Ask Greg McMillan - What are the opportunities for online adaptation of bioreactor models?

Greg's Response:

Model Predictive Control (MPC) can provide the fast and nonintrusive adaptation of bioreactor models. Key model parameters are chosen as the manipulated variables, and associated process measurements are chosen as the controlled variables of the MPC for adaptation of the digital twin. An automated test sequence is then run for the MPC at the highest possible speed off-line, and the MPC models are identified and visually checked as reasonable in terms of the direction and relative magnitude of the effect.

The digital twin is then connected to the actual plant in a read-only nonintrusive setup. The digital twin modes, set points, and batch phases come from the actual plant. The set points (targets) of the MPC’s controlled variables for adaptation are externally referenced to the key respective measurements of the actual plant,

The MPC targets and controlled variables are the actual plant and digital twin manipulated variables (e.g., flows). A key example is how a bioreactor model can be adapted to provide more accurate inferential measurements of biomass growth and production rates that are slopes of the batch profiles for cell and product concentration. The MPC manipulated variables are model parameters (e.g., kinetic parameters) that affect these profiles.

The MPC can be identified and tested without the digital twin connected to the actual plant. The MPC development for adaptation is much faster than that of an actual plant MPC. This independence enables extensive adaptation and gaining associated knowledge of process relationships without any disruption to the plant. The adaptation is done without affecting the actual plant because the plant’s manipulated variables are being read by but not changed by the digital twin. It is critical that the digital twin have the same set points and tuning settings as the actual plant and that the digital twin is started with controller outputs initialized to match the actual plant.

The optimized set points from MPC with inferential measurements of growth and production rate are done in an advisory mode not affecting the actual plant. An MPC is run in automatic mode in another digital twin that is a duplicate of the adapted digital twin running much faster than real-time to study the optimized set points and to prototype new control strategies.

For much more knowledge see 2021 ISA Book New Directions in Bioprocess Modeling and Control – Maximizing Process Analytical Technology Benefits Second Edition.