Integrated Model of Chemical Reactor

In this work, the bond graph approach is employed to develop an integrated chemical reactor model, able to be exploited for supervision and diagnosis of industrial process. The obtained model is evolutive and generic comparing with the first principle methods. Moreover, this methodology is based on systematic approach which allows to represent the dynamic behaviour of complex process under non linear state space equations. The application discussed in this work is a chemical process including CSTR (Continuous Stirred Tank Reactor), where it occurs two exothermic reversible and irreversible reactions. To illustrate the modelling procedure, a simulation results have been performed in the cases when the mixture is considered as ideal and real aqueous solution. INTRODUCTION In the present paper, the bond graph approach is adapted to model chemical reactors, which represents in process engineering, one of the most complex systems to model because of there non linear behaviour due to the interactions between the different energy (hydraulic, thermal and chemical) within the system. The modelling approach used here, reproduce an integrated model using a graphical unified language. Furthermore, the elaborated model is associated readily with industrial equipments models (centrifugal pumps, compressors, heat exchangers, ...etc) to get the overall process bond graph model, needed for supervision and FDI (Fault Detection and Isolation) procedures using model based methods (Iserman 1984). Previously, the bond graph tool has been designed for hydraulic and electrical systems (Paynter 1961), and after that its application has been extended to chemical reactions (Auslander et al. 1972), thermofluid systems (Thoma 1971) (Karnopp and Azerbaijani 1981). However, for chemical reactors, the approach is not well developed yet. Among published works, we cite (Delgado et al. 1999) (Breedveld et al. 2003). In the pseudo bond graph models elaborated in (Delgado and al. 1999), the concentration has been chosen as effort variable in chemical domain to describe the reaction kinetic in the case of irreversible reactions. However, in our case, when the reaction is reversible, the choice of concentration is not well adapted to represent, and analyse the reaction dynamic behaviour, particularly near of the equilibrium. That is why; an adequate variable is required to be employed in such case (reversible and irreversible reactions). Thus, the chemical potential μ is chosen in this paper as effort variable, used before by (Auslander et al. 1972), (Cellier 1991), (Thoma and Ould Bouamama 1999). In contrast to the model proposed by (Breedveld et al. 2003), where the entropy flow S is chosen in thermal domain as flow variable, we have proposed to choose in our application the enthalpy flow H , suitable in the case of open systems (industrial equipments, continues reactors,...etc), where the heat is transferred by convection (Karnopp and Azerbaijani 1981). So to get the integrated reactor model needed, the choice of enthalpy flow H is suggested instead of entropy flow S , in order to use the overall process for supervision and diagnosis applications. The paper is organized as follows: after brief description of the application example, the process word bond graph is presented, and the reactor bond graph model is developed in details to be written after under state space representation generated systematically. In the simulation part the bond graph model is implemented in SYMBOLS software (Mukherjee and Samantaray 2001) to get results which will be discussed and commented. Finally, the conclusions and perspectives are presented according to the results obtained before.