That beautiful shot to the left there is from the basement of the Rookery, courtesy of CAF docent Claudia Winkler. On a recent basement tour their group noticed the sloping walls of the brick pier and wondered whether this was, in fact, one of the infamous pyramidal footings that supported many skyscrapers of the 1880s. Chicago Skyscrapers says otherwise–the Rookery is supported on grillage foundations. What gives?
Glad y’all asked, because this is an important problem in bearing foundation design. Above ground, engineers work hard to collect loads from floors and direct them into columns and piers. This is important, because the columns’ slenderness is the only thing that makes a skyscraper work–you can’t rent out structural area, so you have to find ways to condense loads from the floors above (two-dimensional) into elements that take up minimal floor space (columns are one-dimensional). All well and good, and there’s a good story about the “skeletalization” of brick structures into piers and then into wrought iron and eventually steel columns.
There’s a problem, though, when the columns get skinny. A skinny column can actually support a fair bit of weight, but the connection between the column and the floor is subject to all kinds of intense stresses as the floor loads are redirected into the column. In particular, the column wants to “punch” through the floor due to the shear inherent in the geometry of the problem. Think of pushing a pencil through cardboard–that’s what a thick, brick pier ‘feels’ like. If you try it with a nail, you can see the problem of a skinnier steel column.
That’s Mario Salvadori’s sketch of the problem, along with the most common solution. To prevent columns from punching through slabs, we typically try to maximize the (take a deep breath) interface area between the column and the slab. In other words, project the plan of the column through the depth of the slab. This is the area of material that will resist the shear forces at work–for a round column and a flat slab it will be a cylinder with the radius of the column and the depth of the slab. We can increase this area, and thus spread the stress out, by doing one of two things. We can make the slab deeper, which increases the height of the interface area (the height of the cylinder, e.g.) or we can increase the circumference of the column and thus the perimeter of the interface area.
Here’s how that was typically done in 1910s concrete construction. You can see that the top of each column has two ‘additions.’ There’s a ‘drop panel” that’s roughly square tucked up against the slab–this essentially makes the slab deeper around the column. And there’s a cone-shaped “mushroom cap” that spreads out the cross-sectional area of the column. Both of these are local modifications to the geometry of the slab or the column–they put additional depth or area only in the areas that they’re needed. As a result, the connection is able to spread the shear forces out over a much longer and deeper perimeter. Instead of the interface area being a cylinder whose surface area is the circumference of the column times the depth of the slab, it’s the area of a cylinder whose surface area is the circumference of the top of the cone and the depth of the drop panel. In other words, lots more.
Right, so what does this have to do with foundations?
Chicago engineers faced exactly the opposite of the slab problem below ground, in that they were trying to take the condensed loads in columns and spread them out over a wide area of clay soil–in other words, they were trying to turn the one-dimensional loads in the column back into the two-dimensional loads of the slab in order to distribute them from a point load (which would have sheared right through Chicago’s weak, wet clay) into an area load.
The solution–eventually–was to literally treat the foundation pads as a network of joists and beams just like the ironwork supporting the floors above. Here’s Birkmire’s drawing of a typical “grillage” foundation, (which in Chicago was often built of rejected steel rails, not the beams shown here). Steel’s bending capacity meant that the rails could easily take the bending loads (for those of you keeping score, these work like double cantilevers in reverse), but the punching shear remained a problem that was solved by the trapezoidal column base you can see between the column proper and the top row of steel beams. This is totally analogous to the mushroom caps above.
But before steel came in to the picture, engineers had to spread these loads out through materials–limestone and brick–that weren’t so good in bending. Thus the “pyramid” foundation, basically a big pile of rocks or bricks that very gradually spread the column loads out over the required area. The problems with these were twofold: they were heavy, which put even more load onto the poor soil below, and they took up room–either in the basement where service space for newfangled technologies like elevators and generators was becoming more and more important, or below the surface, which required extra excavation and therefore expense. The grillage foundation solved both of these.
So. The Rookery’s foundations are, absolutely, grillage foundations–here’s Engineering and Building Record reporting on the building’s completion in 1888:
The construction of the foundation is as follows: Under the pier is laid a homogeneous bed of concrete seventeen inches thick. On top of this steel rails are laid quite close together and about two feet shorter than the width of the foundation. On top of these rails is laid a second tier in the opposite direction but standing back at the sides about three feet each way. Above these is a third row of beams which is kept back to about the outer lines of the piers above on the sides though projecting on the ends; and finally there is a fourth row of beams which occupies a spaces a little larger than the area of the pier. These beams are bedded and surrounded with cement, and by reason of their being so thoroughly interlocked, form as it were a solid mass of steel enabling the foundations to spread out as quickly as they do without any defection of the beams, and thus spread the entire weight of the piers over the area of the lowest footing course.”
So what’s going on above? The Rookery is a hybrid structure–most of its structure consists of iron columns but there are also four massive cores that housed the building’s fireproof safe deposit boxes in the corners of the courtyard. I’m guessing that the picture above was taken at the base of one of these, where the masonry pier has to be supported on steel grillage foundations. If that’s the case, it would make sense that the “column base” for the walls would actually be made out of brick, spreading the (really heavy) loads of these four cores out over the steel grills below. You can see that the shape of the brick “spread” in Claudia’s photograph is pretty close to the shape of the column cap in Birkmire’s drawing.
But it was also common for (expensive) iron columns to sit on (cheaper) brick piers in basements, where even if space was at a premium it certainly wasn’t as valuable as the space on the rental floors above–so this could also be a brick pier supporting iron columns above and resting on iron or steel grills below. This would also explain the “column base.”
Either way, punching shear remains an issue that engineers deal with all the time. It’s easier to handle today with high-strength reinforcing steel, but you can still see evidence of this issue in slabs with drop panels, waffle slabs that are filled in around column supports, and in concrete girders that get just a bit deeper as they approach columns. Even Nervi found ways to cope with the problem…rather elegantly: