Rhino Challange Question

[willnewton]

OK, here is the ninth panel you need.

Slice the blimp in half and add one panel cut from a rectangle folded in half at the nose. No tapers, no curves, as long as you bisect.

Also, you might consider a folded isosceles triangle panel for a slenderized tail and you would not have to mod your current tail design.
--- If you generate the triangle from Sweep2rails using a Line across the nose and opening the blimp with a 5º taper from the tail, then you can use UnrollSrf to lay the panel flat. You can check for curvature by using a line snapped to the long unrolled point to see the subtle curve created down the long side of the triangle, but this may no be an issue for your application.

These drawings show half of the new panel removed to give you the idea.

[pyronaught]

The 9th panel to be inserted is identical to all the other panels. The problem was how to generate the new blimp shape once the additional panel gets added, not adding a panel that allows keeping the original curvature. Blimps, balloons and anything inflatable must be symmetrical around the long axis too. The blimp you designed where it is wider on the horizontal axis than it is on the vertical axis would not inflate and hold that shape under pressure. The flatter areas on top and bottom would deform to produce a circular hull, creating undesirable wrinkles in the process.

[pyronaught]

What I'm trying to do is take one of the 2D panels already generated for a blimp designed to use 8 panels and expand the actual blimp being made by sewing in a 9th panel.

. I understand what you are up to now and love the blimp. The seam geometry for 9 panels would change, especially toward the tips. It is a subtle change, but will happen nonetheless.

I do have an idea to add a single panel with no math involved. Gimme a minute to draft it up.

The seam geometry does change, so the challenge is to generate the new geometry. On the actual blimp, the seams are sewn flat, so they will flex to accommodate the new geometry. It's simulating that geometry in Rhino that is the pain.

[jsc]

Why are you trying to go from panel shapes to a 3D model, isn't it always easier to go from the model to panels? Or do you already have a bunch of cut shapes and want to work with those?

I did a little reading, and it turns out (as you intimated above somewhere) that spheres, ellipsoids, and blimps are non-developable surfaces (https://en.m.wikipedia.org/wiki/Developable_surface): they cannot be unrolled onto a plane without warping.

If you treat each panel as a curved longitudinal facet, then the circumference of the shape at each point is 8x (or 9x, depending on how many panels you want) the width of the panel. To get a curve to revolve to recover the shape, you should be able to
1. Slice a panel shape down the major axis to get a half panel.
2. Scale it along the minor axis by 9 / pi. (circumference = 9 * 2 f(x), diameter = circumference / pi, radius = diameter/2).

Rereading your previous post, I gather that's effectively what you did....

[pyronaught]

The exact procedure I use is to take the nose to tail curve (only for the top half, not all the way around) and create a copy of it that is rotated the number of degrees each panel requires. So for an 8 panel blimp that would be 45 degrees. I then do a Sweep2 command using these two curves as the rails, with a straight line between the two curves used as the cross section. This results in a developable surface that can be unrolled without distortion.

And yes the normal procedure is to start with the 3D model and then produce the panels off of it rather than the other way around. The problem is that I came up short on lift and need more volume, and in order not to have to waste the fabric I've already used on the undersized blimp I can just sew an additional panel into it to increase the volume while retaining the same basic shape. This also lets me keep the same G-code files for my fabric cutter, which are kind of a pain to make. The only problem is that the new blimp is then unhinged from the CAD model so I have to work backwards to try and produce a new model that represents the nine panel hull. What I did was repeatedly make polar arrays of 9 panels, adjusting the distance from the center point each time until I finally got the widest point of the panels to just touch. This produced the new diameter, then I went in and created bulkheads at various points using cross sections of the original curved panels (which are still pretty close to the new curvature) at each point to calculate the diameter of a circle at each bulkhead. A 9 sided polygon was fitted to each circle for the final bulkhead markers. Curve fitting could then be used to connect the bulkheads along their edges to get the new curve profile. I used the procedure above to create and unroll a panel from the new curve and then overlaid it on top of the original panel to make sure the two were the same as a method of verifying if the new geometry was valid. It was not 100% accurate, but it was closer than my tolerance at sewing is so in a practical sense it worked.

Normally I don't make this kind of blimp design screwup, but I broke the golden rule of ALWAYS building your hardware to completion FIRST before designing the blimp envelope. There's just no way to estimate the final weight of all the hardware without actually building it. But then some of the hardware is dependent on the hull size too so there is this chicken before the egg scenario that plagues blimp design. You wind up having to make a best guess on the hull, then design the hardware and build it, then go back and adjust the hull for the final hardware weight. This blimp in the pictures is a high speed torpedo design which needs the smallest possible volume to get the job done so there was no room for error. I got within 200g though.

[Jules]

Impressive amount of engineering involved in one of those things.......thas some mad skillz!

[jsc]

Pyro, you should try my geometric construction and see if it is at all correct

A panel outline scaled the short way by 9/pi (with 9 being the number of desired panels) should give you the curve of revolution you're looking for. I think.

[pyronaught]

Turns out I did have to glue these together. Pressure from the thread causes them to pop apart when winding them. Weird thing about this print... just doing one set takes 12 minutes, while doing 36 sets takes 3.5 hours. Note that 12 x 36 is 7.2 hours! I don't see how print speed could double, so more than likely it is Simplify 3D doing inaccurate time estimates on both the single and batch print. I've noticed S3D time estimates are not very accurate. It would be nice if S3D put a print timer on the print page so you could more easily see the actual time each print takes and compare it to the estimate.

[jsc]

Doing one probably slows down for cooling. In PLA you can reduce cooling time per layer down to ~5s without any real harm if you have the fan on.

As for time estimation, try out my service:
viewtopic.php?f=9&t=2785

[pyronaught]

They are printed in ABS though, so I don't think it slows down for cooling.