부산대학교 도서관

In the present research, electric discharge machining (EDM) was performed on Inconel 625, a nickel-based superalloy by mixing graphite micro powder in dielectric fluid during machining. A Box-Behnken type RSM design was constructed for the experimental work varying five input parameters viz. current, pulse on time, duty cycle, powder concentration and mesh size of the powder to study their effects on material removal rate (MRR), tool wear rate (TWR) and surface roughness (SR). Further, back propagation type artificial neural network (ANN) was developed, trained and tested with the experimental data. Multi response desirability function was applied to the RSM model to maximize MRR, optimize TWR at a stipulated target value and minimize SR. The analysis of the experimental result indicates that current and powder concentration have the extreme influence on all the three responses. The obtained ANN model provides a mean error of prediction as 5.95%, 7.30% and 6.40% for MRR, TWR and SR respectively. The overall R value of the model is found as 0.945 which indicates that the ANN model is adequate to predict the responses for other combination of input parameters. From the desirability function analysis, the optimized value of MRR, TWR and SR are found as 26.135 mm3/min, 0.0800 mm3/min and 5.81 µm at current = 8.6 A, pulse on time = 35 µs, duty cycle = 0.5, powder concentration = 8 g/l and mesh size at 35 µm. The study of surface characterization of the Inconel 625 samples have been performed using AFM, SEM and XRD analysis. The AFM and SEM study revealed that the surface irregularities and the width of microcracks increase continuously with the increase of current. [ABSTRACT FROM AUTHOR]