Skip links


I used this photo once before in a different context, but it nicely illustrates the point I want to discuss today. Design in general includes balancing different priorities. The more you emphasize one item, the more you have to compromise on others. The problems the I described years ago as the “Star Trek fallacies” exist in that fiction because design compromise and reality are ignored. Engineering design always has those kinds of trade-offs, and the photo above shows one extreme, where a single priority was taken as far as it can go at the expense of everything else.

The building was a “freight shed” – i.e., a warehouse – in Buffalo, owned by the New York Central Railroad, and photographed around 1900. It’s pretty clearly unheated, which must have been amusing mid-winter in Buffalo, and no building in that era had any cooling other than by ventilation. The clerestory windows don’t look like they operate, although they might and I’m just missing it. The light fixtures hanging from every second truss look to be gas and probably weren’t very bright. In short, “shed” is a good description for this large but minimally-functional building.

The roof structure consists of wood plank supported by tension-rod beams (most clearly seen at the top of the photo); the beams are supported by the roof trusses (double warren or simple lattice, depending on your preferences); the trusses are supported by built-up I-section columns with lattice webs. This is about as light as a steel structure can be made. With the exception of some web plates at the ends of the trusses and tops of the columns, there is no individual piece of steel here bigger than an angle. It’s enough structure to hold up the roof and snow and to keep out the wind, and nothing more. The structure has, in my opinion, been optimized for minimum weight of steel material at the expense of all other considerations.

The first question: why would the designer do that? In 1900, the cost of steel material was more significant than the cost of labor in fabrication and erection. Over the course of the 1900s, the price of metal became less important and the price of labor became more important, so the cost of fabricating all those little pieces of steel and riveting them all together became prohibitive, while the cost of simply using wide-flange beams (much more metal, much less fabrication) dropped.

The second question: why do I take exception to this design, which I find visually stunning? Those very slender trusses and columns have vanishingly small lateral stiffness and strength, and so have little resilience. Any problem that results in unexpected loading – for example, the loss of some of the roof sheathing in a storm – and the trusses will fail sideways like dominos. Those slender pieces of steel have less fire-resistance: small pieces have a larger surface-area-to-volume ratio than large pieces, and so absorb heat faster. And finally, a heavier roof structure could increase the span between the columns, opening up more floor area.