Nitrogen modeling and performance of Multi-Soil-Layering (MSL)bioreactor treating domestic wastewater in rural community

The multi-soil-layering (MSL) bioreactor has been considered in the latest research as aninnovative bioreactor for reducing the level of pollutants in wastewater. The efficiency of theMSL bioreactor towards nitrogen pollution is due to the mineralization of organic nitrogen inaerobic layers to ammonia, and reactivity of ammonia nitrogen with soil and gravel by itsadsorption into soil layers followed by nitrification and denitrification processes when thealternating phases of oxygenated/anoxic conditions occurs in the filter. In this study, we haveexamined the performance of the MSL bioreactor at different hydraulic loading rates (HLRs)and predicted the removal rate of nitrogen. To improve the prediction accuracy of the models,the feature selection technique was performed before conducting the Neural Network model.The results showed a significant removal (p <0.05) efficiency for five-day biochemicaloxygen demand (BOD 5,  86%), ammonium (NH 4 + , 83%), nitrates (NO 3 − , 81%), total kjeldahlnitrogen (TKN, 84%), total nitrogen (TN, 84%), orthophosphates (PO 4 3− , 91%), and totalcoliforms (TC, 1.62 Log units). However, no significant change was observed in the nitrite(NO 2 − ) concentration as it is an intermediate nitrogen form. The MSL treatment efficiencydemonstrated a good capacity even when HLR increased from 250 to 4000 L/m 2 /day,respectively (e.g., between 64% and 86% for BOD 5 ). The HLR was selected as the mostsignificant (p < 0.05) input variable that contribute to predict the removal rates of nitrogen.The developed models predict accurately the output variables (R 2  > 0.93) and could help toinvestigate the MSL behavior.