Comparative investigation of novel dielectrics against cryogenically refined electrodes for modelling and optimizing EDM cutting proficiency using artificial neural network

Superalloys, specifically Inconel 617 (IN617), have distinctive properties, yet it is extremely hard to machine through traditional processes. Electric discharge machining (EDM) has been considered a viable option considering the characteristics and shape intricacy required in the applications of the said alloy. However, EDM’s low material removal rate (MRR) impedes its use. Therefore, the machining performance of five different transformer oil (TO) based modified dielectrics and cryogenically treated (CT) tools have been thoroughly investigated herein, which has never been examined so far. A total of 30 experiments were executed employing a full factorial design considering three kinds of electrodes and five different modified dielectrics. CT Cu (Cu CT ) electrode has given the highest magnitude of MRR (19.65 mm 3 /min) in the blend of tween 80 and TO, and minimum surface roughness (R z ) 11.0 µm in pure TO. The maximum MRR obtained with the CT electrode is 28% greater in comparison to that recorded for non-treated (NT) electrodes. Among the different electrodes, the Cu CT tool performed well in TO-based dielectrics by giving the high MRR compared to the other NT and CT electrodes. CT electrodes yield higher MRR in TO-based dielectrics than kerosene oil (KO) based ones. The highest value of MRR noticed with TO-based dielectric is 20.3% than the maximum MRR noted for the KO-based dielectric. A significant amount of improvement in the MRR (98.56%) and R z (78.88%) has been observed when the mono-objective optimization is utilized to obtain the optimal set of input variables. Finally, an artificial neural network (ANN) has been modelled to accurately forecast the MRR and R z value for various input conditions, obviating the necessity for tests, effectively modelling the complicated and nonlinear phenomenon of MRR and R z .