Intelligent dosing of sodium hypochlorite in municipal wastewater treatment plants: Experimental and modeling studies

Abstract

Sodium hypochlorite (NaClO), a disinfectant extensively employed in municipal wastewater treatment plants (WWTPs), is frequently overdosed to ensure qualified effluent quality consistently. In light of the escalated disinfectant expenses and detrimental environmental impacts, it is critical to have a precise control model in place to enable intelligent dosing of NaClO. The lack of real-time disinfection evaluation indicators presents difficulties in precisely and swiftly regulating the dosage. Herein, the CT value (the product of residual chlorine concentration and contact time) was initially determined experimentally as a metric for assessing disinfection in municipal WWTPs. Fecal coliform (FC) concentrations become less than 1.0 × 103 MPN/L when the effluent CT value is regulated at 0.20 mg·min/L (calculated as Cl2). In addition, a model employing long short-term memory (LSTM) was developed to predict the dosage, utilizing simplified and suitable parameters such as COD, NH3-N concentration, flow rate, and historical dosage of NaClO as input factors, which demonstrated satisfactory performance (R2 = 0.948). Notably, dosage adjustments were made concurrently in response to CT value feedback. The intelligent dosing model under simultaneous feedforward and feedback control was applied at a full-scale WWTP for nearly two months, considerably reducing disinfectant consumption and meeting the discharge standard of FC.