(517m) Supporting Tools for Design and Validation of Process Analytical Technology System

Process analytical technology (PAT) is gaining increasing importance because of the need to obtain high quality pharmaceutical and bio-products. A properly designed PAT system is required to assure the predefined end product quality. The design of a PAT system includes the selection of critical process variables, followed by the proper selection and placement of suitable monitoring and analysis equipments to keep track on the behavior of these critical process variables, and finally coupling of the monitoring and analysis systems to a control system, ensuring thereby, that critical process variables can be controlled. This work focuses on the development of a model-based computer-aided framework through which the methods and tools through needed to design a PAT system for product quality monitoring (and control) can be generated as an add-on to the traditional chemical process design-control methods.

The starting point for the design methodology is the problem definition in terms of data related to the desired product quality and the process specifications needed to achieve it. This starting point is provided by the user (product manufacturer or PAT system designer). Two supporting tools within the framework ? a knowledge base and a model library are employed to generate the PAT system design alternatives. The knowledge base provides the necessary information needed for design steps. It contains information related to process monitoring and analysis systems and this knowledge is structured in terms of unit processes found in pharmaceutical processes, a list of critical process variables that are usually monitored, a list of the corresponding manipulating variables (actuators), and a list of equipments typically used for their measurement (including data related to accuracy, robustness, time, etc.). This knowledge base has been developed through collection of relevant data from various open sources. The model library contains a collection of mathematical models (data-based as well as mechanistic), that support the analysis needed for the design steps. They also help to predict values of process variables which are not measurable (generating additional or missing data needed to obtain the final design). Finally, the framework includes a design template that guides the user through a sequence of steps (design methodology) of generation and screening of alternatives to end with the desired design of a product monitoring and analysis system.

The objective of this paper is to highlight the design methodology together with the supporting tools available within the computer-aided framework for design and analysis of PAT systems. An important feature is the integration of the model library (plus a modeling toolbox) with the knowledge base and the design template that helps to generate, screen and validate the design of a PAT system. Process analysis, identification of critical points and process variables, and, selection of the appropriate monitoring and analysis systems related to the design of PAT systems will be illustrated through two case studies: the first involves a tablet manufacturing process, and the second involves a fermentation process. Both case studies employ collected data to validate, where possible, the generated (model-predicted) data.