Prediction and Surface Roughness Optimization in Die-Sinking EDM of AA 5083 Using RSM and Swarm Intelligence Algorithms

As a non-conventional thermal material removal process, electro-discharge machining (EDM) is preferred when it comes to complex features and with high-precision contours as well as for materials that cannot be processed via conventional material removal operations. Nevertheless, several phenomena may adversely affect surface integrity of EDMed components, thus; they should be considered and experimentally investigated to optimize electro-discharge machining process. This paper experimentally examines surface integrity as regards surface roughness average Ra when using AA 5083 as a workpiece material and copper as electrode. Experiments were designed and carried out according to Taguchi L9 orthogonal design, setting three different levels of the process control parameters namely discharge (peak) current, pulse-on time and pulse-off time. Three algorithms namely moth-flame algorithm, dragonfly algorithm and whale optimization algorithm were employed to minimize the response. All algorithms performed adequately; however obvious differences in convergence speed and optimal solution results were found. The lowest result for surface roughness was found equal to 4.410 μm by DA and WOA algorithms at 35th and 5th iterations respectively, whereas DA was converged to its lowest score equal to 4.736 μm after 35 iterations.