Monday, November 5, 2007 - 10:10 AM

Dynamics and Volterra-Model Based Control of a Tubular Solid Oxide Fuel Cell ( Sofc)

Debangsu Bhattacharyya1, Raghunathan Rengaswamy1, and Finnerty Caine2. (1) Chemical Engg. Dept., Clarkson University, 8 Clarkson Avenue, Potsdam, NY 13699, (2) NanoDynamics,Inc., 901 Fuhrmann Blvd., Buffalo, NY 14203

Solid Oxide Fuel Cells (SOFC) operate at high temperatures. Given the possibility of significant temperature gradients and fast transients, a properly implemented controller is essential for efficient use of the fuel and a high operational life-time. First a detailed transient analysis of the cell is performed and then a controller is implemented.

A two dimensional dynamic model for an anode-supported tubular SOFC is considered. The model includes: (i) mass and momentum transport phenomena in the anode and cathode gas flow channels for reactants and products, (ii) diffusion from the gas flow channels through the porous electrodes to the reaction sites, (iii) activation overpotential through the Butler-Volmer equation with concentration and temperature dependent expression for exchange current density, and (iv) ohmic resistances. Step inputs in cell terminal voltage and inlet hydrogen flow rate are introduced and the transient response of the overall system is analyzed. A MAPLE-MATLAB environment is used to yield the transient response for a number of key variables. The model is validated using data from a commercial SOFC over a wide range of cell temperatures, reactant flow rates, DC polarizations and amplitudes of step. Processing of the data by implementation of the filters for noise reduction and the strategies adopted for validation of the model will also be presented.

We observe that the time constants of the cell change widely depending upon the inputs, operating conditions, the spatial location and the transport field that is considered. We have also studied the varying gains of the system depending upon the operating conditions and the directionality of the step. The results from this study are used in the synthesis of a second-order Volterra-model based controller. Volterra-kernels are derived from the transient response of the well-validated dynamic model of the cell. The servo response and the disturbance rejection properties of the controller are then tested. We will discuss the performance of this controller in comparison with PID and linear IMC controller.