Keywords

1 Introduction

Metal matrix composite is a new engineered material which possesses the inherent properties of matrix and reinforcements [1]. Metal matrix composites have high specific strength most suitable for aerospace and automobile components [2]. Therefore, machining of those components with close tolerance and in desired shape is the major industrial concern especially through the conventional machining process [3, 4]. Wire cut EDM serves this purpose up to some extent. Some of the previous findings on wire cut EDM are as follows: Velmurugan et al. [5] investigated on nickel-titanium shape memory alloy machining in WEDM. They used servo voltage, pulse-on time, pulse-off time, current and wire-speed as an input parameter. Servo voltage is the major influencing factor for MRR and surface roughness. Choudhuri et al. [6] investigated on surface quality of WEDM machined stainless steel 316. Ton, Toff, Ip, Sv, WT were the input parameters for machining 25 sets of experiment. From ANOVA result, Ton found to be the most controlling factor for roughness as well as recast layer. Jain et al. [7] have used ANN for the evaluation machining performances with the simultaneous effect of Ton, Toff, Ip and bed speed on surface roughness and AE signals. The result shows that better R values obtained when network trained with 70% of the data compared to 50 and 60% of data. Nain et al. [8] have evaluation of surface roughness and waviness of the WEDM of aeronautic superalloy with the various process parameters. The results concluded that apart from Ton Toff and Sv wire tension has significant effect on surface roughness. Also, the machining performance of aeronautics superalloy can be efficiently evaluated by BP-ANN model as compared to fuzzy logic method. Devarasiddappa et al. [9] have predicted the surface roughness of Inconel 825 aerospace alloy machined through WEDM through ANN model. The parametric study shows that the lower SR can be obtained at low levels of Ton and Sv. ANN model accuracy recorded as 93.62% and average predicted error recorded as 6.38% at ANN architecture 4-16-1. This ANN architecture found optimum, which were statistically validated by conducting hypothesis tests. ANOVA showed that Ton is the most affecting factor for SR with 76.12% contribution, followed by Sv and Toff, respectively, with 7.18 and 5.3% contributions. Das et al. [10] have used Taguchi L16 OA to conduct WEDM experiment on Al6061/0.5% SiC/0.5%B4C hybrid nano-composite to evaluate the effect of Ip, Sv, Ton and Toff on surface roughness. The experimental data have 96.32% accuracy with predicted values from RSM modelling. Also from the ANOVA Ton is the most influential factor for surface roughness. Garg et al. [11] have machined the ZrSiO4/6063 aluminium MMC using CNC WEDM. In this study, the author has developed the quadratic model for dimensional deviation to evaluate the contemporaneous effect of machining parameters, namely Ip, Sv, Ton and Toff. Experimental results show that the dimensional deviation (DD) is directly proportional to the pulse-on time and peak current. The objective of the present study is to synthesis AA7050/B4C composite through stir casting method and performs machining through CNC wire cut EDM.

2 Experimental Setup

Procedure for the synthesis of AA7050-B4C (7.5% by weight) is explained in my previous paper [12]. A rectangular plate of thickness 7 mm was cut from the developed composite block and used as workpiece materials for CNC wire EDM machining. The pictorial representation with specifications CNC wire cut EDM is shown in Fig. 1 and Table 1.

Fig. 1
figure 1

Wire cut CNC EDM

Full size image
Table 1 Specification of wire cut CNC EDM
Full size table

2.1 Design of Experiments

Orthogonal array (OA) designed by Dr. Genichi Taguchi, the primary purpose was to reduce the number of the experiment by considering whole domain of process parameters [13]. For the present study, three process parameters are selected which varies at three-level each, so as per the full factorial experimental design 27 experiments proposed which is not only time taking rather costly as well. Nine tests can also make a similar study. Therefore, Taguchi L9 orthogonal array experimental design has used as shown in Tables 2 and 3.

Table 2 Design of experiments
Full size table
Table 3 Experimental arrangement as per Taguchi L9 orthogonal array
Full size table

The material removal rate (MRR) has been calculated using Eq. (1). Machining time has been noted using stopwatch with least count of 1 s. The volume of material removed is calculated by (length of machining × thickness of machining × average kerf width).

$$ {\text{MRR}} = \left( {{\text{Volume}}\,{\text{of}}\,{\text{material}}\,{\text{removed}}} \right)/\left( {{\text{Machining}}\,{\text{time}}} \right) $$
(1)

Length and thickness of machining for the present study considered as 5 mm and 7 mm, respectively, and Kerf widths were obtained from the optical microscope for entire machining length of 5 mm, with optical magnification at 5×. Taylor Hobson Profilometer was used to measure surface roughness of the machined surface. Surface roughness has been measured considering the value of average surface roughness (Ra) of the machined surface.

3 Results and Discussions

The experimental results in terms of MRR and surface roughness obtained with different sets of process parameters are listed in Table 4.

Table 4 Experimental results with corresponding SN ratio
Full size table

3.1 Taguchi Methods

In Taguchi method, signal-to-noise ratio (S/N) is a measure of the deviation of the experimental values from the desired benefits. In the term, signal-to-noise ratio signal means the desired benefits, whereas the noise stands for undesired values. There is three definite way to express S/N ratio, namely higher-the-better (HB), lower-the-better (LB), nominal-the-better (NB) [14]. In the present study, both response parameters are of different perspectives. The prime motive of the study was to obtain maximum material removal rate (MRR) with least surface finish. Therefore, MRR was considered as HB, whereas surface roughness considered as LB. The S/N ratios were calculated as per Eqs. (2) and (3), respectively, for MRR and surface roughness [14].

$$ {\text{SN}}\,{\text{ratio}}\,{\text{for}}\,{\text{Higher}}\,{\text{the}}\,{\text{better}} = - 10\log \left[ {\frac{1}{n}\mathop \sum \limits_{i = 1}^{n} \frac{1}{{ y_{i}^{2} }}} \right] $$
(2)
$$ {\text{SN}}\,{\text{ratio}}\,{\text{for}}\,{\text{smaller}}\,{\text{the}}\,{\text{better}} = - 10 \log \left[ {\frac{1}{n}\mathop \sum \limits_{i = 1}^{n} y_{i}^{2} } \right] $$
(3)

where n and yi are the total number of experiments and values of MRR or surface roughness for ith experiments. The S/N ratio of each response parameters listed in Table 4. In order to get maximum MRR, the S/N ratio plot is shown in Fig. 

Fig. 2
figure 2

Main effect plot for mean SN ratio of MRR and surface roughness

Full size image

2. The similar graph has obtained from mean effect plot Fig. 

Fig. 3
figure 3

Main effect plot for the mean of MRR and surface roughness

Full size image

3.

Minimum MRR has derived from lower Ip Ton and Sv. Since MRR in EDM process is a function of spark energy and the spark energy is a function of Ip Ton and Sv as shown in Eq. (4).

$$ E = \int I_{{\text{p}}} T{\text{on}}S{\text{v}} $$
(4)

Therefore, MRR is minimum at the lowest process parameters; it goes on increasing as spark energy rises [15]. But the on higher spark energy, there will be favourable energy loss which reduces the metal removal rate. Figures 2 and 3 show the S/N ratio plot and mean plot for surface roughness.

The main effect plot for mean MRR shows that for maximum MRR the optimised input parameters obtained from Taguchi analysis is Ip3Ton3Sv3, whereas for minimum surface roughness, the optimal combination of process parameters is Ip3Ton2Sv1. In the present study, the purpose of analysis of variance (ANOVA) is to verify the consistency of control variables in the experimental result. ANOVA Table 5 shows that although all the process parameters have significant contribution towards achieving maximum MRR pulses current have maximum contribution compare to other two parameters. For surface roughness evaluation, the pulse-on time and servo voltages are the highly significant factors than pulse current. The similar trend shows in Table 6.

Table 5 ANOVA for MMR
Full size table
Table 6 Analysis of variance for surface roughness
Full size table

3.2 Confirmation Test and Prediction

Taguchi method can be used to predict the S/N ratio, using the optimal level of the process parameters can be calculated as Eq. (5) [16].

$$ \tilde{\chi } = \chi_{m} + \mathop \sum \limits_{i}^{n} (\chi_{i} - \chi_{m} ) $$
(5)

where \( \tilde{\chi } \) is the mean of S/N ratio χm is the total mean of S/N ratio and n is the number of process parameters. Table

Table 7 Confirmatory test
Full size table

7 compared the predicted and experimental values of MRR and surface roughness at optimal setting of process parameters. For MRR, the optimal configuration of process parameters is Ip3Ton3Sv3 that is pulse current 230 A, pulse-on time 110 µs and servo voltage of 40 V. The predicted MRR estimated as 31.0086 mm3/s, whereas the experimental value of MRR at optimal setting of process parameters evaluated as 30.1242 mm3/s. The error calculated from predicted and experimental MRR calculated as 2.85% which is within the acceptable range. Similarly, the optimal setting of process parameters for surface rough obtained from the Taguchi method is Ip3Ton2Sv1, i.e. pulse current 230 A, pulse-on time 108 µs and servo volt 20 V. The predicted surface roughness estimated as 2.1543 μm. Whereas the experimental values of surface roughness at optimal setting of process parameters evaluated as 2.18129 μm. The error calculated from predicted and experimental MRR calculated as 1.24% which is within the acceptable range.

To verify the consistency of the proposed Taguchi method, the predicted MRR and surface roughness obtained at the initial set of process parameters (Ip2Ton2Sv2) compared with experimental one. The errors estimated are within ±5%. Hence, the proposed Taguchi method is consistent (Table 8).

Table 8 Results of confirmation at initial settings of process parameters
Full size table

4 Conclusions

AA7050/7.5% B4C composite fabricated successfully through stir casting method. Wire cut electro discharge machining performed on the composite with pulse current, pulse-on time and servo volt as process parameters to evaluate MRR and surface roughness of the machined surfaces. Following conclusions can be drawn from the results.

  1. 1.

    Analysis of variance for MRR shows that the pulse current is the major influential factor for obtaining maximum MRR followed by servo volt and pulse-on time. For surface roughness, pulse-on time is the major influential factor followed by servo volt and pulse current.

  2. 2.

    Optimal setting of process parameters obtained from the Taguchi analysis for MRR is Ip3Ton3Sv3, whereas for surface roughness optimal setting of process parameters are Ip3Ton2Sv1.

  3. 3.

    The error estimated between predicted MRR and surface roughness with experimental MRR and surface roughness at the optimal setting of process parameters is within ±5%.

  4. 4.

    Proposed Taguchi analysis is consistent as the error estimated between predicted and experimental values at an initial set of process parameters are within ±5%.