Modeling And Optimization of Methane Yield and H2S Reduction from Palm Oil Mill Effluent Using Artificial Neural Network

Abstract

The treatment of palm oil mill effluent (POME) through anaerobic digestion proposes a sustainable corresponding for renewable energy production in the form of methane-rich biogas. However, the performance of this process is highly dependent on several operational parameters including pH, temperature, and recirculation ratio which under industrial conditions can fluctuate. Conventional modelling approaches are limited in capturing complex and nonlinear interactions among these variables. In this study, an Artificial Neural Network (ANN) model was developed and optimized to predict methane yield and hydrogen sulphide (H₂S) concentration based on real industrial data collected from a biogas facility in Pahang (Chen et al.,2023) gathering over two-year period. The ANN model was constructed using MATLAB R2022a and trained using the Bayesian Regularization algorithm. Optimum architecture was identified as a feedforward neural network structure with one hidden layer containing 19 neurons and the log-sigmoid activation function. Excellent predictive performance of ANN model shown with a high regression coefficient (R = 0.99625) and a low mean absolute percentage error (MAPE = 3.53 %) in predicting methane yield and H₂S concentration. Sensitivity analysis revealed that pH was the most influential factor affecting both outputs. Based on the trained model, the optimum operating conditions for anaerobic digestion were determined to be a pH of 7.0934, a temperature of 38.86 ℃ and a recirculation ratio of 1.7817. The outcomes of this research highlight the reliability and practical applicability of ANN models for industrial-scale process prediction and optimization. This model has been developed using real plant data, while its predictions have been compared with actual values from the same dataset as its accuracy evaluation. It further highlights the great potential of models to support decision-making accurately and further improve industrial POME treatment of biogas production system.