Abstract
Solid oxide fuel cells (SOFCs) are receiving considerable interest since they are suited for both stationary and vehicle applications. The reduction of activation polarization, the elimination of expensive catalysts, potential integration with cogeneration systems, and the possibility of being able to consider different composition of syngas as the fuel are interesting technical challenges. The Department of Energy (DOE) of the United States a few years ago initiated a set of research projects (SECA - Solid State Energy Conversion Alliance) with the purpose of increasing the power density, reducing the manufacturing costs, and encouraging commercially cost-effective prototypes. The European Union, through the European Hydrogen and Fuel Cell Technology Platform: Strategic Research Agenda (January 2005), has indicated that the SOFC is a priority choice for stationary applications.

Mathematical models that predict performance can aid in the understanding and development of solid oxide fuel cells (SOFCs). Of course, various modeling approaches exist involving different length scales. In particular, very significant advances are now taking place using microscopic models to understand the complex composite structures of electrodes and three-phase boundaries. Ultimately these advances should lead to predictions of cell behavior, which at present are measured empirically and inserted into macroscopic cell models.

In order to achieve this ambitious goal, the key idea is to numerically simulate the underlying microscopic phenomena in an effort to bring the mathematical description nearer to actual reality. In particular, some recently developed mesoscopic tools appear to be very promising since the microscopic approach is, in this particular case, partially included in the numerical method itself. In particular, the models based on the lattice Boltzmann method (LBM) treat the problem by reproducing the collisions among particles of the same type, among particles belonging to different species, and finally among the species and the solid obstructions.

Recently, a model developed by the authors was proposed which, based on LBM, models the fluid flow of reactive mixtures in randomly generated porous media by simulating the actual coupling interaction among the species. A parallel three–dimensional numerical code was developed in order to implement this model and to simulate the actual microscopic structures of SOFC porous electrodes. The code has been developed in C++ and a free communication library has been adopted (MPICH 1.3) based on MPI technology. The reported numerical results were obtained on two cluster facilities. The first one is System X at Virginia Tech (VT). It is essentially a computational platform made of 1100 dual-processor Apple XServe G5 (2200 total CPUs, each characterized by 2.3 GHz, 4 GB RAM e 80 GB HD), connected by Cisco Gigabit Ethernet and Mellanox switches. The second facility has been recently developed at “Politecnico di Torino” and it is made of 100 Pentium-4 nodes (each characterized by 2.8 GHz, 512 MB RAM and 40 GB HD).