[sadsid]

Originally Posted by kerim

Hi,

I simulated natural convection in a square cavity using OpenFOAM7, installed on ubuntu 18.04LTS. The numerical simulation results were compared against experimental data of Ampofo F. and Karayiannis T.,2003.
The correspondence of the vertical velocity profile at the mid-height of the cavity with the experiment is good. However, the average temperature profile is under-predicted than the corresponding experimental data. When calculating using different versions of the OpenFOAM package(OF5, OF6, OF8), the average temperature profiles are small compared to the experiment. I will be happy if someone can help to solve this issue.
Kerim

Share case setup if possible. It's difficult to suggest anything on this information.

[kerim]

Originally Posted by sadsid

Share case setup if possible. It's difficult to suggest anything on this information.

Dear Sadsid,


I used upwind scheme. Maybe that is a reason of under prediction of T.
Kerim

[sadsid]

Originally Posted by kerim

Dear Sadsid,


I used upwind scheme. Maybe that is a reason of under prediction of T.
Kerim

Pl. share the "constant" folder as well. I want to test the case by running it.

[sadsid]

Originally Posted by kerim

Dear Sadsid,


I used upwind scheme. Maybe that is a reason of under prediction of T.
Kerim

No need to share constant folder. If possible share the data with which you are trying to get a comparision.

[kerim]

Originally Posted by sadsid

No need to share constant folder. If possible share the data with which you are trying to get a comparision.

Vertical velocity profile at mid-height in the cavity was digitized from this article:
https://www.researchgate.net/publica..._square_cavity

[kerim]

Originally Posted by sadsid

No need to share constant folder. If possible share the data with which you are trying to get a comparision.

The vertical velocity profile was attached.

[sadsid]

Originally Posted by kerim

The vertical velocity profile was attached.

My observations are:
1) your mesh coarser and it can be a reason for the wrong profile. Use stretching normal to the surface (I would prefer on the left and right walls as temperature gradients are expected to be steeper at these surfaces).
2) your solution is not statistically converged and ultimately giving wrong results.
3) Instead of 2D, you can test 3D (a more appropriate way to handle turbulent flow). The results you are trying to retrieve are for a cavity with dimensions .75x.75x1.5. So try 3D.

Good Luck!

[sadsid]

Originally Posted by kerim

The vertical velocity profile was attached.

A better result!

[kerim]

Originally Posted by sadsid

A better result!

Thank you FYI. Would you like to give more info about it?
I struggling with two equations models but have small success with LLR and SGG ones.

[sadsid]

Originally Posted by kerim

Thank you FYI. Would you like to give more info about it?
I struggling with two equations models but have small success with LLR and SGG ones.

I formulated 3D-LES model and used 90x80x180 number of cells. I think, a little more dense mesh will give me more closer results. Your boundary conditions are correct. I think you should increase the number of cells.

[kerim]

Originally Posted by sadsid

I formulated 3D-LES model and used 90x80x180 number of cells. I think, a little more dense mesh will give me more closer results. Your boundary conditions are correct. I think you should increase the number of cells.

At first, I thought about LES modelling. However, as a first step, the RANS model turned out to be easier for me. As a first approximation, the simulation was for 2D. Increasing the number of nodes up to 500 in each direction RANS does not give an acceptable result. Both models for large Reynolds numbers and models for low numbers were used. The numerical results in both cases are approximately equally underpredicted.


PS. I came to this problem because of buoyantCavity tutorial:
Please see: OpenFOAM 7 buoyantCavity tutorial-ERCOFTAC test case 79
p { margin-bottom: 0.1in; line-height: 115% }a:link { so-language: zxx }

[sadsid]

Originally Posted by kerim

At first, I thought about LES modelling. However, as a first step, the RANS model turned out to be easier for me. As a first approximation, the simulation was for 2D. Increasing the number of nodes up to 500 in each direction RANS does not give an acceptable result. Both models for large Reynolds numbers and models for low numbers were used. The numerical results in both cases are approximately equally underpredicted.


PS. I came to this problem because of buoyantCavity tutorial:
Please see: OpenFOAM 7 buoyantCavity tutorial-ERCOFTAC test case 79
p { margin-bottom: 0.1in; line-height: 115% }a:link { so-language: zxx }

If you want I can share my mesh so that you can test your case on a mesh with stretching introduced on side walls. I think you are struggling because of mesh!

[kerim]

Originally Posted by sadsid

If you want I can share my mesh so that you can test your case on a mesh with stretching introduced on side walls. I think you are struggling because of mesh!

If you can share with your mesh that will be great for me.

[kerim]

Originally Posted by sadsid

My observations are:
1) your mesh coarser and it can be a reason for the wrong profile. Use stretching normal to the surface (I would prefer on the left and right walls as temperature gradients are expected to be steeper at these surfaces).
2) your solution is not statistically converged and ultimately giving wrong results.
3) Instead of 2D, you can test 3D (a more appropriate way to handle turbulent flow). The results you are trying to retrieve are for a cavity with dimensions .75x.75x1.5. So try 3D.

Good Luck!

Hi,


You were right about 3D. I used graded mesh in the OX and OY directions, and uniform mesh in the OZ direction. And got better kOmegaSST results. May be I should take more finer 3D mesh.


Going ahead, I faced with the next error during LES modelling:



--> FOAM FATAL ERROR:

[U[0 1 -2 0 0 0 0] ] == [-grad(p)[1 -2 -2 0 0 0 0] ]


I never used LES but I know the problem is concerned with dimensions.

Could you help me to resolve this issue?

[sadsid]

Originally Posted by kerim

Hi,


You were right about 3D. I used graded mesh in the OX and OY directions, and uniform mesh in the OZ direction. And got better kOmegaSST results. May be I should take more finer 3D mesh.


Going ahead, I faced with the next error during LES modelling:



--> FOAM FATAL ERROR:

[U[0 1 -2 0 0 0 0] ] == [-grad(p)[1 -2 -2 0 0 0 0] ]


I never used LES but I know the problem is concerned with dimensions.

Could you help me to resolve this issue?

Your dimensions for pressure are wrong. I think it should be: dimensions [0 2 -2 0 0 0 0]. I think you need to correct it in 0 folder. Regarding mesh, it is not possible to upload here due to its size. I can share over email. Let me how can I share!

[kerim]

Originally Posted by sadsid

Your dimensions for pressure are wrong. I think it should be: dimensions [0 2 -2 0 0 0 0]. I think you need to correct it in 0 folder. Regarding mesh, it is not possible to upload here due to its size. I can share over email. Let me how can I share!

Please see the attached picture for all changes I have made for LES modeling. I don't understand why should I change the dimension of the pressure in this case? I switched the turbulence model from kOmegaSST into LES and I have made minor appropriate changes in controlDict, fvScheme, fvSolution files.
Where I have made mistakes?
My email. is: abdikerimkurbanaliev@gmail.com

[kerim]

Originally Posted by kerim

Please see the attached picture for all changes I have made for LES modeling. I don't understand why should I change the dimension of the pressure in this case? I switched the turbulence model from kOmegaSST into LES and I have made minor appropriate changes in controlDict, fvScheme, fvSolution files.
Where I have made mistakes?
My email. is: abdikerimkurbanaliev@gmail.com

Sorry I forgot to attach a picture

[kerim]

Originally Posted by sadsid

Your dimensions for pressure are wrong. I think it should be: dimensions [0 2 -2 0 0 0 0]. I think you need to correct it in 0 folder. Regarding mesh, it is not possible to upload here due to its size. I can share over email. Let me how can I share!

I solved this issue.

[sadsid]

Originally Posted by kerim

I solved this issue.

Good Luck!

[gu1]

Originally Posted by kerim

At first, I thought about LES modelling. However, as a first step, the RANS model turned out to be easier for me. As a first approximation, the simulation was for 2D. Increasing the number of nodes up to 500 in each direction RANS does not give an acceptable result. Both models for large Reynolds numbers and models for low numbers were used. The numerical results in both cases are approximately equally underpredicted.


PS. I came to this problem because of buoyantCavity tutorial:
Please see: OpenFOAM 7 buoyantCavity tutorial-ERCOFTAC test case 79
p { margin-bottom: 0.1in; line-height: 115% }a:link { so-language: zxx }

''After a small correction of buoyantCavity and using buoyantKE model of turbulence I had an acceptable agreement with ERCOFTAC test case 79. Please see the attached pictures'' (from ur post).

Can u share this corrections?

Best.