Hannah Robinson, MEng
Presented by: Hannah Robinson, on behalf of: M. Berry, P. Shapland, P. Clements, M Hughes, S. Ukuser, N. Hodnett, P. Hamilton, I. Areri, M. Dumarey, G. Turner, C. Clarke, C. Mayes, I. Phillips, G. Alford, H. Todd, L. Wong, W. Tan, G. Breen, A. Ochen and A. Richards.
In adopting continuous manufacturing techniques for the production of Active Pharmaceutical Ingredients (APIs), additional levers become available for process control that can enhance those traditionally used for the definition of the control strategy in the batch environment. In the presented example, the development of a flow process including five telescoped chemical transformations in a single solvent system (Figure 1), followed by two batch crystallisations is discussed in the context of building an integrated operational and quality control strategy.
Fig 1 – Process Schematic for Continuous Process System
Fundamental understanding of the chemical reactions through kinetic modeling is combined with equipment characterisation, through methods such as residence time distribution measurements, and process behaviour understanding, to maximise experimentation efficiency and mitigate scalability and operability risks. This modeling approach is used to predefine process parameter ranges, which are verified experimentally to demonstrate control of product quality. The equipment characterisation techniques offer the opportunity to advise decisions on boundaries of material used for start-up, shut down, batch definition and traceability of pedigree, as well as treatment of potential out of specification material (Figure 2). How this strategy is implemented in the factory through use of the automated control system is also discussed.
Fig 2 – Residence time distribution at material boundary
Use of a quarter scale pilot rig to demonstrate process reliability and control methodologies as well as confirm modeled reaction conditions is presented, along with the use of PAT for development of understanding, process trending and operational multivariate modeling (Figure 3). Finally, a proposal will be shared on how the data and frequency of data gathered from the instrumentation within the manufacturing process train can be used to give real time information on the condition of the process and also complement analytical data in the release of material, in the context of the drug substance control strategy, to ensure the process has remained and will continue to remain under a state of control.
Fig 3 – Data trending