Modeling for Process Risk Assessment in Industrial Bioprocesses

The objective of this contribution is to give an overview about model implementations for risk-based decision making in industrial bioprocesses, such as the antibiotic production for the pharmaceutical industry. It focuses on the applications of mechanistic and computational fluid dynamics (CFD) models. The models are built to support the understanding of the process, improve or speed up the process development, and used to monitor and control the production process to achieve the desired product quality and quantity. Uncertainties inherently present in development of these models are considered in many applications, for example, by for example using Monte Carlo simulations, to enable a risk-based decision making when the model results from Monte Carlo simulations are assessed. The sources of uncertainties may include for example process input variations, model parameter uncertainty, assumptions underlying the model structure, and measurement errors, among others. More and more studies combine mechanistic biochemical models with CFD models to investigate especially heterogeneous process conditions at large scale such as substrate gradients. However, on-line applications of CFD models, for example, for process control, are hampered by the long computation times. Instead, CFD modeling efforts are directed towards supporting CFD-based compartment modeling that reduces the spatial resolution, but allows a much faster simulation compared to a CFD model. Compartment models integrating bio-kinetics of the bioprocess, various sources of uncertainties in the system, and heterogeneities in the bioreactor can then be applied as an enabling tool for risk-based on-line monitoring and control systems to achieve optimized bioprocess operations.