SpaceX’s Starship V3—still a work in progress—mostly successful on first flight
- Thread starter JournalBot
- Start date
That doesn't change the underlying fact that satellites of almost every kind are very long leadtime devices, and designs that utilize new launch capabilities don't generally arrive before the new launch capabilities arrive. The market is, at the present time, just not that large.
I'm not at all saying that the market won't change. It will. But the changes required to reasonably utilize Starship will be some time in coming, and SpaceX (or any similar provider) will need to support themselves while the world catches up with its new capabilities. In Starship's case, it may be that orbital launch may only provide launch to LEO, rather than higher orbits that the current satellite makers and users are accustomed to; i.e., satellites will be expected to deliver themselves to their final orbits from one or a few standard LEO orbits. It may be cheaper for them to deal with that by buying and using a disposable orbital tug or use built-in thrusters of one sort or another, and it may be some time before the market decides on the best approach(es).
So... even if it's a raging success, it may need life support until the market catches up. Presumably Starlink is that life support, if the pooch isn't screwed by other factors.
Upvote
8
(8
/
0)
What makes it relevant? The specific problems with the N1 were multiple, and largely derived from things that aren't problems with Starship/SH. Specifically, the engines couldn't be tested prior to first use, they weren't reliable, and the state of computing was such that controlling that many engines and maintaining flight control was beyond state-of-the-art at the time.
Neither of these has been an issue with Starship: each engine has far more computing demand and capability than the N1 rocket had in total. Each engine can be fired prior to use, and always are. And the overall control system has caused very few, if any, problems. SS/SH's scale has clearly caused some problems, and the ongoing development of both the rocket and the engines has caused some issues. But the biggest factor is almost certainly the need to guarantee flight safety with a fully reusable second stage (i.e., a second stage expected to survive re-entry) constrains development in substantial ways that the N1 was not constrained.
That is a major difference, and also due to the development strategy. It's reasonably likely that SS/SH will soon demonstrate both successful micro-gravity engine relight and re-entry. Once those are established, SS/SH will almost certainly start delivering satellites to orbit for Starlink, and development will continue in parallel with operation.
Upvote
23
(23
/
0)
They even did demonstrate this several times already, in flight 6, 10 and 11. It's just that this didn't help much since they immediately moved on to the next version with many changes again, because the payload capabilities just weren't there.
Now v3 is said to finally become the production version. But will it? I somehow have the suspicion that at some point in the near future they will confess that it still doesn't manage the 100 tonnes of payload they want to have and need v4 with stretched tanks for this.
I find it quite alarming that they never stated the payload mass in flight 12 or just confirmed that the Starlink simulators indeed were the same mass as the real things (1.9 tonnes each). Even then this would have been hardly more than 40 tonnes.
And yes, nothing of this has to do much with the number of engines, just with them wanting to have a fully and cheaply reusable second stage. Although the fact that even in flight 12 with the third block of their launcher and the third version of Raptor they had one engine conk out in both stages during the ascent phase despite having made (and tested) hundreds of them now isn't exactly a good sign.
Upvote
-10
(0
/
-10)
No. SpaceX has already successfully used 30+ engines at once, disproving that 30+ is impossible.
We don't even know if the engines themselves were an issue this time. But they have worked in the past, which means they can work, and so N1 failing tells us nothing.
ETA: Ninjas everywhere!
ETA2: How did I get so far in the past?
Upvote
15
(15
/
0)
In addition, larger engines risk combustion instability.
Raptor V3 - 2,750,000 Newtons (Sea level)
RS-25 - 1,860,000 Newtons (Sea level)
These are, already, very large engines by any measure.
Upvote
0
(3
/
-3)
Thomas Cochrane
Ars Centurion
Given how long pad1 will take to rebuild, they're going to have to choose whether to catch starship or the booster on some of the future flights, right? Presumably that'll be starship, so that they can check heat tile longevity over multiple flights.
Upvote
1
(1
/
0)
Bill Swallow
Ars Centurion
No, they do not have to choose one or the other to catch. One pad can easily catch both (assuming the booster does not crash and destroy the tower).
Super Heavy is back and caught within, what, ten minutes or so? Even if you grant an hour for the tower to set Super Heavy down on the launch mount, it can still do that in plenty of time to have the chopsticks open and ready to catch Ship when it comes down an hour or two after launch.
Run the time yourself - from launch to when Ship comes down in the Indian Ocean. Then add the time required for Ship to make it back to Boca Chica.
We've seen how quickly the booster can be cleared from the chopsticks on earlier tests.
Upvote
16
(16
/
0)
On your last, it depends on WHERE on Mars. Equatorial Mars is warming during the daytime than Antarctica is. The appropriate low latitudes in their summers are also quite a bit warmer in the daytime. Night time, yes temperatures drop rapidly and far, even in summer, but the atmosphere is so thin, convective heat losses aren't too bad.
https://en.wikipedia.org/wiki/Climate_of_Mars
Look at the Gale crater temperatures, nighttime temps are only a little colder than Antarctic temperatures, but daytime is a lot warmer throughout the year.
But generally speaking, yes, Antarctica would be easier to survive on than anywhere on Mars (or the moon, or anywhere else in our solar system, probably. POSSIBLY not on some of Jupiter or Saturn's moons under their icy crust. MAYBE. Probably still easier to exist anywhere on Earth (or under Earth's waters).
Upvote
8
(8
/
0)
Does Starship have enough cross-range to do a once-around back to the launch site? I wouldn't be surprised if they have to wait several orbits for the launch site to be back on the orbital track.
Upvote
19
(19
/
0)
Bill Swallow
Ars Centurion
That is a good question, and I'll admit it's one that I hadn't thought to ask myself. Without doing any research whatsoever, my gut feeling is that it does, but as this is Ars, I look forward to some of the extremely knowledgeable people here to show up with all the numbers.
Thank you for pointing out my omission.
(But regardless of whether Ship can make it back in one orbit, or would have to loiter, it still seems like there's no way it can make it back home before the tower can park the booster.)
Upvote
11
(11
/
0)
Of course it can work, no doubt. Mostly. The failing engine during the first stage burn certainly didn't have anything to do with the stage separation.
But can it work reliably enough to enable quick and cheap reuse without lots of refurbishments in between flights? Because this is the point.
I think some people have defended SpaceX against utterly unfair attacks for so long that this has become a habit even with things that can't be defended. I'm certainly guilty with this too.
Look, with Artemis 4 (which is scheduled to happen less than two years from now, in early 2028) survival of the crew will depend on the Raptors for landing them on the Moon (apart from the last 100 meters or so) and launching them back to Orion in lunar orbit and this AFTER the very same Raptors launching HLS into LEO, doing the TLI burn and braking HLS into lunar orbit, with no inspection or refurbishment in between. Do you see this happen? I don't.
Before this launch I thought if from now on with the Raptors in their third iteration they finally are rock-solid with not a single failure in test flights at all anymore SpaceX HLS still has a fleeting chance to happen as contracted, even if delayed. But with even in the 12th flight the third iteration of Raptor with hundreds of them being built and tested still having reliability problems during a test flight HLS is just dead.
It's not enough that the Raptors can work, they have to work.
Upvote
3
(4
/
-1)
We have already seen prototype Tanker docking interfaces (the 'drogues') on Ship 39, in flight 12. They are not androgynous: they're 'female'. To perform docking, there will need to be a 'male' counterpart, with the corresponding 'probes'. Additionally, there will need to be an articulating propellant transfer connector, and that will be on one of the Ships, not both.
In terms of mass to orbit efficiency, it makes sense for as much as possible docking- and transfer-related dry mass to be located on the Depot, so that Tankers wouldn't have to drag that excess mass to orbit and back with every propellant delivery launch.
Similarly, ordinary Starlink Pez Ships don't need to haul the extra mass of the docking hardware, and will fly without it. The same applies to the eventually generic LEO cargo Ships. Currently, SpaceX mounts the docking hardware onto Pez Ships only for testing purposes - because they haven't yet built a proper Tanker to test with; it doesn't mean they intend to put this chimeric configuration into full operation, at scale.
Upvote
6
(6
/
0)
SpaceX says Starship v3 can stay in orbit for 48 hours. And for things like tanker flights it certainly won't be able to land after one orbit anyway.
Upvote
12
(12
/
0)
I think it's more likely they will leave the Ship in orbit for 12 hours or longer, before bringing it back for a landing. That way, the deorbit track can be perfectly lined up to the landing site, with no need for cross-range maneuvering; and it gives them plenty of time to clear and check out the pad before bringing in the Ship - rather than rushing it.
Maybe 90-minute Ship returns are a thing of the future, but that's for when both the rocket and the pad are highly reliable and thoroughly well characterized systems (or conversely, when they have dedicated Ship landing towers.) For the next couple of years, I don't really see that happening.
Last edited:
Upvote
16
(16
/
0)
Again: N1 has absolutely zero useful input for answering those questions. That is the point.
Upvote
19
(19
/
0)
I wouldn't say HLS is "just dead". Delayed, that's for sure. Isaacman's new schedule is still fantasy-land BS.
As for engine reliability, one way to see it is focusing on the engines that failed. Another way to see it, is focusing on the engines that didn't fail. 32 out of 33 on the Booster completed the ascent burn successfully. 5 out of 6 on the Ship also completed a successful ascent burn - and even, a longer one than normal. So, that tells me there is no fundamental problem with Raptor 3's design: when an engine is good, it works as intended. So the problem seems to be with either quality control or manufacturing variances (or both); these will be chased down and addressed to ensure uniform performance and reliability across all engines. This takes time, but it's not a matter of unresolved technical risk; it's more of an operations thing.
Upvote
10
(11
/
-1)
Right. Apart from having to mass-produce engines for as cheap as possible and still doing all the quality control to ensure that every single engine works and not just most of them.
Upvote
-19
(0
/
-19)
Or aim at a landing after the first orbit and reserve a longer loiter for the case of having to check something out or repair with the catching tower. Or range safety or whatever.
Upvote
0
(0
/
0)
Whatever, but they have to solve it or better: Should have already solved it. They have now made how many Raptors? Must be about 800 or so. And still 1 out of 33 engines in the booster failed and 1 out of 6 on the ship failed in the latest flight. These things should be as reliable as Merlins now. How often do these fail? Very, very rarely.
Upvote
-12
(0
/
-12)
Yet again (sigh): This has absolutely nothing to do with N1. Why are you even posting this crap?
Upvote
13
(14
/
-1)
Merlin’s were developed from before 2006 through 2018 and had multiple failures in their early revisions. Raptor has been under full development since 2012 and has had only three real revisions. It’s not a surprise that a much more powerful and complicated engine is taking a little longer to be production ready.
SpaceX built hundreds of Raptors because Elon’s experience with F9 taught him that optimizing the factory to mass produce something is just as hard as designing it and that he will need that capability for Starship to meet its goals. I’m sure a lot do Raptor V2s were thrown in the trash as the factory processes were developed.
Upvote
8
(9
/
-1)
The preceding two generations of Raptor don't have much to say about Raptor 3, since the latter is such a radical redesign. To date, SX seems to have built fewer than 200 production-quality R3's (and some dozens? of development prototypes prior to that).
IMHO the bigger remaining risk with Raptors isn't their initial reliability, which I expect to be ironed out over the next year or two. The one thing that still worries me is, longevity. For how many hours can a Raptor 3 fire, and how many engine restarts can it tolerate, before it's too worn out to remain reliable? (I worry about the Raptor more than I do about e.g. BE-4, because whereas the latter deliberately operates under relaxed conditions, SpaceX's approach with the Raptor seems dead-set on pushing the absolute limits of materials and performance.) That is what will determine the ultimate level of reusability and the steady-state operational costs of Starship.
At least, up until Raptor 4 comes along...
Last edited:
Upvote
13
(14
/
-1)
Because this is about Starship, not the N1. Why are you talking about the N1 all the time?
Upvote
-11
(0
/
-11)
I would like to point out that we do not know that any of the engines that either failed or failed to relight were caused by the engines themselves. They could have all been prop issues before the engine. This is a new rocket, valves and prop line problems seem to be common issues with all new rockets.
Upvote
17
(17
/
0)
You need to pay attention to what you are replying to so you understand context. The post you replied to was me replying to this post:
If you're replying to my post in a context vacuum, that's on you.
Upvote
9
(9
/
0)
We don't know what the problem is yet. SpaceX might. It may not be related to the new engines: it could be a sloshing problem caused by out-of-spec maneuvering; it may be a new failure mode that only shows up in multi-engine situations. It may well be something else entirely.
Upvote
10
(10
/
0)
Not only that, imaginary numbers show up too. Cell phones don't work unless the code keeps the imaginary numbers in the calculations.
I recall a physics experiment (I'm a non-physicist going from an old memory here) where a lasing chamber was excited, then hit with an impulse. As the impulse traveled through the excited chamber, it formed shock waves that aligned with the imaginary parts of the equation.
Google is letting me down right now, but I did find https://forum.centerforinquiry.org/...-part-of-quantum-mechanics-really-exists/7893 and https://www.geeksforgeeks.org/maths/applications-of-imaginary-numbers-in-real-life/
Upvote
1
(2
/
-1)
I believe that was the purpose of the "RTLS banking maneuver" during the simulated landing. It starts at T+1h2m20s.
Upvote
4
(4
/
0)
As far as I understand, the banking maneuver is part of the standard planned descent and approach for landings at Starbase, specifically. The initial approach trajectory targets an overshoot into the Gulf and moreover minimizes or avoids altogether low-speed, low-altitude overflight of Mexican territory as well as population centers (including Starbase itself) on the U.S. side; this is likely a safety precaution in case of a loss of control or some last-minute glitch at the pad, as well as an attempt to minimize the impact of sonic booms upon the surrounding population and infrastructure. The last-minute curve brings it back over the launch pad, approaching from the direction of the Gulf rather than over land in those final moments.
Upvote
9
(9
/
0)
The banking maneuver did seem a little quick for translating 90 minutes worth of earth rotation. But the extra translation just means they need to start the procedure at a slightly higher altitude.
The F9 has a fail-safe landing, where it initially targets open water. Only when the engines fire up does it course correct to target the barge/pad. I could see the Ship banking being related to overflight, but I don't see why they wouldn't use the engines for the fail-safe approach.
Upvote
1
(1
/
0)
I guess they want to drag as little propellants as possible to orbit and back, it's already about 35 tonnes of landing propellants and these come 1:1 out of potential LEO payload.
Upvote
5
(5
/
0)
Yes, of course. Still, if they're still finding new failure modes in their third iteration of the hardware and with the 12th test flight, when will they run out of new failure modes? At some point they will have to arrive at something that reliably works and not just mostly.
And yes, we're used to that now. "Just a single engine, they have more than enough to compensate for this". But this is exactly the kind of "normalization of deviance" that already eats up your safety margins and redundancies you're going to need at some point to save the day.
And I don't think the failing booster Raptor had anything to do with a sloshing problem caused by out-of-spec maneuvering or whatever. They had one out of 33 engines already fail during the first stage burn with basically ideal conditions. Viewed soberly this is a Raptor failure rate of 3% within less than 2 minutes. This is just too much. Can we agree on this?
Upvote
-11
(4
/
-15)
The v3 stack is the first complete design, intended for operational deployment over the next year or two. Right now, you could view flight 12's hardware as a beta release: fully functional and meeting all design criteria, but still somewhat buggy. (Whereas preceding v1 and v2 flights could be seen as using pre-alpha and alpha development builds, respectively). Once it's sufficiently debugged (over the next handful of flights), it'll become the inaugural production release of the Starship product.
Starship v4 is looming in the distance. But at least for now it doesn't look like it'll be as huge of a change as v3 is relative to v2/v1. Starship v4 will use some next-gen Raptor version (Raptor 4, perhaps), distinguished by more thrust (300+ tf) and higher combustion chamber pressures (350+ bar). V4 will also have stretched tanks and hulls on both Booster and Ship. And, Ship will move from 6 engines to 9 (adding 3 more RVacs). So, v4 will probably have some teething pains too, at the start - but not likely as many as v3, and neither v4 nor v3 are likely to have as many problems as v2 did.
Upvote
9
(9
/
0)
Judging from the fact that v3 already has more thrust than it needs and actually has too small tanks for optimal payload: I guess if they don't hit their "100 tonnes of payload fully reusable" target they will very quickly go for stretched versions, call it v3.5 if you want. The only reason they didn't do this already was that the GigaBay isn't high enough and the tower(s) will need to be adapted for it.
Anyway, it's always "this new version now for real!" and after a few flights and problems turning up it will be the next version instead.
I will change my mind if they hit the 100 tonnes mark with v3 and the next flight will work much better than the first one (like landing the booster on the water after a boost back and demonstrate Raptor relight with the ship). I'm not even saying it's impossible, I just can't give them the benefit of the doubt anymore.
Upvote
-7
(4
/
-11)
"...a Raptor failure rate of 3% within less than 2 minutes. This is just too much..."
Are you suggesting that there's a remote chance that SpaceX might settle for this?
From their track record, it's very obvious to me that they won't.
Upvote
16
(16
/
0)
That may be your perception, bit it was actually never the case.
You are probably, like so many others, confusing the "Starship 2" of 2024 with the "v2" of 2025. Whereas, looking at the specs and renderings of "Starship 2", it's very obvious that those specs and renderings refer to what's now called "v3":
(the above is a slide from Musk's presentation back in April, 2024)
The "v2" of 2025 was still using Raptor 2 engines on both stages, which didn't meet the specifications of "Starship 2" thrust. And the first stage (Booster) of v2 was essentially unchanged from "v1": only Ship v2 was significantly different from v1, and it had only a subset of the changes planned for "Starship 2".
The point is that last year's "v2" was just a temporary time-filling side quest where SpaceX evaluated a subset of the planned Starship 2 design, while experimenting with the TPS and Pez dispenser, but above all, waiting for the delayed Raptor 3 to finally complete its gestation.
Note, by the way, that what was called "Starship 3" back in 2024, is now simply relabeled as "v4":
(the above is from August, 2025)
Last edited:
Upvote
9
(11
/
-2)
I'm not suggesting they will settle for this. I'm just saying that this is where they are after three iterations and 12 flights.
I'd love to know their failure rates during ground testing the engines. If it's like 0.1% or so there, it's the booster environment.
Upvote
-5
(2
/
-7)
It does recently seem like, when explaining why Starship will work - imminently - it’s an iterative design process that’s incrementally zeroing in on the finished design, but when explaining why Starship doesn't work - currently - V3 hardware is an entirely new design that should be treated as a whole new rocket. I appreciate that the two things are not completely mutually exclusive, but the net effect does still feel rather dissonant.
Upvote
4
(4
/
0)