tabletop nervi

Salone “B,” Torino Esposizione

In post-Nervi lecture Q&As, this question almost always comes up: what made his work so architecturally distinctive? My answer is that Nervi not only designed the structure of his buildings, he also designed the process (thus the subtitle of the book). In particular, since he was the contractor for most of his early work, he sought ways to reduce his costs by breaking down the large spans he built into smaller elements that could be fabricated by relatively unskilled crews, usually in parallel with excavation and foundation work. Distilling structural form into pieces that could be made and, more to the point, placed by small crews kept his equipment overhead low and it allowed him to telescope construction schedules. But it also imprinted his buildings with a definite grain; if all of his pieces were made at a human scale, their agglomeration into a long span would, inevitably, also express that human scale.

In short, I like to say, Nervi’s structures may have been large, but they were productively simple; in fact, if you had enough space and enough time, you could build the Turin Salone B (above) in your backyard–the basic units were made using the simple process of ferrocemento, or light cement troweled onto cages of steel mesh, and they were designed to be light enough that they could be hoisted into place by three or four laborers and a winch.

That combination of process and product has served me well as a great teaching example and, teaching a structures elective with Rob Whitehead this semester, we had the chance to put it into actual practice. Instead of building a full scale version in the parking lot (someday…) we decided to have two teams see if they could replicate the process at two smaller scales; one team fabricated 1/8 scale ferrocement elements with materials from the local hardware store, while the other used 3D printing to produce smaller scale elements, which they assembled during class in a piece of performance art/architectural gymnastics.

Ferrocement proves to be a lot like a loaf of bread: remarkably easy to make a good one, really hard t make a great one. This group got a lot right, making a mold out of foamcore and bending mesh over it–a pretty precise analog to the actual fabrication process that Nervi’s crews used. Quikcrete proved to be tough to work with, though, as it cured too quickly to get good finished edges or to get a full piece made monolithically in one attempt. But the process was fast, and as you can see from the structural test, they were both light enough to assemble by hand and robust enough to at least handle their self weight and resulting thrusts (assuming proper buttressing–Tara is doing a good job there…) Their appearance bothered some of the team, but the actual Nervi units were surprisingly crude, not all that dissimilar from the finish (or lack thereof) here–but they cleaned up well with a coat or two of plaster and paint on their undersides.

The modeling team had the advantage of farming out their fabrication to the school’s 3d printers, but they ran into similar fabrication issues that were analogous to real on-site problems. It took a few tries to get the orientation right (note the incomplete ones in the back), and to get them produced in a reasonable amount of time–the first efforts took something like eight hours to fully print. Getting the time-per-piece down was key to getting them all ready for class, in the same way that simplifying the algorithm in the job site yard was vital to getting hundreds of ferrocemento elements produced just in time to start hoisting them into place.

Our plan had been to have one team assemble a wooden centering that would replicate the traveling scaffold that Nervi used to hoist and place the units, but Nervi didn’t have to deal with errors in scaling in Rhinoceros, so while there’s a beautiful piece of centering in the pic above, it ended up being purely ornamental. We did have a team pour the buttresses while the 3d printing was going on–telescoping the assembly time–and at this point sort of considered delaying the assembly until we could get the scaffolding done to the right scale. Fortunately, the printing team had come up with a snap-tite connection detail in the units that allowed them to assemble multiple elements “off-site,” giving them just enough structural integrity that they could snap larger pieces onto the buttresses–no centering needed (or, really, human centering applied). It took a couple of tries, but once the pieces were all in place the resulting arch was just monolithic enough that it stood on its own, and even survived some tentative testing.

We’ll call it a modest success–certainly an accurate analog to one slice of the Turin hall, and (we think) a valuable lesson in how process–fabrication and assembly–influence and sometimes determine structural form, alongside actual statics. Lots that we would/will change for the next iteration–better cement, more attention to getting the centering right, and maybe jumping scale. Still, certainly proof of concept in that this structure–seventy plus years old at this point–has much to teach.

The best part? I’d given the class a line about making an “IKEA drawing,” or an instruction sheet, for how the process would work–based on the illustrations done by a then-grad student for the book. And did they take the opportunity and run with it? They did.

Graphic: Tristian Thao