(334ap) Dynamic Process Modeling and System Analyses for Continuous Pharmaceutical Manufacturing

Research Abstract

Given the numerous advantages including improved process efficiency and product quality, continuous manufacturing (CM) has been an active area of research in the pharmaceutical industry. This research focuses on various experimental and modeling strategies, aimed towards process understanding and development of a virtual predictive framework of the overall process. One of the major bottlenecks for successful development of such framework and subsequent commercialization of CM technology is the lack of fundamental understanding of dynamic complexities arising during powder flow. To address this bottleneck, prominent challenges involve detailed understanding and systematic integration of powder dynamics in process operation and system analyses. This is paramount for development of a predictive framework with real-time capability.

Mechanistic, first-principle based particle-scale simulations like Discrete element modeling (DEM) are increasingly employed to explore and extract particle-level information of powder flow and particle dynamics. Such tools are used to investigate powder mechanics during various process conditions and develop predictive models. During the development of such models, there is a balance that needs to be attained between the computational complexity and the model accuracy. Particle-scale simulations, when used for model development, can significantly improve model accuracy, however, at the cost of computational complexity. The computational requirement further increases when integrated with other unit operations in the process flowsheet. Alternatively, computationally cheap models might fail to capture detailed powder dynamics. Thus, there is a need to develop computationally efficient models using mechanistic particle-scale simulations, to be used for real-time prediction within integrated process flowsheets. To further improve the accuracy of these flowsheets, there is a need to develop and integrate real-time material tracking to explicitly track materials along time for lot-to-lot delineation. The process flowsheets, thus developed, can truly be dynamic in nature, while incorporating complexities of powder flow.

Following this ideology, we propose to develop computationally-efficient models of unit operations in combination with DEM, using reduced-order and approximation strategies. Unit operation models, thus developed, are integrated together, within process flowsheets. Dynamic residence time distribution models are also proposed to enable real-time material tracking. Using the improved flowsheet models, we further plan to enhance system analyses tools like sensitivity, feasibility and flexibility analysis to include process dynamics while accounting for uncertainty, due to the variable nature and material properties of pharmaceutical powders. The proposed work would, thus, play a crucial role towards the development of a predictive framework for manufacturing lines, while capturing detailed process and powder dynamics.