Recast Layer Thickness and Residual Stress Analysis for EDD AA8011/h-BN/B4C Composites Using Cryogenically Treated SiC and CFRP Powder-Added Kerosene

• Published in: Arabian journal for science and engineering (2011) 2022, Vol.47 (12), p.15613-15632
• Main Authors: Vivek, J., Maridurai, T., Lewise, K. Anton Savio, Pandiyarajan, R., Chandrasekaran, K.
• Format: Article
• Language: English
• Subjects: Adaptive systems; Aluminum base alloys; Artificial neural networks; Boron carbide; Composite materials; Dielectric properties; Electric discharges; Engineering; Fuzzy logic; Humanities and Social Sciences; Kerosene; Magnesium; Mathematical analysis; Modelling; multidisciplinary; Prediction models; Pulse duration; Research Article-Mechanical Engineering; Residual stress; Root-mean-square errors; Science; Silicon carbide; Silicon dioxide; Statistical inference; Stress analysis; Surface roughness; Thickness; Variance analysis
• ISSN: 2193-567X; 1319-8025; 2191-4281
• DOI: 10.1007/s13369-022-06636-5

The conventional kerosene and powder-added kerosene dielectric performances have been found to degrade in electrical discharge drilling (EDD) efficiency owing to dielectric properties variation, more debris formation, high recast layer thickness (RLT) formation, low residual stress (RS) induced and magnesium and silicon dioxide formation on the machined surface. The cryogenically treated SiC and CFRP powder-added dielectric in EDD on AA8011/h-BN/B 4 C composites have been carried to evaluate the RLT, RS and surface roughness (SR). Hence, the circular holes have been performed. An analysis found that cryogenically treated SiC powder-added dielectric has a higher improvement percentage than that of conventional kerosene and CFRP powder-added kerosene. Cryogenically treated SiC powder-added kerosene shows an improved performance measure of 62.2% in RLT, 22% in RS and 48.8% in SR of composite. Analysis of variance (ANOVA) found that the pulse duration and current are the most significant factors. Second-order and adaptive neuro-fuzzy inference system predictive models have developed. The statistical indices such as mean absolute percentage error (MAPE), root mean square error (RMSE) and coefficient of determination ( R ) are calculated to evaluate the modelling and found that low MAPE, RMSE and high R were obtained in both modelling. The complex proportional assessment (COPRAS) method exhibits an improved performance measure of 46.15% in RLT, 5.71% in RS and 21.42% in SR of composite.