 Open Access
 Total Downloads : 1037
 Authors : Brintha. N. C, Shajulin Benedict, Winowlin Jappes. J. T
 Paper ID : IJERTV2IS60304
 Volume & Issue : Volume 02, Issue 06 (June 2013)
 Published (First Online): 10062013
 ISSN (Online) : 22780181
 Publisher Name : IJERT
 License: This work is licensed under a Creative Commons Attribution 4.0 International License
Machining Parameter Optimization Of Al/Sic p Composite Materials Using Genetic Algorithm
Brintha. N. C1, Shajulin Benedict2 and Winowlin Jappes. J. T3*
1 Department of Computer Science, PonJesly Engineering College
2Department of Computer Science and Engineering, St.Xaviers Catholic College of Engineering
3Department of Mechanical Engineering, CAPE Institute of Technology Tamilnadu, India
Abstract: Composite materials have received potential applications in different fields. The major problem encountered during the machining of composite materials is the flank wear and surface roughness. For this study, metal matrix composite material namely Al/SiCp was considered. Mathematical modeling and experimentation results on the effect of wear and surface roughness were presented. In this work, genetic algorithm has been used for machining parameter optimization. On comparing the results produced from genetic algorithm model and mathematical modeling, the results of genetic algorithm model were found to be more accurate and nearer to the experimental values.
Keywords: Metal matrix composites, Machining Parameter, Optimization, Genetic Algorithm.

INTRODUCTION
Particulate reinforced Al metal matrix composites is one which have SiC particles with aluminum matrix. These composites are rapidly replacing conventional materials in various automotive, aerospace and automobile industries [Allison et al. (1993)]. However, usage of composites in those applications is limited because of the difficulties in the machining [Cronjager et al. (1992)]. The presence of hard ceramic particles makes them difficult to machine as they lead to rapid tool wear [Davim et al. (2000)]. The machining parameters like speed, feed and depth of cut are influencing more in the tool wear and the surface roughness [Manna et al. (2003)]. Monaghan et al. (1994) studied the wear behaviour of carbide tools during the
machining of 25% SiC at speed below 20 m min1. Tomac et al. (1992) developed a tool life relationship for carbide tools during the machining of Al/SiC at speed lower than 100 m min1. Ibrahim et al. (2004) concluded that the percentage volume of SiC and cutting speed are found to be the major factors affecting the tool wear. Gallab et al. (1998) stated the tool wear has direct relationship with cutting speed and indirect relationship with feed. In addition to the tool wear, machining produces poor surface finish due to build up edge formation [Manna et al. (2002)]. Further, Manna et al. (2005) recommended the machining parameters of high speed, low feed rate and low depth of cut for achieving better surface finish. Even though various researchers focuses on the parameters studies on tool wear and surface roughness, effective optimization for the proper selection of parameters based on the required output need to be attended.
Hence in this investigation, in addition to the mathematical modeling, evolutionary optimization technique Genetic Algorithm (GA)is used for optimizing the machining parameters for tool wear and surface roughness. Many researchers have studied the effects of cutting tools and machining parameters on tool wear for example, Chou et al. (2002), Ozel et al. (2005) and many others.
Solving multiobjective problems with traditional optimization methods is very difficult and time consuming. Therefore evolutionary algorithms such as genetic algorithms and particle swarm optimization are more convenient and usually utilized in multi objective optimization problems. These methods are summarized by Zitzler et al. (2003). As for applications of evolutionary algorithms in solving multiobjective optimization problems in machining, studies by Cus and Balic (2003) on metal cutting optimization, Saravanan et al. (2002) on grinding optimization, and Karpat and Ozel (2005) on hard turning optimization, among others, have contributed more on this area.
Even though various researchers focused on the composite materials, a detailed study which combines both theoretical as well as experimental view on process parameters on tool wear and surface roughness need further attention. So the objective of this paper deals about the optimization of machining parameters using neural networks. The performance results are compared and the conclusions are presented.

EXPERIMENTAL ANALYSIS
The test specimens (Al/SiCp) were produced using stir casting technique which is the economical one. It consists of a resistance furnace in which metal is melted in the closed environment. The powder is introduced in to the melt through the powder feeder. The melt is stirred thoroughly. Then the molten metal is taken out at the bottom of the crucible and poured in to the mould. The formed billet is remelted and again shaped casting is made. Green bonded Silicon Carbide particles were used as the reinforcements.
Carbide inserts K10 grade was used for turning operation. The experiment was repeated for the various cutting conditions like % volume of SiC, feed, speed, depth of cut and time of machining. The variables used in this study are given in the Table 1.
Table 1 Selected input variables and Outputs
Sl No
Input Variables
Variations
Output
1
% Volume of SiCp
10, 15, 20, 25 (%)
2
Cutting Speed
50, 100, 140 (m/min)
3
Feed
0.16, 0.32, 0.45 (mm/rev)
4
Depth of Cut
0.5, 1, 1.5, 2 (mm)
5
Time of Machining
5, 10, 15 (min)

Tool wear

Surface roughness
After each experiment, the tool insert was removed from the tool holder and the tool wear was measured using low magnification microscope. For required magnification, specified eye piece is selected (5x, 10x, 20x, 50x) and then the distance between the eye piece and tool is adjusted until to get the clear image of tool wear and the micrometer scale is fixed vertical to the width of the wear land. From the scale the width of wear land is measured and tabulated.
The surface roughness was measured using TalySurf tester. It is a contacting probe type surface tester, in which a stylus is moved along the work piece. The peaks and valleys of the specimen can be accurately found out from the upward and downward movements of the stylus. The average roughness value can be measured from this instrument.

Effect of parameters on Flank Wear
The figure 1(a) shows the effect of percentage volume of SiC on Flank wear. As expected, the percentage volume of the SiC particles showed their impact on tool wear. The flank wear, increases with increases in % volume of SiCp. When the SiC particles are fed while casting process, due to the dispersion hardening effect the possible dislocation movements in the Aluminium are restricted. So basically the hardness of the matrix improved. The tool which produces the compressive stress to the specimen on turning, need to exert more force to move the dislocations. This could be the reason why the tool wear increases with the percentage volume of particles.
The Effect of cutting speed on Flank wear is shown in figure 1(b). The flank wear increases on increasing the cutting speed. The rubbing area per unit time increase when cutting speed increases. This gives more cutting area experienced by the tool which will beone of the reasons for the increasing tool wear. And also the heat generated in the interface between the tool and work piece is directly proportional to the cutting speed. That interfacial temperature softens the tool material which produces more wear. Adhesive nature at elevated temperatures is also the reasons for the increase in the flank wear.
The effect of feed rate on flank wear is given in figure 1(c). As the cutting edge of the tool encounters the hard particles in the aluminum matrix, these particles chip away tiny flakes of the tool edge. Not like other metals, in metal matrix composite materials, the dispersed hard particles give an impact load when the tool contacted it. When feed rate is increased, residual stresses change from compressive to tensile. This also could be the reason for the increased flank wear. In addition to this the hard particles act as abraders and remove the materials from the flank surface. The increase in feed per tools increases force result in high chip tool interface temperature which also leads to increase in flank wear. The depth of cut plays a vital role in the tool wear. The flank wear is increasing on increasing the depth of cut. Since the tool used is a ceramic material, at elevated temperatures chemical wear becomes a leading wear mechanism and often accelerates weakening of cutting edge, resulting in premature tool failure (chipping), namely edge breakage of the cutting tool (figure 1(d)).
Figure 1(e) shows the influence of time of machining on flank wear. The increase in time of machining increases the flank wear. The reason could be the increase in the time of contact between the tool and the work piece. As time increases the chiptool interface temperature will also increase which tend to soften the flank surface of the tool. The rubbing time is also increased which lead to more flank wear due to the adhesion as well as abrasion.
Flank Wear (mm)
Flank Wear (mm)
0.8
0.6
0.4
0.2
0
0 10 20 30
% Vol. of SiCp
10
Surface Roughness (m)
Surface Roughness (m)
8
6
4
2
0
0 10 20 30
% Vol. of SiCp
Fig.1(a) Effect of percentage volume of SiC on Flank Wear
Fig.2(a) Effect of percentage volume of SiC on Surface roughness
0.5
Flank Wear (mm)
Flank Wear (mm)
0.4
0.3
0.2
0.1
0
10
8
6
4
2
0
10
8
6
4
2
0
0
0
50
50
100
100
150
150
200
200
Surface Roughness (m)
Surface Roughness (m)
0 50 100 150 200
Cutting Speed (m/min)
Cutting Speed (m/min)
Cutting Speed (m/min)
Fig.1(b) Effect of cutting speed on Flank wear
Fig.2(b) Effect of cutting speed on Surface roughness
0.6
Flank Wear (mm)
Flank Wear (mm)
0.5
0.4
0.3
0.2
0.1
0
10
8
6
4
2
0
0 0.1 0.2 0.3 0.4 0.5
Feed (mm/rev)
10
8
6
4
2
0
0 0.1 0.2 0.3 0.4 0.5
Feed (mm/rev)
Surface Roughness (m)
Surface Roughness (m)
0 0.1 0.2 0.3 0.4 0.5
Feed (mm/rev)
Fig. 1(c) Effect of feed rate on Flank wear
Fig. 2(c) Effect of feed rate on Surface roughness
0.8
Flank Wear (mm)
Flank Wear (mm)
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
10
8
6
4
2
0
10
8
6
4
2
0
0
0
0.5
0.5
1
Depth of Cut (mm)
1
Depth of Cut (mm)
1.5
1.5
2
2
Surface Roughness (m)
Surface Roughness (m)
0 0.5 1 1.5 2
Depth of Cut (mm)
Fig. 1(d) Effect of Depth of cut on Flank wear
Fig. 2(d) Effect of Depth of cut on Surface roughness
4
Flank Wear (mm)
Flank Wear (mm)
3.5
3
2.5
2
1.5
1
0.5
0
10
8
6
4
2
0
10
8
6
4
2
0
0
0
5
5
10
10
15
15
20
20
Time of Machining (min)
Time of Machining (min)
Surface Roughness (m)
Surface Roughness (m)
0 5 10 15 20
Time of Machining (min)
Fig. 1(e) Effect of Time of machining on Flank wear
Fig. 2(e) Effect of Time of machining on Surface roughness

Effect of parameters on Surface Roughness
Surface roughness in machining is depending upon the geometric factors of the operation, work material factors and vibration and machine tool factors. The figure 2(a) shows the effect of percentage volume of SiC on Surface roughness. This figure shows the increase in the percentage volume of SiC, increases the surface finish. The distributions of the SiC particles are uniform because of increase in the percentage volume which in turns gives improved surface finish.
Figure 2(b) shows the influence of cutting speed on the surface roughness. The surface roughness is gradually coming down with the increasing of cutting speed. At high speed, because of the increment in the interfacial temperature the softening of work piece is taking place which give smooth surface. The feed rate on surface finish does not show a proper trend (Figure 2(c)). The surface roughness remains more or less equal for varying depth of cut because of the less variations in the vibration of the lathe (Figure 2(d)). Figure 2(e) shows the effect of time of machining on surface roughness. Increase in time of machining increases the surface roughness,
due to wear of tool inserts. As time of machining increases, the formation of build up edge is also producing surface roughness.


THEORETICAL ANALYSIS
For the mathematical modeling the basic model which is explaining the influence of various input parameters like percentage volume of SiC, speed, feed, depth of cut and the time of machining on the outputs is used. This basic model is given in equation (1)
Y X *V a * Sb * Fc * Dd *T e
(1)
Where, Y is the Output response like flank wear and surface roughness, V is the Volume of SiCp, Sis the cutting speed, F is the feed, D is the depth of cut and T is time of machining. Where, X, a, b, c, d, e are the constants. These constants are evaluated for each characteristic to develop the models by using SPSS software.

Mathematical Modeling
From the basic model, as in equation 1, the output Y can be calculated from the parameters namely, percentage volume of SiCp(V), speed (S), feed (F), depth of cut (D) and time of machining (T). In that equation X, a, b, c, d and e are constants. These constants are playing a vital role in predicting the desired output factors like Flank wear and Surface roughness.
Through SPSS software, those constants were found. Mathematical modeling was constructed from the results obtained through the experiments. With the help of the Mathematical modeling the value of flank wear and surface roughness can be found for various inputs. Table 2 shows the result obtained through SPSS software.
The equation (2 and 3) shows the models which are used to find out the flank wear and surface roughness with respect to the input parameters. From the equation (2), for varying input parameters, the flank wear could be predicted. From the equation (3) for varying input parameters surface roughness could be predicted.
Flank wear (y1) = 0.001 x V0.8985 x S0.1961 x F0.0677 x D0.146 x T1.672 (2) Surface Roughness (y2) = 373 x V0.682 x S0.4851 x F0.5555 x D 0.121 x T0.1909 (3)
Table 2 Results obtained using SPSS
SOURCE
Flank Wear
Surface Roughness
Degree Of Freedom
Sum Of Square
Mean Square
Degree Of Freedom
Sum Of Square
Mean Square
Regression
6
17.381
2.891
6
289.63
48.21
Residuals
9
0.2010
0.022
9
3.37
0.37
Uncorrected total
15
17.581
15
293.0
Corrected total
14
8.925
14
Figure 3(a) and 3(b) show the comparison of experimental data with the predicted values of flank wear and surface roughness. From those figures it can be seen that the variation between the experimental and the mathematical values are minimum, so the mathematical modeling can be used for predicting flank wear and surface roughness for the given input values.
Fig.3(a) Comparison between Experimental and Mathematical modeling results for Flank wear
Fig 3(b) Comparison between experimental and Mathematical modeling results for Surface roughness.

Genetic Algorithm Based Model Development and Result Discussion
Genetic algorithms (Goldberg 1989; Deb 2002) attempt to mimic the properties of natural selection and natural genetics. They combine solution evaluation with randomized, structured exchanges of information between solutions to obtain optimality. They consider a population of potential solutions and then use a defined survival of the fittest technique to produce a new generation of solutions which are hopefully better approximations to the ideal solution. The purpose of selecting GAs is i) uses probabilistic transition rules, not deterministic ones; ii) search from a population of points, not a single point; iii) work on an encoding of the parameter set, not the parameters themselves and iv) use objective function information, not derivatives.

Overview of Genetic algorithm
A chromosome is an encoded string of possible values for the parameters to be optimized. These chromosomes can be made up of realvalued or binary strings. Often one of the main challenges in designing a genetic algorithm to find a solution to a problem is finding a suitable way to encode the parameters. A set of potential solutions, called a population, is created. Each member of this set is referred to as an individual and they are evaluated by decoding the parameter values from the chromosomes and applying them to the problem to see how well they perform the task at hand (the objective that is to be optimized). The score that an individual achieves at performing the required task is called its fitness.
After the fitness of each individual has been calculated, a procedure known as selection is performed. Individuals are selected to contribute towards creating the next generation, the probability of selection being related to the individuals fitness. Once selection has occurred, crossover takes place between pairs of selected individuals. The strings of two individuals are mixed. In this way, new individuals are created that contain characteristics that come from different hereto relatively successfully individuals. A third operation that occurs is mutation, the random changing of bits in the chromosome. It is generally performed with a relatively low probability. Mutation ensures that the probability of searching a given part of the solution space is never zero.

Implementation of Genetic Algorithm
Genetic algorithm derives their power from genetic operators. A Simple Genetic Algorithm (SGA) that yields good results in many practical problems uses the following three genetic operators: i) Reproduction; ii) Crossover and iii) Mutation.
Selection is an essential component of genetic algorithm, since it determines the direction of search where other genetic operators propose new search points in an undirected way. The selection methods includes fitness proportionate selection, tournament selection, rank selection etc. all these methods use genetic operations in one form or another to create new search points. Crossover and Mutation are the two critical operators in which crossover facilitate exploration, while mutation facilitates exploitation of the space. Crossover is a recombination operator. In the crossover operator, new strings are created by exchanging information among strings of the
mating pool. Mutation is a background operator which produces spontaneous random changes in various chromosomes.
Starting with an initial population, the genetic algorithm exploits the information contained in the present population and explores new individuals by generating offspring using the three genetic operators which can then replace members of the old generation. Fitter chromosomes have higher probabilities of being selected for the next generation. After several generations, the algorithm converges to the best chromosome, which hopefully represents the optimum or near optimal solution.
When applying GAs to solve a particular optimization problem, two main issues must be addressed.

Representation of the decision variables

Formation of the fitness function
While solving an optimization problem using GA, each individual in the population represents a candidate solution. In the binarycoded GA, the solution variables are represented by a string of binary alphabets. For problems with more than one decision variables, each variable is usually represented by a substring.
11001 10111 11101 10111 10011
V1
V2
V3
V4
V5
V1
V2
V3
V4
V5
In GA, there are two ways to generate the initial population namely random initialization and heuristic initialization. All the control variables are represented as binary strings in the GA population. The length of the binary strings is based on their actual value to obtain accurate solution. Using random initialization method, the binary strings are randomly generated as follows.
In this problem, the objective is to minimize the error while satisfying the constraints. The objective function (FT) is given in equation (1). Generally, GA searches for a solution with
maximum fitness function value. Hence, the minimization objective function is transformed into
a fitness function (ft) to be maximized as
f 1 1 FT


Simulation Results
The proposed algorithm was run with an objective of minimization of error. In this study, two different cases namely i) Wear and ii) Surface roughness were considered and run the algorithm under all the cases. The proposed model was implemented in MATLAB with the following parameter setting, as given in Table 3.
Table 3 GA parameter setting
Parameter Setting criteria
CASE I
CASE II
Population Size
30
30
Crossover Probability
0.9
0.8
Mutation Probability
0.01
0.08
Tournament Size
2
2
The optimal values of the control variables obtained by using the proposed genetic algorithm are presented in the Table 4 and it was found that all the variables corresponding to these control variables satisfy their limits which are given in the Table 1. The value of the error obtained is also less than the value reported in the Table 2 for all the cases.
Table 4 Control variables obtained using GA
Control Variables
CASE I
CASE II
Percentage volume of material (%)
10
25
Speed (m/min)
100
135
Feed (mm/rev)
0.16
0.16
Depth of Cut (mm)
0.5
1
Time of machining (min)
5
6



CONCLUSION
Metal matrix composites are widely used in many applications due to their special characteristics. In this paper, Al/SiCp was selected and an attempt has been given to optimize the machining parameters using intelligent techniques. Mathematical modeling was developed
through SPSS software. Experimentation results on the effect of wear and surface roughness were presented and discussed. GA based approach for optimizing the machining parameters of Al/SiCp composites. Verification experiments are conducted to prove the stability and reliability of the modeling and optimization.
REFERENCES

Allison J E and Cole G S. (1992), Metal matrix composite in the automotive industry; opportunities and challenges, Journal of Materials, pp.1924.

Cronjager L and Meister D (1992), Machining of fiber and particulate reinforced aluminum,
Ann. CIRP Vol. 41 (1), pp. 6366.

Davim J P and Baptista A M (2000), Relationship between cutting force and PCD cutting tool wear in machining silicon carbide reinforced aluminum, Journal of Materials Processing Technology, Vol. 103, pp. 417423.

Deb.K.(2002) Optimization for engineering design algorithms and example, Prentice Hall of India.

Gallab M and Sklad M. (1998), Machining of Al/SiC particulate metalmatrix composites Part 1: Tool performance, Journal of Materials Processing Technology, Vol. 83, pp. 151158.

Goldberg, D.E. (1989) Genetic Algorithms in Search Optimization and Machine Learning, Addison Wesley.

Ibrahim Ciftci, Turker M and Seker U (2004), Evaluation of tool wear when machining SiCp reinforced Al 2014 alloy matrix composites, Journal of Materials and Design, Vol. 25, pp.251 255.

Jang, Mizutani. (1987) NeuroFuzzy and Soft Computing: A Computational Approach to Learning and Machine Intelligence, PrenticeHall of India Pvt. Ltd., New Delhi.

Manna A and Bhattacharyya (2002), Investigation for effective tooling systems during machining of Al/SiCMMC, Journal of Materials Processing Technology, Vol. 123 (3), pp. 476 482.

Manna A and Bhattacharyya (2003), A study on machinability of Al/SiC MMC, Journal of Materials Processing Technology, Vol. 140, pp. 711716.

Manna A and Bhattacharyya (2005), Influence of machining parameters on the machinability of particulate reinforced Al/SiCMMC, International Journal of Advanced Manufacturing Technology, Vol. 25, pp. 850856.

MATLAB Optimization Toolbox Users Guide (1997) The Mathworks, Inc., Natick, M.A.

Monaghan J M (1994), The use of quick stop test to study the chip formation of an SiC/Al metal matrix composite and its matrix alloy, Proceedings of Advanced Materials, Vol. 4, pp. 170179.

Sivanandam S.N., Sumathi S., Deepa S.N,. (2006) Introduction to Neural Networks using Matlab 6.0., Tata McGrawHill Publishing Company limited, New Delhi.

Tomac N and Tonnessen K (1992), Machinability of particulate aluminum matrix composites,
CIRP Ann. Vol. 41, pp. 5558.

Yegnanarayana, B. (1999) Artificial neural networks PrenticeHall of India Pvt. Ltd., India.

Chou Y.K., C.J.Evans, and M.M.Barash, (2002), Experimental investigation on CBN turning of hardened AISI 52100 steel, Journal of Material Process Technology, Vol.124, pp.274283.

Ozel T., TK.Hsu, and E.Zeren, (2005), Effects of cutting edge geometry, workpiece hardness, feed rate and cutting speed on surface roughness and forces in finish turning of hardened AISI H13 steel, International Journal of Advanced Manufacturing Technology, Vol.25, pp.262269.

Dimla D.E., Lister P.M., and N.J.Leighton (1997), Neural network solutions to the tool condition monitoring problem in metal cutting A review critical review of methods, International Journal of Machine Tools and Manufacture, Vol.39, pp.12191241.

Ozel T. and Karpet, (2005), Predictive modeling of surface roughness and tool wear in hard turning using regression and neural networks, International Journal of Machine Tools and Manufacture, Vol.45, pp.467479.

Zitzler E., M.Laumanns, and S.Bleuler (2003) A tutorial on evolutionary multiobjective optimization, Workshop on Multiple Objective Metaheuristics (MOMH 2002), SpringerVerlag, Berlin, Germany.

Karpat Y., and T.Ozel (2005) Hard Turning Optimization using Neural network modeling and Swarm Intelligence, Transactions of North American Manufacturing Research Institute, Vol.XXXIII, pp.179186.

Cus F. and Balic. J (2003) Optimization of cutting process by GA approach, Robotics and Computer Integrated Manufacturing, Vol.19, pp.113121.

Saravanan R., Asokan.P. and Schidanandam.M (2002) A multiobjective genetic algorithm approach for optimization of surface grinding operations, International Journal of Machine Tools and Manufacturing, Vol.42, pp.13271334.