View Issue Details

IDProjectCategoryView StatusLast Update
0001413OpenFOAMBugpublic2014-10-30 19:17
Reporteruser1007Assigned Tohenry  
PrioritynormalSeveritymajorReproducibilityalways
Status resolvedResolutionfixed 
PlatformServerOSRed HatOS Version6.3
Summary0001413: Effect of using driftFluxFoam with or without wallFunctions
DescriptionVery recently, issue 0001410 has been resolved in driftFluxFoam OpenFoam-2.3.x. Example simulations using driftFluxFoam were run using wallFunctions for nut, k and epsilon. Results of these simulations look promising.

If a similar simulation is run without wallFunctions - zeroGradient at walls - a clearly different result is obtained, which does not seem physically realistic near top and bottom.
See attached cases and screenshots of volume concentration solids over height.

My questions: is it unavoidable to use the wallFunctions in order to obtain physically realistic results, using the turbulence modeling in driftFluxFoam?
Or is there a shortcoming in the way the "zeroGradient"-BCs are treated within the current turbulence modeling?
Thanks in advance.
Tagsmultiphase, turbulence models

Activities

user1007

2014-10-14 14:32

 

alphasand_driftFluxFoam_3.png (89,676 bytes)   
alphasand_driftFluxFoam_3.png (89,676 bytes)   

user1007

2014-10-14 14:33

 

user1007

2014-10-14 14:35

 

hpipe_driftFluxFoam_3.zip (319,429 bytes)

user1007

2014-10-14 14:35

 

user1007

2014-10-14 14:36

  ~0003249

Note: without wallFunctions the concentration gradient is very steep, whereas with wallFunctions a widely distributed concentration over height is obtained.

henry

2014-10-14 14:40

manager   ~0003250

If you are using the high-Re k-epsilon model you will need to use wall-functions to obtain physical k and epsilon distributions near the wall. If you want to avoid wall-functions you will need to use a low-Re model and sufficient resolution near the wall.

It would be useful to check that the latest formulation provides physically reasonable and stable distributions for laminar cases.

user1007

2014-10-14 15:31

  ~0003251

Thank you for the quick reply Henry.

When taking U_axial = 0.01 m/s, the case should be laminar, Re~1100.
The results are attached, alpha.sand and mag(U) over height.
Also the velocity distribution doesn't look OK.
The problem still occurs, when using no wallFunctions?!

user1007

2014-10-14 15:32

 

user1007

2014-10-14 15:32

 

U_driftFluxFoam_laminar.png (165,163 bytes)   
U_driftFluxFoam_laminar.png (165,163 bytes)   

henry

2014-10-14 16:11

manager   ~0003252

Wall-functions are not appropriate for laminar flow, in fact you will not be able to apply them anyway.

At a Re of 1100 with the high gradients of viscosity it looks like the flow is unsteady, possibly physical, maybe numerical. You will either need to reduce the Re or choose more stable schemes for the momentum equation.

Issue History

Date Modified Username Field Change
2014-10-14 14:32 user1007 New Issue
2014-10-14 14:32 user1007 File Added: alphasand_driftFluxFoam_3.png
2014-10-14 14:33 user1007 File Added: alphasand_driftFluxFoam_wallFunctions_3.png
2014-10-14 14:33 user1007 Tag Attached: multiphase
2014-10-14 14:34 user1007 Tag Attached: turbulence models
2014-10-14 14:35 user1007 File Added: hpipe_driftFluxFoam_3.zip
2014-10-14 14:35 user1007 File Added: hpipe_driftFluxFoam_wallFunctions_3.zip
2014-10-14 14:36 user1007 Note Added: 0003249
2014-10-14 14:40 henry Note Added: 0003250
2014-10-14 15:31 user1007 Note Added: 0003251
2014-10-14 15:32 user1007 File Added: alphasand_driftFluxFoam_laminar.png
2014-10-14 15:32 user1007 File Added: U_driftFluxFoam_laminar.png
2014-10-14 16:11 henry Note Added: 0003252
2014-10-30 08:49 henry Status new => resolved
2014-10-30 08:49 henry Resolution open => fixed
2014-10-30 08:49 henry Assigned To => henry