Automation Aspects of PAT Installation w/ Case Study

1. Automation Aspects of PAT Installation w/ Case Study Pete Miilu, Pfizer Marc Surprenant, Pfizer

2. Intro/Background Automation and Control Strategy: • Levels of Automation • Process Monitoring • Instrument monitors the process to determine when the end point is met • Monitoring and Analysis of data • Process Control • Instrument monitors and adjusts the process based on results • Full implementation of PAT • Ultimate Goal is Real Time Release • Integration with Distributed Control System • Communication • Hardwired Input / Output • Serial • Ethernet (OPC) • Recipe Management • Data Retention

3. Data Strategy Several aspects to consider: • Who are the users? • Operators for process control decisions • Require go/no-go decision [bytes] • Require limited data & trending [KB] • Engineers for process improvement • Require results data [KB] • Require trending capabilities [MB] • Lab personnel for instrument maintenance and calibration • Require raw and spectral data [GB] • Require trending capabilities [MB] • Data retention • Different for each type of data? • Important for long-term supportability when dealing with frequent samples of high-data-volume assays • Limited retention for raw data can be justified using calibration and maintenance procedures which meet requirements for demonstrating instrument performance

4. Case Study Background Operational Setting: • Batch Solvent Recovery Distillation Process • Recovered solvent quality verified with manual sampling and analysis using process lab • In process lab analysis results obtained in one to two hours • During lab analysis distillate is directed to a “slop” tank until quality is verified • Manual sampling does not allow distillate collection at the time it meets specification

5. Case Study Background PAT Solution: • Continuous sampling and analysis required • Achieved by implementing an on-line gas chromatograph on the distillate stream • On-line GC capable of collecting and reporting composition of the distillate every four to eight minutes (near real time)

6. GMP Classification of System • Direct vs. Indirect vs. No Impact • Ultimately this decision is based on the way that Operations intends to use the information produced by the system • CQA provides standard tools for determining classification • Case Study • Classified as an Indirect Impact system because the system is not used to demonstrate compliance with any registered parameters or attributes • Downstream direct impact systems ensure compliance with registered parameters • Benefits of Indirect Impact Classification • Allows commissioning-only approach to testing • Lower project & long-term support costs • Opportunity costs • If/when the business finds that the system is capable of replacing the registered downstream assays, then a system qualification must be performed, in addition to a process validation.

7. Ta – Da! The Final Incarnation

8. Implementation • Qualified: • Direct Impact System’s Critical Components: • Product contact – piping and pump • Analytical Instrument • Critical software functions • DCS Interface • DCS Recipe Control • Vendor Software critical functions • Methods of Analysis • Commissioned only: • PC Hardware • Data Historian Interface • Challenges • Breadth of team’s background and experience with C&Q • Agreeing on System/Component Impact Assessments • Assigning ownership of support responsibilities • OPC & network security • Assigning ownership of components during project • Operations’ learning curve • Learnings • C&Q is valuable, but requires up-front planning • Investment in vendor training for instrument is a must • Once the project is complete – the work begins • OPC is a valuable technology platform for data exchange • Leave a backup control strategy!!

9. DCS to Instrument Integration

10. Epilogue • Experiencing issues with calibration drift – instrument/analytical support are working to resolve • Relying on both PAT and backup control strategy • Expect fully operational and reliable system by EOY2008