The influence of active area and stacking on PEM fuel cell performance: a simulation modelling and experimental investigation
Xue-Song Wu
Abstract
Xue-Song Wu
Abstract
This thesis focuses on the computer simulation of proton exchange membrane (PEM) fuel cells, both single fuel cells and fuel cell stacks. PEM fuel cells as a critical component in hydrogen fuel cell vehicles and hydrogen-based energy storage systems are likely to play an important role in sustainable energy economies based essentially on energy efficiency and renewable energy. The design of PEM fuel cells is a very complex multi-physics task that is greatly facilitated by a multi-component computer simulation model, incorporating all the necessary theoretical equations, to predict cell and stack performance. In this thesis, the ANSYS Fuel Cell Module was chosen to perform the simulation. This Module incorporates finite-element programs (ANSYS Fluent) for computational fluid dynamics to model fluid flow in both liquid and vapour phases, and an electromagnetic suite of programs to model electrical conductivity, both electron and proton, through various materials. The inputs for the ANSYS Fuel Cell Module are three-dimensional geometry and mesh, parameters of different layers, such as density, conductivity and reference current density, and operating environmental conditions such as gas temperatures, pressures and flow rates. The outputs of the ANSYS Fuel Cell Module included total current, and spatial distributions across the active area of current density, gas concentration, water saturation and water content. In the course of the research for this thesis, an unnecessary and unphysical limitation in the allowed input values for charge transfer coefficients in Butler-Volmer equations in the then current version of the ANSYS Fuel Cell Module was identified. Corrections to the program script were therefore made to solve the problem. In this project, the ANSYS Module has been used first to simulate three small fuel cells (5 cm2 active membrane area).The ANSYS simulation output for the VJ curve of the first small fuel cell, using input parameters from an earlier study, compared closely with previous experimental results for this cell. The second small fuel cell used the same geometry of the first one, but different materials for the catalyst and gas diffusion layers. The input values for the material properties were then changed in the simulation to obtain the best fit to the experimental VJ curve obtained in the present project. The VJ curve of a third small cell with a different flow channel geometry was also measured experimentally, and also modelled in the ANSYS Module. Varying input parameters in the simulation allowed a good fit between experimental and the simulation output. The best-fit input parameter values for this cell were used in the later simulation of a larger fuel cell employing the same materials, and also in the fuel cell stack modelling. A large fuel cell (225 cm2 active membrane area) has also been simulated using the ANSYS module, and the VJ curve obtained compared with experimental measurements for the same cell. The gas flow rates were about 45 times those for the small fuel cell. It was found that the average current density of the large cell was lower than the average current density of small fuel cell at same voltage in both the experiment and simulation. The spatial distribution of current density in the large fuel cell showed an uneven distribution of current density with large variations across the active area. The current density distribution from the ANSYS simulation was able to be explained in terms of the corresponding gas concentration distributions. The experimental VJ curve for the large cell displayed clear indications of a sharp turndown at high current density due to mass transfer constraints. The ANSYS Module can represent this turn-down through increasing water saturation and decreasing oxygen concentration, but it has not possible to change the input parameter values sufficiently to get an onset of this constraint at lower current densities as was found in the experimental V-J curve.
OpenAlex reports 2 citations for this work. Citation counts describe recorded attention and do not establish research quality.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
This thesis focuses on the computer simulation of proton exchange membrane (PEM) fuel cells, both single fuel cells and fuel cell stacks. PEM fuel cells as a critical component in hydrogen fuel cell vehicles and hydrogen-based energy storage systems are likely to play an important role in sustainable energy economies based essentially on energy efficiency and renewable energy. The design of PEM fuel cells is a very complex multi-physics task that is greatly facilitated by a multi-component computer simulation model, incorporating all the necessary theoretical equations, to predict cell and stack performance. In this thesis, the ANSYS Fuel Cell Module was chosen to perform the simulation. This Module incorporates finite-element programs (ANSYS Fluent) for computational fluid dynamics to model fluid flow in both liquid and vapour phases, and an electromagnetic suite of programs to model electrical conductivity, both electron and proton, through various materials. The inputs for the ANSYS Fuel Cell Module are three-dimensional geometry and mesh, parameters of different layers, such as density, conductivity and reference current density, and operating environmental conditions such as gas temperatures, pressures and flow rates. The outputs of the ANSYS Fuel Cell Module included total current, and spatial distributions across the active area of current density, gas concentration, water saturation and water content. In the course of the research for this thesis, an unnecessary and unphysical limitation in the allowed input values for charge transfer coefficients in Butler-Volmer equations in the then current version of the ANSYS Fuel Cell Module was identified. Corrections to the program script were therefore made to solve the problem. In this project, the ANSYS Module has been used first to simulate three small fuel cells (5 cm2 active membrane area).The ANSYS simulation output for the VJ curve of the first small fuel cell, using input parameters from an earlier study, compared closely with previous experimental results for this cell. The second small fuel cell used the same geometry of the first one, but different materials for the catalyst and gas diffusion layers. The input values for the material properties were then changed in the simulation to obtain the best fit to the experimental VJ curve obtained in the present project. The VJ curve of a third small cell with a different flow channel geometry was also measured experimentally, and also modelled in the ANSYS Module. Varying input parameters in the simulation allowed a good fit between experimental and the simulation output. The best-fit input parameter values for this cell were used in the later simulation of a larger fuel cell employing the same materials, and also in the fuel cell stack modelling. A large fuel cell (225 cm2 active membrane area) has also been simulated using the ANSYS module, and the VJ curve obtained compared with experimental measurements for the same cell. The gas flow rates were about 45 times those for the small fuel cell. It was found that the average current density of the large cell was lower than the average current density of small fuel cell at same voltage in both the experiment and simulation. The spatial distribution of current density in the large fuel cell showed an uneven distribution of current density with large variations across the active area. The current density distribution from the ANSYS simulation was able to be explained in terms of the corresponding gas concentration distributions. The experimental VJ curve for the large cell displayed clear indications of a sharp turndown at high current density due to mass transfer constraints. The ANSYS Module can represent this turn-down through increasing water saturation and decreasing oxygen concentration, but it has not possible to change the input parameter values sufficiently to get an onset of this constraint at lower current densities as was found in the experimental V-J curve.
Key concepts: Proton exchange membrane fuel cell, Fuel cells, Unitized regenerative fuel cell, Hydrogen fuel, Hydrogen, Nuclear engineering, Stacking, Materials science