View Issue Details

IDProjectCategoryView StatusLast Update
0000134OpenFOAMBugpublic2012-11-05 14:29
Reporterbgschaid Assigned To 
PrioritynormalSeveritymajorReproducibilityhave not tried
Status closedResolutionno change required 
Summary0000134: Unphysical velocity patterns at the interface of a porous zone
DescriptionAs described in http://www.cfd-online.com/Forums/openfoam/82729-poroussimplefoam-oscillating-velocity-porous-zone.html for low permeabilities there is an oscillation at the interface of a porous zone. Investigations show that after the pressure correction phi is correct but the ripples are introduced by U -= rUA*fvc::grad(p) (equivalent formulation for the incompressible solvers) possibly by the differences in the matrix. This can be avoided by doing the complete reconstruction U = fvc::reconstruct(phi) (divide by rho for compressible solvers)
TagsNo tags attached.

Activities

henry

2011-02-10 12:50

manager   ~0000238

Using

U = fvc::reconstruct(phi);

causes a HUGE momentum conservation error but if momentum conservation is not important in your cases then yes it does remove spurious "wiggles" of many kinds.

bgschaid

2011-02-10 17:35

reporter   ~0000239

Point taken. Should have checked that too (pressure drop and velocity profiles were alright)

Anyway: for models that use U these wiggles produce "interesting" results. For instant when taking the angledDuctImplicit/rhoPorousSimpleFoam-tutorial and increasing the Darcy coefficient by two orders of magnitude one sees that each porous interface "produces" turbulence. Probably because the kEpsilon-model takes the gradient of the wiggle. Other stuff (lagrangian) might be affected, too

So it is not a purly cosmetic/postprocessing problem

Issue History

Date Modified Username Field Change
2011-02-10 12:42 bgschaid New Issue
2011-02-10 12:50 henry Note Added: 0000238
2011-02-10 17:35 bgschaid Note Added: 0000239
2011-03-17 15:47 user2 Status new => closed
2011-03-17 15:47 user2 Resolution open => no change required