Musk's comments on in-orbit refuelling of Starship

Starship continues to develop, and its design just keeps on being tweaked.  For flights to the Moon or to Mars, Starship will have to be refuelled in orbit.  When Starship (then the BFR) was first mooted by SpaceX decades ago, refuelling was going to be "belly-to-belly" as it were, with the tanker lying adjacent to the Starship to refuel.  Actually, that "decades ago" is just a dig at SLS, NASA's incredibly expensive and horribly delayed rocket which is supposed to get us to the Moon.  In fact, SpaceX started development of Starship, then called BFR, just 5 years ago, in 2016.  After the initial plans, SpaceX then switched to what is inelegantly described as "butt-to-butt" fuel transfers.  But experience has shown that this is too dangerous, with extraneous fuel lines next to rocket engines just too unsafe.  So we're back to "belly-to-belly" refuelling.

This all came out when Blue Origin, Jeff Bezos's pet project, chucked a wobbly about SpaceX being awarded  part of the Artemis Moon landing project.   This report is from Teslarati.

After a much-anticipated GAO denial of Blue Origin and Dynetics protests over NASA’s decision to solely award SpaceX a contract to turn Starship into a crewed Moon lander, an in-depth (but heavily redacted) document explaining that decision was released on August 10th.

Aside from ruthlessly tearing both companies’ protests limb from limb, the US Government Accountability Office’s decision also offered a surprising amount of insight into SpaceX’s HLS Starship proposal. One of those details in particular seemed to strike an irrational nerve in the online spaceflight community. Specifically, in its decision, GAO happened to reveal that SpaceX had proposed a mission profile that would require as many as 16 launches to fully fuel a Starship Lander and stage the spacecraft in an unusual lunar orbit.

After around 24 hours of chaos, confusion, and misplaced panic, SpaceX CEO Elon Musk finally weighed in on the GAO document’s moderately surprising indication that each Starship Moon landing would require sixteen SpaceX launches.

Confirming many expectations, SpaceX’s solution to sending an entire single-stage Starship to the Moon, landing it on the lunar surface, and returning it to a lunar orbit (and maybe even Earth) goes as follows.

First, SpaceX will launch a custom variant of Starship that was redacted in the GAO decision document but confirmed by NASA to be a propellant storage (or depot) ship last year. Second, after the depot Starship is in a stable orbit, SpaceX’s NASA HLS proposal reportedly states that the company would begin a series of 14 tanker launches spread over almost six months – each of which would dock with the depot and gradually fill its tanks.

Third, once the depot ship is topped off, the actual Starship Moon lander would launch, dock with the depot, and be fully fueled. Finally, the fueled lander would fire up its Raptor engines and head to the Moon, where it would enter a near-rectilinear halo orbit (NRHO) – a weird high-altitude, elliptical orbit only necessary because NASA’s Orion spacecraft and SLS rocket are too underpowered to reach a more normal, functional orbit around the Moon.

After reaching NRHO, Starship would dock with Orion (or vice versa), receive its Artemis astronauts, land on the Moon for several days, and launch back to NRHO to return those astronauts to Orion. After its main mission is complete, it remains to be seen if Starship will have enough propellant left over to return to some kind of Earth orbit, where it could potentially be refueled and reused on future missions to the lunar surface.

In response to GAO revealing that SpaceX proposed as many as 16 launches – including 14 refuelings – spaced ~12 days apart for every Starship Moon lander mission, Musk says that a need for “16 flights is extremely unlikely.” Instead, assuming each Starship tanker is able to deliver a full 150 tons of payload (propellant) into orbit after a few years of design maturation, Musk believes that it’s unlikely to take more than eight tanker launches to refuel the depot ship – or a total of ten launches including the depot and lander. 

[Musk added, in a tweet (how else): 

"Without flaps & heat shield, Starship is much lighter. Lunar landing legs don’t add much (1/6 gravity). May only need 1/2 full, ie 4 tanker flights.  However, even if it were 16 flights with docking, this is not a problem. SpaceX did more than 16 orbital flights in first half of 2021 & has docked with Station (much harder than docking with our own ship) over 20 times."]

But, as Musk notes, so long as Starship gets anywhere close to its design objectives, it would be a non-issue even if each Starship Moon lander mission somehow required 16 launches. A step further, assuming that SpaceX proposed 16 launches per mission out of an abundance of conservatism, it’s fair to assume that a 12-day gap between tanker launches is also an extremely conservative worst-case scenario. Per Musk and SpaceX, Starship’s design goals call for multiple reuses of ships and boosters per day. Even if SpaceX falls a full magnitude short of those ambitious goals, Starship tankers should feasibly be able to launch every few days or maybe every week.

But thanks to SpaceX’s relatively conservative proposal, the company now knows that NASA is more than happy with Starship even if it falls something like 50% short of its payload performance goals and two magnitudes short of its reusability goals.

 

Starship refuelling in orbit.
Render by Erc X

As an aide-mémoire, here is my Mars timetable.

SpaceX to save NASA billions

 From Teslerati

In a move that’s likely to save the US taxpayer several billion dollars over the next few years, NASA has carefully extricated a mission to [Europa,]one of Jupiter’s ocean moons, from the claws of its own Space Launch System (SLS) rocket.

Known as Europa Clipper, the six metric ton (~13,300 lb) spacecraft will instead launch on a SpaceX Falcon Heavy rocket for less than $180M. Had Falcon Heavy not been ready or NASA shied away from the challenge of switching launch vehicles, sending the ~$4.25 billion orbiter to Jupiter could have easily added more than $3 billion to the mission’s total cost. Instead, Europa Clipper will be able to launch one or two years earlier than SLS would have been ready and at a cost that’s practically a rounding error relative to the alternative.

Measuring approximately 3100 km (~1940 mi) in diameter, Europa is approximately 10% smaller and 30% less massive than Earth’s Moon. Both are similar balls of rock with solid metallic cores. However, based on observations taken over decades by spacecraft and Earth-based telescopes, odds are good that Europa also has a vast liquid water ocean insulated by 10-30 km (6-20 mi) of ice so cold that it’s as hard as granite.

Scientists estimate that Europa’s saltwater ocean is dozens to 100+ km (~62 mi) deep, covers the moon’s entire surface, and holds more water than all of Earth’s oceans combined. Signs of a liquid ocean under Europa’s crust (and the crust of numerous other outer solar system moons, as it would turn out) were especially surprising because of the implication that those moons possessed vast heat sources. In the case of Europa, it’s believed that Jupiter’s immense gravitational pull and the moon’s close orbit are balanced in such a way that Europa is heated as those tidal forces violently stretch and squeeze its interior.

In an orbit 30% lower than Europa, tidal heating is so aggressive that the moon Io is littered with titanic volcanoes and lava lakes more than 200 km (~120 mi) across – so large that waves have been spotted on its surface with Earth-based telescopes. In short, because Europa appears to be in the right place to have enough – but not too much – tidal heating, it’s believed to be one of the best potential harbors of extraterrestrial life and Europa Clipper’s primary purpose is to pursue that potential astrobiological treasure trove.

Europa Clipper’s history is a truly bizarre one. Championed almost singlehandedly by fundamentalist Christian and former Republican Representative John Culberson, it’s almost certain that the mission would have never come together and never secured enough funding to proceed. Culberson’s singular goal: determine if humanity is (or is not) alone in the universe. If life can independently evolve twice in the same average solar system, the logic goes, it would practically guarantee that life will be omnipresent anywhere we look.

Culberson’s original vision was an orbiter (Clipper) that would effectively scout Europa for a lander that would follow just a few years later. Incredibly, he appears to have all but guaranteed that Europa Clipper will launch. However, he lost a reelection bid in 2018, casting the lander component into limbo before proper funding or commitments could be ascertained. It now seems likely that the future of Europa Lander will depend almost entirely on what Clipper does (or doesn’t) find.

Europa Clipper is now scheduled to launch on an expendable Falcon Heavy rocket no earlier than a two-week window set to open in October 2024. As part of the politicking to secure the billions of dollars needed to fund the mission, Culberson originally shackled Europa Clipper to NASA’s SLS rocket – now half a decade behind schedule and set to cost more than $23 billion before its first launch. However, it appears that SLS is so mismanaged and uncharacterized that even its infamously zealous, pork-motivated Congressional cheerleaders weren’t willing to put up a public fight to retain the SLS rocket’s only confirmed non-human payload.

Ultimately, on launch alone, Falcon Heavy’s Europa Clipper launch will likely save taxpayers more than $2 billion – the likely minimum cost of a single SLS Cargo launch. Due to issues with the rocket, Ars Technica also reports that Europa Clipper and SLS would have required at least $1 billion in modifications and upgrades to safely fly, meaning that choosing SpaceX will likely end up saving NASA more than $3 billion – equivalent to almost three-quarters of the entire Europa Clipper mission’s price tag.

Of course, by 2024 Starship will prolly be operating, but Falcon Heavy is proven technology, whereas Starship has yet to make it to orbit.

SpaceX assembling orbital Starship

 From Teslarati

SpaceX has begun rapidly assembling the first orbital Starship prototype and the Super Heavy booster set to launch it isn’t far behind.

SpaceX’s Boca Chica, Texas rocket factory seemingly turned a corner in early July as sections of Starship 20 (S20) began to pop up around the site. Though parts labeled Starship “SN20” first appeared as far back as March 2021, the only unequivocal work on SpaceX’s first purportedly orbital-class Starship began in mid-June with the integration of the first engine section with mounts for six – not three – Raptors.

However, in line with SpaceX’s strict focus on maximizing the speed of Starship development and shortening the path to orbit, the company has frequently built Starship hardware before firmly assigning that hardware to any given ship, booster, or tank. In other words, until SpaceX actually begins stacking multiple completed rocket sections, there’s always a degree of uncertainty about the fate of any given ring, dome, or tank barrel. With Starship S20, that process began earlier this month and Super Heavy Booster 4 is likely to follow suit within the next few days – if it hasn’t already.

Since SpaceX unceremoniously rolled Starship prototype SN16 to an empty lot in mid-May, the company didn’t stack a single Starship part until the first week of July – unusual after a frenetic seven months spent building, qualifying, and launching Starships SN8, SN9, SN10, SN11, and SN15 and testing test tanks SN7.2 a nd BN2.1. Around the same time as Starship SN15 became the first prototype to successfully complete a high-altitude test flight and land in one piece, news broke that SpaceX was striving to perform Starship’s first orbital test flight with Ship 20 (S20) and Booster 3 (B3) as early as July.

Eventually, Booster 3’s orbital launch assignment shifted to Booster 4 as it became clear that the former prototype wasn’t meant to fly, but Starship S20 remained. More likely than not, the almost two-month gap between Starship SN16’s instant retirement and the start of the next flightworthy prototype’s assembly can be explained by the significant changes, upgrades, and undecided design decisions required to jump to S20.

Two weeks after the first stack, Starship S20 is already approximately half-assembled and the last section of the vehicle’s tanks is almost ready for installation. What could be Starship S20’s nosecone is also in the late stages of assembly, though SpaceX has yet to even attempt to fully cover a nose in heat shield tiles and getting that process right could take an attempt or two.

Musk has emphasised that there will probably be several mishaps along the way before Starship orbital flights work.  In the first attempt to get Starship orbital, the booster will try to make a soft landing in the sea off Boca Chica,  and Starship will splashdown off Hawaii.  We're so used to SpaceX reusing its rockets that the apparent waste is actually quite shocking, but I expect SpaceX thinks that things could go wrong, possibly badly wrong, and doesn't want to completely destroy its Boca Chica base.   The Super Heavy booster and Starship itself are far too large to land on the drone ships that SpaceX currently uses for its Falcon 9 boosters, so that option is out.  It says a lot about how SpaceX has changed space paradigms that we should be surprised that the spaceships  from the first orbital Starship launch should be dropped into the sea, but of course that's exactly how all other rocket manufacturers still operate.

If the development of orbital Starship seems unbearably slow, that's because we are watching it happen in front of us, day by day.  Musk only announced the stainless steel Starship in January 2019, two and a half years ago.  By contrast, NASA's SLS was announced in 2011, uses old technology, and hasn't flown yet.  Oh, and when it does, the booster will be dumped in the ocean.  Plus each Starship launch will cost roughly $2 million while each SLS launch will cost roughly $1.5 BILLION.

Starship SN1 moves to test site

(Images from SPadre)

SN1 is moved to the test site.
Note the dimple on the hull.  

SN1 is lowered onto the test stand

SN1 lowered onto the test stand

SN1 on the test stand

Note the dimple in the hull in the first picture.  Aircraft skin is 2-4 mm (1 inch = 25 mm) thick, but aircraft have internal bracing.  Is 2-4 mm steel strong enough for Starship?  Will Starship need internal bracing (struts)? 

I'm certain it will.  The stresses caused by the "skydiver" re-entry manoeuvre will be more severe then those faced by aircraft. 

But SpaceX is moving so fast, there'll be time to adjust the design.  It took SpaceX 9 months to build the first Starship prototype.  It has taken it just 4 weeks to build the next one, SN1 (serial number 1).  SN2 is already under construction.  The rapidity of progress on Starship compared with the huge delays and cost overruns of SLS ......  Astounding.  First boots on Mars in 2025!

Living on Mars -- III

Mars with and without a dust storm

I talked here about the problems of living on Mars (temperature, air pressure, UV radiation, cosmic rays, toxic "soil") and about a solution to some of those problems (silicon aerogel, to raise temperatures and reduce UV radiation).  Now we come to the next big issue: energy.

With an glass/silicon aerogel/perspex dome cover, domes on Mars (at least between latitudes 40 N and S)  would be passively heated.  But it is very likely that heating will be required in winter, especially in the southern winter, when Mars is at its furthest from the sun (Mars has a more eccentric orbit than Earth).

That won't be the only need for energy by the first settlers, though.  A big need will be to manufacture fuel for return trips to Earth.   This will involve splitting water mined on Mars into hydrogen and oxygen, then harvesting CO₂ from the atmosphere.  A mixture of the CO₂ and H₂ is then passed at pressure and high temperature over a catalyst and this process (called the Sabatier process or reaction) produces methane.  More competent mathematicians than I have calculated that this will need 17MWh of electricity per tonne of fuel.  [But see below for an update—Robert Zubrin, the scientist who originally suggested propellant manufacture on Mars, has calculated it at 12 MWh/tonne.  About 70% my original information] Let's say each Starship requires 1100 tonnes or so of fuel (the Mars Colonial Transporter, the bigger first version of Starship, needed that), and there are 600 days between landing and relaunch.  That will require 31 MWh [22 on Zubrin's figures] of electricity per day, just to refuel a single Starship.

Average electricity demand in the US is around 12,000 kWh/person/year.  Assuming usage on Mars will be the same, for a colony of 100, that would mean 3.3 MWh of electricity per day.  Only, usage is likely to be higher on Mars than Earth.  If we use the higher consumption data for cold places on Earth (50,000 kWh/person/year for Iceland, 35,000 for Lichtenstein, 24,000 for Norway, 15,000 for Canada and Finland) then we're talking perhaps 10 MWh/day for the whole colony.   We will need electricity to heat domes, to control the air inside the domes (removing CO2 for example), to run rovers, to grow food, to light domes, etc.  So we'll need total output of 44 MWh [32 on Zubrin's calcs] per day—three-quarters of that for fuel production.

So where is this electricity going to come from?

Let's start with nuclear.   It's out of the question to build a large-scale nuclear reactor on Mars.  But NASA has been working on a smaller, simpler, safer reactor, designed specifically for use on spacecraft and on Mars and the Moon.  It's called KRUSTY (Kilopower Reactor Using Stirling Technology), and here's a video which gives a brief explanation of it.  A reactor 10 times larger is planned.  This will produce 10kW of electricity,  will weigh 1500 kg and will contain 44 kg of  U-235.  So each day, one of these reactors would produce 245 kWh of output.  We'd need 180 [130 on Zubrin's data] of the 10 kW kilopower reactors to produce enough electricity for the colony as well as refuelling one Starship.  They'd weigh 270 tonnes [195 tonnes Zubrin].  Just delivering them to Mars would require 3 Starships [2, Zubrin], assuming on current plans 100 tonnes of cargo per ship.

OK, what about wind?  You'd think that with the air pressure on Mars, just 0.6% of Earth's, wind turbines would be useless.  This informative video from Scott Manley shows how wind turbines on Mars could actually work quite well, despite the low atmospheric pressure.  For a start, don't confuse air pressure with air density.  Now on Earth, these two are related.  However, the air on Mars is denser than on Earth at the same pressure because it's 95% CO₂ and because it's much much colder.  This boosts the impact of air density on the output of a wind turbine by about 100% relative to Earth.

Also, average wind speeds on Mars at the Viking 2 lander site were 15 mph (just under 7 metres/second).  In the US, average wind speeds are between 6 and 12 mph, but of course, wind turbines tend to be sited where winds are stronger.  So, back-of the-envelope, 50% of Earth's wind capacity.   Small wind turbines will weigh something like 300kg, but more productive wind turbines are proportionately less heavy, because the power produced is proportional to the square of the blade radius. Let's assume one with a 10 m rotor diameter, twice the size of the rotors discussed in the link.  This will increase the electricity output four fold, but will weigh, say, 600 kgs.   Such a wind  turbine would produce half (on average) of a 10 kW Kilopower reactor at 1/3rd the weight, so we'd need two Starships to provide all the wind turbines you'd need for your  colony on Mars plus fuel production for the return home.  But—and this is key—it will be easy to manufacture small wind turbines on Mars, unlike (at least for the first decade) nuclear and solar generators.

Just as on Earth, the wind won't blow all the time, so you'll need complementary power source—solar.  Thin-film solar is less efficient than conventional solar cells, but they're 100 times lighter, and can be rolled up for transport.  Because Mars is further from the sun than Earth, solar panels there will be 40% less productive than on Earth.  At the equator on Earth (Singapore) 10 kW of solar panels will produce 12,600 kWh per year, or 34.5 kWh/day.  Reduce that by 60% at the Martian equator, and output of 10 kW of conventional solar panels would be 14 kWh/day per 10 kW of panels.  You'd need 32000 kW [23000, Zubrin] of panels.  One kW of solar panels would cover 2.75 metres.  So you'd need 12,000 square metres of panels on Mars to power the colony.  And if you use thin-film solar, some 25% more.  15,000 square metres.  Imagine a metre-wide strip of thin-film panel.  You'd need 15,000 metres in rolls.  15 kms!  It might be much the lightest generation source, but it will surely take up a lot of space inside a  Starship.  Solar output would be almost completely reduced to zero during Mars's periodic dust storms.  The good news is that wind speeds treble during the dust storms, so just as on Earth, wind is highly complementary to solar.

A couple of conclusions:

• It would make sense for all three generation sources to be used.  The nuclear would provide "baseload", i.e., for all the demand for electricity excluding fuel manufacture.  The first priority is maintaining life.   So the first colony would need 60 10 kW Kilopower reactors, enough to heat, grow food, light, air and water purification, rovers, etc.
• 120 10-metre diameter wind turbines, which would on average provide about the same power.  Any surplus energy would be used to make methane and oxygen.
• 15,000 kw of thin-film solar panels.  Again, the electricity they generate will go towards making methane.
• The cargo demands for all these generators, space and weight suggest to me that more than the planned 4 cargo ships will be needed to start colonisation.  Just for electricity generators, five Starships will be needed, one for nuclear, two each for wind and solar.  [Possibly just 3 using Zubrin's estimate]  It won't be a problem once the Mars-Earth trade route is established, because the cost of sending cargoes to Mars will fall precipitously.  As I guess here, the cost of delivering 1 tonne from Earth to Mars will prolly fall to $20K  by the third or fourth expedition, since re-usability is key.  It's only a serious problem for the first expedition. At each subsequent expedition, more wind turbines/solar panels/kilopower reactors will be brought.
• Reducing the number of people doesn't make much difference, since three-quarters of the electricity is needed for propellant manufacture.   The only way to cut the energy needs is to remove the option to return after 2 years, and stretch it out to 4 or 6 years.  Hmmm.  Or, more plausibly, we send ten Starships on the first crewed expedition, two crewed and eight cargo.  But only one will return to Earth (based on my calculations above), so re-usability is in effect reduced, raising costs.  It'll be different after the second expedition, because then there'll be enough electricity generation capacity to make fuel to send two Starships back, and the number will increase with each expedition to Mars.  
• On these numbers, it will take 10 expeditions of 10 Starships at a time for enough fuel to be available to send them all home.   That's 20 years.  
• Even if some of the Starships are in effect not re-usable (because there isn't enough propellant to fly them back to Earth), the cost will still be far below NASA's estimate of $150 billion for a crew of 5.  At $100 million per Starship**, 10 Starships to get the colony started would cost $1 billion, even if they were never used again—and they'd provide shelter to the first colonists while ground-based shelter was built.  Thus the cost will be $1 billion initially, then $500 million per year (Mars is in opposition to Earth only every 2 years)
• If Starship works, NASA will surely ditch SLS and use the $1.5-$2 billion per launch, never mind the $10 billion plus development cost, to send 200 people every 2 years for a permanent Mars base.

As usual, anyone who knows more about this than me, or who spots flaws in my calculations or analyses, is invited to comment below.

See also:

Living on Mars -- I

Living on Mars -- II

Update:

Robert Zubrin (the guy who first suggested we manufacture methane on Mars to reduce the crippling fuel burden involved in bringing it from the Earth) has estimated the energy cost of producing methane in this tweet:

In other words, my calculations are too pessimistic.  Reduce them by 30% to get a more accurate measure.  Just so y'all know.

—————————

**  [Update 27/04/2020] Musk has stated that he's aiming for a total capital cost per Starship of under $5 million, and a cost per launch below $2 million (including the cost of Super Heavy), with a payload of 150 tonnes.  Each launch will use $800 K of fuel.   To get Starship from LEO  to Mars will mean it has to be refuelled in orbit, and that will require 6 launches per flight to Mars, costing say $16 million per Starship to Mars, or $21 million if we add in the capital cost, since the first ships won't be returning.  That means the initial expedition of 10 ships will cost $210 million.  64 cents per inhabitant of the USA.  And a berth on a flight could cost as little as  $210K per ticket.  One tonne to Mars would cost $140 K.  Subsequent flights will be cheaper, because Starship will rapidly get more efficient as SpaceX learns while doing, just as Falcon 9 got better, and because some Starships will return.  Costs per passenger or per tonne are likely to halve over the first 10 years.   SLS, meanwhile, will cost $1.5-$2.5 BILLION per launch.