1. Introduction

Ultrasonic machining (USM) is known as the non-conventional machining process (Kuriakose et al. 2017; Wang et al. 2016). In which the material is removed by erosion mechanism. The selection of input process parameters play an important role in the USM process (Li et al. 2016; Lin et al. 2016). In this paper, the input parameters are abrasive slurry concentration, type of abrasive particles, power rate, grit size of abrasive particles, hydro-fluoride (HF acid) concentration, tool material are selected (Khairay 1990; Choi et al. 2007). The output parameters are material removal rate (MRR), tool wear rate (TWR) and surface roughness (SR).

Machining of polycarbonate bullet proof (PBPG UL-572) and acrylic heat resistant BS-476 (AHRG BS-476) are too tough job, because it have alternative layers of glass, polycarbonate and acrylic material. Acrylic and polycarbonate material are easily machined by conventional processes and glass is machined by non-conventional processes like USM, water jet machining (WJM) and abrasive water jet machining (AWJM) (Choi et al. 2007; Guzza et al. 2004). Other non-conventional processes like Laser beam machining (LBM) is not utilized because it produced heat effected zone, electron beam machining (EBM) is applicable only on conductive materials and conventional machining will damage the PBPG UL-752 and AHRG BS-476. In last USM is best alternative for machining of this material. Some important properties of PBPG UL-752 and AHRG BS-476 are shown in the Table 1.

Table 1. Important properties of PBPG UL-752 and AHRG BS-476

In the experiment, selected parameters having three different level shown in Table 2. Design of experiment is prepared by Minitab 6.7 software in which L₂₇ orthogonal array is used. The levels are selected by pilot experiments. For calculating MRR and TWR the initial and final weight of tools and work sample respectively measured by weight machine and surface roughness of check by Taylor Hobson Surtronic 25 surface roughness tester.

Table 2. Different input or controllable machining parameters & their levels

Experiment are performed on Sonic Mill 500W USM and schematic diagram of machine is shown in Fig.1. 500W Sonic-Mill ultrasonic machine have vibrating spindle kit, feeding system at constant pressure and abrasive slurry flow pump system. Fig. 1 show the schematic figure of USM apparatus. The ultrasonic vibration kit contains an ultrasonic spindle, 25.4 mm diameter cylindrical horn, and power supply (Hofy 2012; Jatinder & Khamba 2010). The power supply unit convert 50Hz electric supply input to sonic frequency 20 kHz output. 500W Sonic-Mill vibrating kit having piezoelectric transducer, it convert electric signal input into mechanical vibration output signal (Lee & Chan 1997; Kanwal Singh & Ahuja 2014; Kanwal Singh & Singla 2014; Singh and Khamba 2008). The amplitude of the vibration is 0.0237-0.0268 mm and frequency of vibration 20 kHz ± 250 Hz. Static load for feeding the USM tool is fixed at 1.757 kg and abrasive slurry flow rate is 30 L/min (Singh & Khamba 2007; Thoe et al. 1998; Vinod & Aniruddha 2008; Zhang et al. 1999).

Figure 1. Schematic Diagram of Chemical Assisted Ultrasonic Machine

2. Material and Methods

Selected parameters and levels are shown in Table 2. For the design of experiment orthogonal array L₂₇ is used and design is prepared by Minitab 6.7 software. The design of experiment is shown in Table 3. All the experiments are performed according to the design experiment. MRR and TWR are calculated by the equation 1 and equation 2, in which density of work material ρ is 8.3 gm/cm³, W_i is initial weight, W_f final weight after processing, t is time take in machining. T_i initial and T_f weight of tool and ρ = Density of D2 Steel 7.83 gm/cm³, Density of HC steel 7.85 gm/cm³, Density of HST steel 8.13 gm/cm³ (Hasani et al. 2012; Hasiao et al. 2008)

(1)

(2)

Table 3. Design of experimentation (Orthogonal Array L27) and their levels

Surface roughness is measured is R_a, it is the universally recognised and most used international parameter of roughness. It is the arithmetic mean of the absolute departure of the roughness profile from the mean line.

After machining the MRR and TWR are calculated and SR is checked, machining data is shown in Table 4. In which MMR and TWR is calculated in mm³/min and surface roughness in R_a.

Table 4. Design of experimentation (Orthogonal Array L27) and their levels

3. Results and Discussion

In grey relation analysis, data pre-processing is necessary to sequence scatter range. Data pre-processing is a process in which original sequence is transferred into comparable sequence. The experiment results are normalized in the range between zero (0) and one (1). Depending on output parameters, data pre-processing methodologies are adopted (Lin et al. 2002; Lin & Lee 2009; You et al. 2017). MRR is the governing output parameter in USM, which decided the machinability of work material under deliberation. “Larger-the-better” characteristics is used for MRR to normalize the original sequence by equation 3.

(3)

Where, is the sequence after the data processing, is the comparability sequence, K=1 and k=4 for MRR; i= 1,2,3………27 for experiment number 1 to 27.

TWR and SR are the important measure of USM, these output parameters are represent the machining accuracy under selected input parameters (Patil & Patil 2016; Das et al. 2016). To get the optimum performance the “Smaller-the-better” characteristic has been preferred to normalize the original sequence date by equation 4.

(4)

Where, is the sequence after the data processing, is the comparability sequence, K=2, K=5 for TWR and K=3, K=6 for SR; i= 1,2,3………27 for experiment number 1 to 27. is the value after grey relational generation, and are the smallest and largest value of . After normalized MRR, TWR and SR of PBPG UL-752 and AHRG BS-476 comparable sequence is shown in the Table 5.

Table 5. The sequences of each performance characteristic after data processing

Now is the deviation sequence between reference sequence and the comparability sequence (Ahmad et al. 2016). Deviation sequence is calculate by the equation 5 and maximum and minimum difference is found, K=1, 2 and 3 and i= 1, 2, 3…27.

(5)

The deviation sequence table is shown in the Table 6, Maximum and Minimum are obtained and shown below.

Table 6. The deviation sequences

After per-processing data, the next step in calculate the Grey relational coefficient and Grey relation grade with the pre-processed data (Lin et al. 2009). It define the relationship between ideal and actual normalized results. Grey relational coefficient can be expressed as equation 6 is shown below.

(6)

Where, is the deviation sequence of the reference sequence and the comparability sequence, is distinguishing or identification coefficient. In this calculation because all parameters are given equal preference (Lin 2012). The Grey relation coefficient for each experiment of the L27 orthogonal array is calculated by using equation 6 and shown in Table 7.

Table 7. The calculated Grey Relational Grade and its order in the optimization process

After obtaining the Grey relation coefficient, the Grey relation grade is obtained by averaging the Grey relation coefficient corresponding to each performance characteristic and represent by and Equation 7 (Manivanna et al. 2011) show the general formula of Grey relation grade and equation 8 is for three output parameters, shown in Table 7.

(7)

(8)

The higher value of Grey relation grade is represent that the corresponding experiment result is much closer to the ideally normalized value. Experiment number 25 get the best multiple performance characteristics among the 27 experiment because it have the highest value of grey relation grade. Now the experimental design is orthogonal, it is possible to separate out the effect of each parameters on the basis of Grey relation grade. Mean of Grey relation grade is calculated for level 1, 2 and 3 by averaging the Grey relation grade of the experiment 1to 9, 10 to18 and 19 to 27 are shown in Table 8. The mean of Grey relation grade for abrasive, power rate, grit size, HF acid and tool material are calculated in same manner. The total mean of Grey relation grade for 27 experiment is also shown in the Table 8. *Level for optimum grey relational grade. Optimum level parameters are find out from response table and shown in the Fig.2. Larger value of Grey relation grade is closer to the ideal value. Therefore, the optimum parameters setting for higher MRR and lower TWR and SR are A₃B₃C₂D₃E₁F₃.

Table 8. Response Table for the Grey Relational Grade

Figure 2. Effect of USM parameters on the multiple performance characteristics

Furthermore, analysis of variance (ANOVA) is performed on Grey relation grade to achieve contribution of each input parameter affecting the output parameters. ANOVA for Grey relational grade is shown in Table 9. In addition, F-test is also used to find out the percentage contribution of each parameters. From Table 9 it is clear that material of tool have the significant role in the machining which have 30% contribution, 25% contribution of concentration, 21% contribution of grit size, 9% contribution of abrasive, 4% contribution of HF acid and 3% contribution of power rate in the machining of PBPG UL-752 and AHRG BS-476.

Table 9. ANOVA of Grey relation grade

After getting the optimum parameters for machining the experiment is performed by those input setting (A₃B₃C₂D₃E₁F₃). Fig.3 show the Scanning electron microscope (SEM) images of PBPG UL-752 machining setting A₁B₁C₁D₁E₁F₁, In which machining by USM is performed and some crack are also found on the work surface. In other hand in Fig. 4 the USM machining of PBPG UL-752 is performed by optimum parameters which are found by Grey relational analysis A₃B₃C₂D₃E₁F₃, there is smoother and crack free surface. Similarly, in Fig.5 show the Scanning electron microscope (SEM) images of AHRG BS-476 machining setting A₁B₁C₁D₁E₁F₁, in which machining by USM is performed and some crack are also found on the work surface. In other hand in Fig. 6 the USM machining AHRG BS-476 is performed by optimum parameters which are found by Grey relational analysis A₃B₃C₂D₃E₁F₃, there is smoother and crack free surface.

Figure 3. SEM image of PBPG UL-752 A1B1C1D1E1F1 experiment

Figure 4. SEM image of PBPG UL-752 A3B3C2D3E1F3 optimum Grey relational analysis

Figure 5. SEM image of AHRG BS-476A1B1C1D1E1F1 experiment

Figure 6. SEM image of AHRG BS-476A3B3C2D3E1F3 optimum Grey relational analysis

MRR and TWR are also compared between optimum Grey relational analysis A₃B₃C₂D₃E₁F₃ and A₁B₁C₁D₁E₁F₁ input parameters. It observed that optimum parameters (A₃B₃C₃D₃E₁F₃) gives 73.02% improved MRR with comparison of A₁B₁C₁D₁E₁F₁ USM experiment setting. TWR is decreased by 37.25%. It is evident from SEM image, optimum parameters setting also give the better surface roughness which is 43.33% improved. Fig 7 shown the percentage contribution of optimum Grey relational analysis parameters.

Figure 7. Percentage contribution of factor on Grey Relational Grade

Confirmation test is carried out to verify the improvement of performance characteristics in machining of PBPG UL-752 and AHRG BS-476 by USM. The optimum parameters are shown in the Table 10. The estimated Grey relational grade using the optimal level of machining parameters can be calculated by using equation 9 (Meena & Azad 2012; Singh et al. 2004; Sreenivasulu & srinivasarao 2012).

(9)

Where, is the total mean of Grey relational grade, is mean of the Grey relational grade at optimum level and n is the number of parameters that significantly affect multiple-performance characteristics. It is clearly show that the multiple-performance characteristics in USM process is greatly improved through this study.

Table 10. Improvement in Grey relational grade with optimized USM machining parameters

4. Conclusions

The optimum machining parameters are identify by Grey relational grade for multiple performance characteristics that is MRR, TWR and SR. This experimental research paper presented the multi-objective optimization of USM machining parameters of polycarbonate bullet proof UL-752 and acrylic heat resistant BS-476 glass for drilling application by Grey relational analysis method. Following conclusion are conclude from the experimentation analysis.

1. Concentration of abrasive slurry, concentration and grit size of abrasive play the significant role for optimum output performance parameters.

2. ANOVA of Grey relational grade for multiple performance characteristics reveals that the concentration have the significant role in the MRR.

3. Based on SEM images, it is evident that optimum parameter improve the surface roughness and give better smooth surface.

4. PBPG UL-752 have improvement in MRR, TWR and SR is 80%, 50% and 40% respectively, based on confirmation test.

5. AHRG BS-476 have improvement in MRR, TWR and SR is 70%, 30% and 25% respectively, based on confirmation test.

6. It proof that, the performance characteristic of USM process like MRR, TWR and SR are improved together by using the Grey relational study and the effectiveness of this method is effectively recognised by authentication experiment.