Studies of the Performance, Stability and Reliability of Various Configurations of the Activated Sludge Process at Full-Scale Municipal Wastewater Treatment Plants

This research presents a performance, stability and reliability analysis of five configurations of activated sludge process, viz., conventional, contact stabilization, complete mix, extended aeration and sequencing batch reactor. Five years effluent records of carbonaceous biochemical oxygen demand (CBOD5), total suspended solids (TSS) and ammonia nitrogen (NH3-N) were used. The impact of hydraulic load classes (HLCes) and seasons on the performance of configurations were also examined. The Univariate General Linear Model with Tukey's Honestly Significant Difference was employed in the performance and stability studies. Performance was determined based on the observed lowest mean effluent concentrations, while the observed standard deviation from the mean effluent concentration was the basis for determining stability. Coefficient of Reliability and the Binary Logistic Regression (BLR) model were used to study reliability in terms of compliance with a daily average and single grab/daily max effluent values, and results were compared. Conventional aeration was the best performer with regard to CBOD5 removal, while sequencing batch reactor and extended aeration were the highest performers in TSS and NH3-N reduction, respectively. Contact stabilization was consistently observed to be the least performer under all effluent limitations examined. With the exception of NH3-N, where no statistical significance was observed, major HLCes were observed to perform better than minor hydraulic HLCes. Although, TSS and NH3-N reduction abilities were not affected by changes in seasons, the highest CBOD5 reduction was observed in summer and the least in winter. No statistical significance was observed with regard to effluent CBOD5 stability. However, sequencing batch reactors were observed to produce the most stable effluent TSS. Major HLCes showed more stability during NH3-N reduction. The Binary Logistic Regression model was compared to the COR and the BLR was found to be the better technique in determining reliability of wastewater treatment plants. Based on the BLR model prediction conventional, sequencing batch reactor and extended aeration were observed to be more reliable with regard to CBOD5, TSS and NH3-N, respectively. Major HLCes were more reliable than minor HLCes and daily average effluent limitations were more conservative compared to single grab/daily max effluent values.