Question: Constraining a ship FEA

[TheMan008]

Question:

Full Global model of a ship. How can a ship be properly constrained in finite element analysis software. A fixed constraint at any node does not properly simulate hydrostatic buoyancy forces.

How can buoyancy forces be properly modeled in FEA?

 

[matkenmd]

Apply a gradient pressure under the waterline driven off the depth from the waterline (.45psi/ft)

 

[SRossiter]

I would use a specialist Marine FEA package like Orcaflex to get fully coupled non-linear response.

Steven Rossiter MEng - University of Sheffield (Aerospace) -

http://www.bpp-tech.com

 

[TheMan008]

If you have a ship hull and you apply a pressure gradient to act as a lifting force, theoretically you could use the self weight from the 3d cad model to a weight.

But if you were even the tensy wensy bit off, your ship would sink to the bottom of the ocean, or fly up into the air.

Thats why I don't feel like any one has answered this question correctly.

Maritime Systems Engineering
Texas, USA
Close the border to foreign engineers.

 

[MikeHalloran]

You would distribute the weight over all the elements, right?

And you would distribute the buoyancy over all the wetted nodes?

And you would model the buoyancy as, what? A nonlinear spring at each wetted node, grounded at the global zero waterplane?

Speaking of which, do you intend to model wave action, i.e. with a nonplanar wetted boundary?


<<
I claim zero expertise in this field, or any other. I'm hoping that you'll refrain from insulting me long enough that the people who actually know this stuff will start to contribute and turn it into a discussion that's useful to you, and perhaps to someone else. When that happens, that's why these forums don't suck.
>>

I'm reminded of a dynamics problem from ~1972. My friends were modeling a helicopter (for a simulator), and the first landing gear bounce on landing took it up to 1600 feet. They had to decrease the time interval for that computation in order to make it bounded. Sort of a sampling problem, really, not the equilibrium problem you seem to be worried about. The whole simulator ran on a computer that would have made a Timex Sinclair (which appeared much later) look like a supercomputer. They may or may not have upgraded the computers, but the simulators are still in service.



Mike Halloran
Pembroke Pines, FL, USA

 

[GregLocock]

You asked this question elsewhere and I suggested a good real reason why a conventional FEA package will struggle, or at least require an interative approach. Hungry horse.



Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[GregLocock]

"But if you were even the tensy wensy bit off, your ship would sink to the bottom of the ocean, or fly up into the air. "

True, but if you don't run it as a dynamic simulation then the stresses will only be a little bit out as well, if you use inertia relief to help ignore the errors. or you could use a 321 arrangement in addition to buoyancy, and then iteratively scal g until the 321 reaction forces are zero.

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[TomaszK]

Hello everybody.

I'm new at the forum, so Hi again.

From my experience. Pressure distribution you usually receive from specialised software for hydro analysis (if you do global analysis, as in this case). All the motions and accelerations are also transferred or read from hydro analysis.
For structural analysis common practice is support model in 3 nodes. First at section of centre line with base plane at stiff structure in the aft (all three translations). Second at the same section, but in fore (transverse and vertical directions). And third node at main deck above first node (only transverse direction).

If I may recommend, please see:

http://exchange.dnv.com/publishing/CN/CN34-1.pdf

- this is guide about direct analysis of ship structures and there is a chapter about FE model.

I've hope that this is quite clear and my English isn't very bad.

regards
Tomek

 

[rob768]

If you want to do it the difficult way: apply a pressure on the wetted surface with a magnitude depending on the depth of the water (within Femap, which i use, you can create a function in which you use the Z-coordinates of the elements you apply pressure on). The total created upwards pressure should compenaste total weight. If not, your waterline is wrong (or the weight incorrect). If the difference is small, you can use a factor on mass or pressure to correct it.
Then fix the ship at both ends and evaluate constraint forces. If one is large (or has a different sign) yo may have to adjust the trim of your waterline.
This way, you can even define hogging and sagging waves.

 

[chbrow10]

What type of stress result are you looking for? Since you have a global model, I assume you are looking for primary longitudinal stresses? If you wanted hydrostatics, you could do a smaller panel ny panel method, right

Assuming you are looking for a longtudinal bending type response:

First, you have to balance the model on the wave, or on calm water, whichever case it is. Buoyancy must equal weight for the vessel. Find the neutral axis, and place your boundary conditions near NA (calm or on wave). Don't over-constrain the model, you just want to stop if from flying off. I use 2 BCs, one forward and one aft. One is restrained in the y, the other in x and z. You know you have it right when your nodal forces at the BCs are very close to zero.