Simulation of Ammonia Nitrogen Removal from Simulated Wastewater by Sorption onto Waste Foundry Sand Using Artificial Neural Network

Abstract

The present study investigated the removal efficiency of ammonia nitrogen from simulated wastewater by waste foundry sand based on 120 batch experiments which were modeled by three-layer artificial neural network technique. Contact time (5-120 min), pH of the aqueous solution (3-10), concentration (400-600 mg/L), sorbent dosage (20-120 g/100 mL) and agitation speed (50-250 rpm) were studied. Results showed that the best values of the above parameters were time of  90 min, pH= 10, 400 mg/L, dosage of 90g/100 mL and 200 rpm respectively with removal efficiency equals to 95%. The sorption process was described in a good manner using ANN model which consisted of the tangent sigmoid and linear transfer functions at hidden and output layers respectively with 8 neurons and the maximum sorption capacity was 0.9 mg/g. The sensitivity analysis signified that the relative importance of contact time equal to 36.9% and it is the influential parameter in the sorption of ammonia nitrogen. However, the relative importance of other parameters was agitation speed of 27.43%, WFS dosage of 17.32%, pH of 9.86% and initial concentration of 9.39%.