Strange Velocity Gradient in the Viscous Sublayer

wcj1n18 · June 24, 2022, 12:05

Hello,

I am simulating the viscous sublayer of open channel flow with a bed of spheres on the bottom wall with a steady, periodic, RANS (k-w) simulation. Now, I have made the cells very small, e.g. perhaps 20 times smaller than than the viscous sublayer. When I run the simulation for a given number of iterations the residuals drop and I take the streamwise velocity along a vertically-aligned probe. The velocity gradient between cells is about what I would expect it to be, however, within the cell, it seems the velocity gradient is constant- as in, it is constant to 15 significant figures. Is there something I need to change to make Star-CCM+ interpolate values within cells?

Domain and Velocity gradient images attached

Thanks for the help in advance

Will

Aeronautics El. K. · June 25, 2022, 04:44

Out of curiosity, are you using wall functions?

wcj1n18 · June 27, 2022, 07:18

Yes, I have the all y+ wall treatment turned on. Perhaps I should set this to low y+ wall treatment seeing as my first cell is much less than y+? I notice I can compute u* iteratively so perhaps I should turn this on too? Any other ideas?

Aeronautics El. K. · June 27, 2022, 07:23

Quote:

Originally Posted by wcj1n18

Yes, I have the all y+ wall treatment turned on. Perhaps I should set this to low y+ wall treatment seeing as my first cell is much less than y+? I notice I can compute u* iteratively so perhaps I should turn this on too? Any other ideas?

Assuming you have some time, I'd be curious to try without any wall functions. If your y+ is so low you're actually solving the viscous sublayer so no need to model it.

wcj1n18 · June 27, 2022, 07:25

I see what you're saying, although I'm also interested in obtaining a varying velocity within each cell so I think without wall functions turned on it probably wouldn't give that?

wcj1n18 · June 27, 2022, 07:35

OK, i'd forgotten I'd run another simulation using, what I thought, was the exact same set up. In this case i've noticed that u* i set to be found iteratively and this seems to have solved the problem! It's also SST (Menter) K-omega rather than the earlier version of the model, although I doubt this would affect it based on my limited understanding of the model

Thank you for the help!

June 25, 2022, 04:44		#2
Aeronautics El. K. Senior Member Lefteris Join Date: Oct 2011 Location: UK Posts: 341 Rep Power: 16	Out of curiosity, are you using wall functions? __________________ Lefteris

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
pisoFOAM (LES) - internal pipe flow - convergence	gu1	OpenFOAM Running, Solving & CFD	0	January 11, 2018 17:39
How to find the velocity gradient and variance of velocity fluctuations	Sharkito	Phoenics	1	March 30, 2016 04:41
Velocity gradient	Stephen	Main CFD Forum	0	April 7, 2003 11:28
Terrible Mistake In Fluid Dynamics History	Abhi	Main CFD Forum	12	July 8, 2002 10:11
velocity of a body fall. th. viscous medium	sohrab arora	Main CFD Forum	3	September 11, 2001 23:33

June 27, 2022, 07:18		#3
wcj1n18 Member Will Crawford-Jones Join Date: Jan 2021 Posts: 43 Rep Power: 5	Yes, I have the all y+ wall treatment turned on. Perhaps I should set this to low y+ wall treatment seeing as my first cell is much less than y+? I notice I can compute u* iteratively so perhaps I should turn this on too? Any other ideas?

June 27, 2022, 07:25		#5
wcj1n18 Member Will Crawford-Jones Join Date: Jan 2021 Posts: 43 Rep Power: 5	I see what you're saying, although I'm also interested in obtaining a varying velocity within each cell so I think without wall functions turned on it probably wouldn't give that?

June 27, 2022, 07:35		#6
wcj1n18 Member Will Crawford-Jones Join Date: Jan 2021 Posts: 43 Rep Power: 5	OK, i'd forgotten I'd run another simulation using, what I thought, was the exact same set up. In this case i've noticed that u* i set to be found iteratively and this seems to have solved the problem! It's also SST (Menter) K-omega rather than the earlier version of the model, although I doubt this would affect it based on my limited understanding of the model Thank you for the help!