Optimizing Fed-Batch Processes with Dynamic Control Flux Balance Analysis

Fed-batch processes are prevalent in biotechnological industries, but design of experiments often results in sub-optimal conditions due to incomplete solution space characterization. We employ a single-level dynamic control (DC) algorithm for dynamic flux balance analysis (dFBA), enhancing efficiency by reducing Karush-Kuhn-Tucker (KKT) condition constraints and adapting the algorithm for predicting optimal process length. In a growth-decoupled plasmid DNA production case study, we predict the optimal feeding profile and switching time between growth and production phase. Comparing our algorithm to its predecessor shows a speed-up of at least a factor of four. When the process length is part of the objective function the speed-up becomes considerably larger.