divScheme div(phi,U) Gauss linear behaves unexpected

docinger · August 16, 2013, 09:18

Hi everybody!
After reading in this awesome forum for months now, i'd like to actively post my first question today

I'm using openfoam / pimpleDyMFoam to solve incompressible flow around NACA airfoils (there'll be more details about my project in the future ...)

And i found a unexpected behaviour in the flowfields when i switch divSchemes in fvSchemes for div(phi,U) from Gauss linearUpwind grad(U) to Gauss linear

divUGausslinear.jpg

In the attached image you can see the completely "chopped" velocity field.
I use a 2D hexa mesh.

Can somebody explain to me why this happens when i use the Gauss linear scheme?

(And also can somebody tell me what the "grad(XXX)" behind the schemes is for? I seem to be incapable of finding the answer myself)

Thanks and welcome to myself,
Andreas

yhy20081016 · October 19, 2013, 03:56

"linear" scheme is unbounded, therefore it is usually unstable, unless the local Peclet number is less than 2. You can refer to the section titled "Divergence schemes" in "UserGuide.pdf". I suggest that you try "linearUpwind" scheme.

docinger · October 19, 2013, 05:15

yes, linearUpwind was what i started with and what I ended up using. The simulations run smooth by now and provide plausible results.

With the Peclet number dependency you mentioned, that would mean that my mesh simply is too coarse to run with linear / central differences, right?

Thanks,
A

August 16, 2013, 09:18	divScheme div(phi,U) Gauss linear behaves unexpected	#1
docinger New Member Andreas Join Date: Aug 2013 Posts: 8 Rep Power: 12	Hi everybody! After reading in this awesome forum for months now, i'd like to actively post my first question today I'm using openfoam / pimpleDyMFoam to solve incompressible flow around NACA airfoils (there'll be more details about my project in the future ...) And i found a unexpected behaviour in the flowfields when i switch divSchemes in fvSchemes for div(phi,U) from Gauss linearUpwind grad(U) to Gauss linear divUGausslinear.jpg In the attached image you can see the completely "chopped" velocity field. I use a 2D hexa mesh. Can somebody explain to me why this happens when i use the Gauss linear scheme? (And also can somebody tell me what the "grad(XXX)" behind the schemes is for? I seem to be incapable of finding the answer myself) Thanks and welcome to myself, Andreas Last edited by docinger; August 17, 2013 at 15:44.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
A turbulent test case for rhoCentralFoam	immortality	OpenFOAM Running, Solving & CFD	13	April 20, 2014 06:32
error on turbulence scheme	immortality	OpenFOAM Pre-Processing	11	August 21, 2013 08:45
no solution in second order schemes	immortality	OpenFOAM Running, Solving & CFD	14	May 23, 2013 01:39
how to modify fvScheme to converge?	immortality	OpenFOAM Running, Solving & CFD	15	January 16, 2013 13:06
solution diverges when linear upwind interpolation scheme is used	subash	OpenFOAM	0	May 29, 2010 01:23

October 19, 2013, 03:56		#2
yhy20081016 Member yehanyu Join Date: Mar 2012 Location: Beijing, China Posts: 48 Rep Power: 14	"linear" scheme is unbounded, therefore it is usually unstable, unless the local Peclet number is less than 2. You can refer to the section titled "Divergence schemes" in "UserGuide.pdf". I suggest that you try "linearUpwind" scheme.

October 19, 2013, 05:15		#3
docinger New Member Andreas Join Date: Aug 2013 Posts: 8 Rep Power: 12	yes, linearUpwind was what i started with and what I ended up using. The simulations run smooth by now and provide plausible results. With the Peclet number dependency you mentioned, that would mean that my mesh simply is too coarse to run with linear / central differences, right? Thanks, A