Model based process analysis and scale-up in membrane chromatography

Membrane chromatography (MC) is gaining wider acceptance in downstream processing of biopharmaceuticals. MC offers the advantage of higher operational flow-rates and thus, provides an attractive alternative to traditionally employed packed bed chromatography. With increased commercial usage, in-depth understanding of different mechanisms within MC capsules is desirable in order to achieve optimum performance. MC capsules are characterized by the sharpness of their breakthrough curves (BTCs). However, BTCs frequently suffer from tailing near the saturation. Furthermore, the degree of tailing differs not only between MC capsules from different vendors but also varies between MC capsules at lab and large scales from the same vendor. A lab scale MC capsule, built as a physical scale-down model of a large scale MC capsule, therefore, cannot be accurately used for optimization of large scale purification processes. Hence, mathematical model-based approaches are required for holistic process analysis and scale-up in MC. Several works have been performed in modeling BTC behaviour of MC capsules in the past. However, although, the traditionally used modeling approach works in performing stand-alone process analysis in MC on one scale of operation, it completely fails in providing a holistic model based scale-up. Hence, in this work, two advanced modeling approaches, the zonal rate model (ZRM) and the computational fluid dynamics (CFD), are developed for providing model based scale-up in MC capsules. Both modeling approaches have been shown to accurately predict the breakthrough performances of large scale MC capsules as compared to the traditionally used modeling approach. MC capsules exhibit several non-idealities at different scales: Non-ideal hydrodynamics, non-ideal protein adsorption and non-linear scaling, all influencing the shape of measured BTCs. The developed modeling approaches holistically capture these non-idealities and decouple their influence on the observed BTCs. Thus, the developed models are universal in nature and can be applied for rational model-based process design employing MC capsules with different flow configurations, membrane types and scales.