A data-driven optimization model for coagulant dosage decision in industrial wastewater treatment

Abstract

Industrial wastewater treatment is one of the critical issues faced by water utilities. With the dynamic and complex characteristics, to properly utilize the wastewater treatment and optimize its use is particularly a challenge. A coagulation-flocculation process is a widely used method in wastewater treatment where the complexity of the coagulant chemical theory shows the difficulties to determine the coagulant dosage efficiently. In the current study, a data-driven approach that combines a genetic algorithm-based optimization and particle swarm optimization technique with the regression model analyses was implemented to optimize the coagulant dosage. By so doing, the effluent quality and the sludge level generated from the printed circuit board manufacturing wastewater plant were improved. To evaluate the performance of the model, statistical analysis to determine the significance of the parameters and the determination coefficient R2 has been used, and the Cu removal regression model showed good agreement (R2= 99.97%). During the optimizing process, the model improved 27% of the emission standard target for outgoing Cu while assuring the quality of sludge level at the same level. The results also demonstrated that the proposed data-driven approach lowers cost by 10% in the examined wastewater treatment plant. Through the proper treatment and the control of the chemical dosage for Cu removal based on practical data, the proposed model can be helpful to the engineering applications.