Best practice and emerging technologies

in process automation and control

To learn more or register for this event, please use the links above or click on the image.

Abstract

The paper from Perceptive Engineering will describe a digital design approach for Model Predictive Control (MPC) of a pharmaceutical twin screw wet granulation (TSWG) process. A workflow involving PharmaMV software, is developed to interact with the parameterised TSWG digital twin, as a virtual plant, implemented in gPROMS Formulated Product software.

The workflow involves first running open-loop step tests on the critical process parameters (CPP), including the Active Pharmaceutical Ingredient (API) and excipient feeders as well as the binder pump, to characterise their response on the critical quality attributes (CQA) including production rate, concentration and particle size distribution (PSD) D50 at the outlet of the TSWG.

The next step is to run an online model adaption routine to automatically identify an MPC control model between the CPPs and CQAs. Then, this MPC controller is tested on the virtual plant by making (simultaneous) setpoint changes on the Concentration, PSD D50 and the Production rate. Results demonstrate that the MPC controller is able to deliver robust setpoint tracking for all the CQAs by efficiently manipulating the CPPs within their respective constraint limits. This is followed by commissioning the designed MPC controller on the actual TSWG process.

The advantages of the proposed scheme include enhanced process understanding, rapid development of the advanced control scheme as well as reduced API consumption and experimental effort. Furthermore, the proposed scheme can be used to investigate what-if scenarios and scale-up effects on the process as well as use as an online soft sensor for any hard-to-measure CQAs.

Suitability