Prediction and optimization of surface quality and material removal rate in wire-EDM cutting of tungsten–copper alloy (W70Cu30)

Abstract

Tungsten-copper alloy (W70Cu30) is widely used in industrial applications due to its high thermal conductivity, melting point, and wear resistance. This study investigates its machinability using wire electrical discharge machining (EDM). Process parameters, including current, servo voltage, wire feed, and wire tension, were optimized to evaluate their impact on Material Removal Rate (MRR) and Surface Roughness (Ra). The Taguchi L16 orthogonal array was employed to design experiments with 16 square samples, each representing a unique combination of process parameters to identify optimal machining settings. Predictive modeling was conducted using Support Vector Regression (SVR) and gradient-boosted regression Trees (GBRT) to assess the accuracy of MRR and Ra predictions. Experiments were carried out on a precision wire EDM setup, followed by microscopic analysis to evaluate surface integrity and machining defects. The optimal parameters for maximum MRR were current 5 A, servo voltage 60 V, wire tension 9 N, and wire feed 12 mm/min. For minimum Ra, the best settings were current 3.5 A, servo voltage 40 V, wire tension 6 N, and wire feed 6 mm/min. SVR outperformed GBRT, with R2 values of 0.977 for MRR and 0.944 for Ra, demonstrating high predictive accuracy. While GBRT excelled for MRR with an R2 value of 0.996, its predictions for Ra were less accurate. Microscopic tests were also performed to inspect the machining surface for flaws and evaluate surface quality. The overview summarizes our systematic approach to improving wire EDM settings, using advanced predictive models, and performing detailed microscopic investigations to enhance production precision.

Keywords

• Tungsten copper alloy (W70Cu30)
• Wire EDM
• Material removal rate (MRR)
• Surface roughness
• Support vector regression (SVR)
• Gradient boosted regression trees (GBRT)