Application of MHE-based NMPC on a Rotary Tablet Press under Plant-Model Mismatch

Area of Science:

• Pharmaceutical Manufacturing
• Process Control
• Chemical Engineering

Background:

• Modern pharmaceutical manufacturing increasingly relies on automation and digitalization.
• The industry is transitioning from batch to continuous operations, necessitating advanced control strategies.
• Active control ensures real-time corrective actions for quality target deviations.

Purpose of the Study:

• To implement a Quality-by-Control (QbC) framework for tablet manufacturing.
• To develop and apply a moving horizon estimation-based nonlinear model predictive control (MHE-NMPC) framework.
• To achieve effective setpoint tracking and disturbance rejection in tablet production.

Main Methods:

• A three-level hierarchical control structure was applied.
• Moving Horizon Estimation (MHE) was coupled with Nonlinear Model Predictive Control (NMPC).
• Nonlinear mechanistic models were used for predicting powder properties and providing physical interpretations.

Main Results:

• The MHE-NMPC framework enabled real-time model parameter updating using historical and sensor data.
• Adaptive modeling compensated for process uncertainties and reduced plant-model mismatch.
• The framework was successfully demonstrated on a commercial-scale tablet press (Natoli NP-400) for controlling tablet properties.

Conclusions:

• The MHE-NMPC framework offers a robust approach for active control in pharmaceutical manufacturing.
• Careful determination of parameter updating frequency and constraints is crucial for control robustness, especially with sensor degradation.
• The developed framework provides practical applicability and enhances the control of critical tablet properties.