The Dragon Investigation

This is good news. Hans Koenigsmann just did a press conference with Kathy Lueders in which they announced the root cause of the explosion. Apparently it was a failed check valve prior to the test as they were pressurizing, that resulted in some NTO setting fire to titanium piping, causing an overpressure which then cascaded to mixing of the hypergolics. At least that’s my preliminary understanding. They’re going to go from check valves to burst disks, and the fix doesn’t seem to be on the critical path to getting to a November flight.

[Update a couple minutes later]

[Update a few minutes later]

Here is the full SpaceX statement. And “hyperbolic” typo fixed in initial post.

18 thoughts on “The Dragon Investigation”

  1. That was not quite clear to me. It says, loosely quoting, “a leaking component allowed NTO to flow into helium tubes during ground processing.” There’s nothing but ground processing prior to the explosion, so I take that to mean that sometime during test preparation, perhaps long prior, such as initial check outs or perhaps post-production fill and drain tests, NTO got into the high pressure helium tubes.

    Then, when they opened the helium valves (I assume to pressurize the tanks), the slug of liquid NTO slammed into the check valve, which is probably only designed for gaseous helium. The momentum broke something, ignited (with what?), and blew the valve apart. I assume the over-pressure from the small explosion in the very limited available space of a full NTO tank caused the tank to rupture, but it doesn’t explicitly say there was a tank rupture, just an explosion. It also doesn’t explicitly say that the hypergolics ever mixed. In fact, it doesn’t clearly say whether it happened in the fuel or oxidizer tank, or in the helium lines, or maybe somewhere between the NTO tank and the engines, just that it happened in a high-pressure NTO environment, which is probably the NTO tank.

    In any event, I’d prefer a switch to tougher check valve that can take a hit because in a hypergolic system, check valves are like the last line of defense against disaster. I’m not sure how they could provide absolute assurance that absolutely no propellant is, or ever will be, in a helium line. “Pinky swear?”

    I think we all prefer graceful failure modes, perhaps a backfiring “pop” or engine-out condition, over what happened.

    * Also, auto-correct got you or there’s a minor typo. “…cascaded to mixing of the hyperbolics.”

  2. Regarding the update, here’s a 1961 NASA report on the impact sensitivity of titanium ignition (both pure and 6AL4V) in an NTO environment. The combustion evolves nitrogen gas but does not self-propagate.

    1. Oh come on. You didn’t learn how to start a fire with dry tinder and a titanium check valve when you were in the Boy Scouts?

      1. Well, you also have to soak the titanium in N2O4 first. The NASA tests I linked only used N2O4 at one atmosphere of pressure. Perhaps high pressures make the problem even worse by causing deeper penetration.

        The other problem with titanium and high pressure NTO was stress cracking, which they eased on the Shuttle by mixing in a very small percentage of nitric acid.

        Anyway, I wonder if they’ll release some photos of the aftermath?

  3. Not clear to me how you would replace a check valve with a rupture disk.

    Possibly they’re talking about a relief valve. Metals exposed to strong oxidizers are protected by an oxide layer. A supersonic flow through a leaking valve could scour the oxide layer and cause ignition. You’re warned to be cautious of mechanical abrasion in such systems. A rupture disk, as long as it was of the proper material wouldn’t allow scouring.

    1. Presumably it’s intended to ensure that fuel can’t leak back into the helium lines when the helium pressure is cut off? Which was important when they planned propulsive landing and would need to turn the thrusters on and off during the flight.

      But now they only need them for an abort, so a burst disk is all that’s required: once the helium lines are pressurized, the thrusters are going to fire and keep firing until they’re out of fuel.

  4. So D2, like Amos-6, turns out to be a second instance of SpaceX being bitten by an exotic and recondite unintended/unanticipated combustion scenario. Looks like scrubbing all current and future designs for such possibilities needs to be a standard development engineering step going forward.

    It’s also noteworthy that all three of SpaceX’s major reverses have been a consequence of vehicles having helium pressurant systems. A co-starring role, in the case of D2, was played by the hypergolic system. I think it bodes well for the future reliability of SHS that neither Super Heavy nor Starship will use either helium or hypergolics.

    1. Helium always was such a Mickey Mouse gas.

      I think another route that would have avoided the D2 incident is if part of the check-out procedure was to slowly crack the helium valves to bleed the system and purge the lines by ramping the pressure. But that wouldn’t necessarily determine if there was a slow leak that would cause problems later unless it could detect any NTO in the lines, and a slow purge couldn’t be the in-flight operational method for starting an abort.

      The best way I can think of to detect NTO in a helium line is with an embedded capacitive sensor. Helium has a dielectric constant of slightly more than one, similar to vacuum or free air, whereas NTO has a dielectric constant of about three, close to mica or mylar. With a solid slug the capacitance would jump threefold, and even a small amount of gaseous NTO should easily register.

  5. Titanium can also do lower valences, but at Ti+4 has potential of -1.899, more negative than Al3+ at -1.662 and making it an enthusiastic fuel. But critically, Titanium is a lousy thermal conductor at around 20 W/m.K vs Aluminum alloys at around 160 to 180. Al is harder to ignite because it can carry the heat away, not so titanium. Breaking or grinding Ti can produce an impressive shower of sparks, and in high pressure liquid N2O4, it’s all over but the crying.

    In a system that would only be used once during abort, burst discs make more sense than check valves- guaranteed leakproof, simple to install, lighter weight, no soft goods to degrade over time…

    1. Another change that probably would’ve saved the check valve is if the helium line ended in a stubbed-off tee, with the check valve at right angles to what had become a gun barrel, so the slug would hit a solid cap instead.

      This failure is kind an odd one where even McGyver would have had trouble figuring out how to make a detonator out of the supplied components, but somehow they managed to do it accidentally.

  6. From what I could see here:
    (pg 66 of 129) Titanium and alloys were considered compatible with NTO, more so than SS. Sources of energy or high pressures probably not considered. I’ve seen chattering valves get hot enough to burn my hand.

    I wonder how the safety math works out with the escape system requiring big tanks of these fun materials right on the capsule. Maybe somebody should write a book?

  7. When you crack the oxide coating of a piece of titanium submerged in liquid oxygen it will ignite, so it’s not just NTO. I’d like to know the manufacture date of the valve, and also how cracked it got (did the metal crack, or just the oxydation layer?). It takes a few years for the coating on titanium to reach the maximum of 25nm, but it starts out at around one nm.

    Somewhere among my artifacts I have a discarded gas turbine vane made of titanium. It was binned because one of the corners somehow got bent under. I have no idea how it happened.

  8. The AIAA has a special report on NTO compatibility, but unfortunately it’s $96.95 for non-members.

    I was looking for information on NTO compatibility of noble metals and alloys. Many of those have great mechanical properties and shouldn’t have any oxidation problems, but many are also used to catalyze reactions, such as platinum, ruthenium, iridium, etc. Rhodium, for example, is most commonly found in catalytic converters where it breaks NOx compounds into nitrogen and oxygen. Hrm… What could possibly go wrong?

    In any event, making critical aerospace components out of ridiculously expensive noble metals is obviously the fastest way to throw money at the problem, and it would promote the recycling of capsules and ocean salvage of expendable stages.

    1. On the paper I linked above, gold and platinum are listed as probably compatible based on actual data of the materials..

      1. There are plenty of interesting platinum alloys used in industry.

        Mechanical properties of some platinum alloys.

        Many have good hardness and very tensile strength, and many are also used by jewelers. It should be quite easy to test a bunch of them.

        I would think you could still make the bulk of a part out of titanium, adding a thick noble alloy coating. On a part that might conceivably break in a particular failure mode, you could use the noble alloy to braze two titanium pieces together along the likely line of failure (building in a weak point), so that titanium wouldn’t be fracturing at all, the inert bonding alloy would. However that might be rather inconvenient to manufacture and test, and it might make more sense to design a part that isn’t going to break.

        Going with solid platinum, gold, or other noble alloys would probably be the simplest, and on a re-usable or returnable vehicle, the extra material costs would be 100% recoverable.

        Also, using precious metals in space vehicles gives the universe a great reason to have space pirates, and those justify military patrol ships, and the abuses of the patrol ships justify rebel warships, and then we have the makings of something interesting.

        1. As it happens, I know a little about testing materials. I wouldn’t describe anything that required working with NTO as simple. The concept is simple, finishing with all the people you started with alive and healthy and your building still standing uncharred is a little more complicated.

          I can imagine something that would let me introduce a small amount of something into a small amount of NTO inside some sort of pressure vessel. But first I have to be sure that the NTO won’t burn or embrittle or otherwise damage the vessel. Then there’s the joy of storing and handling NTO. Processing the exposed specimens, contaminated with NTO promises to be tricky. Whoever wants to know had better have deep pockets.

Comments are closed.