From what I've read, wall functions based around the log law are not compatible for WMLES of bluff bodies.
In my particular case, I am attempting to perform a WMLES with heat transfer, around a square cylinder, involving a stagnation point at the front face and flow separation at the corners: particularly, this case here.
https://hal.archives-ouvertes.fr/hal-01981496/document
As is seen in the article:
"Actually, this bluff-body flow does not verify the wall-function assumption anywhere around the cylinder: the impinging flow with a stagnation point on the front face, the flow separation on the side faces and the backflow on the back face that occur here are not compatible with the constant pressure gradient and steady attached flow assumptionof the wall-function model...
The wall heat flux, however, is directly computed by the wall-function model that fails to mimic the turbulent heat transfer in the wall boundary layerwhen the model assumptions are not verified"
Just like in the case of this article, my Nusselt number profile is globally underpredicted (by about half), despite perfectly fine results in terms of pressure, drag, and Strouhal number. I am using the standard Kader Blend. A few subsequent papers (such as this one: https://pdfs.semanticscholar.org/e7e5/e9277f7a57856b4e96592893ca51543a65cd.pdf) repeated this test case with a PANS approach, and ended up with the same problem.
Would anyone have any good resources or good literature suggestions on how one could better model the thermal boundary layer for bluff bodies, particularly involving stagnation points and flow separation? I've scoured the internet quite a lot and I've only found this one:http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.720.1326&rep=rep1&type=pdf
Here it's suggested that y* is used instead of y+, but as I'm doing WMLES I don't have access to k (is this something that can be easily reached given say, a Smagorinsky model WMLES?). Does anyone have any suggestions?
Hi,
potentially I'm not yet correctly understanding of what you intend to do.
But if you spend the already quite high effort of a LES-like model, than from my perpsective the thermal boundary layers need to be resolved in order to be able to get a highly accurate answer for the heat transfer from the surface of the bluff body to the surrounding flow, and that would mean a mesh with y+
Hello Dr. Thomas Frank
Thank you for your response. The current discretization method I am using precludes the use of high aspect ratio cells, and as a result, getting to y+
Tha answer of Thomas Frank is already focusing on the issue.
Any type of wall-model BCs implies that you are somehow already prescribing the flow conditions develped at the wall, that is stress and heat flux. This way you cannot think to resolve correctly the problem also using LES. The only way to approach the problem to predict correctly the wall quantities is to resolve the boundary layers (dynamics and thermal) using a very refined grid, that is 3-4 node within y+
- Badges
- Science topic
- Similar topics
- Engineering
- Thermal Engineering
- Thermofluid
- Heat Transfer