BIOCONTROL - Model-based strategy for the optimal operation of bioreactors

The proposed project is a continuation of the FFG-funded preliminary study BIOPTI. It aims to design a standardized workflow for optimizing the operational strategy of bioreactors. The project is relevant for the following reasons: Bioprocesses represent the only production route for many pharmaceuticals. To ensure a secure supply of biopharmaceuticals, even in times of crisis, Austria needs to establish itself as an attractive production location. Therefore - to prevent the pharmaceutical producers from relocating - it is necessary to run the processes as cost-effectively as possible. Additionally, reducing the resource requirements for the production of biopharmaceuticals can have a positive environmental and climate impact. These two aspects are addressed by the strategy proposed in the project. The process of interest is a highly complex, two-stage process for nanobody production. Nanobodies are special antibody fragments extremely relevant for pharmaceutical applications due to their small size, stability, and high binding affinity for their target molecules. They are suitable for drug development and diagnostic purposes due to their versatility in targeting various diseases and biomarkers. The workflow proposed in this project covers all critical steps, from selecting an arbitrary process to its optimal operation on the real system. A particular focus is on data-driven AI-based process modeling, model-based fault detection, and model-based process control. The optimization activities are performed carefully, considering process efficiency in production time, product quality, and the related ecological footprint. The optimal process flow is determined and maintained using an integrated process control concept (integrating robust model predictive control supported by a fault detection algorithm). All proposed concepts (including models, control, and fault detection) are integrated into the automation platform. Within this platform, the process operator is involved in the decision-making process through the human-in-the-loop concept. On the one hand, the process operator is supported about the occurring faults through messages and instructions provided through the operator interface. On the other hand, the process operator provides back information (visually assessed plant faults) to improve the controller performance. The proposed approach will not only result in resource-saving (by mitigating the process disturbances and optimal design of the process), but it will also ultimately increase the well-being of biotech personnel and reduce the necessity for highly trained personnel working night shifts.