Prediction of metal ion rejection in electro-cross-flow ultrafiltration using an artificial neural network

We studied an electro-cross flow ultrafiltration system that uses charged and neutral ultrafiltration membranes and changes in pH and voltage to Pb2+, Ca2+ and Fe2+. Simulation of the experimental data was performed with the aid of an artificial neural network (ANN), to obtain a mathematical model to predict metal ion remotion. Analysis of the experimental data indicates that the surface charge of the membrane does not affect the removal process. However, the neutral membrane (SP1) has a higher flux (0.293 L/m2 s) than the charged membrane (AC1) (0.271 L/m2 s). Our results also indicate that the metal ions studied are efficiently removed by applying voltage. In all cases, by applying a voltage of 1.0 V for 30 min to an AC1 membrane, metal ion removal was well over 90%. The ANN model developed properly adjusted the experimental data with a non-linear model and allowed us to predict with a standard deviation no greater than 10% the removal rate as a function of voltage and time. In addition, a simplified mo...