“…Preliminary experiments allowed us to select four factors among eight potentially influent factors (pH, coagulant dosage, coagulation time, mixing speed in the coagulation step, dosage of flocculant, time flocculation, mixing speed in the flocculation step and decantation time). To optimize the levels of the selected variables, one can use the conventional experimentation method that involves changing one variable at a time, and maintaining the others fixed [5,10,11]. However, this experimentation method requires many experimental runs, that are time-consuming, and ignores the interaction effects between the parameters and leads to a low optimization efficiency [12].…”