479684 Advanced Model Predictive Feedforward/Feedback Control of a Tablet Press

Monday, November 14, 2016
Grand Ballroom B (Hilton San Francisco Union Square)
Nicholas Townsend Haas, Chemical Engineering, Rutgers,the State University of New Jersey, New Brunswick, NJ

In continuous pharmaceutical manufacturing, real time precise control of critical quality attributes (CQAs) is necessary for Quality by Design (QbD) based manufacturing and real time release (RTR) with minimum consumption of time, space and resources. Pharmaceutical tablets can be produced through different routes with a common tablet press unit operation always placed at the end of the manufacturing process. Therefore, the tablet press is a crucial unit operation directly influencing the CQAs irrespective of manufacturing routes. Despite this, little attention has been paid to the development of an advanced efficient control system for the tablet press. Process modeling can be used as an efficient virtual experimentation tool to design, compare and evaluate different control systems. We developed a model in Simulink (Mathworks) that includes two master control loops for tablet weight and hardness and a slave feedback loop controlling the compaction force applied to each tablet. We examined the performance of different control strategies based on Proportional Integral Derivative (PID) control and Model Predictive Control (MPC), as well as feedforward/feedback control. We found that a hybrid MPC-PID control strategy outperforms the PID only control strategy. We also observed that the addition of feedforward control further improves the performance of hybrid MPC-PID control strategy.

Extended Abstract: File Not Uploaded
See more of this Session: Undergraduate Student Poster Session: Computing and Process Control
See more of this Group/Topical: Student Poster Sessions