Fuzzy Rule based Modeling of Drilling Parameters for Tool Wear in Drilling GFRP Composites

The most recent development of high performance fibers has led to the availability of composites which can compete and indeed replace metals. Just like any other engineering material GFRP’s have to be machined as well. Amongst the three machining operations of turning, milling and drilling, the latter is the most important. Drilling of GFRP is different from traditional metal machining because of the following concerns. In machining of GFRP the drill bit experiences variable thrust forces that produces excess vibration results in poor tool life. Apart from force fluctuations, cutting tools also experience severe abrasion on the flank surface, due to encounter with hard glass fiber. In this work, experiments were conducted in CNC milling machine for carbide drill bit based on L9 orthogonal array. The process parameters investigated are spindle speed, feed rate, and drill diameter. Fuzzy rule based model is developed to predict the thrust force in drilling GFRP. The results indicated that the model can be effectively used for predicting the response variable by means of which tool wear can be controlled.


Introduction
Glass fiber reinforced composites are perhaps the most commonly used ranging in application from helmets, fishing boats, automobiles body and aircraft structures due to their high strength, high stiffness, low weight, low thermal expansion and good dimensional stability. GFRP composites are characterized by high strength and rigidity coupled with low weight and, in many respects, they are superior to metals Davim (2004). However, the glass fiber constituent often renders the machining of GFRP difficult. Drilling operations involve cutting holes in the raw materials using rotating cutters called drilling machines or drilling presses. It includes drilling holes which enables the GFRP composites to be incorporated and assembled in structural elements. Drilling of GFRP is essential for assembly or fabrication process. During drilling GFRP the tool continuously encounters alternate matrix and fiber materials, whose response to machining can vary greatly, experiencing force fluctuations and consequent deterioration in tool performance stated by Velayudham (2005).
Thrust force is an important parameter related to the drilling tool that is found life of the tool. It also affects the extent of delamination suffered by the composite material. Abrate et al (1992) stated that the thrust force and torque formed during drilling have important information connected to quality of the hole and wear of the drill bit. Lin et al (1996) carried out a study on drilling composite material at high speed and accomplished that an Keywords: Drilling, Fuzzy Logic, GFRP Composites, Thrust Force, Toolwear increase of the cutting velocity leads to an increasing drill wear that in turn provokes an increase in the thrust force. By the research of Saghizadeh and Dharan (1986) who identified that the thrust force is a function of the speed of drilling and the feed rates., when feed rates are increased thrust forces also increase resulting in larger amounts of delamination.
The experimental work conducted by Inoue (1997) involved drilling of a Glass Fiber Reinforced Plastic (GFRP) with a cemented carbide drill (1mm diameter). The author concluded that due to the pre-drilled hole to 0.4 mm or above, thrust force is drastically reduced. Palanikumar et al (2006)  In the present work a user friendly fuzzy logic based system has been designed for the prediction of thrust force in drilling GFRP composites. The experiments were conducted on CNC milling machine. Carbide drill bit of various diameters is used for the investigation. L 9 orthogonal array is used for experimentation. Thrust force and tool wear were obtained as output parameter. The results indicated that the fuzzy logic model can be effectively used for drilling of composites.

Experimental Work
The drilling test were performed on Hass SSYT USA model CNC milling machine with maximum torque 102 Nm@1400 rpm, maximum power rating 20 HP and maximum feed rates 25.4 m/min. Figure 1 represents the experimental setup.

Cutting Tools and Specimen Preparation
A total of nine uncoated carbide drill bits, three drill bits of diameter 6 mm, 8 mm and 10 mm were used in this experiment. The performance of each drill bit was examined and monitored for nine conditions of feed rate and speed of spindle rotation. The dimensional properties of the drilling tool are tool overhang 47 mm, point angle 140°, helix angle 30° and Shank type is cylindrical. In these experiments, the GFRP specimen with dimensions 150 x 150 x 8 mm prepared through Compression moulding technique and used as work piece. High strength 360 CSM OCV Make Glass Mat was used as reinforcement in epoxy resin with araldite as hardener and processed to a temperature of 155 0 C to 160 0 C for approx 35minutes to fabricate GFRP.

Cutting Conditions
The experiment was tested for three levels of feed rates, spindle speeds and Drill diameter. All these three levels are summarized in Table 1. The Taguchi methodology for three factors at each of these three levels was used for the implementation of the plan of experiments. The cutting condition was selected from ASM machining composites data book. The orthogonal array selected was L9 (3 3 ). A total of 9 experiments were carried out considering speed, feed and coolant pressure for drilling each experiment has been repeated for three times and the thrust force was recorded for each experiment.

Thrust Force and Tool Wear Measurement
The thrust force developed in drilling the composite material was measured using Drill tool dynamometer. The drilling thrust force values are stored in laptop for further analyses. Tool wear is determined by measuring the wear as it develops, specifying the effective cutting time elapsed before a stipulated degree of wear is reached. Tool wear were measured using profile projector.

Fuzzy Rule Based Model for Thrust Force in Drilling GFRP Composites
A fuzzy logic unit consists of a fuzzifier, membership functions, a fuzzy rule base, an inference engine, and a defuzzifier. The input and output values are fuzzified using membership functions. The fuzzy reasoning works on fuzzy rules to generate a fuzzy value to be used by inference engine. Finally, fuzzy value is converted into a crisp output by defuzzifier. In this work, defuzzification is done according to the Centre of Area (COA) method. Two fuzzy models namely mamdani and sugeno are available. Sugeno outputs are used for linear or constant; for others the mamdani is preferred. Between the two, Mamdani fuzzy inference is used in modeling the process parameters due to the non-liner relation with outputs. Modeling of thrust force was done by using MATLAB R2011b software 1 .

Fuzzy Expression
For the prediction of output parameters such as thrust force the drilling operation depends on three input parameters, namely Federate , Speed and Tool, diameter.
The first step in establishing the algorithm for fuzzy model is to choose the fuzzy sets of the process variables 2 .

Degree of Membership Functions of Input and Output Variables
The first step in generating a fuzzy logic is to identify the ranges of input and output variables. Then, the range of each process variable is divided into groups of fuzzy subsets. Each fuzzy subset is given a proper name and assigned a membership function 3 . The membership function is assigned without depending on the results of the experiments. In the MATLAB R2011b, the maximum number of membership function levels available is 9.

Results and Discussion
The tensile strength of composites is up to six times greater than that of conventional materials with higher strength to weight ratios 7 . They have high torsional strength and impact properties. Their ability to withstand fatigue is 60% higher than that of conventional materials. In addition they are up to 45% lighter than aluminum structures with low embedded energy levels. The thrust force developed in drilling is an important concern 8 . Table 2 shows the influence of thrust force and tool wear for 10mm diameter drill bit for all the nine experiment 9 . From the above Table 2 it is observed that average thrust force for all the experiment is around 15.5Kg F. The range of thrust force varies from10 KgF to 21KgF. The average maximum thrust force 21KgF was recorded at 9 th experiment, since the speed and feed given is high. In the same way average minimum thrust force 10KgF was recorded at 1 st experiment. Since the speed and feed given is very less when compared to other experiments 10 .
During the experiment numbers 1 to 3, feed is kept constant 5 mm/min and speed varied to 750 rpm, 1000 rpm, and 1250 rpm as a result thrust force and drill wear increases gradually from 1 st experiment to 3 rd experiment due to significant increase in the speed 11 . Similar inference can be referred in the Table 2 for 4 th to 6 th experiment as well as 7 th to 9 th experiment. During the experiment numbers 4 to 9, feed is more and hence the machining time is less as, when compared to experiment numbers 1 to 3 as shown in Table 2. Since the machining time is less the cutting edges of the drill bit is subjected to severe shock leading to high thrust force and tool wear. This shows the significance of feed towards thrust force in drilling 12 . From the Table 3 it is observed the thrust force value for 6mm drill bit in cutting condition 3 is 14 KgF where as for cutting condition 6 and 9 is 19 KgF and 21KgF. This shows the significance of feed in thrust force when the speed is constant. Similar observation is observed for 8mm and 10mm diameter drill bit. Similarly we can observe an increase in drill wear due to increase in cutting condition and increase in drill diameter 13 . This is due to increase in the chisel edge length for larger diameter drill bit compared to smaller diameter drill. Hence we can observe a decrease in thrust force and tool wear in cutting condition I for speed 750rpm and feed 5mm/min. Figure 5 and 6 represent the comparison of experimental and fuzzy results for thrust in drilling composite materials 14 .The results indicated that the thrust force varies with the experimental condition. The graph is drawn for the nine set of experimental condition with respect to thrust force. The tool wear developed in drilling in drilling operation is proportional to the amount of thrust force 15 .
When the tool thrust observed in machining is more it leads to the tool wear. For reducing of thrust force, modeling and optimization of process parameter are required. In this work fuzzy rule based modeling is used for the prediction of thrust force 16 . From the Figure 5 and 6 it can be observed that the results obtained through the fuzzy logic model are almost similar to the experimental results 17 .
To validate the experimental results ,the thrust force was measured for arbitrary cutting condition and validated with the modeling fuzzy based results. It was found that the percentage of error is less than 2% 18 . Hence fuzzy rule based modeling technique can be effectively used for the prediction of thrust force in drilling GFRP composites 19 .

Conclusions
The objective of the present was to identify the negative side effects that occur whilst drilling GFRP's. From the literature review, it was identified that due to increase in thrust force the tools get abraded very quickly as compared to metals. Hence in the present work the thrust force and tool wear developed during drilling GFRP composites has been investigated according to L9 orthogonal array experiments. Fuzzy rule based model has been developed for predicting thrust force in drilling GFRP composites. Based on the experimental and fuzzy logic modeling results the following conclusions are drawn.
• Based on the investigation optimized cutting parameter which minimizes thrust force and tool wear during drilling GFRP composites is speed 750rpm and feed 5mm/min. • Lesser diameter drill bits minimize thrust force and tool wear during drilling GFRPs. • Fuzzy rule based model is best method for predicting the thrust force in drilling GFRP composites.
The present work can be extended to investigate the effect of using different tool materials on the quality of drilled hole in different types of GFRPs.