Runabout developable surface

Rurudyne,

I think that results in the same thing as what I do in freeship. I literally start with a focus point and snap lines to the chine. I can get my first focus point by using the align points function, which will extrapolate a line as well as force middle points into a line. This is not in the book, but you select the two points to reference first and last. All the others will end up inline whether they are between or outside the reference pair.

Once I have a 2 lines running from the chine to the focus, I add a point to the second line to create the next focus, and run a line out to the next point on the chine. And so on. This guarantees the NURB surface will be as close as possible to developable right from the get go. Area errors on plate development are on the order of 10E-6 . Even after pushing points around to get the volumes I want, I'm still usually at order of 10E-5 on edges and areas.

 

If you use the lines of chine and keel to generate a surface by sweeping a straight line between the two lines, is very likely to get a developable surface or "almost" developable.

 

Sometimes.

Lines of a developable bottom surface attached. Note that the station curves are curved, not straight. Station curves of developable surfaces are only straight if the station curves are parallel; ie constant deadrise.
View attachment 93193

 

"Dear all, I'm new to the forums and currently in the beginning stages of designing a runabout that (one day) should be relatively simple to build.

My main question at the moment is how I can be sure that a surface is 'developable'. The first layer over the frames would be marine plywood with a thin mahogany planking finish layer on top."

Several years ago I undertook the same task. I wanted a developable surface runabout which I intended to sheath with an initial layer of marine plywood followed by thin mahogany planking. Rather than learn a new computer program and worry about the algorithms used to approximate a developable surface, I just designed the hull surface using the basic mathematical properties of conic/ruled projections (nothing more than algebra). You can create many different shapes by combining conic and parallel elements connected by ruling lines.

You have been getting some great advice in the other replies to your question. Creating the chine is the best first step. On my most recent project I used a parallel projection (at the bow) and a planar projection aft to create the bottom. For the topsides, a conic projection was used at the bow which transitioned to a second conic projection amidships which then transitioned to a parallel projection aft. The transitions were made using ruling lines. The deck was a combination of parallel projections.

The result is a light, strong, fair hull shape which is relatively easy to sheath. But it does have its limitations; as pointed out previously, the bow of the hull you presented is not developable.

I have some photos and further discussion at developable-surface-boat-designs.blogspot.com

 

Thank you all for some great info! I definetely have something to look into now. I'll give Delftship (we have it here at my university) a try and see what's possible.

My initial plan to use plywood plates was because of the supposed ease of building. To give a little more room to play around with in terms of shapes (for example the bow) would it be a bad idea to go for a hybrid option where the simple parts (bottom, aft, sides) would be developable surfaces made from plywood, and the bow section made by a few layers of diagonal strips?

 

I'm saying same thing than you.
Read my post #6 : a surface with two bends is not developable. This is a simple and fast rule.
In my post # 17 I speak of moving a straight line.
I should also say that when working with NURBS curves is very difficult, if not impossible, to make them go through the required points. The NURBS have control points and fit points, which do not coincide, and often it is impossible act on both. We must act on each of the control points to get the fit point is correct. This job is too hard.

 

When working with nurbs (Freeship) to create developable surfaces, I will do a control surface and matching hull surface. One edge will be coincident and utilize matching edges and control points. This edge is the non-vertex edge of the conic. I'll mess with the vertices until I hunk I have the conic that I want. Once defined, I will visually match the hull surface to the conic. There are two way that I do this. I will make that both surfaces section lines turned on with only the hull surface displayed. I will manipulate the hull surface until section lines are colinear. Sometimes, it takes a few moments because move one control point changes the entire local surface. The alternate method is much more basic. Both the conic and hull layers are visible. The fill display option is selected in the 3D window and the 2D windows are left in wireframe. Control points are selected and manipulated in the 2D windows while the result is monitored in the 3D window. Cross-hatching of the two layers becomes evident when the layers are coincident.

There if a lot of "eye-balling" with this approach, but I am comfortable with the result. Having independent control layers and hull layers allow a bit mor freedom with the control surface and the hull layer only needs to be matched to it afterwards.

 

Henceforth, everything I say is my olpinión, which does not have to be an absolute truth.
When you want to generate developable surfaces, it is best not to use NURBS curves or surfaces. Nurb can simulate a circle or ellipse but will never be real circles or ellipses. Will approach but do not be. When all depends on a lot of "eye-balling" it gets tricky. Two different people will create two different models of the same hull. Moreover, the same person, in different trials, will create two different models of the same boat.
When you make a model of an existing ship is not acceptable that the model looks like the boat, the model must be equal to the ship.
When the hull has one or more knuckles (very common in today's boats), creating sharp edges thereof can be a stressful job if used nurbs.

 

Sorry if you took offense at anything I said. It was not meant as an attack. It was simply an expression of dealing with conics in a nurbs environment. While I referenced your statement, it was used more as a context rather than a criticism. Many of our members use Freeship and Delphship as their primary design software. My comments where directed in their direction in the hopes that it would enhance their ability to build developable hull sections in a nurbs environment.

In reference to "eye-balling", what is the last thing that you do with a design when all is said and done?

I give mine a good "eye-balling".

 

LP, in no time I thought you were criticizing me or you had an unfriendly attitude with me. In all, I feel very comfortable talking to you.
Cheers.

 

A combination of sheet plywood aft and diagonal planking is relatively common.

An example are several Graham Byrnes' designs such as his Outer Banks and Ocracoke series. Byrnes uses sheet plywood planking where the surface is developable, and two layers of plywood strips where the surface has compound curves. The plywood strips run in the same direction with 50% overlap. An inner stripped is applied, then an outer, then another inner, etc which simplifies construction compared to applying the entire layer of inner planking before applying the outer planking.
http://www.bandbyachtdesigns.com/
Ocracoke 20 with photos under construction http://www.bandbyachtdesigns.com/ocracoke.htm
http://www.bandbyachtdesigns.com/custom.htm

 

Conclusions about software which uses NURBS such as Rhino may be mistaken if based on experience with DelftShip or FreeShip which do not use NURBS. They use a sub-division method which is similar to but also different from NURBS.

Edit: Also Software such as Rhino provides considerably more functions to create and control geometry then DelftShip or FreeShip.

 

Creating a NURBS curve which goes exactly though a specified set of points requires a single simple command in Rhino.
If the points already exist then CurveThroughPt builds a NURBS curve through selected points.
InterpCrv builds a NURBS curve through locations which are selected in sequence.
The resulting curves pass exactly through the selected locations.

 

Attached is a drawing of the same developable surface which was shown in post #18 above. This drawing shows both the station curves which are curved in front view, and the straight ruling lines which run diagonally.

 

You can get veneers to have a tiny bit of 3D curvature, so diagonal planking or making the ply in situ is a perfectly reasonable way to achieve the bow shape. This assumes there is not over much distortion from a developable surface. You can also run small sharp knife cuts in the veneer to allow it to slide fractionally open and the gap fills with glue. Or of course the converse where a small cut of a wedge shape allows material to bend slightly tighter.

If done skillfully you can hardly see any difference on the surface veneer, which can be made to look like matching ply as per the rest of the craft.

Very handy for getting decent volume curve transitions on multi chine craft and sweeter forms, mainly in the entry area.