For efficient operation, as well as to avoid operating conditions that can cause damage, fuel cells require a control system to balance fuel and air supply and electrical load. The need to maintain signal constraints during operation, combined with importance of unmeasured variables such as internal stack temperature or fuel utilization, indicate the need for control-oriented models that can be used for estimation and model predictive control. In this paper, we discuss the development of a controloriented dynamic model of a solid oxide fuel cell stack. Using a detailed physical model as a starting point, we demonstrate the utility of a linear parameter varying (LPV) model structure as a mechanism for model reduction. A novel feature is a nonparametric method for determining the scheduling functions in this model.
INTRODUCTIONFuel cells are devices that enable the direct conversion of chemical energy into electrical energy, with a theoretical conversion efficiency much higher than for heat engines [1,2]. The main component of a cell is a anode/electrolyte/cathode structure, which is exposed to fuel on the anode side and an oxidant (usually air) on the cathode side. Solid-oxide fuel cells (SOFCs) utilize a ceramic oxygen-ion conducting electrolyte. Oxygen ions are formed at the cathode, which migrate through the electrode. At the anode, these ions react with hydrogen and carbon monoxide to produce water and carbon dioxide, as well as releasing electrons that flow through an external circuit back to the cathode. A particular characteristic of SOFC fuel cells is their