Taming the Lion: Improving Plant Operations through Inferential Modeling

Inferential property models, often called soft-sensors, estimate process properties that cannot be measured directly or continuously using easily measured variables. Developed using various data analytic techniques, inferential property models enable improved control by providing estimated product properties in real-time. They are widely used in various chemical manufacturing processes at Dow. This presentation will provide some insights gained from the collaboration of soft-sensor development across functional teams in Dow and share implementation case studies and best practices on the use of inferential property models to improve plant operations.