(459c) A Design of Experiment (DOE)-Based Approach for Bioreactor Scale up and Cell Metabolic Reactions | AIChE

(459c) A Design of Experiment (DOE)-Based Approach for Bioreactor Scale up and Cell Metabolic Reactions

Authors 

Nallamothu, S., Ansys Inc
Krishnaswamy, S., Ansys Inc.
Collin, S., Ansys Inc.
Shao, C., Ansys Inc.
Horner, M., Ansys Inc.
Abstract:

Bioreactors are at the heart of biopharmaceutical manufacturing, and they need to be accurately characterized for optimal scaleup. The transition from lab scale to pilot or production scale with the goal of maintaining the same product yield and homogeneity brings lot of challenges. To evaluate the issues related to scaleup in cell culture bioreactors, one must consider mixing time, oxygen mass transfer rates and shear-stress levels among other factor s. As the reactor scale increases, designers would like to ensure consistency in oxygen mass transfer rate as well as exposure of cells to shear because of their influence on the physiology of the cells.

Optimization-based scale up

In this study, computational fluid dynamics (CFD) simulations are used to predict the mass transfer coefficient (kLa) and shear rates for a lab-scale bioreactor. A response-surface (RS) based optimization algorithm is used to optimize the operation of the bioreactor to maximize the oxygen mass transfer rate while maintaining cell exposure to shear under a specified threshold.

The optimized lab-scale bioreactor is then scaled up to pilot-scale. Similarly, a DOE is set up to scan the design space of the pilot-scale bioreactor. Similarly, a DOE will be used to create a response-surface (RS) which is used to optimize the bioreactor operation, maximize oxygen mass-transfer, and minimize shear exposure.

Cell Metabolism System Model

For both scales, the CFD DOE simulations used to construct the response surfaces will also be used to create reduced-order models (ROM’s). The ROM’s will take the bioreactor operating conditions as inputs (e.g. rpm and sparge rate) and providing the mass transfer coefficient (kLa) as an output.

Moreover, a 0D transient biological cell growth model was developed that predicts the cell growth, product formation, and gas consumption using the mass transfer coefficient (kLa) predicted from the CFD-based ROM. The 0D transient cell growth model is then combined with the CFD-based ROM’s in a system model that predicts cell growth, product formation, and gas consumption vs. Time as a function of bioreactor conditions.

The modeling approach proposed herein is an alternative to scale-up laws which may not always be valid (for example when different impeller types are used at different scales). Additionally, the same DOE used for optimization is used to construct ROM’s that are used for system modelling which enables the co-simulation of the flow physics and cell biology.