I have made a little bit of progress and wrote a new ruling line computation program that optimizes a monotonic B-spline which maps the parameter space from one Bezier curve to the other. This program always finds a smooth ruling, even if the shape is not developable and I can look at the twist to see how close to developable the shape is. I have also incorporated penalty functions to trade twist for reduced curvature. This works pretty well as long as the stems are not too steep, but I still have not been able to create a nicely developable shape with plumb stems. I either have a lot of twist (8-12 degrees) or a lot of curvature in the cross sections.

I got curious to see what other programs are available and found a linux version of FreeShip. So, I created a file with 100 data points each for keel, chine and sheer from Bezier curves I had fitted to the Sea Tour 15 offsets. I imported this into FreeShip and generated panel developments. I compared the generated panels to a panel development I had made with my software from the Bezier curves with straight cross sections. I roughly aligned the two panels and plotted them in the attached figure. The two plots show the same data, the top plot has an exaggerated y-scale. This is the panel between keel and chine. FreeShip reports zero edge error and an area error on the order of 3e-3 (units unknown, could not find much information in the manual).

As you can see in the plot, the FreeShip development has a much more concave keel at the stems than the straight line ruling. I got curious and figured out how to show cross sections in FreeShip, and sure enough they are concave at the ends. I am not sure what to make of it, but I cannot imagine getting the concave keel line to be convex after assembly. It looks like the keel has to be pushed up by about 40mm over a distance of 0.5m in an area where the panel is already pretty vertical. There are some free S&G plans on the internet that I have looked at, and some have a concave keel near the stems, but nothing as dramatic as the FreeShip panel shown in the plot.

The problem shown in the plot gets worse if I use plumb stems. I either get cross sections that deviate from a straight line by more than half an inch, or I get a lot of twist resulting in the concave keel line. The only way around that I have found is to use an extra chine that is level with the keel near the ends. This results in nicely developable panels unless I pinch the ends of the bottom chine too much.