[fuzel]

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]

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.

[flotus1]

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.

[lava12005]

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.

[fuzel]

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

[flotus1]

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.

[fuzel]

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]

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!