Examination of possible energy conservation in a biological water treatment process using a multiple regression model

This study examined various controls and operating parameters of a sewage treatment plant by performing multiple regression analyses on the operating parameters and the effluent quality of the sewage treatment plant. For the examination, operating data from April to November, 2012, was collected from a sewage treatment plant using the media anaerobic-anoxic-oxic method. The chemical oxygen demand (COD_Mn) and total nitrogen (T-N) forecasting models for the secondary sedimentation basin effluent were built through multiple regression analysis and showed Mean Absolute Percentage Error (MAPE) 6.6–9.3 and 16.0–23.7, respectively. All models showed similar results to real observational data. When controlling the operating parameters of COD_Mn and T-N, by using the regression model and the standardization regression coefficients without violating legal water quality criteria, operating conditions can be set that save an average of 23% of power consumption. Using this result, an operating guide for low energy consumption can be provided to the operators of sewage treatment plants.