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

DFBI: Wave profile with pitching problem?!

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

Reply
 
LinkBack Thread Tools Display Modes
Old   July 11, 2012, 22:27
Default DFBI: Wave profile with pitching problem?!
  #1
New Member
 
Kons
Join Date: Apr 2012
Posts: 12
Rep Power: 5
fuzel is on a distinguished road
Hey all,

this is my first post here in this forum so a big "Hello to all" first of all.

I have a problem I encounter with a DFBI Problem that I will try to describe:

I have a catamaran in a box shaped domain. I am trying to model heave and pitch motions in headseas at zero speed. The domain is pitching with the catamaran. The thing is, when I look at the wave profile along the domain it seems like the coordinates are pitching as well. So as soon as the Cat/Domain starts to pitch the wave profile just gets really weird. So I dont really know how to fix this. I was hoping it was just a coordinate thing but now I am thinking it is due to the domain moving and not being able to resolve the waterprofile!?
I attached some pictures of the wave profile along the domain and the mesh.

Thanks for any thoughts on this!!

cheers

fuzel







fuzel is offline   Reply With Quote

Old   July 12, 2012, 02:06
Default
  #2
New Member
 
Kons
Join Date: Apr 2012
Posts: 12
Rep Power: 5
fuzel is on a distinguished road
Well, I had a look at some other results from earlier.

Maybe this is useful information for understanding my problem.

I did the same simulation with forward speed and the results are great. The differnce in mesh are that the domain for forward speed is a lot smaller. When I use the same for zero speed I get reflections from the back. So that's why I thought lengthening the domain would help.

But probably DFBI with rotating domain is not a good option for long domains?

Is there another way to get rid of the reflections? I noticed that for the smaller domain it was looking better when i defined the outlet as velocity inlet with wave field functions instead as pressure outlet!?

Thanks again for any thoughts.
fuzel is offline   Reply With Quote

Old   July 12, 2012, 02:27
Default
  #3
Senior Member
 
flotus1's Avatar
 
Alex
Join Date: Jun 2012
Location: Germany
Posts: 1,107
Rep Power: 19
flotus1 will become famous soon enoughflotus1 will become famous soon enough
I think I had a similar problem with a compressible LES in CCM+.
Reflections from the boundary conditions messed up the results.
I wasn't able to resolve the problem because CCM+ still doesn't offer a non-reflecting boundary condition that really works.

There are many "techniques" to get rid of reflections (coarsening the mesh before the outlet and inlet, damping fluctuations with a momentum source) but trust me:
They do not work properly.
flotus1 is offline   Reply With Quote

Old   July 12, 2012, 03:50
Default
  #4
Senior Member
 
KHB
Join Date: Aug 2010
Location: Singapore
Posts: 107
Rep Power: 7
lava12005 is on a distinguished road
As Alexander Stief described, you can either coarsening the mesh before the outlet and inlet, and use the Dampinb Boundary condition to damp the wave..

I did one simulation on a small domain with fixed trim and sinkage with and without the damping boundary condition, it helps to reduce the oscillation of the result for me..

The newest version of Star-CCM+ capable of doing overset mesh, have you tried?
It will help you to reduce the mesh size since not the whole domain is rotating with it.
lava12005 is offline   Reply With Quote

Old   July 12, 2012, 04:13
Default
  #5
New Member
 
Kons
Join Date: Apr 2012
Posts: 12
Rep Power: 5
fuzel is on a distinguished road
Hey,

thanks for your ideas.

The Mesh gradually gets coarser towards the outlet already. And I am just trying the damping as well at the moment.

A question though: is the damping length you specify in starccm+ normal length from the boundary or in domain cooardinates? will i need to put -xy metres if my wave propagetes in -x direction?

Another question: why do you want to coarsen the inlet region? In a free surface case that would kinda mess up my desired wave wouldn't it? Or did u mean for another case?

I ll see what I can get out of overset mesh. Any tips on the meshing strategy? Do I mesh the domain with the free surface? and then the moving body with a bounding sphere?

Thanks

fuzel

p.s: hope ill have some results tomorrow and will let you know what damping did for me
fuzel is offline   Reply With Quote

Old   July 12, 2012, 05:30
Default
  #6
Senior Member
 
flotus1's Avatar
 
Alex
Join Date: Jun 2012
Location: Germany
Posts: 1,107
Rep Power: 19
flotus1 will become famous soon enoughflotus1 will become famous soon enough
Just to make this clear: damping zones and coarse meshes CAN, if applied with caution, reduce errors that are related to reflections.
Nevertheless, reflections WILL occur.

In the case i investigated, these reduced reflections interacted with an acoustic mode in the computational domain and made the results useless, no matter how low the amplitude.

About the damping at the inlet: that was just for the example i had in mind.
flotus1 is offline   Reply With Quote

Old   July 12, 2012, 06:47
Default
  #7
New Member
 
Kons
Join Date: Apr 2012
Posts: 12
Rep Power: 5
fuzel is on a distinguished road
Thanks flotus,

I put some different damping options on to run overnight. I ll see if that does the trick tomorrow but I am also not too optimistic.

I also put an overset mesh on and see how that goes. I think that should give me some more freedom should it go allright. Specially when it comes to higher degrees in pitch or roll respectively.

I ll keep you updated...

Thanks again
fuzel is offline   Reply With Quote

Old   July 21, 2012, 04:52
Default results
  #8
New Member
 
Kons
Join Date: Apr 2012
Posts: 12
Rep Power: 5
fuzel is on a distinguished road
So i finally had time to have a look at my results. With a small domain and not too high pitch angles it worked well with introducing a damping function and having a coarser mesh in the region close to the outlet. I still get reflection after some time but the 25 seconds or so i get before they interfere with the incident wave are enough for my investigation.

However, when angles become to high or with a longer domain there seems to be a problem with the interpolation between the new position of the mesh and the old one. So I think a very small timestep might solve this but makes the computation unnecessary long.

So I tried the oversetmesh approach which takes longer to compute but gives very good results even for high angles and/or a long domain and think that's the way to go.

Thanks for your suggestions everyone!
fuzel 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
How to plot the wave contour in ship flow (free surface) problem deddy1973 CFX 6 February 19, 2013 06:49
2D Profile Data Initialization Problem for Transient Setup deniz CFX 0 July 10, 2012 05:43
wave force problem bojiezhang FLOW-3D 2 January 17, 2011 21:01
Problem with a leapfrog scheme for wave equation Shiranui Main CFD Forum 0 June 22, 2010 09:19
VOF WAVE/ DFBI MAB FLOW-3D 2 November 14, 2008 01:02


All times are GMT -4. The time now is 13:45.