CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   OpenFOAM Running, Solving & CFD (http://www.cfd-online.com/Forums/openfoam-solving/)
-   -   Second derivative (http://www.cfd-online.com/Forums/openfoam-solving/59317-second-derivative.html)

braennstroem November 9, 2007 05:58

Hi, I have a (hopefully not
 
Hi,
I have a (hopefully not to stupid) question concerning the
calculation of the second derivatives in OpenFOAM. Right now, I know of
two ways to compute them using FVM; one is to calculate the first
derivatives at the cell faces and other would use the cell center. One
way to do it in OpenFOAM looks like this:

fvc::laplacian(U)

but how can I know, if the first derivative is calculated at the faces or centers?
Maybe, anybody has any recommandation, which approach to use?

Regards!
Fabian

braennstroem November 21, 2007 13:15

Hi, Hrvoje's answer got lost:
 
Hi,
Hrvoje's answer got lost:

Posted by Hrvoje Jasak on Sunday, November 18, 2007 - 05:43 am:

fvc::div(fvc::grad(U)) calculates the cell centre gradient,
interpolates to the face and calculates the second gradient (well,
divergence) from that.

fvc::laplacian(U) calculates the gradient on the face and does the div
without interpolation.

The difference between the two is the size of computational molecule -
have a look at the difference if you want to see it. I would say it is
very iportant to be CONSISTENT with whatever you are trying to do.

Enjoy,

Hrv

....................

Thanks for your help! I still have a small understanding problem about your being 'CONSISTENT' recommandation...

I would like to run SST-SAS simulations for indoor airflows, which switches between RANS and LES calculations according to the van Karman length scale, which in turn is based on the second velocity gradients. I assume that the first approximation needs, due to the larger molecule, more traffic on parallel setup and could make problems on the boundary!? In addition the accurary would be even worse on a unstructured grid due to the 'double' skewness of the two next cells. Or is it more a problem of convergence?

Greetings!
Fabian


All times are GMT -4. The time now is 12:10.