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

Sky-rocketing bounded omega values - Ship resistance

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

Like Tree1Likes
  • 1 Post By thg

Reply
 
LinkBack Thread Tools Display Modes
Old   July 10, 2014, 04:52
Default Sky-rocketing bounded omega values - Ship resistance
  #1
New Member
 
Andreas
Join Date: Jul 2014
Location: Norway
Posts: 8
Rep Power: 4
ommis is on a distinguished road
Hi!
I have been using this forum quite a lot as an unregistered user, but I think I can learn a lot more by taking part in it

I am in the beginning of learning OpenFOAM, and find it both very exciting and challenging. I am a hydrodynamics student, and find free surface flows around ships interesting, so that's what I'm trying to get up to speed at now.
I'm now doing a summer internship, and the department at work wants to do some resistance simulations in OF. We use Star-CCM+ most of the time, but they are interested in using OF too. I have modified the DTCHull tutorial by mostly changing the model (a medium sized catamaran in model scale), but otherwise the rest is fairly similar. The hull is for now constrained in all degrees of freedom (don't want to get ahead of myself!), so my main motive right now is just to get the calculation going and extract some forces.

The problem I face is rather severe - instantly, the bounded omega values goes through the roof, and stays there. All other variables behaves as expected, and have nicely converging residuals (even omega itself), but this bounding isn't good. After a while though, everything explodes..
I run interDyMFoam with OF 2.3.0. My mesh isn't perfect, but it isn't horrible either - I'm just starting to get a grasp on snappyHexMesh. The mesh could of course be the source of my problem. Especially the prism layer isn't optimal in certain areas, but additional refinement of the overall mesh doesn't help much as far as I can see. My thinking is that the other variables should behave strange as well if my mesh is useless - unless omega for some reason is very sensitive to the mesh. I have tried to run the case with no turbulence model to see what happens then, and it runs very well. I therefore suspect that I have missed something crucial somewhere.

All help is very much appreciated!

This is the contents of my omega boundary conditions, which is more or less copy paste from the DTCHull case:

dimensions [0 0 -1 0 0 0 0];

internalField uniform 1; // sqrt(k)/l

boundaryField
{
//- Set patchGroups for constraint patches
#include "${WM_PROJECT_DIR}/etc/caseDicts/setConstraintTypes"

inlet
{
type fixedValue;
value $internalField;
}

outlet
{
type inletOutlet;
inletValue $internalField;
value $internalField;
}

top
{
type inletOutlet;
inletValue $internalField;
value $internalField;
}

bottom
{
type symmetryPlane;
}

hull
{
type omegaWallFunction;
value $internalField;
}

side
{
type symmetryPlane;
}

symmetry
{
type symmetryPlane;
}
}
ommis is offline   Reply With Quote

Old   July 10, 2014, 09:18
Default
  #2
thg
Member
 
Thorsten Grahs
Join Date: Oct 2009
Posts: 56
Rep Power: 9
thg is on a distinguished road
Hej Anders,

sorry - in the majority of cases it's the mesh,
like you already guessed.

Did you did something special with the mesh?
In Ship simulation quitem often one stech the mesh or hav an addition refinment
splitting of the mesh around the free surface. That coul cause problems inside
the mesh and bring your turbulence quantaties to blow up.

Regarding prisms etc. you should spend more effort in snappy on a good qualatiy
castelated and snapped mesh before you generate the prism. Look at the setting
in the propeller tutorial in OF2.3 to get a clue for the snapping criteria.

Besidem from this whats about you settings/schemes.
upwind schemes for turbulence in fvSchemes can help, but not wehen your mesh
is to bad. Wahts about nonOrtho, skewness etc.?

Ha det!

thorsten
Rafael_Coelho likes this.
thg is offline   Reply With Quote

Old   July 14, 2014, 08:03
Default
  #3
New Member
 
Andreas
Join Date: Jul 2014
Location: Norway
Posts: 8
Rep Power: 4
ommis is on a distinguished road
Hei, Thorsten, og takk skal du ha for svaret!

My meshing procedure is like this:
- A rather coarse background mesh generated with blockMesh
- Anisotropic refinement in z-direction around the free water surface to capture the wave generation - done with topoSet and refineMesh, in many steps.
- Additional refinement in proximity of the hull to get the cells here as cubic as possible (if I don't do this, the snapping is very poor due to the highly rectangular cells emerging from the anisotropic water line refinement). To do this I use topoSet and refineMesh, where I refine blocks around the hull.
- Then I run snappyHexMesh...the snapping looks really good, but I don't know why the layers are so poor. I add the layers with an absolute overall thickness (not relative) and the expansion ratio set to 1,3. I have kept most of the other layer settings as in the DTCHull tutorial, but I've tried to edit some of them with little or no improvement. The layers seem fine around long straight lines, but as soon as there is an angle (it doesn't have to bee sharp), the layers on either side "shrink" gradually to almost zero thickness. There are a few angle settings I don't fully understand, which might help...

When snappyHexMesh is finished, it displays zero bad cells, but when I run checkMesh, 5 highly skewed faces are found. Can only 5 skewed faces really corrupt a mesh consisting of ~500k cells?

My scheme for both k and omega is Gauss linearUpwind limitedGrad.

-Andreas
ommis is offline   Reply With Quote

Old   July 18, 2014, 10:01
Default
  #4
thg
Member
 
Thorsten Grahs
Join Date: Oct 2009
Posts: 56
Rep Power: 9
thg is on a distinguished road
Hei Andreas, ingen aarsak!

Well in priciple your mesh generation proceeing sounds quite reasonable.
I would try to use may a factor less than 1.3 and play more with the snapping parameter in order to get the layers better.
The key to layers is an good quality snapped mesh.

Be extremly carefull with the aniso-refinment in z-direction.
This could cause bad transition between mesh region and this could cause
unbounded turbulence quantaties (at least in my first tries...).

Regarding the schemes for k and omega I would start with upwind.
For turbulence quantaties, this is ok (divSchemes would be not)
and this will stabalize the run. But may be not, if your problems stems
from the problems I mentioned above.

Keep on FOAMing!

Thorsten
thg is offline   Reply With Quote

Old   July 22, 2014, 05:33
Default
  #5
New Member
 
Andreas
Join Date: Jul 2014
Location: Norway
Posts: 8
Rep Power: 4
ommis is on a distinguished road
Thank you for the good advice! They are really helpful
ommis 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
Problem in setting the values of k and omega Naruto OpenFOAM Pre-Processing 9 July 12, 2013 08:38
k omega values mihaipruna OpenFOAM Running, Solving & CFD 0 November 8, 2012 10:41
Ship Resistance problems using CCM+ Dan CD-adapco 9 April 6, 2011 04:27
Values of k and omega at the wall oldah FLUENT 2 February 22, 2008 04:08
Ship Wave Making Resistance Yasser CFX 7 April 24, 2006 14:56


All times are GMT -4. The time now is 02:34.