Saturday, July 21, 2018

The High Frontier: A Technical Critique

The High Frontier: A Technical Critique
Casey Handmer

"The High Frontier," published in 1976 by physicist Gerard O'Neill, remains the text on space industrialization. The culmination of a series of studies through the early 1970s, it presents a compelling vision for life for millions of people beyond the Earth. Today, in 2018, its themes continue to resonate. As several space companies position themselves to launch humans into space on a permanent, sustainable basis, I decided to write a technical critique of "The High Frontier." What works, what doesn't, and how will companies using it as a blueprint, including Blue Origin, have to adapt its ideas after 42 years of intervening discovery?

In this critique, I will:
  • Begin by restating O'Neill's arguments in the strongest possible light.
  • Examine the assumptions that underlie his quantitative estimates.
  • Identify any misperceptions in need of correction.
  • Establish feasibility bounds and figures of merit, in comparison to the original assumptions.
  • Explore ways in which the vision could be extended with a more modern perspective.

What is the High Frontier?
O'Neill paints a compelling image of gigantic cities in space. Consisting of large pressure vessels capable of housing thousands to millions of people, they would rotate to provide artificial gravity. Separate sections would be used for industry or high intensity agriculture with 24 hour sunlight, enriched CO2, and precision hydroponics. The living sections could be landscaped in any desired style - O'Neill is a particular fan of Tuscan architecture. Thus protected from the vacuum of space, people could conduct routine commerce and industry of practically any sort, freed from the constraints of Earth's gravity, limited resources and available space.

In 1976, Earth's resources seemed more pressed than ever. With population growth since 1945 averaging 2% per year, and energy consumption 7% per year, it seemed quite likely that by the early 2000s, Earth would be consumed by the throes of a Malthusian catastrophe. Expansion of industry to space, with its substantial resources, was a relatively cheap alternative. Not long after the book was published, growth stabilized at a more sustainable pace, and in the last few years energy and economic growth seems to have begun to decouple from hydrocarbon extraction, potentially freeing humanity of a major energy constraint.

A gigantic rotating space city would weigh millions of tonnes and could not possibly be launched in one piece, or even in sections from Earth. Once in space, it would continue to require a variety of raw materials to make up for gradual entropic losses, and a versatile, vibrant industry to process raw materials into any of the thousands of products needed to keep this island in the sky functioning.

While launching supplies from the Earth is possible, there are other sources of bulk raw materials in space with much less gravitational binding energy, including the Moon and asteroids. O'Neill envisioned an electromagnetic mass driver system that could launch small slugs of partially processed moon rock from a series of mines on the Moon, to the city somewhere nearby. He and his students even built a series of mass drivers to demonstrate the technology, which is today considered reasonably mature. Extremely high accelerations over short distances reduce the amount of infrastructure needed, while a stream of 1kg packets fired every second or so is adequate to provide a city with enough raw materials to make up for loss and use. Today, we know of numerous low energy transfer orbits which could convey material from the moon to almost anywhere in cis-Lunar space.

With industrial self-sufficiency sketched out, O'Neill turns to the question of economic self-sufficiency. Building and operating a space city would be enormously expensive, and would continue to require shipments of specialty components, such as computers, from Earth, essentially forever. What, then, could the space city offer Earth in order to complete the trade? While the city has a supply of moon rocks, O'Neill recognized that there wasn't concentrations of sufficiently valuable minerals on the Moon to make it worthwhile for selling in an Earth-based market. The city in space had become a solution looking for a problem.

O'Neill settled upon space solar power as the source of space-derived value necessary to balance the trade. Space solar power has certain advantages: The product could be beamed to Earth without having to go through the trouble of re-entry. The product was available essentially 24 hours a day, while on Earth the sun sets at night. Recall that in 1976, energy storage technologies such as batteries were barely on the horizon.

In addition, with 7% per year growth in energy demand, the US and every other country was looking at major infrastructure investments to meet demand for the foreseeable future. The entire grid would have had to have been overhauled every few years, at significant expense.

The final part of the problem that O'Neill examined was bootstrapping. What was the minimum viable space city that would have to be launched from Earth, could support enough people to build most of its structure from Lunar material in space, and then grow from there? Over a series of studies, O'Neill and his team were able to reduce the size of the minimal investment to only 250 shuttle launches, or 6500 tonnes. In 1976, the shuttle had not yet flown and 250 launches was thought to be about five years worth of flights, which seemed quite reasonable even without expected follow-on improvements in launch technology.

Quantification of assumptions
Thus far, O'Neill has sketched a compelling vision, a future for humans in space, unfettered by the limitations of Earth. At no point does settling the High Frontier require any miracles of technology, warp drive, or an investment equivalent to the entire GDP of the Earth for 100 years.

That said, there are still a number of assumptions, all of which need to be true, for O'Neill's vision to be viable. The most crucial of these center around the economic question.
  • Space power needs to be lucrative.
  • Space launch needs to be cheap.
  • Space industry needs to be compact.
  • Overall, the space cost multiplier needs to be low.

I will begin with the space cost multiplier question, since it's the highest level estimate possible in this case. In 2018, the per-person-day-in-space cost of the space program is about $4m. The per-person-day-on-earth cost for people of equivalent qualifications is about $400, so the human space operations cost factor is about 10,000. It is not necessary, or particularly useful, to be more precise than this. On the other hand, the value of the solar resource in space is about 3x higher than on Earth, since about ⅔ of the sun's light is blocked by the Earth during the night, or by terrain, clouds, or atmosphere. For places close to the poles, this number can increase by perhaps a factor of 10.

So even if it were possible to build gigantic solar arrays in orbit, and to capture and transmit the power to where it was needed on Earth with zero losses, the operating costs in space could be no higher than 3x that on Earth, in order to remain competitive. This is the major problem with space-based solar power. Fundamentally, the sun's light is not scarce enough on Earth, nor valuable enough on a per-m^2 basis, to make it worthwhile.

While beyond the scope of this essay, I will point out that there is no practical constraint on available area for solar panels on Earth. Solar will always use a small fraction of the area devoted to agriculture, because eating plants is about 10,000x less efficient than solar for producing mechanical work.

In fact, if we examine current and near-future commercial uses of space, the vast bulk of the industry is supported not by high power transmission of raw microwaves containing electricity, but by low power transmission of structured microwaves, containing information. The per watt value of microwaves is the fundamental question, and beaming TV, GPS signals, phone calls, and internet, is a much more sustainable business model than beaming raw power which is freely falling down anyway!

The problem with space-based communications as a a business model for space cities is that satellites are much cheaper to build and launch from Earth than to build them in space, since they are of quite small size and involve advanced, complex manufacturing.

There is an additional source of risk for long-term space infrastructure project financing, which is that not only is solar power very cheap on Earth, with a record low wholesale price of $0.023/kWh in 2018, it is getting cheaper by leaps and bounds every year. Even if a space solar power concept was valuable on paper in 2018, there is every chance that continued investment and development on Earth will lower the price so much that, like gas-powered peaker plants, a space solar power station will become a stranded asset and major loss.

Let's examine the other assumptions in turn, while bearing in mind that the total cost increase for space based power can be no higher than 3x that on Earth, corresponding to the relative value of the solar resource under the generous assumption of zero losses in transmission.

The cost of space launch must be cheap!
In 2018, the cheapest space launch was by SpaceX, at about $2000/kg to LEO. For much of the previous 3 decades, 10x this was commonplace. Yet, in "The High Frontier", O'Neill speculates that regular flights by the shuttle will reduce launch costs to the order of $100/kg. The math on reusability is much the same today, but despite recent advances in this area, improvement by another order of magnitude remains in the realm of speculation. By comparison, the cost of shipping bulk products or materials around the world on ships and trains and trucks is about $0.05/kg. At that price, an Earth-based solar panel can be shipped through every country on Earth during its production process and still have negligible increase in marginal cost compared to launching it to space.

Space-based industry must be compact!
O'Neill doesn't write from experience with industrial processes either in the 1970s or today. However, our modern economy splits machine and human labor about 1000:1 in energy terms, and as a result, any further developments for space applications will require even more automation/mechanization than is used today. There just aren't that many factories that make complicated things that aren't so big one needs a bicycle to get around. I've written at some length ( about the difficulties of space-based industrialization, and it's not clear to me that any metal-processing nearly self-sufficient city can get by with fewer than, say, 10,000 people. The practical upshot is that throughout the bootstrapping process, the space city will remain dependent on bulk shipments of supplies from the Earth, unless and until a compact primary and secondary manufacturing paradigm can be demonstrated. It is worth stating that such an industrial system would be extremely valuable for nearly any country or city on Earth, and that despite dozens of attempts, no geopolitically isolated country has achieved it.

At this point, the view is rather bleak. It turns out that O'Neill's space solar power gold mine is not viable. Does that mean the dream of millions of humans living and working in space is dead? Not quite. In this section, I explain how future technology development can help bring this vision about, through:
  • Money consumption assumption relaxation
  • Value-added manufacturing
  • Cheap launch
  • Compact industrialization

First, there is an implicit assumption that replicating the entire industrial stack in a pitiless vacuum has to make money, and on the same sorts of time scales as other large projects. Yet there is an obvious difference in scale between, say, a large bridge or chemical plant, and building an industry larger than all but a few nations on Earth. Moreover, large scale infrastructure investments are routinely made by nations with no prospect of any sort of short term return. Expenditure on health, education, defense, and other big ticket items is maintained for a variety of other reasons. Indeed, the largest expenditures by governments promote the general welfare of the nation and wider economic benefits.

I personally think there is no way to mine the moon, asteroids, or build cities on Mars and make money while doing it. I think it's worthwhile to find ways to maximize the results of a given expenditure, but fundamentally any large scale movement of humans into space will be a net consumer of wealth, to the tune of many billions per year, for decades. This expense, which is quite affordable at the national level, will employ and develop many industries on Earth, and will likely involve many of the same contractors and people as the current big ticket defense contracts do! In other words, it will not necessarily involve any new expenditures, just a slight change of course from weapons systems toward space factories.

This is not to suggest that it is pointless for cities in space to sell things in Earth-based markets, even if selling at a loss. The problem is to identify things that are worthwhile enough to bother. This is a tough problem, since nearly everything that humans needs on Earth can be found on Earth and at competitive, commodity prices. Humans in space will need to extract resources, grow food, and maintain their machines with asteroid- or moon-derived raw materials. But while such primary manufacturing is necessary to build a city in space, it's unlikely to be competitive on Earth, which has its own supply chains and breathable air.

What, then, does a good product for production in space and sale on Earth look like? It must be
  • Not readily and cheaply available on Earth. So, not a bulk commodity, like water, dirt, salt, electricity, humans, good TV, or anything in the McMaster-Carr catalog.
  • Not incredibly difficult to make, or requiring lots of human labor. Advanced composite aircraft, cars, aged care, childhood education, cutting edge silicon, medical imagers.
  • High value-added manufacturing. That is, something where the sticker price is much, much higher than the cost of the constituent components or raw materials. The German economy is built around this sort of manufacturing, and it is a comfortable middle ground between bulk commodities, on which there is no margin, and manufacturing nightmares, which require huge scale.
  • Transportable to Earth, if a physical product. So not hugely susceptible to the shocks and forces of reentry. Also, quite small in size, and with a very high value per mass. It is possible to estimate how valuable per kg something would have to be on the Moon to make it worthwhile to transport to Earth, and it's on the order of $10m/kg. This is comparable to enriched plutonium or tritium, and those markets are very elastic. That is, their value is linked to rarity. Also, the Moon doesn't have any bulk deposits of enriched plutonium lying around. With mass drivers and space cities, this cost premium would come down, but not by much. As a rule of thumb, it must be at least as valuable, in bulk, as any vanishingly scarce commodity on Earth. Perhaps the elixir of eternal life?
  • Relatively easier to make in space than on Earth. Perhaps some weird zero G crystals? Optical fibers? Unique tourist experiences?

It is hard to say in advance what advanced manufacturing might be possible and valuable in space. This is another reason to build a city in space with a primary mission of self-sufficiency, and a secondary mission of trying to be useful.

Finally, cheap launch and compact industrialization. These relate to the two main factors for the cost of building a city in space. How much stuff is needed, and how much does it cost to launch?

In my book on Mars industrialization (linked above), I estimated that a million people and a million tonnes of cargo would be needed, over several decades, to achieve full self-sufficiency. For a space city near the Earth, full self-sufficiency is not required, but overall mass requirements are higher as there is no readily available raw material. A million tonnes sounds like a lot but that's only three fully loaded container ships of the largest size. Additionally, it accounts for high technology equipment, and is separate from whatever cargo is launched from mass drivers on the Moon.

A million people and a million tonnes is actually a substantial reduction over the current state of the art, representing a further advance in mechanization of labor equivalent to the total advances to date since the industrial revolution. This industrial compactification is necessary to achieve (near) self-sufficiency in an adversarial environment, in just the same way that nuclear submarines weren't possible before steel was invented.

At current launch costs and technology, a million tonnes to LEO would require on the order of 100,000 launches at a total cost of $5 trillion dollars. If launch is to be, say, 10% of the overall program cost, and the program is budgeted at $10b/year for 50 years, then the total launch budget is $50b. This requires a reduction in launch costs of a factor of 100. To put that in perspective, that's close to the long-term goals of the SpaceX BFR or the Blue Origin New Armstrong, and could only be achieved through complete and rapid reusability of the launcher. This is not forbidden by the laws of physics, but it is a big ask.

Giant cities in space are possible, but require ongoing nation-state level funding, continued aggressive technology development in both launch and industry, and a willingness to think well outside the box when it comes to monetization.