This article aims at demonstrating the interest the design of experiment methodology might present for modeling and optimization of a typical electrostatic separation process. A classical 23 full-factorial design followed by a composite design were employed for carrying out experiments on the selective sorting of conductive and nonconductive particles contained in a granular mixture. The study was performed on a roll-type corona-electrostatic separator, with samples prepared from genuine electric cable wastes, composed of 5% copper and 95% PVC. The response function was the mass of the middling fraction collected after separation, the factors under investigation being the high-voltage level, the speed of the rotating roll electrode, and the angular position of the splitter between the copper and middling compartments of the collector. By using commercial design of experiments software, it was possible to estimate the effects of these factors and then derive the model of the process as a quadratic polynomial function. This model served at predicting the optimal set point of the process. The results reported support design of experiments methodology as an efficient tool for the optimization of industrial electrostatic separation processes.
