Optimization Of Material Removal Rate And Surface Roughness Using The Taguchi Method
Matoorian, Pooria (2008) Optimization Of Material Removal Rate And Surface Roughness Using The Taguchi Method. Masters thesis, Universiti Putra Malaysia.
A non-conventional hybrid machining method called electrical discharge turning (EDT) process is optimized in this research. The EDT process is a suitable method to produce small components with cylindrical geometries. This process is a type of hybrid electrical discharge machining (EDM), hence the material is removed by the action of electrical discharges between the tool electrode and the workpiece. It means materials of any hardness can be removed as long as the workpiece can conduct electricity. This makes the EDT process suitable for machining hard, difficult-to-machine materials. In this process linear geometry of a tool electrode reproducing the same geometry in the rotating workpiece cylindrically. In this study, a dressed copper block (8mm × 10mm × 50mm) serving as the forming tool electrode is fixed on the work table and rotary workpiece uses the rotational motion of 4th (C) axis of the machine. The Taguchi Robust Design method was used to determine the optimum machining performance namely the highest material removal rate (MRR) and the lowest surface roughness (SR) for EDT of High Speed Steel (HSS) 5%-Cobalt. Six control factors namely, Intensity, Pulse-on time, Pulse-off time, Voltage, Servo, and Spindle speed were considered. Based on the analysis of variance (ANOVA) all six factors were influential for MRR but for SR rotational speed did not show any influence. Intensity was the most significant factor for both response of MRR and SR. Signal to Noise (S/N) analysis was performed and optimum levels of the mentioned factors for highest MRR and the lowest SR was achieved based on the S/N ratios. Results of confirmation tests shown the improvement of MRR and SR in optimum condition were 9.17 and 6.54 dB respectively. Finally general linear regression models were derived for 95% confidence interval to predict the output response. The p-value for the used α-level of 0.05 concluded that at least one of the regression coefficients is significantly different from zero and the linear predictors are not sufficient to explain the variation.
Repository Staff Only: Edit item detail