Evaluation of model approximations in simulating pressure swing adsorption solvent vapor recovery

Sixteen different mathematical models, based on all combinations of four major assumptions (i.e., frozen solid phase during pressurization/blowdown, isothermal, equilibrium, and constant gas phase velocity during constant pressure steps), were evaluated in simulating a pressure swing adsorption-solvent vapor recovery process, which was representative of any Langmuirian system utilizing a Skarstrom-type cycle. The evaluation was based on the bed dynamics and process performance predicted by each model, and obtained from a 24 full factorial design. Overwhelmingly, the results showed that the predictions of the process dynamics and performance from the 16 different models varied widely, and depended on which combination of assumptions was applied. Qualitative trends, based on the factorial analysis, indicated that both the constant velocity and frozen assumptions caused an overprediction in the solvent vapor enrichment and the bed capacity factor; and significant interaction effects existed between these two assumptions. Also, all of the models that assumed local equilibrium underestimated both the solvent vapor enrichment and the bed capacity factor, whereas all of the isothermal models, overestimated the solvent vapor enrichment but underestimated the bed capacity factor.