Tails for the Serpent Twins

For this year's Burning Man, we threw in with our good friends Jon & Kystren of Form and Reform . You many know them from their awesome 'how to make a snail car' Instructable they made a while back.

Jon & Kyrsten had a big, big idea : to make a pair of 50' long art cars, two serpents, one light and one dark, based upon ancient mythology with a 'sky viking' theme and full of programmable LEDs. How could one resist helping out with something so awesome?

Anyways, one big thing we helped with was the tails. While most of the Serpents were made via Jon's master blacksmith handmade methods, the tails were fully digitally fabricated. That means made by robots. Well, cut out by robots, and assembled by us!

So in this instructable, we're going to go through step-by-step how we did that, and dive into the deep end of digital fabrication.

Step 1: Figuring Out the Why and What

It all started on paper. Meeting with Jon, we sketched and talked until we got a good idea of what the tail should be. We made some early 3D models, then printed those out and sketched over the top of them. We looked up pictures of classic viking and art deco structures. Back and forth until the idea was solid.

It's important early on to figure out what your goals are for a given project, because it's easy to loose your way when you get into the thick of solving problems.

For example:
- We needed to be certain that the tails were under a certain height, so they could fit into a standard truck.
- Each serpent had to have a generator and a charger fit into the base of the tail.
- We wanted it to be all made of steel so it matched the rest of the Serpents.
- We needed to make them very quickly, for the deadline was really tight.
- It had to fit with the 'sky viking' theme we were going for with everything else.
- Finally, we figured it needed to be able to survive a friendly-yet-aggressive hug, which is likely to happen at Burning Man.

Step 2: Modeling the Initial Form - Drawing the Curves

We took a first pass at modeling up the form. We use Autodesk Revit for most of our projects, and it's got a great toolset for making organic forms like this. While it's intended for big, Frank-Ghery-esque curvy buildings, we've never been above using technology in the wrong way for the right reasons.

If you want to follow along, you can download the free Autodesk Vasari and jump in!

Here's how it went down: I started a new Conceptual Mass Family. In Vasari, this is the default mode, just hit 'new mass'.

First I drew the top and bottom curve of the tail when viewed from the side like so.

Then I drew a series of half circles between those two curves at fairly even sections along the two curves. We only need to model one side of it for now, for it's mirrored on the other side.

Select the circles, and then the top and bottom curves, and hit 'Make Form'. This gives you what's called a 'loft' in the modeling world!

Step 3: Modeling the Initial Form - Getting It Right

Now that I had the overall form, I could add more cross sections to it so that I could get better control of the shape. We needed it to curve and taper towards the end in a nicer way than the initial loft was. While we could have drawn a bunch of curves first, and then lofted exactly what we wanted the first time, I've always found it faster to simply model the general shape and then tweak it into what I'm looking for instead.

So I selected my form, and hit X-ray.

I then used the 'Add Profile' tool to add more curves to my form.

This gives me handles I can pull and push to make the form exactly what I want.

We would print out the form onto paper and meet with Jon. Drawing directly over the print-outs, then going back and editing the model, allowed us to suss out the final form.

Step 4: Planning the Fabrication

Once we got the design down, we needed to plan out how we were going to actually make this thing! We didn't have a lot of time, and we wanted it to be made out of steel. Both to match the rest of the serpent and to be very durable. Our in-house CNC machine (also known as 'Frank') can't cut steel, so we had to figure something else out.

Van Bebber the steel supplier has a CNC plasma cutter . Since that is where Jon gets his steel, it was one stop shopping. A CNC plasma cutter is a tool that takes 2D CAD drawings and cuts them out of steel sheets.Much like our CNC routing table, but with a plasma cutter instead. It's pretty accurate, not as good as a laser or waterjet , but good enough for our tails. Van Bebber also have a Press Break , a big industrial tool for bending sheet metal parts.

This one-two combo of Press Break and CNC Plasma means that we can model our parts on the computer, unfold them flat using special software, then send them off to the steel supplier. Then they cut out the flat parts from sheet metal using the CNC plasma cutter, and then bend those flat parts into their final shape with the Press Break.

It's pretty easy to weld steel, so we opted to have the internal frame slot together and then be welded. Once the frame was assembled, we'll bolt the skins onto that frame, using 'weld nuts' to secure them. Since the skin panels are going to be a thinner sheet metal they were able to just bend to the frame as we bolted them on. The thicker internal frame's tabs and parts could only be folded by the Press Break.

So there was our strategy for getting this done: we'd make the tails out of sheet metal, we'd make it so that all the parts could be made by the CNC Plasma & Press Break combo, then we'd assemble together the parts, weld things into place, and be off to the races!

Step 5: Rationalizing the Form Part One: How the Heck Are We Gonna Make This Thing?!?

Once we got the final shape we wanted, and had a plan for how we'd make the parts, we needed to 'rationalize' it. That's where our design model is turned into a "fabrication model". While design models capture the 'design intent' or what we basically want it to look like, the fabrication model has every nut and bolt and detail needed to actually make the thing.

While it's always good to have in the back of your mind how you're actually going to make the thing your designing, we've found that to focus on that too soon or too much can really stunt how good of an idea you're able to come up with. You don't worry too much about paragraph structure when coming up with your big idea for a novel, right? So we don't worry too much about some of these more specific details when we're designing. Because we have a rough idea along the way of how it could work, and we trust in our abilities to figure out the finer details later!

The internal frame was easy, that would just be steel parts that would get folded into shape and then slot together. We knew that we could model those parts up in Autodesk Inventor, which has a whole toolset for doing stuff like that.

But the skin of the tail was another issue. If we wanted to make it look exactly like our curvy model, we'd have to hammer out the sheet metal or use an English wheel to roll out the sheet metal because it has a 'double-curved' surface. It not only curves up the length of the tail, it's also curving in profile along the way. It's 'dished'. Convex. Very hard shapes to make in metal. They require a huge amount of time that we didn't have. We could theoretically press the sheet metal into shapes like this, but not without a very large and expensive press that only a car manufacturer would own.

We needed to figure out how to make that skin out of flat pieces that could curve in one direction, and even twist somewhat, but wouldn't be dished out or domed.

So we turned to boat hull building techniques. There is a long tradition of making complex forms like this out of flat panels in the boat-building world. By turning the form into a series of curving, twisting, yet still 'flat' panels, we could easily make something this complex from simple sheet metal. Using Carvels we could make the tail the form we wanted, and even make it look 'sky viking' like. Bonus!

Step 6: Rationalizing the Form Part Two: Electric Boogaloo

So we took our original form, and we sub-divided it into equal segments. We did this by selecting our original form, and then using the 'Add Line' tool to add surface lines that ran all the way up the form and divided it into equal panels.

Using that 3D line drawn on the surface of our form, we could then create what's called a two-rail sweep. This gives us a surface that might twist and curve, but is still only curving in one direction, for it's flat between those two curves. We picked two of the curved surface lines at a time, and hit 'Make Form'.

Solving problems like this are what we really, really enjoy about making things. It's a great combination of ingenuity, creativity, and experience that's very addicting when you solve the problem and make things work. It's also the 'hard part' of this, where reaching out to a pro or wise adviser is a great idea if you're just getting started with making things this way.

Then we hid the original form in our view, showing us the final form that could be made from flat sheets of metal!

Now we just needed to 'flatten' the skin down into flat outlines for cutting. This turned out to be very tricky to do properly. We tried a number of things, and in the end wound up using Rhino to do the job. Sadly this was because Revit can't unfold things, Inventor isn't so good at unfolding complex shapes like this, and the nice plugins that can do such work within Inventor are priced out of the hands of mere mortals like us!

Step 7: Planning the Internal Structure & Prototyping!

We needed to make the internal frame strong enough to stand up to the rigors of Burning Man and to give the skin something to attach to. We quickly mocked up some ribs and bulkheads within the form in Revit.

We did this by making a new Generic Revit family, and then nesting our skin model into that. The 'traditional' family editor in Revit is a little easier when it comes to solid-style modeling like this, or when dealing with assemblies of parts. We tend to work in this way often anyway, so it is faster for us than using the more complex mass editing.

One thing Revit is awesome at is making really smart models. I was able to make a parametric 'hole' family, just a cylinder really, that represented where a bolt hole needed to be. It would remain 'true' or 'normal' to the surface, even as I dragged it around, so I could place bolt holes evenly on the complex skin panels really easily. I could then join that 'hole' to the skin, and get perfect holes in a complex surface just where I wanted them easily.

Once we had what we were now calling the 'bulkheads' modeled, we made some prototypes! We try to make them as soon as we can so that we can test our assumptions and ideas as early as possible. It's very easy to make 1/2 or 1/3 scale prototypes when working with digital fabrication. The prototype really helps to bring up problems early, when there is still time to do something about them. We cut the bulkheads out of plywood, slotted them together, then cut the skin from a flexible PVC plastic sheet and just screwed it to the frame with small wood screws.

After looking at the prototype, we tweaked the final model just a little bit, and it was ready for the next step...

Step 8: Modeling the Final Parts

So while we love Autodesk Revit, it's not so hot when it comes to modeling folded sheet metal parts. Sometimes it's better to use a more specific tool for the final fabrication model, for it's got features that can save you a lot of time. In this case, Autodesk Inventor was a better match. It understands sheet metal very well, and has a whole toolset for unfolding parts. So we exported a .SAT solids file our of Revit and into Inventor.

From that model, we were easily able to turn the bulkheads and frame into ready-to-fabricate steel parts. Now, I'm a real novice when it comes to Inventor, so you really should learn it from someone else. But basically what I did was select a face of a imported bulkhead, and then create a new sketch from it.

Then, using the Sheet Metal toolset, I added flanges by clicking on the outside edges. I added holes to the flanges by projecting sketches off of the imported Revit cylinders I'd modeled earlier.

I've included the Inventor file here, but I had to leave out the skin because it was too big to upload to instructables. Also, this is the first time I've used Inventor, so... please don't look to this file as an example of the 'right way' to do this.

Step 9: Isolating the Parts

Once our fabrication model was 100% done, it was time to get it made! And to do that, we have to isolate the parts. Isolation is the process where we take our fabrication model and separate out all the parts and pieces and get them exported out into the right formats for fabrication.

Here's the 'folded' and 'unfolded' versions of a single bulkhead.

In this case, we needed to export our final parts out as 2D DXF/DWG files, and 2004 format ones at that. That's what the CNC plasma torch that Van Bebber has understands, so that's what we had to run with. One layer for the lines that were to be cuts, and another for the lines that marked where things needed to be folded.

So first we unfolded all the simpler interior frame sheet metal parts in Inventor. Then we selected the face of our part, and picked 'Export Face for Fabrication'. This was terribly tedious, as you can imagine, for we had many, many parts. We hope this gets to be as automated in the future as it can be in Revit today.

Step 10: Checking and Cleaning Up the Parts

Once we exported all the parts, we had to check them all. You always, always want to double-check the exports. Software is tricky and can smell fear, and will screw you over when given the chance. We also had to separate out the different exports into different groups, since there were three different thicknesses of metal we were using in the design.

So we imported the files into our CAM software and laid out all the parts next to each other so we're certain we have them all, they didn't get scaled weirdly when exported, and they all look like they are supposed to. We could have used Revit, Inventor, AutoCAD, or even Illustrator / Inkscape for this instead. Just as long as it's a vector-based program it should work just fine.

One big reason we used Vectric Aspire for this was that exports from any modeling software can have a problem: Too Many Nodes. While a circle drawn in CAD can be made up of twenty tiny arcs and look OK, the Plasma torch doesn't cut out such things very well for it's just too much information for it to deal with.

Vectric Aspire has some nice features to reduce the number of nodes in a curve while keeping the curve the same! AutoCAD does too, and maybe Inkscape does as well, but I haven't used either in years and years and don't remember how to do it. But again, either would work just fine in this case.

So after 'cleaning up' the exports from Inventor, and being certain that each part was grouped into the right batch for the thickness it needed to be, it's off to the races: we e-mailed the files over to Van Bebber and they got to cutting.

Step 11: Fabrication!

Van Bebber cut out all our parts, and they showed up about two days later.

If we'd cut the parts in-house, here's how it would have gone down: We would have imported the 2D DXFs into our CAM software. CAM stands for Computer Aided Manufacturing, it's the 2nd half of the 'CADCAM' of old. CAM software is what sets up the jobs for the machine to run to make the parts you want. It's the missing link that lots of people who haven't worked this way don't know about. You import in your parts, arrange them how you want them cut out of the sheet, and tell the machine how you want them cut out and in what order.

Your fabricator already has a favorite CAM software they use. In this case, that's why we had to export our files as 2D DXFs, for that's what works with their CAM software. If we were working with an online service like Ponoko.com, instead we'd need to research what formats they use, and might upload PDFs or 3D models depending upon what we were making. Point is, do your homework, and if you have time, even run a test to be certain things work the way you think they do.

Our favorite CAM software we use with our Shopbot is Vectric Aspire . It's not only got some awesome features for CNC routing, like automated part nesting and joint detailing, it's totally affordable and made by a great group of guys.

Step 12: Assembly!

Van Bebber did an awesome job! They reached out to us, asked questions, and made certain that things were working. So two days later, a stack of parts arrived.

Now, one fun thing about working with digital fabrication is when properly done, everything just fits together like a big puzzle. We didn't even make shop drawings, we just printed out a screenshot of the 3D Inventor model and used that as a guide to put it together!

First we slotted together the spines and bulkheads, then we bolted on two of the outside skins. This pulled the form square. With digital fabrication, you can make things that 'want to be built' and even help you along the way. By bolting on two of the skins first, it auto-aligned all the bulkheads for us, for they can only fit together one way.

After bolting on the two outside skins we tack-welded the frame together.

Then we 'skinned' one side of the tail entirely. We used weld nuts, and so by skinning one side of the tail, we could then weld the flanges of the weld nuts to the flanges of the bulkheads, making so that we could take the skins on and off as needed in the future.

It was a lot of bolts!

Step 13: Mounting Them Onto Trailers

With the tails mostly assembled, it was time to get them rolling!

First, we mounted the generator and the charger in the tail so that they could be secured down to the wooden 'plank' we made to go inside. One fun thing about digital fabrication is because all the parts come from the same model, they can be made by totally different people and machines yet still just bolt together. The wooden 'plank' we cut out on our Shopbot, but the holes lined up perfectly with the holes in the sheet metal. Genius!

Under the barrels that made up the body of the serpent were these little custom-made trailers. The trailers had a 'spine' of 2" tube steel with straps that the barrels bolted onto, an axle for the wheels, a trailer ball for the barrel behind it, and a little trailer hitch to link up to the barrel in front of it. We needed to make a trailer for the tails to ride on, but to do that we needed it to balance just right: not too back-heavy, put enough pressure forward to stay linked up, but no so front heavy to push the trailer down in front of it.

Once the heavy things, the generator and charger, were mounted we were able to find a nice balance point for the tail. That's me balancing the tail on a dowel on the floor. This helped us locate where the axle for the trailer it was going to be mounted to should go.

Once we had that figured out, we welded together the trailer, then bolted the tails onto them.

Step 14: Making the End Fork

We really felt that the very end of the tail, the forked tip, should fully match the heads. We were also worried that making it the same way as the rest of the tail wouldn't be the right look, that it would be too flat. Because it was a smaller piece, Jon, the master blacksmith that he is, could hammer and roll it out, so it could be double-curved and more complex.

So we took the surface of our model for the little tail ends, flattened it out, and had them plasma-cut the shapes. Jon then took those two halves, dished them, and then welded them together with small joining plates.

A 'seam' of thick hammered steel completed the look, and gave us an easy mount point to take them off later on if desired. They just came out great!

Step 15: Finishing the Metal

Now that everything was together, the last step was to give everything a nice finish to match the Serpent heads. One was a light serpent, which was easy: we sanded the sheet metal with 'dual-action' sanders until nice and shiny, and then had them clear-coated. The other serpent was 'dark', which after flaming to prep the surface got a nice metal patina to darken it. It's kind of like gun-bluing, but Jon doesn't share his secret formulas, so that's all we know about it.

We already used stainless steel hardware in the light serpent tail, and black-luster hardware in the dark, so the bolts would match the surface.

Here's one of the heads getting finished to give you a good idea of what it looks like.

Step 16: Rolling!

Here's the finished tail! And here's that great video again, showing what they look like rolling around:

So honored to be included in this project, and very happy with the end results.

Riding around on these amazing, otherworldly, and ridiculous things was a real high point of our lives and made us thankful that we're lucky to have the opportunity to do stuff like this.

A huge thanks to Jon and Kystren and the Crew for including us!