Simulation and optimization of hydrogen production from biogas by integration of DWSIM and artificial intelligence techniques

초록

Hydrogen (H2) production from biogas via steam methane reforming (BSMR) offers a viable route for renewable energy utilization and emission reduction. This study presents an integrated framework combining the DWSIM simulator with artificial intelligence for BSMR simulation and optimization. A validated DWSIM model was represented by an artificial neural network (ANN) surrogate model incorporating operating conditions, feedstock characteristics, and equipment efficiencies to predict H2 production, methane conversion, and energy performance. Bayesian hyperparameter tuning identified an optimal ANN architecture, achieving R2 approximate to 1 and NRMSE below 5.92E-03. Sensitivity analysis revealed biogas flow rate, temperature, and H2 recovery as dominant parameters for productivity, while heat exchange efficiency and water-to-biogas ratio primarily govern energy efficiency. Particle swarm optimization and paretosearch algorithm were applied for single- and multi-objective optimization, respectively. A trade-off between H2 production and energy efficiency was identified; co-optimal conditions yielded an H2 yield of 3.54 kmol-H2/kmol-CH4 and energy efficiency of 91.4%.

키워드