CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   Main CFD Forum (http://www.cfd-online.com/Forums/main/)
-   -   Turbulence model for mixing problem??? (http://www.cfd-online.com/Forums/main/79808-turbulence-model-mixing-problem.html)

nileshjrane September 4, 2010 20:09

Turbulence model for mixing problem???
 
Hello all,
I am trying to simulate air/fuel mixing in a combustion chamber. I need to specify a turbulence model (RANS) for this. My questions are

1) Which model would be better suited for such kind of simulation?? RNG-k-eps or SST-kW or any other??

2) About near wall treatment: I suppose k-w does not need a wall function as it can handle near wall flow on its own, am i right??

3) For RNG k-eps i will have to give some near wall function. Now my dialemma is for coarse mesh (y+ > 30) i have to specify wall function if i use standard k-eps model (which is High Re model in OpenFoam). But what if i use Low Re model like Launder-Sharma?? What about RNG (is it low Re or high Re)??

4) For mixing problem like this, how much importance the "near wall flow" have??? I am currently doing only cold flow simulations.

Thank you.

nileshjrane September 7, 2010 17:44

Help...anyone???:(

artemiss1984 September 9, 2010 15:46

Dear friend,
1. For CFD simulation you should use different models to obtain the best solution.we have not general rule for all of the problems.

2. for near wall treatment you can obtain the details with no slip boundary condition rather than wall function but you should use finer grids.y+<5

3.for k-w model your statement is true.it can integrate from sublayer to outer region and needs no matching.

4.I think for your problem the RSM model is better than other models.However you should check this.The best option is LES which is very time consuming.

pepi33 September 10, 2010 03:05

Hi,

For me, use a k-w SST with 1<y+<3. Indeed k-w doesn't need a boundary layer, but a very thin mesh (so expensive).
If your have experimental result try also other turbulence model (RSM, k-eps realizable enhanced wall). Sometimes it can be better for a compromise time of calcul/ accuracy.

If your model is quite turbulent do not hesitate to use my first advise (k-w SST). Concerning RSM it is true that i didn't use it a lot of time... so i can't say it is better than the k-w one.

pepi

sidspark12 September 10, 2010 05:45

hello ........i think you should go for k-omega SST MODEL........because when u do the analysis, thermal boundary layer will b generated there......that means near wall region is crucial for your analysis........so i think you should go for k-omega SST........

may b this ll work for u frend........:rolleyes:

nileshjrane September 10, 2010 11:43

Hello all,

Thank you for your valuable opinions.

My problem is i dont want to use fine meshes. So i am thinking of using wall functions so i can use coarse mesh at wall.

Also my flowfield is mostly vortex dominated, which means free shear flow. If i am correct k-eps is better for such flow, RNGkeps even so. Anyway i am not much concerned about near wall velocity. So i guess i can leave k-w model.

Now, can anyone tell me how to specify use of wall function in OF 1.7.0??

Thank you all.:)

harishg September 13, 2010 17:56

Before you use wall functions, you need to have a clear idea of their restrictions. Are there any regions of separation or adverse pressure gradients. Older K-epsilon models are known to perform poorly in those regions. If you feel that the wall function will work for your case, I will suggest you to use the k-omega SST model of Menter that is available in OpenFOAM.

nileshjrane September 14, 2010 04:57

Quote:

Originally Posted by harishg (Post 275046)
Before you use wall functions, you need to have a clear idea of their restrictions. Are there any regions of separation or adverse pressure gradients. Older K-epsilon models are known to perform poorly in those regions. If you feel that the wall function will work for your case, I will suggest you to use the k-omega SST model of Menter that is available in OpenFOAM.


Its mixing of air and fuel jets in combustion chamber. So flow separation and adverse P grads. My point is mixing problem like this are not much dependent on the BL on the walls (at least i think so, correct me if i m wrong). The bottomline is i dont want to use y+ ~ 1 mesh, i dont have cluster to solve such big problem. So wall function is a must thing for me. My focus is on jet mixing and if i am not wrong, k-eps model is better for this, and with wall function this should give good enough results. does it seem ok?

Thank you..
Nilesh


All times are GMT -4. The time now is 01:25.