Answers are grounded in this paper and cite their sources.
In this work, improvements to a model predictive controller in an industrial facility is presented. A depropaniser column used to separate propane and propylene from heavier components was experiencing sporadic process instabilities which was impacting product purities. A soft sensor to measure C4 hydrocarbons was inaccurately predicting the actual composition, which caused the controller to behave erratically. A simple linear regression implementation improved the accuracy of the soft sensor significantly, which allowed further optimisation of the controller. Large benefits in process stability and propylene recovery were observed.
No takes yet. Share an insight, caveat, or question.
Lombard et al. (2024) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: