SpaceX Launch Failure
Keith’s note: The Falcon 9 Carrying CRS-7 either exploded or was destroyed shortly after launch a few minutes ago. SpaceX has assembled a team to look into this. NASA has referred to a launch vehicle “failure” and that the last data came down from the vehicle at T+ 2:19. No more news or NASA TV until contingency press conference no earlier than 12:00 pm 12:30 pm 12:50 pm 1:00 pm EDT.
– SpaceX Falcon 9 Rocket Fails on Launch, SpaceRef
Keith’s note: this is Internet rocket analysis and has nothing to do with what actually happened.
Keith’s note: This slow motion video from Astronomy Now shows that the Falcon 9 engines seem to be burning well after the explosion – and also that something shaped like a Dragon is seen exiting the cloud. Note: this is Internet rocket analysis and has nothing to do with what actually happened.
Falcon 9 experienced a problem shortly before first stage shutdown. Will provide more info as soon as we review the data.
— Elon Musk (@elonmusk) June 28, 2015
There was an overpressure event in the upper stage liquid oxygen tank. Data suggests counterintuitive cause.
— Elon Musk (@elonmusk) June 28, 2015
Keith’s update: SpaceX is working toward a launch of the CRS-7 ISS Resupply mission aboard a Falcon 9 rocket this morning at 10:21 am EDT this morning. If all goes according to plan they will make another attempt to land the first stage on one of their autonomous drone barges.
Keep an eye on @SpaceX and @ElonMusk for post-landing news, pics, videos, jokes, etc. We will be live tweeting at @NASAWatch, @SpaceRef, and @Spacecommerce
Damn!
Oh, sh!t.
Blast and damn.
Ouch
“Non-nominal flight”. Big-time serious impact to the ISS? They needed this cargo.
Well, that was unexpected.
Not by everyone.
Failure is always a possibility in this industry, especially when the vehicles are expendable. Not being able to recover 2nd stages after each flight means no possibility of inspecting the flown hardware. No inspections means it’s much harder to find problems since telemetry will never be able to tell you everything,
It blew up near the point of staging.
Looks like the second stage pressure vessel failed. First stage continued to fire. There was no fire initially – just vapor.
Disclaimer – this is only from looking at pictures.
In both timeline and hardware. It looks like the failure started at the top of the vehicle, possibly at the Stage 1 / 2 adapter? Premature stage separation or Stage 2 ignition?
It definitely started at the top of stage 2, not at the 1/2 separator. Click through the videos, and you can see it. The stage 1 engines kept running incredibly long, for many seconds, with the guidance apparently completely stable, until the whole vehicle disintegrated.
You can also see what must be the Dragon capsule falling behind for quite a while, as stage 1 soldiers on….
Early on, I think there is just a massive LOX leak, and what we see is ice crystals all around. I am surprised because the stratosphere at 30 km is very very dry. I suppose it could be dry ice, from CO2? Once kerosine & LOX burn, that will produce a lot of water, and likely some carbon smoke as well.
It looked very symmetrical and slow. Thrust termination system activated?
So what happens to the people on the ISS? They needed these supplies, big-time. Three resupply missions blow up in a couple of years, two in a few months. Can we get new supplies to the ISS soon? How? NASA had to have a contingency plan for this, but I have no idea what it would be.
If they don’t have enough supplies for 6 they would go to 2 or 3 crew. If not that, they’ll have to leave it until supplies can be sent up. They might have to prepare to leave it at some point timed so if a supply ship makes it, they stay. Probably takes a while to prep to leave it unmanned.
There is a Japanese resupply mission flying soon. And another Russian one, both flying before September.
It appeared to lose integrity at the interstage just seconds before MECO. I hate to see this happen to ANY LV provider, just more so for SpaceX .
With the camera next to the second stage engine there should have been video from this area.
The Falcon got eaten by the bear.
this could not come at a worse time. The f9 just got both DoD and NASA certified, and the first DoD open bid is next month. This also must severely impact the human-rating of the falcon.
To quote orwell from the book “1984”, it is double-plus ungood.
There should be excellent photography in addition to telemetry. I think they’ll find the problem quickly. The main thing will be convincing the customers of the fix before the next flight which I would guess could be in 6 months. The main problem now will be how much supplies are on the station and will Progress suffice. If not, the ISS may have to go to a 3 person crew.
Looking at replays again, the first stage did an excellent job keeping on target while the upper stage blew up. There was first the white cloud then a small explosion, more white cloud then major explosions followed by range safety destruction.
I had thought about a scenario yesterday where the first stage lands on the barge, but the Dragon doesn’t make it to orbit. When someone asked me which part of this mission is more important.
I wonder if range hadn’t blown it up if the first stage would have stayed in one piece.
range didn’t blow it up, the signal for self destruct was sent automatically over a minute after it disintegrated
Which raises the question, if the second stage blew a minute a after desperation would range have had to blow the first stage????
There are 3 flights scheduled for August – NASA’s Jason 3, the SES-9 with the enhanced F9, and Orbcomm cluster.
And Sept 2 is the next cargo flight. Plus 5 more over the rest of the year.
This is going to wreak havoc.
The Progess flies next week, and Japan’s HTV-5 flies mid-august, so supplies shouldn’t be an issue.
I wonder if a LOX line broke or detached in the 2nd stage.
Could have been related to 2nd stage prep for startup.
Certainly possible
Why would that have caused overpressure?
I speculated that before i saw anything about overpressure. That being said, a leak can cause the system to compensate or overcompensate.
Because as a cryogenic liquid warms up, it vaporizes and expands. Also, you break a LOX line and any source of spark at all will cause ignition.
Ice from condensation causing a valve to stick. Stuck values have been a perennial problem with cryogenic rockets, historically. STS purged the entire engine compartment with N2, to beat it.
You can also have heaters on the valves I believe.
most likely the pressure relief valves on the O2 tank froze over from the high humidity – Certain four letter agency with lots of space experience recommended that SpaceX put dry nitrogen purges on those valves but their request was ignored
This seems like a possibility since SpaceX has made some changes to the helium pressurization system to bring production “in house”. Unfortunately, launching from the Florida coast comes with obscenely high humidity.
How do you know range safety was activated?
I was just basing that on the fact that the first stage was going nicely and then blew apart. Could have been by an autonomous destruct system on board the rocket or the first stage was damaged by the upper stage. I still haven’t heard if the vehicle had a self destruct that was initiated by breakup of the vehicle.
Thanks John, .. I guess they’ll sort it all out.Whatever the case its a terrible loss, hopefully it will be something simple and quickly remedied.
RSO did not destroy the rocket.
Could be true. This certainly makes ending RD-180 purchases now rather than later look less likely.
Too bad. Looks like something above the first stage exploded, perhaps the second stage. In slowing down the replay, there appears to be a white cloud coming out of the top stage area like an O2 leak. The first stage appears to operate quite well until the range safety package destroyed the vehicle.
An ugly reminder of the value of reuseability: if a thing can’t be tested it can’t be trusted.
Does SpaceX test fire second stages on the ground before assembling the vehicle?
Even if they test on the ground, it’s not the same as flight testing. As a fellow engineer likes to say, “Things that are different, just aren’t the same.”
Understood, but at least it would exercise the system and find gross errors. Early as Falcon is in its development it might be worth the bother.
I think they test the engines, but not all-up stages. I don’t think anyone does that routinely. Just the 3s static test SpaceX does is unusually conservative.
🙁
0 for 3 on ISS. Not good. No Cygnus. No Progress. No Dragon.
Progress was blamed on the launch vehicle, which was newer, more modern variant of Soyuz. The next launch is Friday on a different, earlier variant of the launch vehicle.
Actually, I believe it was blamed on the structural interface between the upgraded Soyuz launch vehicle and the Progress. So, the failure would be unique to this combination of launch vehicle and payload.
How long can ISS keep operating with only HTV-5 for resupply this summer?
There is also a Progress resupply scheduled for before September.
During the press conference, the Station guy said they currently have four months supplies on board. He said that when it gets down to a month and a half, they start planning for crew evacuation. (He said they like to keep six months worth on board normally.)
There’s a Progress scheduled to go up next Friday, plus a Japanese HTV (the last?) schedule for August sometime, so they have two more chances to resupply before things get tight.
Also mentioned: The next Cygnus flight (on an Atlas 5) is currently scheduled for December, but they may be able to move that up to as early as October if there’s a need. (NASA people don’t tend to give such answers unless it’s already been looked at pretty thoroughly, FWIW.)
(Hat-tip to Keith, who asked the question they were answering.)
Thanks. Actually this whole logistics thing is something that the SSF/ISS program has thought long and hard about – we were deep into it back in the early 90s and it drove design evolution, repairs vs replacement, how to stage things in advance, etc. We had lots of DoD logistics folks working it – we stalled a bit during Gulf War when they were all called up for duty and left us civies with all the work.
Interesting historical tidbit. Thanks for sharing.
Times from video:
22:40 Two plumes briefly visible on opposite sides of the upper stage
probably shock wave condensation asvehicle goes supersonic
23:25 “____ chill has begun”
23:44 vapor cloud suddenly appears; second stage LOX tank rupture?
23:44-52 first stage continues firing surrounded by vapor cloud
23:52 First stage disintegrates
Where is the second stage helium tank? Isn’t it in the fuel tank?
What we need to see now is a careful but expeditious investigation by SX with corrective action and design changes implemented quickly enough to avoid further impact to the flight schedule and ready for test on the next launch.
Musk said there was an “overpressure event” in the second stage O2 tank, with data suggesting a “counterintuitive cause.”
Could be as simple as insulation missing that caused external heating to soak through to the tank.
My guess is a pressure relief valve failure. Such a failure is not unheard of, especially if humid Florida air was able to come into contact with a cold valve.
As Tony Stark asked, “How did you solve the icing problem?”
Three failures in a row with three different cargo vehicles? Something not right about this. I’m getting a might worried.
“Once is happenstance. Twice is coincidence. Three times is enemy action”Ian Fleming
True… It’s just that sometimes the enemy is us.
“Keep calm and carry on…..” Go SpaceX !
I live about 6 hours from Jacksonville. In my driveway in a small boat on a trailer ready. I was waiting on the launch to see if SpaceX would stick the landing. I was going to drive to the mouth of the canal and attempt to be one of the first person’s near land to see the booster on the barge. Then I was going to drive to the Dames suspension bridge and catch a picture as it went under. Once it was in port I was going to take pictures as they unloaded it with their crane.
Obviously my plans have changed.
Keep calm and carry on….
Go SpaceX!!!!
Yes, DON’T GIVE UP ! I wish I could be there to welcome the victory parade when it finally happens. (I did fly to HI when I was a grad student to watch Apollo 8 TLI in Dec 2008, and then also to KSFC for Apollo 11, so I’ve had my turns. 🙂 ).
In the final minute before lunch, as they topped off the second stage tank, we heard a puff from the release valve, and then a strange squealing sound that we hadn’t heard during previous launches.
Anyone else pick up on that?
I heard it. Don’t know what it means.
I’m happy that no humans did lost their lives.
Three major malfunctions on supply missions to the ISS within the last 10 months. Bad…
But at the ISS are still more supplies than Apollo 13 ever had to survive.
And I’m sure that SpaceX will come back to their business as soon as possible. It will not be a two or three year delay like NASA did after their own disasters.
Did this Dragon not have a launch abort system?
No, this was a cargo-only Dragon 1 with no LAS. Dragon 2 is the crewed version with the LAS thrusters to move it away from booster problems like today’s in a hurry.
The in-flight test of that is due sometime soon. Date uncertain as of a few days ago, IIRC.
What does LOC and LOV stand for?
Library of congress? and……?
Loss of craft?
Loss of Vehicle?
It’s not hard to type a few words more to be clear
Loss of crew. Loss of vehicle.
Loss of crew, loss of vehicle. But I disagree that LOC is inevitable.
Agreed, with the crewed Dragon Launch Abort System that covers every second from lift off to orbit a loss of crew scenario should not be inevitable at least not one due to launch vehicle failure like this.
Loss Of Crew and Loss Of Vehicle
Thank you Doug
Keith just asked about skip cycles and contingency planning. Sufferdini said they start making decrew plans when they’re at 45 days to zero supplies. Today, they’re at 4 months to zero. Trying to get up to 6 months while simultaneously doing research. Still planning on flying the new crew. Still planning on operating nominally.
It helps that the Japanese are also launching a supply capsule in August.
Nice question Keith.
We need to make sure we are all allowing SpX and the FAA to do their job and get back to flying. Bad days suck and we all have find the root cause, fix it, and PRESS.
We have some time – we will do this right.
At the end of the press conference it was mentioned that telemetry was available from the Dragon after the break up. Was there any chance of the parachutes could have deployed?
I suspect the EDL computers were not even on, so the possibility of parachute deployment is very low. It would have hit the water at terminal velocity, for the capsule that would be about 100 m/s (230 mph). Hard to imagine it not cracking open at that speed.
it’s never a good day for our business when this happens to anybody. I have a lot of sympathy for the guys at spacex who are going to have to go through the investigation/failure process. trust me that is not an enjoyable time at work. if you are thinking Elon will conduct a quick investigation find root cause and start flying again real soon you need to stop drinking the cool aid. NASA and the AF will determine when the investigation is over and it will be thorough. in the meantime there are a lot of politicians sharpening their knives right now. they’re probably going to be after Elon’s buddy McCain as well. sometimes powerful friends have powerful enemies…
No doubt. And that shiny new AF certification just got revoked. No surprise either, SpX has been flying on the edge from day one. A failure was bound to happen soon once they passed the initial flights and were more into the “production line” mode. Quality control has always been an issue for them.
Tell us how they were flying “on the edge” and about how “QC” has always been an issue for them. What facts do you have that support those ideas?
Source on Air Force revoking SpaceX’s certification??
Lt. Gen. Samuel A. Greaves, commander of Space and Missile Systems Center, about SpaceX certification and the effects of the destruction of the CRS-7 mission to resupply the International Space Station.
“Q1. Does this failure prompt review of certification for national security launches?
“A1. No, SpaceX remains certified.”
http://breakingdefense.com/…
Unfortunate launch failure for Space x but they are not the only one’s to suffer a launch loss, those are the perils of launching rockets, however they will have large amounts of telemetry including video feeds etc. to help ascertain exactly what went wrong, we can all speculate what went wrong but at the end of the day Space X engineers should quickly be able to pinpoint the exact problem and take corrective action with minimal effect on their schedule. What I think is also worth noting with regard to manned flights on the F9 is that even in a situation like this the robust Dragon launch abort system would easily have been able to ensure crew survival. Keep calm and carry on Space X.
Good point. This event took place in mid-flight. If it had happened to Apollo, or if it happens to Orion at a similar time, their escape towers might already have been ejected and the crew lost.
That’s not what Spacenut said and I doubt you are correct. I’ll let you look it up, I’m not bothering.
I am really curious if Dragon survived reentry and could have run some kind of abort sequence?
Shotwell stated that the Dragon v.2 escape system would have saved a crew.
I’m just wondering if the cargo version could have made it to a splashdown assuming it survived the mishap event.
Check out @John_Gardi’s Tweet: https://twitter.com/John_Ga…
There is Dragon that point of the arrow
Could the computers have handled reentry and chute deployment in such a mishap?
I suspect the EDL computers were not even on. So the possibility of parachute deployment is very low. It would have hit the water at terminal velocity, for the capsule that would be about 100 m/s (230 mph). Hard to imagine it not cracking open at that speed.
I Watch video on YouTube SpaceX CRS7 launch 1minute 14 second after liftoff notice white plume dot form on second stage very strange not normal.I see Falcon 9 rocket destroy by FLS 2 minute 20 seconds very sad day. So failure start alot earlier lox leaking from lox tank is the source of the Falcon 9 rocket failure.
I think you are seeing a ring of condensation formed by aerodynamic pressure as the rocket goes supersonic, this does happen occasionally.
http://www.space.com/images…
http://www-rohan.sdsu.edu/~…
http://media1.s-nbcnews.com…
Happens to airplanes, too!
http://media1.s-nbcnews.com…
Agree happened just after call “vehicle supersonic”
(& Hug Doug) Prandtl–Glauert singularity
https://en.wikipedia.org/wi…
Edit Condensation bit (facepalm)
LOC or LOV is not and never will be inevitable unless there are supernatural beings that csuse events. there may be high or low probabilities but unless magic exists nothing is foreordained. That makes no logical sense to say it is inevitable.
The real root cause of pretty much every mishap is human fallibility, for which history teaches us there is no perfect cure. Hence Cessna_Driver is essentially correct. Any system built by humans is vulnerable to failure and the lower the margins against failure and the more frequent the operational exposure, the greater the risk of failure. In this business, the risk exposure is high enough that it is reasonable to assume that we will continue to have failures.
Vulnerable yes. Inevitable, absolutley not.
Bernardo/Cessna/Jeff are right, as long as humans are involved, there will be some level of failure. But history also shows that the failure rate in any given technology can drop dramatically if it can be routinized. Space is hard, but then so was flight. If worldwide we performed less than 100 flights per year, and flew each of aircraft only once before discarding, I imagine the failure rate of planes would be pretty horrific also. Reusability is the key to routinization, in part because it lets us examine flown hardware, but primarily because it lowers price and lets us launch more often.
Exactly correct, but also the increased difficulty of spaceflight means that the margins for error are smaller, which in turn means that it must be approached with more rigor. Or by finding more forgiving approaches…
I do not disagree. My only quibble is with the inevitability of catastrophic failure. Def Inevitable: “that you cannot avoid or prevent”
There are failures and there are failures. Only certain unique failures lead to catastrophe and those particular failures that fully penetrate a defense in depth design are in no way inevitable. Possible, even likely, but not inevitable.
Humans make mistakes therefore failure is inevitable. I’m not sure how this could be made more clear.
Mistakes do not lead inevitably to failure. The whole concept of “defense in depth” is predicated on providing robustness.
in any time less then infinity, entropically open systems do not have any inevitability.
let me just ask this again: is it inevitable that an AtlasV will fail. Maybe one will and maybe one won’t.
I see no preordained magical requirement that a lov or loc is inevitable.
There is always the possibility an Atlas V will fail. Unfortunately, we don’t know precisely the odds of failure since it is often the “unknown unknowns” that cause such a failure. On an expendable vehicle that cannot be inspected after flight, it’s quite difficult to know exactly how close any one flight came to failure. Telemetry can only tell you so much.
As an example, the Saturn V had a “perfect” launch history, but anyone at all familiar with the program will be able to tell you just how close some of those flights were to LOV. POGO in the 2nd stage engines/plumbing/tank/stage was a huge issue.
Apollo 6 nearly did itself in.
“Unfortunately, we don’t know precisely the odds of failure”
I agree. Unless the odds are at unity then its only a maybe, not a must be.
Statistically, you’re claim can’t hold. There exists a possibility that failure will not occur, even if that possibility is small. In fact, I’d argue that reusability coupled with a robust inspection program should increase the odds that a failure won’t happen.
There are failures and there are FAILURES. Yes, a hydraulic line may lose pressure because it was connected wrong. But that “failure” does not inevitably lead to “FAILURE”.
Nothing inevitable.
Bad news, the F9 just jumped up to ~5% failure rate, much higher and they start looking like a risky vehicle for launch. Most new vehicles experience at least one failure early on, but it’s almost worse for it to happen this far in. However it could certainly be worse. So far it doesn’t look like an engine failure, meaning that the root cause should be easier to address. The worry is that one failure of this type could point to other problems yet unrecognized. Or not, we are trying to determine reliability with a small number of samples and it’s hard to tell whether we are lucky or unlucky.
There is no luck in space launch; there is only skill and chance. In this case an unanticipated failure mode has been identified prior to initiating manned flight.
I use “luck” to describe the vagaries of chance. Think of rocket launches like a dice roll where you fail on a “1”. SpaceX just rolled a 1 but we still don’t know whether the dice they are rolling is a d20, a d10, or a d100. And we won’t know for quite some while.
Although it seems intuitive that failures are random events, most launch vehicle failures are deterministic, that is they are extremely likely to occur under the right circumstances. Once the sequence of events is understood, the appropriate corrective action is a design change to eliminate the failure mode. How long this will take is unclear, but SpaceX has a record of analyzing and correcting problems quickly.
Understood, and I am not concerned that the failure mode that destroyed this vehicle will be suffered again. The concern is the other failure modes we don’t know about yet, and since we don’t know what they are or how they might be triggered they are best treated as being random. What does this failure tell us about the probability of others? Perhaps not much, but in a Baysian sense, I now trust SpaceX’s quality control less than I did before the failure.
Without a full report from SpaceX, it’s premature to say this is a QC problem.
As Daniel Woodard mentioned, once the root cause is known, corrective action will be taken and one less failure mode will exist. I would only have doubts about SpaceX if they fail to correct the root cause, something I think is unlikely to happen.
At the risk of becoming abstruse… there are two perspectives from which we could consider reliability. The first is the absolute reliability of the vehicle, an unknowable parameter which would account for all failure modes, including those unknown to us, and the likelyhood of them happening. From this perspective, the vehicle that will fly after this problem is addressed will be safer than the one that came before.
The other perspective is the reliability which we would assign to the vehicle based on the information available. From this perspective, I would rate the Falcon as less reliable now than I would have before the accident, even assuming that this particular problem is fixed. One unseen failure mode suggests others, and stumbling across one, particularly in a system this complex, should not increase our confidence in the vehicle. This is not a controversial heuristic when it comes to, say, the Proton, and I don’t think that it should be here.
Very true and very well said.
While what you say is true, because the flight rates of launch vehicles are so low, applying statistics based on flight data is difficult. As with this failure, a single failure skews the data quite horribly. It’s doubly hard when the vehicles are all expendable and it’s literally always the first copy of a vehicle flying each time (i.e. infant mortality).
Agreed, statistics with small numbers is perilous, which is irritating given that so much depends on the reliability of a vehicle. While I have criticized the ponderous Air Force certification process before, I am sympathetic that unless a launch vehicle has a long history, the only valid way to estimate its reliability is to look under the hood.
As IS Chang makes clear in this seminal paper, the actual reliability of the vehicle changes with each flight, can be calculated, and is exactly as you describe it in your first paragraph.
http://www.ewp.rpi.edu/hart…
Your second paragraph does apply to some programs in which the appropriate design changes were not made after a contingency, either because the cause was not understood or because of lack of resources or organizational inertia, or because of the deterioration in quality control that has ocurred in some Russian programs. I don’t think that will be the case with SpaceX.
Thank you for the paper, Daniel. The breakdown of failure by subsystem seems particularly useful.
Quality control is an issue when the failure is the result of poor workmanship or failure to adhere to the design and fabrication procedures. A failure due to an unanticipated failure mode in a vehicle made to spec is a design problem and properly corrected by a design change.
Edit: A QC problem could result in a manufacturing defect in a COPV helium pressure vessel suh as those inside the LOX tank. SpaceX had a leak in one of the first stage helium tanks on a prior launch.
It is no big deal, rocket launch is risky business, only fanboys made an accident like this sounds so embarrassing. The problem with the “software company” is that it always releases before finishing debugging. I am sure there are thousands of bugs in the current version waiting to be found. BTW, some fanboy here promised me a first staging landing in June with certainty, I hope he is properly humiliated.
Who?
🙂
See, we are all fallible…. 🙂
The key is to structure the system so failures drive in a controlled preferred direction – “failsafe”.
That was the beauty of the Westinghouse train brakes. They were pneumatic pressure brakes. But, pressure was not required to make the brakes apply. Pressure was required to release and hold the brakes open. So if any part failed the brakes would be automatically applied.
Might have been me. We cannot count this attempt any more than if the Falcon had been struck by lightning on the way up. The first stage didn’t get a chance to return and land. I expect that the next attempt will succeed.
I think they are extremely likely to succeed, but I wouldn’t bet the farm on a certainty of any single attempt.
1 failure out of 7 missions ain’t bad. On to SpaceX commercial crew!
Failure rate is 1 in 14 at worst, if you’re only counting the flights of the Falcon 9 v1.1, not sure where you get 1 in 7 from.
1 in 18 flights (https://en.wikipedia.org/wi…, so 17 successes of F9 before 6/28/2015, neglecting the single engine failure in CRS-1, which did not compromise the primary objective (& even the secondary was successful, it turned out.)
The ability to tolerate an engine-out in the first stage seems a major reliability plus to me, that has not get sufficient recognition, IHMO.
Counting the Falcon 9 v1.0, yes.
Engine-out capacity is very important, and definitely should be included as a major “pro” when considering SpaceX’s abilities.
A major robustness of the design.
While increasing the likelihood of an engine failure, it simultaneously lessens the negative effects of an engine failure by a serious margin.
No you were not proven right. Unless there are mystic spooks, nothing in an entropic open system is inevitable. Is it inevitable that an atlas v will fail? Of course not. It is voodoo that things MUST happen.
The probability of failure cannot be 0, must be >0. It has nothing to do with spooks, it is just the sea of “little things” that you have not thought of, and which are always there. Probability noise, if you like. So, there will always be failures, but if we deal with the causes one by one as they occur, the expected rate of failures should go down monotonically.
Which is one strong reason for incremental improvements in existing designs (Soyuz, Delta [before Delta III], Atlas [before Atlas V], Ariane V, etc) before rushing to start over. Falcon 9 is getting impressively mature IMHO for a newbie start up rocket – 17 straight good flights, and at least 169 successful runs of the Merlin 1 as testimony.
Even if the probability is greater than 0 on an individual launch attempt (fine with me), that doesn’t mean that a failure will inevitably occur as Cessna says. It does mean though that the probability of a failure occurring will approach 1 (but never reach 1), as the number of attempts approaches infinity.
I hear @Cessna_Driver speaking in generalities while Yale S uses specifics. You’ll probably have to agree to disagree here. At least speak about the same thing when discussing. Yes, it’s true that one particular rocket model may not ever have a failure. But if you think waaaay in the future and for all rocket models, then for all intents and purposes for us humans, Cessna is right that there will be a failure sometime.
I agree with your main idea, but one launch failure does not prove that a failure is inevitable just as a cold day in Rush Limbaugh’s hometown in Florida does not prove that climate change is not happening.
I would rephrase your: “if you think waaaay in the future and for all rocket models, then for all intents and purposes for us humans, Cessna is right that there will be a failure sometime.”
to:
“if you think waaaay in the future and for all rocket models, then for all intents and purposes for us humans, … there will extremely likely be a failure sometime.
There a are a couple of reasons why I seem to be so obstinate about the distinction, but it will take some time for me to compose the rationale. later.
Because a probability is greater than zero it does not REQUIRE that failure occurs. We as animals will fail (die) because our components (without technological breakthu) MUST stop functioning. Our chromosomes have parts called telomeres which shorten every time a cell divides. When the last telomere is gone, the cell line ends.
It is a closed system where entropy drives the end.
Technology does not REQUIRE failure, either in part or catastrophically in total, Failures MAY occur, but not MUST occur. That requires deliberate action by supernatural intervention.
MUST an Atlas V INEVITABLY fail?
MUST an Airbus 380 INEVITABLY fail?
Can they? Might they? Sure. MUST they? No.
In a totally unexpected coincidence the telomeres I described above were being investigated with an experiment lost on the falcon loss:
http://www.spaceflightinsid…
Hmmm… maybe supernatural stuff does exist… or not.
There is always risk, but not inevitability. That is mysticism.
It’s mathematics. If your probability of failure is p per launch then after N launches the probability if NOT having a failure is 1 – (1-p)^N, which in the limit of N -> infinity is 1. So given enough launches the probability of failure is so close to “inevitable” that not even you should quibble about the difference.
Actually, I think I finally see what Mr. Simkin is talking about. You are describing a case where the probability of failure is constant. As others have pointed out, it isn’t. It decreases with experience and elimination of failure modes (and increases is reuse implies wear and maintenance issues.)
If, hypothetically, the probability of failure droped by a factor of two every for every event, that infinite product you describe becomes 1 – (1-p0)*(1-p0/2)*(1-p0/4)*… Which converges to a number less than unity. So in some technical, mathematical sense, there is a case where failure isn’t inevitable. I don’t see any real application of that. I’d believe 20% on a first test flight could drop to 10% on the second and 5% on the third. But I can’t see that going on forever, which is what the mathematics would require.
I heard one discussion where you have the probability of failures due to a design error plus those due to a process error. The more times you fly, the design error probability goes down while the process error is dependent on your QA and people doing the work.
Unless I misunderstood what you are saying, this sounds like the old coin flipping misconception. If we consider heads as success and tails as failure, the odds of failure on any particular flip are one in two. If I have managed to land it heads ten times in a row, the probability of tails on the eleventh flip is still one in two.
I used to hear numbers like 1/200 for Shuttle towards the later part of the program. Some might say that was too optimistic so let’s say 1/100. Assuming 100 Shuttle launches had flown successfully, the odds of failure on the 101st launch would still be 1/100.
Of course that assumes that p does not change. In fact p can change over time, due to factors like complacency from past success, inadequate knowledge transfer as people change positions or retire, failure to anticipate the effects of changes to the system, etc. In the other direction experience, improved testing methods, etc. can reduce the probably of failure, as mentioned by fcrary. Those are the things that change your probability of failure, not how long your “lucky streak” has run up until now.
You’re talking about the “gambler’s fallacy”, and I think you’re right. Without a change in probability, the probability for failure on flight n has to be the same as on flight n+1, unless the probability has itself changed. In any case, each launch should be statistically independent.
Yes, at INFINITY. Because a definition of infinity is that time in which everything happens.
The production of any technological device must exist over a time less than infinity (because it must include the time it is not produced) so failure is NOT inevitable.
I was wrong; you _would_ quibble about the difference between 0.0001 and zero. Let’s put some numbers to this. Suppose probability of failure is 0.1, which seems to be a pretty average batting average recently. After 100 launches the probability of there being no failure is 0.00003. That’s not zero but I’m certain you wouldn’t risk your life on those odds as your death would be suitably close to “inevitable”; and 100 is a very long way from infinity. You could argue that those engineers just aren’t doing a good enough job and they should reduce the odds of failure to 0.01 or 0.001. But then you may as well demand they reduce the cost to $10 per tonne to orbit and make the rocket dance a jig before it leaves the pad. One has to be pragmatic.
Who’s probability of failure is 0.1? You’re making a mistake if you think that calculating SpaceX’s probability of failure, based on 19 flights, is an accurate proposition.
Hard to imagine how this mishap doesn’t affect SpaceX’s already compressed schedule to support both human and DoD missions, its ridiculous for NASA to say otherwise (like they did). It might ultimately make the Falcon safer, but there is no argument for making them faster to get there. Yes its a set back, but SpaceX can rebound … however if this happens again the company would be in serious trouble before all their leaders and early employees can cash on their IPO stock with a crazy +10 billion evaluation.
Spacex is not a publicly traded company and has never issued stock.
It issues stock .. It just is not available to the public, it is restricted..
Yes, if you are considering internal equity positions by private investing. I was referring to the statement above: “can cash in on their IPO stock “.
IPO means Initial PUBLIC Offering.
Spaceflight is difficult, but spaceflight is doable. It’s the not letting your ego make you stupid part that is hard.
Elon Musk has known for a long time that failure is always a possibility and has reportedly agonized over the possibility of a launch failure. What will be telling is how SpaceX, NASA, and the Air Force all deal with this failure and how they choose to move forward from it.
That is the interesting question at such junctures…
part of NASA’s commercial crew effort is technical support, KSC has teams of engineers assigned to special advisory and review boards that will answer questions and help resolve any problems commercial launchers like SpaceX have – of course it is up to the people at those companies to ask the questions and accept the answers.
I was away from NASAWatch with this happened, so depended on whatever news I could get on the radio. Lots of emphasis on ‘NASA having to re-consider the commercial program’ and similar crap. Nothing about the nuts and bolts.
NPR reported that SpaceX flew 18 in a row before the explosion, an important bit of information. But they also conflated the Falcon with Orbital’s problems; there is very little overlap as far as I know.
One wonders how many ships were lost by the great Henry the Navigator before they figured things out. How many were lost as he developed his crazy-good maps?
Right; right; very right.
can someone explain the phrase “counterintuitive cause” ?
failure rate 1:8 or ~11% – unacceptable in any industry
It means something happened the opposide way of what they expected.
Failure rate is 1 in 14 at worst, if you’re only counting the flights of the Falcon 9 v1.1, not sure where you get 1 in 8 from.
I’ve seen that number being thrown around. I assume it is in reference to failure rate of just the COTS missions?
Maybe, but the only reason that metric would be used is to deliberately misguide and misinform people.
No launch vehicle has a constant failure rate, since almost all catatrophic launch vehicle failures are the result of unanticipated failure modes which are (in a properly managed program) eliminated by design change as soon as they are revealed.
The first delta 2 exploded. 100% failure rate. Yet it is revered as a classic workhorse.
big difference, SpaceX has 50 years of history and engineering knowhow to utilize, NASA had ZERO!!!
Orbital and sub-orbital flight is extremely difficult, it being “High-performance” by its very nature, providing little to no room for error. In the past year, Orbital, Progress, SpaceX and Virgin have have all had Catastrophic Failures. There is no Cheap (or expensive) sure formula to access space safely and consistently, it requires the utmost diligence in every aspect from design to launch. SpaceX and Orbital are not exempt, not a criticism just a fact.
Modern jet engines are more complicated than rockets, and handle gasses which are hotter than the melting points of any the materials used in the engines. Jet airplanes look easy only after seventy years of intense experience. We Can Do This. We just need a little more practice.
Its the operational environment and conditions that make it “Extremely difficult” not necessarily the hardware. Jet engine failure do not always result in a catastrophic event. Not so with Rockets.
It looks that the actual event has few mysteries to it – The upper stage LOX tank ruptured and the whole vehicle progressively disintegrated thereafter. The question that SpaceX have to answer, clearly and promptly, is: ‘Why‘?
As a total amateur, the only two options I can think of are:
1) Manufacturing fault of some kind in the U/S;
2) Payload shifting and damaging the structure of the U/S.
I’m sure that real experts will have far more clues than I ever could.
The real issues here are the damage this does both to commercial space-flight and to the ISS Programme.
Could be as simple as a system relief failing or freezing closed and not relieving as required causing the over pressure event. Although cryogenic systems usually have more than one pressure safety device built into the system i.e. burst discs, pressure reliefs.
Whatever the cause of the anomaly , it was not the 9 first stage engines . You can observe them continuing to fire into the LOX cloud. But what was that powered projectile racing out ahead of the explosion in a straight line along the ascent vector , leaving it’s own LOX white plume/contrail?.
I’m hoping the rocketcams on the vehicle were broadcasting till the bitter end.
To put a positive spin on the accident: This is an non-counterintuitive educational moment.
Can you give the clock time in the video where you saw this?
The last call about 15 seconds before the overpressure was to begin second stage engine chilldown. This would require pumping the LOX through a circuit including the engine. Since there is obviously no hot fire test of the second stage (unlike the first stage) could an undetected problem in the LOX chilldown circuit have triggered the overpressure? Is additional helium released into the LOX tank at this point?
Edit: I have been told the chilldown is open-circuit. If so the only source for the overpressure is the helium tanks. When chilldown starts the LOX tank pressure should drop, instead it rose, thus counterintuitive. SpaceX has had trouble with the COPV helium tanks before. In the LOX tank there is the possibility of LOX permiation into the composite overwrap. If so, the drop in pressure aschilldown begins might trigger boiling in the LOX permiating the composite, causing delamination. An He leak could be at a rate too fast for the pressure relief valve to compensate. Other possibilities include a frozen pressure relief valve accompanied by a regulator failure.
I totally agree with much of what you say sadly accidents are always going to happen and sometimes this will involve loss of life. Space travel is extremely risky and we should never underestimate or undersell that risk and never overreact when fate deals us a tragic hand, simply understand what went wrong and try to ensure the failure mode that led to loss of life can not occur again. However I don’t think we should ever just accept a loss of life as inevitable as we then start to view crew as just another asset like the cargo on CRS7. We must always strive to ensure that loss of life is never inevitable even if that is truly an impossible goal.
What SpaceX needs to do, to get back on schedule.
https://screen.yahoo.com/th…