Evaluating the Impacts of Leachate Co-Treatment on Biological Nutrient Removal in a Full-Scale Municipal Wastewater Treatment Plant Using Machine Learning

Despite early evidence indicating that co-treating leachate with municipal wastewaters is beneficial, over the last decade more plants refusing to accept leachate for co-treatment. Leachate exhibits great temporal and spatial variations in physical and chemical properties and strength making predictions of process upsets difficult for operators. It can adversely affect water resource recovery facilities (WRRFs) biological nutrient removal (BNR) processes due to the presence of inert, nonbiodegradable chemical oxygen demand (COD) and recalcitrant dissolved organic nitrogen (rDON), which are not treated but mainly diluted increasing the difficulty of compliance with increasingly stringent discharge standards, particularly for total nitrogen (TN) limits. There is currently no framework that informs WRRF operators how much leachate they can accept without causing process upsets. Our objective was to evaluate the impacts of leachate co-treatment on a full-scale municipal wastewater treatment plan (WWTP) by comparing plant performance at varying levels of leachate contributions and hydraulic loadings. We used the K-Nearest Neighbors (KNN) machine learning algorithm to determine at which leachate volumes BNR was impacted. Leachate BOD:COD ratio was 0.08 +/- 0.07. The ratio indicated a stabilized, old matrix. Zinc, iron, aluminum, chloride, and sulfate concentrations were 0.174, 38, 1.47, 1803 and 119.1 mg/L, respectively. The average volumetric leachate ratio (VLR%) was approximately 0.01% corresponding to a daily volume of 30 meters cubed (m3) but reaching a maximum of 270 m3 (VLR% = 0.1%) and fluctuating on a daily basis. The cluster analysis revealed 5 VLR% groupings that were used for subsequent analyses: no leachate, 0 < Low ≤ 0.001, 0.001<Medium ≤ 0.02, 0.02<High ≤ 0.05, and 0.05< Very high ≤ 0.2. The completed study results showed that treated effluent concentrations of TKN, ammonia, fecal coliforms (FC), E. coli (EC), TSS and TP experienced a trend where effluent quality was improved at low and medium VLR% compared to no leachate addition, but deteriorated in high and very high VLR%. Treated effluent UVT% and EC were not statistically significantly different at varying VLR%, but FC was. Plant hydraulic had a significant impact on removal rates. Ammonia removals and nitrite concentrations improved in high flow conditions, while TP, BOD and cBOD removals deteriorated. Finally, VLR%, leachate COD, TKN ammonia, chloride and arsenic had significant relationships with plant performance. The study's findings mean that for leachate with comparable age and strength, VLR% should not exceed low to medium contributions (0 and 0.02%) during co-treatment at this WWTP.