Simultaneous Optimization of Product Purity and Yield in a Simulated Moving Bed Chromatography for Nystatin Purification

Abstract

In this study, we accomplished the task of optimizing the product purity and yield simultaneously for a four zone simulated moving bed (SMB) chromatographyaiming at the purification of nystatin, which is one of the well known antifungal antibiotics. For this work, a multi-objective optimization principle was adopted while employing the purity and yield of product (nystatin) as the objective functions. The results from such a multi-objective optimization task were obtained in the form of a Pareto optimal set, which comprises a group of multiple solutions that have equal optimum status in terms of both the nystatin purity and yield. The optimal solutions in the Pareto set showed a trade-off between the two objectives, which could be well interpreted using the equilibrium theory. The effect of adsorbent particle size on the above optimization results was investigated. With the increase of the particle size, the nystatin purities and yields of the SMBs corresponding to the Pareto set were found to increase in the region of small particles (pressure limiting region) but decrease in the region of large particles (mass transfer limiting region). As a result, the best Pareto set, which is defined here as the Pareto set surpassing all the other ones in both the nystatin purity and yield, occurs when the particle size falls on the boundary between the pressure limiting and the mass transfer limiting regions. The effect of throughput (or feed flow rate) on the optimization results was also examined. The results showed that a decrease in the throughput improved the nystatin purity and yield while narrowing down the distribution region of the two objective values on the corresponding best Pareto curve. Consequently, the highest purity and yield of nystatin (99.9% each) were attained when the best Pareto curve was converged to only a single point as a result of a significant decrease in the throughput.