tortured plywood unfold software

Discussion in 'Software' started by jpquattro, Apr 29, 2012.

1. Joined: Oct 2009
Posts: 29
Likes: 0, Points: 0, Legacy Rep: 10
Location: Pisa Italy

jpquattroJunior Member

I have written my first program in 1972... I can do it...
I have searched the forums, I see that there is some interest on the subject, but no free software...
I am looking for suggestions and comments...

I want a program that:
-get the hull's coordinates
-make a "raw" unfold of the panel
-redistribute the stretch taking into account the E of the material
-evaluate the amount of stretch and return a warning if the amount of sheet deformation exceeds some given values...

to obtain this I will write a C program, that:
-read a coordinates file, representing the half hull,
-place on it a triangular mesh,
-make one raw unfold (that is: divide the hull in stripes unfold it and assemble the strips)
-run a relaxation algorithm that must smoot the stresess and deformations.
-analize and print the result, with warnings where the deformation exceeds the limits (maybe 1% in extension an 5% in compressions?)

The question is:
Is the approach reasonable?

Paolo

2. Joined: Oct 2009
Posts: 5,030
Likes: 515, Points: 113, Legacy Rep: 1485
Location: Midcoast Maine

DCockeySenior Member

Plywood does not have a single "E" value. The stress-strain properties of plywood depend on the direction the load is applied which means it is not an isotropic material. To approximate plywood as isotropic will introduce an error, thought the error may be small compared to errors from other sources.

What will be used for the hull coordinates? What will the spacing of the coordinates be? A typical set of offsets are too widely space to provide a reasonable approximation of a hull surface with triangles using only the offsets. Fitting a smooth surface to the offsets would be needed to obtain points in between the offsets so that the mesh can be refined.

The shape of the "unfolded" strips will vary depending on the orientation selected for the strips on the surface. The edges of the strips will not match so deformation of the strips will be required to assemble the strips. The choices made on how to deform the strips will change the shape of the assembled strips.

Try it and see but I will be very surprised if it works for arbitrary tortured shapes. It is probably possible to make it "work" for one test case when the answer is known in advance.

3. Joined: Oct 2009
Posts: 29
Likes: 0, Points: 0, Legacy Rep: 10
Location: Pisa Italy

jpquattroJunior Member

I know. Actually I have inserted one E value for fiber direction and another for cross fiber direction.
Do you think that I must use different values also for extension and compression?
The real problem can be... Where to find reliable value for E? Do you have a source?(metric please...)
The source will be the output of free hull design programs, like Hullform or Delftship. I can specify the number of points.
But I agree, a "best fit" spline will be a second step of refinement.

I will allign the strips from gunwale down, but, taking into account the building scheme, the reference line should be the keel (the mass of glue there make that line almost indeformable) your opinion?

I dont expect that work for arbitrary case... I try to make it reasonably reliable for simple compound curvature multihull panels...

Any constructive criticism is help for me.

Paolo

4. Joined: Oct 2009
Posts: 5,030
Likes: 515, Points: 113, Legacy Rep: 1485
Location: Midcoast Maine

DCockeySenior Member

It is more complicated than just two "E" values. There is also interaction between tension and shear to be considered. If you pull on a sheet of plywood at angle to the fibers the panel distort in shear. A thin laminate of orthotropic layers would be a reasonable approximation and that requires five coefficients to characterize. But I don't think that will be your major source of error.

Why do you need "E"? "E" along with Poisson's ratio provides the link between deformation and stresses. You appear to be working only with the deformation, not the stresses so it's not clear where "E" will be used.

Why do you think that is the correct way to align the strips? You could also align the strips perpendicular to the centerline. The results will probably be different. Is one correct and the other wrong?

Try a developable surface first for which you know the unrolled shape.

5. Joined: Oct 2009
Posts: 29
Likes: 0, Points: 0, Legacy Rep: 10
Location: Pisa Italy

jpquattroJunior Member

I want to disregard some aspects, in fact thickness is small with respect the other hull dimensions.
I plan to simulate ply just like a thin sheet without thickness.

After unfolding I plan to run one relaxation algorithm on the panel, to equate stresses.
I try to mimic the building proces, in reverse, I cut the panels, stich and glue it at the correct angle at the keel, at this point I have a V shaped hull, now I "torture" it to the corret hull shape.
If I follow this sequence in reverse, the reference line is the keel, because of glue thickness, it is almost impossible to strech it.
But is difficult to mimic this sequence in code, then i follows the opposite direction, use gunwale as reference, it is simple and the relaxation algoritm should correct the panel shape, producing a similar result... at least in my mind....
I have this one... do you tink that is good enough?

Paolo

6. Joined: Aug 2003
Posts: 275
Likes: 11, Points: 18, Legacy Rep: 45
Location: Edinburgh

AndySenior Member

Hi Paolo, did you finish this? I would just add a couple of stretch factors to the software that users can set themselves, perhaps after doing some basic testing on the ply they intend to use. You could could have default stretch factors for those who don't want to experiment. I think the software is a great idea, and it will only improve with use. Do you have something to show us yet?

7. Joined: Oct 2009
Posts: 29
Likes: 0, Points: 0, Legacy Rep: 10
Location: Pisa Italy

jpquattroJunior Member

Hi Andy.
I have suspended the work because I am busy with some other duties.
Up to now I have done the following:
Taken the output of hullorm, after some manual editing, the program can read the 3d coordinates of points uniformly spaced on hull.
The program flattens the grid and after iterates to balance the deformations and stresses.
The problem arrives here, sometimes the iteration arrives to some convergence, but usually don't...
I have introduced different refinement to iteration to improve the convergence, but the reuslt aren't satisfactory up to now.

I used as input patterns with many points, many more than the usual, to reduce the size of grid's elements, but no satisfactory results.
Any way the problem is the program itself, non the representation of ply or something related to real process.

Paolo

Forum posts represent the experience, opinion, and view of individual users. Boat Design Net does not necessarily endorse nor share the view of each individual post.
When making potentially dangerous or financial decisions, always employ and consult appropriate professionals. Your circumstances or experience may be different.