Modeling of a gas phase polypropylene polymerization in a horizontal stirred bed reactor

In the industry when someone wants to modify the way a chemical reactor works; it is not possible to try new improvement without prior studies. The first thing to do is to model the problematic. In this optic, this paper aims to develop a kinetic model for a polymerization reactor and to use this model to develop new ways of managing this reactor to produce different polymer that are created now. This document contents the development of a steady state model for the polymerization of impact propylene-ethylene copolymer using a horizontally stirred bed reactor. This model is implemented in the advanced software tool Aspen+ Plus. Moreover, a quite simple CFD model has also been built to describe the gas phase in the Ansys Fluent software tool. We first considered the components and the existing processes used to convert them to polymer. We then characterized the process in question and present several ways of modeling. We also focus on the kinetic reactions scheme of polymerization and the end-user properties. The model was developed with a multi steps methodology. The first part detailed the homopolymerization section composed of a prepolymerization loop reactor and a vertically stirred liquid-gas reactor. The second part focuses on the copolymerization section which is mainly composed of the horizontally stirred bed gas phase reactor. This document includes discussions about thermodynamic properties, reaction kinetics, polymer properties and other modeling issues. The three reactors were modeled using the CSTR model. Except for the prepolymerization reactor, recycling loops have been included. The CFD simulation used a turbulent model and the powder of polymer is set as a fixed bed with the corresponding parameters. The catalyst used in the reaction scheme is of type Ziegler-Natta. The multiple catalyst sites allow to model the wide distribution of molecular weights of the polymer produced. To identify the parameters of each sites, we used deconvolution techniques. Concerning the properties of the products, some estimation has been conducted for the melt flow index and the xylene soluble. The results are based on expression found in the literature. The model is validated using plant data. The parameters of the desired products are satisfied for the several grades we use to fit the kinetics reactions parameters. We succeeded in predicting production rates and molecular weight distributions. Eventually we developed several models with modifications in the process. We managed to show that with these modifications it is possible to modify the products of the process. At the end, we present the leads for improvements and what remains to be doneMaster [120] : ingénieur civil en chimie et science des matériaux, Université catholique de Louvain, 201