PEM Fuel Cell Analysis, Circular Geometry - Solver Crashing

  • Last Post 21 December 2018
AndrewNMurphy posted this 14 December 2018


I need some help with the fuel cell simulation that I'm attempting to run.

The geometry is flat circular profile with a disc shaped membrane stack. The reactants are flowed over each side of the membrane stack. I've reviewed the ANSYS how-to video on YouTube and tried to configure my model similarly but when I run the calculation, there are numerous errors and the solver crashes. I created the attached PDF to try to describe what I'm trying to model and what I've done so far. If anyone can see something obvious that I've set up, wrong, I would be grateful for the help.

After encountering all the trouble with the fuel cell module, I tried running the model with the fuel cell module turned off (just running air through the flow channels). With that configuration, the model converges well in less than 100 iterations with no errors so I'm thinking the mesh is at least adequate. 

Thanks for the help,


Attached Files

Order By: Standard | Newest | Votes
rwoolhou posted this 17 December 2018

Please can you post images into the thread, as staff we're not allowed to open attachments (don't ask....). What error messages are you seeing, some aren't overly useful but several can be very diagnostic. 

AndrewNMurphy posted this 20 December 2018

Thanks for the reply. Below is an image of the fuel cell I was attempting to run. The gas flows into/out of the small cap-like circular bodies and the gas flows over (anode) and under (cathode) the element stack in the middle. I have since updated the geometry to simplify it by removing parts of the flow channels that don't contribute to the reaction.

Old geometry (meshed) shown below. The geometry below is a multi-part assembly with each part sweep meshed with contact regions to link all the parts.


Errors encountered when running the solver (fuel cell module):

Error Messages (Old Geometry)

Most of the errors I was seeing were the divergence detected for the various species and the reversed flow at the outlets. 



After the solver failed, I tried making some changes to the geometry to simplify the bodies. I eliminated the vertical members with the circular tops (gas inlets/outlets) and trimmed the gas flow paths to be as short as possible. At the moment, I'm finished with the geometry but have not yet completed the meshing. My plan is to make the assembly a multi-body part to achieve conformable meshing.





I have two specific questions that I can use some help with:

1. Along with the simplified geometry, I'm attempting to make sure the mesh is both conformable and swept to maximize quality and efficiency. Do you know if the fuel cell module requires a conformable mesh to function? Or are defined contact regions permitted?

2. I read an article that suggested that shadow wall pairs showing up in the solver should be eliminated. Would you be willing to help me understand that a little better? Do we need any shadow walls or should they all be deleted in the solver?


Thanks for all the help,


rwoolhou posted this 21 December 2018

I'm not sure about the fuel cell model, but having a conformal mesh is likely to help. You also need to make sure you have 10+ cells across a flow channel and that the cell quality is good. 


A wall - wall:shadow pair means you have a thin wall. Ie the wall is a single surface and has cells on both sides. They're not "good" or "bad", they're just part of the boundary definitions. If you need to you can give a coupled wall a thickness in the solver for heat transfer purposes. 

  • Liked by
  • peteroznewman