[bouclette]

Hi Vincent,

Are you using snappy to do the meshing?

I'm still getting those waves around the hull and in the wake, even with a finer mesh (1.3M cells).

My residuals don't go down at all, they remain flat at around 5e-2. Is that something to be worried about ?

Regards,

Ben

[pablodecastillo]

Hi Ben and Vincent,

Can you post your fvsolutions and fvschemes dictionaries?? Sometimes, depends the mesh i got the ugly waves around the hull, but never in all the domain (allways i runned with interFoam).

[bouclette]

Hi Pablo,

Here are the 2 files plus a pic of the latest I get. This run was with interFoam to see if it makes an difference. Doesn't seem to.

I will try with more refinement at hull surface to see if that makes a difference...

Regards and thanks a lot for your help.

Ben

[desert_1250]

Hi foamers, i hope that you are well,
i try to add dynamic mesh motion to one solver. used this link :
http://www.cfd-online.com/Forums/blo...foam-code.html

but when i compile it with wmake, this error apears :
'correctPhi' was not declared in this scope

i used OF-1.5-dev

tanx
_______
Rasoul

[chripp]

Hello!

bouclette and vince_44 what did you use for meshing? SHM? If so can you upload the dict file so i can test bouclette's case with some other ship?

Thanks!

[bouclette]

Hi there,

Hereś my sHM dict file.

I'm trying now using blockMesh to refine everything and only use snapping of sHM to see if it helps...

Regards,

Ben

[chripp]

Thanks a lot Ben!

I'll try it later and let you know!

As for the blockMesh changing the mesh density and simpleGrading only affects the mesh quality. I attach the blockmesh and SHM files that worked fine for a full scale tanker i'm running so you can check them out.

Hope that helps!

Best regards
Ippokratis

[bouclette]

Hi All,

So I've made a mesh with the refinement in blockMesh and only snapping in sHM (+ level1 refinement on hull) and it seems that it has gotten rid of those parasitic waves. There are still some little ones near the hull where I guess sHM messed up my nice structured blockMesh but I can't do much about that I think( maybe reduce nCellsBetweenLevels to 1 ?) .

Here's a shot of the run (interFoam) after 6s (so not converged yet) and a shot of a closeup of the mesh.

I've made the blockMesh refinement so that the cells are close to a cube shape at the hull and stretched away from it.

Mesh is around 650k cells. CheckMesh reports a max skew of 3.38 (higher than what I had using sHM refinement) and max non orthogonality 59.4 and average 3.25. I can post the full mesh report if anybody is interested.

Thanks very much for all the help.

Regards,

Ben

[chripp]

Hi Ben!!

That is soo nice! Well done!

I'm interested in the mesh report!

I tried to run the case you posted before and in the shm i got a missing groovybc files error. What's wrong? How do i install it?

Also can you post the complete final case folder you did? I would really appreciate it!

Thanks!

[bouclette]

Hi Ippokratis,

I'm running the case until the forces stabilize and if its still fine by then, I will post the files but without the stl because it is way too big.

I still haven't tried with extra layers on the hull so maybe that will be another issue later...

I will also try with LTSInterFoam to see if there is any difference.

Regards,

Ben

[bouclette]

Hi All,

I've refined the mesh to about 1.1M cells to see if it was still working and, after some tweaking, it now runs stable without those parasitic waves anymore. Meshing is definitely the most critical part.

I tried to add surface layers to the hull in order to get a better friction drag value but the mesh ended up being too crap to make the solver to work. It blew up after about 1000 iterations both with LTSInterFoam and interFoam.

I'm attaching the files of the case although they are pretty much the same as the ones in the tutorials and also a pic of the run at 7s (interFoam).

Anybody has a trick on adding clean surface layers to hulls? They fail to extrude mostly around the bow but also randomly along the hull... Any magic settings in sHM?

Next step is to have free sink and trim (2DoF). which tutorials should I study for that?

Regards,

Ben

[bouclette]

I'm also posting some shots of the mesh.

For sure it can be optimized better than this...

B.

[chripp]

Hi there again!

I see you're doing some good work there!

Can you post the blockMeshdict as well?? I want to check what you did with the grading cause i'm having troubles with it.

Thanks!

[daveatstyacht]

Ben,
Good work getting it to run. My experience with sHM (in 1.7.1 mind you) is that it can't seem to produce a good boundary layer mesh for a hull particularly as you have found near the bow. Relative size=true seems to be a little more robust but obviously it is becomes very difficuly then to size (particularly because it is sensitive to the local mesh size from what I can tell). For tutorials for getting familiar with the 6dof solver the pitching wing tutorial under pimpleDyMFoam is not a bad place to start. Alternatively, in interDyMFoam there are som sloshing tank tutorials but I think the pitching wing one is a little more clear what every thing is doing. My experience with interDyMFoam has never been particularly satisfying since it tends be more difficult to make a stable run on a fine mesh.

[bouclette]

Hi Dave,

Thanks for the response. Indeed sHM seems to be a real pain to get it to make clean surface layers... I've been trying to do this for a simple wing and couldn't get it to mesh properly although it was doing a great job for a rectangular section. the refineWallLayer doesn't seem to do a much better job although I haven't tried it for a hull...

As for the dynamic stuff, my goal would be to run dynamic on a coarse mesh in order to get the sailing trim / sink of the yacht and then do a fixed run with the hull trimmed correctly.

I've been doing the wingMotion tutorial as well as the floating body one. Not everything is clear and there isn't much documentation about it but hopefully I'll figure out what to do.

I'm not very clear with the pointDisplacement file, especially with the orientation matrix, the fixedAxis and fixedLine things. Could you enlighten me on that? I want free Ry and Tz with all the rest fixed.

I have taken pointDisplacement file directly from the wing motion tutorial and modified it to suit but not quite sure on my actions... And the motions constraints are not set at all yet...

Btw, attached is the blockMesh file.

Regards,

Ben

[daveatstyacht]

Ben,
Although it has been a while since I touched interDyMFoam. I seem to recall that fixedLine constraint is allowing translation movement only along that axis (in your case heave). The refpoint defines the point of the object that is actually constrained to move along that axis (since other points must move in other axes if rotation is to occur). Presumably you would want to match this to your LCG unless you were trying to match the attachment point of a tow tank. The fixedAxis constraint only allows rotation about that axis (presumably you will want (0 1 0) for trim). There are other constraint types that allow multiple directions of motion (I think Plane was the one for motion in a plane, but I can't remember the other options off hand).

As an aside, if you are attempting to replicate the running trim of a sail boat, don't forget the trim moment of the sail which require you to either modify things to apply a moment equivalent to the driving force acting at the center of effort or to estimate your resistance before hand and modify your LCG to replicate the moment (how tow tanks often approximate it).

[bouclette]

UPDATE: I found the reason for that error message: I was missing a library in the controlDict file... Seems to be running ok so far.

Hi Dave,

Thanks for your post, it made things clearer. I will apply a sails trimming moment at a torque since it seems to be possible in the pointDisplacement file.

I am running into a problem now: it seems that the patchFields used by interDyMFoañ are not the same as for other dynamic solvers. Here is the error msg I am getting:

--> FOAM FATAL IO ERROR: 
Unknown patchField type sixDoFRigidBodyDisplacement for patch type wall

Valid patchField types are :

23
(
angularOscillatingDisplacement
angularOscillatingVelocity
calculated
cyclic
cyclicSlip
empty
fixedNormalSlip
fixedValue
mixed
nonuniformTransformCyclic
oscillatingDisplacement
oscillatingVelocity
processor
processorCyclic
slip
surfaceDisplacement
surfaceSlipDisplacement
symmetryPlane
timeVaryingUniformFixedValue
uniformFixedValue
value
wedge
zeroGradient
)

So it looks like both fixedAxis and fixedLine are not an option in that solver... Any idea why?

Here's my pointDisplacement file.

Regards,

Ben

[bouclette]

Ok,

I managed to make my case run but, after 1s, it seems that only sink is "working". It is sinking a lot at the beginning then bounces back up and looks like it will eventually stabilize somewhere although I haven't carried on solving because it is going very slowly.

Looking at the angular motions, they are all very small (between 1e-6 and 1e-13) but always the trim axis (y) has much lower value than the 2 others. I think that trim isn't solved at all in fact.

I have tried inverting the fixedAxis fields to see if that made some changes but it wouldn't run, telling me that it has reached the maxIterations value (set to 500 but even set to 5000 doesn't work)... Bellow a typical print of motions:
Linear velocity: (3.89761584351e-13 -3.50710150586e-14 -0.319708649006)
Angular velocity: (-6.33825536553e-06 1.42023324474e-12 -1.26833537355e-06)

I will try to fiddle around more to see if I can get something out of it (add damping) but if you guys have some suggestions on what to improve, it would be greatly appreciated!

Attached are the relevant files.

Ben

[daveatstyacht]

Ben,
Looking at your pointDisplacement file you appear to have 2 fixedAxis. When I did free to heave and trim it was with one fixed axis with the axis as (0,1,0). The fixedAxis specifies the axis about which you want freedom of rotation. Likewise the fixedLine specifies the axis you want translational freedom. By sepecifying 2 fixedAxis I believe you have effectively contrained all of the rotational degrees of freedom. You will probably want to add dampening, particularly for trim if you are not concerned with sea keeping since I always found trim to be particularly sensitive.
As an aside, looking at your fvSolution, you may want to slightly underrelax U but if it runs stably as is then no point slowing your convergence down.
Regards,
Dave

[bouclette]

Hi Dave,

I have corrected this after posting last night. I was thinking these constraints would lock the motion and not free it...

I have also added a lot of damping because case would crash due to extreme sink rate.

I still can get it to have a decent sink / trim speed so it is still crashing, but after a bit longer now.

Am I getting someting wrong with the units? The mass should be he half hull displacement in kg right? I've put some quite random numbers for the inertia but that is probably not too much of an issue as long as they are not 0 ?

I will reduce a bit the relaxation factors, maybe it'll help.

Did you manage to obtain some reasonable convergence speed with this or was it very slow? Maybe my plan of running a coarse mesh in interDyMFoam and then a finer one at corrected trim / sink with LTSInterFoam or interFoam is flawed?

Regards,