The longest object I've ever had printed

Hi,

I wanted to print the longest object possible with Shapeways. I though the main limitations were:
- the bounding volume (15x15x15cm for White Strong and Flexible Polished)
- the number of triangles (1,000,000 now, but I remember a time when you could bypass this limit)

I needed something that could be unfold to get bigger than this initial volume, so I choose to design a simple chain.
I first designed a simple link made of two circular loops, intersecting in a perpendicular direction by their sides (two tori to be precise).
I reduced the number of polygons as much as I could without scarifying the shape, and the result was an object with 1176 triangles.
Then I put as much of these links as I could to make a closed loop.

My first attempt was a loop of 824 of those links (969,024 triangles). As you can see, in theory, I could put 26 more links but it was not esy because the chain had to form a closed loop, so the end of the chain had to join the beginning of it, whereas staying into the maximum bounding volume. In other words, I had to fold the chain several times to fit it into this volume and thus the total length had to be a multiple of a certain basic length.
Unfortunately when I uploaded the model, I got an error, saying basically that the Meshmedic program (that check if all is OK with your model) failed to merge the shells. This was quite frustrating because there is no shell to merge: each shell (separate part) is interlocked with two other, but non is intersecting another.

But this means that the number of shell (probably also depending on the number of polygons) is an undocumented limit.

So I reduced the number of shells to 758 then to 692 and finally the 626-link version was uploaded successfully.


So the version available is made of 626 links and has 736,176 triangles, which is not bad at all. Each link add a length of 6 mm so the total length is in theory 3.76 meters. As there is a clearance of 0.5mm the actual length is more that 4 meters (that you have to fold into two since it is a closed loop).

Here is a video of the result (with a bit of me for the scale) :
http://youtu.be/yfSM1TBtBrc

And you, did you manage to print something bigger? In this case how many shells and how many triangles?

cool
its always interesting how designers push the possibility of the 3D-Printing technology.
would be interesting where the problem comes from or where is the limit.
I am not sure, but IMO was in there a blog about folding a chain armor or other clothes so it can be printed? IMO a chain armor has much more shells.

Quaternary tree mobile (level 6) - 1365 shells, 341200 triangles. If there are fewer polygons then it gets through the spurious shell merging more quickly - before it times out and fails the model. Yes, it is very annoying to have to downgrade the resolution of a perfectly good model to get it through the upload process.

Tres cool Vincent

The limit is not the number of shells, but how the number of shells vs the number of triangles affects the ability of MeshMedic to verify the model before it hits a pre-determined time-out limit (20 minutes iirc) - 2034 (or more) shells is possible, however the model has only 73728 triangles, a bit of a bummer really when 3D printed chainmaille clothing is doable.

Paul

One wonders how much longer a chain could be made by using flat faced links instead of toroidal ones. MeshMedic might have less trouble with the quantity of shells, too?

Using the largest print size for WSF, links of two sizes could be made, one in the region of 640mm long the other maybe 8mm long, each size link could be constructed in a similar manner with 24 triangles per link using a triangular cross-section for the 'wire' making the links. How many fits into the printer size of 650x350x550mm, I'll leave up to Magic

Paul

If the 640mm links can be aligned on the horizontal plan of 550mm by letting 8mm from link to link you can put 68 of them horizontally and then make 43 identical layers. This is 68x43 = 2924 links of 640mm plus the same number of 8mm. the length would be 648x2924 = 1'894'752mm, that is a little less than 2km (for only 2924x2x24 = 140'352 triangles). if the density is, say, 20% the volume would be 65x35x55x0.2 = 25'025 cc and the price a little more than $35'000.
Am I right?
Oops, I forgot the discount for large volume WSF...

Mathematics isn't my strongest area, so I'll just agree

It would be an interesting experiment - are there any Guiness records for 3D printed items yet?
(My idea on the world record bit if Shapeways decide to take it up as a publicity stunt)

Paul

Here's some competition on chain length: Hyperform - Longest Chain.

There are two problems with our theory:
- at 64cm long the a link made of a wire of 2 or 3 mm diameter will not be very sturdy.
- the weight of the whole chain would be 25kg. The first link must support all this weight...

I'm thinking the same, not only long, but strong too - I have some chain of 0.7mm WSF 'wire' rings, time for a test.

Paul

For a same quantity of matter (that is for a same weight) a hollow cylinder is sturdier than a solid one. Should we try hollow links?

Here is a hollow link that could be a first try:

It is 192 triangles (it seems that the 1 million triangle limit is not the most important one)
Its external size is 18x12x3 mm with an internal hole of 12x6mm
The wire is 3x3 mm with a hole of 1x1 mm
Each time you add a link you add 12 mm

If we were not limited by the bounded volume (or the number of shells) but only by the number of polygons, we could make a chain of 1000000/192*12/1000 = 62.5 meters. That's much less than our initial calculation because I did not consider the two sizes of links.
How long could this link be made without compromising its sturdiness?

@Magic: nice chain!

@stop4stuff: with regards to 0.7mm WSF links: I don't think it's worth trying that, I had this moebius strip printed in WSF with 1mm links and many of them were broken by the time I got them:

You can reduce it to 112 triangles if you close the holes on the sides:

The side holes were for an easier cleaning of the internal powder. I am wondering if we could have a "U" section, instead of a hollow square section. It would be lighter and even easier to clean.

A little bird told me that this problem of shell limits will soon be fixed.
So, it is time to think bigger (well longer).
So I have designed a new link with less triangles (272 instead of 1176).

If the only limit is the number of triangles (1 million) then I can print 3676 links each on adding 6 to 6.5mm of length, which would give a chain whose total length will be 22 or 23 meters. If I am not wrong the weight would be 128g and the price something like $195...

Just an update: with the latest weekly release, the number of shells accepted at the upload is now higher (thanks Alan!).
This means that I was able to upload my original model (824 links made 1176 triangles each). So, for this model, the limit is now the number of triangles only (less than 1 million). Of course I ordered the Double Loop Chain - 824 today for myself. It should be more than 5 meters long!
And I still have to try something 4 times bigger using the low polygons link.

The Double Loop Chain - 824 arrived. It is a loop measuring twice 2.67m that is 5.34 meters.
Here are two pictures:



So I finally broke my own record.

Who's next?

Instead of a design file describing each individual chain link (with all vertices and faces) is there another file format that can define the link geometry once as an instance that is repeated x times with simple position and angular offsets for every other link? This type of step and repeat is a feature of many engineering software tools to reduce file sizes.