On June 2nd, at AIA San Francisco’s NEXT/EVOLUTION conference at the San Francisco Art Institute, a panel composed of Yes Duffy of David Baker Architects, acting as moderator; developers Patrick Kennedy, owner of Panoramic Interests, and Fei Tsen, founder of Windflower Properties; and Stanley Saitowitz, principal of Natoma Architects, explored the state of modular housing today.
Tsen’s Windflower project in Union City is being constructed in two phases, the first scheduled for completion early this fall. Both phases use conventional wood frame construction. The second phase will employ the largest legally transportable units—74 feet long by 16 feet wide, by 12 feet tall—dimensions that accommodate the depth of two apartments, flanking a double-loaded corridor.
In the edited transcripts that follow, Duffy begins the discussion with a look at the history and promises of prefabrication. He goes on to frame the underlying question for the panel, “What about a factory can allow us to build things differently?” Kennedy presents an alternative construction method, based on the techniques used to make shipping containers. Saitowitz discusses an approach that uses metal stud walls but eliminates the duplication of walls and floor/ceiling assemblies common to most stacked-box systems. He concludes with a call to architects to “completely rethink the elements of buildings and, rather than relying on craft tradition components, take full advantage of contemporary production techniques.” In their own words:
Today’s panel is a select crew of developers and designers and architects leading the charge on a variety of fronts, utilizing modular, which I’ll also call factory-built housing—those terms are one and the same—fiercely testing out the promises coming from the modular industry. This early in the game, the results are mixed, but the momentum is moving forward.
When you type in “modular,” “prefab,” or “factory-built” into Google, you get a plethora of single-family, sexy little modern cabin/cabana things, which everybody’s touting as the new game changer. We won’t be talking about these today. We’ll be speaking about dense urban multifamily modular housing, which is a very different beast. I’m talking stuff that stacks up to 32 stories high—larger, denser stuff. Prefab single-family housing has been around for 80 years, and it’s really tested. There’s a million ways to do a single-family house, and there’re even more ways to do a little cottage in the backyard. The technology, innovation, prototyping testing that’s happening now is in the multifamily market.
Safdie Architects’ Habitat ’67 got built 50 years ago for the Montreal World’s Fair, using prefabricated concrete forms. It was a government-sponsored effort; they did actually deliver it on time, many millions of dollars over budget. Safdie was 25 years old at the time. Can you imagine why Phase 2 was never built? It’s enormously complicated and expensive, and that promise of modular was blown out of the water for many decades. The ambition of this level of prefabrication was stifled.
Modular makes so many promises: faster to build, safer, more sustainable, less disruptive, time-predictable, cost-certain. The big thing is construction schedule. They say, if you can do site development and foundations at the same time you’re building the units, you have this enormous time savings. You can save up to 40% of the schedule. That makes sense; you have two construction sites building at once. This is the modular appeal, which is very enticing for all of us who are thinking about building more efficiently, building more affordably, and just building more, period.
But very quickly these promises get broken, as we get into permitting, construction trade skillsets, unions, site issues, financing, insurance—all this stuff creeps into what otherwise would have been a project delivered on time and on budget. The question is, can these promises be fulfilled today?
From a design standpoint, the nuances of the factory are a major factor, when there is no real standard out there yet. Not all factories are the same, and not all construction types are the same. The constant that applies to all modular projects is the maximum wide-load freeway dimensions: 16 feet wide—you need a full-on escort—12 feet tall, 74 feet long. That’s the box we get to work in if we are not driving them down a residential street. If we are driving them down residential streets, alleys, or other unique conditions, then the allowable mod size can drop significantly, or not even be a viable option.
We start with the size of the unit. We think of this unit like a cabin of a nice sailboat. The more you can do in the factory the better. If we can build everything in the factory and deliver these things finished on site and never go in the unit, that’s the goal. That’s efficiency at the unit level.
We also look at efficiency at the scale of how they will aggregate on site. We have different studio types, and they can go together, they can be one bedrooms, they can be two bedrooms, and so on. They deploy in a building scheme that can be mixed and matched to meet the unit mix. Every architect has a different stab at how we approach this, and every factory has different efficiencies that are part of the design.
For example, Guerdon in Idaho noted that the tooling of their chassis for the modular units was labor intensive and a significant added cost for each different module size, so we designed a 110-unit building with only two jigs, an 11-foot wide jig and a 15-foot wide jig. That’s how the whole building lays out. We do everything in the factory, except the exterior skin and corridor finishes. Even the elevator shafts are done in the factory.
Which gets to an interesting point in terms of details. There are two contractors on the project, the factory builder & site builder. Our drawings are the only interface to share who’s doing what and to be able to keep everything on track. The sequencing is critical—for designers, it becomes extra work. The efficiency of modular is not that efficient from the design and documenting standpoint. Because here we are on every detail, literally saying, OK, you’re doing that, and you’re doing that.
We’re invested in the role of software in this coordination effort. Can we take the architect’s BIM model from my desk, and can that be the shop drawing on the shop floor for them to build? That’s what were working on now. And that, to me, is what would be exciting. That way, the architectural model is the shop drawing.
In some ways, the factory-centered approach is nothing new. In most factories today, we’re building the same way we would build it onsite for the past 50 years, just in a controlled factory with no weather delays and less skilled labor. If we can achieve just this, we’re onto something. But there’s another way to think about it: to rethink the factory process. What about a factory can allow us to build things differently?
Can we have new wall types being tested and prototyped in a factory, new floor assemblies, new structural connections tested and prototyped in the factory? Can we innovate and still have it be structurally rigorous enough to be able to be trucked down the highway X number of miles and be plucked up by a crane and set down? That’s a tall order.
So in some ways it’s back to 1967: We’re dreaming big, innovating like crazy, but actually this is a little bit different. We now have a wider range of maverick developers, architects, and contractors that are trying things, all over the United States and around the world. They’re working collectively to see this promise of prefab come through. And so it’s not just a one-off project called Habitat subsidized by the government, but a number of real-world tests that are reshaping how we build housing more affordably for the future.
I have been developing high-density housing in Berkeley, San Francisco, and Oakland since 1990. When I first started in Berkeley, it was described as a place torn between a desire to overthrow the US Government and a quest for the perfect croissant. Holds true today. They hate developers. They don’t like infill housing. But they accepted it over time, and over the course of twenty years we built a number of projects, low-rise, mid-rise, high-rise.
Now we’re focusing on entry-level urban housing for students, workforce housing, and supportive housing. Supportive housing is a project that particularly lends itself to prefab development.
It has been exasperating for me as a developer to reinvent the wheel every time we want to build entry-level rental housing. I know architects don’t want to hear this, but I’m looking for a more off-the-rack design, rather than bespoke design, if we’re going to produce serious numbers of affordable housing.
In 2010, I acquired property on Harriet Street in San Francisco, between Folsom and Howard, and decided not only to build micro-apartments—350 square foot apartments—but also prefab. It was a beautiful project, it was successful, it was fully rented, and we sold it. There’s nothing better than the quality of a well-designed prefab place produced under factory conditions. All of the walls were perfectly plumb, we knew where all the wiring was, there were no call-backs on the plumbing, We didn’t have any sheetrockers working with a hangover, so we had no visible nails, anything like that. The quality was first rate. We were heartened by it, but we’re not going to do it again using this technique.
The module is 65 long, 12 feet wide. We brought it down Harriet Street, and it required military-like mobilization to get the building in place. We did it successfully, and we had 4” to spare, which was a little close for comfort, but we had an excellent contractor and excellent preparation. We completely closed the street and had 30 or 40 people working on it. You can imagine the complications that would arise if we didn’t have a first rate truck driver backing that into Harriet Street.
After this experience, which went flawlessly—we had a superb contractor, Pankow, we had an excellent designer, Lowney, and a very good prefab firm, Zeta—we concluded that this was too risky to try again.
We were sold on the idea of prefab, but we concluded we needed to have a smaller, more agile, and easier approach. So we have embraced a prefab design that is all steel modular. It resembles a shipping container, but it’s not a shipping container. It happens to use the same intermodal transportation system as the international shipping system. It’s 8’ wide by 45’ long. It does not require escort cars, it does not require special permits. It doesn’t require calling out the National Guard to bring it down Howard Street. And it can be duplicated on a very large scale.
A fun fact: it’s cheaper to send an 8’ x 45’ module from Shanghai to San Francisco than it is to truck it from Sacramento. That makes a massive difference when you have hundreds or thousands of modules. That is crazy, but that’s international trade; it’s a scale issue, as well.
Our new modular design—the MicroPAD—is based on an idea that we can put a post-tensioned slab on any parking lot in any area. If the city for example wants to give us a lot, we’ll give them back their parking, we’ll build our units on top of it.
The connections are the key. The only connections we have to make on our steel modules are on the outside corners. That’s a typical connection for a container and we added the lateral connection to deal with seismic issues. There’s also moment frames inside these units.
And this is not a technology, by the way, that’s new or untested. Holiday Inn, Courtyard Inn, Marriot have built 22,000 hotel rooms using this technology worldwide.
Our first project using this technology is going to be over in Berkeley. It’s going to start in August of this year. We’re not going to win a Pritzker Prize for this, but it will produce a lot more housing a lot quicker for a lot less than any other technology that I’m aware of.
I want to leave you with is this image, to give you an idea of the capacity that this kind of technology represents, and that is: 56% of that boat—and that boat, by the way, came through the Golden Gate on December 31st of 2015—can deliver 10,000 MicroPADs on one trip.
Before the industrial revolution, most manufactured products were made individually by hand. Craftsmen would create parts, which were then assembled into final product.
The industrial revolution led to a proliferation of manufacturing invention. Many industries—textiles, clocks, carriages, locomotives, bicycles—saw improvements in material handling, machining, and assembly during the 19th century. But it was in the 20th century, with the automobile industry, where the assembly line concept was formalized. The moving assembly line was developed at Ford for the model T in 1913 and had an immense influence on the world. It reduced production time for a vehicle to 93 minutes in 45 steps, delivering cars quicker than the paint could dry, which led to the development of Duco, and hence a single color for the Model T.
This method had great interest for the building industry, and for more than a hundred years we have been trying to implement assembly line technique for construction, but it still remains more partial, and for the manufactured components of buildings, rather than the means for general production.
Garden Village in Berkeley is an example of this hybrid production, where large components of the project, building modules, were built assembly-line style, and delivered finished and furnished to the site, where they were assembled and connected using traditional construction techniques.
This is student housing, which is a quite specific demographic. There were two basic modules, A & B. A is 12’8” x 30’0”, which is living and kitchen. The B module is a bathroom with two bedrooms, 9’8’ x 30’0”. By the assembly of three of these modules, 1 A and 2 Bs, we produce a 4-bedroom student unit, stacked 4 and 5 levels high and then arrayed on the site. Two bedroom units, with an A and a B module stacked in the same way, fill in other areas of site. The other element was the connective tissue, which is a circuit, inserted independently of the modules. That is the structure of the project. What’s special is that it’s a form that derives very directly from the process of making. It’s not using modular to make a building like every other building. It’s actually trying to develop a character of architecture out of the process of making.
The necessities of student housing with the number of bedrooms and need for excessive light exposure also gave us a logic. The first question people always ask me is, “Don’t you have a lot of exterior wall?” Yes, we do. The value of that exterior wall is that it provides a lot of girth for light and creates the possibility of this amount of bedrooms on this size of land. There’s also a rooftop area in the middle, which is surrounded by housing to keep the noise out of the Berkeley neighborhood. If you’ve ever worked in Berkeley, you know what Berkeley neighbors are like.
The project was done by Nautilus, the developer, and Nemo Building Systems is the factory that was established to build the project. This was the first time they’d done a project like this, the first time this factory had made modules. We took on some challenges that other projects don’t do. We eliminated the redundancy of duplicating walls and ceilings. It was partially necessitated by the height restrictions in Berkeley, within which we couldn’t double the floor and ceiling with a complete box. We don’t duplicate any walls, either, so the transport of these partially completed objects was quite complicated.
On Macarthur in Oakland, we built a 5-unit test project to work out the problems we were going to have to deal with at this large scale, the stacking issues, the fact that the project came with the exteriors completely finished with FunderMax and these complicated alignments. We only had one chance to drop the one on top of the other and very little opportunity for correction. So there was a lot of learning we went through in this first trial project.
Some thoughts that emerged from this project:
I don’t think the question revolves solely around means and methods. Building craftsmen type objects with assembly line techniques may never really be feasible. There are other critical and fundamental design evaluations we need to make to invent more feasible products to build through assembly line processes.
The lessons from the automobile industry for construction continue. From Ford to Tesla, the next leap in automotive technology. The replacement of the internal combustion engine with the electric motor is a giant step, not only in terms of performance, where the Tesla can out accelerate a Carrera GTS with ease, but also in terms of resources, efficiency, environmental responsibility, user interface, safety – the list goes on.
There are more than 200 moving parts in a typical internal combustion engine drivetrain; in a Tesla S there are fewer than 20. No belts, air filters, spark plugs, coils, balance shafts, harmonic balancers, pistons, rings, camshafts, valves, lifters, valve springs, fuel pumps, injectors, pressure regulators, idle mixture regulators, throttle bodies, catalytic converters, mufflers, gudgeon pins, con-rods, end bearings, timing belts, distributors, rotors, ignition leads, fuel filters, alternators, tensioners, clutch or torque converters, multi-speed transmissions, to name a few. The EV engine that runs circles around the internal combustion engine in terms of performance is little bigger than a household vacuum cleaner.
The question is “What is architecture going to do in the face of this scale of transformation that’s going on around us?”—not how to build the same old buildings using factory-manufactured processes, but how to radically simplify and greatly enhance building performance simultaneously. This would mean far more standardization with many fewer components and products than the hundreds and thousands of parts we currently waste energy producing, time selecting, and money using. Elements, materials, components, and designs could be streamlined to a few highly efficient and sophisticated models with radical new standard designs that perform at the highest level and are implemented ubiquitously, like cars.
Like Tesla, Apple’s campus may provide a clue for a new direction, not in terms of means of production, but in terms of conceptualization of a minimal language of maximal performance. Here a family of buildings are reduced to two essential elements, floors and walls, concrete and glass, which are deployed for numerous types and programs and create an elaborate universe of high performance which, because of their universal simplicity, could be mass-produced with the most sophisticated industrial processes.
What I’m suggesting is that architects completely rethink the elements of buildings and, rather than relying on craft tradition components, take full advantage of contemporary production techniques.
All images courtesy of the authors, except as noted.