@anajob You can send me a personal message through this site by clicking on my name or avatar.
Dear David, I just wanted to say that I have not put any work in that surface, so it’s not up to what can be achieved. I know how to design that surface, but I would start with some new and clean curves, instead of rebuilding anything.
@Joerg.H Completely understand.
@Joerg.H Your surface looks nice but did you change the edge curve shapes? For a given pair of edge curves there is only one developable surface. If your surface and my surface are on the same side, what is the difference between edges?
How are you using to create the surfaces?
I may be stating what you already know here!
Plating a hull is best done with conical development, this negates any twisting and stressing of the plate.
If you do not use conical development, the twisting of the plate particularly in the forefoot can be imposable if to Sevier. Forcing the plate is not expectable, as stressed plate isn’t happy.
“Conical surfaces” are one specific type of developable surfaces. Developable surfaces are curved in one direction and do not have any twist.
Conical surfaces are popular because they are easy and simple to design, particularly using manual drafting. But other, more general, types of developable surfaces can be designed using various computer algorithms such as in the DevSrf plug-in for Rhino V4 and V5, and the DevLoft command in Rhino V6. D.LOFT is a third party plug-in for Rhino.
Then why does devloft let you define initial ruling lines? The resultant surface depends on the starting point for the ruling lines.
I’m not connected with McNeel so I can’t answer the “why” question.
One shortcoming of DevLft iscreates a surface which runs the full length of both edge curves, with a ruling line (isocurve) at each end. But in general the true developable surface for the pair of end curves may not and often does not have a ruling line which matches the ends of the edge curves. The result from DevLoft may be “close enough” to a true developable surface depending on the acceptable tolerances for the particular situation. DevSrf, the plug-in from McNeel for use with V5, truncates the ends of the surface at ruling lines corresponding to a true developable surface (or close to a true developable surface).
The work-around with DevLoft is to extend the edge curves as needed, and then determine the ruling line which passes through the end of one ruling curve at each end. Then use DevLoft to create the surface with the extended curves, and trim as needed. The side surface I posted above is an example of this proceedure.
My experience so far with DevLoft is it usually does a good job determining the developable surface after the ends of the curves are adjusted, provided a sufficient ruling density is used in the -DevLoft version of the command. Sometimes the surface can be improved by using “Untwist” to specify the location of one or more ruling lines, provided the location of the ruling lines is known.
I’ve developed a simple procedure for determining “exact” ruling lines for developable surfaces based on the edge curves only. It uses Rhino commands, and does not require any iteration. I’ll try to post a brief tutorial about it in the near future.
1 Like
It would help tremendously if DevLoft were history-enabled.
1 Like
I must admit that the white surface was not made with rhino, but with an old yacht design software that I bought in the middle of 1990s. The edge curves were modified slightly. I know that many boats were build succesfully with that solution, so it looks quite right to me. Back to my original comment to Pascals devloft solution: It unrolls quite nicely ( on the screen), but I have seen people to try to build something with rulings running about these direction, and it´s absolutely impossible to plate like this. So I don´t know if devloft is really a better solution than devsrf. And as you already pointed out: the only way to modify the ruling lines is to modify the defining curves. Even this possibility is gotten lost now. Instead you can define rulings in any direction you like, but I don´t believe that this is any help to find a buildable plate developement. I would love to hear comments from other builders / designers
Yes. One should know that Rhino will unroll any surface which is linear in one direction. Then it does not care about any negative curvature. It will only throw out a warning if the area of the unrolled surface deviates too much from the curved one. But there are cases where stretching and compressing compensate. The only way to check is _CurvatureAnalysis. How much compound curvature is acceptable depends on the shape, the size and where the compound curvature happens. Last not least it depends on the material, the available techniques and the people in the shop. I’d say Pascals model is not possible without “special” tools 
I agree with @Jess that area change is not a reliable check on whether a surface is close enough to developable.
In addition to CurvatureAnalysis to look at Guassian curvature I use Curvature to look at the principle curvature and check that one principal curvature is zero or close to zero. Curvature analyzes at a point on the surface so I move the cursor over the surface to see if there are any problem areas.
can’t wait to see this,