Implementation of filtration models for filter performance prediction in bioprocess dead-end membrane filtration

Abstract

Mechanistic modeling is frequently applied during process development and scale-up in bioprocessing. For normal flow filtration (NFF, also called dead-end filtration) processes, such as aseptic filtration, harvest clarification, and virus removal etc., the modeling based on a theoretical fouling law has long been used to assess filter performance and filtration area requirements. This case study describes how an appropriate fouling model is selected and provides guidance for using the fouling model for bioprocess application. There is general agreement that the modeling and process simulation enables accurate filter size calculations for scale-up. Their predictions can then be used to identify optimal filter formats for each unit operation. However, because feed characteristics and conditions vary between processes, fouling models need to consider several filter fouling mechanisms for different bioprocess applications and unit operations. This study reviewed several fouling models and considered various bioprocess filter fouling scenarios, as well as the accuracy of filtration area predictions from the models. It was determined that filtration runs in small scale sizing studies should be run to at least 40% flux decay to minimize the risk of prediction error due to extrapolation.