CFD Online Logo CFD Online URL
www.cfd-online.com
[Sponsors]
Home > Forums > CFX

Turbulent SST model

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

Reply
 
LinkBack Thread Tools Display Modes
Old   October 28, 2010, 17:56
Default Turbulent SST model
  #1
New Member
 
Join Date: Oct 2010
Posts: 15
Rep Power: 6
fergal is on a distinguished road
Hi,

I'm quite new to CFX, i am currently modelling laminar and turbulent incompressible flow over a smooth flat plate. I have completed the laminar model, however when i increase the velocity (selecting the SST model) the solver crashers during the first iteration.

the error given is a floating point error; zero divide.

I'm using CFX-Meshing and i have conducted a sensitivity analysis (the laminar option works but the turbulent option doesn't using the same mesh).

thanks in advance for any help

Fergal
fergal is offline   Reply With Quote

Old   October 28, 2010, 18:11
Default
  #2
Super Moderator
 
Glenn Horrocks
Join Date: Mar 2009
Location: Sydney, Australia
Posts: 10,958
Rep Power: 85
ghorrocks has a spectacular aura aboutghorrocks has a spectacular aura aboutghorrocks has a spectacular aura about
Steady state or transient?
ghorrocks is offline   Reply With Quote

Old   October 29, 2010, 05:58
Default Turbulent SST model
  #3
New Member
 
Join Date: Oct 2010
Posts: 15
Rep Power: 6
fergal is on a distinguished road
Hi ghorrocks,

The model is steady state, i'm unsure of what aspect of the model is giving issues. The mesh i am using is an extruded 2D mesh with an expansion of 5% from the plate.

Thanks for your help.

Fergal
fergal is offline   Reply With Quote

Old   October 29, 2010, 06:50
Default
  #4
Super Moderator
 
Glenn Horrocks
Join Date: Mar 2009
Location: Sydney, Australia
Posts: 10,958
Rep Power: 85
ghorrocks has a spectacular aura aboutghorrocks has a spectacular aura aboutghorrocks has a spectacular aura about
What is the error message? Please post the final sections of the output file.

But the answer almost always is to use a better initial guess to start you off. Use a laminar model as an initial condition. Even if it was done at a different flow velocity it is better than nothing. And you will probably need a smaller physical timestep to start off. Once it has started and is reliably converging then increase the time step size.
ghorrocks is offline   Reply With Quote

Old   October 29, 2010, 09:34
Default
  #5
New Member
 
Join Date: Oct 2010
Posts: 15
Rep Power: 6
fergal is on a distinguished road
please see message below,

+--------------------------------------------------------------------+
| Convergence History |
+--------------------------------------------------------------------+

================================================== ====================
| Timescale Information |
----------------------------------------------------------------------
| Equation | Type | Timescale |
+----------------------+------------------------+--------------------+
| U-Mom | Auto Timescale | 1.59399E-04 |
| V-Mom | Auto Timescale | 1.59399E-04 |
| W-Mom | Auto Timescale | 1.59399E-04 |
+----------------------+------------------------+--------------------+
| K-TurbKE | Auto Timescale | 1.59399E-04 |
| O-TurbFreq | Auto Timescale | 1.06266E-04 |
+----------------------+------------------------+--------------------+

================================================== ====================
OUTER LOOP ITERATION = 748 ( 1) CPU SECONDS = 4.843E+04 (1.994E+01)
----------------------------------------------------------------------
| Equation | Rate | RMS Res | Max Res | Linear Solution |
+----------------------+------+---------+---------+------------------+

+--------------------------------------------------------------------+
| ERROR #001100279 has occurred in subroutine ErrAction. |
| Message: |
| Floating point exception: Zero divide |
| |
| |
| |
| |
| |
+--------------------------------------------------------------------+

+--------------------------------------------------------------------+
| Writing crash recovery file |
+--------------------------------------------------------------------+

Details of error:-
----------------
Error detected by routine POPDIR
CRESLT = ILEG

Current Directory : /FLOW/NAMEMAP

+--------------------------------------------------------------------+
| An error has occurred in cfx5solve: |
| |
| The ANSYS CFX solver exited with return code 1. No results file |
| has been created. |
+--------------------------------------------------------------------+

End of solution stage.

+--------------------------------------------------------------------+
| Warning! |
| |
| The ANSYS CFX Solver has written a crash recovery file. This file |
| has been saved as |
| C:\Users\Administrator\AppData\Local\Temp\Turbulen t_rev01_5440_Wo- |
| rking\dp0\CFX\CFX\Work1\CFX_002.res.err and may be an aid to |
| diagnosing the problem or restarting the run. More details should |
| be available in the solver output section of the output file. |
+--------------------------------------------------------------------+


+--------------------------------------------------------------------+
| The following user files have been saved in the directory |
| C:\Users\Administrator\AppData\Local\Temp\Turbulen t_rev01_5440_Wo- |
| rking\dp0\CFX\CFX\Work1\CFX_002: |
| |
| mon |
+--------------------------------------------------------------------+


This run of the ANSYS CFX Solver has finished.


i forgot to mention that i got a k-eps model running under similiar conditions. i tried a similar problem using a mapped face meshing and SST model and it worked fine.

Thanks again
fergal is offline   Reply With Quote

Old   October 30, 2010, 05:46
Default
  #6
Super Moderator
 
Glenn Horrocks
Join Date: Mar 2009
Location: Sydney, Australia
Posts: 10,958
Rep Power: 85
ghorrocks has a spectacular aura aboutghorrocks has a spectacular aura aboutghorrocks has a spectacular aura about
OK, so it is diverging in the first iteration on the first set of equations. Then you need a better initial guess or smaller timesteps to start off with.
ghorrocks is offline   Reply With Quote

Old   October 30, 2010, 17:47
Default
  #7
New Member
 
Join Date: Oct 2010
Posts: 15
Rep Power: 6
fergal is on a distinguished road
Hi

Thanks again for the quick response, i tried reducing the timestep and ran a K-eps model for the initial guess but it still keeps giving the same error.

i set up two models, both to model turbulent incompressible flow over a flat plate (sst).

1) the flat plate is a smooth no-slip wall (works)
2) the flat plate is a smooth no-slip wall with a small free-slip plate at the inlet (this is not working (see crash recovery file))

thanks again

fergal
fergal is offline   Reply With Quote

Old   October 31, 2010, 05:37
Default
  #8
Super Moderator
 
Glenn Horrocks
Join Date: Mar 2009
Location: Sydney, Australia
Posts: 10,958
Rep Power: 85
ghorrocks has a spectacular aura aboutghorrocks has a spectacular aura aboutghorrocks has a spectacular aura about
Is the no slip region planar? You might be able to make it a symmetry plane.

Also try making the timestep smaller still. How much smaller have you already gone, and what is the flow timescale? Also try starting with local timescale factor.
ghorrocks is offline   Reply With Quote

Old   October 31, 2010, 08:03
Default
  #9
New Member
 
Join Date: Oct 2010
Posts: 15
Rep Power: 6
fergal is on a distinguished road
hi ghorrocks,

Thanks again for the hasty response and your help, i just ran the sst model using the results from the k-eps model and a local time-scale factor of 0.00001.
It is running . do you suggest stoping the run and changing the timescale back to automatic??(i tried this and the same problem persists)

thank you so much for the help.

Fergal
fergal is offline   Reply With Quote

Old   October 31, 2010, 18:14
Default
  #10
Super Moderator
 
Glenn Horrocks
Join Date: Mar 2009
Location: Sydney, Australia
Posts: 10,958
Rep Power: 85
ghorrocks has a spectacular aura aboutghorrocks has a spectacular aura aboutghorrocks has a spectacular aura about
If it needs such a tiny time step to start off then you will need to ramp it up to a sensible value slowly as the simulation progresses.

But if you need such a small value to start it off something is probably wrong anyway. Is your mesh quality good? Tried double precision numerics? What about using a symmetry plane rather than a slip wall?

And why do you want to put a slip wall in anyway?
ghorrocks is offline   Reply With Quote

Old   November 1, 2010, 09:42
Default
  #11
New Member
 
Join Date: Oct 2010
Posts: 15
Rep Power: 6
fergal is on a distinguished road
Hi,

I set up a new model without the free-slip wall and it works fine. I was using the free-slip wall to reduce any negative effects at the leading edge of the plate (inlet condition).

I'm still unsure why the model runs with no free-slip region and it doesn't when the free-slip region is added to the model. This problem only exists for the SST model as the k-eps model works fine.

Is there a method within CFX of checking mesh quality?

Again thank you so much for the help.

fergal
fergal is offline   Reply With Quote

Old   November 1, 2010, 17:41
Default
  #12
Super Moderator
 
Glenn Horrocks
Join Date: Mar 2009
Location: Sydney, Australia
Posts: 10,958
Rep Power: 85
ghorrocks has a spectacular aura aboutghorrocks has a spectacular aura aboutghorrocks has a spectacular aura about
I have not come across a problem with free slip walls like this before. But I do not use free slip walls very much as they are not physically possible, and from your description of what you are doing I don't think you should use them either.

You can check mesh quality in the output file as a summary, or in detail using CFD-Post. Read the documentation to get a guide for how to interpret the mesh quality numbers.
ghorrocks is offline   Reply With Quote

Old   November 2, 2010, 08:45
Default
  #13
New Member
 
Join Date: Oct 2010
Posts: 15
Rep Power: 6
fergal is on a distinguished road
Hi ghorrocks,

I have been going crazy with this model. I think one of the problems was drawing the geometry in mm in the design modeller (i use workbench) it seems to be sensitive to a change in units. A re-drawing of the model in metres seems to have worked. But would this have caused the problem.

I take your point on the use of free-slip walls.

Thanks for your patience and help.

Fergal
fergal is offline   Reply With Quote

Old   November 2, 2010, 19:15
Default
  #14
Senior Member
 
Attesz's Avatar
 
Attesz
Join Date: Mar 2009
Posts: 355
Rep Power: 8
Attesz is an unknown quantity at this point
Quote:
I think one of the problems was drawing the geometry in mm in the design modeller (i use workbench) it seems to be sensitive to a change in units. A re-drawing of the model in metres seems to have worked.
I think it doesn't matter which unit you use. Your problem is propably related to your mesh or the boundary conditions. I did simulations on a centrifugal compressor with a very coarse mesh, and it converged in 200 iterations with automatic timescale and SST. Your model is more simple. Can you post images about the geometry or the mesh?
Attesz is offline   Reply With Quote

Reply

Thread Tools
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 On
Pingbacks are On
Refbacks are On


Similar Threads
Thread Thread Starter Forum Replies Last Post
Low Reynolds Number SST Model Josh CFX 39 February 19, 2014 09:47
Missing Information in SST Transition Model Formulation Josh CFX 2 September 14, 2010 11:49
Low Reynolds k-epsilon model YJZ ANSYS 1 August 20, 2010 13:57
difference of the laminar and turbulent model duaiduaihu FLUENT 0 August 13, 2010 23:40
SST model help please! Sheila CD-adapco 2 September 6, 2004 22:39


All times are GMT -4. The time now is 03:13.