Prefabricated or factory-built construction is often touted as a more advanced, more modern way of building. “Everything else you buy was built in a factory, why should your home be any different?” is a common talking point from prefab enthusiasts. But more often than not, prefabricated buildings are built in the exact same way as their site-built counterparts. The “factory” is often not much more than a big warehouse, where framers, electricians, plumbers, drywallers, etc. manually build homes one or two at a time, using the same assembly method, materials, and tools they’d be using on-site. Highly automated production facilities like AutoVol are the exception (and even there, the automated work is likely a relatively small portion of the total effort).
Most prefabrication starts with an existing building system (generally light framed wood or light gauge steel), and tries to figure out a way to build it in a factory and ship it to the jobsite. But you can also do the reverse - start with a particular manufacturing technology or production process, and figure out a way to get it to spit out a building system.
The original balloon frame is such a system - instead of a better way to build post and beam construction, outputs of industrialized production processes (dimensional lumber and nails) were combined into a totally new building system. 3D printing is another - 3D printed houses are having something of a moment.
Another such system is the CNC plywood frame. And while the balloon frame is ubiquitous, and I can’t go a week without seeing a story about 3D printed houses, plywood frames are mostly unknown, tucked away in a corner of the building industry. Which is a shame, as it’s a fascinating bit of technology.
Plywood frames use thin sheets of plywood  as the material for the entire structural system. Sheets of plywood are run through a CNC router, which cuts the plywood into specially designed shapes. The shapes are then attached together using slots and tabs (helped into place with large mallets) to form a variety of structural elements - walls, box beams, portal frames, columns - which then get assembled (using more slots, tabs, clips, and wedges) to form the superstructure of the building. The tight tolerances of the machined plywood mean that everything fits snugly together, which allows simple friction connections to do a great deal of the work (though in practice most of the folks using the system are throwing in some mechanical fasteners as well).
It’s occasionally described as large-scale legos, but a better description might be large-scale U-Gear.
Plywood has a long history of structural use - stressed skin panels and plywood box beams date back to the 1930s, and it was a key element of the famous De Havilland Mosquito. But CNC plywood frames are a relatively recent invention - as best I can tell, they were first described in 2006 by Larry Sass, an MIT professor of architecture. His paper “A wood frame grammar: a generative system for digital fabrication” describes a method for decomposing a 3D structural model into a series of 2D shapes that can be assembled from cut plywood. Subsequent developments have explored the best structural elements to use, and how best to connect them together, but haven’t strayed much from this initial idea.
Plywood may seem like a less than ideal material for supporting the weight of an entire building, but the system has some compelling features. It's wood, which means it's light, cheap (until this year!), easy to throw a fastener into, easy to modify with simple hand tools, and carbon friendly. It’s available nearly everywhere in large quantities, and can be packed tightly and shipped dirt-cheap. Because it's a manufactured material, it's more precise (sizes for plywood or OSB go in increments of 32" of an inch), dimensionally stable, and stronger than typical dimensional lumber. If you can keep it from buckling (and much of the development on the system involves figuring out how to prevent that) a section of plywood can support thousands of pounds per foot. It’s more resistant to supply chain issues (until this year!!!) - plywood is produced at dozens of mills around the country, and if the truck it's on breaks down you can pick some up at Home Depot.
Critically, the machinery for creating the frames (a CNC router) is both widely available and comparatively inexpensive. While a CNC router capable of cutting plywood will run into the tens of thousands of dollars or more, something like a CFS roll-former will be in the hundreds of thousands, if not millions, and is far less flexible .
To the extent that it exists at all, enthusiasm for plywood frames has been focused around the relative ease with which they can be fabricated and built. It requires comparatively little expertise to obtain the material, get it cut, and assemble the structural frame. Much of the development effort (such as Sass’s later work, or a similar effort at University of Clemson) has been centered around giving communities and individuals the tools to fabricate and build their own homes.
Of these sorts of efforts, by far the most developed is wikihouse. Founded in 2011 in England, wikihouse is a non-profit that aims to “put the tools & knowledge to design, manufacture and assemble beautiful, low-cost, low-carbon buildings into the hands of every citizen, community and business“. It provides a full set of open source 3D model files for a plywood-framed house, and instructions/scripts for exporting them to a CNC router, all of it hosted on github.
It’s unclear how many wikihouses have been built, there seems to be at least several dozen scattered around Europe. The house design has gradually evolved over time as they’ve worked the kinks out of the system (they’re currently rolling out what seems to be the third iteration), and appropriately for an open-source project, there are several parallel forks being developed by affiliated groups.
Commercial interest in the system has been fairly limited. I’m aware of only two companies that will sell you a plywood framed house - U-Build and Facit Homes, both located in the UK.
U-Build, appropriately, will ship you a CNC plywood frame kit that you can assemble yourself (taking advantage of its ease of assembly, as well as the dense-packing potential of flatpack components). You place your order, U-Build sends you a truckload of cut plywood, a set of assembly instructions, and you’re off to the races. They have several off-the-shelf models, for small sheds and tiny homes, but will also design (and ship!) a full single family home.
The most developed version of the plywood frame system is used by Facit Homes. Founded in 2007 by architects and industrial designers, Facit’s goal is to help architects more accurately translate their visions into reality. By using software and digital manufacturing, they aim to strip out the “site interpretation” step that takes place when contractors and site crews read a set of plans, and ensure that what gets built is what they designed.
Facit has essentially designed their building system to make maximum use of components that can be produced via programmable equipment. Not only the structural frame (which consists of plywood box sections for the walls, floors, and roof), but also sunshades, stairs, doors, flooring, cladding, and anything else that can be built by a machine reading a dxf file. Often these are made of plywood, but Facit also uses floor trusses, structural steel, ICF, and a variety of other systems and technologies, sometimes performing their own fabrication, sometimes subcontracting it out.
For large projects, Facit will actually deliver a mini-fabrication workshop (complete with a CNC mill) directly to the site inside a shipping container, and build components right there. If a piece is missing, or a change needs to be made, they can simply fabricate a new component immediately, without being slowed down by shipping.
To ensure accurate realization of their architectural vision, Facit acts as a single point of contact with the client, overseeing and incorporating all the building services into their design. This level of control means they know exactly where the systems and equipment will run, preventing them from having to figure out or rework things on-site. Walls are assembled with cable chaises in them, and recesses are left for the mechanical systems, the lights, the windows, etc. This level of coordination is supported by a library of custom components, which adjust parametrically based on the dimensions of the building (unlike many manufactured building systems, every Facit home is custom designed).
Thanks to this level of control, Facit has greatly reduced the uncertainty and variability in their construction process. This allows them to give upfront pricing on their projects, and build using fixed-price contracts (impressive for a builder that only does custom homes). Their current prices are just over $300/ft2.
This is a high-end price, and Facit are delivering high-end homes. They have high energy performance (aided by the tight fit of the machined plywood boxes), capable of hitting Passivehaus. Triple glazed windows, heart recovery ventilation, underfloor heating, and high-end finishes are all standard. This makes them somewhat similar to Bensonwood (and probably a million European builders I’m unaware of), another firm using factory production methods to build high end, high performance homes.
Facit Homes is a relatively small builder - they build fewer than 10 homes a year, all of them custom. And though they’re using advanced manufacturing techniques, they don’t seem especially interested in any sort of mass production. Their goal is quite the opposite, to use them as tools to help architects to achieve their vision, and free them from the constraints of traditional building.
The Potential of Digital Manufacturing
Facit Homes shows much of the potential that digital manufacturing has to offer the construction industry. In typical construction, the architect first creates their drawings of the building. Those get distributed to the various engineering disciplines, who produce their own set of drawings for the various building systems - structural, electrical, mechanical, etc. All those drawings get distributed to the various subcontractors, who then produce ANOTHER set of drawings - shop drawings that can actually be used to fabricate the components, pour the concrete, etc. Each material or discipline will have their own set of shop drawings, which must be coordinated with all the others.
All those drawings together are the set of instructions workers will use to construct the building, on-site. This repeated process of translation and mapping the chain of dependencies is incredibly time consuming and error prone, and is a big part of why buildings take so long to finish. Whenever something doesn’t match between various sets of drawings, or the intent isn’t clear, it has to propagate back up the chain for a change to be made, and then back down again.
Facit Homes shows a potential way to avoid all that. For them, the drawing information seems to live in one place, and gets interpreted by their software (known as the D-process) directly into fabrication instructions read by a CNC router, laser cutter, etc. Those components can then be installed without anyone needing to figure out what exactly needs to be built.
This enables a lot of interesting opportunities. It allows the creation/fabrication of components that would be too complex or expensive to build manually. It theoretically lowers the cost of coordination between disciplines (less effort spent ensuring things are built as intended), allowing labor-saving things like built-in electrical chaises, blockouts for mechanical systems etc. It can streamline assembly via good DFMA - on Facit Homes, the connectors between pieces are arranged such that things can only be attached one way, preventing incorrect installation. And it allows for dramatically reduced cycle time . Instead of taking weeks to implement changes, Facit can make changes to their models in an hour, and be building them that same day.
Cutting the Prefab Knot
Traditionally, prefabricated construction has been limited by several difficult-to-overcome constraints. You either build for a small, local market (within about a days drive), with low-volume, manual production, or you build a large, expensive factory, and try to fill it with enough volume to justify the expense, hopefully manufacturing things inexpensively enough to offset the transportation cost. Neither of these methods has thus far resulted in dramatically lowered construction costs.
But digital manufacturing techniques potentially allow a way to sidestep that. The equipment is inexpensive enough and ubiquitous enough that you can manufacture a large fraction of your building either yourself at relatively low cost, or via a local fabricator. And because the process is flexible (a CNC router will cut whatever it tells you to cut, and can cut a different shape every time) it to some extent frees you from needing to produce a high volume of identical items. Instead of staking out a place on our component size vs complexity chart, it rethinks the entire premise.
 - I’ll use “plywood” as shorthand for any sort of structural sheet wood production - plywood, OSB, etc.
 - CFS rolling machines can usually just produce a narrow range of sections.
 - This is an underappreciated benefit of this sort of system I think. The faster you can roll out design changes, the less design work-in-process you have, lowering your design “inventory” costs, increasing your throughput, and making the overall design process much more efficient.