[mnobrega]

Comments and questions are welcomed on the article Weber et al., Open-source Computational Model for Polymer Electrolyte Fuel Cells, OpenFOAM Journal, February 2023, https://doi.org/10.51560/ofj.v3.50

[Pratik_21]

Nice work


I have been working on this model getting to run.Just one observation
In the Makefile, section regionMesh fails because of no constant folder in the subRegions directory.
Its quite easy fix.


My goal is to get this model running for PEM electrolyzers.
I would like to know from developers about possibilities of it.

[dl6tud]

Thank you for this hint. I corrected it.

[jose.ceballos]

Originally Posted by mnobrega

Comments and questions are welcomed on the article Weber et al., Open-source Computational Model for Polymer Electrolyte Fuel Cells, OpenFOAM Journal, February 2023, https://doi.org/10.51560/ofj.v3.50

Thanks for sharing this work, definitely will read it. I'm currently working with the J.P. Kone code and this paper expands my universe about open-source codes for fuel cell simulations.

I'll try to run it hoping to work with this tool in the future.

[Matt7]

Thank you for sharing the wonderful solver for PEFC! I would definitely like to working with this model!

While I'm still trying to understand the model, I would like to share some of my own opinions.

1. I noticed some disagreement with my understanding in the spices transfer equation in YairEqn.H and YfuelEqn.H. The velocity is the intrinsic velocity, so that it was timed by the porosity. And the diffusion term of the equation was also timed by the porosity and also (1-s). If I understand it correctly, the diffusion coefficient is already renewed (i.e. timed by porosity/tortuosity and (1-s)^1.5) to be the effective diffusivity before you solve the equation. Is there any possibility that you underestimated the diffusivity in your codes?

2. I'm also trying to run your example cases after I fixed the error of constant folder, as mention by Pratik_21. However, I found that the initial residual of the velocity didn't drop, even after I ran thousands of the time steps. I guess it can be fixed by changing the settings in fvScheme and fvSolution. I'm trying to change the PISO control to PIMPLE control in fvSolution of air and fuel regions. Would you mind providing me any suggestion for that?

3. I totally agree with the electrolyte model that you neglected the water transport in the in-plane direction, since it seems to have little impact on the results. Maybe the reason is that the charge transfer can be omitted, and the cathode catalyst layer field can be reduced to a boundary condition consequently, so that we don't need to solve the electrical potential field, which is really a computational sources saver. Please forgive my greed, but do you have any plan to compile the 3D water transfer model in electrolyte and charge transfer model in the future? Since the ohmic resistances may be critical in some other electrochemical system.

I would definitely appreciate your reply!
Thanks in advance!

[dl6tud]

I am very rarely in this forum, and do not get a message for new posts. In case of comments or questions I am always happy to receive an email.

[jose.ceballos]

Greetings;

Could anyone, please, indicate the steps to compile and run the application?

For example, I tried to compile with ./Allwmake inside of the directory .../fuelcellfoam-master/solver but I got errors.

Then at .../fuelcellfoam-master/testcase/ciemat I type make and start to iterate Surface refinement but it never ends.

In the case of the juelich folder, I can run make mesh to visualize the cell in Paraview, but when I type make run it says that there are not enough slots available in the system to satisfy the 16 slots that were requested by the application.

I would appreciate the help. Thanks in advanced.

[dl6tud]

Originally Posted by Matt7

Thank you for sharing the wonderful solver for PEFC! I would definitely like to working with this model!

While I'm still trying to understand the model, I would like to share some of my own opinions.

1. I noticed some disagreement with my understanding in the spices transfer equation in YairEqn.H and YfuelEqn.H. The velocity is the intrinsic velocity, so that it was timed by the porosity. And the diffusion term of the equation was also timed by the porosity and also (1-s). If I understand it correctly, the diffusion coefficient is already renewed (i.e. timed by porosity/tortuosity and (1-s)^1.5) to be the effective diffusivity before you solve the equation. Is there any possibility that you underestimated the diffusivity in your codes?

2. I'm also trying to run your example cases after I fixed the error of constant folder, as mention by Pratik_21. However, I found that the initial residual of the velocity didn't drop, even after I ran thousands of the time steps. I guess it can be fixed by changing the settings in fvScheme and fvSolution. I'm trying to change the PISO control to PIMPLE control in fvSolution of air and fuel regions. Would you mind providing me any suggestion for that?

3. I totally agree with the electrolyte model that you neglected the water transport in the in-plane direction, since it seems to have little impact on the results. Maybe the reason is that the charge transfer can be omitted, and the cathode catalyst layer field can be reduced to a boundary condition consequently, so that we don't need to solve the electrical potential field, which is really a computational sources saver. Please forgive my greed, but do you have any plan to compile the 3D water transfer model in electrolyte and charge transfer model in the future? Since the ohmic resistances may be critical in some other electrochemical system.

I would definitely appreciate your reply!
Thanks in advance!

I pushed these bugfixes. Many thanks for pointing out these mistakes!!!
If anyone needs a more detailed documentation than in the article, please feel free to contact me.

[Matt7]

Dear Jose,

Regards to your question, firstly, I think the code should be run under the environment of OpenFoam v2112. There may be some troubles to run with the different version of OpenFoam and something they are difficult to be fixed.

Then, when I compile the solver, I go to solver/, type the commands below in order

cd solver/libSrc
wmake libso
cd ../fuelCellFoam
wmake

Running (./Allwmake) in solver/ also works, but the permission to access the files is required, which can be procced by running like

chmod -777

And some errors can be fixed by running "wclean" before "wamke", if you have previously tried to compile the solver.

Next, when the solver is successfully compiled, I try to run e.g. case/juelich. I go to the case directory, and type (make All). However, (make All) is for parallel run. If your system is not 16 cores, you need edit the file in system/decomposeParDict, change "numberOfSubdomains" and "simpleCoeffs" if "method simple" is employed. For example, I run with 4 cores and my setting is like

numberOfSubdomains 4;

preserveFaceZones (iFaces);

method          simple;	

simpleCoeffs
{
    n               (2 2 1);
    delta           1e-16;
}

The product of 3 numbers (2 2 1) in simpleCoeffs/n in my case is 2*2*1=4, which should exactly equal to "numberOfSubdomains".

Or if you don't want to run with parallel, you can add a new line in Makefile in case/juelich as below, and run "make Alls" in command window.

Alls: mesh srun view

meanwhile the "srun" in lines 112-113 should be edited as

srun:
	(fuelCellFoam | tee log.srun);

p.s. I guess that the author wants to write "fuelCellFoam" instead of "uftFoam" originally.

Matt

Originally Posted by jose.ceballos

Greetings;

Could anyone, please, indicate the steps to compile and run the application?

For example, I tried to compile with ./Allwmake inside of the directory .../fuelcellfoam-master/solver but I got errors.

Then at .../fuelcellfoam-master/testcase/ciemat I type make and start to iterate Surface refinement but it never ends.

In the case of the juelich folder, I can run make mesh to visualize the cell in Paraview, but when I type make run it says that there are not enough slots available in the system to satisfy the 16 slots that were requested by the application.

I would appreciate the help. Thanks in advanced.