[wc34071209]

Hi all,

For omegaWallFunction, OpenFOAM employs the strategy as follows

omega = sqrt(omega_vis^2 + omega_log^2)

which seems to be able to automatically take care of first layer thickness no matter it falls in viscous sublayer (y+ < 5) or log-law region ( y+ > 30 ) or even in the buffer layer ( 5 < y+ < 30 ).

However, for epsilonWallFunction, OpenFOAM seems to apply the following formula:

epsilon = Cmu^0.75 * k^ 1.5 / (kappa *y);

which is reported to be valid only when the first layer thickness falls into log-law region.

So my question is what will happen if I use epsilonWallFunction (k-epsilon model) with a grid where y+ < 5.

[gregjunqua]

You answered by yourself, you will use an invalid boundary condition since some wall function are only valid for the logarithmic layer position and when y+<5 you are in the linear sublayer.


For low Reynolds model I personally use for the wall condition :

k fixed value 10e-12
omega fixed value 10e9
nut nutUSpaldingWallFunction
U fixed value 0

If you need more information i suggest you to read : On the wall Boundary condition for turbulence models
from Jonas Bredberg

You would find many answer in it in all your questions about wall condition

[wc34071209]

Originally Posted by gregjunqua

You answered by yourself, you will use an invalid boundary condition since some wall function are only valid for the logarithmic layer position and when y+<5 you are in the linear sublayer.


For low Reynolds model I personally use for the wall condition :

k fixed value 10e-12
omega fixed value 10e9
nut nutUSpaldingWallFunction
U fixed value 0

If you need more information i suggest you to read : On the wall Boundary condition for turbulence models
from Jonas Bredberg

You would find many answer in it in all your questions about wall condition

Thanks. Actually I am also wondering why kqRWallFunction treats k just as zeroGradient. Apparently at the wall k should be zero.

By specifying k to a very small number and omega a very big number, I think you have to use very fine grid inside the boundary layer, e.g. y+ ~ 1 or even smaller.

nutUSpaldingWallFunction, as far as I know, is kind of continuous wall function which can work for both y+ < 5 and > 30. However, nutLowReWallFunction sounds, by its name, to be more suitable if you resolve the boundary layer.

Could I conclude as follows:

For y+ > 30,

k: kqRWallFunction
omega: omegaWallFunction or epsilon: epsilonWallFunction
nut: nutkWallFunction


For 1 < y+ < 5,

k: kqRWallFunction
omega: omegaWallFunction (k-epsilon is not suitable)
nut: not so sure


For y+ < 1,

k: very small number
omega: very big number (k-epsilon is not suitable)
nut: nutLowReWallFunction

For non-uniform y+ (between 1 and 100), however, is it better to use the following?

k: kqRWallFunction
omega: omegaWallFunction
nut: nutUSpaldingWallFunction

[kaaja]

For non-uniform y+ (between 1 and 100), however, is it better to use the following?

k: kqRWallFunction
omega: omegaWallFunction
nut: nutUSpaldingWallFunction

Does anybody know if the above is correct?