Sensitivity analysis and artificial neural network-based optimization for low-carbon H2 production via a sorption-enhanced steam methane reforming (SESMR) process integrated with separation process

Abstract In this study, a sensitivity analysis was performed for an integrated SESMR process, and an optimization approach was formulated by developing an artificial neural network-based optimization (ANN-based optimization). The process comprised a cyclic fluidized bed (CFB), pressure swing adsorption (PSA), compressor, dehydrator, and other units. The PSA variables considerably affected product quality, while the CFB variables mainly contributed to other performance parameters. From the data analysis and domain knowledge, three main objectives and five main variables were selected for the process optimization. Thereafter, the ANN models were integrated with the economic model to formulate a SESMR-driven model for optimization. At the optimum conditions, the cost (1.7 $/kg) of the H2 (+99.99% purity) with 90.3% CO2 capture from the integrated SESMR process was 15% reduction compared to that of the SMR process, which agreed well with the US Department of Energy prediction (15–20%). These results suggest that the integrated SESMR process is valuable for the production of blue H2, and the ANN-based optimization is very effective for a complex integrated process.