PROCESS PARAMETER OPTIMIZATION FOR AMYLASE-ASSISTED BIOGAS GENERATION FROM ORGANIC WASTES: INTEGRATING RSM AND ANFIS FOR PREDICTIVE MODELING

Abstract

This study addresses the optimization of process parameters for amylase-enhanced biogas production from organic wastes, leveraging the synergistic application of Response Surface Methodology (RSM) and Adaptive Neuro-Fuzzy Inference System (ANFIS) for robust predictive modeling. The valorization of poultry droppings, pig manure, and brewery spent grain both as individual substrates and in co-fermentation was investigated under oxygen-free anaerobic digestion, with exogenous amylase serving as a hydrolytic promoter to improve substrate bioavailability. Comprehensive physicochemical characterization of the feedstocks was conducted using standard APHA protocols, while X-ray diffraction (XRD) analysis elucidated the crystalline structural phases, providing critical insights into the recalcitrant fractions influencing enzymatic hydrolysis efficiency. Batch-mode experiments systematically evaluated the interactive effects of four independent variables: fermentation temperature, hydraulic residence time, enzyme concentration, and pH.
Optimal reaction parameters for individual feedstocks were established at 42.5 °C, 30 days, 4 μmol/L amylase, and pH 7.0, whereas the co-fermentation of combined substrates required more intensified conditions, with optima recorded at 50 °C, 20 days, 6 μmol/L, and pH 8.0, reflecting the complex synergistic interactions within mixed organic matrices. The comparative optimization performance of RSM and ANFIS was rigorously validated through coefficient of determination (R²) analysis, demonstrating exceptional concordance between experimental yields and model predictions. For mono-substrate systems, ANFIS (R² = 0.981) exhibited marginally superior predictive accuracy over RSM (R² = 0.974); similarly, for co-digestion setups, ANFIS (R² = 0.973) outperformed RSM (R² = 0.965). These high correlation coefficients affirm the reliability of both modeling approaches, with ANFIS demonstrating heightened aptitude for capturing the intricate, non-linear biological dynamics inherent in enzymatic digestion. Collectively, these findings substantiate that the integrated RSM-ANFIS framework constitutes a powerful, decision-supportive tool for precisely forecasting fermentable biogas yields. This optimization strategy not only enhances methane generation efficiency from diverse agro-industrial residues but also provides a scalable, data-driven blueprint for sustainable waste-to-energy conversion, advancing both renewable energy recovery and circular economy objectives.

Keywords