Wear Resistivity of Al7075/6wt.%SiC Composite by Using Grey-Fuzzy Optimization Technique

The application of SiC particulate reinforcement impact greatly for making aluminium matrix composite because of its superb heat conductivity, oxidation stability and high resistance to mechanical erosion. Present work based on dry sliding wear analysis of Al7075/6wt.%SiC composite fabricated by liquid state stir casting method. To acquire a productive wear rate, three major process parameters viz. load, sliding speed and covering sliding distance were compared at four different levels. ANOVA analysis showed that the probability rate of the load is less than 0.05 that revealed their significant factor. From the study, the highest GRG and GFG values are found 0.914 and 0.854, respectively, for the optimal operating parameters of 10 Newton load, 2 m/s sliding speed and 500-m sliding distance. Finally, it is revealed that the grey-fuzzy technique effectively authenticates the decision making of wear performance characteristics rather than a plain grey relational grade.


Introduction
Wear resistance capacity is a most significant aspect to be considered appropriate for making design elements that secure superb performance for various tribological applications [1]. MMC reinforced with particles reinforcements carried out excellent strength, better wear performance, high hardness and impact strength, which enhanced its application automobile and aircraft industry [2]. MMC obtained the ability to survive heavy tensile and compressive stresses during the transfer and dispersion of the concentrated load from unreinforced matrix to reinforced composite [3]. Compared to steel, aluminium is a good choice for its low density, high corrosion capacity and better mechanical properties. Aluminium is eminently used as a transporting medium for its lightweight and low cost [4][5][6]. Mechanical strength of aluminium alloy enhanced with the addition of some hard particulates reinforcement such as SiC, TiB 2 , B 4 C, Al 2 O 3 , TiC, ZrB 2 , fly ash etc. [7][8][9]. Ceramic particulates filled aluminium metal matrix composites reveals greater wear resistance capacity compared to base alloys [10]. Reinforcement gives preferred s t r e n g t h t o c o m p o s i t e i n a s u p e r i o r d i r e c t i o n . Reinforcement distributed in matrix according to different weight percentage in the formation of whisker or particles [11]. The non-homogeneous distribution of reinforcement particulates enhances the agglomeration behaviour and porosity that decreases the hardness strength of composite material [12]. The stir casting method is usually used fabrication method because of its low cost, ease of operation and flexibility. Stir casting is generally used to melt aluminium, copper etc. Stir casting process preparing composite by mixing matrix and reinforcement at a proper speed [13][14][15]. Stir casting is a satisfactory fabrication method for making composites up to 30 wt.% of reinforcement particles that reduce porosity and uniform distribution of reinforcement. A vital drawback of stir casting is the segregation of filler particles due to the settling of filler particulates during melting and spheroidal grains generated with enhancing stirring time [16]. At the preliminary stage of stirring, few coarse dendrites are observed in microstructure and reinforcement weight percentage, type, and size exhibit minor changes in wear rate of metal matrix composites [17,18]. Reinforcement particle size plays a vital role in developing properties of the composite, generally when particulates are microscale to nanoscale sizes [19]. Prasad et al. [20] investigated the mechanical and wear behaviour of SiC/TiB 2 particle reinforced Al6061 matrix composite fabricated by stir casting. It's noticed that TiB 2 reinforced composite acquire more hardness and wear resistance as compared to SiC reinforced composite. Akbari et al. [21] established the nature of nano and micro-grain size TiB 2 particles reinforced aluminium matrix composites and observed that TiB 2 particles significantly dispersed throughout the molten aluminium matrix. It was concluded that UTS was significantly reduced with enhancing the temperature of casting done, and nanoparticle TiB 2 reinforced composite has better ductility in nature as compared as microparticle. Bhowmik et al. [22] investigate mechanical properties and fracture behaviour of SiC/ TiB 2 micro-sized powder reinforced aluminium matrix composite fabricated by stir casting and shown that TiB 2 reinforced composite has additional strength related to SiC reinforced composite. Poria et al. [23] discussed the wear performance of TiB 2 reinforced Al-TiB 2 aluminium matrix composite and revealed that wear reduced with incorporation of TiB 2 powder. Meti et al. [13] analyze tensile strength and fractography of TiB 2 dispersed AA7075/ TiB 2 composite fabricated by ultrasonic-assisted stir casting. The result observed that TiB 2 particles homogeneously distributed all over the matrix and ductility notably deuterated with enhancing TiB 2 content. Raju et al. [24] reported the tribological performance of Al/CSA composite by applying the grey-fuzzy technique and noticed that wear rate reduced from 5.556 to 1.936 as per grey fuzzy grade. Chamarthi and Nagadolla [25] reported the machining performance of AL6082/SiC/Gr composite by applying a grey-fuzzy approach and revealed that GFG notified best factor combination instead of a simple grey relational grade.
Based on earlier research work, it is observed that very few numbers of experimental work obtained about multi-objective optimization of wear behaviour on SiC reinforced aluminium matrix composite and a huge scope remaining to work in this field because SiC particle reinforced Al7075 matrix composites having a good range of applications in erosive and excessive temperature zone due to the mixture of superb thermal conductivity and low wear rate. SiC particle dispersed Al7075/SiC composite performed as excellent structural applications in the area of aircraft transport, automotive sector caused by its better fusion of less density and superior thermal conductivity. The present research work is to evaluate the effect on SiC reinforced Al7075/6wt.%SiC composite fabricated by a stir casting technique and analyses its dry sliding wear performance with the variation of three different factors and their four different levels by applying grey-fuzzy optimization technique.

Materials
A certain amount (900 g) of Al7075 has chosen for matrix material and 6 wt.% SiC selected for ceramic agents with an average grain size of 20 μm. Chemical compound of aluminium alloy 7075 are Cr:-0.21%, Fe:-0.22%, Si:-0.04%, Mg:-2.58%, Mn-:0.03%, Cu:-1.65%, Zn:-5.75%, Ti:-0.03% and Al-Remaining. Physical properties of SiC particles are: Density-3.21 g/cm 3 , Melting point-2730°C, Purity-99%, Formation-powder, Size-20 μm. Micrograph and EDX elemental mapping of SiC ceramic particles depicts in Fig. 1. Stir casting liquid state fabrication technique used for the present experiment to prepare aluminium matrix composites with the assist of mechanical stirrer which mixes both matrix material and reinforcement particle homogeneously. The schematic representation of the stir casting fabrication setup is shown in Fig.2. A graphite crucible is used for melting the aluminium ingot for composite preparation. After reaching 750°C pouring temperature, the actual weight percentage of titanium diboride poured within the molten matrix material and began stirring for 10 min. Mould and SiC reinforcement particles are preheated at 400°C and 450°C temperature, respectively, in a muffle furnace for reducing porosity, oxide formation, and material shrinkages. After mixing, the composite slurry is poured into a preheated mould and reach room temperature. Casted composite fabricated by stir casting displays in Fig. 3. After preparing Al7075/ 6wt.%SiC aluminium matrix composites fabricated by stir casting, the sectioned specimen was cut, and the surface polished up to mirror finish by using consecutively 400, 600, 800, 1000, 1500 and 2000 graded emery paper and also polished with 0.1 μm diamond paste. The composite specimen was etched by Keller's reactant (solution of 2 ml HF, 3 ml HCL, 5 ml HNO 3 and 190 ml H 2 O). Microstructural analyses were used to observe the microstructural evaluation of aluminium matrix composites and the distribution of the reinforced particulates of silicon carbide throughout the aluminium matrices. Pin-on-disc dry sliding wear runs were done in DUCOM's; TR 20LE-M5 depicts in Fig. 4. Casted specimens were machined to organize 6 mm diameter and 40 mm gauge length pins as per ASTM G99 standard that rubbed within the counter friction plate EN31 steel disc, which has a hardness is 62 HRC.

Design of Experiment
Taguchi's ideology is a powerful tool to optimize the pattern of superior quality aspects. Dr. Genichi Taguchi was invented this technique depend on orthogonal array experiments, which introduce minimum deviation for the experiment with optimal factor setting. Therefore, the combination of the design of experiments according to the optimal combination of the process to acquire appropriate results is established in the Taguchi technique. Based on previous literature [8,26,27], it is observed that the load, sliding speed, and sliding distance was recognized as best process parameters for wear test and assorted with four different levels of each factor shown in Table 1. Putting load quantity (10, 20, 30 and 40 Newton) were chosen to depend on earlier research output to found the influence on the wear rate of casted material. Sliding speed levels (0.5, 1, 1.5 and 2 m/s) were chosen because of the shelter of the friction plane that preserved by the oxide layer. At 3 m/s sliding speed, the contact period in between contact planes is very low that produced a wide oxide layer. Sliding distance levels (500, 1000, 1500 and 2000 m) chosen because the highest limitation of sliding distance was retained 2000 m due to the wear testing machine limitations and after 2000 m sliding distance shown a steady graph. For precision, take four wear rate values of each experiment and find the average result. The standard deviation of wear rate is mentioned in Table 2. This work has been conducted by Taguchi's DOE to decrease the total number of runs. For minimizing the number of experimental runs for three parameters and their four levels, Taguchi's L 16 orthogonal array used shows in Table 3. Frictional force responses are collected from machine data. Wear rate and coefficient of friction for sample pin were measured by Eqs. 1 and 2 as used earlier [9,28].

Grey Relational Analysis (GRA)
According to grey relational analysis, measured readings of response characteristics are being normalized, ranging from 0 to 1. This method called grey relational generation. After normalization, determine the grey relational coefficient to interconnect between different factors and their multiple levels.
Next, find the overall grey relational grade by determining the mean value of the grey relational coefficient. These three steps converted multi-objective optimization to singleobjective optimization [29,30]. The optimal factor setting then analyzes the maximum grey relational grade by applying the Taguchi method.

2.3.1
Step-1 Output raw data have been normalized first (Grey Relational Generation) by using Eqs. 3 and 4 described below [31]. The normalized results for all factors related to wear analysis have been furnished. In this experiment, all responses or outputs such as wear rate, friction force and COF are applicable for smaller is the better criterion.
Where i = 1 to p and m = 1 to k; p represents all experimental runs, and k represents the total number of process parameters. y i (m) represents the original response, max y i (m) and min y i (m) represent the maximum and minimum result of the original output result, and α * i (m) denotes the normalized value after calculating output result attributes.

Step-2
Grey Relational Coefficient (GRC) for all responses of Taguchi's L 16 orthogonal array after data processing or normalizing is calculated by Eq. 5 described below [32].
Where Δ 0i (j) represents the variation of the original responses and η represents the differentiate coefficient 0 ≤ η ≤ 1 and is generally chosen 0.5 because of the equal weightage specified in all factors. Δ min and Δ max represent the minimum and maximum value of data preparing responses.

Step-3
Grey relational coefficient for every value has been collected for analyzing Grey Relational Grade (GRG) that indicates the overall characteristics of wear behaviour. Thus grey relational analysis generally performs for converting single objective   optimization from a multi-objective optimization problem [33]. The maximum values of GRG are too closer to the optimized parameter combination. Grey relational grade lies between 0 and 1. Grey relational grade for all responses of Taguchi's L 16 orthogonal array after data processing is calculated by Eq. 6.

ð6Þ
Where ɣ i represents the overall GRG; m is the total number of output results; ɛ i (k) represents each response value of GRC.

Fuzzy Interface System
Fuzzy logic contains various steps includes fuzzification of input data, fuzzy rule interface, and defuzzification of crisp value [34,35]. In this study, Grey Relational Coefficient (GRC) values are fuzzified using fuzzifier applying membership functions [36]. Steps include the grey-fuzzy approach depicts in Fig. 5. After that fuzzy interface engine executes a fuzzy interface to form fuzzy rules for single fuzzy value instead of three various GRC values. At last, the defuzzifier transforms this single fuzzy value move into GFG. Initially, input response characteristics of GRC values transform into a rhetorical fuzzy set. Triangle type membership functions are applied at the time of fuzzification of three various GRC values by applying small (S), medium (M), and large (L) fuzzy subsets. By using triangle type membership functions, the fuzzifier fuzzified grey relational coefficient dataset ranges between 0 and 1.
Response characteristics of fuzzy interference method are a fuzzy grade ranges 0 to 1, which transformed by using triangle type membership functions into the rhetorical fuzzy subset depicts in Fig. 6. Seven various sorts of subsets are allocated for output results viz. very very low (VVL), very low (VL), low (L), medium (M), large (H), very large (VH) and very, very large (VVH). Figure 7 displays the quantity of fuzzy rule executed during this controller.
According to the fuzzy rule, IF-THEN rules are wont to prepare the boundary report. The fuzzy rule retains a gaggle of IF-THEN formula by applying 81 several sorts of fuzzy logic that describes below, Rule 1: If α 1 is E1 and α 2 is F1 and α 3 is G1 then k is H1 else Rule 2: If α 1 is E2 and α 2 is F2 and α 3 is G2 then k is H2 else Rule n: If α 1 is En and α 2 is Fn and α 3 is Gn then k is Hn else Where α 1 , α 2 , and α 3 are input values of fuzzy logic. En, Fn, Gn and Hn are different fuzzy subsets, and k represents output results. These E x , F x , G x and H x are fuzzy subsets described by the denoting membership functions, viz. μE x , μF x , μG x , and μH x . Multi-objective response fuzzy values alongside membership function denoted as, After that, defuzzification process obtained here to form fuzzy interference output μD i (y i ) [37] into an un-fuzzy value y k , A wider grey fuzzy grade of the upper un-fuzzy value y k gives excellent results.

SEM and EDX Analysis
Microstructural analysis of Al7075/6wt.%SiC composite conducted by Scanning Electron Microscope (Model -Sigma 300, Carl Zeiss) depicts in Fig. 8(a). The solidification process regulates the dispersal of SiC particles in Al7075 matrices. The density difference between the matrix material and reinforcement particles plays a vital role at the stage of solidification. If the reinforcement particle is denser than the aluminium matrix, then start to submerged in the melt and vice versa. Microstructure reveals that silicon carbide particles homogeneously distributed throughout the matrix Fig. 6 Structure of seven fuzzy subsets for GFG Fig. 7 Fuzzy logic membership function for GFG material that obtained better mechanical bonding between matrix and reinforcement due to hard and rigid SiC particles induce shear stress in the melt that boosts split up and mixing homogenization. EDX (Energy Dispersion X-ray spectroscopy) result of aluminium matrix composite reinforced with silicon carbide microparticles is shown in Fig. 8(b). EDX analysis identified some major elements are Al, Si, C, Cr, Fe, Mg, Cu and Zn. EDX spectra curves of SiC reinforced composite indicated the peaks of Si and C that revealed the existence of reinforcement particles in the matrix. EDX basically detects the atomic and weight percentile of each and every element that present in the composite and identified by the formation of peaks.

Grey-Fuzzy Relational Analysis
Matlab statistical software used to treat the Grey-Fuzzy technique for occurring grey fuzzy relational grade. Clarified grey fuzzy grade developed by using the Fuzzy technique. Developed GFG response with minimum untrustworthy has been observed in between fuzzy rule and GRA. Grey relational results using fuzzy make a better qualitative result as compared to plain GRA. By applying Eqs. 3 and 5, GRA and GRG value for all responses are determined displays in Table 4. Triangle type membership functions along three fuzzy subsets for grey relational coefficient and seven fuzzy subsets for response grey relational grade is applied. GFG  values of all experiments and their rank displays in Table 5.
The variation of GRG and GFG value for all experimental runs are shown in Fig. 9. By matching Tables 3 and 4, a modern grading with minimum unpredictability observed for each sample. The main effects on the input parameters of GFG are displays in Table 6. Delta value represents the deviation between the maximum and minimum values. From the main effect, plot depicts in Fig.10, the optimal process control factors have been found. The optimal parameter setting for maximum grey fuzzy grade becomes L 1 S 4 D 1 (10 N load, two m/s sliding speed, and 500 m sliding distance). According to Table 5, it is concluded that sample number 13 was ranked first with a maximum GFG value of 0.854.

ANOVA
The objective of ANOVA is to determine the significant factor from the design of experiment control parameters. Table 7 represents the ANOVA (Analysis of Variance) analysis for GFG. ANOVA is an arithmetical tool that can be applied to identify factors impact. It is noticed that the probability (p) value for load is 0.027, which is lower than 0.05. So it was evident that load is the only significant factor. R-sq and R-sq (adj) values were 99.11% and 97.77%, respectively, which very closer to unity which means the model equation was better fitted with the actual value.

Regression Analysis
Predicted grey fuzzy grade as per stipulated levels of factors by applying regression analysis, a quadratic polynomial regression equation for grey fuzzy grade shown in Eq. 9.    Where L is load in N, S is sliding velocity in m/s, and D is the sliding distance in m. Adequacy of the model was executed by the residual plot of grey fuzzy grade depict in Fig. 11. Variation in between estimated value and equivalence fitted result represent as residual. It is usually used to check model fitting or not. A normal probability plot displays every response closely fitted during a line. Versus fits was shown, the formed regression model was close-fitted. Histogram and versus order shown that the residual points are distributed on the specific side.

Confirmatory Test
Optimized estimated result of GFG (y Predicted ) is measured [38] by using Eq. 10 are as follows. Where y m is the mean result of grey fuzzy grade, y i is the mean of the optimized value of grey fuzzy grade, and "n" is the number of parameters.
A confirmatory test was performed on the optimal process parameter combination of L 1 S 4 D 1 found on the MINITAB statistical tool. The experimental value of the grey fuzzy grade for the optimized parameter combination is lower than the predicted value. A minor percentage error of 2.79% is acquired between the predicted and experimental value that impacts a good correlation shown in Table 8.

Worn Surface Morphology
Worn micrographs of Al7075/6wt.%SiC composite collected from the unlubricated wear test examined by field emission scanning electron microscope. The impact of variation on applied load with different sliding distance and constant sliding speed on the pin surface is required to study and also it has great importance for analyzing wear characteristics depicts in Fig.12(a-d). Figure 12 (a) and (c) depicts several plastic deformations due to the formation of parallel deep grooves. Stress generated during rubbing has maximum sharp edges at higher loads that form cracks in the pin surface. From the worn surface analysis, it's perceived that a slicker layer of pin surface produced with enhancing applied load. As it is seen in Fig.12(a-d), asperities of pin surface make deformed shape at maximum load and formation of the smooth contact layer and due to continuous sliding process, a huge amount of heat form in between the sliding contacts that generate tribo oxide layer [39][40][41]. The microstructure of worn surface of Al7075/6wt.%SiC composite at 40 N applied load with the variation of sliding speed and sliding distance depicts in Fig.12(d-f). Figure 12(f) shows the formation of a mechanically mixed layer and shallow grooves that indicates maximum material removal because of more heat formation and maximum sliding speed. Mechanically mixed layer generally forms due to the combination of oxide surface just after the crucial plastic deformation of pin contact surface at very high temperature, and it is occurred due to high sliding speed [8]. Figure 12(d) and (e) identified the ploughing effect and smaller grooves that generally form due to the stiffness of reinforcement particulates. Smaller grooves show the minimum material removal because of the low pressure generated between the pin surface and counter disc plate [42,43].

Conclusions
Depending upon the experimental value of response characteristics, the Grey-Fuzzy analysis was done to find optimize parameter on the wear rate of composite. Based on these experiments, the following observations can be concluded below: (i) TiB 2 particles reinforced Al7075/6 wt.% SiC aluminium matrix composite successfully fabricated by the stir casting process. (ii) EDX analysis revealed the existence of SiC particulates in the aluminium matrix. (iii) SEM micrograph reveals that SiC particulates were uniformly distributed throughout the composite with minimum agglomeration. (iv) The grey-fuzzy inspection noticed that the optimal parameter combinations of control factors are 10 N load, 2 m/s sliding speed, and 500 m sliding distance. (v) As per residual plots, individual data was observed to follow a far better distribution as every dot was nearer to the line. (vi) Worn surface revealed that abrasive wear and grooves reduce with enhancing applied load. It is also perceived that higher sliding speed formed a mechanically mixed layer.