Mechanistic modeling of primary depth filtration in downstream bioprocessing
Date
2023
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
University of Delaware
Abstract
Depth filtration (DF) is a common method of removing solid particles from suspension; in the biopharmaceutical context, these often consist of culture cells and cellular debris, although DNA and protein molecules and their aggregates can also be removed. It is frequently the principal method of removal for impurities 1 µm or larger, particularly when employed to filter raw bioreactor effluent in “primary” depth filtration. Published analysis of the depth filtration of cell suspensions is quite sparse, and the studies that are available often employ models of fouling in membrane filtration, as well as purely empirical methods. These methodologies ignore some of the defining characteristics of DF which are captured by mechanistic models formulated specifically for the process; these features include the capture of particles along the filter’s depth instead of solely at the upstream face, from which the operation's name is derived. ☐ Another persistent issue in the prediction of depth filtration performance is the lack of characterization of the relationship between the filter microstructure and filtration properties. The structural properties of porous materials, such as porosity and pore size distribution, are well known to affect their permeability and particle entrapment characteristics but are often scarcely characterized for depth filters. The nominal pore size or retention rating provided by manufacturers falls short of a full description of the filter pore space, and no standard methodology has been established for independent investigation. ☐ This work seeks to adapt an existing mechanistic DF modeling framework to analyze filtration in the biopharmaceutical context. The constitutive relations that govern pressure drop and particle deposition within the filter are the primary focus of model development, as previous applications of the model have shown that the most appropriate expression for either depends highly on the characteristics of the system, e.g., the properties of the filter and particle size distribution. The model is initially adapted for and evaluated against literature data from the filtration of a yeast suspension and subsequently applied to analyze the filtration of a mammalian cell culture fluid. The model is found to provide semi-quantitative accuracy for the data on yeast filtration and captures several phenomena of interest, including the transient and lasting decrease in turbidity observed with several filter samples. In contrast, the model predictions do not comport with the results of the mammalian cell culture filtration, indicating that relevant features of the process are not properly reflected in the model’s structure and specification. ☐ Chapter 3 details the microstructural characterization of several filter samples using X-ray computed tomography (X-ray CT) and pore network modeling. The filter structure is digitally reconstructed with submicron accuracy, and several characteristics of the pore space are analyzed and compared to manufacturer-provided retention ratings. The retention ratings are found to correspond to the most consistent percentile of the distribution of inscribed pore diameters as compared to the volume-equivalent pore diameter.
Description
Keywords
Depth filtration, Fibrous media
, Mechanistic modeling, Pore network modeling, Bioprocessing