Predicting plywood bending behavior for stitch and glue design

[CET]

Not sure whether this is the appropriate forum to ask these questions in, but here goes...

When designing a stitch and glue hull for plywood composite construction and all developable panels, how can one predict how the plywood will bend? For example, if a panel of a given thickness is “stitched” to another panel, and the meeting edges of the two panels are curved, are there calculations or tools available to predict what shapes the panels will assume? Also, are there calculations or tools available that predict the maximum bend a panel of a given thickness will take before failing?

Modeling designs in Rhino is one thing but I have no idea whether the wood will conform to the developable shapes I am modeling. Does Rhino have functionality for this, or are there plug-ins that can be used with Rhino to accomplish this?

Any info would be much appreciated.

CET

 

[marshmat]

I don't know of any computer tool that can do this.... although a lofted surface in Rhino does, I think, come pretty close to what the sheet would actually do. It should be safe to assume that the plywood will, if allowed to find its own shape, form a section of a cone.

Maximum curvature before failure seems to be a bit trickier. You could try modelling it in terms of the compressive / tensile strain in the inner and outer veneers, respectively, just from the geometry of the curve (use the panel dimension when flat as the arc length on the neutral axis, and compare to the arc lengths at +/- 1/2 the thickness either side of that). But failure could occur in either the outer veneer or the glue plane, and if you're bending axis is parallel to the grain of the outer veneer, then perhaps the second veneer will fail first. And that's not accounting for steaming or hot water, which can really let you curl ply up tight.

Phil Bolger is, by a long shot, the master of plywood prediction. His stitch-and-glue plans fit so well that the wire stitching is completely unnecessary, hence why they're called "tack-and-tape". I think part of the trick is in the matching of the curvature of the two panels, measured parallel to the seam.... Bolger reportedly says this also helps reduce vortex formation at the chines.... but don't quote me on that.

 

[dougfrolich]

Also a look at the gougeon Bros. book on boat construction will give you some limits on min bend radius vs thickness---

 

[Gary Baigent]

bending ply

Here's a jpeg of the bow sections of the CBSkimmer - in 4mm meranti plywood. Other plys will bend even further, like Hoop pine and of course the expensive aircraft plies in 1.5 and 3mm guages.

 

[CET]

Thanks for the replies.

Gary - that looks like a great design. How did you get the plywood to conform to those shapes? I would love to see more pictures.

 

[Gary Baigent]

CET
There are other images of the Cox's Bay Skimmer in the backblocks of boatdesign.net - but here is one more.
Bending ply requires some patience and slowly increasing clamp pressures, plus a few judicious saw cuts in a couple of high tensioned areas (but watch out the whole thing doesn't explode like a spiked balloon).

 

[alan white]

To calculate how far a panel will bend, rip a strip of it and bend. Stay within a healthy percentage of destruction (say, 2/3) as a maximum radious. For greater accuracy, rip a series of plywood strips. Twisting is the same as bending except for the direction of bend relative to plywood grain direction.
You can test that too, by twisting a strip. Generally, breakage through twist isn't an issue.
The nice thing about wood is, it sets eventually into an induced shape, so that while the original bend came within, say, 2/3 of breaking strength, after some time it relaxes and the breakage radious decreases from the original calculation. Same with solid wood.
I've done coamings without steaming by laying them between two supports for a couple of weeks in the summer with weight bowing them down. To one degree or another, all wood and all plywood can be induced to bend further this way.

Alan

 

[sholt85]

using rhino or rhinomarine there is a command under the "surface" drop down menu that says unroll developable surface. if a surface is not developable then the command will not work.

 

[CET]

Thanks for the replies. I’ve used the Unroll Developable Srf command and it works great as far as I can tell.

I’m also interested in things like cupping (don’t know if this is the correct term) of panels that are twisted to form a planning hull. Not sure whether I can explain this adequately, but I’ll give it a shot. When two panels are “stiched” together to form the bottom of a v-hull, as the twisting becomes more extreme toward the bow, the panels become somewhat “cupped”, or concaved when viewed from above. Here’s an image of a cross section of a hull to serve as an example. Note the slight curvature of the bottom panels. Are there any tools or calculations that can predict this behavior?

 

[CET]

Can anyone offer any insight into the phenomenon I described in my last post? That is, the lateral cupping/bending of the bottom panels of a planing hull as they are bent into position, especially near the bow end. I understand why it happens (the wood is behaving as the section of a cone), but I'm interested in ways to predict it so it can be accounted for when designing/modeling a hull.

I would also be interested to know if there is a commonly used term for this phenomenon, which would allow me to search on it to find more info.

Any help would be greatly appreciated.

Thanks!

 

[Landlubber]

CET, not sure if you realise it but the curvature is induced by the framing of the boat, not necessarily natural forces acting on the bending of the ply, frames can be structural of temp, but they control the amount of bend to suit the design specs

 

[mick_allen]

Can anyone offer any insight into the phenomenon I described in my last post?

If you take a cut across a simply curved surface in any direction other than perpendicular to the curvature plane, you will be cutting across some portion of the original curvature. This apparent ‘phenomenon’ happens in every panelled vessel at just about every station, section, or form cut, but is usually so small as to be not shown or is assumed not to be there.

For prediction, I would imagine the amount is directly related to the angular difference between the section cut and the conic (if present) radial line or surface 0 isobend (don't know the term - 'contour' where bending = 0 ) factored by the amount of curvature or bending (perpendicular to the surface) that the panel has at that averaged (if twist) curvature location. ie the greater the curvature and the more the section cut is aligned with it, the larger the shown curvature – to the maximum of the panel curvature itself.

i/m sure others will chime in with a more mathematically precise answer.

 

[CET]

Thanks for the additional responses. I may be wrong, but I don’t think the curvature I am referring to is a result of any framing. Here’s a photo that may better describe what I’m thinking of. The photo shows the assembled bottom panels of a planning hull that were stitched and glued without use of any frames to influence their shape, or at least with no frames that had any curvature to them. I’ve added a blue line roughly parallel to a scarf joint that shows the curvature I am interested it. It’s this sort of curvature I would like to be able to predict. I would think it is influenced by the factors of plywood thickness, degree of twist applied to the panels, length of the panels, and perhaps other factors as well.

Thanks again,

CET

 

[Landlubber]

one of the main factors of bending plywood is the laminates, if for instance it is 3 ply, then the wood will bend easier the direction of only one ply (naturally)....3 ply is also not marine grade anyhow, it has to be 5 ply plus many other factors to qualify here.....no joint gaps, waterproof glue etc etc

 

[alangluyas]

Conical development of sheet material

CET,

The situation you are describing sounds to me like the conical development you will get anytime you induce twist into a rolled sheet. Any sheet material with conform to what is called a developable surface, whic is essentially a section of either a cylinder or a cone (or combinations of cylinders and cones). When you bend a sheet of ply, alloy or steel around the chines and centreline of most boat shapes, you will induce some twist into the sheet. Most of the twist is where the chines lead up toward the bow, which is why you notice the effect most there. From midships aft, the shape is often more of a cylinder, at least im hulls. If you look at the frame sections superimposed on each other, as they are in the sections plan, then you will notice that the aft sections are more likely to be parallel and the front ones tend to fan apart.

I use several surface development software packages to design hulls, and I normally export the design data into Autocad to finish the drawing. I use Autocad to do a conic development of the hull sections forward to make sure that the shape is "developable". Some shapes just won't go. The process of performing a conic development looks scary to a novice but gets earsier with practise. Autocad makes it much simpler, as the origin points of the cones can be dragged around to find the best origin points.

I normally draw a fair chine in elevation and plan and then adjust the keel profile to suit, while adjusting the origin points of the conic projection lines. Sometime, not matter how much I adjust, things don't line up and I have to move the chine. It is not always a very intuitive process but as I said, it gets easier.

If I haver had to alter the lines in Autocad, I then go back to Prosurf or Rhino and change the lines accordingly so I check the hydrostatics.

What I have found is that performing a conic projection produces a possible conic solution to how the plates will sit, but it is not always the solution that the plates will find. I generally allow a bit of "trim" so that I can scribe the frames/bukheads to the actual hull sections during assembly. On a 22 to 30 foot hull, this is normally only about 20mm.

There may be (certainly are in fact) much smarter people out there who know things that I don't about developing accurate cross sections.

I live to learn.

Cheers

Alan