Prediction of Torque in Drilling Woven Jute Fabric Reinforced Epoxy Composites Using the Adaptive Network-Based Fuzzy Inference System and Response Surface Methodology

Abstract

Jute fiber reinforced epoxy (JREp) composites were prepared by the compression moulding technique by varying the fiber content (0, 20, 30 and 40 wt%). Fabricated JREp composites were subjected to a drilling study to observe the impact of factors such as spindle speed (rpm), feed rate (mm/min) and fiber content (wt%) on the output response – torque. A set of experiments were designed and conducted as per Taguchi’s Design of Experiment. The obtained torque results were found in the range from 14.84 to 32.28 N-m. The minimum value of torque was achieved for the composite drilled using an HSS twist drill (90°-point angle) at a high spindle speed (3000 rpm), with low feed rate (25 mm/min) on low fiber loaded JREp composite (20JREp). ANOVA analysis showed that the developed regression model was fairly significant and torque was mainly influenced by the feed rate. Mathematical models were developed for drilling JREp composites using response surface methodology (RSM) and adaptive neuro fuzzy inference system (ANFIS), and compared for their efficacy. The coefficient of determination (R2) values for RSM and ANFIS were 0.9778 and 0.9982, respectively, which conveys that both models were beneficial to predict the torque. The average checking error percentage (0.0000222) was obtained for the ANFIS model trained using ‘gbellmf’ membership function with 100 epochs. FESEM images of the drilled surface were captured to analyse the mode of failure endured by the JREp composites.