[yfjma]

Hi, All
I'm a new user of OpenFoam. I just found her several days ago and was interested in its open and robust attributes. After plays with it, i think it has a bright promising if it's help documentation can be strengthened.
Now, comes my question:
I run the LES case:channel395 and get several results written as files in different directory,e.g. in directory 1000, there are files of P,U,u_0,k,.....etc.
My exact question is what's the meaning of U_0,Umean,R,Bmean.
Does Umean means the average U with time??

Can anybody help me??

[grtabor]

Umean is a running average of the velocity field and R is a running average of the 2nd moments (velocity variance av(u'u'). Bmean is a running average of the SGS stress tensor. By running average I mean a continuously-updated time average of the quantity.

Most of this is in createAverages.H, which is fairly easy to read once you get used to the syntax. I actually can't find where U_0 is being defined : I'm wondering if its redundant. I'll keep looking.

Gavin

[eugene]

U_0 and the other *_0 fields are the old time values (i.e. timestep previous to U). They are necessary to restart calculations that use second order accurate schemes in time like Crank-Nicholson.

For the rest,"p" for a incompressible code is pressure devided by density. For a compressible code it is just the pressure (same for mean quantities). "pPrime2Mean" is the mean of the square of the deviatoric part (p-pMean) of the pressure and is a good indicator for noise. "nuSgs "is the subgrid-scale viscosity and "k" is the SGS turbulent kinetic energy. "nuTilda" is a working SGS viscosity that is only used with the Spalart-allmaras model.

Have I missed anything?

[santos]

Hello

Does anyone know what is the definition of the 9 components of Reynolds stress R in OpenFOAM?

Thanks!

[kumar2]

Hi all,

I was looking at what the components of R is and how they are calculated in channelOodles.

In createAverages.H, R is created as a symmetric tensor. so the 6 components of R written out by the program are Rxx,Rxy,Rxz,Ryy,Ryz and Rzz. As Dr. Tabor remarks above , R represents the variance of velocity thus the components of R should ; Rxx=ave(u'u') , Rxy=ave(u'v') , Rxz=ave(u',w'), Ryy=ave(v'v') , Ryz=ave(v'w') and Rzz=ave(w'w'). so the dimensions of R should be [0 2 -2 0 0 0 0] .

But if you look at the output file of R generated by the solver (say channelOodles.C) , then its dimension is [ 0 1 -1 0 0 0 0]. Is this correct? Also in createAverages.H, R is calculated as sqr(U)-sqr(UmeanPtr()) . thus R seems to be a function of velocity squared and should have dimension of m^2/s^2.

and can someone tell me how to get u'u' from sqr(U) - sqr(Umean) . It seems to me this should be square(U - Umean)

Thanks a lot

kumar

[hadi]

Hi Kumar,

About your last question i can tell you the following (PLEASE SOMEONE CORRECT ME IF I AM WRONG):

-R is a running average of u'u'
so R=<u'u'>=<(u-umean)(u-umean)>
=<(u-umean)^2>
=<u^2>
=<u^2> - 2<u.umean> + <umean^2>
=<u^2> -2<u>.Umean + Umean^2
=<u^2> -2.Umean^2 + Umean^2
=<u^2> - Umean^2.

Cheers,
Hadi

[hadi]

Hi Kumar,

About your last question i can tell you the following (PLEASE SOMEONE CORRECT ME IF I AM WRONG):

-R is a running average of u'u'
so R=<u'u'>=<(u-umean)(u-umean)>
=<(u-umean)^2>
=<u^2-2u.umean+umean^2>
=<u^2> - 2<u.umean> + <umean^2>
=<u^2> -2<u>.Umean + Umean^2
=<u^2> -2.Umean^2 + Umean^2
=<u^2> - Umean^2.

Cheers,
Hadi

[panda60]

Very good!
Thanks for engene, kumar and hadi's explanation.
hadi's explanation is especially excellent.
This thread shouldn't be forgot.

[arunsmec]

Originally Posted by eugene

U_0 and the other *_0 fields are the old time values (i.e. timestep previous to U). They are necessary to restart calculations that use second order accurate schemes in time like Crank-Nicholson.

If it is necessary only for restarting calculations, why we have to store this data? I am doing LES and is experiencing difficulty with disk space. If I dont save this *_0 fields, I can save about 10 GB space. Would you recommend removing these fields?