On March 20, we organized a scientific seminar titled “First-principles Modelling and Parameter Estimation of an Industrial Propylene Polymerization Reactor”. Jozef Vargan, a PhD student at the Slovak University of Technology in Bratislava, led the seminar, organized within the FrontSeat project as part of the “Research Seminar on Smart Cybernetics” series.
Abstract:
First-principles modeling is a cornerstone of chemical process engineering, providing physically consistent and predictive descriptions of complex systems. However, its industrial application is challenging due to the model complexity, parameter uncertainty, and limited data quality. To address these challenges, this presentation highlights the procedures and obstacles involved in estimating kinetic and polymer-property parameters for a first-principles model of an industrial propylene polymerization reactor. An industrial data set is presented, including process variables and polymer properties, with the focus on the data pre-processing. An extended kinetic model is explained simply to reflect industrial operating conditions. Local and global sensitivity analyses, including Sobol’s indices, are discussed to quantify the influence of parameters on model outputs, followed by an orthogonalization-based estimability analysis to detect collinearity and rank parameters. A multiple local sensitivity analysis is proposed to address issues in the computation of Sobol’s indices. A reduced set of parameters is selected and estimated using optimization-based methods, while 99% uncertainty intervals are computed to quantify the parameter values. Model predictions are validated on both training and testing data. It will be demonstrated that the identified model demonstrates a strong predictive performance on industrial measurements.
This project has received funding from the European Union’s Horizon under grant no. 101079342 (Fostering Opportunities Towards Slovak Excellence in Advanced Control for Smart Industries).
0 Comments