bug in the AMI option for the solver chtMultiRegionFoam

[andprad74]

Dear all after different tests seems that the AMI interface for the solver chtMultiRegionNFoam have a bug or something like that. The problem appear in my cases and also for the CHT case of the baracfd tutorial. The CHT case run work well in single core, in multi core it crash, please see the below picture. The problem appear in the multicore solution, sometime the 2 cores solution run but the convergence is difficult, with the polyedra mesh of fluent or with baramcfdmesh the single core doesn't run or the converge is difficult.

For the buoyantSimpleNFoam solver i made only 1 test and the solution in single core and with 8 core was perfect. The mesh was polyedra made with barammesh.

The picture in attachment is referred to the error message in output fot the CHT case of the tutorial.

After that, i'm intersted to know if the radiation with the FvDOM algorith is planned for the chtMultiRegionNFoam solver and when.

Thanks in advance. By. Andrea.

[andprad74]

The bug seems limited to the 25.1 version because the 24.4 it's ok. Thanks by. Andrea

[kdg5424]

Hello @andprad74. The error doesn't occur in our PC.

I'm not sure why this problem happens.

Could you please re-try with this file?

https://drive.google.com/file/d/1apNFLINZQ5PGILVm6Xakzkgmuk94vn0u/view?usp=sharing

Best Regards

[andprad74]

Hello kdg5424, i tested your file with

• 2 core and it works
• 4 core doesn't work

Here in attachment 2 cases one is the CHT case with 4 cores and the other is my working case for introduce the fvDom radiation. About the prism_mesh case it work weel with1 core, with 2 cores work but the residuals and results are not good, with 4 cores doesn't work.

Please can you tell me when is planned to introduce the radiation model in the chtMultiRegionNFoam? Only the P1 as inthe past or also the fvDOM?

Thanks in advance. By Andrea

CHT.zip

prism_mesh.zip

[andprad74]

hello any news about that? Thanks A

[kdg5424]

Hello @andprad74. Sorry for the delay.

There were same repots about this bug. We're doing test but unfortunately, we even haven't find the reason why this problem happened. The commons with the repots, there is a "interface" like fluid-solid thermo interface or rotating interface. If we find the reason, I'll let you know. I'm very sorry for this bug.

Before we found the reason if this kinda problem happens, please do iteration 1,2 step with single core and then simulate the case with parallel cores.

I'm so sorry

Best Regards

Donggyu Kim

[Ma77h3w]

I'm doing a simple natural convection thermal model and have the same issue. It works on 1 core, but fails immediately when I tried 12 core. The result looks about correct on 1 core but the residuals remained quite high, I'm not sure if that's relevant. Reducing the Energy relaxation factor from 1 to a slightly lower number helped improve the residual a bit. I'm still a beginner really and it's been a long time since I studied it. Here's the last bit of the log if that helps:

Solving for fluid region AirInner DILUPBiCGStab: Solving for Ux, Initial residual = 0.999317, Final residual = 0.0115277, No Iterations 1 DILUPBiCGStab: Solving for Uy, Initial residual = 0.999499, Final residual = 0.012114, No Iterations 1 DILUPBiCGStab: Solving for Uz, Initial residual = 0.999592, Final residual = 0.0114677, No Iterations 1 GAMGPCG: Solving for p_rgh, Initial residual = 0.999678, Final residual = 0.0206356, No Iterations 1 AMI: Patch source faces: 89792 AMI: Patch target faces: 89792 AMI: distributed

job aborted: [ranks] message [0-2] terminated [3] process exited without calling finalize [4-8] terminated [9] process exited without calling finalize ---- error analysis ----- [3,9] on XXXXXXX C:\Users\XXXXXXX\AppData\Local\Programs\BARAM\solvers\openfoam\bin\chtMultiRegionSimpleNFoam ended prematurely and may have crashed. exit code 3 ---- error analysis -----

I tried 2 core, and it got further. Looks like it solved for the fluid region, but failed in the same way as above for the Solid region.

[kdg5424]

Hi @Ma77h3w. It will be solved in BARAM v25.2 And the newest version will be released in this month. Could you try after the new version is released?

Best Regards

Donggyu Kim

[andprad74]

Hello i have made a test with the new release 25.2 in comparison to the 24.4. The test is the shell an tube heat exchenger, case of the tutorial, with the 25.2 it possible to start with 2 cores and after update the solver to 4 cores after around 100 iteration. The solution with 4 core from the 0 iteration will stop. With the 24.4 not problem to start with 4 cores. The impression is that the 25.2 is better than the 25.1 but not as the 24.4. Thanks for the great job andrea

[kdg5424]

Hello @andprad74. Sorry for the late reply. My github account was blocked for unknow reason.

Could you try again with BARAM the new mesh which we modify? The Mesh quality was so bad. And we fixed it. If you try the new mesh, you will be able to see that there is no problem.

Best Regards

Donggyu Kim

[andprad74]

Hello Donggyu, i try the new mesh and i can confirm the improvent, the parallel solution with 8 cores was good. Thanks to you and to the all BaramCFD team. Best Regards. Andrea.

[kdg5424]

@andprad74 Thank you!!

I'll close the issue.

Best Regards

Donggyu Kim