1 Introduction

In general, composite material (CM) is the term that refers to all solid materials composed of more than one material/element in which the elements are in different phases. High stiffness, strength, and low thermal expansion and weight are the unique features of CM [1]. Metal matrix composites (MMCs) have a broader scope in manufacturing industries due to its strength, hardness, resistance to wear, and corrosion [2]. MMCs are made of either continuous or discontinuous metallic matrix reinforcing phases. Among all reinforcement shapes, particulate metal matrix composites (PMMCs) provide high ductility and poor anisotropy [3]. A wide choice of metals and their alloys is used as matrix elements in MMCs. The most important are titanium, magnesium, copper alloys, and aluminum (Al) alloys. Aluminum alloys are mostly preferred in the industry since they possess excellent mechanical characteristics such as outstanding strength, resistance to corrosion, toughness, and high machinability index [4] with low density.

MMCs are prepared by mixing a reinforcement material into a metal matrix. But the big challenge in casting MMCs is bringing a homogeneous distribution of reinforcement materials in the alloys since it influences the quality and properties of composite materials. Recent studies have shown suitable methods of preparing PMMCs to ensure the proper distribution of reinforcement materials. Powder metallurgy [5], squeeze casting [6], and stir casting [7] techniques have been adapted to produce quality composites. Out of these techniques, stir casting is the most commonly used method for PMMC preparation because it is simple, economical and ensures the uniform distribution of reinforcement particles [8,9,10,11].

Different reinforcement materials have been used such as SiC, TiB2, aluminum oxide (Al2O3), boron carbide (B4C), zirconium oxide (ZrO2), graphite (C), and silicon nitride (Si3N4) to study the machinability characteristics of Al-based PMMCs [10, 12,13,14]. Pawar and Utpat [15] prepared a composite by dispersing SiC particles in Al alloy using a stir casting method. They added the SiC particles by 2.5%, 5%, 7.5%, and 10% into Al alloy to produce three different compositions. Excellent improvement in hardness and toughness was reported for automobile components application. Similarly, Reddy et al. [16] emphasized the importance of surface quality to avoid failures. They investigated the influences of machining parameters on the surface integrity of the components machined in end milling of Al/SiC PMMCs. Though the increasing percentage of reinforcement improves the core mechanical properties, surface roughness and tool life are affected negatively. Hence, a proper selection of cutting parameters should be practiced for getting surface quality of these composites.

Al7075-based PMMCs have gained interest among the researchers since these composites have a wide range of applications in automobile and aerospace engineering. Rajeswari and Amirthagadeswaran [17] investigated the machinability characteristics of Al7075/SiC composites using response surface methodology (RSM)-based grey relational analysis (GRA). The authors studied the effect of machining parameters on surface roughness, cutting force, and material removal rate (MRR). They reported that spindle speed and reinforcement addition are the most influencing parameters. Ramakrishnan et al. [18] experimentally attempted to achieve better machining parameters while machining Al7075/Al2O3 nano-composite using seimens CNC lathe. The significance of experiments was checked using analysis of variance (ANOVA). Further, the technique for order of preference by similarity to ideal solution (TOPSIS) was used to find the best combination of machining parameter in order to obtain better values of material removal rate (MRR) and surface roughness. Bhushan et al. [19] investigated the influence of turning parameters on the surface roughness of Al7075/SiC composites. The relationship between tool wear, process parameters, and resulting surface roughness was reported. Similarly, Kumar et al. [20] performed an experimental work to optimize the machining parameters on the milling of Al7075/SiC/B4C composites in order to obtain a better surface finish. The authors used Taguchi, ANOVA, and GRA techniques to optimize the machining parameters.

Researchers have focused on optimizing machining parameters for machining composite materials by incorporating various decision-making techniques such as GRA, TOPSIS, preference ranking organization method for enrichment of evaluations (PROMETHEE), failure mode and effects analysis (FMEA), and analytic hierarchy process (AHP) [18, 21,22,23,24,25]. Among these techniques, GRA has been widely adapted for selecting optimized machining parameters while machining composite materials [26,27,28,29]. Many researchers have extended applications of GRA and Taguchi techniques for other manufacturing processes also. Tang and Du [30] combined the Taguchi technique and GRA to optimize the process parameters in electric discharge machining. Open-circuit voltage, peak current, inter-electrode gap, pulse on/off time, and width of pulse were considered as process parameters while machining Ti–6Al–4V material. Lin et al. [31] studied a similar work on the same material with the combined application of Taguchi, ANOVA, and GRA. They additionally considered material removal rate (MRR) and over-cut as additional responses. Application of GRA and Taguchi methods have also been found in grinding [32], plasma arc welding [33], plastic injection molding [34], face milling [35], and so on.

From the literature, it is clear that studies related to machinability characteristics on Al7075 reinforced with ZrO2 and graphite are significantly less. The present research work aimed to obtain a combination of optimal machining parameters for a better surface finish, minimum cutting force, and increased MRR on the end milling of Al7075/ZrO2/C PMMC. The following objectives were framed to conduct the experimental study:

  • i) To use the Taguchi technique for designing and experimenting and to use ANOVA for evaluating the results.

  • ii) To identify the optimal machining parameters to concurrently minimize surface finish, cutting forces, and maximize MRR using GRA.

2 Materials and Methods

2.1 Materials

The present work uses Al7075 as a matrix material for preparing MMCs. ZrO2 and graphite are employed as reinforcement materials. The elemental composition of matrix material is shown in Table 1, and its properties are shown in Table 2 [36]. The particle sizes have desirable effects on the mechanical properties of PMMCs. The lesser the particle sizes, the smaller the spaces between them, which contributes to increased mechanical strength and hardness [37]. The density of ZrO2 particles is higher than graphite particles. Hence, the percentage inclusion of ZrO2 is set as 2.5% by weight for all composition, and the average size of ZrO2 particles is limited to 45 µm. The inclusion percentage of graphite particles by weight is varied as 2.5%, 7.5%, and 12.5%, and its size is limited to 15 µm [6]. Stir casting process has been adapted to cast the specimen. Using an induction furnace, aluminum alloy in a crucible is heated up to 750 °C, and the mixture of reinforcement elements is added. Further, the molten alloy mixed with reinforcements is stirred at 500 rpm for about 5 min while the temperature is maintained at 750 °C. Then, the melt is poured into a metal mold of the required dimension and then is taken out after complete solidification [38]. The cast specimens are roughly machined with face milling operation to get a uniform cross section of 38 mm × 27 mm.

Table 1 Elemental composition of Al7075 by weight
Full size table
Table 2 Properties of Al7075
Full size table

2.2 Experimental Design

The challenge associated with experimentation is deciding on the number of runs. The design of experiments (DOE) is a successful tool to reduce the number of runs effectively for solving real-time problems. DOE with Taguchi technique involves specifically built tables called 'orthogonal array' to reduce the experiment runs effectively. Due to this, experimentation becomes economical and effortless. This can also reduce energy consumption and experimentation time [39]. Taguchi and ANOVA are the promising methods due to their perfection in modeling and validation of experimentation and are generally used for linear interactions [40].

In the present work, Taguchi L27 orthogonal array has been used for DOE to analyze how the machining parameter affects the responses of the machining process [41]. The spindle speed (S), feed rate (F), and depth of cut (D) are regarded as input parameters while force components, surface roughness, and MRR are considered as the responses of the milling process. Along with these parameters, the inclusion percentage of reinforcement material by weight is also considered as a parameter. Table 3 provides all the input parameters and corresponding levels for performing experiments. The levels of input parameters such as spindle speed, feed rate, depth of cut, and inclusion percentage have been carefully selected based on the similar works performed on Al-based composites [17, 19, 21].

Table 3 Input parameters and corresponding levels
Full size table
Table 4 Responses measured from experimental work
Full size table

2.3 Experimental Procedure

The experimental flow chart is presented in Fig. 1. The prepared composite materials are initially machined with conventional face milling operation to obtain identical cross section, and they are cut into bars. The composite bars of three different compositions (Al7075 + 2.5%ZrO2 + 2.5%C, Al7075 + 2.5%ZrO2 + 7.5%C and Al7075 + 2.5% ZrO2 + 12.5%C) are machined using vertical milling center L-Mill 55 with maximum spindle speed of 6000 rpm, spindle motor power of 11 kW equipped with Fanuc 0i Mate-MC type CNC system. Microstructures of the prepared specimens are observed, as shown in Fig. 2, and it ensures even distribution of reinforcement particles in all the three compositions. The prepared composites are placed over a piezoelectric-based cutting force dynamometer and milled along the width of the bar, which is about 38 mm. For milling operation, XCELL end mill cutter of 8 mm diameter with four flutes and a right-hand (RH) helix angle of 30° has been used with the help of BT40 end mill tool holder. The entire experimental setup and process flow are represented in Fig. 3.

Fig. 1
figure 1

Experimental flow chart

Full size image
Fig. 2
figure 2

Distribution of reinforcement particles in Al7075

Full size image
Fig. 3
figure 3

Experimental setup and process flow

Full size image

2.4 Measurement of Forces and Surface Roughness

The strain gauge-based milling tool dynamometer has been utilized to measure the force components. The accuracy of the dynamometer is ± 1% from full-scale voltage. It is capable of measuring forces varying from 0 to 5000 N. The machining forces measured during milling operation are acquired and stored in a personal computer with the help of an N1 USB 6221 data acquisition card. The sample signals of 1000 per second are acquired, and noises are filtered with an appropriate filtering process. The cutting forces during milling are measured by clamping the workpiece using a fixture mounted on the dynamometer, which is positioned on the CNC table. The average values in all three axes are taken for analysis after considering the fluctuations in cutting forces during machining.

The column-type Mitutoyo Surf test SV-2100 has been used to measure the surface roughness. The measuring range of the surface roughness tester used in this study is 5 mm, and it is capable of measuring at the speed of 0.25–0.5 mm/s. Since the machining is characterized by end milling operation, the bottom surface of the milled slot is considered for surface finish. The average of 3 measured values is taken for roughness with a cutoff (λc) of 0.8 mm as per the standards ISO 4288-1996 [42].

2.5 Computation of MRR

The MRR of milling operation is estimated using Eq. (1), as shown in Fig. 3 [17].

$${\text{MRR}} = {\raise0.7ex\hbox{${wDd}$} \!\mathord{\left/ {\vphantom {{wDd} t}}\right.\kern-\nulldelimiterspace} \!\lower0.7ex\hbox{$t$}}$$
(1)

Here \(w\) is the width of a sample workpiece in \(\mathrm{mm}\), \(D\) is the depth of cut in mm, \(d\) is the diameter of milling cutter in \(\mathrm{mm},\), and \(t\) is the machining time in \(\mathrm{min}\) for performing a complete slot as shown in Fig. 4.

Fig. 4
figure 4

Milled specimen showing slots along the width

Full size image

2.6 Grey Relational Analysis (GRA) Procedure

The purposes of this work are to improve the Ra, reduce the cutting forces, and MRR by optimizing the input parameters. This is to meet the requirements of both customers and manufacturers like quality needs and improved productivity. The GRA procedure has been applied in this study after performing ANOVA to handle this multi-response problem. First, the responses such as cutting forces (Fx—main cutting force, Fy—feed force, and Fz—thrust force), surface roughness, and MRR are normalized by normalization process using Eqs. (2) and (3) [43]. Since the objective is to minimize the cutting forces and roughness, for smaller the better attributes, Eq. (2) is used. MRR is to be maximized; hence, for larger the better attributes, Eq. (3) is used. The normalized values vary from zero to one, and they are dimensionless values.

$${X}_{nj}=\frac{\mathrm{Max}\left\{{R}_{nj},n=\mathrm{1,2},\dots ,27\right\}- {R}_{\mathrm{nj}}}{\mathrm{Max}\left\{{R}_{nj},n=\mathrm{1,2},\dots ,27\right\}-\mathrm{Min}\left\{{R}_{nj},n=\mathrm{1,2},\dots ,27\right\}}$$
(2)
$${X}_{nj}=\frac{{R}_{\mathrm{nj}}-Min\left\{{R}_{nj},n=\mathrm{1,2},\dots ,27\right\}}{\mathrm{Max}\left\{{R}_{nj},n=\mathrm{1,2},\dots ,27\right\}-\mathrm{Min}\left\{{R}_{nj},n=\mathrm{1,2},\dots ,27\right\}}$$
(3)

where \({X}_{nj}\) is the normalized value of jth response for nth experiment. \({R}_{nj}\) is the actual value of jth response for nth experiment. Using Eq. (3), normalized values of all the responses are calculated, as given in Table 10. The most manipulating trial on optimal responses will score ‘1,’ and the least manipulating trial will score ‘0.’ Further, the determination of grey relational coefficient (GRC) is done with the help of Eq. (4). GRC is calculated to compare the correlation between reference sequence (best sequence or ideal value) and comparability sequence (GRC sequence).

$${\mathrm{GRC}}_{nj}=\frac{{\Delta }_{\mathrm{min}}+\zeta {\Delta }_{\mathrm{max}}}{{\Delta }_{nj}+\zeta {\Delta }_{\mathrm{max}}}\mathrm{ for }n=1, 2,\dots ,27 j=\mathrm{1,2},\dots ,4$$
(4)

where \({\mathrm{GRC}}_{nj}\) is the GRC of nth trial of jth response. \(\zeta\) is the distinguishing coefficient, which is subjected to \(\zeta \in \left[\mathrm{0,1}\right]\). In this study, \(\zeta\) is fixed as 0.5. \(\Delta\) is the difference between the present normalized response to the normalized reference response. GRC of all the trails for all the responses is computed, as shown in Table 11. Equation (5) is used to compute the average of GRCs, and this is termed to be the grey relational grade (GRG). The calculated GRGs of all the 27 trials are shown in Table 12.

$${\mathrm{GRG}}_{n}=\frac{\sum {\mathrm{GRC}}_{nj}}{\mathrm{j}}\quad \mathrm{ for }\quad n=1, 2,\dots ,27$$
(5)

where \({\mathrm{GRG}}_{n}\) is a GRG for nth trial and ‘j’ is the total responses. After the GRGs are calculated, the experiment numbers are graded based on GRG value. The highest GRG values attribute to the optimal input parameters.

3 Results and Discussion

3.1 Experimental Results and ANOVA Analysis

To meet the objectives, 27 experiments are conducted on the workpieces with different levels of input parameters as per the Taguchi design. Three repetitions of measurements for each experiment have been made, and their average is taken for the analysis. The values measured for force components Fx, Fy, and Fz using milling tool dynamometer are shown in Table 4. The values of force components presented in Table 4 are the averaged values acquired from the measurement. Similarly, the surface roughness values and calculated MRR values for all the 27 experiments are shown in Table 4. The main effects plots of the responses as a result of ANOVA are shown in Fig. 5. To find the influences of machining parameters and their interactions, ANOVA has been carried out by setting a 95% confidence level for all parameters.

Fig. 5
figure 5

Main effects plots of responses such as a force component—Fx, b force component—Fy, c force component—Fz, d Ra, and e MRR

Full size image

Tables 5, 6, 7, 8 and 9 show the ANOVA results. It is noted that, in all the cases, the inclusion of graphite particles accounts for the lowest contribution and does not show a significant influence on any of the responses. From Table 5, the next less significant parameter is the feed rate since it accounts for only 16.31% of contribution trailed by speed and depth of cut as 33.72 and 35.28, respectively, on Fx. From Tables 6 and 7, it is noted that the feed rate is the most significant parameters on cutting forces. Feed rate accounts for the contribution of 77.89% on Fy and 82.10% on Fz. Similarly, speed and depth of cut contribute to 1.26 and 7.54 on Fy and 2.20 and 2.73 on Fz, respectively. On the whole, feed rate dominates the other force components by its contribution percentage.

Table 5 Results from ANOVA for the force component Fx
Full size table
Table 6 Results of ANOVA for Fy
Full size table
Table 7 Results of ANOVA for Fz
Full size table
Table 8 Results of ANOVA for Ra
Full size table
Table 9 Results of ANOVA for MRR
Full size table

For surface quality, spindle speed seems to be the most important parameter by its 48.22% contribution on surface roughness followed by the depth of cut with a contribution percentage of 20.70% and feed rate with a contribution percentage of 13.94%. Since the MRR for all the experiments is derived values, the contribution of all input parameters accounts for the same level of contributions as 28.35%. From Table 4, it is noted that higher values of these input parameters lead to higher MRR during the machining process. Main effects plots are predicted to analyze the influence of milling parameters on the responses, as shown in Fig. 5.

3.2 Main Effects Plots

3.2.1 The Effects of Spindle Speed (S)

From Fig. 5a–c, it is observed that spindle speed influences the force components in different ways. For Fx, increasing spindle speed constantly reduces the force component Fx. Since the cutting force (Fx) perpendicular to the direction of tool movement is the main component that leads to cutting action, it is considered to be a predominant force component in milling. But the influence of spindle speed on the other two force components first decreases with spindle speed up to 1000 rpm and increases with further increase in spindle speed. Surface roughness reduces while increasing the spindle speed. The maximum surface quality is obtained at 1500 rpm. Similarly, MRR is peak at 1500 rpm. Hence, from these plots, 1500 rpm could be the significant parameter level.

3.2.2 The Effects of Feed Rate (F)

Feed rate shows a similar trend in all the responses. For all the force components, increasing feed rate values raises the cutting forces. Hence, to decrease the cutting forces, feed rate has to be minimal as possible. From Fig. 5d, surface roughness values are also minimum at a minimal level of feed rate. The effect of feed rate on MRR is shown in Fig. 5e. Maximum MRR is obtained at a feed rate of 0.3 mm/rev. Hence, the most suitable feed rate needs to be selected by considering other input parameters in achieving the objectives of this study.

3.2.3 The Effects of the Depth of Cut (D)

The depth of cut shows a similar tendency to feed rate. From Fig. 5a–c, increasing depth of cut increases the major force component Fx but reduces the other two force components, whereas minimum surface roughness is obtained with a minimum depth of cut of 0.3 mm as shown in Fig. 5d. Similar to feed rate, a higher depth of cut yields higher MRR; hence, the depth of cut also needs to be selected based on interaction analysis with other parameters.

3.2.4 Effects of Reinforcement Percentage

Deviations in means due to reinforcement inclusions by weight are very less compared to the other input parameters. For force components, Fx increases with increasing reinforcement inclusion. This is due to the higher shearing action of hard and brittle particles rather than the plastic deformation of the matrix material. From Fig. 5b, c, Fy and Fz initially decrease up to 7.5% of reinforcement inclusion and further slightly increase when reinforcement percentage is increased from 7.5 to 12.5%. The same trend is obtained when MRR is calculated. Surface roughness also increases with respect to reinforcement percentage. Hence, the lower level of reinforcement percentage improves both forces developed and surface roughness generated compared to MRR as shown in Fig. 5e.

3.3 Inferences from Contour Plots

The two-dimensional contour plots are used to find the optimal levels of each input parameter and to understand the relationship between parameters to achieve maximum responses. The contour plots are the representations of regression equations [44]. Each contour plot represented here provides the effect of different combinations of two input parameters to achieve the maximum response value. Since spindle speed has influence majorly on responses as observed from main effects plots, contour plots are developed between spindle speed and other input parameters. Since interactions of parameters do not show appreciable variations for force components Fy and Fz, only contour plots of Fx are discussed. Figure 6a illustrates the effect of interaction between spindle speed and feed rate on force component Fx. The region with minimum force values is found between higher speed levels and lower feed rate levels. Similarly, from Fig. 6b, it is observed that minimum values of Fx are found at higher feed levels with lower feed rate levels.

Fig. 6
figure 6

Contour plots showing the interaction between a S and F on Fx b S and D on Fx and c S and reinforcement percentage on Fx

Full size image

Figure 6c depicts the influence of spindle speed and reinforcement addition percentage on Fx. The lower values of Fx are found at higher speeds for all the inclusion percentages. At the highest level of inclusion percentage (12.5%) of graphite particles, force values slightly increase but do not significantly vary. From the contour plot, it is clear that the weight percentage of hard particles does not make any significant difference in the force component.

Contour plots from Fig. 7a, b illustrate the interactions of the spindle speed and the feed rate on the surface roughness and MRR, respectively. From Fig. 7a, the region which represents the minimum surface roughness values (i.e., Ra < 0.20) is found at a higher level of spindle speed with lower levels of feed rate. Similarly, from Fig. 7b, maximum MRR is found at higher levels of both spindle speed and feed rate.

Fig. 7
figure 7

Contour plots showing the interaction of S and F on a Ra, and b MRR

Full size image

Figure 8a, b illustrates the contour plots showing the interactions between spindle speed and depth of cut on the surface roughness and MRR, respectively. From Fig. 8a, it is noted that minimum surface roughness values are present in the region where a higher level of spindle speed interacts with the depth of cut. But the region becomes wider with minimum levels of depth of cut. Hence, maximum spindle speed with minimal depth of cut helps to attain minimum surface roughness values. Figure 8b shows the interaction of spindle speed and depth of cut on MRR. Here the maximum MRR is found in the region where a higher level of spindle speed and depth of cut interacts.

Fig. 8
figure 8

Contour plots showing the interaction between spindle speed and depth of cut on a Surface roughness b MRR

Full size image

Figure 9a, b illustrates the interaction of spindle speed and reinforcement percentage on surface roughness and MRR. Minimum surface roughness values are obtained at higher levels of spindle speed through all the reinforcement inclusion percentages. Also, MRR shows a similar kind of region when spindle speed and reinforcement percentage interacts. Hence, the inclusion of hard particles yields the best results to meet the objectives when machined with higher levels of spindle speeds regardless of inclusion percentages.

Fig. 9
figure 9

Contour plots showing the interaction between spindle speed and reinforcement percentage on a surface roughness b MRR

Full size image

3.4 GRA analysis

GRA has been carried out to identify the most optimal levels of input parameters. The response values from experiments are first normalized to the range varying from 0 to 1 using Eq. (3). The normalized values are shown in Table 10. Each response holds a value of 1 if the corresponding experiment is most influential and holds 0 if the experiment is least influential. Other experiments may produce normalized values between 0 and 1 based on their influence levels.

After normalizing the responses, GRCs are calculated using Eq. (4). The computed GRCs for all the responses are then used to calculate the GRGs using Eq. (5). Based on the GRGs, the experiments are ranked. The highest value of grade is ranked as 1, and the lowest value is ranked as 27, and other grades are ranked based on the descending values. The calculated coefficients and grades are shown in Table 11. The grades are also illustrated in the graph, as shown in Fig. 10 for a better understanding of variations in grades.

Fig. 10
figure 10

Variation graph for GRG of all experiments

Full size image

3.5 Wear Behavior of Composites During Machining

Since the responses for different combinations of reinforcement percentages during machining have not shown significant variations, it is necessary to perform analysis on the milled surface microscopically to expose the pattern of metal wear. SEM images of the milled surface are taken at × 500 magnification for all three composites at the optimal level of machining parameters.

Figure 11a–c show the SEM images of machined surfaces for all the three compositions. Figure 11a shows the machined surface of 5% total composition in which narrow grooves with furrows due to plowing effect are visible in the direction of cutting. This plastic deformation happens due to the lower inclusion percentage of reinforcement particles. This smoothens the material flow along the cutting edge and hence results in lower surface roughness value, as illustrated in Fig. 5d.

Fig. 11
figure 11

SEM images of machined surfaces of a Al7075 + 2.5%ZrO2 + 2.5%C b Al7075 + 2.5%ZrO2 + 7.5%C c Al7075 + 2.5% ZrO2 + 12.5%C

Full size image

Figure 11b shows the machined surface of 10% total composition in which sheared particles are visible as a result of increased reinforcement inclusion. This increased shearing action leaves the sharp projections of hard particles; hence, there is a little bit increase in surface roughness, as shown in Fig. 5d. But in 15% of the total composition level, many visible particles with smeared surfaces, as shown in Fig. 11c, reduce the roughness again slightly as in the mean effects plot. From the images, it is observed that increasing hard particles increases the shearing effect and thus results in increased cutting forces, as shown in Fig. 5a.

4 Confirmation Test

A confirmation test has been carried out to check the improvements in the performance characteristics of the end milling process on selected material with optimal parameters combination. Based on GRGs, the optimal level of machining parameters is found at experiment number 19 with the spindle speed of 1500 rpm, a feed rate of 0.1 mm/rev, and a depth of cut of 0.1 mm with graphite inclusion of 12.5%. The same experimental setup with the same milling cutter has been used to perform a confirmation experiment. The levels of parameters set to the optimal combination found from GRA and three repetitions are carried out to have a mean value. The confirmation test results are shown in Table 12. The GRG for the confirmation experiment is calculated using Eq. (6) [17].

$${\mathrm{GRG}}_{\mathrm{predicted}}={\mathrm{GRG}}_{m}+ \sum_{i=0}^{N}{(\mathrm{GRC}}_{0}- {\mathrm{GRG}}_{\mathrm{m}})$$
(6)

where \({\mathrm{GRG}}_{\mathrm{m}}\) is the total mean of GRGs and \({\mathrm{GRC}}_{0}\) is the mean GRG at the optimal parameters level. N is the total number of input variables that influence the objectives. The confirmation test results are compared with initial values at optimal parameter combinations (S3D1F1, i.e., spindle speed of 1500 rpm, a feed rate of 0.1 mm/rev, and depth of cut of 0.1 mm) as shown in Table 12. The difference between the initial and test experiment values does not radically vary. On the whole, only 2.02% of error is noticed in GRG. Hence, test experiment confirms the validation of performance measures.

Table 10 Normalized values of responses
Full size table
Table 11 GRCs for normalized responses and grey ranking
Full size table
Table 12 Test results comparison
Full size table

5 Conclusion

The present study investigated the influence of milling parameters on the machinability characteristics and quality of the surface produced. The experimental work is performed on a composite material Al7075 reinforced with 2.5% of ZrO2 and varying graphite percentages developed through the stir casting technique. Taguchi L27 design and ANOVA techniques are used to check the significance of experiments conducted and to analyze the responses. GRA is performed to ascertain the optimal levels of input parameters using the response values.

Based on the analysis, the optimal combination of milling parameters has been found to concurrently reduce the cutting forces, surface roughness, and increase the MRR. The optimal combination of end milling parameters is found as spindle speed of 1500 rpm, a feed rate of 0.1 mm/rev, depth of cut of 0.1 mm, and 2.5% ZrO2 + 12.5% of graphite by weight. The confirmation test has validated the optimum milling parameters with less than 5% error. Analysis from the ANOVA reveals that spindle speed and feed rate are the most influencing parameters, respectively, for surface roughness and cutting forces. Spindle speed is the most influencing parameter for force component Fx also. From the contour plots, interactions of input parameters are analyzed and found that higher level of spindle speed with a lower level of feed rate and depth of cut helps to attain the objectives more closely. In contrast, the inclusion percentages of reinforcement particles do not have any significant effect on the responses.

6 Limitations and Scope for Future Work

The work has been carried out on the Al alloys of different combinations using end milling operation. Different tool materials with varying sizes of tools may be used in the future to perform the same analysis to generalize the results of this study.

Only reinforcement percentages are varied in this work but not the particle sizes. Future works may include the analysis using different reinforcement particle sizes to check their influences on the responses.