The feasibility of direct internal reforming in SOFCs has been widely investigated, but at present it is not a state-of-the-art option for SOFC systems, which, instead, embed a Methane Steam Reforming (MSR) reactor prior to the SOFC stack itself [1]. Despite these reactors implement an established technology, they still present some issues. In particular, carbon can deposit under the form of whiskers and coat the active sites of the catalyst, reducing its activity and limiting gas diffusion through the catalytic bed. Even if it is well known that coke deposition is inhibited by a high steam to carbon (S/C) ratio, generally in the range 2-6, nevertheless, the problem of carbon deposition is still not solved, as widely discussed in the literature [2]. In case coke deposition occurs in the MSR reactor, the reformate fuel contains high levels of methane. This could force the SOFC stack to perform direct internal reforming, which could then lead to dangerous SOFC stack operation [1], with high solid temperature gradients associated to mechanical stresses, fractures and eventually break-down. Early detection, isolation, identification and possibly correction of a coke deposition fault in the MSR reactor is essential in view of reducing the maintenance cost of SOFC plants, and to increase of their lifetime. To this end, Fault Detection and Isolation (FDI) tools, under development for SOFC systems, can be suitably tailored to be applied specifically to the MSR reactor. A number of FDI methods have been proposed and applied to a variety of systems. A survey can be found in [3], where they are classified as: (i) quantitative model-based ; (ii) qualitative model-based, and (iii) process history-based methods. In practice, the classification regards the approach used to produce the fault data for the FDI: in the first two cases, the data are obtained from a model developed for the system under analysis, while the latter approach is based on historical experimental data. In the present work, we propose a first-principle model based on a quantitative and detailed simulation developed for the MSR reactor, which is intended to provide the basis for the development of an FDI tool based on pattern recognition techniques, which will be developed in a subsequent phase. Modelling of industrial MSR reactors has received wide attention so far [4-5], mainly aimed at optimizing operating conditions and identifying innovative geometries, particularly with reference to the arrangement of the burners within the combustion chamber where the MSR reactor tubes are situated. These studies include the kinetic equations of the MSR reactions, but to date there are no literature papers in which the kinetics of the coke deposition reaction is embedded into the MSR reactor model, and this is a distinctive feature of the present work. Our MSR reactor model is based on microscopic balances of mass, energy and momentum, written under transient form. The evaluation of heat and mass transport phenomena within the catalyst pellets is included as well, together with, as already mentioned, chemical kinetics. The equations of the model form a PDAE (Partial Differential and Algebraic Equation) system, with appropriate boundary conditions, which is integrated numerically using a finite element method, implemented through COMSOL Multiphysics. Then the result is a 3-D and time-dependent distribution of all the chemical-physical variables within the MSR reactor, in particular composition, pressure and temperature of the reactant gas. An example of simulated temperature distribution in a full size industrial MSR reactor tube is reported in Fig. 1. In a subsequent development of this work, the MSR reactor diagnostics will be addressed through the classical model-based philosophy, based on the comparison between the chemical and physical variables measured in the reactor, and the same variables calculated by the model, under the hypothesis of unfaulty reactor operation. [1] Sorce A., Greco A., Magistri L., Costamagna P., “FDI oriented Modelling of an Experimental SOFC System, Model Validation and Simulation of Faulty States,” Applied Energy, vol. 136, pp. 894-908, 2014. [2] S. Helveg, J. Sehested, J.R. Rostrup-Nielsen, “Whisker carbon in perspective,” Catal. Today, vol. 178, pp. 42-46, 2011. [3] Venkatasubramanian V, Rengaswamy R, Yin K, Kaviri SN. “A review of process fault detection and diagnosis. Part I: Quantitative model-based methods,” Comput. Chem. Eng., vol. 27, pp. 293-311, 2003. [4] J. Xu, G.F. Froment, “Methane steam reforming: II. Diffusional limitations and reactor simulation,” AIChE J., vol. 35, pp. 97-103, 1989. [5] M. Behnam, A.G. Dixon, P.M. Wright, M. Nijemeisland, E.H. Stitt, “Comparison of CFD simulations to experiment under methane steam reforming reacting conditions,” Chem. Eng. J., vol. 207-208, pp. 690-700, 2012. Figure 1
Read full abstract