SpaceX has been working on some variant of the Big Falcon Rocket for almost a decade, with a publicly announced architecture for three years. The target performance figures are on the Starship website, endlessly dissected on Twitter, Reddit, and NASA spaceflight forums, and there’s even a livestream of construction.
Yet none of the oft-published mainstream articles seem to capture the magnitude of the vision that Starship embodies. Starship prompts superlatives, but by the end of this post the reader will understand not only how big Starship is, but also that it’s as small as it can possibly be.
Since Starship was first unveiled as BFR/BFS at the IAC in 2016, the rocket has undergone a number of design changes. Together with Elon’s rather cryptic statement that the hardest part of the Starship design process was understanding exactly what question it would answer, we have a golden opportunity to employ reasoning backwards from design to implicit requirements.
Let’s start with a quick recap of the essential numbers.
Starship is the upper stage vehicle. It has a dry mass of 200 T, a fuel/ox mass of 1200 T, and a nominal payload of 150 T. Combined with high performance methane-oxygen vacuum engines, Starship is capable of over 7 km/s of Δv, which is very important.
Starship is boosted for Earth launch by Super Heavy, which is capable of lifting Starship to about 4 km/s before returning to the launch pad.
Both stages are designed to be fully reusable, enabling both high reliability and very cheap launch cost. Indeed, the marginal cost per flight could fall to $5m or below, reducing launch costs to the neighborhood of $35/kg, or 1000x less than Shuttle.
In 2016, it wasn’t immediately apparent how SpaceX intended to fund Starship development, but by 2017 the development team seemed to have an answer. Even though Starship has a launch capability greater than any other rocket ever built, its launch cost is comparable to the smallest launch vehicles currently in service, such as Electron.
If flights were sold at the going rate, Starship could launch any payload currently in the market, while making money. In practice, of course, Starship will aggregate launches to spread cost, but it still was not entirely clear how Starship would not saturate the global launch market immediately.
SpaceX clearly intends to build dozens of Starships. With an eventual flight rate of once per day per Starship, we’re looking at roughly a million tonnes to orbit per year. That exceeds the current launch market of about 500 T/year by a substantial margin.
Economically, there is an answer to this question in the form of Starlink, SpaceX’s communication constellation. In 2012, SpaceX realized that their customers, primarily communications satellite operators, had better margins than they, the launch provider, did. And this was despite the generally usurious launch costs prevalent in the industry. Unlike space mining or solar power, space-based communications satellites fulfill a genuine market need. If satellite manufacturers can make money launching $100m satellites on $100m rockets, how much money could SpaceX make launching their own satellites for much, much less? More on this in a future post, but in short, SpaceX believes (correctly) that there is almost infinite upside to flying satellites to meet global telecommunications demand, provided you have a monopoly on launches that are orders of magnitude cheaper than the competition.
With the money question out of the way, we can ask ourselves: What is Starship for? SpaceX could make plenty of money with incremental improvements on the Falcon launchers, and even build up Starlink without Starship.
Starship is for building nations in space. I don’t mean Project Artemis or some version of “The Martian,” though Starship could easily do both. I mean serious logistics.
The Starship atmospheric test prototypes are deep into construction, and I still routinely hear space visionaries adapting old exploration architectures for Starship. With a Starship or two, they say, I could build a really sweet Moon mission. With a Starship or two, I could upgrade the space station.
Think bigger. Much bigger.
In 10 years, Starship flights will be sold by the dozen. Starship is only cheap if it gets used as much as possible. The only meaningful barrier to production today is engine manufacture, and that will be through the hardest part of the learning curve well before commercial flights begin.
When we think about how to use Starship, we can’t analogize to Christopher Columbus or Captain Cook. In fact, let’s avoid analogies entirely. But if you must use an analogy, think instead of the Berlin Airlift. In 1948, allied forces in Berlin were blockaded by Soviets to force capitulation and withdrawal. Instead, the remnants of the demobilized airforces began shipping in food, coal, and other supplies in 3.5 T (and later 10 T) increments – the cargo limits of the available planes. Daily demand was 5,000-10,000 T as the city neared winter, but the operation scaled rapidly. Over 15 months, 278,000 flights moved 2.3 million tonnes of cargo into the city, breaking the blockade. The Berlin Airlift functioned as a virtual conveyor belt, moving cargo from one point to another.
Starship is a cargo conveyor system. The Starship is comparable in complexity to a 737, and so it’s not unthinkable to have a construction rate of 500/year. If each Starship manages 300 flights per year, each carrying 150 T of cargo, then we are talking a yearly incremental cargo capacity growth of 22 million tonnes to orbit. At this point, the most meaningful constraint on launch capacity might be launch pad construction rate.
Not all this cargo will be bread and cheese. In fact, by mass around 90% will simply be more methane and oxygen to refuel the LEO-parked cargo-laden Starships for flights to more distant destinations. Much of the remainder will be heavy industrial equipment as the purpose of these missions is to replicate the industrial stack of Earth.
How will these missions work?
The Starship mission profile is somewhat dependent on destination. Starship is designed to be a versatile cargo vehicle, capable of landing on any terrestrial body in the solar system, but the details vary somewhat. The destinations I’ll cover here include Earth, Moon, Mars, and deep space.
More Starships will land on Earth than any other planet. In fact, for each Starship loaded to the gills with cargo and launched to LEO, another eight or so are needed to completely refill its tanks for further flights. Each of these Starship tankers will transfer fuel directly (there are no gas stations in space) and then return to the launch site for another load.
Starship has a superficial resemblance to the Space Shuttle, with its stubby delta wings. Unlike Shuttle, however, these wings provide no lift, and they hinge along the body. Starship is designed to control its entry using differential drag, much like a sky diver. It can angle its body to produce a small amount of body lift necessary to avoid sharp deceleration and heating in the lower atmosphere. Unlike Shuttle, which glided subsonically to an absurdly long runway, Starship will relight its core engines, turn around, and execute a propulsive pinpoint landing. Just like a futuristic spaceship should.
Since Earth’s atmosphere is nice and thick, Starship can slow down to perhaps 150 mph before lighting its engines, requiring only a tiny amount of fuel to land.
The Moon is a different story. Unlike Earth or Mars, it lacks an atmosphere so requires propulsive landing and the fuel to slow down from orbit. Despite this drawback, which in alternative architectures requires multiple stages to execute, Starship has an enormous Δv and can fly significant cargo from LEO to the Lunar surface and back to LEO, all without refueling on the Moon.
This is vitally important. While converting Lunar water to rocket fuel is all the rage right now, it won’t be available for early missions, if ever. Oxidizer would perhaps be easier to obtain and, being 4/5 of the fuel weight, more worthwhile. Nonetheless, this graph shows how a Starship, refueled in LEO and optionally boosted to GTO, could transport 150 T of cargo the Moon and still have ample fuel to bring another 100 T of cargo back. A LEO-GTO refilling operation would require 15 tanker flights, and would take about a week to execute from TLI burn to Earth return.
In other words, the baseline Starship architecture is capable of moving enormous quantities of cargo to anywhere on the Moon with little to no preplaced infrastructure. Assuming a $35/kg LEO launch cost, cargo could be delivered to the moon for perhaps $500/kg. This sounds like a lot, especially when the cost of shipping anything in a container, anywhere on Earth, is about $0.10/kg. But $500/kg is comparable to the cost of specialty industrial equipment, which is a decent baseline for the cost of Lunar-customized gear. In other words, for the first time in any Lunar mission reference architecture, the shipping cost could be a minority of total costs.
Next, let’s consider Mars. Mars is the underlying design case for the Starship, although Starships that go to Mars will probably be of a customized nature. While the Moon takes about 3 days to get to, Mars takes 4-10 months. Launch windows to Mars only open every 2 years, so even if Elon has 1000 Starships ready to go, they can only go every 2 years. If there is copious fuel available on Mars, they could potentially return in time for the next launch window, but even so, 2 years is an awfully long time to wait for a reusable vehicle that depends on high flight rate. Starship needs to refuel on Mars from locally-produced methane and oxygen to return to Earth.
The economics shift here so much that some people even advocate employing Starship as an upper stage alone. Under this model, Starship would boost some Mars-bound payload onto the appropriate orbit, then turn around and come back to Earth, ready to fire off another payload as soon as it was refueled, perhaps a week later. Such a system could employ a given Starship 100 times as much.
There are two reasons why staging off Starship is a bad idea.
The first is that Starships are not fundamentally scarce. By the time we are launching thousands of people to Mars every launch window, there will be so many Starships we’ll be giving them away. There may not be much point in flying them all back to Earth, but a few will return to bring people who want to come back.
The second, and most important, reason is that Starship is not just a huge fuel tank and some kick ass engines, it’s also a Mars landing system. It turns out that landing on Mars is really hard. Of 17 attempted landings, only 8 have been successful, all built by JPL. Here’s a great explainer of the general problem.
In short, the atmosphere is really thin. Thick enough to cause a thermal and guidance problem, but not thick enough to slow the spacecraft down enough, especially if it’s a really big (>2 T) lander.
Mars Direct originally called for a blunt aeroshell and parachute landing system, but such a system has several fundamental issues. First, the aeroshell produces inadequate levels of lift, greatly restricting landing sites to very low altitude areas that aren’t downrange of mountains. Second, the aeroshell’s low lift causes the spacecraft to experience a very high G loading, potentially at levels that would incapacitate a human crew. Third, parachutes have proven very tricky to get right on Mars even for rover-sized payloads. Parachutes are problematic because a) their mass scales much faster than their size and b) their opening speed reduces as their size increases. Well below the size needed to be useful for human landers, parachutes weigh too much and open much too slowly to successfully arrest a lander.
Fortunately, there is a solution and Starship leverages that solution. By bringing the spacecraft in sideways, the vehicle maximizes its area and lift, reducing g loading and ballistic coefficient. It can also perform a guided entry and precision rocket landing. Starship is not just a huge fuel tank and some kick ass engines, it’s also a reusable honest-to-god Mars landing system for >100T payloads that doesn’t require any parachutes. This is a big deal.
The Starship is designed to enable the rapid construction of an enormous, self-sufficient Mars city with a population measured in the millions. This can’t be done by analogy with western pioneer expansion in the USA. 20 acres and mule isn’t enough to survive on Mars. Mars is a deadly adversarial environment and humans need copious advanced technology and production capacity to survive. Mars city building isn’t about Mark Watney lovingly raising 500 sqft of potatoes in his hab. Mars city building is about robots building potato farming robots and humans facilitating a nearly-automated industrial stack. Mars city building is about tenting and pressurizing 10,000 sqft of the Martian surface, per person. Mars city building is about maintaining an exponential growth curve for 6 orders of magnitude. It’s an enormous logistical enterprise.
What is the optimal size for Mars Starships? It turns out that the practical limitation for Mars Starships is set by entry, descent, and landing. Without terraforming the atmosphere, this tops out at somewhere around 1000 T of cargo. Still, a 1000 T cargo Starship would be 25m wide and perhaps 70m tall. The 150 T Starship (the current version) is just barely viable for Mars cargo missions, but useful for proving out the architecture, testing the systems, and getting things started. Fortunately, welded stainless steel is much easier to increase in diameter than virtually any other rocket manufacturing technique. This is what I mean when I say that while Starship is bigger than any of our other puny rockets, it’s actually quite small compared to the magnitude of its destined task.
Finally, deep space, by which I mean any destination that isn’t a major planet or moon. Starship’s pressurized volume is about double that of the International Space Station. A space station built from Starships, its tanks converted to habitable space and equipped with electric thrusters, could be used for any imaginable purpose. Earth orbit space station? Asteroid exploration? Deep space telescope chassis? Outer solar system exploration? I’ve written before that modular space stations are not cost effective. Perhaps what we need are fewer, much larger modules. After all, there’s nothing cheaper than developing a new heavy lift rocket!
To finish this post, I’m going to revisit its starting point. Starship is still seen by many in the space media community as a slightly overgrown version of any other rocket, with reusability tacked on. This is an error of analogy. Starship fundamentally changes our relationship with space.
Starship is a devastatingly powerful space access and logistical transport mechanism that will instantly crush the relevance of every other rocket ever built.