2019 - PI World - San Francisco - Life Sciences / Specialty Chemicals
It’s The Data Stupid: How PI AF Analytics Rescued Our Real-time Multivariate Process Monitoring (Bristol-Myers Squibb)
Bristol-Myers Squibb implemented the capability to provide operations with a proactive, multivariate tool to monitor batch manufacturing processes in real-time. This capability highlights the value of the enhancements made to our underlying data infrastructure and data availability from various manufacturing data sources, including OSIsoft PI.
Multivariate process monitoring is critical to the business to ensure optimized process performance within the process design space. It also provides operations and technical support teams the capability to monitor multiple variables of the manufacturing process simultaneously against historical performance and take proactive action.
One of the challenges to deploying multivariate monitoring is providing the required batch context and continuous data in real-time. Creation of offline multivariate models uses manually curated historical data sets that incorporate batch time context and time series data. Transitioning to online monitoring requires providing the same batch time context and time series data solely from real-time data sources such as PI. This required the implementation team to develop innovative design choices to circumvent the limitations of the vendor-supplied PI integration package and constraints imposed by BMS network infrastructure. This presentation provides details of how the team applied PI AF Analytics to enable real-time process monitoring and realize on-going value to manufacturing operations.
A chemical engineer by trade, Matt enjoys applying creative technological solutions to streamline operations, helping to deliver medicine to patients faster. Since joining Bristol-Myers Squibb, Global Product Development and Supply IT in 2013, Matt led teams to deploy PI across the biologics and pharmaceutical sites. He partnered to devise technology standards for the organization. Matt also co-architected and integrated the Enterprise PI system, which serves as a foundational data source for the product robustness and data lake initiatives. Matt is currently leading cross-functional teams to expand data analytics in the Global Biologics and Stability spaces. Lastly, he is collaborating with the Data Systems, Monitoring and Analytics group to expand the PI footprint that enables real-time multivariate process monitoring. A future area of interest includes discovering use cases and technologies in predictive analytics that could be applied to further optimize drug manufacture.
Bing Zhang is an Engineer from the Data Systems, Monitoring and Analytics team at Bristol-Myers Squibb and is based in Devens, MA. She leads projects in Real-time Multivariate Monitoring on Bioprocesses and Equipment. Bing specializes in Process Analytical Technology and Chemometrics. She holds a PhD in Chemical Engineering from Louisiana State University and a MS in Chemistry from Tsinghua University.