very strange velocity profiles for kOmegaSST model (with or with out wall function)

atul1018 · February 26, 2022, 08:09

Hello Community

Let me explain what I want to do: I want to simulate particle-infiltration phenomena in Flume. I am using CFDEM coupling (OpenFOAM+Liggghts) for this purpose. Firstly, I perform pure DEM simulation where I fill large particle in a cavity section within the channel and then let the small particles gets injected with water from inlet of channel (here both CFD and DEM are active). As per experiment I am simulating, the bed remians stable (large particles in cavity doesn't get transported) and only small particles get transported and infiltrated in the bed (i.e cavity). The flow is turbulent therefore flow needs to be reresolved using turbulence modeling.

As flow is turbulent, I want the flow at the inlet of channel as fully developed turbulent flow and for the same purpose I am using mapped bc. By mapping flow fields from a cetrtain offset from inlet back to inlet I should get fully developed turbulent flow. In the third dimension (front and back plane) I am using symmetry bc. top and bottom channel is treated as no slip walls. The initial turbulence parameters are estimated as per guidelines provided here in CFDEM wiki by assuming turbulence intensity of 4%. The estimated YPlus is around 100-200. I tried both kEpsilon and KOmegaSST with and without wallfuncitions and found out only KEpsilon (without wall functions) produces velocity profiles similar to fully develped channel flow (peak at the centre of channel). When wall functions are used with KEpsilon the peak get shifted near to upper wall. Similar observations are also found in KOmegaSST (with or without wall functions) where is is observed either near to upperwall or lower wall but not at the center. I wonder why KOmegaSST performs in this stange way?

As I am having higher Yplus values I must use wall functions (in both case KEpsilon and KOmegaSST) but even with wallfunctions results are very strange and incorrect velocity profiles (peak near tro walls) is observed. What could be the reason for this?

Looking farward to hear from seniors.

Best Regards
Atul Jaiswal

JulioPieri · April 5, 2022, 09:10

Dear Atul,

Sorry for the delay. By now, you might already have solved your problem.
Have you tried changing your mesh resolution a bit? One possibility is that you are in the exact transition yPlus value so that your first cell keeps changing from viscous sublayer to outer layer.

Do a mesh sensibility study on a simpler case. Get the channel with same properties and try simulating your problem with different near-wall mesh resolution (thus yPlus). Check the different possibilities (enhanced wall treatment, different laws, etc), without randomly changing it to see if the results are as expected. Study the different wall treatment options and see which one suits your problem.

February 26, 2022, 08:09	very strange velocity profiles for kOmegaSST model (with or with out wall function)	#1
atul1018 Senior Member Join Date: Jun 2020 Posts: 100 Rep Power: 6	Hello Community Let me explain what I want to do: I want to simulate particle-infiltration phenomena in Flume. I am using CFDEM coupling (OpenFOAM+Liggghts) for this purpose. Firstly, I perform pure DEM simulation where I fill large particle in a cavity section within the channel and then let the small particles gets injected with water from inlet of channel (here both CFD and DEM are active). As per experiment I am simulating, the bed remians stable (large particles in cavity doesn't get transported) and only small particles get transported and infiltrated in the bed (i.e cavity). The flow is turbulent therefore flow needs to be reresolved using turbulence modeling. As flow is turbulent, I want the flow at the inlet of channel as fully developed turbulent flow and for the same purpose I am using mapped bc. By mapping flow fields from a cetrtain offset from inlet back to inlet I should get fully developed turbulent flow. In the third dimension (front and back plane) I am using symmetry bc. top and bottom channel is treated as no slip walls. The initial turbulence parameters are estimated as per guidelines provided here in CFDEM wiki by assuming turbulence intensity of 4%. The estimated YPlus is around 100-200. I tried both kEpsilon and KOmegaSST with and without wallfuncitions and found out only KEpsilon (without wall functions) produces velocity profiles similar to fully develped channel flow (peak at the centre of channel). When wall functions are used with KEpsilon the peak get shifted near to upper wall. Similar observations are also found in KOmegaSST (with or without wall functions) where is is observed either near to upperwall or lower wall but not at the center. I wonder why KOmegaSST performs in this stange way? As I am having higher Yplus values I must use wall functions (in both case KEpsilon and KOmegaSST) but even with wallfunctions results are very strange and incorrect velocity profiles (peak near tro walls) is observed. What could be the reason for this? Looking farward to hear from seniors. Best Regards Atul Jaiswal

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Ansys CFD-Post Application Error	rushiCFD	FLUENT	0	March 21, 2021 08:51
Table bounds warnings at: END OF TIME STEP	CFXer	CFX	4	July 17, 2020 00:44
Radiation in semi-transparent media with surface-to-surface model?	mpeppels	CFX	11	August 22, 2019 08:30
Question about adaptive timestepping	Guille1811	CFX	25	November 12, 2017 18:38
A girl fail to plot velocity profile when mesh changes + Wall function	asherah	STAR-CCM+	0	February 19, 2010 18:45

April 5, 2022, 09:10		#2
JulioPieri Senior Member Julio Pieri Join Date: Sep 2017 Posts: 109 Rep Power: 9	Dear Atul, Sorry for the delay. By now, you might already have solved your problem. Have you tried changing your mesh resolution a bit? One possibility is that you are in the exact transition yPlus value so that your first cell keeps changing from viscous sublayer to outer layer. Do a mesh sensibility study on a simpler case. Get the channel with same properties and try simulating your problem with different near-wall mesh resolution (thus yPlus). Check the different possibilities (enhanced wall treatment, different laws, etc), without randomly changing it to see if the results are as expected. Study the different wall treatment options and see which one suits your problem.