[chandra shekhar pant]

Dear Foamers,


First of all sorry for this query, I know that many threads are there in this regard, but yet I am not able to find it explicitly.

I am simulating the flow over an aerofoil, in which I am using the k-\omega with wall functions for k at the wall (or aerofoil) as "kqRWallFunction" and for \omega "omegaWallFunction", the results looks fine when compared with the experimental data. Now, when through post processing I am calculating the value of y(plus), I am getting the values of y(plus); min as 1.34, max as 578.05 and average as 153.47.

Could please any one can explain are these simulations correct/incorrect etc? Should I use the wall function or not? Should I not use at all the wall functions with the k-\omega ?
Any help/suggestion or some reference is highly welcomed.

Thanks!

[Rasmusiwersen]

Originally Posted by chandra shekhar pant

Could please any one can explain are these simulations correct/incorrect etc? Should I use the wall function or not? Should I not use at all the wall functions with the k-\omega ?
Any help/suggestion or some reference is highly welcomed.

Thanks!

The value of the yplus refers to a non dimensional distance from the wall. The velocity profile varies in the boundary layer (the flow of which a wallfunction models) as a function of yplus. The reason that different thresholds are present for different wall functions, is that the viscous and logarithmic sublayers in the boundary layer are modelled best at certain intervals for each wallfunction (I've attached a figure which hopefully visualizes this). So for example for a nutKwallfunction, i believe the recommended threshold is 30<yplus<300. So in order to answer your question, some research on the "threshold" for the wallfunctions you use would probably help (i don't know them by heart unfortunately).

Whether you should use the wallfunction or not depends on the flow problem. Wallfunctions are only needed for turbulent flows as they are capable of solving both the turbulent disturbances as well as the laminar flow in the viscous sublayer. If you don't have any turbulence (hence very low Re), flow separation doesn't occur and there is no need to model turbulence anymore. Hence no need for wallfunctions.

Hope this sheds some light on your problem.

[Krao]

If you are using k-Omega model it is recommended to have y+ values under 1. In worst case, not more than 5. If it is k-Epsilon then optimum yPlus value would be between 30 to 100 (max 200). For k-Omega SST anywhere between 0.1 and 200 except (5-30 buffer zone). This is what I have learnt attending CFD lectures at my University.

If wrong please correct me.

K. Rao

[chandra shekhar pant]

Hello, Rasmus Iwersen and K RaoThanks for your help and wonderful explanations.


I am still wondering that what is the prime value to be concerned, min max or average, since my min max and average are: 1.34, 578 and 153 respectively. If I understood you correctly, then you both suggested the 20<yplus<200, excluding the buffer zone, for k-\omega SST.


So what should I emphasize min, max or average ? I hope I am clear in my question

[Rasmusiwersen]

Originally Posted by chandra shekhar pant

Hello, Rasmus Iwersen and K RaoThanks for your help and wonderful explanations.


I am still wondering that what is the prime value to be concerned, min max or average, since my min max and average are: 1.34, 578 and 153 respectively. If I understood you correctly, then you both suggested the 20<yplus<200, excluding the buffer zone, for k-\omega SST.


So what should I emphasize min, max or average ? I hope I am clear in my question

Hello again,

Usually i lean towards the average value. yPlus is calculated for each cell, meaning a single cell with a large value of yPlus does not mean you won't resolve the turbulence properly. The average value is more representable of the entire boundary layer.

I would imagine that if you have a structured boundary layer grid, the maximum, minimum and average values might align more compared to if the grid is unstructured.. This is nothing i know, just a quick thought

[Krao]

Take a look at you yPlus values in paraview, hope there you can figure out which part of the grid has high yPlus values. As Rasmus said, avg is what we should look into. Lets say if you have 1/3rd of the grid at yPlus 500 and on the rest at yPlus 1, the avg is 200. But it does not make any sense. So try to achieve non drastic values. Use yPlus prediction tools available online, it saves your time! Although the predictions are not 100% accurate, at least you will get closer values, which you can improve within the next simulation. https://www.cfd-online.com/Tools/yplus.php https://www.computationalfluiddynami...t-cell-height/http://www.pointwise.com/yplus/.

The third link have some explaination too!

K. Rao