An optimization strategy to improve performance in electrochemical discharge machining of borosilicate glass using graph theory algorithm and desirability index

Electrochemical discharge machining (ECDM) a combined version of electrical discharge machining (EDM) and electrochemical machining (ECM), is an emerging alternative method to shape low machinability material like borosilicate glass. Since the material removal rate (MRR) decreases with the machining depth due to insufficient electrolyte at the tool tip, the radial over cut (ROC) is found at the entrance due to accumulation of electrolyte. To overcome these issues, a hybrid electrolyte (HE) of NaOH+ 5% KOH is proposed in the present study, because the HE has low viscosity and high electrical conductivity compared to individual conventional electrolyte (CE). Experiments are conducted with an in-hose developed ECDM set-up integrated with metal oxide semiconductor field effect transistor (MOSFET) assisted power module. Experimental plan developed based on L9 orthogonal array with three process variables, electrolyte concentration, applied voltage, and duty factor at three levels, taper angle (TA) and ROC considered as quality characteristics and MRR is considered as performance characteristics. In this paper, a novel simultaneous multi-response optimization technique by combining Graph theory algorithm (GTA) and desirability function approach (DFA) is presented. Weights for the responses are calculated using the GTA and are used to evaluate overall (composite) desirability index (ODI) and it is treated as the multi response performance index (MRPI) to determine optimum levels for the process variables. ANOVA test indicated that DF has more influence on machining performance with 47.16% for CE and 57.06% for hybrid electrolyte. An increased MRR, and a reduction in OC and TA was noticed with HE compared to CE. In addition, SEM images revealed a smooth machined surface texture with CE and a feathery-like machined surface texture with HE. The average of the response values from the confirmation tests were found to be within the ±5% of predicted mean value, and the optimal values obtained from the proposed optimisation technique are confirmed.