Buoyancy - Stability simulation - nonlinear analysis

[quequen]

Non linear analysis applied to a simple stability - buoyancy exercice.
I made it using a game-physiscs simulator, with a reasonable accuracy.
I think this is a good sample of what a good nautical software should do for us:

http://www.youtube.com/watch?v=aFTQByZ4cx0

Blender (GNU licenced) can manage many of the issues shown here.

[daiquiri]

Originally Posted by quequen

I made it using a game-physiscs simulator, with a reasonable accuracy.

Very nice video and a good job done with 3D modeling of that sailboat.

However, without intention to dismiss the amount of work you have put into this simulation, its practical usefulness depends on how accurate is the model of hydrodynamic forces involved (for example, those acting on the keel, on the submerged mast or on the submerged corner formed by the hull-deck joint). This doesn't look like a CFD software, so you obviously have to model these forces manually and separately for each component. That's a guesstimate-intensive task, hence the results of it have to be verified against known data.

To make it brief, I'd say that it is not possible to judge how "reasonable" is the accuracy of this simulation without comparing its numerical output with the numerical data from full-scale tests.

Cheers!

[quequen]

Daikiri, I agree 100%, but would add some thoughts:
This is, by now and as it is shown on this example, just a sofisticated toy, plenty of little mistakes, but essentialy could be much more accurate than any other traditional-linear calculation metod, just because it works as real physics do in reality. It uses just mass, CG of bodys and CB of submerged volumes, newton and archimedes, and a sophisticated math to simulate water behavior (including viscosity and friction), and a lot of iteration and processor work, that's all. It don't need TKM, LKM, AftPerp, FrontPerp, etc. wich are mathematical conventions to simplify linear calculations.
Certainly, game-phisics simulators are really easy (and fast) to use to the final user. The sailboat used was modelled using Delftship free, but any 3D software cappable of exporting .stl .dxf or .wrl could be used. After setup of well done limits, to check buoyancy of your models can take from minutes to just seconds.
By the way, I used Reactor, from Havok, over an old 3DStudio version (9). Havok sells his work to developers, not to final users.
Reactor has many limitations, but could be easily evolved to a professional tool (not only nautical).
Blender has many of this cappabilities, unfortunately I don' know Blender yet.
Regards

[quequen]

In addition, I've found that Google Sketchup can (roughly) manage this problem using Sketchyphysics (a GNU licenced Game-Physycs simulator that uses Newton Game Dynamics, GNU licenced also). By now, this are non-scientifical approachs to the problem, but we should be aware, because they could evolve rapidly.
To manage this problem, game-physycs envelope volumes into a mesh. Havok can use a very accurate mesh, resulting in a very well calculated Center of Buoyancy. Sketchyphysics seems to use a very simple mesh, with low accuracy.
In both cases, the major limitation resides on how they manage object's Gravity Center: Reactor and Sketchyphysics fix CG at the mass center of each object. This works well with solid objects having only one material, but is not enought accurate when objects are hollow or thick surfaces. This problem could be easily avoided allowing manual control of CG position for each part of the assambly.
Plug-ins for Sketchup are growing fast. May be in a short time some new developments on this issue will appear.