search

karimimp / PUFoam

This repository contains PUFoam solver, an OpenFOAM solver that is tailored for polyurethane foams expansion by coupling CFD with population balance equation.
6 stars 5 forks source link

Parallel running issue-PUFoam #1

Open sahrishNaqvi opened 2 years ago

sahrishNaqvi commented 2 years ago

Hi Karimi, this is Sahrish Batool Naqvi, I am working with your open-source code PUFoam and have facing some issues regarding parallel running. Does it have the capability to run parallel? Please shed a light in to this issue.

karimimp commented 2 years ago

Hi, The solver was a proof of concept for the PU modeling and test cases that I tried were rather simple. So at that stage the need for the parallelization was not realized. However, I think it would be quite interesting to invest some time on that in case you might have time to work on the parallelization.

sahrishNaqvi commented 2 years ago

Dear Mohsen thank you so much for your quick response, I am trying to handle it. In case I am able to do it I will write you. Or I will post it on my wiki page and will share a link. Stay in touch, please. It is very nice to have a conversation with you. I hope if you don't mind I can write you whenever needed, right?  Sahrish Batool NaqviPhD, Fluid DynamicsPostdoctorate at UMinhoPortugal On Tuesday, 29 March 2022, 17:54:32 BST, Mohsen @.***> wrote:

Hi, The solver was a proof of concept for the PU modeling and test cases that I tried were rather simple. So at that stage the need for the parallelization was not realized. However, I think it would be quite interesting to invest some time on that in case you might have time to work on the parallelization.

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: @.***>

karimimp commented 2 years ago

Sounds like a really good plan and please do not hesitate to write to me, I'd be glad to help.

sahrishNaqvi commented 2 years ago

Hi Mohsen Yesterday, I tried to run a simple cavity case parallel using your PUFoam solver. It is working with some limitations; for example, I divided cavity case (2D)  into equal subdomains (4 rectangular strips using the "simple method" with Coeffs (4 1 1)). With this division, it works well. However, for other divisions (nonuniform), the simulation did not converge and got stuck at one point where we have an uneven division of the domains (I should say that foam formation stops progressing at that point, and I believe that this depends on the how you are solving the PBE). I am looking at this issue, and as soon as I configure this out, I will share it with you. If you have comments at this point, please share. I am exploring this solver in details step by step. Please stay in touch; I have a lot of things to discuss and share.  P.S. I am adding Prof. J. Miguel Nobrega (University of Minho, Department of Polymer Engineering, Guimaraes, Portugal) in cc to keep him up to date, and of course, he can participate whenever needed.  Note: I attached a pdf document pointing to the issue and debugging (for time being).  Sahrish

On Tuesday, 29 March 2022, 18:09:02 BST, Mohsen ***@***.***> wrote:

Sounds like a really good plan and please do not hesitate to write to me, I'd be glad to help.

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: @.***>

karimimp commented 2 years ago

Hi, That's pretty good test case to try. Have you also tried a simple cup test case to run it in parallel. I would say to look at the moments and check if the gelling point is reached in one or two cells in the domain Can you please email me (mohsenk@outlook.com) the pdf file, I do not see any attached doc here

sahrishNaqvi commented 2 years ago

Yes, I tried simple cup case (provide by you on GitHub together with the code) and parallel running is working perfectly fine in that case. 

Let me have a deep look into the moments and to follow the gelling point, and actually this is what not happening, that's why solver got diverge. Thanks for this suggestion, it will save my time in order to track the hidden needle.

I have attached that document again and included your alternative email. Please let me know if you find it, else I will copy paste it here.

BR

Sahrish

On Wednesday, 30 March 2022, 14:43:01 WEST, Mohsen ***@***.***> wrote:

Hi, That's pretty good test case to try. Have you also tried a simple cup test case to run it in parallel. I would say to look at the moments and check if the gelling point is reached in one or two cells in the domain Can you please email me @.***) the pdf file, I do not see any attached doc here

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: @.***>

sahrishNaqvi commented 1 year ago

Thank you so much. Sure I will keep disturbing you. Sahrish On Tuesday, 29 March 2022, 18:09:02 BST, Mohsen @.***> wrote:

Sounds like a really good plan and please do not hesitate to write to me, I'd be glad to help.

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: @.***>