Excerpted from Eric Hunting, our p2p urbanistic correspondent, who monitors modular architecture:
“Skyler Tibbits’ concepts are really intriguing. At first glance my feeling was that this was something with huge potential, but all very distant. But the more I saw the more it seemed like there were some possibilities for demonstrating experimental architectural-scale structure in the here and now if we rely on relatively simple designs. If I’m interpreting this correctly, the general building system he proposes as Logic Matter is like a chain of blocks–self-linking or carried on a cable–that are picked up from a packing/transport configuration and then dropped into place on a cable using the energy from gravity to compel them to fall, nest, and link in the desired 3D orientation. 3D structures are thus ‘knit’ together like a rigid amigurumi figure. The cartoon in his larger paper on this shows a crane being used to load and drop blocks. Presumably, the crane is roughly following the ‘knitting path’ of the linking blocks to keep them falling vertically while at the top is a towing mechanism. Programming of the blocks seems to be done in their packed configuration–implying an integral electronics–or this might be achieved by a mechanism on the crane. Either the blocks rely on their nesting topology for rigid interconnection or some integral bonding or locking mechanism is also used.
So what we basically have is a space-filling structural system based on a more-or-less solid module. A kind of ‘brick’ structure that doesn’t need human hands to lay the bricks.
How well can this space filling geometry approximate the useful shapes we want for a structure? This relates to my comment you noted earlier. This space filling geometry would be the ‘microstructure’ I eluded to. The building form is the ‘macrostructure’. The larger the modules of the microstructure–or shall we say the lower the system’s ‘resolution’–the more the macrostructure is limited to paralleling the topology of the microstructure. Conversely, the smaller the microstructure modules the more it can approximate any desired macrostructure topology but with the compromise of increasing the number of parts. Previously, I had noted this in the context of labor. It’s a hassle to manually assemble structures that need a lot of parts. But since we’re automating that, the issue is now the general size of module we need to achieve a useful structural performance relative to the production costs and methods of the modules. That then determines the limit in effective ‘resolution’ of the structure and the range of building forms we can use. Tibbits implies a relatively small module size/high resolution (maybe soccer ball sized sized and smaller?), which means many tens of thousands–maybe hundreds of thousands–of units in something the size of a conventional house. Their production would need to be highly automated, cheap, and reliable.
What materials can we use? The larger the module size the more materials options we have but also the more complex and labor-based fabrication becomes. The smaller the module the more limited our choice of practical materials and the more things favor automated processes like injection molding. Imagine the cost of a house if, by equivalent volume, you were making it from the same materials we use in the typical 3D printer. (not to mention the time that currently takes) It could be astronomical. This is why Tibbits was rotomolding his experimental blocks. But most people would not be interested in a house made out of plastic. (even if, frankly, that’s pretty much what the modern suburban house increasingly is. Plastic and high-tech papier mache…) They might tolerate structural foams, but are they strong enough? How about cast/injection molded wood composite? Milled alloys would also be problematic, even if we’re relying on milling machines. It would really demand a process like injection molding, which largely limits us to aluminum. (injection molded glass-alloys is probably too exotic a process to work with near-term) High performance ceramic is likely a very good choice, especially if we used a two-stage process creating a foamed core to lighten the block and make them more insulating, but its still potentially expensive. CEB is probably too brittle, too irregular without a highly refined source material, and would need a large, heavy, block size to accommodate the topological features of these Logic Matter shapes. Concrete would probably have similar issues, though geopolymer could be much better because of its higher density and more ceramic-like properties. Would bio-stabilized sandstone do better? Any other possible materials?
How strong and rigid must the interface between assembled modules be, especially if we are using this to make floor decks? This could be a critical problem for putting this concept into practice. The same problem has also hampers modular/cellular robotics. Either the blocks rely on some aspect of their nested topology to achieve a rigid connection or they must have some other, automatic, means of sticking strongly together. At a relatively small unit block size as implied, automatic mechanical connections become difficult. There are some new possibilities, like magnetically driven screws that have recently come to market in the furniture industry;
But it’s still difficult to deliver sufficient energy for such mechanisms to many small elements, especially if deeply buried in the structure. Building-in a motor to drive such mechanisms would be impractical unless the modules were very large and, consequently, expensive. Tibbits’ prototype Logic Matter blocks are completely passive objects that rely on orientation to program the basic assembly modules. (physical NAND gates) They have no bonding/linking mechanism and rely on human labor. Useful as a way to study the information/computing theory of the concept, but you can’t make anything useful with this. So we’d have to reduce this to practice with a functional block of new design.
How do we finish such structures and integrate utilities? Obviously, these would not be waterproof structures by themselves, though they might be weather-resistant. They would need separate roofing and cladding materials which, ideally, sync-up with the module topology to allow them to be modularized. Can we integrate things like radiant floor heating? Is there interstitial space to route wiring and cables?
If this is possible near-term, I think we’re probably relying on a rather larger scale of block and pretty simple architectural forms. Dome-like forms or pavilion structures where everything else is largely independent of the main free-standing structure leaving relatively large free-span spaces for open-plan use. This may generally be limited to single floor structures at first, but such minimalist forms could well exploit the topology of the building system aesthetically because there may be no broad load-bearing walls to cover-up, just floor and ceiling. I personally consider open-plan pavilion-based housing to be the most practical form of free-standing housing anyway, in part because there are so many ways you can create a self-supporting roof structure and many kinds of hardware that can be repurposed for that. There’s an old Polynesian saying that a good roof and a good floor make a good house. Notice the lack of ‘walls’ in that…
So this does seem possible, but it’s so new we would definitely be in uncharted territory working with it. It could certainly garner a lot of attention, though, if we could come up with a nominally functional system.
More at-hand possibilities may lay in systems employing robotics with larger structural elements. Tibbits notes the problem with the commonly seen modular/cellular robots as a building system. They generally have a very high overhead in integral mechanical systems and electronics relative to the volume of structure they can make, which means a large manufacturing overhead and an astronomically high building cost per square meter. But there are ways this can be ameliorated, leveraging active hardware over much larger areas of structure.3