24 thoughts on “SpaceX’s Latest “Failure””

1. Well if this is how you measure success, by confusing progress with spectacle, I guess it would a “spectacular success” if SpaceX dropped a flaming F9 stage onto the fireworks barge in the Charles River for the upcoming Boston Hatch Shell 4th of July concert. Seriously though, I prefer not to draw too many conclusions where we are currently in the “recoverable stage” story. SpaceX needs more time. I am concerned that the difficulty of what SpaceX is attempting will cause people to lose patience and if after (only) 10 failures, just as easily draw the wrong conclusions. Edison claimed the path to the electric light bulb was not paved by failures, but by lessons of what didn’t work. I’m not convinced SpaceX is as close to success on vehicle recovery as they’d like us to think. But that doesn’t mean they should give up. Personally, I’d prefer a more go-slow approach with better controls (ala F9Dev2/Grasshopper). To get incremental progress. It’s hard to gage if we’re getting good experimental results that lend towards practical systemic solutions with the current launch and crash approach. We’ll see. Probably will take a few (3-5) more failures before it’s necessary to take the “systemic approach” step back rather than tweak and tweak. The PR problem for SpaceX is the very hype these attempts generate. Where otherwise spectacularly successful flights are still judged failures. I am forever grateful that SpaceX hasn’t cloaked these recovery attempts in total press blackout. But the PR guy in me says conducting these tests in the open is a mistake.

   1. They have nothing to lose from their “crash and burn” approach. The vehicles will either be dumped in the ocean, or used for flight test.

      1. Only usable for flight test once a recovery has been successful. Here’s where we disagree a bit. There’s a tendency to go with what I’d call “flight hubris”. Since actual flying F9 hardware has to be able to do this. However, we are still dealing in the area of engineering flight test. In that domain, we might be better off with a special-purpose instrumented craft that can yield better data on the flight regime during recovery. Better data hopefully yields better solutions. The problem with trying to do the same on flight hardware is that there is a tendency to go with flight hardware bias. Systemic solutions that might require redesign and higher expense are back-seated with the prevailing belief is that the flight hardware can achieve the goal with a tweak here, then a tweak there. Pretty soon the tweaks begin to add their own variants into the equation, etc. I’m sure you get my drift…. You may think you are getting free testing, but what you may be getting instead are non-free dead ends.

         1. Sorry I should clarify my first sentence. I thought you were referring to the plan to re-test recovered boosters in NM. Yes, as stands I agree with your original statement. My qualm is with data quality and the hubris behind testing flight hardware that may lead to false conclusions and remedies.

         2. There are issues involved with trying to land on the barge that would be difficult to replicate on the ground. For one thing, the barge may not be 100% stationary due to wind and waves. Regarding the wind, from the looks of the flag from the still image released yesterday, it looks like the wind speed was in excess of 15 MPH. I doubt any of their previous land-based tests were performed in winds that high.

            It appears to me that they’re going to have to tweek their control laws at the transition point where the airspeed is too low to use the grid fins and they have to use the nitrogen thrusters. They can do this on land, but it costs them very little to do these tests on operational missions. The biggest cost is the cost of operating the barge and towing it to the landing location. At the moment of separation, the first stage has fulfilled its contractual purpose and essentially has no value. On any other rocket (or a Falcon 9 mission heading for GTO), the first stage is thrown away. It costs very little to attempt a landing compared to simply letting the stage drop into the ocean.

            SpaceX has proven the ability to learn from problems and make corrections. Other than their helium pressurization system, they rarely seem to have the same problem twice. I suspect it won’t be long before they’re able to stick the landing – something no one has ever done before. After that, they’ll have to examine and test the recovered stage to learn how much it’ll cost to refurbish it for another launch. At this point, no one (including SpaceX) knows whether it’ll be economically viable to recover, refurbish, and relaunch the stages. I suspect it will be but no one knows for sure.

2. Here’s a better video than the vine

   1. Oops, you may want to delete the link Rand, apparently it is unshared(?) Maybe SpaceX doesn’t want people see this in its entirety yet for some reason. I thought I saw it on Jurvetson’s site earlier this morning, but the link is gone there. I grabbed it from my youtube history.

      1. Ya. Oops. Video is private now. Sorry SpaceX.

   2. OK, I’ve updated the link.

   3. Watching this vid again, I think the pendulum action critically damaged the left most landing leg (by swinging it into the pad under the full weight) which is why there was the slow tip-over in that direction after the engine is cut. If the leg was still functioning it wouldn’t have fallen in that direction.

      1. I was wondering about that, too. With the weight of the engines so low down, I can’t see why it would have toppled if that leg was still intact. Is the centre of gravity really that high?

         Hopefully the video from the barge will show the problem, one way or another.

3. It’s a failure in so far as they didn’t get the stage 1 rocket back in toll. They still got paid by NASA for the groceries run, and that’s pretty much what NASA cares about. Sure NASA would be interested if this pans out and lower the cost of launch, but it’s still competitive.

4. That poor little RCS thruster, just working its heart out trying to keep the stack upright…

   I’m glad to see this high-quality, longer version of the Vine video, for sure. It answers a lot of my questions about gimbaling, RCS thrusters, and others’ questions about the ultimate fate of the stack (LOX\Kerosene, and all that, after all). I also noticed that the RCS thrusters don’t fire at all until the stage is almost on the deck. The waggling tail of the flame from the engine seemed to indicate an inadequacy on that method to completely control the yaw, and perhaps an RCS thrust or two when the stage was 1-2 lengths from the top of the deck would have helped, especially when the Merlin over-compensated.

   Are the RCS thrusters reliable enough to use as an “anti-skid” mechanism at that altitude, or is that what the fins are intended to do?

   Is it possible that there’s a speed at which the waffle-fins are no longer effective but the RCS thrusters are not *yet* effective, leaving a rather risky gap that’s fully dependent on the Merlin and gimbal to compensate? If so, I would think that narrowing THAT gap would be the quickest route to gain increased control over the attitude of the stage.

   It also looked like there was a point at which the stack was completely vertical and could have potentially dropped nicely on to the barge before it went too far over the other direction, but I don’t know the height at that point and whether or not the legs and RCS thrusters would have kept it upright if “dropped” like that.

5. To Zimmerman’s point, it’s amazing what a little real competition can do, isn’t it?

6. Video available at https://twitter.com/elonmusk
   Freaky. That ship just looks too frisky – I’m guessing that’s because you can’t throttle down. It’s like trying to parallel park coming backwards off a freeway at 70 mph, when all you can do is slam full on the brake at _just_ the right time. Good thing we have computers. But I’d imagine that it would be real helpful to be able to throttle down.

7. My take;

   Like so many aviation crashes, this one seems to be a cascade of problems, not just one issue. In other words, multiple causes.

   The f9’s sudden tipover during flight seems to have introduced a rate that persisted to touchdown. That, IMHO, is cause #1.

   The sea state was problematic. Per the vid, we see the drone ship pitching. More importantly, we also see whitecaps. The latter gives is a gauge of the windspeed; looks about force 5 to me, which is 17 to 21 knots. It could be a bit more, but I highly doubt it is less.

   I’ve seen a “zoomed in” edit of this video. In it, you can see the motion of the drone ship. It’s pitching a fair bit. Not enough to affect a perfect landing, but a bit. The key, IMHO, is how the pitching rate interplays with the stage’s rate at touchdown. The ship is not merely rolling (changing angle on deck) but also, per the wave motion, osculating side to side a bit (from our frame of reference). IMHO, the touchdown occurred at the worst possible moment vis-a-vis the drone ship’s motion. That’s cause #2.

   Cause #3, wind direction. It’s from the same general direction that the F9 tipped away from. The F9 teetered for several long moments, top thrusters firing, before succumbing and toppling. Had the wind been less, or from any other direction, IMHO the stage would not have gone all the way over.

   So, IMHO, we had three factors. The lack of any one of them, IMHO, would have resulting in a safe landing, because it took the interplay of all three to cause the topple.

8. They’re making progress- the failures are more complicated and show advancement in the state of their art. They’ll make it work, though it might be a bit painful along the way-

   “When I first came here, this was all swamp. Everyone said I was daft to build a castle on a swamp, but I built one all the same, just to show them. It sank into the swamp. So I built a second one. That sank into the swamp. So I built a third. That burned down, fell over, then sank into the swamp. But the fourth one stayed up. And that’s what you’re going to get, Lad, the strongest castle in all of England.”

9. I’m wondering if I might prefer -barbed- landing legs. Or barbed spikes extending below the weight bearing ‘feet’ of the stage. And cover the deck with tight chains, or something.

   That effectively hands some of the responsibility for immediate stability to the barge – which doesn’t care about mass ratios. A few winches could turn than into active participation and a “completely strapped down” stage.

   That is: Ignore all the difficulties with the -landing- part that this video demonstrates, and just think about the (future) difficulty of balancing a completely landed stick on the pitching deck for an indeterminate amount of time. There’s no one aboard to run out and use chains and binders to restrain the craft for the trip home. Is the nitrogen expected to be sufficient for -that-? Or active hydraulic balancing by shifting the deck or the legs themselves?

10. Response to the main engine TVC looked rather underdamped, which might indicate it was rate limiting, which could have many reasons, from hydro system problems to unmodeled dynamics to just too much external disturbances. Some odd high frequency motion out of the TVC prior to strut deploy too. If it were my GNC design, I’d be feeling a bit anxious about now, despite the learning opportunity…

    UNLESS either (a) I knew there was a HW failure that hasn’t been made public yet; or I knew the odds of success were poor anyway. I wonder what the Monte Carlo analysis showed for the wind and sea conditions that existed at the time?

    (I know it’s bad form, but I can’t refrain from the armchair speculation; it’s too interesting a problem in flight control, and the video shows some tantalizing dynamics to chew on…)

    1. +1 Yes. This is a flight dynamics problem par excellance. I’m in the same place as you. This *is* armchair quarterbacking, but the problem is just too interesting to ignore.

11. I could be wrong but it seems to me that if you watch the video in HD in full screen mode that:

    when the thing has landed and is clearly tipping to the left (as we view it)

    and you see the thruster on the left trying hard to keep it upright

    that at the 14 second mark, you see a thruster on the RIGHT fire.

    If that’s the case, I wouldn’t think that would help.

    1. Yeah that right thrust firing was just weird. Reminded me of one of the failed rocket launches in the 1960s.

    2. That thruster burst just before it fell looked to me more like a shutdown purge than an intentional thrust. It appeared that someone hit the shutdown switch (or it hit a programmatic shutdown) once that long plume of nitrogen started to shorten and give up trying.

12. I think they need to improve their software tests. A lot. They also need to do more realistic testing with Grasshopper in Texas than the highly controlled simple vertical up and down tests they’ve done so far.

    I’ve heard some people claim that they think the problem is the sensors and software don’t properly take into account that the thing is a long cylinder that you need to straighten during the touchdown. That you need sensors on top and on the bottom. It certainly looks like that. It kept wiggling a lot. Those grid fins are probably a partial cause. I don’t think its the fuel sloshing around as it is supposed to be nearly empty. You would expect the grid fins to kind of straighten out the top while the bottom would just follow gravity because the engines weight so much but that ain’t what happened. The bottom swings wildly like its a pendulum.

Comments are closed.