Optimizing the PSA process of propylene/propane using Neuro-Fuzzy modeling

Abstract

Cryogenic distillation is the common method for separating propylene/propane mixtures, but this is highly energy intensive. Some 8-ring silica zeolites, especially pure silica chabazite (SiCHA), are known to show high diffusivity ratio for propylene over propane. In this work, we study the separation of propylene/propane using pure silica chabazite (SiCHA) in a simple 4-step pressure swing adsorption process. An isothermal isobaric micropore diffusion model with concentration-dependent diffusivity has been developed to simulate this kinetically controlled separation. It is first developed and implemented in the multi-physics software COMSOL to simulate different modes of PSA process. In this study, we present a sequential optimization strategy based on neuro-fuzzy model and genetic algorithm (GA) with synergistic combination of COMSOL simulation model to maximize the purity of propylene and propane productions. © 2012 Elsevier B.V.