Modelling of Biomass Gasification Through Quasi-Equilibrium Process Simulation and Artificial Neural Networks

Abstract

In the past two decades, advancements in thermochemical technologies have improved biomass gasification for distributed power generation, enhancing efficiency, scalability, and emission control. This study aims to optimize syngas production from biomass gasification by comparing two computational models: a quasi-equilibrium thermodynamic model implemented in Aspen Plus and an artificial neural network (ANN) model. Operating at 850 °C with varying steam-to-biomass (S/B) ratios, both models were validated against experimental data. Results show that hydrogen concentration in syngas increased from 19.96% to 43.28% as the S/B ratio rose from 0.25 to 0.5, while carbon monoxide concentration decreased from 24.6% to 19.1%, consistent with the water–gas shift reaction. The ANN model provided rapid predictions, showing a mean absolute error of 3% for hydrogen and 2% for carbon monoxide compared to experimental data, though it lacks thermodynamic constraints. Conversely, the Aspen Plus model ensures mass and energy balance compliance, achieving a cold gas efficiency of 95% at an S/B ratio of 0.5. A Multivariate Statistical Analysis (MVA) further clarified correlations between input and output variables, validating model reliability. This combined modelling approach reduces experimental costs, enhances gasification process control and offers practical insights for improving syngas yield and composition.