Kinetic Parameter Estimation for Linear Low-Density Polyethylene Gas-Phase Process from Molecular Weight Distribution and Short-Chain Branching Distribution Measurements

Abstract

Kinetic parameter estimation for a complex copolymerization process has always been a challenge in the modeling procedure. This study aims at the kinetic parameter estimation for a linear low-density polyethylene (LLDPE) gas-phase process from molecular weight distribution (MWD) and short-chain branching distribution (SCBD) measurements. First, experimental MWD and SCBD are simultaneously deconvoluted to obtain intermediate model parameters as output variables. Then, appropriate nominal values of the kinetic parameters are provided by solving an optimization problem. This procedure plays a significant role in narrowing down the range of the nominal values. The determined output variables and nominal parameter values are used to form a sensitivity matrix for parameter estimability analysis. After that, a new parameter ranking strategy is proposed using hierarchical clustering. Based on the determined nominal values, the ranking results obtained using the proposed strategy in a robustness test are more robust than those obtained under random nominal values. Lastly, the hierarchical clustering is combined with Wu's mean squared error-based method to determine an estimable parameter subset, during which the selected kinetic parameters are estimated by matching the intermediate model parameters. The gas-phase copolymerization process model based on the estimated parameter values is further validated with different MWD and SCBD measurements. Model predictions show good agreement with experimental data.

Kinetic Parameter Estimation of the Polyethylene Process by Bayesian Optimization

“…Touloupidis introduced an effective methodology capable of estimating all pertinent parameters and reverse engineer experimental results. Feng , carried out parameter estimation centered on a non-standard ethylene/a-olefin gas phase copolymerization reaction mechanism, utilizing MWD and short chain branch distribution (SCBD) as objective functions. However, relying on MWD or microscopic chain structure information as the sole evaluation function for optimization might be insufficient.…”

Simple Method for Incorporating Prior Parameter Information during Model-Based D-Optimal Design of Experiments for Multi-Response Chemical Process Models

“…Often fundamental models for novel chemical systems contain many unknown parameters, such as kinetic and transport parameters, that require estimation from experimental data. Typically, only limited data are available for parameter estimation, making it difficult to estimate all of the unknown parameters reliably. − …”

“…When modelers cannot reliably estimate all of the unknown parameters in their fundamental models, they may choose to estimate a subset of the model parameters, ,,,− or they may adopt a Bayesian approach, where prior information is provided about some or all of the model parameters. − When modelers wish to obtain more accurate estimates for model parameters, they may choose to perform new experiments to obtain additional information. Performing additional experiments can be time-consuming and expensive, so researchers want to carefully select operating conditions so that as much information as possible is gleaned from the new experimental runs.…”