CFD Online Logo CFD Online URL
www.cfd-online.com
[Sponsors]
Home > Forums > Software User Forums > OpenFOAM > OpenFOAM Running, Solving & CFD

Inaccurate RANS simulations when using Wall-Functions at Re<= 4 10^6

Register Blogs Members List Search Today's Posts Mark Forums Read

Like Tree1Likes
  • 1 Post By andreaRocca

Reply
 
LinkBack Thread Tools Search this Thread Display Modes
Old   April 22, 2019, 11:42
Default Inaccurate RANS simulations when using Wall-Functions at Re<= 4 10^6
  #1
New Member
 
Andrea
Join Date: Dec 2017
Posts: 9
Rep Power: 6
andreaRocca is on a distinguished road
Hi Foamers!
This is the first time asking for help in FOAM community, I hope someone out-there could help me.

I' am currently simulating the flow over a submarine. The mesh is structured and is high quality. Re number, for the case #1, is 10 M. I used firstly wall-resolving approach. The results are pretty good and well-aligned with literature. In this case no problems.
I modified the mesh in order to use wall-functions and, at this Reynolds, I got very good results compared with wall-resolving approach. The only thing to mention is initial residuals of p-equations do not go down more than 4 10^-05.



I want to run LES on this geometry and for this reason I decided to move to lower Re numbers.



In the beginning I moved to Re= 1 10^6. The simulation "wall-resolved" converges fine and results are ok. When I use wall-functions p-equation residuals do not go down 10^-3. Wall functions seems to have problems to properly resolve the flow close to the sail (looking at y+ contours), and the forces differs from the case "wall-resolved" of 10-20%, both the pressure and viscous component. I've tried many many combinations of numerical schemes for the calculations of gradients without any improvement.
I though the problem was the low Reynolds: a different physics which could be more difficult to catch using wall-functions. I think it makes sense.

So I switched to Re 4 10^6. Same story.



So, for Re 10 10^6, no problem when using wall-functions. For Re 4 and 1 million I can't get a good solutions using wall-functions.



Let me give you some more details about the simulation. We are talking about steady-state RANS simulation. Angle of attach 0 deg.

When using wall functions I use Spalding's continuous formulation and <y+> is 25 on the submarine's hull; on sail and rudders has peaks of 40.

Turbulence is modeled using a classic k-omega SST. I use SIMPLEC (so consistent set to "yes" in SIMPLE dictionary), but I tried also SIMPLE or PISO p-v-coupling.

In attachments you can find fvSchemes, fvSolutions and checkMesh Log file.



I'm going to try some different schemes for convective terms, LUST or linearLimited 0.5 /1 , for instance.



Do you have suggestions? Is someone an expert in using wall-function for such flows and could correct me if I'm doing something wrong? Did someone already faced this problem?
andreaRocca is offline   Reply With Quote

Old   April 24, 2019, 10:35
Default
  #2
Senior Member
 
Join Date: Mar 2014
Posts: 108
Rep Power: 9
mzzmrt is on a distinguished road
An interesting situation.

I have done some submarine validation work with suboff and got very good agreement at Re 14e06 with a high Re mesh in the pasat. But I did not test it at lower Re values.

On the other hand, I think the wall functions must work also well at lower Re values unless yPlus distribution on the hull and sail is very bad. You said y+ was 25 on the hull, was it average or max value?

There may also be a problem on the boundary and the inital conditons for the wall-function case.

Could you try any other turbulence model, specially SpalartAllmaras and LienCubicKE?
mzzmrt is offline   Reply With Quote

Old   April 24, 2019, 10:51
Default
  #3
New Member
 
Andrea
Join Date: Dec 2017
Posts: 9
Rep Power: 6
andreaRocca is on a distinguished road
Yes, I said <y+> is 25, which is the average. The max y+ is about 40. "There may also be a problem on the boundary and the initial conditions for the wall-function case." // They are the same (except at the wall) for wall-resolved case. There is any chance I need to change my BCs for lower Re and Wall-functions? I think not. I can try a SA. In the past I've done few simulation on this geometry using SA but it is less accurate compared to k-omega sst. I also tried a realizable k-epsilon without success (it means there was something wrong in setup for epsilon magnitude) and I should conclude that case. I'll let you know with SA if there is any improvement.

By the way, LUST scheme for convective term is unstable. I have switched on limiters (coefficient equal to 1) for gradients and also for laplacian schemes and in that case residuals of p-equation decreased till 5E-05 and solution converged. Forces were 35% bigger that WR. I would like to understand what's going on and how to properly set-up the case.
Santiago likes this.
andreaRocca is offline   Reply With Quote

Reply

Tags
numerical schemes, rans, reynolds number, wall-functions

Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are On
Refbacks are On


Similar Threads
Thread Thread Starter Forum Replies Last Post
Divergence in AMG solver! marina FLUENT 20 August 1, 2020 12:30
Enhanced Wall Treatment paduchev FLUENT 24 January 8, 2018 12:55
help with wall functions Nick Georgiadis Main CFD Forum 10 January 17, 2017 11:03
Wall functions tutlhino OpenFOAM Pre-Processing 0 July 2, 2007 06:04
Wall functions Abhijit Tilak Main CFD Forum 6 February 5, 1999 02:16


All times are GMT -4. The time now is 07:05.