A linear-interaction model for electrostatic separation processes

Previous studies based on Box-Wilson and response surface methods have proven that the design of experiments is a powerful tool in improving electrostatic separation performances by controlling the two main parameters of the process: the high-voltage level and the roll speed. The aim of the present paper is to analyze the possibility of deriving a mathematical model capable to reflect the effects of a larger number of factors, as well as their main interactions. At first, the main variables of the process were listed and classified in accordance with the ease of controlling them. Then two experimental designs were chosen, in accordance with Taguchi's methodology. The objective was to minimize the middling fraction. The experiments were carried out on a laboratory roll-type electrostatic separator, provided with a corona electrode and a tubular electrode, both connected to a DC high-voltage supply. The samples of processed material were prepared from genuine chopped electric wire wastes (granule size >1 mm and <2 mm) containing 25% copper and 75% PVC. The first experiment consisted of 16 tests, which enabled the derivation of a linear-interaction model comprising 7 variables and 8 interactions. The second experiment consisted of only 8 tests, as the corresponding model took into consideration 4 variables and 3 interactions. A good agreement was obtained between the two models.