I get the reasoning behind designing Starship to land upright. It doesn't require landing gear and the turnaround is expected to be much faster. But I'm beginning to question that logic.

I think the cost/benefit of an upright landing is acceptable for unmanned vehicles like the booster. Most of the time it will land just fine, and if it craters once in a while, that's ok. But for a manned vehicle, it's far too risky. If those engines don't light in time, you and your 100+ fellow passengers are dead.

Is the turn-around time that much better? The main difference is that it's horizontal, right? Surely SpaceX can find a quick and easy way to get Starship vertical and in position for the next launch.

Is it really saving any weight? How much more or less does landing gear weigh compared to header tanks? Could you remove an engine or two if you don't need any to land with? How much could you save in fuel weight?

  • 14
    $\begingroup$ Are you suggesting adding wings to it? $\endgroup$ Commented Feb 11, 2021 at 19:44
  • 47
    $\begingroup$ So basically, a Space Shuttle? $\endgroup$
    – Star Man
    Commented Feb 12, 2021 at 0:19
  • 5
    $\begingroup$ I think you just answered your own question. $\endgroup$ Commented Feb 12, 2021 at 3:47
  • 27
    $\begingroup$ Your question is based on a false premise that landing horizontally would somehow be significantly safer. The example of the Space Shuttle shows that this is very much not the case. Without sufficient engine power and fuel to perform a go-around and powerful control surfaces to correct for off-nominal situations (wind shear, asymmetric thrust due to engine failures, energy management issues), you only ever get one shot at the landing and if anything goes slightly wrong, crater again. $\endgroup$
    – TooTea
    Commented Feb 12, 2021 at 7:52
  • 8
    $\begingroup$ As far as I know Starship (unlike Falcon 9) can throttle the engines far enough to hover. So for additional safety in manned landings one could light several engines for landing and/or start lighting them early enough to bring another engine up if the first one fails to light. $\endgroup$
    – Michael
    Commented Feb 12, 2021 at 10:03

8 Answers 8


You're basically describing the Space Shuttle.

The Space Shuttle wasn't even a good solution when it was designed. It had precisely one goal - to look like a plane for the image of the Air Force. As far as engineering goes, the Big Dumb Booster was already well proven, and is what every other solution to space has used. But in order to get Air Force funding, it had to look like a plane. The justification at the time was reuseability - but even at the time, the level of reuse and cost of refurbishment made this questionable, and it was never a problem which actually could be solved because the design simply didn't make it possible.

Your first problem is reentry. Wings are really bad for that, because they're exactly the kind of shape you don't want, especially around the wing root. The leading edge of the wing is also a very vulnerable point. Wing-shaped heat shields basically aren't a good thing, and Columbia demonstrated what happens when you get unlucky. How many Columbias can your winged spaceship program tolerate?

Your second problem is landing it. The Space Shuttle was actually a glider - and as every glider pilot can tell you, landing is stressful. If the pilot of a powered aircraft gets approach wrong, they can push the engines up and go round again. If a glider pilot gets approach wrong, something is going to get broken, and that something is often the occupant(s). There are no second chances. So instead of a highly-automated system of engines and sophisticated control systems, you've replaced this with a pilot who can and will screw up, with no backups or any way to recover the situation. Sweet.

Ah, you'll say, but gliders don't crash that often. Firstly, you're probably not going to hear about it on the news unless someone dies. And secondly, gliders have a glide ratio of around 30:1, are ridiculously manouevrable, and land at about 40mph (fast but not too much) or slower if they can land into a headwind. The Shuttle had a glide ratio of 4.5:1, was notoriously hard to control (because a wing which forms even a partly-acceptable heatshield is not a good wing to fly), and landed at over 200mph. For reference, a glide ratio of 4.5:1 is substantially worse than any hang-glider - in fact it's about the same glide ratio as a skydiving parachute - and that landing speed is faster than an F-16 touches down. This is not a happy place to be, and it's a testament to the insane skills of the pilots (and a healthy dose of luck!) that none of them were lost on landing.

Thirdly, you need somewhere to actually land the damn thing. The big bonus of landing vertically is that you only need a flat patch of ground the size of your landing pads. The Shuttle needed a 3 mile runway to land on. That's 3 miles of perfectly flat ground, with the Shuttle initially rolling at 200mph when it touches down. Good luck with that on Mars.

And fourthly, you need an atmosphere. Earth has a thick enough atmosphere that wings work fairly well. Mars's atmosphere is a whole lot thinner, and designs for aircraft on Mars (there's a helicopter due to land next week!) need to pay serious attention to this. Glide ratios and speeds would be correspondingly worse - as if they weren't already bad enough for the Shuttle on Earth. And the Moon of course has no atmosphere at all.

If you really, really want a wing, then you can go back to Rogallo's work instead. Re-entry uses a normal heatshield, atmospheric braking with drogue chutes gets the speed down, and then a flexible wing is deployed. Since the invention of the ram-air parafoil, it's probably more practical to use that instead though. It may not look as cool, but you can still fly it as normal - it is perfectly practical, fairly robust, and easy to control. It even lands slowly. On the downside though, you have an extra thing to deploy, and any skydiver can tell you that chute deployments do fail, even before you add the extra ways that something can fail when you need to deploy it mechanically from a space craft. SpaceX did consider this, but their assessment was that firing the rockets (which after all are known to work, because they got you off the ground in the first place) is more robust than adding something else which can go wrong.

TLDR: It's not because of the weight of the landing gear at all!

  • 23
    $\begingroup$ This is factually incorrect "It had precisely one goal - to look like a plane for the image of the Air Force. " -1 Also it's factually incorrect that the X-37 is "not designed to land using the wings". WTF? You veer off into crazy land in that paragraph. $\endgroup$ Commented Feb 12, 2021 at 17:13
  • 5
    $\begingroup$ X-37 seems to land using the wings in youtube.com/watch?v=IoailiPGTZQ $\endgroup$
    – Suma
    Commented Feb 12, 2021 at 18:16
  • 7
    $\begingroup$ "The whole program is still very secret" – While that is true, and there are many mysteries surrounding the X-37 program, there is one thing it is not: invisible. You can literally watch it land as a glider with your own eyes, at least if you live on the East coast of Florida or in California, depending on whether it lands at Kennedy Space Center's Shuttle Landing Facility or at Vandenberg Air Force Base. $\endgroup$ Commented Feb 12, 2021 at 19:51
  • 5
    $\begingroup$ Gliders don't land at 60-70mph, it's kmh. Big difference. Most gliders have their optimal performance at about 100kmh, and a stall speed well under 70kmh. $\endgroup$
    – vsz
    Commented Feb 12, 2021 at 20:59
  • 7
    $\begingroup$ The purpose of the Shuttle's wings was not to look like an airplane, it was to give the Shuttle sufficient cross-range capability to take off from Vandenberg into a polar orbit, do something nefarious, and land back at Vandenberg having completed a single orbit. $\endgroup$
    – Mark
    Commented Feb 13, 2021 at 4:03

There is not enough air on Mars. You would need absolutely humongous wings. There is no air at all on the Moon.

Surely SpaceX can find a quick and easy way to get Starship vertical and in position for the next launch.

Starship is not structurally capable of being in a horizontal position. It will simply crumple and/or break in half.

Could you remove an engine or two if you don't need any to land with?

Considering you need only one engine to land but 6 to launch, obviously not.

  • 1
    $\begingroup$ SpaceX is already planning to have different Starship variants, e.g. moon and deep-space. I'm only considering a variant that would need to land back on earth. I think they're going to need Earth, Mars, Moon and deep-space variants. $\endgroup$
    – loweryjk
    Commented Feb 11, 2021 at 22:29
  • 9
    $\begingroup$ The one that takes off from Mars and lands on Earth will be constructed on Earth. So there has to be a model that can land on both Mars and Earth. $\endgroup$
    – Robyn
    Commented Feb 12, 2021 at 4:16
  • 13
    $\begingroup$ @loweryjk: None of those variants require significant changes. The Earth-to-Earth version will replace the vacuum Raptors with sea-level Raptors – all that needs is a different thrust puck. The deep-space variant will remove the recovery hardware (body flaps and landing legs). The "Moonship" variant will remove the body flaps and add landing thrusters in the nose cone – the nose cones are designed to be different anyway. Your proposed design needs a completely different body because Starship is not structurally capable of being in a horizontal orientation. It needs a wing, which likely … $\endgroup$ Commented Feb 12, 2021 at 5:59
  • 11
    $\begingroup$ … needs a wing spar that goes through the entire body – but the body is the tanks and the tanks are the body, so that means the wing spar has to go through the tanks, or you need to completely redesign the tanks. The aerodynamics are completely different. You have to put the landing gear somewhere, you can't put it inside the body because that's the tank. The added drag alone will make it impossible to reach orbit, and I'm not even talking about the huge weight increase. $\endgroup$ Commented Feb 12, 2021 at 6:04
  • 6
    $\begingroup$ @WayneConrad the belly flop maneuver has it steadily supported by air pressure distributed across its entire surface and fins. Supporting its entire weight on landing gear (and more importantly, handling shock and steering loads) would take a completely different internal structure, a much more complicated and heavier structure. And then there's the wings needed for a glide landing which would also put heavier loads on their anchoring structures than the current stubby flaps. $\endgroup$ Commented Feb 13, 2021 at 4:22

In addition to the other very valid answers, at the moment we see Starship flying alone and can easily visualize adding wings, but in orbital mode it will be stacked atop the superheavy booster. Adding large wings to the top of a rocket makes it very unstable, akin to making a dart fly backwards source, and would need to be stacked at an odd angle to zeroise the wing lift during ascent.

Space shuttle and Buran solved these problems by stacking the winged element on the side of the rocket to keep center of lift and drag close to center of mass. The X-37B launches inside a fairing. Both of these solutions would have substantial side effects on the intended Starship design if incorporated.

It would theoretically be possible to use a very fast acting control system to manage the dynamic instability, but this starts to look very similar to the problems being solved to do a powered landing in any case.

  • 1
    $\begingroup$ Important to note "large wings" were there to meet a military specification not needed for a commercial shuttle, the ability to glide a large distance from the plane of its orbit. A "modern" configuration, without this requirement (and avoiding a large cargo bay within the shuttle), would need only slightly more "wing" than the Starship already has. The old Shuttle orbiters do have the ability to return a large payload from orbit. $\endgroup$ Commented Feb 12, 2021 at 12:10
  • $\begingroup$ More examples: the Dreamchaser spacecraft is to launch in a fairing similarly to the X-37B. The aerodynamic issues are a major problem for making a version for carrying humans, as aborting is problematic with a fairing enclosing the vehicle. The old DynaSoar would have flown without a fairing, but required adding huge fins to the Titan it launched on. $\endgroup$ Commented Feb 13, 2021 at 4:33
  • $\begingroup$ This is why the Air Force encloses their shuttle in a shroud (which also protects it during ascent). Again, the (more primitive) DynaSoar was designed for military use. The X-15 showed smaller wings were viable and safe if the landing area size is sufficiently large. Modern computer and GPS precision makes "one shot" glides quite safe, provided the weather cooperates. $\endgroup$ Commented Feb 14, 2021 at 13:10

Wings are heavy. Surprisingly so.

As Jorg pointed out there is no air on the Moon, and Mars's atmosphere is pretty thin.

In fact, Starships payload to the Moon is surprisingly low, because it has to carry all the fuel to land entirely propulsively.

Cheating by using air resistance is important.

  • 2
    $\begingroup$ They plan to have a moon variant with no wings or ability to land back on earth. $\endgroup$
    – loweryjk
    Commented Feb 11, 2021 at 22:31
  • 4
    $\begingroup$ @loweryjk Yes But it does not have wings. It has flaps. :) And you mean also no heat shield. Agreed. $\endgroup$
    – geoffc
    Commented Feb 12, 2021 at 0:12

On Earth, you need only build a suitable runway. Have a look at the Fly Back Booster concept from the 1990s. Gliding can lower rate of descent even better than parachuting, and at high landing speeds not a lot of wing is needed.

One can only imagine a time traveler going back to the 1970s and meeting von Braun.

We might have some idea what the NASA SLS would look like today.

But one may also expect SpaceX to forge ahead with their groundbreaking work, and, in an age of advanced computers reacting in milliseconds, the "HoverSlam" may yet reach an acceptable safety level for passenger transport.

  • 2
    $\begingroup$ The last paragraph of this answer touches on an important point I don't see in the other answers: plenty of crazy, dangerous tech has been made safe over the years. E.g., li-ion batteries are one little puncture or overheat away from being an inferno, but most folks keep one in their pocket $\endgroup$
    – BThompson
    Commented Feb 12, 2021 at 15:19
  • 2
    $\begingroup$ @BThompson: Indeed. Elon Musk has said he wants hundreds of successful flights without crew before putting people on it. That is orders of magnitude more testing not just than any other spacecraft in history but even airplanes and cars. The very first time ever that a Space Shuttle Orbiter was mated to an External Tank and SRBs and launched was the first operational crewed flight. It didn't have a single test or demonstration flight of the whole stack, only some glider tests of the Orbiter. $\endgroup$ Commented Feb 13, 2021 at 14:29
  • $\begingroup$ If they get to those hundreds of flights, then this will be settled. But I'm thinking there is a good possibility of never reaching that goal. $\endgroup$
    – loweryjk
    Commented Feb 16, 2021 at 19:58

A better question to ask might be: why do planes not land vertically, and instead require a long runway? The answer is that the engine (or more precisely the propeller or fan stage it drives) lacks sufficient thrust to overcome gravity, and therefore the plane relies on air passing at speed under its (large) wings to maintain lift.

Rockets have enormous thrust, so lack of engine thrust absolutely isn't a problem. The nine engines of Falcon 9 have sufficient thrust to lift the rocket with a full load of propellant off the launch pad and accelerate it up to thousands of km per second. When it lands, it uses only a single engine, and even when this is throttled right down, it produces so much thrust that the booster is unable to hover. Instead the engine must be controlled carefully to ensure it decelerates to 0 m/s just as it reaches the ground - and then the engine is shut down at just the right moment. I understand the engines on Starship will be able to throttle down low enough to be able to hover (indeed this was demonstrated already with the initial Starhopper prototype.)

Using the rocket equation we can work out the weight penalty for a powered landing. Rearranging the formula as shown below we get the following

$dv$ = required change in velocity (assume "wing" flaps slow it to 50m/s fall speed, 180km/h, 113mph)

$v_e$ = exhaust velocity (3200m/s according to Raptor article on Wikipedia)

$\frac{\text{m}_o}{\text{m}_f}$ = mass ratio before and after burn

$\frac{\text{m}_o}{\text{m}_f} = \text{e}^{dv/v_e} = \text{e}^{50/3200} = 1.016$

That means that if the flaps slow Starship to a fall rate of 50 m/s, the weight penalty for landing propellant to reduce velocity from 50 m/s to 0 m/s is 1.6% of the total dry craft weight.

Making the flaps big enough to act as proper wings to enable a safe landing would be a massively higher weight penalty.

It should be remembered that the header tanks are larger than required just for landing, as they will also contain propellant for re-entry burns, which require higher $dv$ (change in velocity.)


This is not a happy place to be, and it's a testament to the insane skills of the pilots (and a healthy dose of luck!) that none of them were lost on landing.

The fact there were no incidents on landing the Shuttle is a tribute to the genius of the design engineers and the skill and training of the pilots.

Luck had nothing to do with it.


But for a manned vehicle, it's far too risky. If those engines don't light in time, you and your 100+ fellow passengers are dead.

There are plenty of ways to kill 100+ passengers dead in a horizontal, airplane-style landing as well. The right answer is to make the time and effort to make the system safe and reliable, which will happen after they dial in the final design. The development flights happening in Boca Chica right now are just figuring out how to make the system work - once they've figured that much out, they can spend time refining and bulletproofing.

We've done the Frankenrocket with the Shuttle, and we've learned some painful lessons from it. We don't need to learn them again.


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.