Best Marine Design Software for Hull Modeling? (2007)

I disagree with the need for expensive software for hull design , I can get CNC machined parts from Rhino to match other CNC machined parts with tollerances less than 0.5 mm, just a matter of getting the settings right. Even the cheapest software has lower tollerances than the average builder can handle, if used properly. How much more accuracy do you need, even for scientific purposes?

 

I have it from the developers of ShipShape, but it's many years ago and Napa has probably improved. At that time Napa didn't include the extra length due to curvature, but used a linear interpolation between sections.

 

multihull modeling

For the modeling I would start by evaluating Delftship with their free demo. They have many multihulls in their database for you to download and play with. For the structural design I would look at FEA softwares.

 

better than NAPA!

How do you say it is better than NAPA.
Have you tested it with some exact figures?

 

NAPA now

NAPA has undergone many changes. Now it is so user friendly and definitely has a lot of features , because of which it is one of the costliest ship design softwares around. Like TRIBON, FORAN etc.

 

That might just give slow convergence to the "real" answer sometimes. Then again, what you lose on the convex portion of a hull might be made up on the concave portions.

Any ideas on how it performs on something that can be verified objectively?

Regards,
Leo

 

No, sorry.
I can ask.

 

You might not need them for fabricating hulls, but there are definite advantages to exact solutions when testing hydrodynamic codes. Sometimes you need 15 decimal place accuracy or more for that type of work. That is clearly a ridiculous accuracy for real hulls (unless someone here has a radical quantum-effect hull they aren't telling us about). You can't test these sorts of codes using output from Freeship or most other codes because they just aren't accurate enough.

There are many examples where very high accuracy is required. For example, some lifting surface codes break down at the wingtips of highly skewed (crescent) planforms because they require more than 15 decimal place accuracy.

 

I asked the developer of Shipshape again.
They simply uses a wireframe modell of the hull and the distance between wires (frames). Napa did (many years ago) not use the y component, only the spacing between frames...

 

Tri-Star

MacSurf has been Maxsurf for more than 10 years, ever since it became windows compatible.

Multihulls are no problem in Maxsurf, including hull design, hydrostatics, stability, resistance (using slender ship theory), seakeeping or structural design.

Maxsurf is used by Austal for its big cat and trimaran ferries and warships, and is also widely used by sailing multihulls (including Ellen MacArthurs), so I would not be too concerned at its ability to do what you want.

 

To smartbright:

Thank you very much for the advice. It is appreciated.
Will let you know how it goes.

Cheers !

 

Also.... thank you Mr. Mason.

I am aware of the vessels being designed by Austal and that they use
Maxsurf.
Also, if it's good enough for Ellen's designer - I'm sure it does the job.
So Maxsurf will be on our short list.

Cheers all !

 

Leo

I don't agree that ships, when defined by a surface model by programs such as Maxsurf, Freeship, Rhino etc., are not mathematical objects. I agree that they may not have analytical solutions for surface area, however NURBS and subdivision surfaces are capable of being successively refined to give solutions for surface area to within any precision required, up to the limits of floating point accuracy. This may not be exact in a mathematical sense, but for all practical purposes it is sufficiently accurate.

 

I have to agree with both Andrew and Leo. There's no real practical solution to calculate the exact surface area (or volume for that matter) of freeform surfaces like ships, neither using NURBS nor subdivision surfaces. It indeed depends on the gridding. For most hydrostatics programs this gridding is in the form of crossections (stations) and the results of the calculation greatly depends on the number of stations chosen and their distribution. Although DELFTship uses a different method that enables exact computation of volume and areas the outcome still depends on the precision-setting of the model and subdivision surface. Although both methods may not give results accurate to 15 digits, results have proven to be accurate enough for the intended use. I don't think any of the users of this kind of software has sleepless nights over the fact that results are only accurate up till, let's say 6 or 7 digits rather than 15. And even so, any user can always increase the precision by either inserting more stations or adding another subdivision step.

 

Of course it depends on your "practical purposes". NURBS and subdivision surfaces are limited by the accuracy of the hull surface that is supplied. If I supply you with the offsets of, say, the DTMB 5415 destroyer hull, NURBS can only give an approximation of the surface area. If I define the hull with too few, or with poorly chosen spacing, NURBS cannot correct for that.

I'm only arguing that it is difficult to find scientific criteria for comparing various codes. For example, how many nodes (or panels or whatever) are required to give the value of the surface area to (for sake of example) four significant figures for a sphere and a parabolic strut using Maxsurf or Rhino? Does the answer depend on where the nodes are placed? How sensitive is the answer to the gridding? If four sig figs are attainable using a particular grid structure, can six figures or more be achieved using exactly the same spacing?

Regards,
Leo.