Fast and reliable computational strategy for developing a rigorous model-driven soft sensor of dynamic molecular weight distribution. (August 2017)

• Record Type: Journal Article
• Authors: Kang, Jiayuan; Shao, Zhijiang; Chen, Xi; Gu, Xueping; Feng, Lianfang
• Is Part Of: Journal of process control. Volume 56(2017)
• Page Start: 79
• Page End: 99
• Publication Date: 2017-08
• Subjects: Soft sensor -- Grade transition -- Molecular weight distribution -- Finite elements -- Collocation method; Process control -- Periodicals; Fabrication -- Contrôle -- Périodiques; Process control; Periodicals; Electronic journals; 660.281
• DOI: 10.1016/j.jprocont.2017.05.006
• Languages: English
• ISSNs: 0959-1524

Abstract

Highlights: A systematic framework for a rigorous model-driven soft sensor of dynamic MWD is developed. A bi-layer surrogate thermodynamic model is developed to improve the online performance. A novel dynamic MWD model is proposed to improve the MWD computation efficiency. Moving finite element method with advanced starting point strategy is designed to improve the online performance of the large-scale nonlinear dynamic systems. The fast and reliable computation strategies are demonstrated through plant data. Abstract: Key polymer properties are substantially directly related to the polymer molecular weight distribution (MWD). On-line monitoring and prediction of dynamic MWD profiles are highly important for on-line quality control of polymerization processes. In this study, a fast and reliable computational strategy for an equation-oriented model-based soft sensor for the high-density polyethylene grade transition process is developed. The simultaneous collocation approach is adopted to discretize the dynamic model. A novel moving finite element method is proposed to improve the on-line performance of the derived large-scale nonlinear equation systems. The sensitivity information of the nonlinear equation systems contributes to a convergence enhancement strategy for the sensor. The prediction accuracy and computational efficiency are demonstrated using industrial data. A potential application to extend the polymerization process with changeable flowsheet is also tested through simulation.