Per Wikipedia, SpaceX has landed boosters successfully 79 out of 90 times. Yet Starships seem to be reliably exploding upon landing.

What is so different about the two that Starships are so much more difficult to land? Is it something to do with the multiple engines? Or just that it's a larger scale? It seems like the starships are trying to accomplish something which they've done successfully many times already.

  • 15
    $\begingroup$ They're testing very, very early in the maturity lifecycle of the vehicle. Those boosters failed a bunch of times before they started working. $\endgroup$ Apr 21, 2021 at 18:59
  • 9
    $\begingroup$ If they were missing the pad, or over-hovering, or that kind of thing, that would be a head-scratcher. But I think all these problems they've run into are new problems. $\endgroup$ Apr 21, 2021 at 19:07
  • 6
    $\begingroup$ Hmm, good call, the belly flop and "relight flip" are different, aren't they... I guess the general idea seemed similar enough that I was confused as to why they seem to be having so many failures now... but that makes more sense. Thanks for the clarification, feel free to make that an answer! $\endgroup$
    – sǝɯɐſ
    Apr 21, 2021 at 19:16
  • 8
    $\begingroup$ Falcon 9 landing attempts failed 6 times until the 7th succeeded. The next 2 failed again. (To be fair, there are a couple of ocean splashdowns in there, which may or may not have resulted in a landing.) Starship now has had 5 successful landings in 9 flights, and only 3–4 failures (depending on whether you count SN11's explosion as a landing failure or not – note that there are also a couple of Falcon 9s that didn't land because they failed before they had the chance to land, e.g. CRS-7 and Amos-6). $\endgroup$ Apr 21, 2021 at 20:47
  • 4
    $\begingroup$ Don't forget that SpaceX are also using new engines etc here and have had lots of problems with them... $\endgroup$
    – Moo
    Apr 21, 2021 at 22:10

2 Answers 2


There are a few reasons:

The most obvious but easier to overcome is that the Starship is new. They haven't had the time to perfect everything. For example, the Raptor engines use two separate turbopumps, one for liquid oxygen and one for methane. This improves efficiency but makes the engines far more complex. The Raptor is the first engine of its kind to ever fly, so its current track record is surprisingly good considering everything it has to do.

The main and most difficult issue is the way Starship is landing. Unlike the Falcon 9 which falls engines first, the Starship falls on its side to maximize drag. While this improves efficiency because the rocket is not going as fast and allows for safer re-entry, it creates a problem in that the engines are pointing sideways. While a Falcon 9 just has to light one engine at the right time, a Starship has to light three engines, gimbal them to flip it vertically, then shut off two and land. It is much more difficult and complicated. Combined with how untested the rocket is, it's a marvel that SN10 actually landed without immediately exploding.

  • 2
    $\begingroup$ What I find really amazing is that it's worth bringing it in on it's side until just before landing. Bringing it through the fire on it's side makes sense (and I've done it many times in KSP), but I'm amazed that the difference in terminal velocity is enough to make it worth the games they are playing. $\endgroup$ Apr 22, 2021 at 2:45
  • 2
    $\begingroup$ @LorenPechtel It's not only about landing speed. Much more important is how rocket enters atmosphere. Falcon has sub-orbital speed and uses engines to slow down. Starship will be entering atmosphere with near-orbital speed and it would be very good not to use fuel to slow down. That's why you want to have more drag, and heat resistant tiles (engines would not withstand that amount of heat). $\endgroup$
    – Volvox
    Apr 22, 2021 at 12:35
  • 9
    $\begingroup$ Someone else (I trust) did this math, so double check me, but I hear moving the flip point just a kilometer higher would add take 4-5 tons of additional propellant. It's about maximizing the capabilities, and leaving lots of room for things like life support in the mass budget. $\endgroup$
    – Saiboogu
    Apr 22, 2021 at 12:59
  • 2
    $\begingroup$ @Saiboogu It's not just moving the flip point, it's also burning the engines the whole additional kilometer down. The highly-gimballed Raptors' burn initiates, and ends, the flip. $\endgroup$
    – dotancohen
    Apr 22, 2021 at 13:35
  • 1
    $\begingroup$ @LorenPechtel See the two comments directly above your latest comment. The longer you have increased drag, the less time your engines have to spend burning propellant to slow you down. They could flip as soon as the rocket enters the atmosphere and is no longer 'burning', but then they'd have to fire the rockets for a much longer duration, which means more fuel, which means less space for food or life support systems. $\endgroup$
    – TylerH
    Apr 22, 2021 at 15:46

The experience of landing a booster for SpaceX is actually more relevant directly to the Super Heavy booster, which has yet to fly. (BN1 was built as a pathfinder, then disassembled. BN2 is under construction as of Apr 2021).

Falcon 9 lands its first stage, which is analogous to Super Heavy. However, it does not land its upper stage, which is more similar to Starship.

The landing approaches are different. It is worth noting they did land SN10, though a Raptor underperformed leading to a rough landing that caused an explosion after landing. Getting it that close on their 3rd try demonstrates that the past Falcon 9 experience very clearly resolved a lot of risk.

The engines are very different and start differently. (Starts: TEA-TEB hypergolic 'lighter fluid' for Merlin, and igniters for Raptor). It also uses a completely new type of design with two turbopumps, adding complexity but increasing efficiency. The Raptor engine is still very young in its design life. As of Apr 2021 we have seen serial numbers as high as 66 which means they have built a large number of them (For rocket engines... In the car world, this would be 1 hours output of a factory). There has already been a major revision change (as seen in tank watcher photos) at the SN54 mark (which is mounted on Starship SN15).

The landing profile of the two vehicles is different with Starship, putatively reentering from orbital speeds, needing to burn off more velocity, so using aerodynamic friction to reduce speed more than a Falcon 9 first stage. Though if you watch a Falcon 9 when they show the speed/altitude numbers on the live cast (Like on the last NRO mission, where they were not allowed to show the upper stage for secrecy reasons) it is quite amazing how much velocity is lost just by aerodynamic friction. My recollection was almost 1500-2000 km/hr was lost just through friction, before the landing burn scrubbed off the final velocity.

This means the flip is needed. This means header tanks are needed to avoid sloshing fuel and ingesting bubbles.

That means reigniting a new engine design, while on its side as opposed to the usual upright. (Which is interesting since they use horizontal test stands, and often the worry is that vertical usage might differ).

All these things are new, and they got very very close with the first 4 landing attempts.

Each time they fail, they learn something new, and clearly since the vehicles are cheap enough, and are being built fast enough it is worth trying them out as soon as they are able.

  • 5
    $\begingroup$ AFAIK SN10 did not land correctly. Engine thrust was lower than it should, so landing speed was much higher than required for proper landing. So it crashed, but without explosion (fuel leak was caused by this crash). $\endgroup$
    – Volvox
    Apr 22, 2021 at 12:43
  • $\begingroup$ @Volvox That is correct. I have submitted an edit for this response fixing that error and a few other problems, but it hasn't been approved or denied yet. $\endgroup$ Apr 22, 2021 at 17:09
  • 5
    $\begingroup$ @Volvox Any landing you can run away screaming from, before the vehicle explodes is a good landing. $\endgroup$
    – geoffc
    Apr 22, 2021 at 18:21
  • $\begingroup$ Header tanks are needed for weight balance during re-entry, not to avoid fuel sloshing. Though it's important that sloshing is controlled to avoid ingesting bubbles, that's true of both tanks $\endgroup$ Apr 22, 2021 at 20:31

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.