Application of An Advanced Process Controller to a Continuous Mixing, Direct Compression Process

A Model Predictive Controller (MPC) has been applied to a pilot-scale continuous mixing , direct compression manufacturing process.  This process consists of three gravimetric feeders , an in-line powder mixing device and a rotary tablet press.  A near-infrared (NIR) probe was inserted into the feed frame of the rotary tablet press to continuously monitor blend potency immediately prior to tablet compression.   A series of process response tests were applied to the system.  Process predictors , such as impeller rotational speed , excipient and API mass feed rate set-points , were varied.  The corresponding quality attribute responses , such as blend potency by NIR , mixer hold up mass , and tablet press hopper level were observed. Using this dataset , a dynamic process model has been developed. This model has been used in a Model Predictive Controller to provide multi-variable control of the blend potency prediction by NIR , hold-up mass and total mass flow rate. The prototype predictive controller was challenged by changes to either the total mass flow rate through the system or set point changes to the target blend concentration levels.