Incompatibility: Curtain Walls

by Don Friedman on October 11, 2016


The 1918 Hallidie Building in San Francisco, from the HABS survey.

I’ve studied extensively the early years of skeleton-framed buildings – the 1890s, 1900s, and 1910s – which inevitably means that I’ve studied early curtain walls. Curtain walls are defined in the negative: they are exterior walls that do not carry structural load. They can be heavy masonry (and were in the early years) or they can be thin sheets of glass and metal. Both are capable of performing the non-structural function of an exterior wall: providing separation between the indoor and outdoor climates.

If we think about the curtain wall as a whole over an entire facade, the difference in thickness between an old masonry wall and new glass one is not very important. Looking at a wall as a whole, what we have structurally is a building element that is not very strong for load in any direction but is quite stiff when loaded in plane. As a reminder, strength is a member’s capacity to carry load and stiffness is how much the member moves when loaded. The relation between the two is complicated to the point where it’s easiest to say there’s no correlation unless you are comparing only members of the same material and general geometric properties. If you look only at rolled steel beam sections, the correlation between strength and stiffness is pretty good. If you try to compare rolled steel sections to plywood sheets, the correlation is terrible. So if we look at curtain walls on a large scale, they are generally weak (although that varies widely with material), they are very flexible when subjected to out-of-plane forces, and they are quite stiff the subjected to in-plane forces.

Why does this matter? Because structural frames move sideways under wind load and vertically as interior live moves and snow arrives and goes away. Or more to the point, structural frames are often more flexible than the curtain walls that surround them. If we imagine a situation where a building’s curtain walls are subjected to little thermal movement (a tropical island with high temperatures in the 80s and lows in the 60s; a building shaded from direct sunlight by a nearby cliff) we still need expansion joints in the facade because the facade and the frame are incompatible. The facade is stiff and the frame is strong, and that’s a recipe for facade failure. If there are no expansion joints, the walls will tend to carry load they are not supposed to, until they crack and release the load back to the frame.

The oldest skeleton-frame buildings have curtain walls that tend to perform okay, despite having no expansion joints, because the heavy masonry walls are strong enough to carry the load that is accidentally imposed on them. But this is good performance by luck – not a recommended engineering strategy – and there have been notable failures. The newest skeleton-frame buildings have curtain walls that perform okay for structural purposes because they have a lot of expansion joints. It’s the middle range of buildings – 1930 to 1970 – where the walls are too thin to work as accidental structure but lack enough expansion joints. Incompatibility of movement eats away at them, causing crack after crack, spall after spall.

Previous post:

Next post: