FALCON — USA, 2008

Finally we get to the good stuff — the brand new wunderkind rocket which is obsoleting all the rusty crap which the space industry has been putting up with for too long. Elon Musk — who, regardless of what you may think of his sometimes appalling behavior, is still easily the most important tech visionary of our time — once tried to buy some services from existing rocket companies, and found them so unhelpful and uncompetitive that he decided to take the money he was going to offer them and start a rival company with it instead. In short, he found a market ripe for disruption. The company found ways to undercut their prices, and in the process, almost by accident, produced some of the most rapid advances in rocketry that had been seen in decades. Many other space startups emerged around the same time, but so far only Space Exploration Technologies Corporation, better known as SpaceX, has made the new venture a sound commercial success on a large scale. They now boast a long track record of successful and profitable launches.

SpaceX is far from Musk’s only company, of course. In all of his ventures, what separates him from other tech moguls is that he approaches things first as problems of basic physics, then of engineering, and finally of manufacturing processes, rather than in terms of marketing or finance. This is, I believe, why other billionaires who try similar ideas fail while he succeeds.

The company started with a small two-stage launcher called the Falcon 1, which had a gas-generator engine called “Merlin” on the bottom and a small pressure-fed motor called “Kestrel” on top, both burning kerosene. (Merlins and kestrels are both birds of genus falco.) The Merlin’s turbopump was designed by an outside company named Barber-Nichols. The first three launches in 2006-8 all failed, but they got a dummy payload to stay up on the fourth try. With the fifth they launched a real satellite... and then Elon cancelled the Falcon 1. Once the company was saved from going broke, he put all the resources into making the larger Falcon 9.

Falcon 9

Falcon 9 rockets, unlike any from previous eras of spaceflight, are designed to be highly reusable. (Even the Falcon 1 included a parachute, though it came to nothing.) The first stages soft-land themselves and are ready to go again after some refurbishment. They’ve pretty much perfected the “suicide burn” landing, in which the rocket stops exactly at ground level as smoothly as if you were watching a takeoff in reverse.

But the reuse process is still kind of slow, with time between landing and takeoff only gradually decreasing from months to weeks, with a turnaround in days still far away. In the period when they were developing this reusability, I guessed that if they succeed in getting a whole fleet of Falcons to do ten launches each, we could see prices falling through the floor, like under two million dollars per ton to orbit... if any competitor ever gave them a reason to lower their profit margin. (We’ve seen Tesla raise prices and then cut them depending on demand and competition, so I would expect SpaceX to do likewise.) And with no such competition coming forth yet, and this level of reuse now solidly achieved, it appears that SpaceX feels no need to lower any of its prices yet. In 2020 they reused some boosters over five times, and one of them went up to eleven uses in 2021, and two more did in 2022 (forming a group which Scott Manley called “the Nigel Tufnel club”), but the turnaround time is still weeks rather than days. Four weeks was the record for a while, then three, which by the ambitions the Falcon was designed for is a big success, but it’s still nothing like the quickness that might be possible with methane-burning engines.

The pace keeps quickening nonetheless — in 2024 booster reuse counts reached the twenties, and the rate of launches was around two a week. By that point Falcons were hucking about twice as much mass to orbit as all other rockets combined. So of course SpaceX plans to keep pushing that even further, to see if they can get thirty or forty flights per booster and three or four launches per week. They especially want to raise the pace from the west coast, meaning that Vandenberg might need to add more launch pads.

But getting to thirty flights might not work — one booster’s luck finally ran out and it broke a leg while landing, after perhaps touching down a bit too hard. It was the oldest booster still in service, completing a record twenty-third flight. This mishap resulted in a brief suspension of launches. It didn’t help that this occurred shortly after the first upper stage failure in years, which had also caused a suspension, and shortly before two nearly simultaneous crewed launches (Polaris Dawn and Crew 9)... all at the same time that the Boeing Starliner was going through its stranded astronaut humiliation. The summer of ’24 was not a cheery time for American spaceflight. Even the Cygnus cargo capsule had a scary moment when it looked like it might fail to match orbits with the ISS.

The Falcon 9 has nine Merlin engines, all fully gimballed, and now far more advanced than the primitive early versions. Today’s Merlins produce more than double the thrust of the 2006 original. The earliest Falcon 9 design had the nine Merlins arranged in a square, with the ones in the corners sticking out past the edges of the fuel tank, but they soon changed this to an octagon with one in the center. That square-bottomed version was not particularly tall, but once the engine thrust was increased, the revision stretched the tanks considerably, lengthening the first stage from around thirty meters to over forty, while retaining the original diameter of 3.7 meters (some early sources say 3.6 — apparently the actual size is twelve feet, which was chosen for being transportable by road). This severe elongation left the Falcon as one of the skinniest rockets in use, in proportion to its length. Then they crammed in even more fuel by chilling the kerosene and lox to near their freezing points.

Its second stage uses a single Merlin, with an oversized bell for efficiency in vacuum — not much smaller than the fuel tanks it’s attached to. The extended part of the bell is made of exotic niobium alloy, and glows orange in use.

Even when reuse was still marginal, the Falcon 9 made SpaceX the number one launch provider in the world, except for the Chinese government at times — twice as busy as any of their commercial competitors. In 2017 they averaged one and a half launches a month, and said they were facing the challenge of speeding up toward a weekly pace, because customers were still backed up in a waiting list. The plan was to do thirty launches in 2018, but this slipped, as things often do with SpaceX due to their “agile” approach to engineering and their “aspirational” scheduling of innovative achievements. For a while, the limiting factor was the rate of manufacturing enough of the latest “Block 5” booster stages. As a result, SpaceX fell behind the pace set by China with the Long March 2, 3, and 4.

But the number of customers has been flattening out. In fact, in 2019 their overall launch rate decreased, and they actually had a three month dry spell with no launches at all. It appears that the queue of customers awaiting the launch of heavy satellites has emptied out. (ULA’s launch cadence has also decreased.) I think they were hoping that somewhat lower prices would stimulate demand, but so far, that isn’t happening. Maybe if they were a lot lower. As things are, most launch customers have needs which are fairly inelastic as to price, or they wouldn’t be in the market in the first place. It may be that no amount of price lowering significantly increases the demand for heavy satellite launches, as usually the craft themselves already cost plenty more than the launch does nowadays. Perhaps there will eventually be new opportunities for budget-priced large satellites, but if there are, nobody has figured them out yet. It may be years before the demand catches up to the supply.

Starlink

Well, if no one else has a use for cheap mass-produced launches yet, Musk will have to come up with one himself. And he has: Starlink. It’s a plan to provide fast internet wirelessly using thousands of satellites — tens of thousands, if they grow to the scale they claim they could. At sixty quarter-ton sats per launch, they’ll be doing dozens of flights for this project for the next decade... and once they have the full set aloft they’ll be continuing afterwards with replacement launches, as the early versions of the sat are designed to stay in service for only five years and then ditch into the atmosphere, so that the technology can be continually refreshed. They planned to do about 25 launches a year for this project, starting in 2020. That brought their cadence numbers up... they did 26 launches in 2020 despite covid, and in 2021 they did 31. By this time they had plenty of boosters available, making five or six new ones per year and only losing one or two. In 2022 they passed the cadence of a launch per week, completing sixty flights, and pushed the number of uses of three of their boosters to fourteen, and one booster to fifteen. Even if you don’t count Starlink, 2022 had 27 launches (plus one Falcon Heavy). The Long March 2 was the only other rocket to come close to that, with 24. In 2023 they did an incredible 91 launches of the Falcon 9 plus five of the Falcon Heavy, of which 33 were non-Starlink... and 33 is still comfortably more than all Long March launches of 2023, or all other launches from the USA, or all launches from Russia, or all combined launches from the second-tier and minor spacefaring powers.

But astronomers are not happy about this large increase in the amount of “space junk”, and they’ll be even less happy when competitors start putting up their own satellite swarms. (We ought to build and launch several Hubble-sized telescopes — that might relieve the feelings of the astronomical community significantly.) Several outfits have announced plans for such networks, including Amazon and the Department of Defense. But competing won’t be easy... an outfit called OneWeb launched about three loads of sats and then went bust. They then got bailed out and resumed launching, though it’s hard to imagine any success coming out of this effort. The Pentagon’s swarm, of course, doesn’t need to compete, nor does one that the Chinese government is going to start putting up. The first launches of the Pentagon swarm will be on Falcons.

With all this makework, the logistics of the Florida space coast are struggling to keep up with that pace. Many companies and agencies have to share common infrastructure there, and increasingly often, a delay experienced by anyone affects everyone. SpaceX has been reviewing the sticking points and pushing for changes to support busier schedules.

other plans

SpaceX is also now organizing ride-share launches which carry sats from many small customers instead of from a few large ones. If you’re willing to share your orbit, this will underprice being launched on something like an Electron. They do about three of these a year.

At one point they said that their goal was to put a heat shield onto the second stage and soft-land it too, but they backed off from that. That sounded dubious anyway unless they cut the payload capacity way back, but I can see why they wanted to, as they now say that with reusability of the booster, the loss of the second stage accounts for something like one third of the cost of each flight. They might cost around $20 million or so apiece, by one guess I’ve heard. Musk admitted that this would only be feasible for “some payloads”, but he still wanted to see if they could do it. They also worked on landing and reusing the fairings which cover the payload, bringing them down on parachutes. They tried guiding them to a ship with four arms holding up a big circus-net made of cloth straps, but despite occasional successes, this turned out to be unreliable and hazardous. Trying to land them in that net was very difficult, like trying to throw a layup with a potato chip — it’s just too aerodynamically unstable. So they added floats and just let them splash down into the water. Though those fairings are very simple things compared to the second stage, they still cost like $5 million, and they have to buy them from a vendor, who apparently has taken advantage of SpaceX’s inability to get them anywhere else to not only keep prices high, but obstruct progress on making a larger size. One could speculate that this might be due to pressure from their other customers who compete with SpaceX.

Anyway, all of these hopes and plans for improving the Falcon are now fading away, as Musk becomes increasingly committed to building an all-new rocket to obsolete it — the Starship.

Falcon Heavy

They’ve put together a triple booster “Falcon Heavy” version... about five years after they thought they would. With the Block 5 upgrades (which they now say will be the final version of the Falcon 9), even the basic Falcon now nearly matches the payload capacity of the Delta IV Heavy, so the Falcon Heavy is a huge leap in lift capacity, and for a few years it was the most powerful rocket in operation. (The Delta looks bigger, but that’s because it’s full of hydrogen. The rocket which supplanted the Falcon Heavy as most powerful was the SLS, which is the size of an Apollo.) The Falcon Heavy’s claimed capacity is over sixty tons, without even using a bigger second stage. But that’s if you’re willing to pay them to discard stages instead of recovering them, as the payload limit is quite a bit less if the boosters need to land afterwards. In expendable mode, they can not only burn all the fuel for ascent instead of saving some, they can leave off unnecessary parts such as the landing legs and grid-fins. A regular Falcon can be reused up to about two thirds of its max expendable payload, but the Heavy can only manage about half. But the real point of the Heavy is not to lift giant stuff, but to make it so that they don’t have to expend boosters anymore, or at least only have to expend the center one for the worst cases. By losing the center core they can reach around 90% of the capacity they’d have by expending all three. They occasionally use this kind of launch for heavy geosynchronous payloads — the Pentagon hardly notices the cost of an expended core booster.

The rocket may not actually be able to support a sixty ton payload. (Their standard payload adapter is only designed to support eleven tons — just half of the amount a non-heavied Falcon can supposedly lift to orbit. What the limits of the stack might be with a better adapter, I have no idea — all I know is that the heaviest loads actually launched are somewhere around twenty tons.) The main benefit of the Heavy is probably for reaching higher orbits. If the destination is geostationary transfer orbit (GTO), for instance, the Heavy would increase the nominal capacity from 8.3 tons to 26.7 tons, without reuse. They claim it could send 16.8 tons to Mars, or 3.5 to Pluto without using a gravitational slingshot. One could do even better by adding a third stage, but they don’t sell one — you’d have to bring your own, and fit it into fairing, which is smaller than that of competing rockets. (A bigger one is on their shopping list, or was before they decided that the answer to all such problems would be the Starship.) I will note that these GTO numbers are not that impressive, compared to the huge loads that it can lift to low orbit... the use of kerosene as fuel in the upper stage, instead of hydrogen, reduces its efficiency for reaching distant trajectories. With full booster reuse, the capacity to GTO is actually less than that of the Delta IV Heavy.

Some customers still fork over willingly for a nonrecoverable overweight launch. They hope the availability of the Heavy will make such launches largely unnecessary... but in May 2023 a customer actually paid them to expend all three Falcon Heavy boosters, in order to get a comsat into geosynchronous orbit without needing to bring its own fuel for lifting out of the transfer ellipse. Making the Heavy proved far more difficult than they anticipated, with practically the whole first stage needing a redesign to be able to support the additional thrust and vibration of adding side boosters. This highlights SpaceX’s general philosophy of shaving things as close to the limits as they can reasonably get away with, and leaving only a minimum of safety margin... when they decided to ask more of their booster, the margin they needed wasn’t there. The rebuild job was so challenging, and so hard to test on the ground, that they almost cancelled the project three times, and Musk said that if the rocket just managed to get a safe distance away before it exploded, and not destroy launch pad 39A in the process, he would “consider even that a win, to be honest.” But even the first test flight of the Falcon Heavy put reused boosters on the sides. They used a “silly” payload on the first test — they threw a Tesla roadster into what they hoped would be a Mars transfer orbit, and which ended up going out all the way to the asteroid belt, with the stereo playing David Bowie, and a dummy named Starman in the driver’s seat with its arm on the window sill, wearing one of their new space suits. (In 2010, they tested the first Dragon capsule by launching a wheel of Le Brouère cheese in it, so a degree of silliness is not unprecedented.)

With the Falcon Heavy they considered pulling off a trick known as “asparagus staging”, or as it’s more properly termed, propellant crossfeed. How this works is that with all three boosters burning at full power, the two side ones are also pumping fuel to the middle one’s engines. They run low and drop off in less than two minutes, and hey, the single remaining booster still has nearly full tanks. The traditional approach, as used in the Delta IV Heavy and its ilk, is to just run the center booster at minimum throttle for the later part of the initial burn, until the side ones are gone. But they’re not doing the asparagus trick. The reasons for caution are obvious, with all that might go wrong with such a scheme, and it simply isn’t needed — to be honest, more lifting power is the last thing the Falcon Heavy needs. They originally thought the Heavy would be used for about half of their business, but they overestimated the demand for such heavy lift, and every time they improved the regular Falcon 9, it covered more of the weight range which they previously thought would require a Heavy. Nowadays, it looks like this hulking rocket might struggle a bit to find enough work to justify its presence.

final revision?

Meanwhile, as the Heavy was making its debut, the familiar Falcon version that made it famous, which was apparently known in full as v1.2 “Full Thrust” Block 3, got upgraded to Block 5 (or “Fuller Thrust” as it has sometimes been unofficially termed), increasing its claimed payload capacity by 30% while also meeting NASA’s toughest safety standards for crewed flights. (Block 4, a transitional version with a minor increment to the engine thrust, was used for a few flights in late 2017 and early 2018.) This upgrade included lots of fixes to the various weak points that have been revealed by inspecting landed boosters, such as replacing the aluminum grid fins with titanium, which is a lot more expensive but should last forever. They hoped that the Block 5 will be able to refly up to ten times with no refurbishment at all, just an inspection and cleaning. The rocket business in general is prone to the dashing of such hopes, but in this case it worked out pretty well. They pinned all their dreams of rapid and inexpensive reusability onto the Block 5, and started throwing away used Block 3 boosters, rather than saving them — they pretty much considered the successful Block 3 landings to just have been part of the program for learning how to make the Block 5, and didn’t feel like collecting old models that need heavy refurbishment to fly again. After this, they did no further development on the Falcon booster.

One of NASA’s hurdles for crewed flight was that they should make seven consecutive flights with no changes to the design, so they wanted this version to be truly final. By the end of 2018 they completed ten Block 5 launches with six boosters, one of which was the first to make three flights. But it wasn’t until 2020 that the Dragon 2 crew capsule was ready enough for NASA to put their astronauts into it.

The block 5 version has been amazingly successful. Having gotten all the bugs out in previous revisions, it quickly started racking up dozens, and then hundreds, of successful flights, becoming the go-to launcher for almost everyone who just wanted their rocket to be reliable and affordable rather than having some political reason to use something else. Over the next six years it launched almost three hundred payloads without any errors, and over that time it cemented the Falcon as the nearest thing to a monopoly that a single rocket could possibly become. Then in the summer of 2024 (as briefly mentioned above), a routine Starlink launch saw the upper stage engine spring a lox leak, and when it was time to relight it, it went kerblooey. And suddenly everyone had their schedule thrown into chaos when Falcon launches had to be put on hold. Spaceflight needs redundancy between systems as well as within them if you want to have it reliably available, and while governments are mostly very cognizant of that and try to keep alternate options open, the free market would rather just dump everything into whatever is the best deal of the moment, with little thought for whether that might bite them in the ass later. As a result, the many outfits which have been working on new rockets to compete with the Falcon have not only been struggling to get their hardware working, but also struggling to appeal to customers. Hopefully that will start to change now, and bring back a bit of balance.

SpaceX quickly announced that they’d found the cause: a small tube that may have been not quite tightened down enough, which vibrated too much and developed a crack. And instead of improving it, they just decided it wasn’t needed, and for a short term fix they just took it out. They were able to resume their own Starlink launches quickly, and other launches resumed not long after. Even crewed ones were pronounced OK to proceed, which I would probably have been more hesitant about.