Hello all,

I am using Standard Wall Function with k-epsilon model. I understand that wall y+ value should be 30-300 to get a good results. I tend to get y+ around 1-3 which is veru small even after coarsen the mesh. If the valus is less than 30, is my results wrong?

To get 30-300, I really have to make the mesh REALLY coarse and I am not sure as to the appropriateness of this.

Hope someone can clarify.

Thanks. Sham.

Check the overall Re number and take the appropriate decision of modelling turbulence or not?

Do not change the mesh, use Enhanced Wall Treatment instead of the standard wall functions (this is valid for y+ around 1). Or change the turbulence model (standard or SST k-omega with "transitional flow" option activated).

All the best,

Razvan

If I use EWT, will my results be accurate. I read that SWF is well-established and has been used widely.

Well, that depends on what you are trying to model. Describe your problem in detail and I should be able to suggest you an appropriate setup.

Razvan

Razvan,

I am modelling a cylinder in free stream with structured mesh. Re=10000. I am interested in Cd, Cl and also vortex shedding. I am using RKE model. Hope you can give some suggestions here.

What FLUENT version are you using?

Razvan

Hey guys,

I am trying to create a model around a car geometry provided in the GAmbit tutorial guide to model the turbulent flow past the simplified sedan. I am thinking of using the standard k-epsilon model for this purpose. Howeevr I do not know hot to create a proper mesh around the car. CAn someone help me on how to create a good mesh. I am using FLUENT 6.3.26

Thanks Regards, Arjun

Fluent 6.3.

OK. If you are interested in accuracy, then you MUST use LES. Vortex shedding behind the cylinder is a very difficult case to simulate correctly, and only LES models can give you good results for Re=10000. The LES models I would recommend you are the dynamic Smagorinsky model or the turbulent-kinetic-energy transport model. Of course, all this has to be done in 3D.

If you are not familiar to LES, then a second good approach would be the low-Re omega-based RSM. I personally tested it on a cylinder configuration, and the results were qualitatively good, actually quite similar to LES. Quantitatively, I cannot say, because I did not run the simulation for long enough to be able to apply a time-averaging. But this was also done in 3D, unsteady pressure-based solver.

The simplest way to try this is 2D, with a SST-kw or low-Re omega-based RSM turbulence model, of course using unsteady solver. But I cannot guarantee you that the results will be good. The 3-dimensional effects cannot be neglected. You will obtain vortex shedding, no doubt about that, but Cd and Cl might not be correctly calculated.

Razvan

Razvan,

Thanks very much for yourt reply. At the moment though I'm simulating combustion with RANS as a part of research so I don't really have a choice with LES. I am moving onto LES towards the end of my PhD though.

The results are good escept for when I inject the fuel from the wall in question. The fuel seems to stick to it and it causes inaccuracy.

For second order which will give better results in my unstructured tet mesh (fuel injection) segment, when I put it onto EWT it crashes every time.

When injecting from an area far back inside the swirler section, the distribution of fuel in the combustor is largely accurate, so EWT isn't required. It's just this damn wall injection model.

PS - I was thinking of trying to get a much much smoother transition between the dense and coarse areas to see if it helps. IF ANYONE KNOWS IF THES OR A MORE 'STEPWISE' SOLUTION PROCEDURE IS WORTH PURSUING THEN PLEASE LET ME KNOW.

Thanks again Razvan! Phil

Sorry for some reason I thought this was a different post. Pleases igonre my message.