Free-radical polymerization engineering—IV. Modelling homogeneous polymerization of ethylene: determination of model parameters and final adjustment of kinetic coefficients
✍ Scribed by Pascal Lorenzini; Michel Pons; Jacques Villermaux
- Publisher
- Elsevier Science
- Year
- 1992
- Tongue
- English
- Weight
- 773 KB
- Volume
- 47
- Category
- Article
- ISSN
- 0009-2509
No coin nor oath required. For personal study only.
✦ Synopsis
A mathematical mode1 for free-radical polymerization of ethylene was established in a previous paper. The present paper deals with the determin ation of kinetic parameters of this mode1 by optimal fitting of experimental data obtained in a bench-scale CSTR. Overall polymerization rate coefficient and initiator micromixing time were previously determined. The other rate coefficients (transfer to polymer, intramolecular transfer, /&-and /?,-scission, transCer to solvent, to initiator and to chain transfer agents) are adjusted to represent best molecular-weight distribution and structural characters (short-and long-chain branching, double bonds). The obtained valuer agree with literature data. Simulations of the influence of pressure and temperature on polymer quality are presented. Starting from a standard set of parameters, a method is proposed to force the model to fit data obtained under new experimental conditions by a fine tuning of rate coefficients based on a sensitivity analysis. An example is presented which shows the practical interest of the method for application in the plant on a microcomputer. INTRODIJCTION In a previous paper of this series (Lorenzini et al., 1992), a mathematical model was established to represent free-radical polymerization of ethylene in the high-pressure, high-temperature range. Owing to these relatively severe conditions, scission and transfer processes are involved which make it especially difficult to obtain the molecular-weight distribution (MWD) and structural characters such as branching points by standard analytical methods. Original methods were derived to overcome these difficulties. In particular, it was proposed to obtain the MWD by distorting a log-normal distribution in such a way as to make its first three leading moments equal to those predicted by the model. This method was shown to be simple and effective and thus makes it possible to compare experimental and model-generated MWD. The aim of the present paper is to determine optimal values for the kinetic coefficients involved in the model by fitting theoretical predictions to experimental data. EXPERIMENTAL STUDY IN A BENCHSCALE REACTOR In order to determine the kinetic parameters of the model, a series of 26 runs were carried out at the Mazingarbe Research Center of CdF Chimie S.A. (now taken over by SOFRAPO). The experimental setup comprises a bench-scale CSTR for kinetic t So&t6 Franqaise de Poly&hylZne (SOFRAPO),