Starship is Still Not Understood

Another entry into my blog series on countering misconceptions in space journalism. I discussed this post on The Space Show on November 5 2021.

It has been exactly two years since my initial posts on Starship and Starlink. While the Starlink post has aged quite well, Starship is still not widely understood despite intervening developments. As usual, this blog represents my own opinions and I do not have any inside information.

There is now an audio version of this blog.

[Update Oct 2023: Ars reports that senior JPL and NASA observatory advocates have testified that Starship’s capabilities must be exploited for next generation telescope design.]

To catch you up, two years ago SpaceX unveiled their boilerplate full scale mockup of Starship. Starhopper had completed two untethered flights. SN5 and SN6 hopped to 150 m in August and September of 2020, followed by 10-12 km flights of SN8, SN9, SN10, SN11, and SN15 between December 2020 and May 2021, the last of which stuck the landing.

Two years ago, Raptor was unproven, aero flaps had never been demonstrated, and stainless steel rocket construction was still troubled. Today, these major programmatic risks are largely retired. SpaceX has qualified their full flow staged combustion engine. They’ve done a full system test of the landing process, and they’ve ramped up QA in construction. There are still major risks on the critical path between now and a fully reusable Starship, but no miracles are required to solve them. For example, many mature heat shield (TPS) designs already exist. SpaceX can try to make a better, cheaper, lighter one but if it doesn’t work out, they can always trade some mass and just use PICA, like Dragon. In just two years, practically all the low TRL science projects have been solved.

SN9 on the pad (wikimedia).

As of late October 2021, SN20 and the booster SB4 have performed basic fit checks and individual static fires, while the ground support equipment and the launch tower are being assembled with truly gigantic cranes. The Boca Chica rocket factory and launch site are now enormous ongoing operations, as seen in this video tour with Tim Dodd, the Everyday Astronaut.

While I am 100% certain that the Starship design will continue to evolve in noticeable ways, the progress in two years cannot be understated. Two years ago Starship was a design concept and a mock up. Today it’s a 95% complete prototype that will soon fly to space and may even make it back in one piece.

The odds of Starship actually working in the near future are much higher today than they were two years ago. Across the industry, decisions are being made on a time horizon in which Starship operation is relevant, and yet it is not being correctly accounted for.

Starship matters. It’s not just a really big rocket, like any other rocket on steroids. It’s a continuing and dedicated attempt to achieve the “Holy Grail” of rocketry, a fully and rapidly reusable orbital class rocket that can be mass manufactured. It is intended to enable a conveyor belt logistical capacity to Low Earth Orbit (LEO) comparable to the Berlin Airlift. That is, Starship is a powerful logistical system that puts launch below the API.

Starship is designed to be able to launch bulk cargo into LEO in >100 T chunks for <$10m per launch, and up to thousands of launches per year. By refilling in LEO, a fully loaded deep space Starship can transport >100 T of bulk cargo anywhere in the solar system, including the surface of the Moon or Mars, for <$100m per Starship. Starship is intended to be able to transport a million tonnes of cargo to the surface of Mars in just ten launch windows, in addition to serving other incidental destinations, such as maintaining the Starlink constellation or building a big base at the Lunar south pole.

The fact that Starship flown expendably would be perhaps 10 times cheaper, in terms of dollars per tonne, than even Falcon is not relevant. For the last two years, space community responses to Starship can often be summarized as “Starship would be awesome! I can customize one or two and do my pet mission for cheap.” This is true, but it misses the point.

First, SpaceX is unlikely to spend a lot of engineering effort doing custom one offs for otherwise obscure science missions. Find a way to fit the mission in the payload fairing and join the queue with everyone else trying to burn down their manifest as quickly as possible.

Second, and more importantly, shoehorning Cassini 2.0 or Mars Direct into Starship fails to adequately exploit the capabilities of the launch system. Not to pick on Cassini or Mars Direct, but both of these missions were designed with inherent constraints that are not relevant to Starship. In fact, all space missions whether robotic or crewed, historical or planned, have been designed with constraints that are not relevant to Starship.

What does this mean? Historically, mission/system design has been grievously afflicted by absurdly harsh mass constraints, since launch costs to LEO are as high as $10,000/kg and single launches cost hundreds of millions. This in turn affects schedule, cost structure, volume, material choices, labor, power, thermal, guidance/navigation/control, and every other aspect of the mission. Entire design languages and heuristics are reinforced, at the generational level, in service of avoiding negative consequences of excess mass. As a result, spacecraft built before Starship are a bit like steel weapons made before the industrial revolution. Enormously expensive as a result of embodying a lot of meticulous labor, but ultimately severely limited compared to post-industrial possibilities.

Starship obliterates the mass constraint and every last vestige of cultural baggage that constraint has gouged into the minds of spacecraft designers. There are still constraints, as always, but their design consequences are, at present, completely unexplored. We need a team of economists to rederive the relative elasticities of various design choices and boil them down to a new set of design heuristics for space system production oriented towards maximizing volume of production. Or, more generally, maximizing some robust utility function assuming saturation of Starship launch capacity. A dollar spent on mass optimization no longer buys a dollar saved on launch cost. It buys nothing. It is time to raise the scope of our ambition and think much bigger.

Apollo was limited by the lift capacity of a single Saturn V to use a lunar orbit rendezvous architecture, in which just two astronauts sortied to the surface for a few hours. Every NASA mission to any planet has to be a marvel of miniaturization, just to cram as much science as possible into a severely mass constrained space craft. The Artemis program to the Moon requires a Gateway and separate Human Landing System (HLS) because even the SLS doesn’t have enough lift capacity to execute the mission on its own. The HLS request specified performance requirements that only make sense if the launchers are not Starship, and are objectively inadequate for any kind of serious base building or long term sustainable presence.

Starship changes this paradigm. Starship won the HLS contract because of the three bids only it delivered a system that actually closed. But more than that, Starship could be used for the entire Artemis program, and probably will if the program continues. Indeed, for the same annual cost Starship could deliver perhaps 100x as much cargo to and from the Moon, meaning that instead of two or three dinky 10 T crew habs over the next decade, we could actually build and launch a base that could house 1000 people in a year or two. We probably won’t, but we could.

This cuts to the core of the problem. Why won’t we upgrade Artemis to actually use the capacity of Starship? Because Starship is somehow less proven or likely than SLS and Vulcan? Please! No, Artemis is still trapped in a pre-Starship paradigm where each kilogram costs a million dollars and we must aggressively descope our ambition. This approach is evidently self defeating.

To make this concrete, compare these two bat charts for pre- and post-Starship Artemis conops.

Conops as envisioned in original Artemis HLS RFP. The two unsuccessful bids followed this model. Each 12 T lander cycle costs at least $6b.

Artemis designed around Starship capability looks completely different, because it is. Each 100 T lander cycle costs less than $100m.

Even though Starship was selected for HLS, Artemis hasn’t been redesigned, because Starship is still not understood at the organizational level.

Nowhere was this clearer than the September 26, 2021 NASA press conference where Administrator Senator Bill Nelson spent 45 minutes discussing the future of Human Spaceflight at NASA. The town hall was to announce the reorg of Human Exploration and Operations Mission Directorate (HEOMD) into the Exploration Systems Development Mission Directorate (ESDMD) and the Space Operations Mission Directorate (SOMD), reversing an org chart change made about a decade ago.

HEOMD reorg NASA townhall September 26 2021.

My main takeaway from this wasn’t speculation as to whether Kathy Lueders had been demoted, but the observation that in 45 minutes of conversation about the future of human space flight at NASA, Starship wasn’t mentioned once. The gigantic rocket that is poised to improve our access to space by three orders of magnitude just didn’t come up.

I know that SpaceX and Starship are controversial in certain circles at NASA, but what purpose does it serve to maintain a policy of quietly ignoring it forever? I know dozens of people in the US space industry who basically agree with everything I’ve written about Starship, and yet the official policy sails serenely on as though Falcon has never even landed.

Starship will change the way we do business in space, and now is the time to start preparing. Pretending that it doesn’t exist isn’t an adequate strategic hedge, whether Starship flies in 2022, 2025, or never.

What do I mean by strategic hedge? There is a steadily increasing chance that Starship will succeed and total certainty that if it succeeds it will change the industry, therefore the appropriate hedge is to take actions somewhere between total panic that it is already flying, and complete inaction. The cost of preparing and Starship not eventuating is lower than the cost of Starship flying while NASA is still unprepared. As of today, continuing inaction by the legacy space industry continues to accrue fundamental structural risk. Starship is mostly good news. It certainly doesn’t have to be a harbinger of doom, but acting as though it can never change anything serves only to increase the chance that it does bring about negative changes in future.

What sort of negative changes am I referring to? The US space industry has a strategic blind spot in this direction. Ask a room of engineers and scientists what they can do with Starship and the response will be enthusiastic, to say the least. 100 T of science instruments on Titan in just four years? Sign me up! Ask a room full of program managers how they will avoid negative programmatic consequences due to Starship launch capability and you will probably get blank stares.

Let me explain the fundamental issue. NASA centers and their contractors build exquisitely complex and expensive robots to launch on conventional rockets and explore the universe. To take JPL as an example, divide the total budget by the mass of spacecraft shipped to the cape and it works out to about $1,000,000/kg. I’m not certain how much mass NASA launches to space per year but, even including ISS, it cannot be much more than about 50 T. This works out to between $100,000/kg for LEO bulk cargo and >$1,000,000/kg for deep space exploration.

Enter Starship. Annual capacity to LEO climbs from its current average of 500 T for the whole of our civilization to perhaps 500 T per week. Eventually, it could exceed 1,000,000 T/year. At the same time, launch costs drop as low as $50/kg, roughly 100x lower than the present. For the same budget in launch, supply will have increased by roughly 100x. How can the space industry saturate this increased launch supply?

I doubt Congress is going to increase NASA’s budget to a trillion dollars, so NASA and industry will have to find a way to produce 100x as much stuff for 1/10th the price. Rovers will have to be $1000/kg and we will need 100 T of them every year. This is comparable in terms of costs and volumes to Ferrari manufacturing, so we’re not necessarily talking about replicating Toyota’s automated production lines, but we are definitely talking about finding ways to drastically increase the productivity of the current work force, while shifting its skill focus away from mass optimization and towards mass generation. Since the mass constraint really doesn’t matter anymore, there isn’t much point devoting hundreds of person-years of effort into assembling the whole thing from custom machined titanium parts.

This is where the risk to the space industry originates. Prior to Starship, heavy machinery for building a Moon base could only come from NASA, because only NASA has the expertise to build a rocket propelled titanium Moon tractor for a billion dollars per unit. After Starship, Caterpillar or Deere or Kamaz can space qualify their existing commodity products with very minimal changes and operate them in space. In all seriousness, some huge Caterpillar mining truck is already extremely rugged and mechanically reliable. McMaster-Carr already stocks thousands of parts that will work in mines, on oil rigs, and any number of other horrendously corrosive, warranty voiding environments compared to which the vacuum of space is delightfully benign. A space-adapted tractor needs better paint, a vacuum compatible hydraulic power source, vacuum-rated bearings, lubricants, wire insulation, and a redundant remote control sensor kit. I can see NASA partnering with industry to produce and test these parts, but that is no way to service the institutional overhead embodied by a team of hundreds of people toiling on a single mission for a decade. There is a reason that JPL’s business depends on a steady stream of directed flagship missions with billion dollar price tags. Hordes of PhDs don’t come cheap and need a lot of care and feeding.

Even if the space industry fully understood Starship, I think it would be very difficult for them to plan and adapt rapidly enough to match the coming explosion in launch capacity. But it has been two years since my earlier post and the implications were obvious enough even then. Yet I have seen almost no evidence that, on an organizational level, any of the prime contractors or senior NASA leadership have internalized the full implications of the coming change.

History is littered with the wreckage of former industrial titans that underestimated the impact of new technology and overestimated their ability to adapt. Blockbuster, Motorola, Kodak, Nokia, RIM, Xerox, Yahoo, IBM, Atari, Sears, Hitachi, Polaroid, Toshiba, HP, Palm, Sony, PanAm, Sega, Netscape, Compaq, Enron, GM, DeLorean, Nortel. In many cases, such as with Kodak and digital cameras, these powerful corporations even invented the technology that eventually destroyed them. It was not a surprise. Everyone saw it coming. But senior management failed to recognize that adaptation would require stepping beyond the accepted bounds of their traditional business practice. Starship, like Falcon, is built on a foundation of fundamental rocketry research funded and performed by NASA, Roscosmos, and other government agencies. SpaceX has found a powerful new synthesis but they didn’t invent rockets from scratch. Either the incumbent space industry adapts to Starship by finding ways to produce much more space hardware for much lower cost, or dozens of other new companies, unbound by tradition, entrenched interests, and high organizational overhead, will permanently take their business.

Just two weeks ago, former NASA Associate Administrator for Exploration and current Boeing consultant Doug Cooke, gave a presentation on his vision for lunar exploration, as reported by Jeff Foust.

Doug Cooke’s slide on Lunar Exploration Oct 12 2021 (Jeff Foust).

The washed out yellow on black can be hard to read, so I’ll copy the text below [grammatical errors and typos uncorrected].

Logical Early Lunar Architecture and Mission(s)

Architecture

• 130 mt SLS (Block 2) as envisioned in the 2010 Authorization Act.
• Orion as presently configured.
• Develop two-stage, storable propellant lunar lander with not-to-exceed mass of 33 mT.
  • Lander requirements – include cargo mode to land hab(s), rovers, surface infrastructure – separate from crew landings.
• Develop Lunar Orbit Injection (LOI) stage capable of delivering the lander to Low Lunar Orbit (LLO) using efficient Liquid Oxygen/Hydrogen fuel. Same LOI stage design for delivering Orion and service module to LLO.
• Enhance Ground Systems to support this architecture with sufficient flight rate.

Lunar Mission

• Fully fueled integrated lander is launched as cargo on the SLS Block 2 and injected by the LOI stage into LLO to await the crew.
• Crew is launched on SLS to LLO in Orion using the same LOI stage design as for the lander.
  • Several tons of margin for additional cargo
• Orion performs the rendezvous with the lander in LLO
• Crew and additional equipment and provisions transfer to the ascent stage on the lander.
• With the crew onboard, the lander descends from LLO and lands on the lunar surface.
• The crew executes its surface mission
• The crew launches back to LLO in the ascent stage to rendezvous and transfer to Orion.
• The crew returns to Earth from LLO in Orion, using the Orion Service Module to perform the Trans-Earth Insertion (TEI) maneuver.

Follow-on Crew and Cargo Missions to fulfill lunar exploration objectives

Allow me to fill in the gaps. This is 98% similar to the original Constellation lunar program. It requires SLS Block 2, which has a new, upgraded upper stage. This was always meant to be part of Ares V and it’s what has always been required to make SLS actually useful, with real cargo capacity to LEO and beyond. Of course, this Exploration Upper Stage (EUS) is still in the preliminary design phase and may never actually be built let alone flown. In addition to the EUS, which is essentially a whole new rocket, this architecture also requires a Lunar Insertion Stage, also originally called for in the Constellation architecture but long since cancelled, and without which Orion can’t even make it to Low Lunar Orbit (LLO). It also requires a new two stage lander, which is still being treated almost as an afterthought.

When it’s all put together, we have an architecture rather similar to Apollo, only heavier, more expensive, slower, with more moving parts, and with about the same net cargo capacity to the surface. That is, another decade or so of incredibly expensive clean sheet development of four new space vehicles, and for what? The ability to get “several tonnes” of marginal cargo to the surface for two launches of the SLS Block 2, and to finally deliver the Lunar part of Constellation two decades late and at ten times the price, as though it was never justifiably cancelled in the first place?

Consider the two critical metrics: Dollars per tonne ($/T) and tonnes per year (T/year). Any effective space transport cargo logistics system must aggressively optimize both these metrics simultaneously. Starship is intended to reach numbers as low as $1m/T and 1000 T/year for cargo soft landed on the Moon. Apollo achieved about $2b/T and 2 T/year for cargo soft landed on the Moon. Constellation 2.0 as described above would be more like $4b/T and 2 T/year.

Not only is this architecture obviously worse than Starship, it’s also significantly worse than Apollo or any existing lunar delivery system. For example, the Blue Moon lander could be flown on Falcon Heavy, delivering perhaps 10 T to the surface for <$200m. Indeed, the Constellation architecture is worse than the current state-of-the-art by roughly the same factor that Starship promises to be better. That is, it takes the key metrics of $/T and T/year and runs as far as possible in the wrong direction. It is also a programmatic dead end, since none of the individual components can be upgraded in a meaningful way without restarting development of the entire system from scratch. It’s an expensive, interlocking failure. What “lunar exploration objectives” can be “fulfilled” with such an architecture? There is no possibility for a sustainable program, no possibility for continuous human presence or base building. Just tens of billions of dollars on obsolete hardware serving ill-defined programmatic goals that lost their geopolitical relevancy on July 24, 1969.

Obviously it is NASA, Cooke, and Boeing’s prerogative to propose programs that serve their particular respective interests, but what I don’t understand is how they can seriously think that ignoring Starship can help them. Indeed, Boeing is in prime position to greatly increase the scale and revenue of their space hardware business if they can scale production to saturate Starship’s launch capacity. Boeing can make much more money building Lunar cargo for Starship transportation, because they’ll be shipping thousands of tonnes a year while building an expansive future and opening a new economic frontier. Would they prefer that SpaceX be compelled to verticalize in the Lunar base hardware space and own yet another colossal tranche of future value creation? At this point, the real fear of other industry players should be that SpaceX won’t even ask them to try. Instead, they’ll wake up one morning and find that all their ambitious junior engineers have taken a pay cut and moved to Texas, while no-one can work out why Starliner’s valves refuse to work properly.

This is why I think Starship is not understood. Understanding the risks and benefits of Starship would drive very different adaptive behavior than what we can see, ergo Starship is not understood, ergo I write yet another blog about it.

In October 2019 I explained why Starship and Starlink were such a big deal. In October 2023, looking back, what may have taken place?

It is hard to predict when the Starship design will stabilize, but I predict that SpaceX’s efforts in this area will only accelerate. As incredible as the progress at Boca Chica seems today, in two years time today’s rocket factory will look like the lonely tents of 2019. We’ll have Starships lined up along the beach, multiple launch towers reaching into the sky, and a series of high bays doing serial production. As SpaceX methodically retires programmatic risk in terms of Starship performance and reusability, engineering focus will shift towards the next constraints on the critical path, but not before. These constraints include deep space life support, robotics, and human-focused Lunar and Mars surface habitation. If NASA and other industry players don’t rapidly shift into high gear to provide the nine key needed space technologies, expect to see SpaceX spool up internal R&D in these areas. The earliest signs of this occurring will be obscure-looking job postings and quiet recruitment efforts, so if you notice your friends and colleagues inexplicably moving to South Texas or Austin, that’s why.

Meanwhile, it is reasonable to expect that the SLS will eventually attempt a launch, perhaps even with people on board. As Starship design converges, other launch companies (in particular Relativity, Blue, and Rocketlab) will adapt the design for their own reusable launchers, eventually driving down launch prices for third parties. Artemis will continue to limp awkwardly on with occasional half-hearted press releases, Eric Berger scoops, and middling budgets. At some point Starship will demonstrate an automated Lunar landing and return with a few tonnes of Moon rocks and either NASA will have branding rights, or they won’t. Starship will launch robots to Mars for landing site surveys and selection. While it is likely that NASA will be involved in this mission, I doubt they will pay for it or provide much/any hardware, unless there is a ride-along payload that would ordinarily have launched on an Atlas, or a few cubesats. Some (dozens) of these robots will be VTOL aircraft to perform extended surveys, building on the legacy of the Ingenuity Mars Helicopter but otherwise designed and operated very differently.

Perhaps JPL will continue to produce a flagship mission every decade or so. Perhaps the ice giants of Uranus and Neptune will get some attention, along with continuing efforts towards Mars Sample Return and participation in the Titan octocopter. These will expand our knowledge of planetary science in important ways, but as it stands neither JPL nor other NASA centers are well positioned to be the natural producers of any large subset of necessary Lunar/Mars base infrastructure, so I don’t expect to see them there, except perhaps as ride-along tenants.

In the meantime, other companies will spring up to exploit Starship’s improved access to space, procuring rides to the Moon, Mars, or asteroids for prospecting, entrepreneurship, services provision, national prestige missions, giant space stations, orbital factories, LEO constellations, and anything else one can dream up.

In my opinion, this is a huge tragedy. NASA is in the midst of the biggest opportunity since its founding in 1958. Starship can catalyze the organizational shifts necessary to once again align NASA’s workforce towards a technically coherent vision. We could have every NASA center churning out world-building machines by the truckload, building critical infrastructure that forms the backbone of humanity’s leap to a multiplanetary civilization. For example, JSC is the natural place to leverage decades of human spaceflight experience and develop futuristic life support machinery. Ames and JPL should be building fully automated construction management machinery. Glenn should partner with midwestern machinery manufacturers to build and operate Lunar and Mars environmental test systems and qualify a catalog of space-compatible commodity parts and retrofits. Marshall and KSC should build out containerized space power plants and enable launch cadence increases from ~1/week to ~1/hour. Goddard and Langley should oversee development of ambitious scientific research programs to be conducted from permanently occupied Lunar and Mars bases. Armstrong should coordinate supporting development work by the specialist contractors doing Lunar surface operations.

It should be impossible to not see a NASA logo anywhere on the coming generation of space stations and planetary bases, but this outcome is far from guaranteed. It certainly will not occur if the Artemis program continues to steadfastly ignore architectural economies offered by Starship. It certainly will not occur if NASA squanders these valuable years of transition waiting forlornly, as it has for decades, for Congress to accidentally turn the money supply up to eleven.

It may take a year or three, but Starship will happen and it will change everything. While the major industry players continue to not take Starship seriously, it is safe to say that Starship is not understood.