(623c) Nonlinear First-Principle Model-Based Control of a Continuous Pharmaceutical Manufacturing Process

Authors: 
Yazdanpanah, N., U.S. Food and Drug Administration
O'Connor, T., U.S. Food and Drug Administration
Pavurala, N., Office of Testing and Research, U.S. Food and Drug Administration
Mazumder, S., Office of Testing and Research, U.S. Food and Drug Administration
Ashraf, M., U.S. Food and Drug Administration
Cruz, C. N., U.S. Food and Drug Administration
Taylor, C., U.S. Food and Drug Administration
Xu, X., Office of Testing and Research, U.S. Food and Drug Administration
Model-based process control is one of the pillars of continuous pharmaceutical manufacturing process by the QbD approach. The first-principle approach, which comes from in-depth understanding of the governing transport phenomena and performance of unit operations, yields developing of predictive models with which the continuous processes can be manipulated, controlled, and optimized. Despite empirical modeling, where limited operation conditions and process status are being mapped within a DoE, other unseen nonlinearities in the process can be captured in a nonlinear first-principle model to optimally control the process.

In this work, a nonlinear first-principle distributed dynamic model was developed to control a drying process for continuous manufacturing of orally dissolving films loaded with a pharmaceutical compound. The drying model is the core of active control for the process, in which the process parameters (such as air temperature, or production rate), feed compositions (like solid content and concentration), or final products’ properties (for instance the required final moisture content) can be the manipulating variables. Simulation of transport phenomena, diffusion controlled mass transfer (mutual solvent-polymer vs. moisture sorption isotherm), the spatial distribution of solvent concentration in the film and the moving boundary on the interface due to the thin film shrinkage during drying are some of the complexities of the model development. The final quality attribute of the product is set as the control set point and other process variables were manipulated in the nonlinear model control (MIMO) to converge the process in the design space of variable inputs. The model funds the steps for future phases of PAT implementation, RTRt, and process optimization.