A Bayesian approach to improving production planning. (May 2023)

• Record Type: Journal Article
• Authors: Santander, Omar; Kuppuraj, Vidyashankar; Harrison, Christopher A.; Baldea, Michael
• Is Part Of: Computers & chemical engineering. Volume 173(2023)
• Publication Date: 2023-05
• Subjects: Bayesian modeling -- Production planning -- Fluid catalytic cracker; Chemical engineering -- Data processing -- Periodicals; 660.0285
• DOI: 10.1016/j.compchemeng.2023.108226
• Languages: English
• ISSNs: 0098-1354

Abstract

Abstract: In this study, we present an efficient Bayesian framework for improving production planning decisions. The framework consists of a Bayesian modeling section that accounts for time correlation and mass balance, and an improved production planning formulation that considers the effect of model uncertainty, correlation, disturbances, process control, extra degrees of freedom and process limitations. The proposed Bayesian framework is implemented on an industrially relevant fluid catalytic cracking process model and is compared to the production planning process traditionally followed in the refining industry. Simulation results demonstrate that the proposed Bayesian model is 60% more accurate and requires half the training time of traditional industrial approaches. The resulting production planning structure has robust performance due to considering uncertainty in model predictions. Highlights: A new framework for improving production planning is proposed. Bayesian modeling accounts for time correlation and mass balance. Production planning formulation considers uncertainty, disturbances, constraints. Implementation to large-scale FCC case study.