So Very Close

The renovation of a 1910s apartment house (technically a New Law tenement, but somewhat nicer than the image that phrase might being to mind):

What you’re looking at: two spans of floor joists between the (barely visible) wall on the left and the (out of frame) wall on the right, with a steel girder where the spans meet. The girder sits on a granite block where it bears on the side wall, with the bearing point (correctly) not over a window. The wood is overall in good condition although there’s some (apparently) dead mold on the underside of the subfloor above.

It’s the form of the steel girder that I want to talk about, mostly because it’s driving me crazy. That funny-looking bottom flange is the result of having a pair of shelf angles bolted to the web, one on each side, to carry the wood joists. I understand why there’s a mismatch in height: the joists were designed (or, more likely, had their sizes pulled out of a table for the residential load and the span) and the steel girder was designed for its load carrying the end of two spans of joists, and the depths of the joists and girder didn’t match. Even if the depths did match, fitting the ends of the joists into the space between the top and bottom flanges of the girder would have been impossible without trimming the joist ends slightly, which would have been an enormous pain. (One of the first things that I learned when I started working was that people will go to some length to avoid doing things that are enormous pains, so there’s no point in including such designs on your drawings.)

So the reason for the shelf angles is clear, but…wasn’t there an easier way to do this? Mount a 2×4 on the top surface of the bottom flange? Or a 1×4 if a 2x was too thick? That’s part of the problem the original designers and builders faced: they didn’t have the ready availability of plywood that we do. If I want a wood shim, I can ask for it in any thickness by sixteenths of an inch increments; they had to go with 1xs (3/4 of in inch thick), 5/4 plank (1 inch thick) and 2xs (1-1/2 inches thick). Maybe none of those was quite right.

There’s another problem. We now automatically provide 4 inches of bearing for wood joists. Some hundred years ago, 3 inches was often used. But that beam flange isn’t 6 inches wide, so the joists wouldn’t have had full bearing on the half-flange. If you put a piece of wood there, bolted to the bottom flange, it would have to cantilever out to provide at least 3 inches of bearing (the way that the shelf-angle leg is cantilevered out) which is less than ideal. You could (of course) use a larger steel beam, but that would likely weigh more, since the top flange would also be large; since a crane was almost certainly not used in the construction of this building, weight mattered for logistics.

TL;DR: Those weird, redundant-looking shelf angles may be the optimal solution given the building technology of the time of construction.