Arcosanti was Paolo Soleri's project to build the city of the future in the Arizona desert. The Venus Project is Jacque Fresco's project to build it in the Florida swamps. They have been noticeably less successful than other efforts such as Burning Man, which gives the appearance of having been inspired more by Mad Max: Beyond Thunderdome than Star Trek, or Dubai. One of the cornerstones of Soleri's vision was the "arcology", a kind of building-sized self-sufficient city.
Here I try to imagine what kind of arcology I would like to live in.
Some of this will probably seem outdated in a short decade or two, in particular my concern for energy conservation. Rather than seeing a movement toward energy self-sufficiency in small groups, I think we'll see something like another decade of energy scarcity, ending around 2026, as bigger and bigger fractions of the globe are devoted to harvesting solar energy, now that it's finally become cheaper than fossil-fuel or nuclear energy.
A basic question about such structures is how much population density they can accommodate. For self-sufficiency, they must necessarily use no more energy than they can harvest from their environment; in the simplest case, with no deep drilling and fracking to harvest geothermal and no nuclear reactors, this is merely the solar energy available to them.
The NREL Solar Resource Maps show a photovoltaic solar resource of 125–250W/m² in the US, which they have unfortunately chosen to state in non-SI units as 3–6 kWh/m²/day.
This is necessarily the energy supply before the inefficiencies of conversion to electricity; the solar constant above the atmosphere is only 1400W/m², the sun only shines at most half the time for 700W/m², and it's not direct all of that time; reaching 250W/m² after photovoltaic conversion would imply panel efficiencies of 50%, which has not yet been achieved.
A human being normally needs to eat 100–120 W (unfortunately typically rendered in non-SI units as 2000–2500 kcal/day), equivalent to about 5–6 mg/s of carbohydrates or protein, in addition to micronutrients; but micronutrient needs can be satisfied at an insignificant energy cost.
(For brevity, I'll frequently use the Chinese 人 to mean "person" in what follows.)
Dividing these two numbers, we have an absolute maximum population density of about one or two people per square meter in temperate areas like the US. A standard 100-meter-square city block could then accommodate some ten or twenty thousand people, which works out to one or two million per km². Wikipedia tells me this is much higher than any real city:
| City | 人／km² |
| maximum for self-sufficiency | 1500000 |
| [Friendship Village, Maryland][2] | 32000 |
| Delhi | 29000 |
| Ahmedabad | 22000 |
| Manhattan (in New York City) | 27000 |
| Chennai | 26000 |
| Mumbai | 23000 |
| Paris | 21000 |
| Cairo | 18000 |
| Buenos Aires | 17000 |
| New York City | 10600 |
| Taipei | 9600 |
| Shenzhen | 8600 |
| San Francisco | 6600 |
| Hong Kong | 6500 |
| Tokyo | 6000 |
| Los Angeles | 3000 |
However, real solar plants (see the "Utilities → Food" section below) don't even approach 100% conversion efficiency, nor does any known means of conversion of sunlight to electrical, chemical, or mechanical energy. Typical large-scale electrical efficiencies are 16% for the sunlight that hits the panels, but the panels are placed some distance apart so they don't shade each other when the sun angle is suboptimal — since panels are so much more expensive per m² than sunny land, emphasis is placed on not wasting panels, rather than not wasting land — so typical yields are more like 11W/m² for Optisolar's Sarnia project, 6.25 ac/MW = 40W/m² for Nevada Solar One, 15.5MW/85 acres = 45W/m² for SunEnergy1's Duke Energy project.
There are currently-shipping mass-producible thin-film panels that hit 20%, space-deployed very expensive multijunction semiconductor panels that hit 40%, and concentrating solar can hit the usual 35–40% for large fixed steam turbines. There are some potential nice synergies here: photovoltaic panels, including multijunction panels, are nearly black, so the sunlight that photovoltaic panels fail to convert to electricity is mostly converted to heat rather than reflected, which means you can use it to make steam, and concentrating the light means you can justify the use of much more expensive photovoltaic cells, but if you're concentrating more than one or two suns on your cell, you'll need some kind of active coolant to keep the cell from melting — such as steam! So in theory you might be able to hit 50% or 60% conversion efficiency with the combination, but nobody's done it yet.
Modern life involves using energy beyond the bare necessity for survival, generally quite a lot of it, mostly bought on the market rather than harvested for direct use. A self-sufficient arcology would probably have to be less efficient in its material energy use than the average person in a modern society, who takes advantage of economies of scale and specialization in many kinds of goods. The USA DOE EIA AER estimates 2011 marketed energy use in the USA at 97.30 quadrillion Btu/year, another absurd non-SI unit; this quantity is actually 3.243 terawatts; if we figure 311.8 million people, that's 10.4 kW per person, the equivalent of about 100 slaves per person, Buckminster Fuller's "energy slaves".
Let's assume that the overall solar conversion efficiency hits 30%, rather than 50% or 65% or staying at 16%, and that the overall electrical energy use of the inhabitants of the arcology is one-third lower than current US marketed energy use, for the following reasons:
If we need 10.4 kW per person, each person would need 83 m² at 125 W/m² or 42 m² at 250 W/m², giving a density of some 12000–24000 人/km². If we double this, on the assumption that we'll lose a factor of 2 in efficiency to generalization, we get 6000-12000 人/km².
XXX what about the factor of 5 in efficiency you lose to current PV cells? Huh? I don't think the NREL figures cover that! RECALCULATE EVERYTHING. unless NREL does cover that.
XXX if I just multiply everything by 5 it stops being an arcology at all. Possible optimistic assumptions:
With these two 2:3 optimistic assumptions, you use only 4/9 of the current energy and therefore need only 4/9 of the land area.
XXX recalculating again from first principles: current low-cost PV panels are 16% efficient, and current utility-scale PV capacity factors range from 10% (in Germany) to 29% (in California) or a bit more, with the US average being around 24%. (See Japan can achieve energy autarky via solar energy, but not much before 2027.) Those factors are normally calculated relative to a nominal “solar constant” of 1000 W/m², so a 20% CF means 20% · 16% · 1000 W/m² = 32 W/m². This means that 10.4 kW is 325 m², quite a bit more than 42. This gives a density of only 3000 人/km². So it looks like the NREL figures didn’t account for PV inefficiency.
This is 10× the threshold population density where the EU defines an area as urban but about one sixth the population density of Buenos Aires.
XXX food refrigeration is perhaps a fundamental utility service!
42 m² by itself is a fairly comfortable, if smallish, apartment. So this kind of "arcology" could comfortably be constructed, at first, as a city block of single-story construction, which means you could bootstrap it fairly incrementally; you don't have to start with a megastructure.
But if you want a megastructure — especially a pleasant one — it's far cheaper to build parts of it all at once, due to certain economies of scale.
Contrary to the usual Cold-War-era utopias, I don't envision an above-ground mass of megalomaniacal architect-ego-worship, nor some kind of giant right rectangular parallelepiped with sad little rectangular holes in its bare gray concrete walls, nor gleaming cubes and towers — I'd prefer to leave the above-ground surface as a park, with crops, meadows, trees, perhaps a forest with some babbling brooks, and keep the human habitation beneath the ground. Sunlight can easily be brought downstairs with the occasional translucent artificial stone cemented in place below the dirt-line, and ventilation requires a few wide ventilation towers, which can be constructed in organic shapes and beautified with vines.
Subterranean dwelling is not a new idea, going back of course to Paleolithic hunter-gatherers, and featuring in many science-fiction stories, but here I explore why it's historically been marginal (mostly confined to extreme conditions and military surplus reuse), why we now have the option to change that, and how completely fucking awesome it could be.
Underground construction is tricky and expensive, particularly in wet areas where the water table approaches the surface. Here in Buenos Aires, each construction site excavation vents a continuous stream of seepwater into the gutter, proceeding, I suppose, from a pump down at the bottom of the pit. Water tends to trickle and seep in through porous materials like concrete, needing continuous pumping to keep it from eventually filling your living space. Less porous building materials (e.g. concrete sealed with water glass or resins) reduce the seepage, but they don't eliminate it. Even missile silos in the Great Plains fill up with water when left unattended.
There are also risks like collapses, poisonous and asphyxiant gases, and explosive gases; these three can be minimized during construction by digging a pit instead of a cave, but in many places, further measures must be taken after construction to prevent them.
But underground construction has advantages. Aside from the benefits of being able to enjoy the land surface as a park, underground construction is silent, naturally temperature-stable, and much less vulnerable to terrorist bombing — perhaps not a concern wherever you live at the moment, but that's what people here in Buenos Aires thought until Hezbollah blew up the Israeli Embassy and AMIA in the 1990s.
Suppose, though, that instead of a single story of underground construction, we have ten stories, but still one person per 80 or so square meters of ground area (6000 人/km², comparable to the density of Tokyo, in between Los Angeles and San Francisco or Hong Kong). Now the average person has 800m² of live/work/storage space, some of it shared with others, plus 80m² of park; the average couple has 1600m².
This is enough for a sort of underground termite mound of palatial mansions. A normal bedroom may occupy some 16m²; 50 of them will fit into 800m². Every man, woman, and child could easily own their own 12-bedroom mansion.
Hearst Castle, that classic of robber-baron gilded-age excess with its 56 bedrooms, occupies some 8400m², so 800m² of floor per person is enough for truly palatial living. The question is whether we can make that kind of thing affordable to everyone.
Summary: it looks like it costs about US$30k/人 to dig the hole using current mining methods, although that's still really far from the fundamental efficiency limits.
If the average height of a story is 4m — with some ceiling height variety to allow some rooms to be cozier and others more majestic — this is 400k(m³) of excavation, some 3200m³ of excavation per person, perhaps 7000 tons of dirt and stone to remove. I have no idea how to estimate the economic cost of this, except that it is one of those things that has truly enormous economies of scale, since big strip-mining steam shovels are vastly cheaper per ton of rock than little backhoes; but the energy cost of listing 7000 tons by an average of 20 meters is some 1.4 gigajoules.
Energy is commonly priced in yet another non-SI unit, the megawatt-hour (MWh), at typically around US$40 wholesale (US$11/GJ), and two to five times that retail. A megawatt-hour is about 3.6 gigajoules, so we're talking about an energy cost of excavation of some US$20 per person.
(Looking at it another way, at 125 W/m², 1.4 gigajoules per 80m² is about 39 hours' worth of energy production.)
Some random web site says that typical excavation costs are US$299.64 to US$423.77 per cubic yard (another non-SI unit of some 0.764 m³), which would place the cost of excavation of the above at some US$1.3–$1.8 million per person. However, I'm fairly confident that that's a backhoe cost, not a strip-mining cost, while what I'm envisioning is more a strip-mining kind of operation (or "surface-mining", as the miners like to call it these days), followed by construction and surface restoration.
The Global Surface Mining company web site boasts of a trial where one of their surface-mining machines removed 830 tonnes of limestone per hour of operation. At this rate, you could excavate the 7000 tons for the habitation of one person by using one machine for a single ten-hour shift; excavating an entire 100-meter-square city block, digging self-sustaining luxury homes for 125 people, would take some one to four months.
But how much would that really cost? An online Open Pit Mine Model seems like it might be more relevant:
This mine is an open pit mine producing 5,000 tonnes ore and 5,000 tonnes waste per day. The total resource to be mined is 18,715,000 tonnes. Ore is hauled 1,068 meters to an ore stockpile. Waste is hauled 535 meters to a waste rock dump. Rock characteristics for both ore and waste are typical of those of granite or porphyritic material. Operating conditions, wage scales, and unit prices are typical for western U.S. mining operations. ... November 2007
Some further figures drawn from that model:
| Hours per shift | 10 |
| Shifts per day | 2 |
| Days per year | 312 |
| Total Hourly Personnel | 38 |
| Total Salaried Personnel | 15 |
| Supplies & Materials | $/tonne ore | $2.07 |
| Labor | $/tonne ore | 1.96 |
| Administration | $/tonne ore | 0.82 |
| Sundry Items | $/tonne ore | 0.48 |
|-----------------------+-------------+-------|
| Total Operating Costs | | $5.33 |
| Total capital costs | $15,988,500 |
More than half the total capital costs are equipment, including a single 4½-cubic-meter hydraulic shovel. There are two four-kilowatt pumps included, presumably because of the water seepage I mentioned earlier; that's a heck of a lot of seepage.
So these guys are spending US$5.33 to remove two tonnes of rock (one tonne of ore and one tonne of waste), plus their capital cost. If we were to use the same equipment and methods to dig a 40-meter-deep hundred-meter-square hole, removing 875000 tonnes of rock and soil — let's just call it a million tonnes — then it would have an operating cost of US$2.7 million, and at 10 000 tonnes per day, would take 100 days, almost four months. If you divide by the 312/365 days they work, you get 117 days.
If you divide that among 125 people, it's US$21000 per person. Not insignificant — certainly much greater than the US$20 per person that is the inherent cost of the energy to just lift the soil up — but not an overwhelming cost in the world of construction, either.
But wait! I haven't accounted for the capital costs yet. If we figure an interest and depreciation rate of some 20% per year — high for the US, but probably reasonable for much of the world — we end up spending about 6% of the capital cost on interest and depreciation. That adds up to about US$960k, bringing the total cost to about US$3.7M, or US$30k/人. (This is assuming, perhaps optimistically, that capital line items like "Engineering & Management" and "Buildings" represent things you could sell without too much loss at the end of the operation.)
About a third of that cost is labor, and the population of the dwellings is more than double the size of the digging crew, so it seems likely that you could reduce the money cost by a third by means of DIY, or perhaps half or more, if you use methods that are more labor-intensive and less capital-intensive.
It seems possible that a larger-scale excavation operation would be more efficient, but given the number of equipment items of which only a single item is present, it seems likely that a smaller-scale one would not reduce the cost by much.
These fairly enormous capital equipment investments probably explain why underground construction is not traditional, because without them, it's not only dangerous but ruinously expensive.
As an alternative that involves moving less dirt, and might help with water seepage, you could build your arcology essentially above ground, but then cover it with dirt. The dirt sloping away from the center will encourage the water to run off, and so the water table won't rise toward the center of the hill as much as you think it might; and you can run drain pipes underground to provide passively-safe drainage and seep protection for the buildings inside.
This is basically your classic science-fiction pyramid-shaped arcology, but with a layer of dirt on top of it, and with its outer walls restricted to a slope of at most 30° above the vertical in order to keep the soil from slumping off. Since in this case the base will be four times as wide as the height, the bh/3 pyramid volume formula in this case reduces to 16h³/3, times or divided by some minor deviation from squareness. If you want it to have the 125 × 80m² × 4m × 10 = 400k(m³) volume of the previous excavation, it needs to be about 42 meters tall and 170 meters wide at the base; if its surface is covered in three meters of soil (vertically, not perpendicularly), that's about 85000m³ of soil, or about 677m³ per person, less than a tenth of what you'd have to remove for the excavated version.
A circular mound is slightly taller, narrower, and with less surface area; a triangular or kidney-shaped one would be the opposite.
While this might seem more handicapped-accessible than the purely-underground version, you could of course add small hillocks to that version to provide ramp or elevator access, or ordinary small buildings, as in the Terra Vivos entrance near Barstow.
In the USA, it used to be that once a hill was 1000 feet (almost 305 m) tall, you could call it a "mountain" (although the USGS no longer has such a hard-and-fast rule). So if you scale this proposal up by a linear factor of 8 (an area and population factor of 64, a volume factor of 512), you get 8000 people living inside an artificial mountain, each with an average of 25600 m³ (6400 m²). The construction cost per person would remain similar.
Once you have a gigantic hole in the ground, or a building site for a hill, before you can landscape its roof into a park and start building mansions and cathedrals inside, you have to build that roof, and also stop up the seepage from the walls. As far as I can tell, the essentially universal technique at this point in history is to use reinforced concrete, made with portland cement, for floor, pillars, and ceiling, and either concrete or something similar for the walls to keep the water out.
It's conceivable that some other kind of construction technique could work and be cheaper, but I don't know of a candidate.
The surface area of a 100m×100m×40m hole in the ground would be (20000 + 4*4000)m² = 36000m², so each cm of thickness of concrete means 360m³, or about 900 tonnes, or 7.2 tonnes per person, of concrete that will be needed. The surface area of a 170m×170m×42m pyramid would be (170×170 + 170×170×√5/2)m² = 61000m², 610m³/cm, 1.47Gg/cm, or 11.7 tonnes/cm/人.
Unfortunately, I don't have any idea how thick concrete needs to be for this kind of construction. If I guess (I hope reasonably?) that it needs to be some 20cm thick, we end up with:
| | total concrete | total concrete | concrete mass/人 |
| square pit | 7200m³ | 18000 tonnes | 144 tonnes |
| pyramid hill | 12200m³ | 29000 tonnes | 230 tonnes |
These numbers are smaller than the amount numbers for the amount of rock and dirt that need to be removed from the hole, by about two orders of magnitude, so I assume that the labor cost to put this amount of concrete in place will be small compared to the labor cost to dig the whole. Nevertheless, concrete costs US$100–200/tonne, so we're looking at a materials cost per person of US$14000–45000.
The cement in concrete costs about 330–660kWh/m³, which is to say 1.2–2.4GJ/m^3, to produce. At US$11/GJ, this is a very small fraction of the dollar cost of the concrete. At 10.4kW/人, that's about 33–66h/人/m³, so we're looking at about 80–160 days of the arcology's population's usual total energy usage to produce the cement; or, if the arcology were to produce its own cement, 40–160 days of its own energy harvest. In practice, it's probably not reasonable to expect the arcology to be self-sufficient before it's built!
You can probably strengthen the walls substantially and avoid the need for internal support pillars by making them not flat — like an eggshell, they should be everywhere convex outward, except possibly t the bottom — and veining them, like a leaf, an insect's wing, or a plastic injection-molding.
Once you're done with the excavation and sheathing it in concrete, you're done with the parts of the traditional construction process that benefit enormously from economies of scale. Everything past this point can be done incrementally without dramatically increasing its unit cost.
I was interested in Magnus Larsson's 2009 "Dune: Arenaceous Anti- Desertification Architecture" proposal a few years back to cement desert sands using urea, Bacillus pasteurii bacteria that ferment it into alkaline carbonate, and calcium chloride to form calcite from the carbonate; but nothing seems to have come of it, and the process seems to produce hazardous quantities of ammonia, as anyone who's inadvertently fermented urea can tell you. So it might be a cheaper alternative at some point, but isn't yet; watch Ginger Krieg Dosier's startup "bioMASON" to see if it works out! (Dr. Dosier Tweeted me that in her process, the "by-product is captured in a closed loop system", by which I assume she means they make their bricks in a hermetically-sealed chamber and bubble the ammonia through a sulfuric-acid solution, or maybe just dissolve it in water and recycle it into new urea.)
If it does work out, we can expect the result to be as hydraulic as portland cement, slightly less porous, and perhaps significantly cheaper. You don't need a cement mixer, the raw materials are urea and calcium chloride, and you may be able to use aggregate in-situ by soaking it with the water-soluble cement components. Portland cement costs US$110/ton, and comprises between a quarter and half of the mass of concrete, and therefore about a third of its cost. Urea costs about US$400/ton, while calcium chloride costs about US$200/ton; but the resulting calcium carbonate will not include most of their mass, since the chloride ion and the majority of the urea (both its amines, which is to say, everything except its carbonyl) are merely waste products.
Even if it's possible to substitute raw piss for industrial urea, it would not reduce the cost substantially. Piss is only about 1% urea, So you'd need 100 tons of piss per ton of urea, and I think you get 1 mole of carbonate per mole of urea, which is to say 60g of carbonate or 100g of calcium carbonate per 60g of urea — that's assuming the other oxygen comes from dissolved O₂, not from more urea molecules — so you'd need 60 tons of piss per ton of cement, or 8600–14000 tonnes of piss per person. At a rate of 1–2 ℓ of piss per person per day, this would amount to 12–38 millennia of piss production. Gathering the piss of millions of people for the construction project would cost more than simply buying industrially-produced urea.
Reinforced biocement would seem to pose a couple of major problems. The first, which may not matter in this application, is that calcium carbonate expands and contracts very little with heat, while portland cement has an expansion coefficient similar to that of the reinforcing steel, so reinforced lime cement will tend to crack when exposed to thermal stress — which hopefully subterranean construction won't experience! The second is that chloride ions can corrode the reinforcing steel even in extremely alkaline environments, so to reinforce biocement with steel, you probably need a source of calcium other than calcium chloride. Unfortunately, the other soluble calcium salts that occur to me (calcium bromide and calcium iodide) are much more expensive, and furthermore, I'm not sure they wouldn't cause the same corrosion.
An alternative that's been sometimes explored in the past, which might solve both problems, is reinforcing concrete with bamboo rather than steel. You need a higher fraction of bamboo than steel to reach the same strength, but I'm pretty sure chloride doesn't degrade the cellulose that gives the bamboo its strength.
As reported in the March/April 1980 Mother Earth News, you can get a concrete-like substance by electrolytically extracting minerals from seawater on submerged wire armatures; futurist architect Wolf Hilbertz founded companies in the 1970s and 1980s to commercialize the project for, among other things, healing reinforced concrete that had been damaged by corrosion. The resulting accretion on the cathode, principally composed of aragonite and brucite, is known as "seacrete" or "biorock"; it can grow about 5cm/year, and the efficiency is about 400–1500g/kWh (100–400kg/GJ). The strength is comparable to concrete when accreted at 5–100mA/ft² (50–1000mA/m²); typically you use 12V.
If it were logistically feasible to use seacrete to construct the arcology, the 144–230 tonnes needed per person would therefore cost 350–2000GJ, or US$4000–23000.
However, discussions on the Seasteading fora debate this, claiming the real efficiency is closer to 50g/kWh (14kg/GJ). If this is correct, it would push the cost an order of magnitude higher than concrete.
To keep the structure fit for human habitation, it's necessary to provide light and clean air at comfortable temperature and humidity; to exhaust or consume CO₂; to prevent the growth of toxic or allergenic molds; to haul away garbage; to provide food and drinkable water; to provide means of ingress and egress; to limit the spread of vermin such as cockroaches, fleas, mosquitoes, mice, and rats; to contain, ventilate, and neutralize eruptions of noxious chemicals (say, when some poor dummy foolishly mixes bleach with ammonia while cleaning — or just when someone has a night of farts with a lot of hydrogen sulfide, or burns his toast); and to deal with sewage, ideally by composting and refining it. It's also highly desirable to provide electrical energy, internet access, individual climate control to taste, hot water, compressed air, and vacuum suction for cleaning, not to mention some way to wash your laundry and stuff.
Many of these problems are simply slightly-larger-scale versions of the problems that face large contemporary apartment buildings, but some are not, and some admit better solutions than the traditional ones.
The compact shape of an arcology makes such utilities easier to provide. In the 100×100×40 pit, for example, you have 400k(m³) of space within a 74-meter radius; you never have to lay more than 74 meters of pipe or duct to reach anywhere. The corresponding distance in the pyramid is 119 meters.
The most obvious question about living underground: "Won't it be too dark?"
It depends. It can be a great deal brighter than many current dwellings. Aboveground, nobody lives in a glass house, because if you stop convective and radiative cooling with glass, you get a massive greenhouse effect, and you can rapidly reach dangerous temperatures — think of a toddler locked in a closed car in the sunshine. In above-ground passive solar design, according to Daniel D. Chiras's The Solar House: Passive Heating and Cooling, typical allowable glazing on sun-facing walls ranges from 7% to 12% of the floor area, depending on issues like thermal mass, overhang, and local climate.
That means that the above-ground passive solar houses we think of as "luminous" and "bright" are already between 88% and 93% dimmer than direct sunlight, just in order to remain at a livable temperature without outrageous amounts of heating and air conditioning!
And traditional construction techniques (the picturesque clapboard and adobe houses we all know from landscape paintings) more or less adhere to these limits.
Bringing 10% of the sunlight underground would more or less imply that 10% of the land needs to be skylights. (You can also illuminate electrically, but this adds the 80-85% inefficiency of the photovoltaic panel to the 85% inefficiency of the LED or fluorescent tube, giving you a total of 97-98% loss and therefore requiring 30–50 times as much land area; better to bring the light inside directly when you can.)
However, first of all, the skylights don't need to look like windows. They can, and should, take the form of massive glass or glass-like sculptures, frosted luminous stones, clear-bottomed waterways, or gigantic artificial formations of quartz, sapphire, fluorite, or salt crystals; quartz crystals can be grown at a millimeter per day with the hydrothermal method, and giant thick, but hollow, glass or acrylic vessels can be filled with a cheap material with a similar refractive index, such as water. LiTraCon-style embedding of optical fibers in an opaque matrix can even produce mostly-opaque rocks which nevertheless conduct a substantial fraction of the light that strikes them down into the ground.
Second, we only really need to illuminate the rooms we're in. If each of our 125 inhabitants is, most of the time, in a room of some 3×6 meters, we only need some 2m² of skylight per person — let's say 4m²/人 to be safe, for a total of 500m² out of some 10000m² — and beam the light down to where we want it using light pipes, either fiber optics or air-core metallic lightguides. They don't have to be very high-quality fiber optics, since the light only has to travel some 40 meters at worst, and losing half of it over that distance is tolerable, as long as it doesn't damage anything (e.g. melting it). That is, you could lose 3dB/40m, or 75dB/km, and be fine.
Generally it's a lot easier to increase illuminance than to decrease it, so it might be a good idea to have a few skylights that focus sunlight to provide more than one sun of illuminance for high-illuminance applications.
Lightguide illumination has been deeply investigated already as a way to illuminate buildings, and it works.
For solid fiber-optic lightguides, I think the circumference of the lightguide must be proportional to the absorptivity of the material, or a bit more, to prevent overheating; this would seem to suggest bundles of thin lightguide cables so you can blow air through them. The total power absorbed in the fibers will be quite significant, at least if you don't use high-quality glass: if you're collecting 500m² of sunlight at the surface, that's about 500kW, and if you absorb half of it in the fibers, you need to dissipate 250kW from the fibers. Liquid-core lightguides could use the liquid itself as the coolant, providing hot-water heating from the light absorbed from the lighting system.
Typical optical fibers used today for communication have attenuation of about 0.3dB/km, which is about 0.012dB in 40 meters; so if you managed to use them, they would absorb only about 0.3% of the sunlight, or 1.5kW, so you could use about 30× less of them, by circumference, than of some cheaper material. However, they cost a lot more than 30× as much, they're a lot trickier to connect, and they'd need complicated light-gathering and great care taken with the highly concentrated sunlight within them, so it wouldn't be worth it except perhaps for special applications, like solar surgery or welding.
Ocean water attenuates at about 10dB/75m or 5dB/40m, despite its particulate content, so transparent pipes full of pure water might work fine as light pipes for this purpose if you don't bend them too sharply, as long as the pipe material's refractive index isn't too much higher than the water's. (On the other hand, other published attenuation coefficients for seawater point at more like 10dB/40m in even the clearest waters; apparently it depends on the wavelength range you consider. Some book on the Chesapeake Bay says, "Lorenzen (1972) estimated the attenuation due to water alone to be 0.038 m⁻¹, though his measurements were for deep ocean conditions, in which measurements generally commence at depths > 5 meters."; this works out to be 0.17dB/m, or 6.7dB/40m. Pope and Fry's 1997 measurements find absorption coefficients varying by wavelength from 0.00442/m at 417.5nm, which they point out is "more than a factor of 3 lower than previously accepted values...unquestionably a result of contamination by [the other experimenters'] stainless steel cell", up to 1.678/m at 727nm; I don't even know how to interpret an absorption coefficient >1, but if the other one meant what I thought it did, it's 0.019dB/m or 0.77dB/40m, which is quite acceptable. If too much of your light at depth were blue, you could use fluorescence to reconvert some of it to yellow; I used to have a dark green plastic bottle that fluoresced bright yellow in the light of my blue LED, but wouldn't fluoresce in ultraviolet light at all.)
Fiber optics have the advantage that you can "switch" them by filling junctions with liquid water, which allows light to flow instead of suffering total internal reflection.
I don't know if you can get reasonably cheap glass, acrylic, or polycarbonate of sufficient clarity. Apparently Toshikuni Kaino et al. (Applied Physics Letters 41, 802 (1982)) describes an acrylic (PMMA) plastic with a 50dB/km loss, which would be ample. Apparently this is what is now known as "POF", is widely available, and costs US$0.25/m in 1mm inner fiber diameter, but that stuff has unacceptably high losses of up to 200dB/km; so I suspect the answer is no.
Alternatively, you might be able to use an air-filled pipe lined with aluminum, aluminized mylar, silver, or gold, to a thickness of a few dozen atoms. The trouble is that you lose energy on every reflection. Aluminum is only, at best, about 92% reflective in the visible, so you lose 0.36dB per bounce, while gold reaches 97% in the red, thus losing only 0.13dB, but absorbs a lot of blue and even yellow light. Silver nearly hits 97% across the visible. With sufficiently wide light pipes, say bigger than 50cm, these could maybe perform better than fiber-optic light pipes full of water.
The actual luminaires illuminated by that the sunlight delivered through the lightguides can be of almost any form. You can illuminate the water in an indoor fountain or pool, a crystalline object such as a large salt crystal, or anything made of frosted glass. A planar lightguide made of glass could even out the illumination over a large surface made of something like marble bonded to it. Or you could backlight a water-cooled LCD display as a sort of virtual window.
Fluorescence and filters (dichroic or otherwise) can be used to alter the colors of the lighting.
You might prefer to have dimmer lighting some of the time, even during the day; for example, many computer screens are more readable in dim light. Direct sunlight, like sunbathing on the beach, is around 100klux; full daylight out of the sun is around 15klux; a cloudy day is around 1klux; a typical office is around 400 lux; a typical living room at night is around 50 lux; you can see when things are illuminated with around a millilux; and you can see light sources down to a few nanolux.
Accordingly, if you choose to light your office like a typical office during the day, instead of using 4m² of skylight for yourself, you'll only be using 0.04m², leaving more for other arcologists.
You can, of course, use electricity to get lighting at night or when you want more light than is available from the sun. The best LEDs are around 200 lumens per watt, and a lux is a lumen per square meter, so to reach 15klux with those you'd need 75 watts per square meter, and about US$75 per square meter. Fluorescent lights are a little lower in efficiency, but also in cost.
10.4 kilowatts is enough energy to run quite a lot of incandescent lights, too, but you probably want to avoid doing much of that, just because it will contribute undesirably to heating, adding extra heat that needs to be removed.
One of the key features of historical luxury homes has been their gardens. But, aside from the park at the surface, it might seem that we don't have sufficient lighting to maintain hundreds of square meters of lush gardens per person; after all, to preserve the surface, we're only letting a small fraction of the sunlight into the arcology.
But, if our purpose is lush vegetation rather than high agricultural productivity, a small fraction is plenty.
Photosynthesis and productivity in different environments, vol. 3, (ed. John Philip Cooper), has on p. 20 a graph of "relative illuminance", showing that on the floor of an evergreen oak forest, illuminance was only some 5% of full sunlight, while in the ten meters above the floor of a tropical rain forest, illuminance never exceeded 2% of full sunlight, only reaching 5% at some 20 meters; at the forest floor it was more like 0.5%. (He also provided absolute numbers for PAR, or "photosynthetically active radiation": 5-50 "cal cm⁻² d⁻¹", which I suppose is cal/cm²/day; these non-SI units, sadly, are ambiguous, as there are two separate units of energy known as the calorie, which differ by three orders of magnitude). Tarsiers: past, present, and future, (ed. Wright, Simons, and Gursky) cites "Grubb and Whitmore, 1967" for the observation (p. 39):
At the forest floor, rain forests have low relative illuminance of visible light, often less than 1%.
Frances Baines, in the Daintree ancient tropical lowland rainforest in Queensland, measured 2klux in the shade, where she was examining ultraviolet illuminance on reptiles, compared to 90klux "in the sunlit area". She notes, however:
It is interesting to note the wide range of temperatures, light levels and UVB light levels available on the forest floor - all within feet of each other, and constantly changing as the sunlight moves through the canopy above.
This is consistent with some 2% average illuminance.
So, plants adapted to life in forests below the canopy could be gardened productively with minimal amounts of light, if its spectral composition isn't too far off. Every square meter of light captured at the surface could be distributed to feed 50 or 100 square meters of underground gardens simulating a rain-forest understory, with its ferns, mosses, shrubs, insects, and reptiles.
You'll probably need substantial ultraviolet, though, so soda-lime glass and water won't cut it as light-pipe materials.
People and fires produce carbon dioxide, which acidifies your blood, makes your lungs feel like you're suffocating, and would eventually replace all the oxygen in your air and suffocate you. But for plants, CO₂ is an essential nutrient, which they struggle mightily to extract from the air; normal air is only 0.039% CO₂ (up from 0.032% CO₂ in 1960). If you feed them air with 0.1% CO₂, they can grow 50% faster and produce 12% higher yields. You can deal with sustained breathing of air with 1% or 2% CO₂, but normally you want to try to keep it to 0.2% or so. Your own exhalation is about 5% CO₂, totaling about 1kg/day (12mg/s); 1% CO₂ is enough to kill some insect pests.
Also, in some places, CO₂ will seep out of the ground, and you need to ventilate spaces fast enough to keep it from building up. This is particularly tough underground, because CO₂'s molecular mass of 44 is much higher than the molecular masses of nitrogen and oxygen (30 and 32). It's so much denser that you aren't going to be able to do it with convection at any kind of safe temperature. You have to do it with chemicals or air pumps.
Compost also produces CO₂.
So you'd ideally want to route a lot of CO₂-rich air to surface greenhouses, both to reduce the CO₂ emissions of the arcology and to promote the productivity of its crops, ultimately using nearly all of the produced CO₂ (average 1.4g/s across all 125 people, not counting sources other than respiration) into the biomass in the greenhouse. But that turns out not to be possible, as explained below.
The numbers above suggest that you need to change the air in a room after you've breathed at most a twenty-fifth of it. If the room is 3m×3m×6m, it contains about 54m² and 54kg of air, and it needs to be replaced every time you emit 0.2% × 54kg = 108g CO₂, which at 12mg/s, is 150 minutes — an air change every two and a half hours. However, as shown below, other considerations require an air change several times more often than that!
Nuclear submarines and space stations sometimes have "scrubbers" consisting of metal oxides (CaO, LiO) which absorb CO₂ from the air, producing carbonates; and houseplants can also absorb CO₂. These might be useful for resiliency purposes, but they probably aren't necessary for day-to-day living.
It's necessary to replace the air in a room several times an hour to keep it habitable, due to accumulation of other things than CO₂: humidity, bad smells, dust, and heat. A list of uncertain provenance gives numbers that range from two air changes per hour (that is, 2m³ of air per m³ of room volume per hour) for a warehouse, up to 15–60 for kitchens, or 20–30 for spaces like bars, taverns, clubhouses, and repair garages.
(Dust, aside from being potentially allergenic, can also be explosive.)
This suggests that, for luxury, you need at least 10 air changes per hour in the rooms where you are, and more if there's a crowd or things are on fire; and at least 4 air changes per hour in the rooms where you aren't, which will be the vast majority of the arcology and therefore dominates the total ventilation need.
If you figure that all of this air needs to come from the outdoors, which seems like a conservative but perhaps not unreasonable assumption, that's the full 3200m³/人 of air, every 15 minutes (≈1mHz), or roughly 3.2m³/人/s, or 400m³/s across the whole arcology. In the non-SI units usually used for HVAC work in the US, that's 850 000 cfm, which is a fucking hell of a lot of air, all of which has to go through highly-efficient heat exchangers to prevent excessive heat loss. That's not hard to do, since you can easily lay hundreds of meters of ducts, but it's something to keep in mind.
The Passive House guidelines require not exceeding 30m³/人/hour, which means 0.008m³/人/s, "to avoid overly dry air", because "humidifying the air within the ventilation system is to be avoided for reasons of hygiene". You're obviously going to be violating the crap out of both of these recommendations, along with most of the rest of the Passivhaus program, because although it's awesome, it's designed for ordinary buildings, not underground cities.
Changing the air so often means your breath will be useless for feeding plants unless you concentrate it further, but whatever.
A small US$80 six-inch fan does 400CFM, which is to say, 0.2m³/s. You need sixteen thousand times that, which would, linearly extrapolating, cost US$1.3 million of fans.
More digging around suggests that you can order custom-built 3.5PSIG 50 000CFM blowers, of which you'd need 17, but if you have to ask, you can't afford one. At 3.5 PSIG, which is to say 24 kPa difference input to output, each of these would be outputting 570 kW, for a total of 10MW, or 77kW/人 — way outside our energy budget, even before factoring in motor losses!
That means that if you really want to do 400m³/s, you need to do it at more like 1kPa, not 24 kPa. A table of duct sizes suggests that for 50 000 cfm (24m³/s), you should use a 41" (104cm) round duct, which will have a frictional loss of 0.67 inches water (160 Pa) per 100' (30.5 m).
If you used 17 such ducts for 40 meters of input and 17 more for an equivalent distance of output (a combined total of 104cm * π * 80m = 4400m² of galvanized sheet metal), then you'd have 426 Pa of pressure dropped in those ducts, for "only" 171kW. That gets us down to just over 1kW per person, which is manageable but still seems high. Despite the chances of infiltration by movie villains, it seems like it might be worthwhile to use still larger ducts in order to further reduce the energy usage, noise, and surface winds; and perhaps very wide, flat ones, so you can make them work effectively as heat exchangers. (They have to be wide way out of proportion to their flatness.)
XXX figure out ducting and heat exchanger heat loss
It may also be necessary to filter air coming in from the outside to remove smoke, volcanic ash, carbon monoxide, ozone, or other pollutants.
XXX
XXX
XXX
XXX
The traditional approach to garbage in big apartment buildings was the garbage chute: a nearly-vertical shaft with a door on each level into which you would dump your garbage bags, emptying into a bin in the basement. This seems to have gone out of style; XXX I don't know why.
The approach taken in the apartment buildings I've lived in in Buenos Aires in the last few years is, instead, either for each inhabitant to carry their garbage separately to the curb for pickup, or (especially in larger buildings) to have a garbage can on each level, which the building caretaker empties each day, typically dirtying up the freight elevator.
Garbage in general is a pretty big problem for would-be self-sufficient communities; simply stated, a group that generates garbage is not really self-sufficient. The universe has no garbage, only recycling. But full recycling in a modern society is quite difficult, and probably requires a group much bigger than 125 people.
Nevertheless, we can make substantial improvements over the typical state of modern urban life, within the scale of this single modest building.
A 2008 New York State study found the following composition of the garbage of the residential and commercial/institutional sector:
| paper | 33% |
| glass | 4% |
| plastics | 14% |
| metal | 7% |
| "organics" (food?) | 23% |
| textiles | 5% |
| wood | 3% |
| other | 11% |
The total is 18.3 million tons; if the state contained 19 million people, that's one ton per person per year, or 28mg/s/人. With that in mind:
| | | mg/s/人 |
| paper | 33% | 9 |
| glass | 4% | 1.1 |
| plastics | 14% | 3.9 |
| metal | 7% | 1.9 |
| "organics" (food?) | 23% | 6.4 |
| textiles | 5% | 1.4 |
| wood | 3% | 0.8 |
| other | 11% | 3.0 |
Of these, paper, wood, textiles, and "organics" can all be composted for fertilizer, a process which dramatically reduces its mass over the course of some six months to two years. Plastics can be recycled into other plastic (especially if PET), melted into structural material, or burned for energy and CO₂ for the greenhouses. Metal and glass can be profitably sold outside the arcology for scrap, or recycled within. This leaves only "other", much of which could also be used for structural material; but if not, we're left with 3mg/s/人 of dumpit, 375mg/s in total, or almost 12 tonnes per year for the arcology as a whole, perhaps 6m³.
Since this "other" contains nothing that could plausibly rot, and hopefully nothing that's chemically unstable, it doesn't need to be taken out immediately. You could quite reasonably take out the arcology's garbage once a month instead of once a day, to keep it from piling up; or if you decided you had to let it pile up, you would have enough space to pile it up inside the arcology for 68 millennia. Given the likely future strategic value of material resources in garbage, this is probably a good idea.
In my book, that essentially removes garbage as a problem for self-sufficiency, as long as composting can be made to work, which turns out to be pretty easy.
From the above, we get an estimate of 64% of garbage as compostable, a total of 18mg/s/人, or 2.2g/s. Of this, some 6.4mg/s/人 is "organic", which presumably means "wet"; that probably shrinks by some 75% if you dry it out, which you probably want to do before you compost it, reducing it to 1.6mg/s/人, and reducing the total to 13mg/s/人 or 1.6g/s.
This compost, if managed as "hot" compost and watered and turned properly, takes some 1-3 months before it's ready for use as fertilizer. If you take the high end of that range, 3 months, you can see that you have some 13 tons of compost (say, 13 m³), plus water, constantly fermenting. This is an eminently, almost trivially manageable quantity, and it will diminish further once I get to discussing sewage, since much of this compost will be diverted to sewage treatment.
This is, however, a size of compost heap that will benefit from a mechanical compost tumbler, regular temperature and humidity checks, and XXX
As I mentioned before, you, individually, need to eat 100–120W or 5–6mg/s of carbohydrate and protein to survive.
Soy, one of the highest-yielding crops, yields on average about 45 bushels/acre/year (12 picometers per second), and each bushel yields 11 pounds (5kg, 141 g/l) of oil and 48 pounds (22 kg, 620 g/l) of meal, which is about 80% carbohydrate and protein. 620 g/l * 12 pm/s * 80% = 6.1 micrograms/m²/s, which means you need about 1000m² of soy to feed you, a quarter of an acre. Not much!
Well, that's unfortunate, because you only have 80m², and that has to include your solar panels and skylights too. You're too low by a factor of about 12.
Sugarcane is one of the most energy-efficient crops. Experimental small plots in Brazil, says WP, have yielded 250 tonnes/hectare/year (25 kg/m²/year, or 790 micrograms/m²/s). Suppose you can raise it in greenhouses with enriched CO₂ and fertilization and everything; how much can it feed you? Suppose, like Louisiana's farmers in 2011, you get 230 pounds of raw sugar per ton of cane, 11.5%, and suppose that's basically 100% carbohydrate. That's 91 micrograms/m^2/s, which is 15 times better than soy! You're saved! Barely!
But, well, you do need some protein. And 25 kg/m²/year is way beyond normal; normal is 7. Those Louisiana farmers were delighted to get 9.6. 25 was the NPK-fertilized trial in Brazil reported in Bogdan's 1977 Tropical pasture and fodder plants, and it's anyone's guess whether you can really reproduce that.
How can it be that 80m² provides 10kW of XXX oh, of energy, not of electricity.
You may be able to supplement your diet with fungus. Quorn, for example, is made from the hyphae of the Fusarium venenatum soil mold, processed to remove RNA and DNA.
Somewhat to my surprise, it appears that it's actually possible for an arcology to comply with, say, the Pennsylvania fire code, because the fire code doesn't actually require you to be able to get out of a building in a reasonable time; it just requires you to be able to get to the other side of a firewall (an "exit") in a reasonable time, and from there to outside the building (an "exit discharge").
The fire code's guidelines on number and width of doors seem reasonable:
The fire-code occupancy guidelines for housing occupancy is to assume at least one occupant per 125 square feet (11.6m²) of bedroom. If each person did actually dedicate a quarter of their 800m² to bedrooms, you'd end up with 200m² of bedroom per person, with the rather absurd result that you'd need to build means of egress for some 17 times the sustainable occupant load of the arcology — almost 2200 people. That means you'd need 36 "units of width" on the exit stairways. The "units of width" measure is a little bit nonlinear, and it turns out that, with four exit stairways, you could manage it if each one were 4.6 meters wide. Four stairways of 4.6 meters wide, by themselves, should be no problem. However, should someone decide to set up a classroom or dining area or something in one of their rooms — something with fewer square feet per person — they could easily exceed the 2200-people "occupancy" limit!
Certainly, exits sufficient for emergency use by thousands of people should be adequate for regular use by 125 people.
Moving-sidewalk ramps carrying you rapidly up from the lower stories to near the surface, in the case of the below-grade arcology, would seem to be a good idea.
XXX
XXX
XXX
XXX
XXX
XXX
A bacteriostatic water tank at the top of the arcology, filled with chlorinated water, can provide on-demand pressurized clean cold-water supplies to all residents.
Where do you get the water? Well, maybe you can use a municipal water supply, but that's not very self-sufficient. You can use a well, if you're in a zone where the groundwater is safe and abundant. Or you can harvest rainwater or moisture from the air.
As mentioned later, the usual Burning Man figure is that you need about 8 ℓ/人/day, or 93 microliters/人/s. Spread over 80m²/人, that's 1.16 nm/s of rain you need, or in the non-SI units typically used by meteorologists, 37 mm/year or 1.43 inches/year. There are parts of the world that get less than that: much of the Sahara and the Atacama, and small parts of Arabia, the Australian Outback, Nevada, Greenland, and Siberia, and presumably Antarctica; but most of the Sahara, of Arabia, of the Gobi, and so on, get more rain than that.
But suppose you don't? Suppose you're in the middle of the Sahara? You have to condense water out of the air, either using a thermal store (radiate away heat at night and store the cold in a water tank, then use the cold water to condense more water on the surface of a pipe, warming up the cold water until the next night) or, in the worst case, using an electrical heat pump, aka an air conditioner. The electrical heat pump needs to be able to lower the temperature of the air to its dewpoint, at which point its coefficient of performance is probably closer to 1 than the usual 2 — it's burning 1 J for every J of heat that it manages to pump away from its cold side. [Condensing water takes] about 2.26 MJ/ℓ; at 93μℓ/人/s, that's 210W/人, an easily payable energy cost.
So you can totally build a viable arcology in the middle of the Sahara, just as long as the Tuaregs are cool with it.
XXX
XXX
XXX
XXX
It seems inevitable that some of the people attracted to such a community will be interested in resiliency against external shocks, such as terrorist attacks, economic collapse, or military occupation. So it might be worthwhile to investigate how much storage space you need for emergency supplies.
Of all the groups who store up for emergencies, the Mormons are the most extreme; they are recommended to keep three months' worth of food on hand, and normally a year's worth. Of all the supplies you need, oxygen is the bulkiest, but for the time being, let's exclude the emergencies that require you to use stored oxygen! Water is next. How much water do you need?
At Burning Man, in the desert, you needed about 8 liters of water per person per day. A year's worth, then, is 2.9m³/人, which is an almost insignificant fraction of the 3200m³/人 we're talking about. You could probably get by with less, since arcology air isn't desert air, but why bother?
If you had a centralized water tank with storage for the whole arcology, you'd need only 360m³, which is about 7 m in diameter. While this would provide great resiliency against external supply disruptions, it would only provide limited resiliency against internal prisoner's-dilemma problems; see "Community issues" below. A centralized water tank can also provide gravity-feed water pressure for the majority of the arcology, as described above under "Utilities".
As anyone who's attended a condo association meeting knows, cooperating with your neighbors can be by far the hardest part of living in a community. And the more common resources you share, and the less trusting your neighbors are, the harder this is.
So it seems that the problem of growing a community to create the arcology is likely to be at least as difficult as the physical architecture; you could say that's what sunk Arcosanti.
XXX