Panel Development

[philSweet]

Steve, this post from an earlier thread may help you with FreeShip and developability.

http://www.boatdesign.net/forums/bo...on-two-cones-two-planes-51910.html#post713716


<Edit> I seem to have missed your earlier question. As far as my experience building goes, I did things entirely backwards. I built a couple of plywood boats in my 20's before I had a clue about any of this. They were fairly close copies of other boats, built ply on frame, and I wondered why the ply didn't want to lay flat on the straight frames. So I looked into the matter and bought some books on boat building and lofting. Much later, I got hold of Plyboats Software and that has lead me into the curious world of free software and boat forums. I basically grew up in a woodshop, so my interest in the shapes that the software produces has a lot more to do with the feel of the panels than the exact shape. I use the panel shapes from software for designing plate nestings, and I part out the panels with an allowance even when doing stitch and glue. In particular, I will leave a generous length allowance. The width allowance can be pretty small - 1/4" is plenty. The final panel shapes are governed by the frames.

What I can tell you is that it can work just fine, but getting your head around the math behind it all is certainly worth doing. As I said, I went into this with a pretty good idea of how wood behaved, so I was looking more for assistance than epiphanies and cut files. It's not that I don't trust the software, but things move around in the shop, and plywood is not as tweakable as a plank on frame, especially after it has been epoxied in place. I believe the vast majority of the deviation from design to build occurs in the shop. These methods are not idiot proof, but they can be adequate in skilled hands, and that is what engineering is all about.

For small boats, such as a stitch and glue dinghy or skiff, I would have no problem at all precutting 5 panels based on panel offsets from FreeShip if I had done the mesh, leaving the laps 1/4 wild and cutting the faying edge exact, and assembling over some scrapwood temporary frames. Plan on a 1" allowance on length. Start stitching amidship and work to the ends. Stitch the entire boat if it is smallish, then dab, pull, and fillet. This is not the easiest way for a first-timer to build a boat, though. I think a ply on frame build from an established design is far easier and more intuitive.

 

[smjmitchell]

Phil,
Thanks for the link to this old thread. A long read, but very interesting - particularly the contributions of Michael Storer. Good to know that others have had the same issues with Freeship/Delftship developments in the bow area that I have noted. In fact it is not just the plate developments that I feel don't look right but also the frame cross sections in the bow (they seem to lack enough curvature on each face where the panels sit). The fact that this is attributed to twisted panels is something I need to give some throught to to fully understand. Not surprising as plywood is orthotropic. This brings up the question of how to control the twist during design. Some twist seems inevitable but obviously needs to be minimised.
The design I am working on is ply on frame (as opposed to stitch and glue), but still I want to minimise the stresses in the panels for easy skinning. Also panel sizes are close to sheet sizes in some dimensions and thus I want fairly accurate developments to make sure everything fits on the sheets before cutting wood.
Steve

 

[philSweet]

Okay, so your primary concern is getting the frame shapes right, so that the plywood will lay well. You're not too concerned with the plate expansions. I haven't tried to use FreeShip for this. It would require that two or three edges running across the panel (and along ruling lines) cross over each frame. So you need a pretty dense population of control points along the chines, and some careful attention to control curves and realistic chine curvatures. You probably need 4 or 5 control points per frame bay on each chine. In this way, it should perform much as Rabl's method does.

 

[philSweet]

One other point about using FreeShip to construct developable surfaces. It's a bit of a cobble, but you can do what Mr E is doing in his sketch directly in FreeShip. You need to add a point and position it at the guestimated focus. Then to project the sheer up from the chine, use the Align points function. This function will extrapolate lines as well as interpolate them. Select the focus point, then the sheer point, then the chine point, and the sheer will be moved into alignment. Now add another point and position it in line with the those focus and chine points and repeat with the next set of control points. Instead of a point, you can start with a tiny cylinder and extrude it to get the next focus point. This way you can see your directrix in all views. You can also use the transforms on it and apply control curves to it and thus (somewhat) quickly adjust the model volumes at a coarse mesh level, then subdivide to get better frame definition. Most of these tricks aren't explained in the manual. And I still don't completely understand the effect that control curves have on the software operation. But you need to establish them for all chines and be aware when they are turned on or off. I generally leave them all on.

 

[SukiSolo]

I have used Rhino since before it became commercially available - it was Beta full working in case you wondered. The developable surface tool is very good. However you do need to understand how to create the original surface to some degree. I have unfolded aluminium aircraft wings and the stress curvature tool will show you how much the metal is stretched. Similar for sailcloth if you model a sail.

There are also one or two 2 or 21/2 D tools I used one called Boat Design as an addon to FastCad some years back in DOS. It could also develop panels but from 2D views, and showed all the node connections so you could see no cross over. Having built some racing sculls and a number of boats with double curvature where ply can take a little 'stress' it is still pretty hard to get an exact handle on how a specific sheet will bend....

Concentrate on the hull shape, way more important than developable form. If you have a problem, just scarf a section across a transverse frame and build in thinner ply or veneers (laminated) to achieve the desired form. This also has the added benefit of much less stress in the build and consequently less stress in the entire hull so the shape stays where you put it. Same with any stringers if you use them, laminate or heat bend until they stay where you want them not where they want to bend to....

 

[smjmitchell]

Okay, so your primary concern is getting the frame shapes right, so that the plywood will lay well. You're not too concerned with the plate expansions. I haven't tried to use FreeShip for this. It would require that two or three edges running across the panel (and along ruling lines) cross over each frame. So you need a pretty dense population of control points along the chines, and some careful attention to control curves and realistic chine curvatures. You probably need 4 or 5 control points per frame bay on each chine. In this way, it should perform much as Rabl's method does.

Yes plate expansions is a secondary consideration for this design. In in the general case I would like to get that right as well for other designs I have on my drawing board.
Your comment on needing "two or three edges running across the panel" gives some insight into what the problem might be with Freeship - I had just assumed that internally the code it would do the right thing if you selected a panel as "developable". Perhaps this is not the case .... The problem with any of these programs is that it is impossible to know what is really going on inside. It is not even clear if Freeship does the developments correctly by placing edges in a plane that is tangent to both lines and passing those edges through the tangent points on each line or whether it is making some sort of Rabl like approximation. I did notice some improvement when the density of the control points was increased but I never considered that it might be because there were more edges and I certainly did not have as many points you have suggested might be required. Obviously the complication with using a lot of control points is smoothing the lines ......

 

[smjmitchell]

One other point about using FreeShip to construct developable surfaces. It's a bit of a cobble, but you can do what Mr E is doing in his sketch directly in FreeShip. You need to add a point and position it at the guestimated focus. Then to project the sheer up from the chine, use the Align points function. This function will extrapolate lines as well as interpolate them. Select the focus point, then the sheer point, then the chine point, and the sheer will be moved into alignment. Now add another point and position it in line with the those focus and chine points and repeat with the next set of control points. Instead of a point, you can start with a tiny cylinder and extrude it to get the next focus point. This way you can see your directrix in all views. You can also use the transforms on it and apply control curves to it and thus (somewhat) quickly adjust the model volumes at a coarse mesh level, then subdivide to get better frame definition. Most of these tricks aren't explained in the manual. And I still don't completely understand the effect that control curves have on the software operation. But you need to establish them for all chines and be aware when they are turned on or off. I generally leave them all on.

Interesting idea. I will try this ... something tells me it won't be easy, but I will try it. Of course this assumes that the bottom is developable by a single conic which, at this stage, I am not sure it is.

 

[Wayne Grabow]

Your question is one that I first considered in 1976. Since then I have built nine self-designed developable-hulled boats. I have a website with information describing my technique. I use the mathematics of developable surfaces to create the boat design; thus, the shape is by definition developable, and the dimensions are exact. See developable-surface-boat-designs.blogspot.com
The most recent entries describe my latest design for a runabout. Entries from December 2008 and July 2013 describe more completely the approach used. No computer program; no fancy computer; no drafting board; basically some algebra & geometry with a pocket calculator.

 

[smjmitchell]

I have used Rhino since before it became commercially available - it was Beta full working in case you wondered. The developable surface tool is very good. However you do need to understand how to create the original surface to some degree. I have unfolded aluminium aircraft wings and the stress curvature tool will show you how much the metal is stretched. Similar for sailcloth if you model a sail.

I too have used Rhino with some success to loft aircraft sheet metal. It seems to work fairly well for the simpler shapes in a metal aircraft. However it does have its limitations and crashes quite regularly with some of the more complex lofts that I have attempted (wing / fuselage and canopy / fuselage intersections). But I won't elaborate here because it is outside the scope of this forum.

Of the CAD software that I have experience with I would say that Rhino definitely has the most powerful surface modelling capability (which includes developable surfaces). CATIA is also quite good. As discussed above though there is still the problem of not knowing exactly what is going on inside the code and you can only build confidence by validating it against other proven methods based on the fundamental geometry (i.e. conic lofts).

I am going to do some manual lofts of my hull in Autocad (as I outlined above). I will model the same hull in Rhino to see how the developments and frame cross section shapes compare.

 

[Robjl]

Software suggestion ??

I know how to to develop surfaces (learned for sheetmetal-work many years ago) but I ask why would I do it by hand when good programs are readily available that will work to greater accuracy than I can achieve. I have used a program to design several round-bilge and one multichine yacht.
A free download is still available (I think, with limited saves) at:
http://www.newavesys.com/products.htm
I agree with others who suggest that the real difficulty is producing a hull shape that works well.
For my hard chine yacht I built a number of scale models using the developed panels from the program, ... always worked for me.
You may also find some info here: http://www.newavesys.com/RulingLines.htm

 

[Mr Efficiency]

Your question is one that I first considered in 1976. Since then I have built nine self-designed developable-hulled boats. I have a website with information describing my technique. I use the mathematics of developable surfaces to create the boat design; thus, the shape is by definition developable, and the dimensions are exact. See developable-surface-boat-designs.blogspot.com
The most recent entries describe my latest design for a runabout. Entries from December 2008 and July 2013 describe more completely the approach used. No computer program; no fancy computer; no drafting board; basically some algebra & geometry with a pocket calculator.

Am I right is assuming from that, that your curves ( the chine in particular) are defined by equations, and don't need lofting ?

 

[Wayne Grabow]

"Am I right in assuming from that, that your curves ( the chine in particular) are defined by equations, and don't need lofting ?"
Yes, exactly so. It is the key to the entire approach.

 

[smjmitchell]

Your question is one that I first considered in 1976. Since then I have built nine self-designed developable-hulled boats. I have a website with information describing my technique. I use the mathematics of developable surfaces to create the boat design; thus, the shape is by definition developable, and the dimensions are exact. See developable-surface-boat-designs.blogspot.com
The most recent entries describe my latest design for a runabout. Entries from December 2008 and July 2013 describe more completely the approach used. No computer program; no fancy computer; no drafting board; basically some algebra & geometry with a pocket calculator.

Interesting approach. I have read your Dec 08 and Jul 13 posts a couple of times today and I can (more or less) follow your approach. I need to study the finer details some more. Do you have any more pictures of hulls generated using this method (or drawings of hulls) you would like to share ? (I saw a few in your blog)

This is a great approach since it defines smooth curves for the keel, shear and chines exactly. No fiddling around trying to smooth splines like in the various hull design software products. However I wonder if the functions you are using (parabolic etc) are sufficiently general to model all hulls. Do all your hulls use parabolic equations and straight lines or have you used other equations ? Have you tried using a second degree conic equation ?

From the description of the method it sounded like your lower surface used a straight ruled surface (not necessarily a conic development or multiconic loft) so that would not be strictly developable .... assuming I understood correctly. Did you have any issues getting the skin to fit on the frames ?

Thinking some more about this, it seems like smooth mathematically defined lines (not splines but functions) combined with the exact Kilgore multiconic method for developing the plates might be a very good way to loft a hull. However this all hangs on the ability of a function to define the keel profile, chines and shear in space.

 

[TANSL]

No computer program; no fancy computer; no drafting board; basically some algebra & geometry with a pocket calculator.

Surely you know that a computer and a handheld calculator can perform the same mathematical operations. There is, therefore, not reason for not using the many advantages that a computer has in terms of data manipulation and graphing.

"Am I right in assuming from that, that your curves ( the chine in particular) are defined by equations, and don't need lofting ?"
Yes, exactly so. It is the key to the entire approach.

I think there is no curve that can not be represented (or approximate) through a series of equations. So I do not see the difference between working with equations or lofting. Any curve traced by a loftman can be represented by a set of equations, and vice versa. And also a curve defined by equations may require lofting. Nobody works for the lofting with a single curve, you have to work with a set of them.
As, for now, I am not able to understand it, I would appreciate, if possible, to have more information about your method. Thanks.

 

[Wayne Grabow]

SMJ: All of my hulls have used parabolic equations and straight lines. I looked at other equations, incl. second degree conic, but found that the parabolic gave me the best results with the least complication. A trajectory curve is a form of parabolic curve. Think of a water molecule displaced at some angle by a passing hull and then returned to a rest position by gravitational forces; similar to a projectile trajectory. The parabolic curve can be differentiated to calculate the slope at any point: 2B/L where B is the offset (from maximum beam) to that point and L is the length to that point. The equation can also be integrated to calculate the length of the curved chine to any point (or between points). When a curve is described, it can be a series of shorter curves spliced together with common slopes at the point of juncture. With familiarity, such work is rather simple. On a runabout bottom, I usually make the aft section a straight run; I did do one hull with a variable deadrise, but discovered that most of the time I run a boat at either planning speed or very slow; the variable deadrise was a waste of time. Did you look at the website entry for May, 2006, titled "Maiden Voyage"? It is a double-chine rowing hull with a wineglass stern. I haven't considered whether this approach would work for all hull forms, but it has suited my purposes. Chines usually have a fairly simple shape; beyond that comes visualizing the best projection to create the desired shape.