SpaceX launches Starship/Superheavy on ninth test flight, but experiences issues in orbit
Starship in orbit before losing its attitude control
SpaceX today was able to successfully launch Starship and Superheavy on its ninth test flight, lifting off from SpaceX’s Starbase spaceport at Boca Chica.
The Superheavy booster completed its second flight, with one of its Raptor engines actually flying for the third time. Rather than recapture it with the launchpad chopsticks, engineers instead decided to push its re-entry capabilities to their limit. The booster operated successfully until it was to make its landing burn over the Gulf of Mexico, but when the engines ignited all telemetry was lost. Apparently that hard re-entry path was finally too much for the booster.
Starship reached orbit and functioned successfully for the first twenty minutes or so. When engineers attempted a test deployment of some dummy Starlink satellites, the payload door would not open properly. The engineers then closed the door and canceled the deployment.
Subsequently leaks inside the spacecraft with its attitude thrusters caused the attitude system to shut down and Starship started to spin in orbit. At that point the engineers cancelled the Raptor engine relight burn. The spacecraft then descended over the Indian Ocean as planned, but in an uncontrolled manner. Mission control then vented its fuel to reduce its weight and explosive condition. It essentially broke up over the ocean, with data was gathered on the thermal system until all telemetry was lost.
Though overall this was a much more successful flight than the previous two, both of which failed just before or as Starship reached orbit, the test flight once again was unable to do any of its objectives in orbit. It did no deployment test, no orbital Raptor engine burn test, and no the re-entry tests of Starship’s thermal protection system. Obviously the engineers gathered a great deal of data during the flight, but far less than hoped for.
SpaceX has a lot of Superheavy and Starship prototypes sitting in the wings. I expect it will attempt its next flight test, the tenth, relatively quickly, by July at the very latest. I also do not expect the FAA to stand in the way. It will once again accept SpaceX’s investigative conclusions instantly and issue a launch license, when SpaceX stays it is ready to launch.
As Starship reached orbit as planned, I am counting this as a successfully launch. The leaders in the 2025 launch race:
64 SpaceX
30 China
6 Rocket Lab
6 Russia
SpaceX now leads the rest of the world in successful launches, 64 to 49.
Readers!
My annual February birthday fund-raising drive for Behind the Black is now over. Thank you to everyone who donated or subscribed. While not a record-setter, the donations were more than sufficient and slightly above average.
As I have said many times before, I can’t express what it means to me to get such support, especially as no one is required to pay anything to read my work. Thank you all again!
For those readers who like my work here at Behind the Black and haven't contributed so far, please consider donating or subscribing. My analysis of space, politics, and culture, taken from the perspective of an historian, is almost always on the money and ahead of the game. For example, in 2020 I correctly predicted that the COVID panic was unnecessary, that the virus was apparently simply a variation of the flu, that masks were not simply pointless but if worn incorrectly were a health threat, that the lockdowns were a disaster and did nothing to stop the spread of COVID. Every one of those 2020 conclusions has turned out right.
Your help allows me to do this kind of intelligent analysis. I take no advertising or sponsors, so my reporting isn't influenced by donations by established space or drug companies. Instead, I rely entirely on donations and subscriptions from my readers, which gives me the freedom to write what I think, unencumbered by outside influences.
You can support me either by giving a one-time contribution or a regular subscription. There are four ways of doing so:
1. Zelle: This is the only internet method that charges no fees. All you have to do is use the Zelle link at your internet bank and give my name and email address (zimmerman at nasw dot org). What you donate is what I get.
2. Patreon: Go to my website there and pick one of five monthly subscription amounts, or by making a one-time donation.
3. A Paypal Donation or subscription:
4. Donate by check, payable to Robert Zimmerman and mailed to
Behind The Black
c/o Robert Zimmerman
P.O.Box 1262
Cortaro, AZ 85652
You can also support me by buying one of my books, as noted in the boxes interspersed throughout the webpage or shown in the menu above.
Some amazing milestones on this flight. 2nd flight of the gigantic 1st stage, 2.5 x Saturn V, all 33 engines including 29 reused worked fine through staging. No engine explosions during 2nd stage burn. But 2nd stage is still a problem. Elon is going to sleep on the factory floor, guaranteed.
DO NOT bother reading the comments. Pointless. And I didn’t even go there. Probabilities.
https://arstechnica.com/space/2025/05/elon-musk-turns-his-focus-back-to-space-says-starship-and-mars-matter-most/
This is all very exciting, and while I am always looking forward to the next launch, I am beginning to wonder if SpaceX needs to slow down – on its own – just a bit for more testing. Is everything else in the launch process humming along so efficiently that there hasn’t been sufficient time to get all the other pieces to where they need to be for complete.y successful launches?
Thoughts?
Eric Berger landed himself a valuable interview. I guess Elon didn’t take the final chapter of REENTRY too badly.
This answer in particular: Eric asks him if they think they have solved the issues that caused the explosions in the last two flights in the upper stage engine bay.
Elon also reaffirms the plan — hope? — to use Raptor 3’s in flight by the end of the year. I guess we’ll see.
Like the Everyday Astronaut, live viewing made the ascent seem slower than the replay –that I assumed was from this launch.
Still, there was just a touch of an Astra Powerside as it rose.
They gave the upper stage every chance –and still botched it.
But the SLS bashers who call this a “success” won’t bat an eyelash.
Deprogramming their only hope.
Jeff,
It’s a step forward. They got to SECO. They reused a booster.
And, of course, they got a lot of data.
Again: SLS should be cancelled even if Starship and New Glenn were terminated today. It’s an active hindrance to our future in deep space.
A little disgusted with the repetitious fuel leaks that are repetitiously “mitigated”.
What use is a vehicle without fuel? Seems like elliminating all fuel leaks should be priority number 1.
Maybe Elon could submit a consulting contract for some Honda engineers from Torrance.
SpaceX is getting data on issues that probably can’t be obtained by ground testing. I suspect they’ll address those issues and get to test flight 10 ASAP.
Has any other 2nd stage failed while sub orbital on 3 consecutive flights? If Starship fails like this while orbital doesn’t it damage other satellites?
Steve Richter: The history of rocket is replete with many similar failures. SpaceX’s track record is actually remarkably good, especially considering what it is trying to accomplish.
Don’t get spoiled by its success, and remind yourself of the astonishing accomplishment here. They reused a Superheavy booster on only their ninth test flight. If anyone had told us that they would do such a thing before the first flight we would said they were insane. Superheavy is a monster doing things considered impossible by the entire rocket industry only three years ago, and on top of that they have already flown one twice!
As for the risk to other satellites, there is none. The orbit chosen is very low, about 100 miles in altitude, so that the atmosphere will bring the spacecraft down even if everything fails — which is exactly what happened today. That orbit is thus not used by any other satellites, for the same reason. Most orbit considerably higher.
Patrick Underwood,
I don’t think the factory floor is where the trouble is. Elon needs to set up a cot in the Starbase engineering offices.
Your advice about the Ars comments is worth following. I read the first seven as a statistical sample. Based on that, there isn’t an edible apple in that whole barrel.
F,
SpaceX needs to speed up, not slow down. It did vast amounts of ground testing between IFT-8 and IFT-9. Some of it paid off, but there are still obvious deficiencies in the Block 2 Starship design – obvious once in-flight anyway, but not so obvious on the ground. The quickest way to run all of these issues to ground is to work 24/7 on design fixes and get IFT-10 launched ASAP. If there are still problems, lather, rinse, repeat. The Starship football game has to be won, whether via a Hail Mary pass or a lot of three-yards-and-a-cloud-of-dust ground gains does not, in the long term, matter.
Richard M,
Over at Rand Simberg’s blog there was some recent speculation that Berger might have worn out his welcome in the highest levels at SpaceX. I took the view that he had not, but that it wouldn’t likely affect his coverage much even if so. Seems I was right on the first point and we’ll not have to find out about the second.
Elon proved accurate in assessing the efficacy of the mods made to fix both the IFT-7 and IFT-8 failures. But he was also right that the thing most needed right now is TPS re-entry performance data and IFT-9 probably provided only marginally more such than did IFT-7 and IFT-8. On a preponderance of evidence basis, I have to disagree with our host and call IFT-9 a failure even if it wasn’t as much of one as IFTs 7 & 8. Leaks in the main tankage are definitely a step backward.
Super Heavy B14 performed very well on its second outing. The controlled flip induced by the hot-staging worked perfectly and all 13 inner-ring engines relit for the boostback burn. It seems that the more aggressive angle of attack cranked in for descent was just a bit beyond the actual envelope of the current design but perhaps the next Super Heavy block upgrade will prove more resistant. A bit of dial-back seems indicated for any such testing on upcoming launches using the current Super Heavy design.
Elon’s remarks about the Moon can, I think, be reasonably paraphrased as, “Either get serious or stay home.” I entirely agree. I hope for an early announcement of an unfunded Starship-based SLS-Orion replacement once Isaacman is confirmed a couple of weeks hence. I’d also be pleased to see the same for something based on New Glenn. This seems a good time to get NASA entirely out of the development funding game anent new manned spacecraft.
We will, indeed, see if Elon’s prediction of Raptor 3-based Super Heavies and Starships flying by year’s end comes to pass. I would put the odds of that happening at better than 50% right now.
Jeff Wright,
I always watch the SpaceX webcast. The IFT-9 ascent looked like most of its predecessors. The only really pokey ascent was IFT-1. The “powerslide” is deliberate to save wear and tear on the Pad A infrastructure. The IFT-9 controlled boostback flip was a definite improvement on the randomness of previous hot-stagings. Boostback relights also seem to be a solved problem now. And, of course, Starship made it to SECO this time with all engines intact even if things rapidly went downhill from there.
All that said, as I noted above, this was still an overall failure even allowing for some definite successes and progress. The main objectives were the Starlink test deployments, the Raptor re-light and an intact descent to gather TPS performance data. The first two were complete failures and the last was probably all but a complete failure. The telemetry connection appeared to last long enough to send back at least a bit of TPS/re-entry data, but the uncontrolled nature of the re-entry may not render this of much use. We shall see.
I think you and the dwindling cadre of SLS obsessives who sneer at Starship failures are a bit like the elderly lady who is alleged to have once upbraided Winston Churchill for being drunk. “Yes, Madame,” he is said to have replied, “I am drunk. But you are ugly. And tomorrow I shall be sober.” Starship isn’t quite on the wagon yet, but it’s getting there because its existing defects are fixable. SLS’s two biggest defects are its cost and its glacial cadence. Neither of those is fixable.
Believing otherwise or simply ignoring these intrinsic shortcomings marks out the SLS fan as the cultist in this picture.
Deplorable Dave,
The leaks on the last three Starship IFTs have all been in different places and have had different causes. Those that afflicted IFTs 7 & 8 appear to have been remedied. The leaks on IFT-9 were in the main tankage, something we have not seen before. That fact makes them retrogressive so we have here a literal case of two steps forward, one step back. That isn’t good, but it also isn’t the end of the world, just of this mission.
Anent Honda, I daresay Elon may well have a shot at picking up some of its engineering talent before long. None of the Japanese car companies are in what one could call radiant good health these days. About all one can say for Honda is that it is not doing quite so badly as Mitsubishi, Subaru and Nissan but that is, at best, damning with faint praise.
Steve Rogers,
Exactly. Nice to know I see things eye-to-eye with Capt. America.
Dick Eagleson: You say in two comments that the Starship fuel leaks were in the main tankage, but that was not what I understood based on what the announcers said during the flight. To me it sounded like there were leaks in the system for the attitude thrusters, which I have assumed is an entirely different system.
If I am wrong I very much would like to be educated.
Did Musk postpone his Mars presentation?
Gary: You need to refresh your browser. Do so, and look at the updated post where Musk’s speech live stream is embedded. He did.
“”Superheavy is a monster doing things considered impossible by the entire rocket industry only three years ago””
“”Rather than recapture [SuperHeavy] with the launchpad chopsticks, engineers instead decided to push its re-entry capabilities to their limit.””
“”SpaceX needs to speed up, not slow down.””
Some of my first memories are just before the Mercury 7 astronauts actually flew. I have vivid memories of rocket tests where the rocket either blew up on the launchpad, or just after launch.
During President Trump’s first term, SpaceX was launching/testing almost once a month. I remember some “test to failure” events where they did exactly that. Fill a fuel tank until the tank metal fails. To the uninformed, it might look like an accident, or a setback. The Drive-By Media would definitely take the low road with any reporting. .
The ability/audacity/chutzpa to “push the monstrous SuperHeavy to its limits” during reentry/return is simply, well, the best phrase that comes to mind is: waaaay cool.
Wow, I just remembered one of my favorite Robert Heinlein short stories: Blowups Happen.
wouldn’t worry too much about the leaks, the Raptor 2 interface is obsolete anyway, the Raptor 3 Starships should perform better since their plumbing is all integrated
honestly it’s sort of funny to me that they bother to launch these Raptor 2 ships at all, but of course it’s the only way to get data
Musk has often commented on the drawbacks of the bolted flange piping interface, and the Raptor 3 design eliminates them wherever possible, which is part of its remarkably clean and simple configuration.
But for the necessity of connecting an engine to the propellant supply, it is difficult to imagine an alternative that respects the need to change out an engine rapidly and efficiently.
Did anyone else notice that just before the first stage failed on reentry the whole engine area flared and caught fire?
I do not think this was just some captured fuel but instead reentry heat. If it burned through the cooling jacket of the nozzle, when the engines restarted ………. Boom. If it was still dumping LOX at that time the extra O2 environment could have helped a burn through.
Until then the booster was coming in just fine.
DD says: “A little disgusted with the repetitious fuel leaks that are repetitiously “mitigated”.
What use is a vehicle without fuel? Seems like eliminating all fuel leaks should be priority number 1,… ”
Don’t substantively disagree. Still ……
One of the drawbacks is that they are working with cryogenics for both fuel and oxidizer, I’d think. RP-1 is a lot better behaved with a half century long experience base in rocketry. The Starship stack is breaking new ground with a new fuel/oxy mix. And the real fun isn’t going to happen until they start developing ship to ship refueling in a year or so. Not so difficult with hypergolics, really exciting with cryogenics. They need refueling to reach anything out of LEO.
Also note that there are a lot of test articles under construction or already built at Starbase including what may be the first Block 3 Booster / Ship vehicles. They may have enough to ramp up testing cadence to a flight every couple weeks as we get thru summer and into fall. Cheers –
PS: You guys probably already knew this, but I just realized that Superheavy has more engines than the Soviet N-1 did (33 v 30).
Robert Zimmerman,
On the IFT-9 video at spacex.com, just before the 56:00 minute mark, the POV changes from the engine bay to a camera on the trailing edge of one of the upper flaps looking down along most of the length of the Starship. There is pretty obviously something venting to space from a place on the lee side (no TPS tiles) of the hull that looks to be slightly above the level of the top end of one of the aft flaps and somewhat to leeward of it.
The POV during the last couple of minutes of Starship’s ascent burn switches back and forth between the engine bay and this upper flap camera. There are brief views of this same area of hull shown starting at approximately the 54:12 and 55:05 marks. The first of these is while all six engines are still firing and there is no evidence of a leak. The second of these starts just before the three RaptorVac engines shut down but cuts away to the engine bay view again before the three sea-level Raptors also shut down. Again, there is no evidence of the leak following the RaptorVac shutdowns.
The view stays in the engine bay through the shutdown of the three sea-level Raptors. There seem to be a number of small pieces of debris that tumble past the engines and out the engine bay skirt during this sequence. There is also some pretty significant venting from engine bells, and perhaps elsewhere, in the engine bay.
The POV switches to the upper flap cam again at 55:56 and the leak is now clearly apparent. The leak appears to come from a featureless section of hull that has no obvious fittings, pipes or anything else on it. That makes it seem to be either a small crack in a joint where two hull rings are welded together or a small crack in the material of one of the hull rings itself.
The leak is continuous and significant. That also appeared true of the engine bay venting. Thus, it is unsurprising that the combined venting eventually overwhelmed the attitude control thrusters – which, themselves, work off of ullage gas in the main tanks – and the whole Starship began to tumble.
Having now looked much more closely at certain portions of the IFT-9 video than I did upon first viewing in real-time, I have come to suspect that perhaps the demons of IFT-7 and IFT-8 have not been entirely exorcised. A less energetic, but still significant, hard shutdown of one or more of the sea-level Raptors could have caused a “burp” back up the engine plumbing and into the tankage that opened a hull crack or cracks and also tweaked some plumbing joints sufficiently to start them leaking.
I say “crack or cracks” because a bit later in the video, where we are shown the cargo bay interior, there are a lot of small white blobs moving around. These are most likely bits of methane or oxygen ice or slush caused by an internal tank or plumbing leak. An event, of whatever kind, energetic enough to cause even a couple of small cracks in hull metal might also have tweaked the whole hull enough to account for the inability of the payload bay “Pez” door to open.
It will be interesting to see what SpaceX eventually reports about IFT-9, but I think the company has more work to do than seemed, initially, to be the case.
Ronaldus Magnus,
As a fellow Boomer old fart, I also remember the entire Space Race era and was also a major fan of the late, great Robert Anson Heinlein. Blowups do happen, especially when engineers are wrestling with, as RAH put it, “the fitful demons that live in their engines.” It’s a shame he isn’t around to see what the NewSpace era has wrought. He was right about so much, including not least the need for single-minded SOBs like his own Delos D. Harriman – of whom Elon Musk is a real-world instantiation.
Dick Eagleson: Thank you for the detailed analysis. It suggests SpaceX still has some serious engineering changes to make Starship work.
A good point. It is a possibility.
I thought they didn’t open the door and launch the Starlink simulators because the ship had a substantial yaw rate.
This started shortly after the engine shutdown. I saw it early on and remarked to my wife that this didn’t look good and they were going to lose the ship when the yaw didn’t stop.
Tall Dave,
You will likely be proven right. SpaceX’s own explanation of the IFT-7 & 8 mishaps implicates engine bolted-flange joints as the source of initial leaks. I speculate that off-nominal energetic events caused by such leaks could send shock waves back up the feed plumbing and also cause leaks, or even breakage, well upstream of the engines.
Any flange joints above the engines would not be designed to withstand internal engine pressures, just those much lower pressures associated with propellant feeding. This is the whole point of having pump-fed engines in the first place, to allow propellant feed pressures to be vastly less upstream of the pre-burners and pumps than downstream of them.
It’s exactly these downstream pressures that the V3 Raptor is designed to more effectively contain than the current V2/2.5 models seem able to do. From a pressure containment standpoint, the V2/2.5 Raptor design seems to be in a “sour spot” where the higher-than-V1 internal pressures that allow the greater performance of V2/2.5 engines vs. V1 engines are still not fully under control as the V2/2.5 design still has a lot of the bolted-flange interfaces carried over from the V1s.
With V3, this should, indeed, go away. The V3 design still has a few bolted-flange joints, but they are all smaller than the average size of those on the V1/2/2.5 series. If this eliminates any future unintended “energetic events” – especially at engine shutdown – then there will be no “water hammer” effects to upstream plumbing and tankage.
Let us hope we reach such a happy state of affairs by year’s end or even earlier.
Ray Van Dune,
You are correct. But if I am correct that problems with upstream Starship low-pressure plumbing are entirely the result of initiating events that occur due to deficiencies in high-pressure downstream plumbing, then eliminating the latter will also automatically eliminate the former and there will be no need to re-invent Starship’s low-pressure plumbing.
Let’s all wish fair weather and following seas to the Starbase engineers.
pzatchok,
What you saw there was entirely normal and has occurred on every Super Heavy re-entry that got far enough into the atmosphere. You should be able to find plenty of footage from previous Starship IFT missions to confirm this.
There is a layer of TPS material that covers all of the area in the Super Heavy engine bay that is not actually occupied by engines. Once the descending booster hits sufficiently dense air, this material glows orange, with the engines – that are not yet running at that point – appearing conspicuously black in contrast.
agimarc,
In some quite important ways, RP-1 is not better behaved than cryo methane. When kerolox engines run fuel-rich, as all rocket engines typically do, RP-1 generates a lot more soot than methane. RP-1 is also subject to generating coking deposits in an engine’s tight spots that need to be routinely flushed out with solvents as part of reuse preparations. When all rockets were expendable, this didn’t matter. For reusable rockets it matters.
RP-1 is also a lot warmer, even when chilled and densified, than is LOX. This makes the design of common tank bulkheads thermodynamically trickier than for methalox tankage. As cryo methane and LOX are far closer in temperature, especially when both are further chilled below their boiling points and densified during loading, the thermal contraction of tankage is nearly identical over most of the vehicle, unlike the situation with kerolox.
There are also the matters of natural gas being far cheaper to refine into cryo methane than even ordinary kerosene is to refine into RP-1 and of methalox having a better achievable Isp than kerolox, both at sea level and in vacuum.
Nor – other than boil-off mitigation – do cryo propellants present any particular challenges regarding their bulk transfer in zero-G. Hypergolics are always attended by the possibility of unplanned mixing at scales ranging from negligible to catastrophic. Bits of methalox finding one another don’t have the same potential for damage causation.
You are correct that possible Block 3 componentry is beginning to be seen around Starbase. There is precedent during previous phases of Starship development for already-built articles to be scrapped in favor of moving newer construction items to the head of the line for testing. Perhaps we are about to see another such chapter of Starship development history be written. “Excitement guaranteed,” as Elon likes to say.
Super Heavy does, indeed, have three more engines than N-1’s first stage did. And it is slated to get a couple more. But Super Heavy has already exceeded N-1 in pretty much every other relevant metric as well, including number of flights and level of success. And, of course, Super Heavy has also demonstrated both recovery and reuse capability – two items that were never on N-1’s to-do list at all.
Mike Borgelt,
The commentary on the IFT-9 mission webcast indicated that at least one unsuccessful attempt had been made to open the Pez door. Perhaps the flight controllers also saw the deteriorating attitude control situation and were trying to salvage at least one of the major goals of the mission to whatever extent proved possible. In the event, it proved impossible, but them’s the breaks of the game. Sometimes you eat the bear and sometimes the bear eats you.
Boostback seems fine for Falcon–then too that core is narrow.
On Mythbusters, they tried to have extinguishers and a flamethrower’s emission meet.
Could narrow kerolox boostback plumes be more effective than methalox emissions?
I really like the old idea of Space Freighter.
Scale up SS/SH, but use wings.
That might allow Raptors a more benign environment.
I know Elon doesn’t much care for wings–but there might be some aviation outfits that could help out.
“So after the first firing, it turns out that’s what caused some of the bolts to loosen a little bit; like some of them, some of the time, would loosen and that would allow basically fuel and gas to combine. ”
Can they not wire the bolts on these things? I thought that was a common practice in some aerospace applications.
Not an AS engineer, so correct me if I am wrong.
FAA Advisory Circular 43.13-1B, 7-42
tps://www.faa.gov/sites/faa.gov/files/2023-08/AC_43.13-1B_Ch7.pdf
Steve Richter asked:”Has any other 2nd stage failed while sub orbital on 3 consecutive flights? If Starship fails like this while orbital doesn’t it damage other satellites?”
So many people forget that these are developmental test flights. The first stage was tested as a reused booster stage, and while its reuse was being tested, they also used the booster to test the limits of methods for the booster’s reentry.
Starship is still undergoing basic research. What spacecraft has ever had such a large hatch as the “Pez dispenser” — which has to close for reentry? No other spacecraft has been that large, and large structures behave differently than smaller ones. They have a satisfactory thermal tile, but they are still optimizing, testing a wide variety of concepts, including failure modes (failure of tiles, not of stability of attitude control). Later, SpaceX will have to develop other new concepts, such as transfer of huge quantities of propellants and how to prevent boil-off losses; a manned capability of a huge volume, including docking and survival during the wild landings; and a resealable clamshell fairing and release mechanism for other payloads.
They are also going to change the engine design in future flights. The development is not nearly close to almost being complete.
If Starship fails like this while on an operational orbital flight, then it is little different than the hundreds of upper stages that have been left in low Earth orbit after launch. SpaceX has been careful about its upper stages, and they launch into low initial orbits so that if the upper stage fails to perform the retro burn for a controlled reentry, then the upper stage will reenter anyway within a few months. Many operational Starship flights might be performed in a similar manner.
Spaceships don’t need wings.