[wc34071209]

Hi Foamers,

Looking at the source code, the difference between

alphatWallFunction

and

alphatJayatillekeWallFunction

is apparent.

For alphatWallFunction, alphat = mut/Pr, where Pr is the turbulent Prandtl number.

For alphatJayatillekeWallFunction, it is more involved, as it distinguishes linear law region and log law region.

What I am wondering is the scope of application and the pros and cons of these two wall functions.

Does appliation of these two wall functions for alphat depend on low-Re or high-Re turbulence model, high-resolution mesh or low-resolution mesh, or others?

Thank you.

[arvindpj]

alphatWallFunction - for compressible cases
alphatJayatillekeWallFunction - for incompressible cases

Cheers :-)
-Jay

[wc34071209]

Originally Posted by arvindpj

alphatWallFunction - for compressible cases
alphatJayatillekeWallFunction - for incompressible cases

Cheers :-)
-Jay

Hi Jay,

Thank you for your reply. But both can be applied to compressible cases, provided that you put 'compressible::' before them.

[arvindpj]

What would you use in "alphat" in case of a compressible low-re case ?? Is it

 
wall
{
            type             calculated;
             value           $internalField;
}

Thanks,
Jay

[wc34071209]

Originally Posted by arvindpj

What would you use in case of a compressible low-re case ?? Is it

 
wall
{
            type             calculated;
             value           $internalField;
}

Thanks,
Jay

Hi,

As far as I know, 'calculated' b.c. does not apply to mut/nut.

[arvindpj]

Originally Posted by wc34071209

Hi Jay,

Thank you for your reply. But both can be applied to compressible cases, provided that you put 'compressible::' before them.

But,

compressible::alphatWallFunction is strictly for compressible cases

[khedar]

alphatJayatillekeWallFunction performs much better than alphatWallFunction for various y+ meshes (I have tested on a channel flow for y+=10 to 50 with BSL-EARSM model) and would advise to use the same if using high y+ meshes.

[wc34071209]

Originally Posted by arvindpj

But,

compressible::alphatWallFunction is strictly for compressible cases

Hi,

Yes, if you use for incompressible, you use

alphatWallFunction

or

alphatJayatillekeWallFunction

and if you use for compressible, you use

compressible::alphatWallFunction

or

compressible::alphatJayatillekeWallFunction

I am talking about OpenFOAM 2.3 and I am not certain about OpenFOAM 3.0 and later versions.

[wc34071209]

Originally Posted by khedar

alphatJayatillekeWallFunction performs much better than alphatWallFunction for various y+ meshes (I have tested on a channel flow for y+=10 to 50 with BSL-EARSM model) and would advise to use the same if using high y+ meshes.

Hi,

Thank you for your reply. But much better performance, you mainly mean the profile of T at the near wall region or the convective heat flux or both?

[khedar]

By much better performance I mean the convective heat fluxes. I am not sure how you would get a profile at the near wall region since due to the wallfunction usage, there is only one value of temperature which in turn is calculated using some thermal universal profile by alphatJayatillekeWallFunction. So in other words WallFunction uses the profile to get the temperature at the near wall cell. Do correct me if I am wrong.

[wc34071209]

Originally Posted by khedar

By much better performance I mean the convective heat fluxes. I am not sure how you would get a profile at the near wall region since due to the wallfunction usage, there is only one value of temperature which in turn is calculated using some thermal universal profile by alphatJayatillekeWallFunction. So in other words WallFunction uses the profile to get the temperature at the near wall cell. Do correct me if I am wrong.

I mean how the value of temperature at the 1st, 2nd, 3rd, ... layer of the mesh agrees with DNS or experimental data.

Have you ever compared these two wall functions for alphat on a grid leading to y+ < 5?

[khedar]

I have not done cell wise check for the temperature values since it would not make any sense. First cell value with this wallfunction for say y+30 mesh cannot be same to that of y+1 mesh without any wall function.

I have compared the heat fluxes with a simulation on y+1 mesh in OpenFoam and CFX. I get a good match for these.

[wc34071209]

Originally Posted by khedar

I have not done cell wise check for the temperature values since it would not make any sense. First cell value with this wallfunction for say y+30 mesh cannot be same to that of y+1 mesh without any wall function.

I have compared the heat fluxes with a simulation on y+1 mesh in OpenFoam and CFX. I get a good match for these.

Hi,

I mean you compare the values of temperature at different wall distances (not at different y+) with DNS or experimental data.

So you mean you have obtained quite close results to CFX on a y+=1 mesh with alphatJayatillekeWallFunction? Did you simulate a compressible or incompressible case?

[khedar]

I simulated compressible case using buoyantSimpleFoam solver.

[calf.Z]

I am wondering whether alphatWallFunction is suitable for both high Re and low Re? Thank you.

[arvindpj]

Originally Posted by calf.Z

I am wondering whether alphatWallFunction is suitable for both high Re and low Re? Thank you.

For LowRe cases, I prefer using alphatJayatillekeWallFunction

[Gerhard]

Originally Posted by arvindpj

What would you use in "alphat" in case of a compressible low-re case ?? Is it

 
wall
{
            type             calculated;
             value           $internalField;
}

Thanks,
Jay

For low-Re cases, would a fixed value of zero not be better?
The turbulence viscosity, nut, is set equal to zero in the nutLowReWallFunction.
Now, for alphat, looking at alphatWallFunction, for example, the thermal diffusivity is calculated as
alphat = nut / Prt.
In other words, for low-Re application, if nut = 0 at the wall, I would think alphat = 0 should also be the case at the wall?


Regards,
Gerhard