 5 (Ref 46).

  • Fuzzification of input data

  • Determination of rule base

  • Decision making based on rule

  • Defuzzification of data

Membership functions are established via database that are used to produce fuzzy rules. The implications of the developed rules are obtained by a decision-making unit. The fuzzy interface then turns the input into linguistic items based on their corresponding degree. The defuzzifing unit then transforms the fuzzy outputs into crisp result (Ref 46, 47). The generation of fuzzy rules is governed by if-then control principles. Table 4 portrays fuzzy technique for 2-input and 1-output system (Ref 46, 47 ).

Table 4 Fuzzy rules
Full size table

Xi, Yi and Zi are fuzzy subsets that are clearly derived through their membership, for example μXi, μYi and μZi. Table 4 shows the fuzzy rule that if the inputs (A1, A2) are X1, Y1, the output B becomes Z1 and the same n number of rules can be created. The Mamdani fuzzy inference engine employs fuzzy logic technique to develop crisp results through fuzzy rules.

2.5.3 Analysis of Variance

The signal-to-noise (S/N) ratio is an efficient analysis tool for the Taguchi method to depict a quality characteristic, and higher S/N value stands for the preferred process parameter level. GFRG results were feed as single response and optimized by Taguchi method. Larger the better condition was selected for analysis grey fuzzy grade. ANOVA was used to estimate the contribution of individual parameters for selected inputs on the determined output responses. ANOVA observations could be used to determine the responsible parameters for the specified process output and can analyze parameters for best outcomes (Ref 47).

3 Results and Discussion

3.1 Characterization of Coating

The cross sections of TiAlN-, AlTiN- and TiAlSiN-coated inserts are represented in Fig. 6. The deposition of coating on cutting tool is clearly evident in the cross-sectional SEM image. From Fig. 6, coating thicknesses were measured using Image J software. The average thickness of TiAlN, AlTiN and TiAlSiN coatings is equivalent to 22.81 μm, 27.71 μm and 30.08 μm, depending on the cross section correspondingly as listed in Table 5. Figure 7 depicts the XRD patterns of deposited TiAlN, AlTiN and TiAlSiN coatings. TiAlN and AlTiN coating reveals the crystal structure of NaCl. However, there was no nitride phase observed regarding TiAlSiN coating, which is owing to the low content of Si that could replace Ti atoms in TiN lattice (FCC) and forms metastable phase of Ti(Al Si)N (Ref 48). It may also result in the formation of amorphous Si3N4 accumulated at the grain boundaries of nanocrystalline TiAlN (Ref 49). TiAlN and AlTiN have adhesion strengths of 45 N, 47 N, and TiAlSiN coating has a greater adhesion strength of 53 N as compared to other coatings.

Fig. 6
figure 6

Cross section FESEM image of coated tools (a) TiAlN, (b) AlTiN and (c) TiAlSiN

Full size image
Table 5 Properties of deposited coating
Full size table
Fig. 7
figure 7

XRD of base and coated inserts

Full size image

Table 5 indicates an excellent bonding between the substrate surface and the deposited layer, thereby improving adhesive strength. This shows that physical trapping and mechanical locking have been developed and play a significant role in bonding TiAlSiN coating with tool substrate (Ref 49).

Figure 8 portrays the hardness of as-deposited TiAlN, AlTiN and TiAlSiN coatings. It was evident for Fig. 8 that TiAlSiN-coated insert possesses higher hardness of 33.67 GPa as compared to TiAlN- and AlTiN-coated inserts. The augmented hardness of TiAlSiN-coated insert could be linked with the development of nanocomposite structure consisting of nc-TiAlN embedded in a Si3N4 amorphous matrix that can refine grain structure (Ref 50). Furthermore, solid solution hardening can also contribute to improved hardness. The enhanced hardness of the TiAlSiN coating can be due to the combined solid solution hardening and grain boundary refinement (Ref 48).

Fig. 8
figure 8

Variation of hardness

Full size image

3.2 MCDM - GFCT Method

The output responses like Ra, Rz, wear rate and MRR were measured and are listed in Table 6 based on the designed array. Each experiment was repeated three times in order to evade the inaccuracy of measurements, and the average value is reported in Table 6. In order to optimize several responses like Ra, Rz, wear rate and MRR, MCDM technique has been selected. At first, GRA has been employed in which the measured output responses (surface roughness—Ra, Rz and tool wear) were preliminary normalized using Eq 2, whereas MRR was normalized using Eq 1. Subsequently, absolute values were estimated. By substituting normalized values for all the responses in Eq 5, grey relation coefficients were obtained as listed in Table 6. The average of all the grey relational coefficients was determined, and it is termed as grey relational grade (GRG). The experiments had been ranked based on calculated GRG. From the table, experiment 8 - A3B2C1D3 (TiAlSiN, cutting speed: 180, feed rate: 0.1 and depth of cut: 1.5) attained highest rank and it was found to be the best combination of input parameters for the outputs. Moreover, the weight assignments for individual responses in GRA may result in inappropriate values. The grey fuzzy relational grade (GFRG) was subsequently implemented to mitigate these consequences. The machining parameters were set as input, and GFRG was considered as output for fuzzy system as depicted in Fig. 9. Linguistic membership functions such as low, medium and high were employed to define the input variables. Also, output of the fuzzy logical system was set by averaged grey grade, and the performance of related grey grade is represented by their attributes as depicted in Table 7. In this investigation, the triangle membership function was utilized for input and output variables. A series of fuzzy guidelines were allocated based on rules as shown in Table 4.

Table 6 Record of experimental values and GRG
Full size table
Fig. 9
figure 9

GFRG system—Mamdani approach

Full size image
Table 7 Intervals for sorting grade
Full size table

Maximum and minimum synergistic action is generated by managing the fuzzy logic rules. In the end, the defuzzifier translates the anticipated fuzzy outputs into GFRG using fuzzy logic toolbox in MATLAB (R2016b). Table 8 encapsulates the acquired GFRG data and sample outputs from GFRG.

Table 8 Comparison of GRG and GFRG
Full size table

As described in Table 8, the optimum parameters were perceived for experiment number 8 - A3B2C1D3 (TiAlSiN, cutting speed: 180, feed rate: 0.1 and depth of cut: 1.5). The relative assessment for all GRG investigations and the accompanying GFRG is illustrated in Fig. 10. In comparison with GRG, the GFRG is enhanced by 3.53%, which reduces fuzziness. The results are consistent with the previous work (Ref 2, 51).

Fig. 10
figure 10

Relative evaluation of GRG and GFRG

Full size image

3.3 S/N Analysis for GFRG

The data from GFRG analysis were considered as the output response for selected process parameters, and it was optimized by Taguchi method. Since it is the most appropriate tool for optimizing single responses, augmented GFRG was optimized through Taguchi’s S/N analysis.

The optimum range of process variables is the highest ratio of S/N (Ref 52). The impact of every input factor was shown more clearly in the response graph of S/N ratio. As depicted in Fig. 11, the feed rate and coating have a substantial influence on GFRG and its S/N ratios. These findings are consistent with the earlier report (Ref 33). From the S/N ratio plot, the parameter setting—A3B2C1D3 (TiAlSiN, cutting speed: 180, feed rate: 0.1 and depth of cut: 1.5)—was estimated as the best optimal parameter for GFRG and in turn for the output responses.

Fig. 11
figure 11

Mean effects plot of S/N ratio

Full size image

3.3.1 ANOVA for GFRG

The percentage contribution for GFRG was estimated, and it was observed that feed rate contributes 42.74% on the measures of GFRG as shown in Table 9. It is the most influencing parameter, which predominantly determines the value of output responses. Followed by feed rate, coating material contributes about 32.05%, whereas cutting speed and depth of cut nominally influence GFRG by 11.97 and 13.90%, respectively. From the results of S/N ratio and ANOVA, feed rate and coating have played a substantial role in enhancing the GFRG, thereby minimizing wear rate, Ra, Rz and increasing MRR. The R2 value of ANOVA for regression of GFRG was found to be 93.47%.

Table 9 Analysis of variance for regression of GFRG
Full size table

It is attributed to the higher hardness of surface coating (TiAlSiN: 33.67 GPa) as illustrated in Fig. 8. The findings are in line with the Arcades rule, which stipulates that wear is exactly proportionate to the hardness of cutting tools (Ref 53). Regression equations of GFRG for TiAlN-, AlTiN-, TiAlSiN-coated inserts were determined as listed in Eq 6, 7, 8.

$$ {\text{TiAlN}}:{\text{GFRG}} = 0.605 \, + 0.00012 {\text{Cutting}}\,{\text{speed }} - 1.44 {\text{Feed}}\,{\text{rate }} + 0.0164{\text{ Depth }}\,{\text{of}}\,{\text{cut}} $$
(6)
$$ {\text{AlTiN }}:{\text{GFRG}} = 0.639 \, + 0.00012 {\text{Cutting}}\,{\text{speed }} - 1.44 {\text{Feed}}\,{\text{rate }} + 0.0164 {\text{Depth}}\,{\text{of}}\,{\text{cut}} $$
(7)
$$ {\text{TiAlSiN }}:{\text{GFRG}} = 0.697 \, + 0.00012 {\text{Cutting}}\,{\text{speed }} - 1.44 {\text{Feed}}\,{\text{rate }} + 0.0164 {\text{Depth}}\,{\text{of}}\,{\text{cut}} $$
(8)

Figure 12 depicts 3D surface plots of GFRG with respect to cutting speed, feed rate and depth of cut. From Fig. 12a, GFRG is higher at higher depth of cut with respect to lower and higher cutting speed. Feed rate at lower level tends to produce higher GFRG as shown in Fig. 12b. It possesses higher value at lower feed rate for low and high cutting speed. Figure 12c portrays the relation between feed rate and depth of cut with reference to GFRG. With higher depth of cut, GFRG possess higher values with respect to low and high feed rate.

Fig. 12
figure 12

3D surface plot of GFRG vs. input parameters

Full size image

3.4 Confirmation Test

An experiment on the optimum combination (A3B2C1D3) of parameters was conducted to validate the outcomes of MCDM method (TiAlSiN, cutting speed: 180, feed rate: 0.1 and depth of cut: 1.5). The results were measured and summarized as presented in Table 10.

Table 10 Confirmation test
Full size table

The next step following the identification of optimal process variables is to forecast and assess the improvements in performance of optimum process parameters. Table 10 shows a comparison of several objectives with initial and optimal processing parameters. The initial level of optimal machining parameters from GFRG and GRA is A3, B2, C1 and D3, and it matches with experiment number 8 in Table 6. The optimum results were predicted, and the anticipated test results with an average error of 3.54% and an accuracy of 96.5% are validated.

These findings substantiate and validate the adopted MCDM–GFCT method. Multiple objectives have been optimized simultaneously in the turning process, and this strategy is clearly displayed. These results are consistent with the prior investigations (Ref 51).

4 Conclusion

Hard TiAlN, AlTiN and TiAlSiN coatings were successfully deposited on the surface of cutting tool insert through HiPIMS system. As-deposited coatings were confirmed by XRD analysis, and their coating thickness was determined by FESEM. The average thickness of TiAlN, AlTiN and TiAlSiN coatings was equivalent to 22.81, 27.71 μm and 30.08 μm. From the results of hardness measurements, TiAlSiN-coated insert exhibits higher hardness of 33.67 GPa. MCDM technique has been adopted for optimizing multiple responses like Ra, Rz, wear rate and MRR. The measured data were normalized and ranked through GRA, and the same has been fuzzified through GFRG. From ANOVA results, it was found that feed rate and coating have played a substantial role in enhancing the GFRG, thereby minimizing wear rate, Ra, Rz and increasing MRR. It contributes 42.74% on the measures of GFRG, followed by coating which contributes about 32.05%, whereas cutting speed and depth of cut nominally influence GFRG by 11.97% and 13.9%, respectively. The optimized parameters of MCDM–GFCT are TiAlSiN-coated insert, cutting speed: 180 m/min, feed rate: 0.1 mm/rev and depth of cut: 1.5 mm. The confirmation experiment has been conducted, and the results of MCDM were validated with a prediction accuracy of 96.5%. These findings confirm and substantiate the implemented MCDM–GFCT method.