The National Renewable Energy Laboratory (NREL) and Plug Power Inc. have been working together to develop fuel cell modeling processes to rapidly assess critical design parameters and evaluate the effects of variation on performance. This paper describes a methodology for investigating key design parameters affecting the thermal performance of a high temperature, polybenzimidazole (PBI)-based fuel cell stack. Nonuniform temperature distributions within the fuel cell stack may cause degraded performance, induce thermo-mechanical stresses, and be a source of reduced stack durability. The three-dimensional (3-D) model developed for this project includes coupled thermal/flow finite element analysis (FEA) of a multi-cell stack integrated with an electrochemical model to determine internal heat generation rates. Sensitivity and optimization algorithms were used to examine the design and derive the best choice of the design parameters. Initial results showed how classic design-of-experiment (DOE) techniques integrated with the model were used to define a response surface and perform sensitivity studies on heat generation rates, fluid flow, bipolar plate channel geometry, fluid properties, and plate thermal material properties. Probabilistic design methods were used to assess the robustness of the design in response to variations in load conditions. The thermal model was also used to develop an alternative coolant flow-path design that yields improved thermal performance. Results from this analysis were recently incorporated into the latest Plug Power coolant flow-path design. This paper presents an evaluation of the effect of variation on key design parameters such as coolant and gas flow rates and addresses uncertainty in material thermal properties.
Continuous Adjoint Approach for the Spalart-Allmaras Model in Aerodynamic Optimization
