CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   OpenFOAM Running, Solving & CFD (http://www.cfd-online.com/Forums/openfoam-solving/)
-   -   Oscillating velocity and porous zones (http://www.cfd-online.com/Forums/openfoam-solving/58636-oscillating-velocity-porous-zones.html)

 alberto August 16, 2008 20:54

I noticed that in simulations

I noticed that in simulations with porous zones, the velocity oscillates at the exit of the porous zone. Moreover something unexpected happens also close to the inlet, as shown in the picture.

http://www.cfd-online.com/OpenFOAM_D...ges/1/8749.png

The system is a 2D channel, 1m long (diameter 0.07), with a porous zone in the first 0.5m (porosity = 0.5, d = 10^7, f = 0). The inlet velocity is 1m/s. All solver settings are taken from the rhoSimplePorousFoam implicit tutorial case.

The result shown here is obtained with a coarse grid. The behaviour improves significantly with a very grid with cells of some millimeter, but some issue is still present at the end of the porous zone, as shown below (cell 0.001 x 0.001).

http://www.cfd-online.com/OpenFOAM_D...ges/1/8750.png

Is this a known issue or are there particular requirements for the mesh around the porous zone?

With kind regards,
Alberto

 alberto August 16, 2008 20:59

Sorry for the typo in "The

Sorry for the typo in

"The behaviour improves significantly with a very grid with cells..."

which should be:

"The behaviour improves significantly with a very fine grid with cells..."

 dbxmcf August 17, 2008 16:16

Hi, here is just a guess, when

Hi, here is just a guess, when the density changes significantly for the inlet and outlet, a shock wave case probably occurs, to overcome this, some type of artificial viscosity should be added to the momentum equation, see Von Neumann and Richtmyer (1950) and Jameson et al

 alberto August 18, 2008 15:25

Thanks Roy, I agree because

Thanks Roy,

I agree because the porous zone introduces a discontinuity in the resistance at the borders of the porous zone. However I think the right way to solve this is to work on the interpolation schemes/solution algorithms to manage the shock.

What gives me something to think about is the too high sensitivity to the mesh refinement. Even on a very refined mesh, the velocity profile changes significantly.

For example:

Grid with squared cells of 3mm:

http://www.cfd-online.com/OpenFOAM_D...ges/1/8771.png

Grid with squared cells of 1mm:

http://www.cfd-online.com/OpenFOAM_D...ges/1/8772.png

Regards,
Alberto

 All times are GMT -4. The time now is 16:27.