The First High-Altitude Starship Test

…coming this week.

Regardless of the outcome, it will be exciting.

[Wednesday-afternoon update]

It could be as soon as Friday.


[Tuesday-morning update]

Today seems to be the day. It’s being covered by NASA Spaceflight.

[Bumped again]

[1440 PST update]

Raptor abort. They aren’t going partway to space today.

[Wednesday-afternoon update]

Fuel is loading, T-26 minutes to launch.


[Post-flight update]

Well, that was kind of spectacular. As I said, guaranteed to be exciting. The exhaust looked a little weird toward the end. It was greenish, as though it was maybe running fuel rich. If so, Perhaps the thrust wasn’t up to spec, and insufficient to control the landing on final.

[Update a while later]

Elon said that the header tank pressure was low on landing. I’m guessing that this maybe resulted in bubbles in the cooling channels, overheating of the nozzle, and injecting copper into the flow, for that green effect. And reduced thrust, of course, which is why they came in too fast.

[Update Thursday afternoon]

Am I the only person who thinks it would be pretty unpleasant to be in the nose of that thing at final (even with a successful landing)?

99 thoughts on “The First High-Altitude Starship Test”

  1. My guess is that Starship SN8 won’t survive this test (I hope I’m wrong). There’s so much of the mission profile that’s a “first”, that has never been tested; re-ignition of the Raptors on demand, the dynamics of the “skydiving” maneuver, the actual effect of the flaps, etc.

    There’s also the software; this profile is vastly different from what the F9 does, so, it’s new ground there, too, in several ways.

    However, if SN8 completes the “skydiving” part of the profile, that alone is a huge step forward.

    This rapid iteration/embracing failure style reminds me of what I’ve read of the WWII era and shortly after (except they risked test pilots, then).

    1. Reignition of the Raptors shouldn’t be a problem as that happens internally, in the pre-burners. The Raptor being a staged-combustion engine, there is no need for a separate ignition process for liquid propellants entering the main combustion chamber – a job performed by slugs of TEA-TEB in the Falcon’s Merlins. That’s because what enters the main combustion chamber of a Raptor are two streams of already partly-combusted hot gas – oxygen-rich and fuel-rich, respectively – which are mutually hypergolic.

      This is an inherent feature of staged-combustion rocket engines and one which – in addition to their inherently superior Isp – especially recommends them for use on places like the Moon and Mars where extreme reliability is paramount along with no dependence on chemicals that can’t be locally sourced.

      The dynamics of the skydiving maneuver are key, to be sure, but the dynamics toward the end of said maneuver are far trickier than those of its earlier phases. Once SN8 is laid over on its side at the top of its trajectory, all it has to do for some time is maintain, more or less, that attitude. This allows time for a bit of modest “wiggling” of the aerosurfaces to be done with results to attitude detected by angle, rate and altitude sensors aboard and comparisons made to pre-computed behavioral expectations. This allows needed adjustments to be made in real time to the more critical aerosurface position change commands needed as the “belly flop” transitions to the “swoop” needed to establish a final more or less vertical attitude just before landing.

      Nobody does better or smarter avionics than SpaceX and SN8 will be carrying its best work to-date.

      That’s not to say the odds of total success are actually close to 100% rather than the 33% Elon has quoted. But Elon has a long history of public declarations that prove, in general, to be too pessimistic in the short term.

        1. The Raptor engines have duel-redundant electrical igniters (a.k.a. spark plugs) to light the pre-burners. I don’t know how they’re emergized. I’m guessing batteries.

          1. There’s a Tesla Model S battery pack mounted atop the upper tank dome on each Starship prototype. Model S motors drive the aerosurface actuators. I’m sure there are a few watts left over to run the igniters.

          2. I would use the batteries to spin an electric motor that runs a magneto, probably with gap type points instead of carbon brushes, since brushes wear and would be hard to fabricate in remote landing sites. Designing a maintainable engine depends on knowing what Johnny Jetpack can understand and work on, and Amazon isn’t going to ship in a fancy circuit board because, well, Bezos.

          3. It’s even simpler than that. Short an inductor to ground then open the switch, the current has to go somewhere; if you provide a spark gap with a lower corona potential than any other path in the circuit then you’ll get a spark. The switch can be controlled by a simple Schmitt trigger to be a bistable oscillator. It’s about $2 in non-moving parts and a small PCB.

      1. @Dick Eagleson;

        The raptors having another abort today concerns me. It was an auto abort from one or more of the Raptors, so, my guess is one or more of them was out of spec during preburner ignition.

        I know (or at least, most fervently hope) that during flight the specs are ignored for startup, but what’s worrying me here is that Raptor is both new and very advanced. Such things often have teething problems. Raptor seems to be no exception (such as the Raptor running hardware-rich during Starhopper’s 150M hop).

        My current wild guess is that Elon’s 33% is on the conservative side (like his FH estimate was). I think he learned a lesson on that from the first Falcon 1 launch (he predicted 95% chance of success on that one).

        I very much agree with you in the main, and thank you for your very knowlegable post (It reassured be quite a bit). Fingers crossed they’ll be able to launch it tomorrow.

        1. Thanks. That was quite spectacular. Ascent good. Engine shutdowns good. Tip-over at apogee good. “Skydive” good. Engine reignitions good. “Swoop” good. Didn’t quite stick the landing. As Maxwell Smart used to say, “Missed it by that much!”

          1. I’m inclined to rate the flight as a 95% success, especially as what I thought would be the hard part (modeling the skydive) looks like it worked perfectly. That bodes well for reentry.

            Also, it looks like I was wrong about the Raptors being a likely source of trouble; they appear to have worked perfectly. (well, aside from them not working too well when not getting fuel, hardly an uncommon flaw with engines of any sort). The Raptor performance alone would rate this test a big success to me.

            And, that flight was even more spectacular than I’d thought. 🙂

          2. I quite agree. Truly an amazing performance.

            The Raptors are incredibly well-engineered as is the entire ship. Like football players and warships, rocket engines show their true grit when playing hurt. Even when getting inadequate propellant in radically sub-optimal proportions, the Raptors still managed to function in degraded fashion rather than blowing up.

            More than a bit of this demonstrated grit is also owed to the Starship’s superlative avionics as well. SN8 seemed almost to be a living creature during its mesmerizing flight, especially in the last few seconds when that Raptor sole survivor refused to give up trying to save the situation even as it was consuming its own guts in the effort. SN8 had soul and went down swinging. Taps should be played for her.

          3. Elon Musk has agreed with a suggestion to save SN8’s nose as a memorial to SN8. I definitely agree.

            I very much agree that the Raptors not doing a RUD was impressive as heck.

            The avionics… indeed, that they did so incredibly well while dealing with areas of the flight envelope with zero real-world experience was stunning, and something I did not expect. IMHO, this bodes well for reentry.

            I also can’t help but to be impressed by the massive improvement in build quality since MK1; they appear to have pretty well slayed the dent/dimple issue that would have certainly made applying a heat shield, um, challenging.

  2. “Regardless of the outcome, it will be exciting.”

    Well, it would seems it could indicate how starship flies, and after that maybe we can then see how starship re-enters the atmosphere. Which would indicate whether starship could go to Mars.

    If tax payers which paid a billion dollar for X-33, had been able to see it go to 50,000 feet. it might been worth the money wasted on it.

  3. Other than an unpredictable engine RUD, the main waypoints are the drag brakes staying on through the brief transsonic phase, transitioning to horizontal, and then going tail town. Landing they’ve already tested. The interesting thing, iirc, is the transition back to powered flight requires the engines to reignite so thrust vectoring can turn the ship tail down (that is per a Musk tweet). That’ll be interesting to see. Here’s hoping (or hopping).

    1. Popping a parachute from the nose might also work to initiate a flip, but would create another critical failure point. If the engine won’t ignite for the flip, a nose parachute would just delay smacking into the ground by a fraction of a second.

      I could see using a thruster to initiate the flip, but since they don’t want to be coming down directly at the pad (or drone ship) prior to ignition, they need horizontal translation for the touchdown anyway, so the main engine is doing double duty during the maneuver.

      1. My main concern was ullage, since the rocket is sideways. If the downcomers from the header tanks are full and pressurized, that would do it. My original expectation was, the “elonerons” would execute the pitch up manuever. since they have 3-axis control over the vehicle, but I guess not. And I don’t know how much control authority the cold gas thrusters have. I guess you’d have to have a pretty good nose chute to pop it after a 9 mile free fall!

        1. I’m a big advocate of keeping rockets horizontal, to make their handling and maintenance simpler and cheaper.

          In that vein, instead of a flip-turn and tail landing, I’d leave it horizontal all the way to touchdown.

          As a quick though experiment, I’ll use the existing dual Super Draco clusters as currently used as abort motors for Crew Dragon. Given a 90 ton dry weight for Starship, I’ll set it up for 2.5 G deceleration.

          I’ll mount the Super Dracos with the existing dual-pod arrangement, mounted slightly above the centerline of the rocket (on the non-tiled topside) to avoid direct re-entry heating), and angle them out at 30 degrees. This gives each motor a 13,850 lb vertical thrust component, or 27,700 lbs per cluster, and since there will be a port and starboard cluster. That comes to 55,400 lbs vertical per quad-pack installation.

          It would take 8 such installations (totaling 32 motors) spaced out from nose to tail, to give it 2.5 G’s of decel. The vertical thrust component is 443,400 lbsf, almost as much as one Raptor.

          For the moon, hovering a completely empty Starship would need only 8 motors in total, so four positions along the length of it, each firing only one of the dual motors, and each motor throttled down to 27%. Thankfully the Super Draco throttles down to 20%.

          Of course nobody would ever land an empty Starship on the moon because there wouldn’t be any point to it, but it shows that the method would allow both Earth landing and lunar landings, and of course takeoffs. At full throttle, the Super Dracos would lift a 665 ton Starship off the lunar surface with the same acceleration as the Apollo ascent module, which is 575 tons of fuel and payload.

          Another advantage is the the motors being on the sides and angled out 30 degrees shouldn’t produce cause any debris kicked up to slam into the rocket itself, and assuming the landing legs are, say, four meters long, the landing motors should be about 10 meters above the ground, with about 12 meters (40 feet) from the motor to the center of where its exhaust hits. That should put a lot less wear on the landing pad, or whatever surface the ship lands on.

  4. Regardless of the outcome, it will be exciting.
    In general that has been my experience with rockets both scale model and otherwise.

  5. I’d have built a dynamically similar scale model and dropped it from a helicopter to test the Elonerons. Has this been done?
    OTOH DC-X successfully completed the “swoop of death” first try. I think computing and simulation has gone a long way since 1995.

  6. All the discussions of belly landings and SuperDracos, etc., miss the point that all the extra stiffening and tanks for non-methalox fuel (SuperDracos use hypergolics) ignors the fact that all that comes out of payload. Pretty soon you’ve got a Starship that can’t make orbit, much less make orbit with a hundred tons of cargo. Well, we could beef it up with an external fuel tank and solid-rocket boosters… oh. Wait a minute.

  7. I’m in a hotel in South Padre Island with a great view of the launch pad. Not going to miss this one if it goes this week. We got here Thursday for possible Friday launch, delayed to Monday and now looking for something on Tuesday. Had a very nice walk along the beach today.

  8. Rand, shows a watch link for “Starship | SN8 | High-Altitude Flight Test”, scheduled “Live in 12 hours – December 8, 6:00 AM”. This as of about 5:45 PM today (Dec. 7).


  9. Looks now like SpaceX’s coverage will begin at 8:00 AM PST. (Their web page is counting down.)

  10. A NASA WB-57 is heading to Boca to try and film the launch, should it happen today. Whenever the launch takes place, the results should be spectacular.

    1. I was trying to figure out what that plane was, never seen one quite like it before. It was fairly easy to make out in 8×40 binoculars. Only saw the first pass as it was creating a contrail for a bit.

  11. I’m seeing 50K folks on the Nerdle Cam.
    20K on one of Lab Padre’s other cams.
    48K at Everyday Astronaut.
    106K at NASA Spaceflight.
    and maybe 40K at other assorted live streams.

    About 260K-270K watching, at probably an hour prior, or maybe 110K watching with multiple tabs open.

    1. I think Everyday Astronaut said he hit 150k viewers, an all time high for him.

      I had 4 streams going and other people probably did too.

  12. The Spacex youtube just says that liftoff time is TBD. Been saying that all day.

    It’s now 5:06pm Eastern.

    Any info on what’s going on?


  13. Raptor abort. Standing down for the day.

    I’m inclined to think the engine start reliability may not be quite ready for landings.

    1. I suspect the start sequence will have different operating envelopes for in-flight versus pre-flight. At this point it should be all about the software, the Raptor engines must have hundreds of hours on the test stand by now.

      Sure hope it flies tomorrow!

  14. Embedded ignition controllers have corrupted boot partition and couldn’t boot up. Oh wait, sorry, wrong thread….

  15. Starship is much heavier and fatter than the Falcon first stage. What kind of sonic boom are they expecting during landing?

    1. Late answer, but I think Starship will go subsonic much higher and earlier than Falcon; terminal velocity in bellyflop mode seemed to be only around 100m/s!

      But I don’t think SN8 ever got supersonic.

  16. Well, damn, they almost made it. The 10 o’ clock engine ate itself (green flame) during the landing and the 2 o’clock engine shut down, so it landed hard. They nailed most of the flight profile.

    1. The fault, dear Elon, is not in our Starships but in our Raptors.

      (I have been waiting years to deploy that one.)

  17. Stuck everything but the landing — looked from the SpaceX feed as though only 2 Raptors reignited.


      1. Being light on propellant, they probably didn’t intend to use all three engines for the landing, just as the Falcon 9 often only uses one engine for landing. Lighting all three engines on SN8 would have required them to throttle down a great deal or to pull of a big hover slam on the first attempt.

  18. Well, that was utterly spectacular.

    Take a close look at the single engine burning prior to touchdown. The exhaust looked definitely green.

      1. Elon reports that the propellant flow from the header tank was too low. One engine shut down and the other went to higher power with diminished propellant flow (probably methane), causing the engine to go oxygen and metal rich. The engine was sacrificing itself in an attempt to pull off the landing. It would out like a hero.

  19. I would count this as a success. The belly flop procedure seemed to go well. SN9 is pretty much ready to go once they analyze the data from this attempt. Maybe another launch this month?

    1. The cynic in me wonders if the engine shutdown right before landing was an intentional test because it adds drama and the audience likes a boom more than a boring landing.

  20. Pretty cool. One step closer to making it irrelevant to who controls government and from a much healthier perspective than Alphabet facing the same government irrelevancy.

  21. A tweet from Elon Musk about the cause of the slightly rough landing;

    “Fuel header tank pressure was low during landing burn, causing touchdown velocity to be high & RUD, but we got all the data we needed! Congrats SpaceX team hell yeah!!”

    My take; the green seen from the Raptors is due to poor fuel flow (and maybe gas bubbles) causing the Raptors to run a bit hardware-rich. My guess; the header tank issue might be due to the dynamics of the flip. They pulled a few Gs with that.

    My further guess (based in part on Elon’s enthusiasm) is that they learned what they needed, and thus the test is a massive success.

    I’ll even go way out on a limb and predict that SN8 will not fly again.

  22. Re-watching the landing attempt from the aft facing camera it looks like it landed astraddle the edge of the landing pad, partly on and partly off. Not that it really made a difference in the end result.

  23. Some questions:

    1. On the ascent it seemed that first one then a second engine shut down. Was this planned?

    2) Just prior to the shutdown all the nozzles seemed to gyrate wildly. Was this planned? Was this to re-orient the thrust vector just prior to an engine shutdown?

    3) I take it that once the fuel is burned, only one engine is required for landing?


    1. 1. I think so. After the first shut down, it looked like some wiring in the engine compartment burned to a crisp. I doubt that was planned.

      2. Yeah, I saw that, too. I don’t know if it was planned. Given the flight continued under control, even if it wasn’t planned, the controller compensated properly.

      3. Two engines lit for the landing but one shut down prematurely due to low propellant flow from the header tank. The one remaining engine gave it all it had but it wasn’t enough.

      1. 1) The have to shut down engines sequentially for that flight profile. When it gets to the top it’s almost an empty vehicle, with a dry weight that should come in a bit over 100 tonnes, or 220,000 lbs. For hovering, one Raptor at less than 50% throttle is all that’s required. Run three at 50% and you’d be pulling 3 G’s of acceleration, which is not what they wanted to do.

        2) Whenever an engine is about to shut down, you’ll see quite a bit of motion because the rocket is shifting so that the new thrust vector is aligned with the rockets center of mass. With three engines running, the thrust vector is dead center of the engine bay. When it shifts to one engine, the whole rocket has to be tilted, as we saw during the earlier single engine hops. The engines that shut down need to move out of the way of the engine(s) that are still running, so they don’t impede the maximum travel of the running engine.

        3) Starship probably can’t touch down with no payload and two engines lit, unless it goes for a precision slam landing like Falcon. But it does need two engines for that final pivot and breaking maneuver, where it needs to slow down quickly prior to touchdown.

      2. I don’t think any wiring burned, looked like flammable gases to me. There was some plastic sheeting inside the bay that wasn’t even melted/burned by that brief flash of flame.

        1. That’s what it was. The engine bay of a rocket is always a low pressure area in flight. When a Raptor shuts down, there’s a bit of left-over methane-rich gas from the nozzle that gets sucked back up and which flares when it mixes with atmospheric oxygen.

  24. I suspect 80% of NASA watched that, laughed at the explosion, and then went back to making fun of the SpaceX flightsuits while moving the SLS first launch data a month or two to the right.

    1. As a former NASA contractor I’d have to say that most NASA employees were probably indifferent (NASA is just a job), but a large fraction were enthusiasticly cheering SpaceX on. At long last something exciting is happening that is massively moving human spaceflight forward!

  25. I saw a video later from a mile or two away, and I was amazed how close to the ground SN8 got before going vertical again. l’m guessing 3 to 5 thousand feet.

    That would be a white knuckle landing. I wonder if coming in from orbit it would go to vertical higher up after shedding velocity, and be more like a typical Falcon landing.

    1. I don’t think so. I think what we saw yesterday is going to be pretty much SOP – a “last mile” landing evolution that completes pretty close to the ground. Starships will land fed by their header tanks. Those aren’t very big so one can’t dawdle about landing.

      1. Plus, the lower it falls, the slower it goes due to lower terminal velocity in thicker air. That combines with gravity losses from a powered descent, so the lower it pivots, the less landing fuel it needs.

        However, I don’t like the location of the LOX header tank. It’s probably 130 feet or so above the engines, with a long propellant line running between. At the start of the flip, gravity hasn’t added any extra head pressure at the engine. All is calm in the downcomer pipe.

        So, a quick bit of math based on some assumptions:

        1) The top of the LOX in a full header tank is 47 meters from the engine system’s pressure regulators (pogo dampers, etc). (based on eyeballing a basic Starship diagram)

        2) The pre-flip weight of a Starship is 114.72 tonnes (fairly close, given that their individual dry weights are still not finalized, and that number makes the subsequent math really easy).

        3) The full thrust of a SL Raptor is 505,000 lbs (again, very close, and chosen to make the math easy).

        4) The LOX density at landing temperatures is 1.141 kg/m^3, which is that standard value, but not for superchilled LOX.

        So, the output of all that is:

        a) Given the height of the LOX column, the pressure increase is 76 psi per G of acceleration. (Basic calculation of head pressure in a fluid column).

        b) Given the Starship pre-flip mass (assumptions above) and Raptor thrust, the number of G’s with two engines running is 4.00 * throttle setting. (4 G’s with two engines at 100% throttle. 2 G’s with two engines at 50% throttle).

        c) So the LOX pressure at the engine regulator inlet, with two engines running, is header tank gas pressure + 4 G’s * throttle setting * 76 psi / G.

        That’s LOX header tank gas pressure + 304 psi * percent_throttle.

        Running through the same quick analysis (based on eyeballing things), the CH4 header tank, with 31% the height to the top of the methane, and with methane only being about 37% the density of LOX, produces an engine system pressure that should be about:

        CH4 header tank gas pressure + 34.7 psi * percent throttle.

        So the LOX pressure is about 8.8 times more sensitive to acceleration affects than the CH4 pressure.

        That isn’t necessarily a problem, it’s just extra engineering work that has to be applied to solve a pressure variation caused by the remote LOX header tank location. I’d move the LOX header tank to somewhere near the bottom of the LOX tank, so that box LOX and CH4 pressure at the base of the rocket would increase by about the same amount under acceleration.

        What I haven’t addressed is momentum and water-hammer effects in a 47 meter pipe full of dense LOX.

        1. They tried that and then found out they needed to move it forward to help trim the ship. I would bet on some non-fundamental failure in this instance. You may remember we talked long ago about mounting a Dragon in the nose for early-days LES. That’d trim it nicely too.

  26. I think any humans onboard will need their barf bags (and Depends) strapped on during that final pull-up-to-vertical maneuver…

    1. It’s not worse than a decent roller coaster. A few people do ralph on roller coasters, but I never have, and I get vertigo. My Dad rode on a 3g coaster when he was 80 and a hundred pounds overweight. I think most people headed for space will have some training and will do fine. I also don’t have much faith in the point to point idea.

    2. I did a not so quick moment of inertia calculation to check into that.

      Based on the cross sectional shape of Starship (popping a diagram into Paint and then feeding the points into polygon centroid and area calculators), the center of area is about 21.5 meters up from the base, about a foot below the center of the main CH4 tank. Since it was stable in free fall, that should also be the center of mass.

      After a bunch of calculations of the moment of inertia of the fuel and engines, and likely moment of inertia without those, I figure that the during the skid-turn, the rocket will feel like it’s both accelerating in the usual direction, and pivoting about a point located about 14 feet above the top of the LCH4 tank.

      If you seat was lower down than that point, you would feel some blood rushing to your head, not quite like you were pulling negative G’s, but more like having the head of your bed lower than the foot of your bed.

      If you were seated higher in the rocket, more toward the nose, you would feel a bit of force pushing you down toward your feet. The closer your are to that pivot point (the center of percussion of the rocket, relative to the location of the main engines), the less of this force you would feel, and the more it would seem like your chair is just being tilted forward or back, with no unusual forces on your straps, with no significant G’s pushing blood to your head or to your feet.

  27. There’s a few seconds of rapid rotation, some extra acceleration and then you are stable on the ground. I doubt there’s enough time to get sick and barf. The stable on the ground after a few seconds is key. It is like the old sailor’s 100% cure for sea sickness – sit in a old Church yard, under the shade of a tree, in the country.

Comments are closed.