(223f) Experimental Validation of PSA Optimization Techniques

Advancements in numerical and optimization techniques, and the availability of cheap computational power have had a significant impact on PSA design. It is now very common to see rigorous optimization techniques being used in conjunction with detailed process models. These techniques are being employed to compare different process configurations, screen materials etc. However, there is a dearth of experimental studies that validate these optimization results. This presentation seeks to bridge this important gap.

In this work, the vacuum swing adsorption (VSA) separation of CO₂ from a mixture of 15% CO₂ and N₂ on Zeolite 13X is considered. Volumetric and gravimetric methods were used to measure the pure component isotherms, while breakthrough experiments were used to measure the competitive isotherms. Based on these input parameters that were obtained from a few milligrams of material, a detailed model was developed to simulate two cycles: A simple 4 step cycle consisting of feed/co-current blowdown/counter-current blowdown/ feed pressurization; and feed/co-current blowdown/counter-current blowdown/ light product pressurization. A genetic algorithm based optimization algorithm was used to calculate the Purity-Recovery Pareto curve. The Pareto curves clearly demonstrated the superiority of the cycle with the light-product pressurization over feed-pressurization. Thee points from each Pareto curve were chosen and the operating parameters, as predicted by the model, were translated to an experimental two-column PSA rig that contained 324g of adsorbent. The results showed that the performance indicators (Purity and recovery) and the transients of various process variables (flows, compositions, temperatures) predicted by the model were well reproduced by the experiments. The study demonstrates the reliability of optimization techniques for the design of PSA systems.