Economic Nonlinear Model Predictive Control of a U-loop Bioreactor

In this paper, we present an algorithm for economic nonlinear model predictive control (NMPC) of single-cell protein production in a U-loop bioreactor. The model of the U-loop bioreactor consists of both stochastic ordinary and partial differential equations. Using a typical finite-volume discretization, the model contains 87 state variables. The NMPC algorithm is based on the continuous-discrete extended Kalman filter and a simultaneous collocation method. We present a closed-loop simulation which demonstrates the computational feasibility of real-time implementation of the NMPC algorithm for startup and steady state operation of the U-loop reactor.