468259 Virtual Plant Environment Incorporating Data for Predictive Modeling and Control in Bio-Pharmaceutical Manufacturing

Thursday, November 17, 2016: 3:51 PM
Monterey II (Hotel Nikko San Francisco)
Yingying Zheng1, Tony Wang1, Christian Kunert2, Chris Garvin1 and Cenk Undey3, (1)Process Development, Amgen, Thousand Oaks, CA, (2)Process Development, Amgen, Cambridge, MA, (3)Amgen, Thousand Oaks, CA

A virtual plant environment which incorporates data-driven models was developed for bio-pharmaceutical manufacturing. The objective of the virtual plant is replicate the control strategies implemented at bio-pharmaceutical plants in a virtual operating environment. This enables application and testing of data-driven model-based process simulation to gain insight into process and control system design while minimizing the number of wet experiments required. A DCS system was used as the platform to build the virtual plant, and Matlab was used to develop data-driven models. Data used in Matlab was taken from first principles simulations as well as empirical data. Bi-directional data communication was established between the DCS system and Matlab via OPC connection. A nested model scheme was developed to work within virtual plant. A case study was performed on a chromatography unit operation for drug substance manufacturing. The plug-and-play modular design of the virtual plant gives it the flexibility to simulate existing automation setups at many bio-pharmaceutical manufacturing plants.

Extended Abstract: File Not Uploaded
See more of this Session: Process Control Applications II
See more of this Group/Topical: Computing and Systems Technology Division