Aromatics/Alkanes separation: Simulated moving bed process model development by a concurrent approach and its validation in a mini-plant

Abstract

Abstract SMB process development for multicomponent systems is often challenged by the difficulty of optimizing and validating a complex adsorptive separation process whose performance is strongly dependent on multicomponent nonlinear adsorption behavior. We recently proposed a “concurrent” iterative approach to simultaneously optimize SMB process design while determining the adsorption characteristics, wherein a multi-component Langmuir isotherm approximates more complex adsorption behavior in a certain range of liquid phase compositions that are relevant to the SMB design. In this report, the concurrent approach was applied to the separation of a model aromatics/alkanes mixture that approximates hydrocarbon streams often encountered in petroleum refining. In particular, we consider a nonlinear multicomponent system of toluene, dodecane and cyclohexane which have strong competition for adsorption in a mesoporous silica adsorbent. Using an approach that combines adsorption measurements, model fitting, and SMB model predictions that allow targeted selection of compositions for adsorption measurements, we demonstrate convergence of the concurrent method in three iterations with only a small number (32 in total) of multicomponent adsorption measurements. Then, we demonstrate a second iterative methodology to refine the SMB model by estimating the non-thermodynamic (i.e., mass transfer) and other (dead volume) system parameters, using only a very small number of runs on a 16-column SMB “mini-plant” designed and constructed in-house at the Georgia Institute of Technology. The final process design, validated in the SMB mini-plant, maximizes the throughput while maintaining the desired aromatic purity and recovery within an error of 1.5%. The present work thus demonstrates the idea that the SMB process can be developed and experimentally validated in an accelerated manner without requiring prior knowledge of the “global” multicomponent adsorption characteristics.