[Giannis]

Hello Everyone!

I have a strange problem, concerning my CFD project. The project is associated with the flow of a liquid inside a tube whose shape is U (U pipe). At this, I want to examine some temperatures and generally the heat transfer data between the fluid and the pipe. However, my results seem to be far away from what i expected for.
At the following ''wall y+ scene'' you can notice some ''gaps'' at the interface between the fluid and the pipe wall.
I have used k-omega model, since my flow is transitional and I also know that wall y+ value should be less than 1 (y+<1). However, I do not know whether that gaps mean something or are created due to I do not have an appropriate Y+ value. Is it possible to have any other issues with the boundaries? (although i am pretty sure that they are correct).


Any advice would be really helpfull!
Thank you a lot in advance!

Giannis

[eRzBeNgEl]

Could u please give us more information about your mesh setup (prism layers, reference size,..,) and also the mesh report is very interesting

[Giannis]

Hello eRzBeNgEl and thank you for your reply!

I noticed that the main problem that I have concerns the quality of the mesh. I used Surface Remesher, Trimmer and Prism Layer properties/models.

I changed the ''Gap Fill Percentage'', ''Minimum Thickness Percentage'' and ''Layer Reduction Percentage'' values, and then I receive somehow better result but still not the excepted one. I also increased the number of prism layers, from 2 (by default value) to 6.
One more problem that I face concerns the computer's power since my geometry is huge (80x6x6 m) and sometimes mesh development is impossible to be implemented.
At this note I would like to ask you sth else if you know about. Is it possible to have any effect if I change the mesh model of the fluid from trimmer to polyhedral, but the pipe's mesh remain trimmer? Currently I have used trimmer model for both fluid and pipe.

Thank you again for your help.

Giannis

[flotus1]

Is it really necessary to do a conjugate heat transfer simulation here? In most "flow throug a pipe"-cases, it is not...
That would make the case much easier both in meshing and solving.

[Giannis]

Hello flotus1!

Actually, I investigate heat transfer phenomena between that geometry and the ground. I work on a Ground Source Heat Pumps problem and I consindered that would be a worth considering approach to investigate many things using CFD.

So the problem is associated with heat transfer, from the fluid to the pipe (convection heat transfer), from the pipe to the (grout - part of Ground Heat Exchangers) and then, from the grout to the ground.

[abdul099]

When you really run your case with y+ <1, you need to significantly increase the number of prism layers. About 15 layers is the minimum, while it doesn't harm to have some more.
So I recommend to run with wall functions and 30 < y+ < 150.

With your huge domain, I would use the generalized cylinder mesher when possible or stick to the extruder for straight sections of your pipe. Or when you're using a new version (7.04), you can use directed meshing. It might be a little tricky to get a small, good mesh, but it's worth not just to hit the mesh button using a trimmed mesh.