Selective acid gas separation from diatomic nonmetal gas via ZIF-8 membrane: Taguchi analysis and neural network modeling

Abstract

ZIF-8 membranes are an option for separating acid gas from diatomic nonmetal gas, proposing an alternative technology for combating increasing greenhouse gas emissions and reducing climate change's harms. Synthesis and operational parameters are the critical factors that contribute to the upward trend in membrane performance in gas separation applications. In this study, the L8 (23) orthogonal array of the Taguchi method was adopted to identify the optimum conditions for separating acid gas from diatomic nonmetal gas. Three key parameters - seeding duration, growth time, and operating pressure were investigated at two levels each. From Taguchi analysis, the SN ratio and means are influenced by growth time, with a delta of 2.266, for CO2 flux. Meanwhile, the SN ratio and means for CO2/N2 ideal gas selectivity are impacted by seeding duration, with a delta of 4.190. Additionally, a feedforward artificial neural network (ANN) with three inputs, one hidden layer, and two outputs is employed to develop a predictive model. The findings indicated that the ANN successfully projected the CO2 flux and CO2/N2 ideal gas selectivity, with an R-value of 1 for training, validation, testing, and overall, respectively suggesting the validity of the model. Overall, customizing synthesis and operating parameters using the Taguchi method improves membrane performance and reduces variation, while the ANN model provides insight into forecasting acid gas separation from diatomic nonmetals application.

Keywords

• ZIF-8
• Acid gas
• CO2/N2 separation
• Taguchi analysis
• Artificial neural network