[aarvay]

I'm trying to create a mesh for a PEM fuel cell using ICEMCFD and I'm having a heck of a time getting it to converge. I feel like there is something obvious I am missing and so I'm appealing to the community to help me find my mistake. This is all in version 12.1

I've gone back to the demo fuel cell which is described in this tutorial(Modeling a Single-Channel, counter-flow polymer electrolyte membrane (PEM) Fuel cell). I got it from the fluent site somewhere, but I can't find where I originally found it. This document talks about meshing using gambit or something, but I was able to mostly replicate the results by using workbench meshing. I've found the workbench mesher to be inadequate for more complicated geometries so I've switched to using ICEM CFD.

now I am trying to do the tutorial using ICEM to mesh the geometry and its not working. I'm able to generate a mesh (using cartesian grid method) and it seems like a good mesh to me, but it will not converge in Fluent. I have attached case files for each type, workbench mesh and the ICEM mesh. You will need to have the fuel cell license available to actually intialize the solution. The Workbench mesh does converge (well atlesat it doesn't immediately diverge) while the ICEM mesh immediately explodes.

I'm trying to understand the differences between these two meshes and why one works and one doesn't. I'm asking anyone that knows anything about the FC module to load up these two cases and tell me why the Workbench one works and the ICEM one doesn't. And also some suggestions on how to fix it.


Workbench case file
ICEM case file

I can provide related geometry/mesh files if you think that will help you.

Thanks for any info.

For future reference you can find the fuel cell module manual here

[aarvay]

the one thing I can notice that seems to be important is the creation of shadow walls. In the workbench mesh, shadow walls are only created at the boundary between solid and fluid parts. In the ICEM mesh, shadow walls are being created at boundary of every part. I suspect this is the cause of my issues but I do not know how to resolve it.

[aarvay]

Well I figured out my problem. It turns out it was something trivially simple. The two meshes had different physical sizes. I suppose the ICEM meshes default to meter scale, while the other was in mm. So the non-working mesh was 1000x larger than the other. Hopefully someone else will find this helpful.

edit for my own future reference: There was in fact a problem with those shadow walls. I was able to work around it by setting my material properties accordingly (solids and liquids) then turning one of the wall/shadow-wall pairs into an internal. I did this to all shadow walls in the geometry. I could not change shadow walls at liquid-solid interfaces into internals which is exactly how it should be.

[LWX]

I think I have encountered the same difficulties as you have by modeling the same case(Tutorial: Modeling a Single-Channel, Counter-Flow Polymer
Electrolyte Membrane (PEM) Fuel Cell),While I use the Gambit to get the mesh file.There is still something wrong with the mesh after several modifications and now I can't find where the problem is as I think my mesh is perfect.Utill last night,I read your post and I think you can help me.The words below illustrate the warnning when I import the mesh file into the FLUENT.
Warning: Thread 18 has 2 contiguous regions.
Warning: Thread 19 has 2 contiguous regions.
creating wall-gdl-c-shadow
creating wall-gdl-a-shadow
creating wall-ch-c-shadow
creating wall-ch-a-shadow
shell conduction zones,
Done.

I am a Chinese and my English is poor,so expect you can understand the words above.What's more,this is my first time to loggin this forum.

[aarvay]

Originally Posted by LWX

I think I have encountered the same difficulties as you have by modeling the same case(Tutorial: Modeling a Single-Channel, Counter-Flow Polymer
Electrolyte Membrane (PEM) Fuel Cell),While I use the Gambit to get the mesh file.There is still something wrong with the mesh after several modifications and now I can't find where the problem is as I think my mesh is perfect.Utill last night,I read your post and I think you can help me.The words below illustrate the warnning when I import the mesh file into the FLUENT.
Warning: Thread 18 has 2 contiguous regions.
Warning: Thread 19 has 2 contiguous regions.
creating wall-gdl-c-shadow
creating wall-gdl-a-shadow
creating wall-ch-c-shadow
creating wall-ch-a-shadow
shell conduction zones,
Done.

I am a Chinese and my English is poor,so expect you can understand the words above.What's more,this is my first time to loggin this forum.

Hey,

Fluent will automatically create shadow walls at interfaces between solid and liquid zones in your mesh. This is normal and you should be able to run your simulation without any issues.


If you are having problems, I don't think the shadow walls are the cause. What are problems are you having?

[LWX]

I specified all the interfaces as "interface" in the boundary condition panel while meshing in the Gambit first time and the Fluent gived me the warnings above and checked the mesh failed with the warnings like:Unassigned interface zone detected for interface 11. Then I modified all the interfaces boundary conditions to interior boundary condition and the warnning still appeared while there was nothing wrong when checking the mesh. So I want to know how you specified the boundary conditions especially the face between the flow channel and the GDL.

[aarvay]

Originally Posted by LWX

I specified all the interfaces as "interface" in the boundary condition panel while meshing in the Gambit first time and the Fluent gived me the warnings above and checked the mesh failed with the warnings like:Unassigned interface zone detected for interface 11. Then I modified all the interfaces boundary conditions to interior boundary condition and the warnning still appeared while there was nothing wrong when checking the mesh. So I want to know how you specified the boundary conditions especially the face between the flow channel and the GDL.

As you know, I generated my mesh using ICEM CFD and I didn't bother to assign any boundaries during mesh generation. By default in ICEM all the interfaces were walls. After I imported it into fluent, all of the interfaces were converted to wall/shadow wall pairs automatically by fluent. I left the solid/liquid interfaces as walls and changed the rest of the interfaces to internals.

[elemad1987]

Hi Adam,
I had find your master thesis and read it, it was very useful to me.
I'm Emad from Egypt study master degree in PEM fuel cell in Assiut University.
I want to ask you a lot about your thesis, can you send an email to me, to know your mail.
Or can you send me your final case to understand how you did your mesh.
I want to ask you about mesh using ANSYS.
Regards,
Emad
elemad1987@gmail.com

[Diab]

Hi,

I am a starter in the field of modeling the fuel cell using ansys fluent. May anybody provide me some sources from which I can learn how to perform my simulation

Thank you in advance

[aarvay]

Hey Diab,

I did take a stab at writing an intro manual for doing PEMFC modeling using fluent based on the notes I took while I was going through the process. I had planned on making this a much cleaner document and making it available but apparently i couldn't be bothered to. I don't like to release low quality work, but it will have to suffice since I don't think anything else exists along these lines. I've passed this thing around to a few people in emails and stuff, but I don't think I ever posted it online. Hopefully it is helpful.

Ansys fluent PEMFC manual

[A CFD free user]

@ Adam, here's an opptunity for me to thank you very much for the intro manual you gave me. Thank you again and it helped me so much.
good luck

[Diab]

Dear aarvay,

Thank you very much for the introduction manual. It seems very beneficial. I am really thankful

Best Regards

[leily]

Hi
I want to simulate single PEM fuel cell using fluent 6.3 for study concentration and temperature distribution in the flow channels of the bipolar plates and performance of the PEM fuel cell. But I don't know, what do I need data to simulate?
Do I write the new UDF for this study?
can anyone explain to me step by step process of the simulation?

[leily]

Hi
I want to simulate single PEM fuel cell using fluent 6.3 for study concentration and temperature distribution in the flow channels of the bipolar plates and performance of the PEM fuel cell. But I don't know, what do I need data to simulate?
Do I write the new UDF for this study?
can anyone explain to me step by step process of the simulation?

[A CFD free user]

Originally Posted by leily

Hi
I want to simulate single PEM fuel cell using fluent 6.3 for study concentration and temperature distribution in the flow channels of the bipolar plates and performance of the PEM fuel cell. But I don't know, what do I need data to simulate?
Do I write the new UDF for this study?
can anyone explain to me step by step process of the simulation?

If you want to model what you mentioned in above statement, I should say no, there is no need to employing UDF in that case. But, it surely depends on what you looking for. If you want to know a more correct answer, you have to offer more details about the problem.
Regarding to the second question, there is a PEMFC tutorial you can find easily On web. I recommend that you do it first to get some insight of performing a real simulation in PEMFC system. doing the tutorial, you can understand at least what type of question you face in a typical PEMFC system and how to run a simple case.
Hoe it helps

[leily]

Originally Posted by A CFD free user

If you want to model what you mentioned in above statement, I should say no, there is no need to employing UDF in that case. But, it surely depends on what you looking for. If you want to know a more correct answer, you have to offer more details about the problem.
Regarding to the second question, there is a PEMFC tutorial you can find easily On web. I recommend that you do it first to get some insight of performing a real simulation in PEMFC system. doing the tutorial, you can understand at least what type of question you face in a typical PEMFC system and how to run a simple case.
Hoe it helps

Dear A-A Azarafza
Thank you very mach.
I want to compare the concentration and temperature distribution in the two types of the serpentine flow channels. I draw geometry in the Gambit and I run it by fluent according to manual.
I think, I entered incorrect boundary conditions. this photos clearly show the boundary conditions that I have set.
I don't know, these boundary condition are true or not true????
I don't know, why I can't see any result of residual in the curve???
I don't know, which areas should be related to the voltage jump?
these are my problems for draw geometry and run of a single PEMFC.
could you give me your email address that I sent photos and help me for solve these problems?
Thanks you so much.
best regards

[vinhgsyb]

My email vinhgsyb@yahoo.com

[SAM_R]

I'm using PEMFC addon module of fluent and model of single pem fuel cell channel was made in Gambit 2.4 according to below open online tutorial file:
ssadeghi.ir_tutorial_modeling_pem_fuel_cells

only my cell having active area of 16 cm2 & hence channel length is 40 mm.

with same flow rate or by lowering flow rate solution is converging on 0.02254 A/cm2.

where may I have made mistake?

I have gone through both threads of Adam Arvay, I have also read appendix of TUI of obayopo's thesis for gambit and Adam's for fluent help.

Almost all my problem have solved but above mentioned problem

[SAM_R]

@ Adam Arvay
please mail me PEMFC Manual
mail:ranasaumy@gmail.com

[SAM_R]

FLUENT's PEM add-on module supports the use of non-conformal grid interfaces. In such cases, it is recommended that the INTERFACE boundary type be used halfway between the membrane continuum zone, and that the mesh on opposite sides of the interface have similar size, aspect ratio, and orientation. In such cases, the membrane will consist of two fluid zones.

I can't understand above line.

Please help.