Last night was the (virtual) ceremony for the 2021 Lucy Moses Awards. As I have mentioned before, we were the engineer of record for two winning projects, and had a small role on a third. Obviously, it’s gratifying to be associated with winning projects and to be in the company of so much good work by so many talented people.
The award for the Serbian Orthodox Cathedral of St. Sava – a project I’ve discussed here at some length – got me thinking about what structural restoration means in the context of changing requirements for safety. I first worked on the cathedral about 25 years ago, designing repairs to the wood roof trusses. The repairs mostly used steel plates to supplement wood that had lost significant amounts of material to rot. (The steel was painted black, on the theory that it would blend into the background in the unlit upper reaches of the sanctuary, and that seemed to have worked.) There was a problem then that had a answer that was simple in execution but complicated philosophically: the trusses we were repairing didn’t make sense. There was no ceiling, so the trusses were exposed and decorative; they had what looked like a hammerbeam arrangement, but the probably did not function as one because of the long unsupported length of upper chord directly below the roof. In short, there was no way to analyze the old wood-truss roof that would show it to be adequate for anything like code levels of wind load. (Because the roof is very steep, snow load is not a problem, but that means it is more exposed to the wind.) Of course, it was built in the 1850s, long before the building laws of New York included any information on structural design, so it was not designed in an engineering sense, so my attempts to analyze that roof in the 1990s were an anachronism.
The roof performed reasonably well until the 2016 fire. The only damage was rot at the ends of the upper chords and hammerbeams, and that was because the exterior gutter system did not drain well. So, some 150 years of empirical evidence says that the roof was not overloaded by wind. There are a few possibilities as to why: most of the building’s existence, it has been partially sheltered from the wind by taller buildings and it got lucky with wind pressures in its early years; the old-growth wood used in the trusses was significantly stronger than modern lumber, so my analysis underestimated strength; my analysis was wrong and the shallow main arch contributed more to helping the top chords than I thought; the main arch, hammerbeams, and top chords were more fixed against rotation by the wall masonry than I thought; and so on. This is fairly esoteric stuff and the simple answer is: when numerical analysis disagrees with observable reality, reality is correct and the analysis is wrong.
We designed the new, steel-truss roof using current code wind pressures, and it takes advantage of the much greater strength of steel relative to wood. It feels almost like cheating. For the repairs twenty-five years ago, the simple answer was to replace the strength and stiffness lost to rot and not worry about how it was working because we knew that it was working; for the new roof after the fire, we had to know how it worked.