Design of simulated moving bed chromatographic processes for biochemical purification
Abstract
A model-based design approach was first developed for a linear SMB system for two amino acid fractionation. A PVP resin was chosen as the adsorbent with water as the desorbent. The solute isotherms and the mass-transfer parameters were estimated using batch chromatography. Pilot scale SMB experiments were then conducted using the operating conditions calculated from a standing wave analysis. The estimated parameters were validated by comparing the SMB experimental data with theoretical predictions. After the validation, the parameters were used to explore various design alternatives. This approach significantly reduced the number of trial-and-error experiments in SMB process design and development. The design with the consideration of mass-transfer effects gave higher product purity and recovery. A systematic optimization strategy was then applied to another linear SMB system for the separation of paclitaxel and cephalomannine. Four design parameters—desorbent composition, column configuration, column length to diameter ratio, and adsorbent particle size—have been analyzed to maximize throughput and minimize desorbent consumption. The effects of flow rate and pressure drop limitations were also considered in this analysis. The comparison between batch elution and SMB chromatography showed that SMB can achieve higher production rate and higher yield with less solvent usage. By using the model-based design method and the systematic optimization strategy, a SMB process for the separation of paclitaxel from four major impurities was developed. Two SMB steps were adopted in this process. Experimental data were in close agreement with the theoretical predictions from VERSE-LC. Finally, VERSE-LC was extended for the dynamic studies of nonlinear SMB systems. It was first benchmarked with the experimental data for large-scale sugar separation, and then used to study the dynamic behaviors and mass-transfer effects for this system. A CSTR model has been applied to simulate the extra-column mixing and the theoretical predictions were in close agreement with the experimental data.
Degree
Ph.D.
Advisors
Wang, Purdue University.
Subject Area
Chemical engineering
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.