Application of a combined response surface methodology (RSM)-artificial neural network (ANN) for multiple target optimization and prediction in a magnetic coagulation process for secondary effluent from municipal wastewater treatment plants.

Abstract

In this study, an enhanced coagulation-flocculant process incorporating magnetic powder was used to further treat the secondary effluent of domestic wastewater from a municipal wastewater treatment plant. The purpose of this work was to improve the discharged water quality to the surface water class IV standard of China. A novel approach using a combination of the response surface methodology and an artificial neural network (RSM-ANN) was used to optimize and predict the total phosphorus (TP) pollutant removal and turbidity. This work was first evaluated by RSM using the concentrations of coagulant, magnetic powder, and flocculant as the controllable operating variables to determine the optimal TP removal and turbidity. Next, an ANN model with a back-propagation algorithm was constructed from the RSM data along with the non-controllable variables, raw TP concentration, and raw water turbidity. Under the optimized experimental conditions (28.42mg/L coagulant, 623mg/L magnetic powder, and 0.18mg/L flocculant), the TP and turbidity removal reached 88.79 ± 5.45% and 63.48 ± 9.60%, respectively, compared with 83.28% and 59.80%, predicted by the single RSM model, and 87.71 ± 5.74% and 64.62 ± 10.75%, predicted by the RSM-ANN model. The treated water were 0.17 ± 6.69% mg/L of TP and 2.46 ± 5.09% NTU of turbidity, respectively, which completely met the surface water class IV standard (TP < 0.3mg/L; turbidity < 3 NTU). Therefore, this work demonstrated that the discharged water quality was completely improved using the magnetic coagulation process. In addition, the combined RSM-ANN approach could have potential application in municipal wastewater treatment plants.