Dynamic Model Reduction and Predictive Control of Hot-Melt Extrusion Applied to Drug Manufacturing

This study aims at designing model-based controllers for a hot-melt extrusion process designed for the manufacturing of pills. The control objective is to regulate the output mass flow rate and the active pharmaceutical ingredient (API) concentration. The first part of this study is concerned with the derivation of reduced-order models from a detailed distributed parameter (DP) model involving mass and energy balance partial differential equations. The parameters of these reduced-order models are estimated based on experimental data from a laboratory device presenting a specific screw geometry, and validated against the detailed DP model. The latter is too computationally expensive with regards to process control, but can serve as a process emulator to test control strategies. The second part of this study focuses on several model predictive control strategies, ranging from Smith predictor to NEPSAC controllers. The effectiveness of these control strategies is assessed using numerical simulation and performance indicators such as disturbance rejection and computational cost.