The skeleton dance

17 September 2009, from Housetruck

One of the most complicated parts of the housetruck is the interior frame. This is as it should be, as the frame is the skeleton that supports the structure – the floor, walls, and roof. The housetruck is a small structure, but because it is designed to move, it is like designing a house to withstand the equivalent of earthquakes and hurricanes as a climatic norm. But the structure also needs to be light enough to be driven by a reasonably-powered vehicle.

A while back I’d discovered torsion boxes, in which a very strong panel is created from a grid of lightweight members, faced with sheets of equally lightweight panels. The combination, if designed properly and glued well, has far more strength than the individual pieces. It’s the same principle behind foamcore, an incredibly strong material belying its basic ingredients of paper and styrofoam.

I figured that if I could somehow make wall and floor panels as torsion boxes, I’d have a building system that would function as a combination of frame, insulation (by using ‘blueboard,’ a popular rigid building insulation), and perhaps both interior and exterior walls. It turns out, of course, I wasn’t the first person to invent this: there is an industry producing structural insulated panels. However, the more I researched the materials, the less enthused I became: to avoid exposure to off-gassing, I didn’t want plywood or OSB on the interior; the synthetic foam is both polluting and requires a greater thickness than I wanted to gain the needed insulation factor; and engineering my home-made SIPs seemed a mysterious art that even my architect friends couldn’t help with.

I believe it was Christiaan, a friend who is a kinetic sculpturist, who suggested I look into building in steel. This was reinforced by Ignacio, whose architectural experience showed him that steel was strong, lightweight, and relatively easy to use. I sketched out a minimal steel frame – basically, a post-and-beam structure – to give the basic shape and strength to the structure, but still felt it necessary to build minimal SIPs to fill in the void. This seemed a reasonable and buildable design, so I started looking for a steel fabricator.

I didn’t have to go very far: I was pleased to learn of a skilled fabricator right here in Silverton: Byron Miller, of Miller Industries. It turned out Byron had built many structures I knew about around town, from stairways and railings to ceramic kilns and even small bridges. Everyone who mentioned him to me said he did great work.

By this time I’d built a foamcore model of my house (in its former design of a trailer, not a housetruck). I showed the model to Byron and his eyes lit up – this was clearly something he thought would be fun to work on. As I explained my idea for the frame, he immediately suggested that there was a far easier way to do it: a simple system of vertical posts every three or four feet, plus horizontal beams dividing each area, much like a series of H’s. Instead of individual modules, the entire frame would be welded together, a monolithic monocoque network of 2” square-tube steel.

Byron explained that the steel itself, if welded well, should provide the needed strength. To make sure, I could attach shear panels of plywood on the outside of the frame. With this basic structure, the structure would have no trouble bouncing down a bumpy road or rocketing down a freeway.

Now knowing the materials and construction of my housetruck’s skeleton, I spent another couple of months tuning the design: placing windows and door, lengthening then shortening the overall size, and finally, what turned out to be the most difficult part to conceptualize: designing a skein of struts that supported the curved, stacked, ‘mollycroft’ roof. I am by no means a structural engineer, but I was pleased to discover that I have an intuitive, if primitive, feel of weights and stresses. That, coupled with an attempt to always simplify, led to a slow but certain development of the skeleton.

Most of the design work was done in SketchUp, a wonderful (and free!) 3D design program. Once I learned the basics of SketchUp, I was amazed at how fast I could incrementally design the house. I was constantly zooming in and out from the model, flying around it, trying out views from different vantages and angles. A simple, solid wall would eventually be divided into the core faced with interior and exterior skins; then the core would be made into 2” square tubes. Different parts of the structure were put on different layers, so it was easy to see only the frame, and then with a few clicks, see the full house complete with windows and furniture.

One day I found myself saying, ‘I think I’m done.’ The structure was the right size, the windows were placed where they seemed best, all furniture and systems were accounted for. The roof support was both simple enough and effective, and it was clear how the overall structure would fit on top of the chassis of the Isuzu NPR truck that was sitting in a friend’s field, waiting for the house.

I converted the 3D shapes to 2D mechanical drawings with dimensions (not an easy task, it turned out), and took it over to Byron, nervous at what he would say. ‘I’ve never seen such a detailed specification for such a small project’ was his first reaction. ‘This’ll work’ was his second. Apparently my naive engineering had been accurate, and Byron didn’t see much that needed to be changed.

I still had an uneasy feeling about the scale. After all, I’d been creating this shape on a computer, not in the physical world. I decided to pay a visit to the truck and try to mock up the design. With measuring tape and an old scrap of plywood, I measured and propped and squinted… and worried. The whole structure seemed just too wide, too tall: the overall height from the ground would be nearly eleven feet! What had I done? I went home feeling dejected.

I glanced at specs for other RVs – surprisingly similar. I re-checked my assumptions about physical dimensions – the same. I considered, briefly, whether the mollycroft roof was really that important – it was. Finally, almost just to make myself feel better, I shaved a few inches off the width, a few more off the height of the main roof, and then for good measure increased the height of the mollycroft to ensure reasonable windows. Overall, I adjusted by no more four inches: but I was pleased. (A few days later, I met someone who suggested that buildings and other structures should be made slightly too small, to encourage a childlike feeling when within.)

Finally, I delivered my revised drawings to Byron, and let it go.