Artificial Neural Network-Based Modeling of Membrane Contractors for Industrial Gas Treatment

Gases produced after the combustion of fuels and other industrial processes contain H2S gas, which has several adverse effects on human health and the environment. Thus, H2S needs to be removed from flue gases before disposal. In this study, the artificial neural network-based modeling method is used to analyze the performance of membrane contractors for the removal of H2S from flue gases. H2S selectivity, H2S concentration in outlet stream, and percent CO2 removal were selected as output parameters, and CO2 concentration in feed and gas to liquid ratio was selected as input parameters to study the performance of the absorption unit. First, ANNs were trained with the experimental results, and the number of nodes in trained ANNs was optimized for minimum square mean errors. Trained ANN was then successfully validated with the experimental results. Simulation results showed that higher CO2 concentration in feed and higher gas to liquid ratio improved the H2S selectivity. However, higher gas to liquid ratio also increased the amount to H2S in the output stream. Both input parameters did not significantly affect CO2 removal as absorbent Na2CO3 is selective for H2S gas only.