Major Proton launch failure

[Kilmeister]

Note the 'lift-off switch'-0.4 seconds before nominal, exactly what happened on this flight. And of course nothing failed-that's my point.

Let me clarify what I meant. This is just a log of major events, not every technical detail.

For the sake of an example, lets say this was a car.

The event in the log would be "Changed gear 1 - 2"

not something like : "Throttle Reduced, Clutch plate opened, dog gear detached from 1st Gear, Dog gear moved to 2nd gear, dog gear attached to 2nd gear, clutch plate closed, throttle increased"

So if during our mission we were scheduled to shift from 1st to 2nd at 1 second, but due to a strong headwind we didn't shift until 1.2 seconds we would still have a successful shift to 2nd gear, but it would have been a +0.2 second variation from the flight plan.

Might not be the best example, but it was the first I could come up with to explain what I meant.

[PakledHostage]

Using the crash site coordinates given in the article that Kryten linked to (46° 3.633' N, 62° 59.717' E), I plotted the crash location and some other info in Google Earth. Helps put things into perspective:

[Shuttle]

Angular velocity sensors wired up with the wrong polarity remind me of the second Energia launch when the gyroscope block was installed upside down. The most interesting thing is that the block couldn't be installed in a wrong way but some smart guy made a special adapter for it. This is the info from the russian Energia wiki page, I haven't found any info about this accident in english wiki

[magnemoe]

Angular velocity sensors wired up with the wrong polarity remind me of the second Energia launch when the gyroscope block was installed upside down. The most interesting thing is that the block couldn't be installed in a wrong way but some smart guy made a special adapter for it. This is the info from the russian Energia wiki page, I haven't found any info about this accident in english wiki

That made sense, some fool designed an adapter to be able to make an mess of an foolproof design.

Almost scary.

[Bunsen]

http://www.interfax.com/newsinf.asp?pg=2&id=429110

Anatoly Zak is corroborating that, but doesn't give an original source: http://www.russianspaceweb.com/proton_glonass49.html#culprit It certainly fits the otherwise-unexplained aspects of the rocket's behavior, though. Specifically,

wait, rewatched it. Its still strange one engine is behaving differently but I've just noticed something. There doesn't appear to be any differential control going on here. when the rocket is nearing straight up its still thrusting into the turn, something it absolutely should not be doing.

An amplifying oscillation like that makes me think somebody dropped a negative sign or plugged a gyro in backwards.

[TouhouTorpedo]

I think we have a bright career ahead of us as Space Crash Investigators Bunsen.

Though, I've gotta admit I'm not entirely buying the "Gyros backwards" thing here. It doesn't make sense why it would correct its course rather than just pile into the ground upside down. It could be a result of Kalman filter feedback saying "I expect to be here" and finding its gone the other way giving such a command, but I'm not 100% sure if it would behave that way. Whats for sure if there is such a filter in the system, it should have been able to spot the sensor errors, giving the option to discard those inputs via software.

Theres pretty much implications across the whole board regarding the Proton rocket and program after this.

Edited July 10, 2013 by TouhouTorpedo

[Bunsen]

Yeah, it only makes sense if the angular rate sensors aren't the only source of attitude information. It seemed to be exerting correct proportional control, but the differential part was backwards. That would require having an absolute attitude reference (horizon sensor, maybe?) that it trusted for knowing its orientation, but using the rate gyros as the dominant source of angular velocity information. I can see how that would be useful; gyros could give more precise instantaneous readings of angular rates, and the absolute reference would let you null out integration errors.

I would tend to think that a huge conflict between the model and a sensor would ring some alarm bells and tell the filter to ignore that sensor, but I don't write those programs so I don't really know. Maybe they depend on redundancy to spot errors, and only notice if there's disagreement between the same type of sensor. But if they all get installed by the same crew...

[Kryten]

They're talking about one bank of sensors out of the whole system; my guess is there were too many agreeing for it to ignore the information, and too many working in other systems for it to just go immediately. Looking at the videos, I think it was the increasing rate of roll that ultimately overwhelmed the rest of the system.

[maro]

....and fell not far from a triple UR-100NU missile silo facility at Site 175.....

upps.....lucky they are even in there mishaps

[Dooz]

Velocity sensors installed backwards

[Bunsen]

Roscosmos is finally confirming what we've been reading on RussianSpaceWeb's increasingly detailed account of the angular velocity sensor issue. It sounds like installing them upside down was impossible without damaging the units, but somebody managed to flip them over and destructively jam them into place anyway.

I'm sure there will be some very unpleasant company seminars and more layers of documentation and verification in the future, but it still means that somebody thought the hammer-and-crowbar approach to assembly was appropriate for rocket avionics, and was so comfortable with it that he never said "Man, I really had to bash that sucker good to make it fit. Is it supposed to be like that?" to anybody. Or, worse yet, did ask that and was told "Yeah, happens all the time with that subcontractor's parts. Go ahead and bash the rest in too" or "It's 4:30 on Friday and I can't go home until you're done. Less talk, more bash." That's got to give potential customers some second thoughts.

[Kilmeister]

Once again, all the training and technical expertise in the world pale in comparison to the might of human stupidity. Nice to see an official statement, I had just opened this forum and was about to dig to see if there was an update.

oh and "Even a small fault in the launch command that was given 0.4 seconds earlier did not affect the performance of the engines." (From another page linked on that site) So it was indeed an error, but not the cause of the failure.

I had a similar(albeit much less costly) incident when I was doing some IT project management. I had one of my workers successfully insert an RJ-45(Cat5 Ethernet) cable into an RJ11(Telephone Cable) wall jack. My best guess is that when it didn't fit he used a combination of a hammer, pliers, and a screwdriver to wedge that bad boy in there. I gathered all my workers around and told them to marvel at the achievement. It took a great deal of focus and determination to accomplish this feat without ever thinking "hrm, maybe this doesn't go here?" This was the handywork of a "Trained" employee.

[Nuke]

well thats what you get for putting your accelerometer in backwards.

[koshelenkovv]

If you read source http://www.federalspace.ru/main.php?id=2&nid=20216 you discover, that it says that sensors was flipped 180 degrees.

It does not say that they was placed upside-down.

And people, who knows process more closely, says that sensors will not work upside-down (rotated 180 deg. over horizontal axis) and this will be discovered straight away at first check.

It's more likely that they were rotated 180 deg. around vertical axis (so right become left or face become back)

Disregard this, there just too much contradictory information about accident.

Which only confirms that we cannot know for sure if this is real failure cause or they just found some scapegoat.

Edited July 19, 2013 by koshelenkovv

[Flixxbeatz]

Ground control to major... Proton?

[Shiny]

You know, I once put on my cockpit backwards on a ship in KSP. When I used mechjeb to orbit it, the results were similar.

[Thorbane]

this is why you don't place your probe core on backwards.