Resolving stability issue of thermophilic high-rate anaerobic palm oil mill effluent treatment via adaptive neuro-fuzzy inference system predictive model

Abstract

Abstract Palm oil mill effluent is a highly polluting and corrosive liquid waste generated from crude palm oil processing. Its varying characteristics depending on the crop season and loading rate of the milling process complicate the handling of anaerobic digestion of palm oil mill effluent. To ease monitoring and aid decision making to handle operation anomalies, this study adopts Adaptive Neuro-Fuzzy Inference System with hybrid learning algorithm and subtractive clustering to predict the output from a thermophilic high-rate anaerobic digester treating palm oil mill effluent. Influent pH, chemical oxygen demand, organic loading rate and historical data of effluent pH, chemical oxygen demand and total suspended solids were utilized to build the model for output prediction of treated effluent (pH, chemical oxygen demand and total suspended solids). Trained models were validated with experimental data. The models were able to predict the effluent pH, chemical oxygen demand and total suspended solids satisfactorily with average error of less than 8.32% and standard deviation of less than 7.65% with closely followed pattern between the actual values and simulated results. The viability of adaptive neuro-fuzzy inference system in modelling palm oil mill effluent using thermophilic high-rate anaerobic digester is further demonstrated by coefficient of determination, R2 of 0.82, Root Mean Square Error of 0.0377, Index of Agreement of 0.95 and d-factor of 0.0496.