Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

Huge congratulations to the engineering team on this one! It's been quite interesting to watch the evolution of Starship. Had they put another couple of tonnes of propellant and oxidizer on the Booster they might have caught both of them. It is still unclear to me what the final mass fraction to orbit will work out to be, but something that is 100% re-usable is better than something 60% reusable[1].

[1] Estimate of Falcon 9's booster + fairings being 60% (by cost, not by mass) of the launch package.

 help



It looks like they delivered 50 tons today. Even if it stays this way, it will still reduce price to orbit significantly.

Unfortunately the payload margins continue to be much lower than what they were aiming for at various points.

Originally they aimed for 150 tons of payload to LEO for Starship V1. Then they lowered that goal to 100 tons. But then Starship V1 only managed to achieve approximately 0 tons, which prompted them to target 100 tons for a larger, more powerful Starship V3 (and 200 tons for Starship V4). But now Starship V3 only achieves around 50 tons to LEO, based on the estimated weight of 26 Starlink V3 satellites.

I think the main issues were that Starship dry mass continued to be heavier than expected, and that the Raptor engines performed lower than they thought they would. Elon Musk recently said the Raptor V3 engines could only reliably achieve 230 tons of thrust per engine (a measure of force) instead of the originally projected 280.


I have every confidence they will work this out. I would hate it if people benchmarked my code whilst it was still in development and extrapolated the future performance from that.

Not being able to achieve arbitrary number (e.g. 200 tones) does not detract from achievements of engineering.

But not everything is achievable just because we want to achieve it. Physics/chemistry also have a say in it.


Well said. I'd also add not everything is unachievable just because it hasn't been done yet. A lot of people won't even update their priors once something is done (belief perseverance). I've lost count of the number of older people that tell me electric cars won't work whilst there is a perfectly functional one sat on my driveway.

A rocket engine can't be optimized during its lifetime like software can.

I want you to look at the evolution of the capacity of Falcon 9 over its various design iterations. There certainly was a lot of optimization leading to substantial growth in capacity.

Chamber pressure, mixture ratio, and throttle limits can be adjusted operationally within the hardware’s margins. Engines on reusable vehicles are regularly swapped, inspected, and replaced. Process changes, minor part upgrades, and material improvements get introduced into the production line.

Flight data drives continuous tweaks to control software (throttle profiles, mixture ratio schedules, startup sequences, health monitoring). Those are software updates


But those are very limited changes. You can't eke out the missing 50 tons of payload this way. That likely requires a significant redesign of the current Starship hardware, in which case they probably have to test a lot of things again (as we have already seen when they went to V2 and V3).

Due to the tyranny of the rocket equation, actually relatively small changes to the engine can result in substation payload gains by allowing more thrust and reducing gravity losses. Even just minimizing the amount of fuel left for boostback can result in tons of extra payload mass. I'm not saying that every rocket could do that, but Starship was designed for 100T+ of payload, and if it's not capable of that at the moment it is likely due to being overweight or engines not being able to put out the amount of thrust in the specs. From what I've seen, it sounds like a bit of both right now. The V3 engines are obviously a massive change (look at the comparison pics against V2 if you haven't seen them!).

> Even just minimizing the amount of fuel left for boostback can result in tons of extra payload mass.

Yes, but this time they actually had too little fuel for landing left in the booster, as confirmed by Elon Musk on Twitter. It ran out of methane, and they couldn't catch it. So next time they have to be more conservative, which would reduce payload mass even further rather increase it.


They don't necessarily need to be more conservative. They need to be more efficient. The two are not the same. A more aggressive approach to the tower (e.g. doing the suicide burn late) might actually result in less fuel usage not more.

Elon Musk said they would do the booster main engine cut off slightly earlier. Which would result in more fuel in the booster but less in the upper stage, which could lead to a reduced payload margin. Certainly it wouldn't lead to an increase.

https://x.com/elonmusk/status/2121648438144024943


They already went from 20 mass simulators to 26 sats, and it's very likely that they have more headroom. For this specific flight, they clearly prioritized Starship fuel margins over just about everything else.

It is highly unlikely that they have anywhere near enough headroom for the missing 34 satellites of the 60 advertised satellites.

> they clearly prioritized Starship fuel margins over just about everything else.

That is not clear. Indeed, there is no indication that this is true.


They left S1 with hair thin margins and got burned by that. In the meanwhile, S2 had fuel to spare and had to vent down during reentry. It's very clear where the priority was.

Yes the mass fraction is much lower than they were aiming for and the question remains, what is the eventual cost per ton to orbit? There are many paths where Starship can fail and few where it can succeed.

This is part of 'watching the evolution' I find so interesting.

Getting the booster up and back, that was just 'scaling up' F9, but it also many 10s of engines firing at once which was a problem that vexed the Soviets. With the last 3 launches have 33 engines start and nearly all of them staying lit, that seems manageable? Certainly within reach of becoming reliable.

"Flipping" from drag flight to powered landing for a propulsive landing. The power of the rockets works against you, how do you keep from shooting off sideways? Not to mention significant control surfaces encumbered by heat shielding.

Heatshielding itself, the bigger the mass you're returning the more energy you have to dissipate with the heat shield, we knew you "could" do this (shuttle proved it) but the shuttle's tile system was hard to maintain and delicate. Could you make that resusable?

Separating the booster and the Starship while minimizing velocity loss. Hot staging was known, it caused a couple of Russian rockets to blow up, and NASA put it on the shelf, SpaceX made it work. (at least well enough for now)

Propulsive landing on a piece of equipment. This was, for me, one of the more interesting adaptations of F9, landing on a barge stationed to catch you out at sea. "Catching" the rocket was a huge step, but necessary because the mass cost of landing legs on Starship is really really high. (only works on the Moon or Mars if you can refuel in orbit since you need do decelerate those legs to landing speed.

There are still a number of challenges to go, on orbit refueling is likely the biggest one. You both need an orbital cryo-fuel tanker and the means to send fuel to it, and get fuel from it.

Big picture though is:

* IF you can reuse the stack a number of times

* IF you can lift at least enough mass to assemble in orbit.

* IF you can refuel in orbit

Then you have the basis for building infrastructure in orbit. There is another "IF" which is whether or not you can do that without damaging the planet. Specifically what are the externalized effects of rockets going up out of the atmosphere on the planet's ability to remain stable (atmosphere, climate, Etc.)

As others have pointed out, doing all of these things was on the NASA 'roadmap' at one time (my grade school self read about how I was going to be able to take a vacation at a Lunar hotel by 2025, and on Mars by 2050. But yeah not so much) the reality it complicated and I love watching a talented engineering team knock down obstacles.


Regarding the shuttle heat shield, it was reusable, but required extensive refurbishment. The question remains how much refurbishment the Starship upper stage will need. For true "rapid" reusability, it arguably couldn't require any real refurbishment at all, only a visual checkup. This might not be possible with ceramic tiles, which are brittle and tend to fall off during launch.

A passive (like VentureStar) or active metallic heat shield (or some of both) would be more robust. But those have never been used in production, and would require considerable research time and effort, apart from open questions like the mass penalty.


> Starship V1 only managed to achieve approximately 0 tons

Are all of these comments comparing hypothetical production loads to small test scenarios?


Typical rocket tests will have dummy mass to simulate a payload (if not some high risk tolerance university project). This is ideal because you want to test as close as possible to the real thing, including things like stress on the vehicle supporting the payload mass. So a vehicle barely making to orbit without any payload mass isn't a great place to be. Starship is being developed differently than most rockets, so they have time to get to the better place, but they're still not where they have stated they expected to be. Today's achievement was incredible, but it seems reasonable to temper the lofty statements previously made about this vehicle's immediate impact with statements of its actual capabilities.

Why? This is the worst version of starship and booster that will ever fly. It isn't unreasonable to look at the giant rocket factory they have produced and think "yeah, I reckon they will continue development at the lightning quick pace they always have done"

> This is the worst version of starship and booster that will ever fly.

That is true for any rocket.

Remember that Starship V1 was announced to have a payload margin of 150 tons to LEO, by around 2021 if I recall correctly. Now it's late 2026 and the payload is something like 50 tons. You may still call this "lightning fast", but it is far less progress than what we thought would happen several years ago.


The difference is the cadence. Also the rapid pace of development, they aren't afraid to change things and go through multiple blocks, getting data by flight testing and making improvements. Look at Vulcan Centaur and how long it has taken and compare it to Starship.

"ULA began development of the new launch vehicle in 2014, primarily to compete with SpaceX’s Falcon 9 and to comply with a Congressional mandate to phase out the use of the Russian-made RD-180 engine that powered the Atlas V. "

"The Vulcan Centaur conducted its first launch on January 8, 2024, carrying the Peregrine lunar lander as part of NASA's Commercial Lunar Payload Services program. Its second flight, conducted on October 4, 2024, as part of the U.S. Space Force's National Security Space Launch (NSSL) certification process, achieved its planned orbit despite the loss of a nozzle on one of the GEM-63XL solid rocket boosters, which resulted in reduced and asymmetrical thrust. Following a five-month review, the Space Force certified Vulcan for NSSL missions in March 2025. In February 2026, a second issue involving a solid rocket booster occurred, after which launches were paused pending investigation."

Centaur has flown 4 times since 2014. SpaceX began active development of the Starship program in 2018.


Vulcan is not relevant to what we were talking about. Again, you can call it "rapid pace of development", but the fact remains that Starship performs much lower, at a much later date, than what we originally thought it would. Since hardly anybody really expected they would be that far off. 50 tons to LEO in 2026 vs 150 tons in 2021. The only lesson is that their announcements about future capabilities have to be scaled downward by a substantial degree, and their timelines strongly upward.

What material are they using for dummy mass?

Unsold Cybertrucks?

    > Had they put another couple of tonnes of propellant and oxidizer on the Booster they might have caught both of them.
Can you explain this part?

They called it off because it was running lower on fuel than they expected - the margins are crazy thin. You can see a green burst of flame near the end, which is what happens when you run oxygen rich/fuel poor (also called "engine-rich" - iirc the green flame is copper from the engine as the oxidizer-rich mix causes it to run hot).

I think it's a supposition that the reason they didn't catch the booster was that it couldn't get back to the chopsticks. I don't think SpaceX described the issues as that ... it would seem sort of strange if it was because the seemingly planned a catch for it so you'd think they would have a good handle on the fuel required. It's possible they didn't fully understand that requirement though.



Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: