Model Predictive Control of a Multi-Effect Evaporator for Robust Discharge Concentration Regulation in Biomanufacturing

초록

Multi-effect evaporators are widely used in biomanufacturing to create concentrates of fermentation-derived products prior to or during crystallization, where the discharge concentration directly governs crystal product quality. However, bioprocess feed streams are subject to large and irregular disturbances arising from batch-to-batch fermentation variability and cell separation operations, making stable concentration control essential. In this study, a dynamic differential-algebraic model of a three-effect evaporator was developed, in which product composition evolves according to mass balance ordinary differential equations, while vapor and concentrate flows are determined algebraically from energy balances. Using this model, the discharge concentration was controlled against a representative feed composition disturbance scenario by comparing proportional-integral-derivative (PID) control with model predictive control (MPC). The PID controller failed to suppress the disturbance and exhibited sustained oscillations due to the large structural time delay and composition-dependent nonlinearity of the cascaded process. In contrast, the terminal-cost MPC predicted future behavior from the process model and compensated for disturbances preemptively, maintaining the discharge concentration almost exactly at its set point. The integrated absolute error decreased from 0.3872 for PID to 0.00145 for MPC with a 99.6% improvement. These results demonstrate that MPC enables robust product quality control in disturbance-rich biomanufacturing processes.

키워드