Application of Grey Entropy and Regression Analysis for Modelling and Prediction on Tool Materials Performance During EDM of Hot Pressed ZrB2 at Different Duty Cycles

Abstract

In this paper, the regression analysis (RA) and grey relational analysis is used for evaluating performance of the various tool materials at two different duty cycles. Objective is to maximize the material removal rate (MRR) and to minimize the roundness, surface roughness (SR), tool wear rate (TWR), weight wear ratio (WWR) and taper angle during Electrical Discharge Machining (EDM) of hot pressed ZrB2. Most of the ceramic components are manufactured through powder metallurgy route at net shaped production, but some feature like holes of smaller diameter or complex features cannot be produced by this technique and hence needs additional machining. In this work, a ZrB2 disc of 100 mm diameter and 5 mm thickness is manufactured by means of hot pressing technique and this disc is machined using diamond load grinding to have parallel surfaces. Then, 2 mm diameter holes are machined on the disc using EDM spark erosion machine with different tool materials. Mathematical models are proposed for the modelling and analysis of the effects of Pulse on time and tool materials on the performance characteristics in the EDM using regression analysis (RA). Absolute fraction of variance (R2) for MRR, TWR, WWR, roughness, roundness and taper angle are 0.7622, 0.9939, 0.9796, 0.6810, 0.5049 and 0.7017 respectively. Since the variation of the roundness is more among the experiments it produces very low R2 value. Entropy method is employed to find the weights of the individual responses in Grey Relational Analysis (GRA) and they are 0.166056, 0.170019, 0.169181, 0.166227, 0.166205 and 0.166529 corresponding to MRR, TWR, WWR, roughness, roundness and taper angle respectively. The performances of tools are rated using entropy based grey relational analysis and best suited value of pulse on time also found out. Interaction of Pulse on time with various tool materials is investigated by using analysis of variance (ANOVA). The results have also been verified by running confirmation tests.