ASTRA — USA, 2021

While most New Space ventures try to drum up as much hype as possible, this one tried to keep its doings as far out of the public eye as it could. For the first three years its hiring page on LinkedIn only said “Stealth Space Company”. It was reported that the actual name of the company was Ventions LLC, but this turns out to have been a past venture by one of the founders, Adam London. (The other founder is Chris Kemp, who is mainly a software entrepreneur but has held a post at NASA. He’s the CEO.) It was also widely reported that the company’s real name was ASTRA Space, this being short for Atmospheric and Space Technology Research Associates, but this was false — that is a separate unrelated company. And note also that neither of them is to be confused with Ad Astra Rocket Company, which is working on ion propulsion and the microwave-powered VASIMR engine. I’m starting to think that it would be a prudent idea to change the company’s name to something less overloaded than “Astra” is. It doesn’t help that the rocket itself has no official name, so the company name is the only thing we can call it. They just designate their current design with a version number, such as “Rocket 3.3”.

While keeping publicity to a minimum, they attempted two suborbital launches from Alaska. These apparently consisted of test flights of their first stage with a dummy second stage, and neither of them managed to get clear of the spaceport grounds before falling back. Once they felt ready to attempt orbit, they went a bit more public. Unfortunately that attempt failed — they scrubbed a minute before launch, and then while pumping the fuel out, problems cascaded and Rocket 3 was lost in a fire. This sent them back to the shop for months, after which their Rocket 3.1 managed to get part way through the first stage burn before going off course and being forced to self-destruct, again failing to clear the spaceport grounds.

They build their rockets at the former Navy base in Alameda, California, which is an island next to Oakland. They picked the spot because one of the old buildings is set up for testing engines indoors. This lets them iterate rapidly without having to schlep parts between a high-tech hub and an empty desert for testing, as most rocket companies need to do. With this asset, they hoped to set a speed record for how quickly a small rocket startup could actually reach orbit, but that hope was dashed.

What they were working on was an orbital rocket smaller than the Electron, with a 150 or 200 kilogram payload capacity. The idea was to make it cheap, simple, and mass-producible, with no carbon composite or 3D printing, so once they have it working, they can quickly ramp up to producing and launching a rocket a day, if sufficient demand ever exists. They regarded the Electron as over-engineered, and were hoping that their aluminum rocket would only cost a third as much, even in small production volumes. Once they started making them in bulk, there was talk of bringing the cost down to six digits.

When their second test flight experienced a Rapid Unplanned Disassembly shortly after liftoff, we heard that the booster uses five engines... all of which failed together, apparently, due to a fire that destroyed the wiring. Even after they’ve emerged from stealth, some of our best further technical detail came from information released by NASA about R&D payments made to the company. These describe particular development tasks which include a kerosene-burning engine with electric pumps, similar to the Electron’s Rutherford engine, but apparently of lower performance. They also mention work on thrusters for satellites, probes, and interplanetary landers. Yep, they’re apparently involved peripherally in developing sample-return probes for the moon or even for Mars.

Anyway, the rocket is a stubby two-stage design in which the booster has five engines with rather small bells, and the second stage appears to be very small — the interstage and fairing apparently swallow it from both ends, leaving it as basically bare tanks for minimum weight. It weighs maybe one ton fully fueled. The fairing is small enough that each half can be easily lifted into place by one guy. The enclosed payload volume looks like maybe one cubic meter. I don’t know if the tapered interstage comes off in halves, like the fairing, or just stays attached to the booster. Eventually we learned that the booster engine is named “Delphin” and the upper stage one is called “Aether”. The Aether is pressure-fed.

They are building this little rocket in respose to the Pentagon’s desire for a rapidly responsive launch capability, because the military services are tired of needing to schedule launches months in advance. Astra tried to get a $12 million prize from DARPA if they could make a launch with only a few weeks notice, and then a second launch within a few weeks after that. The prize also required that they do not require a fixed launchpad. They say they’ve got portable launching gear that they can set up on any flat piece of concrete, and the rocket itself can be transported in a shipping container. The DARPA prize expired, but they already had about fifteen flights booked, with rockets for them already semi-assembled. They meant to transition quickly into outright mass production of these babies. They also want to start building satellite buses, and have bought up a small outfit that makes ion engines. And they’ve gone public raising their total captial intake to around two billion dollars — an amount they’re unlikely to earn back without doing a big volume of business.

getting it operational

But first the rocket has to work. With the fourth test flight (Rocket 3.2), they reached space and nearly reached orbit, which meant their goal of reaching orbit in three tries looked likely. All they needed to do was fix a mixture issue that made the upper stage run out of fuel with lox left over. But flight attempt five with Rocket 3.3 — the first with a real payload, and the first version to get more than one test before being upgraded — was a step backwards, with the fuel hose spilling propellant and starting a fire, which made the rocket lose an engine, so it ended up wallowing sideways before starting to gain altitude. At least they got it over the ocean, with the majority of the fuel gone, before aborting. In November 2021 they finally had success with flight six, but it had no satellite, only an instrument package which stayed attached to the upper stage in order to monitor it. For their next try they launched from Florida with a NASA payload, and lost it when the fairing wouldn’t open and the upper stage ignited anyway, apparently because they mislabeled two lines in the wiring diagram. Only on their eighth launch, in March 2022, did Rocket 3.3 finally put up actual satellites... after which the next launch had an upper stage failure 500 kilometers up.

I hoped they would make it; they’re local to my area, and it would be great if launches could be done as inexpensively as they’re aiming for. But maybe orbital launching just doesn’t work if you’re this cheap. Investors and analysts have disparaged them for thinking they could use non-aerospace-grade parts, and for unrealistic market expectations (though the latter is something you will find in every small rocket company). They further claim that unless they find a customer who wants hundreds of launches, they’ll lose money on every flight.

back to the drawing board

But wait, what are they doing now? They signed a deal with Firefly to license their Reaver engines and build them in their Alameda factory. This was signed shortly after both companies had test launches fail due to one booster engine quitting early. It may be no more reliable than the engine they already had, but it does allow them to make Rocket 4.0 bigger, so it competes head to head with Electron. The goal is to use two Reavers and a fatter second stage to lift 300 kilos for four million dollars, while still allowing the rocket to be transported by shipping container. They also want the new rocket to support a higher cadence, and to be automated enough so mission control can be just two people. (They are already down to eight.)

In mid-2022 Astra essentially acknowledged that Rocket 3 sucks, and announced that they would discontinue it instead of persisting in the struggle to get it to orbit reliably. That rocket probably holds the world record for the poorest success rate of anything that has shown itself able to actually put something in orbit. In their strategy of using cheaply available parts, they went too cheap. They’re now essentially going back to square one, and betting everything on the future success of Rocket 4, which they now expect to carry 600 kilograms.

A year later the layoffs started getting serious, and they started moving people off of Rocket 4 development to work on ion thrusters for satellites, which do have some customers and might have a chance of making money. It may be that Rocket 4 work is essentially halted, with only enough of a pretense of ongoing effort so they can avoid admitting publicly that it’s dead. By autumn they were defaulting on a loan. And I was thinking: place your bets: the day we move them to the Failed Plans section may be only a few weeks away. But then someone came through with new funding. I don’t think it was very much, though, so the timetable may not be extended by all that much.

By early 2024 the company was on the brink of shutting down. From being valued at $2 billion when they went public via SPAC three years earlier, it had dropped to $13 million. The founders, Kemp and London, bought it back at 50¢ a share, valuing it under $12 million, to give it one more try.

It was this point that I decided that Astra should have been moved into the Failed Plans section two years ago.

Why did it fail? Because they tried to be cheap. Aerospace is extremely unforgiving of tiny flaws and minor shortcuts, and orbital service does not let you try to make up thin margins with volume.

Rocket 3: mass <10t, diam 1.2m, thrust 140 kN, imp unknown, electric pump (kerosene), payload < 0.2t, cost hopefully <$5M/t to $10M/t, record 1/3/5 (final).