Solving the Mass Problem

The key to accelerating space settlement might not be any clever architecture or new technology, but simple brute force.

The vision of human settlements in space has been clear to many people for a while. Practical designs for Mars bases abound, and there have been many iterations of free floating habitats since Gerard O’Neill wrote The High Frontier in the 1970s. These concepts have been studied and found to be technically feasible, but today we don’t seem to be much closer to seeing them realised than we were in the 60s and 70s, when NASA was hoping to land men on Mars by 1981 and Libertarian Bernal Spheres would save us from energy crises. The only part of the current space sector that really offers a connection from the current state of things to that long imagined future is SpaceX Starship, and many people I talk to seem to very skeptical it can do what Elon Musk says it can.

One thing that would be required to move from the present situation of human spaceflight - two space stations with 10 crew between them, each typically staying for a few months - to a more expansive future is simply being able to throw more mass into space. This is something SpaceX understands, and is already achieving through increasing the cadence of their Falcon 9 rocket. Musk himself has explicitly stated that mass flow to space is a driving metric of the company

What I will be arguing here goes a little further; not only is a key requirement for moving out into space, it is in fact the only one that really matters. Almost all other problems that have to be solved getting from where we are now to the space future we want are reducible to mass problems.

Cosmic radiation? Sending about 2 tonnes of shielding for every square metre of spacecraft hull will sort that out, so its a mass problem. Closed cycle life support? Don’t bother closing the cycle. Just send more food, air and water to make up the shortfall. Weightlessness? Easy - just build a more massive spacecraft that is 100m wide and rotate it. If stability is a problem with your rotating spacecraft, send more fuel for control thrusters. In principle, you can always give your spacecraft enough mass that it generates its own gravity, but at that stage you are building a planet.

A lot of hardware in space can be made simpler and cheaper by lifting mass requirements - a good example is the Soviet Zenit spy satellites, which rather than developing a camera and electronics that could function in the vacuum of space, kept essentially off the shelf hardware all inside a large (and massive) pressurized return capsule - one that shared a design with the manned Vostok spacecraft that carried Yuri Gagarin. American spy satellites were much less massive, but couldn’t take this shortcut.

Focusing this tightly on mass might seem like a very reductive way to look at things - but being the sine qua non of space activities, the need for mass is one thing that can be confidently predicted about the future of spaceflight without knowing any other details. When good engineering of payloads matters, it usually matters to the extent it reduces the mass required for present or future launch - and that is, I believe, a good way to measure its utility.

Mass as a Metric

The Moon race was primarily a race to send mass to the Moon. The largest (functional) Soviet rocket, the Proton, could send about 4 tonnes to TLI, whereas the largest US rocket could sent 45 tonnes to TLI. Even though there were other technological advantages NASA had in flight computers and rendezvous experience, the Soviet program simply did not have the lift capability to compete. They would have at least needed the N-1 rocket to function (which could send about 30 tonnes on TLI) to even have a shot at beating the Americans.

Mass delivered to orbit was a useful metric in this particular case. But it doesn’t capture everything that we might be interested in. Future space missions are likely to use mass from in-situ resources. SpaceX intends to refuel Starships on the surface of Mars using methane and oxygen derived from the CO2 in the Martian atmosphere. There are various plans to extract propellant and other resources from the surface of the Moon and from asteroids. To capture such things within the same metric as launch capability, I use mass value defined as

The mass that would have to be placed into LEO to transport the mission payload to its final destination, using a standard basic method

The calculation is detailed in this paper - the main difficulties being ensuring that close to LEO the mass value converges on the payload mass, that a mass value is calculable for any destination regardless of the Δv required to get there, and that it does not reach extreme values for any destination that we might be interested in.

I discuss the details more here, but as an example, consider the Mars Direct proposal. This would have enabled a human Mars landing using 2 heavy lift launches, equivalent to about 240 tonnes to LEO. Performing the mass value calculation though, it would yield over 5,000 tonnes equivalent, by virtue of the fact that the Earth Return Vehicle generates around 100 tonnes of propellant on the surface, mainly from local resources.

This principle will also be applied to Starship, but the larger vehicle will need (using these estimated figures for Starship supplied by GrandpaJoe) about 500 tonnes of propellant for the return trip - equivalent to 25,000 tonnes in LEO using the same calculation above, or an additional 250 Starship payloads. The merit of this approach compared to carrying propellant from Earth should be clear.

Exponential Growth

Technologies which generate mass value also consume mass value. This means that, potentially, the rate at which new mass value capacity is added would be proportional to the current mass value rate. That would mean that the rate would increase exponentially.

Let us say current mass value, globally, is 1 million kg equivalent/yr. A feasible near term space habitat, the Kalpana One design, has a mass of 7 billion kg - lets call it 10 billion kge to account for the fact it needs to orbit at the higher end of LEO to avoid drag. Throwing mass up using current capability it would take 10,000 years committing all of our resources to just send the construction materials to the site.

However, our capacity to throw mass into space is not static. The Falcon 9 for instance has shown substantial growth in its annual mass flow to LEO - below I have hand fitted a rough exponential to their launch data (multiplied by a payload mass estimate of 15 tonnes per launch)

The bar in orange is a guess for the total launches this year. It seems an anomalous outlier so I don’t worry about the curve hitting it. This is only a very rough estimate anyway. The line corresponds to a growth rate of 35% per year - a little bit slower than Moore’s law, and doubling roughly every 2 years. If - presumably by moving on to Starship - SpaceX were to maintain this pace for the long run, how long would it take until the mass of Kaplana One were comparable to the annual mass rate of SpaceX launches? It’s easy enough to answer, but we have to switch to a log scale for the current launch output to be even visible

At 35% per year, the “affordability” point in terms of mass alone comes in 2055. This is a massively oversimplified model, but 33 years is a strikingly different result from 10,000 years as a timescale. What if the growth rate is not 35% though? That was an eyeball fit to the current launch data. What if the actual long run rate were 30% or 40% instead?

At 30% the mass rate would be reached in 2061. At 40% it would be reached in 2050. Rates within this range are arguably compatible with the launch data, depending on which years you want to consider outliers for whatever reason.

Predicting the Future

An accurate long run growth rate can’t be predicted from just counting Falcon 9 launches. The point I wish to make here is that if there is a continuing exponential growth, it radically changes the rules of what is possible in space, on timescales that we care about. I was born in 1981 and reasonably expect to still be around in 2055. My children will be in their early 30s, and may wish to live or work in space. Whether or not this capability is available by that date impacts our lives.

But as I also showed, such predictions are sensitive to inputs. Reasonable fits to the existing data can shift the timescale by a decade. It is also by no means certain that the current trend continues. Certainly it can’t go on for much longer with Falcon 9, due to limitations such as the manufacturing rate of the expended second stage - so SpaceX would have to have Starship pick up the baton and drive further growth. Other companies may then introduce their own fully reusable launch vehicles. Further on, space resources and recycling of used materials may provided additional sources of mass without requiring additional launch. Each innovation is likely to follow a sigmoid function - an exponential-like rise and then leveling off - but the accumulation of them could produce an overall exponential increase in useful mass.

Its worth revisiting the comparison with Moore’s law. Each doubling of transistor count was not just a repeat of the same technology - each node entailed a new R&D program. Yet the cumulative effect of them was a single exponential that held over decades. There are several ideas as to why this is - one common one being that it is a self-fulfilling prophecy. Once that pace was established, competing manufacturers knew that they had to achieve it in order to stay in the market. Such a pace setting effect could occur in the space industry - not just in launch, but in space services that save launch mass.

I would go so far as to consider this quite likely unless SpaceX fails to realise Starship - because after all, anybody asking for investment in space manufacturing or ISRU will have to answer the question “Why not send stuff up from Earth instead?”. If the pace set for general mass value increase is anywhere close to the current pace in launch mass increase, we are standing on the cusp of a revolution.