Ranking bioprocess variables using global sensitivity analysis: a case study in centrifugation

Abstract

The selection of appropriate operating conditions for bioprocessing is complex due to the large number of interacting stages and variables. Bioprocesses also operate under tight regulation and therefore tools to assist bioprocess design are of significant utility. Conventional approaches for the analysis of variable sensitivities are inadequate. We propose the use of global sensitivity analysis to determine the level of importance of each variable and their interactions. Once key variables have been determined, the designer may focus on the most significant subset. Two case studies are used to demonstrate the applicability of the approach. Each is based on centrifugation and determines the impact of flow-rate, feed viscosity, density difference and particle size, while performance is assessed by supernatant clarification. Significant differences in sensitivities were found between the two studies due to the different feed material properties. Variable sensitivities were found to be system-specific and provide insight for potential operating strategies.

A Model‐Based Approach to Determine the Design Space of Preparative Chromatography

“…Sensitivity analysis [17][18][19][20][21] is used to determine which disturbances affect the performance of the process most. Uncertainty analysis [22][23][24][25][26] is used to estimate how the performance of the process and the quality of the product are affected by process disturbances.…”

Designing Robust Preparative Purification Processes with High Performance

“…The first question is answered by performing a sensitivity analysis [7][8][9][10][11] on the studied operating point. This will rank the importance of the process parameters and determine the critical process parameters [3].…”

Global sensitivity analysis for the determination of parameter importance in the chromatographic purification of polyclonal antibodies

“…To determine sensitivities at other parts of the function using LSA requires calculation of partial derivatives at every point of interest -a potentially time consuming and computationally intensive task. Additionally, LSA only analyses one input at a time, 5 such that characterisation of the model would necessitate separate calculations for every variable. Global sensitivity analysis (GSA) is a more powerful method that assesses all input parameters simultaneously over their full ranges of uncertainty and ranks them on a zero to one scale to signify the average contribution that each one makes to the variance of the model function.…”