Oil & Gas Science and Technology (Nov 2006)
DIMG92 : un simulateur de formation pour le procédé DIMERSOL Dimg92: a Training Simulator for the Dimersol Process
Abstract
DIMG92 est un simulateur de formation sur le procédé pétrochimique DIMERSOL (procédé IFP de dimérisation du propylène). En simulant le réacteur et ses équipements annexes, les opérateurs s'exercent à conduire l'unité et comprennent les effets des divers paramètres à prendre en compte. Le simulateur est composé de deux logiciels : - Un logiciel de modélisation, basé sur la représentation mathématique de connaissances physico-chimiques et physiques. Aucun empirisme n'est utilisé dans le modèle. Ce logiciel a été développé par l'Institut Français du Pétrole (IFP). - Un logiciel d'environnement de simulation dynamique pour la formation. Ce logiciel à interface graphique et interactive est distribué par la société française Réalisation en Systémique Industrielle (RSI). Industrial units are tending to become more and more complex and are characterized by a very high degree of integration. The operating of such units requires highly skilled operators. This is why dynamic simulators have training as their prime goal. The DIMG92 simulator falls within this category. It simulates the reaction section of the DIMERSOL G process in a steady state or not. The DIMERSOL G process is designed for the dimerization of propylene into hexene. In the modeled unit, the reaction section includes a liquid-phase reactor and a pump-around on which a heat exchanger is installed. The heat exchanger, which uses water as the cooling fluid, evacuates the heat of reaction. The reaction temperature is between 45 and 55°C. The reactor pressure is in the vicinity of 22 effective bars. The catalyst is in a liquid state. A standard feedstock is made up of 30 weight % inert matter and 70% propylene. Modeling is broken down into three parts : the kinetics, the reactor, the heat exchanger. Mathematical modeling of the oligomerization kinetics was performed at Institut Français du Pétrole (IFP) in 1987. By taking into consideration the initiation, prolongation and termination reactions starting with propylene and hexene, we can understand the oligomerization kinetics in the DIMERSOL G process. The deactivation of the catalyst is represented by a first-order kinetics. With a reduced number of constants, the model takes the following aspects into consideration :(a) The production of heavy oligomers (no propagation reaction is neglected). (b) The formation of oligomers combined with the catalyst. (c) The deactivation of the catalyst in time and the temperature. The kinetic constants and activation energies were determined so that the model can restore the different kinetic results of pilot tests. The modeling of the reactor considers that this reactor is perfectly stirred and always full of liquid. The principal equations reflect the balances of the molecular species. There are also molar balances for the olefins, for the olefins combined with the catalyst, for the inert matter, for the active catalyst and for the deactivated catalyst. An overall mass balance and a mass balance for the sum of olefinic species are added on. A heat balance is used to predict the temperature of the reactor. Other calculations are performed to obtain the selectivities, the conversion and the outflow of each constituent. The modeling of the heat exchanger is done by a static model. The outlet temperatures of the fluids and the power of the heat exchanger are calculated by making an estimate of the efficiency of the heat exchanger. The programming of the model is modular. This makes for a standardized and coherent whole. The modules are written in FORTRAN 77. The user has access either to a static solution to find a stable operating point or dynamic calculating during which the console operator can introduce various disturbances in the variables of the process. In the static module the balances for the reactor form a system of nonlinear algebraic equations that is solved by a Newton-Raphson algorithm. The dynamic module mainly includes the integration, in relation to the processing time of the system of differential equations for the balances relative to the reactor. The above two modules make use of various joint subroutines. The modules containing the modeling of the reaction section are integrated in the SORYA-MX application. This application is a graphic and interactive interface designed for training simulators. SORYA-MX is distributed by the French company Réalisation en Systémique Industrielle(RSI). SORYA-MX has a time management system capable of controlling the accelaration of the simulation at will. It is a multitask application. SORYA-MX has two levels of use :(1 ) A designerlevel for configuring the simulator by creating, for example, mimic diagrams, alarm levels, operating data sheets, controllers, disturbance scenarios, etc. (2) A userlevel for activating the functionalities defined by the designer level, for controlling the controllers, for safeguarding, etc. As the result of this environment, learning on the simulator is much like operating the unit via a centralized control system. This type of simulator has already shown proof of its effectiveness :(a) At the level of research, as a working support for the designing of expert systems or for designing advanced controls. (b) At the level of training, for operators of units operating under an IFP process license. By a fairly fine kinetic model for oligomerization and careful programming, the model thus developed can be used as a basis for other studies. In the future, it seems desirable to : (a) Increase the degree of modeling (reactors in cascade, modeling of impurities and of start-up). (b) Extend modeling to similar processes. Nevertheless, for pursuing this project it is absolutely necessary to arouse greater interest among customers during the selling of processes for this training tool and among process engineers on the spot for obtaining a better understanding of internal phenomena.