Tutorials Guide

General structure

Consider the cross-section of a fuel cell as an example. The computational domain provides:

In a PEM fuel cell or other types, there are typically several regions or zones: air, fuel, electrolyte (membrane), and interconnect (bipolate or monoplate). You can find more information about this on the openFuelCell repository. However, additional regions such as phiEA, phiEC, and phiI are also necessary to account for electron, ion, and dissolved water transfer.

To consider the coupling transfer problems in a fuel cell, a parent mesh (main region) and several child meshes (local regions) are used. In the main region, the global energy/enthalpy equation is solved, while in the local regions, corresponding partial differential equations are discretized and solved. During simulation, material properties such as density and thermal conductivity are mapped from local regions to the global region, while the temperature field is mapped from the global region to the local regions.

The local regions can be classified into three different types: fluid, solid, and electric regions. Refer to /src/libSrc/fuelCellSystems/regions for more details.

• Fluid region:

  This region represents the space where fluid flows. In a fuel cell or electrolyzer, it consists of gas channels and/or porous zones. The following processes are addressed in this region:

  • Fluid flow (single/two phase)
  • Species transfer
  • Electrochemical reaction
  • Heat and mass transfers

  For example, in a fuel cell, the following parts apply to this region,

  • Air/fuel flow paths + diffusion layer + electrodes
  • Cooling channels

---

• Solid region:

  This represents the solid components in a fuel cell or electrolyzer. In the current framework, no equations will be solved here. However, stress analysis during assembly and thermal effects may be implemented for future applications.

  For example. in a fuel cell, the following components apply to this region,

  • Electrolyte/membrane
  • Interconnect/Bipolar-plate
  • Endplate

---

• Electric region:

  This region accounts for the electric-conductive components. It is designed to consider electron/ion transfer specifically. However, it is found that dissolved water takes place along with the region where proton transfer transfer takes place. Therefore, a switcher is enabled in the code to turn on/off the dissolved water transfer model. The following equations will be solved:

  • Potential equations (Possion equations)
  • Dissolved water transfer equations (diffusion and electro-osmotic-drag)

  For example, in a fuel cell, the following components belong to this region:

  • Bipolar-plate, GDL, MPL, CL -> electron transfer regions
  • Catalyst coated membrane, CCM, -> ion transfer and dissolved water transfer region

---

• main region:

  The heat source/sinks in local regions will be mapped to this region, and the obtained temperature is mapped back to the local regions. The heat source/sink includes:

  • Joule heat from the electric regions.
  • Condensation/evaporation in the fluid regions.
  • Electrochemical reactions in the fluid regions.

How to mesh

You have the freedom to choose any meshing tool, as long as it can reproduce mesh files in OpenFOAM format. In our test tutorials, you can utilize either make mesh or make salomeMesh to access the built-in meshers, blockMesh or salome, respectively. These will help you create the initial mesh and run pre-processing scripts to generate the required primary and sub-region meshes.

If you opt for using other meshing tools, it is recommended to name sub-domains (or zones/regions) as illustrated in the figure below:

• interconnect : the solid regions
• cchannel : channels at the cathode side
• cathode : cathode porous layer
• cmpl : cathode micro porous layer
• ccl : cathode catalyst layer (electrode)
• electrolyte : the ion-conducting electrolyte/membrane
• acl : anode catalyst layer (electrode)
• ampl : anode micro porous layer
• anode : anode porous layer
• achannel : channels at the anode side

Creation of the meshes

The mesh generation process originates from the openFuelCell project and involves two sets of meshes: the main (parent) mesh and sub (child) meshes. Sub-meshes are closely linked to the main mesh. In each sub-mesh, several polyMesh files, cellRegionAddressing, faceRegionAddressing, boundaryRegionAddressing, and pointRegionAddressing, indicate cell, face, boundary, and point IDs within the main mesh. These IDs facilitate data transfer between sub-meshes and the main mesh, as well as between different sub-meshes.

To obtain these files, use the splitMeshRegion tool. Before running this command, categorize all cells into non-overlapping cell zones. The sub-mesh creation involves three steps:

• Use topoSet to create the cellZones file in constant/polyMesh, containing four zones: air, fuel, electrolyte, and interconnect. Since these zones do not overlap and cover the entire mesh, splitMeshRegion can be executed successfully. You can overwrite the existing sub-meshes with the -overwrite option or manually copy them from the result folder.
• Delete the cellZones file from constant/polyMesh and generate new cell zones, including phiEC, phiEA, and classify the remaining cells as phiE0. Sub-meshes phiEC and phiEA can then be created with splitMeshRegion.
• Delete the cellZones file and generate new cell zones, including phiI, and classify the rest as phiI0. This will allow you to create the phiI sub-mesh with splitMeshRegion.

Following these steps, you can create the sub-meshes: air, fuel, electrolyte, interconnect, phiEC, phiEA, and phiI. To create zones within each sub-mesh, execute topoSet with the source set to masterRegionToCell in each action defined in topoSetDict. masterRegionToCell collects cells in the corresponding cellZone from the main mesh and filters them to match the sub-mesh.

How to decompose

In the context of parallel simulations or high-performance computing, domain and mesh decomposition are crucial. Various decomposition methods, such as simple, hierarchical, metis, and scotch, are commonly used, and they can significantly impact parallel performance.

Decomposition in single-domain computations is relatively straightforward, but it becomes complex when multiple regions exist and are interconnected. In such cases, data transfer and sharing occur between regions on each core. If these regions are coupled through boundaries, one solution is to have the main core store all coupled boundaries, allowing sub-cores to access boundary values through the main core.

However, if regions are coupled through overlapped zones, like electrodes, it becomes challenging. To address this issue in the current implementation, we employ manual decomposition:

As depicted in the figure, the domain can be divided evenly, e.g., into nx=8 and ny=8 sub-domains. Cells within each corresponding sub-domain are labeled the same. This labeling ensures that cells in the same locations across different regions share identical processor IDs. Thus, in overlapped zones, one region can access data from another. You can generate decomposition cell IDs using make decompose.

The steps for creating the main mesh and sub-meshes for parallel simulations are similar to those outlined in the mesh creation section above, and you can find them in the parallel.sh script.

Tutorial Guides

• Conjugate heat transfer
• High-temperature proton exchange membrane fuel cell
• High-temperature proton exchange membrane electrolyzer
• Proton exchange membrane fuel cell
• Proton exchange membrane electrolyzer
• Solid oxide fuel cell
• Solid oxide electrolyzer
• Hydrogen pump