SpaceX Falcon 9 - AMOS-6 - (Pad Failure) - DISCUSSION THREAD (2)

[SWGlassPit]

  There's no reason to assume that happened simultaneously with the tank breach.

Yes it does, because the avionics sits on top of the tanks and all data and power harnesses run along side of it.

I'm only talking about a difference of tens of milliseconds (likely only two video frames).  The disturbance created by the breach is still going to have to propagate to whatever failure path created that cutoff.  It would only be perfectly simultaneous if the initial location of the breach coincided perfectly with either the boxes or the data/power lines.  Otherwise, it would take a small -- but finite -- amount of time for the failure to propagate far enough to kill the data stream.

[glennfish]

Is the assumption valid that the cessation of data coincided with the emergence of the fireball?  Why couldn't the data cease, say, 30-40 ms after the fireball appears?  Or even longer?  The data doesn't stop just because the tank gets a big hole in it -- either the avionics equipment sending the telemetry or the transmission path has to fail.  There's no reason to assume that happened simultaneously with the tank breach.

My assumption wasn't about a tank breach, but the USLR first frame of the visual event.  That frame may have occurred many milliseconds after whatever initiated it.

I wouldn't be surprised if they were getting 1st stage engine data for many seconds after the fireball started, but for any sensors near the initial event, things are departing initially at around 1,000 fps before they start slowing down to a fast fire.  Divide measured (horizontal, not vertical) radius (not diameter) of initial fireball by 16.x milliseconds to get a speed check.

I would suspect without knowledge, that sensors nearest the initial fireball were most likely providing the most relevant data, and that they would have started departing for the nether reaches trailing their cables within a few milliseconds thereby giving up their ability to contribute to the data store.

[SWGlassPit]

Is the assumption valid that the cessation of data coincided with the emergence of the fireball?  Why couldn't the data cease, say, 30-40 ms after the fireball appears?  Or even longer?  The data doesn't stop just because the tank gets a big hole in it -- either the avionics equipment sending the telemetry or the transmission path has to fail.  There's no reason to assume that happened simultaneously with the tank breach.

My assumption wasn't about a tank breach, but the USLR first frame of the visual event.  That frame may have occurred many milliseconds after whatever initiated it.

I wouldn't be surprised if they were getting 1st stage engine data for many seconds after the fireball started, but for any sensors near the initial event, things are departing initially at around 1,000 fps before they start slowing down to a fast fire.  Divide measured (horizontal, not vertical) radius (not diameter) of initial fireball by 16.x milliseconds to get a speed check.

I would suspect without knowledge, that sensors nearest the initial fireball were most likely providing the most relevant data, and that they would have started departing for the nether reaches trailing their cables within a few milliseconds thereby giving up their ability to contribute to the data store.

Their statement says 93 milliseconds from "first signs of an anomaly" to "loss of data."

Unfortunately, there's a lot of room to interpret that.  Does "loss of data" mean loss of all telemetry from the second stage or does it just mean loss of some data channels?  On the flip side, are the "first signs of an anomaly" the dropout of channels for sensors near the initial failure?  Without clarification, it's hard to precisely place those endpoints with respect to the video frames.

[glennfish]

Their statement says 93 milliseconds from "first signs of an anomaly" to "loss of data."

Unfortunately, there's a lot of room to interpret that.  Does "loss of data" mean loss of all telemetry from the second stage or does it just mean loss of some data channels?  On the flip side, are the "first signs of an anomaly" the dropout of channels for sensors near the initial failure?  Without clarification, it's hard to precisely place those endpoints with respect to the video frames.

LOL, very good questions.

My interpretation would be that at least one sensor goes off nominal and then 93 ms later it goes offscale low.  But they didn't say that. 

Going back to my point that lead us down this path, 90+ milliseconds is a very long time.   All kinds of things can happen in that time, which means I get to see if I can find anything I missed in those last 5 frames of data, like maybe this isn't a statistical anomaly or algorithm created artifact after-all, whatever it is.

[Wolfram66]

Their statement says 93 milliseconds from "first signs of an anomaly" to "loss of data."

Unfortunately, there's a lot of room to interpret that.  Does "loss of data" mean loss of all telemetry from the second stage or does it just mean loss of some data channels?  On the flip side, are the "first signs of an anomaly" the dropout of channels for sensors near the initial failure?  Without clarification, it's hard to precisely place those endpoints with respect to the video frames.

LOL, very good questions.

My interpretation would be that at least one sensor goes off nominal and then 93 ms later it goes offscale low.  But they didn't say that. 

Going back to my point that lead us down this path, 90+ milliseconds is a very long time.   All kinds of things can happen in that time, which means I get to see if I can find anything I missed in those last 5 frames of data, like maybe this isn't a statistical anomaly or algorithm created artifact after-all, whatever it is.

to better align with events of countdown in context w/ Glennfish analysis

Launch Sequence timeline
T-0:09:15   Stage 1 Helium Topping
T-0:07:45   MVac Fuel Trim Valve Setup
T-0:07:30   Engine Chill Readiness
T-0:07:00   Engine Chilldown (Bleed Valves Open, both Stages
T-0:07:00   Spacecraft on Internal Power
T-0:06:45   Stage 2 Helium Transition to Pipeline
T-0:06:35   MVac Hydraulics at Bleed Pressure

~~ Algorithms are always wrong... because Al Gore has no Rhythm !!! ;o Have a great weekend gang!

[Mader Levap]

I'd imagine that even if they did have strong indication of what process failed to allow this event, that they wouldn't announce it w/out buy in from major customers like spacecom, ses, iridium, nasa, etc.     all takes time.   that they're reiterating nov is interesting in this context.

I don't really see it as meaningful. During fallout of their previous failure they had unrealistically optimistic claims about date of RTF too.

They won't do RTF in november.

[bstrong]

Now that an overpressure event has been confirmed, I'll allow myself one more speculative post:

As the LOX tank and helium COPVs fill, you've got several interacting components changing in length, but at different times, rates, and ratios: the tank, the struts, the COPV's, and probably some fittings and helium lines, as well. They'll be shrinking mostly, but there may be times in the process when the COPVs increase in temperature/length.

I'm no expert, but I would imagine the changes in the lengths of the of the tank and struts would be pretty straightforward to model and are properly accounted for. The COPVs and helium lines, on the other hand, would be challenging, since you have to account for helium temperature and pressure, and it's probably hard to perfectly predict at what time the helium will be at what temperature. Plus you've got variations in overwrap to account for. Defects which may not cause problems with structural integrity may still change the insulative properties of the overwrap, which may then change the timing and rate of shrinkage/expansion.

So, the tensile (and maybe even compressive) loads on the struts and COPVs are going to be varying constantly during LOX and helium loading, and I wonder if the change in timing for this test caused them to be significantly different than modeled because of some non-obvious interaction (e.g., maybe the loads weren't too high, but they cycled too many times). Also, the loads would presumably be different on first and second stages, since the ratios of COPV and strut length to tank length are different, which could explain why the second stage is more fragile.

I can even imagine a scenario where they made the problem worse after CRS-7 by replacing the struts with beefier ones that shrink at a different rate. Apologies if this has already been discussed and dismissed. I'm pretty sure I missed some posts in the thread.

[sewebster]

So does this definitively rule out the pre-blast groaning sounds in the USLR video being associated with the rocket?

[Rocket Science]

When I attempted to analyze the event occurring in the sequence in thread 1, I was admonished by a member for even considering it, who stated it it occurred at 8 minutes period... Just sayin'...

[Maine1]

Does the LoX tank vent to atmosphere during filling?  If there was an over pressurization
would we see a change in the vapor plume, if even for a video frame or two before the 'explosion'?

[AbeJ]

Summarising what some knowledgeable NSF members posted in the last thread:
- Falcon is unusual (possibly exceptional) in having LOX-immersed COPVs
- some voices at NASA were not wholly convinced that SpaceX had truly got to the root of CRS-7, therefore it could be premature to completely rule out any connection
- SpaceX's COPVs are unusual in having Al liners, rather than Ti, which has the potential for differential thermal expansion issues.

The thing that strikes me as very odd is that there has not been a He system failure in a first stage yet. These stages have been fired many more times than the second stages have, so you would predict that failures would crop up there instead- unless there are fundamental differences in the He systems?

I don't think aluminum liners are unusual in COPVs.  I know of at least one example that has been flying for years and is still in use.  Of course, its not submerged in LOX, though!

[glennfish]

Does the LoX tank vent to atmosphere during filling?  If there was an over pressurization
would we see a change in the vapor plume, if even for a video frame or two before the 'explosion'?

If there was an over pressurization, if there was a change in the vapor plume, it was not visible in any video frame prior on the USLR video.

Trust me on this.  If not, download the 1,300 odd stabilized frames and look for yourself, or look at the prior delta analysis provided in thread one.

While there may have plausibly been a change in the vapor plume, it's not detectable in the USLR video.  I'd have sucked on that like honey plated gold if it were there. 

The only video change I have seen, is a statistically insignificant but intriguing change in the last few frames that I could not testify in court are real.

[bstrong]

One last thought on shrinkage rates: I wonder how much the shrinkage rate of the tank is affected by how quickly the insulative ice layer forms, which is in turn affected by humidity.

The kind of situation I'm postulating is: 1) Tank shrinks faster than usual because ice layer builds up faster due to high humidity that day. 2) COPVs shrink slower than usual because helium is hotter at the time they are submerged due to changes in prop and helium load timing. 3) All of a sudden a system designed for tensile loads is under compression instead. (I'm not proposing this as an actual probable scenario, just illustrating how hard-to-predict circumstances could have changed the relative rates of shrinkage, which could substantially change the loads on the COPVs and other components of the helium system.)

[john smith 19]

Their statement says 93 milliseconds from "first signs of an anomaly" to "loss of data."

Unfortunately, there's a lot of room to interpret that.  Does "loss of data" mean loss of all telemetry from the second stage or does it just mean loss of some data channels?  On the flip side, are the "first signs of an anomaly" the dropout of channels for sensors near the initial failure?  Without clarification, it's hard to precisely place those endpoints with respect to the video frames.

Indeed. it sounds straightforward, but is actually quite vague if you know something about the subject.

To get total telementry loss exactly at one moment you need to destroy the data link carrying the information. I think F9 telemetry is by FO landline somewhere near the base (I'm sure it's been mentioned but can't find it).

That suggests data loss will be gradual as more and more sensors, and their comms hardware, are destroyed.

I would suspect SX had a 3d film of that loss pattern within a few days of the event.

Unfortunately that gives you the source of the event, but not the chain that caused it nor the specific cause. It could be the data that would decide one over another would only be generated by sensors that had already been destroyed.

OTOH this situation gives SX something they did not have on CRS7, a very nearly complete set of debris. In principal you can track the ever escalating damage back to it's source (which is likely been turned to dust). That will take some time.

Please note from the Updates thread SX have now specifically ruled out the struts (or at least that's my reading of the sentence in the update thread press release).

[ccicchitelli]

I just want to chime in here to state that there seems to be some confusion about "cause and effect" among some members' posts (not naming names). All SpaceX has stated is they know the effect: "a large breach in the cryogenic helium system of the second stage liquid oxygen tank took place". They are still tracking down the cause. As such, it's premature to exclude potential causes such as pipelines, fittings, valves, etc. We've narrowed the potential failure paths, but have not isolated it yet.

[Nigeluna]

This sounds like good progress to me though not yet there.

Wierd causes such as explosive bird droppings etc seem thankfully to be most improbable.

Telemetry indicates 93mS of anomaly - three frames of video with the first probably not showing a problem because once the deflagration becomes visible data cables will have died as I understand it.

The cause is declared localised to the helium pressurisation system and I understand at this time it was not fully pressurised. CRS 7 failed with the top of the LOX tank being popped off - Amos 6 had a failure near the common bulkhead. The symptoms do not look similar in addition to SpaceX saying the cause is different.

As it was still being charged with helium cpuld the fault lie in the helium charging plumbing within the vehicle rather than the COPV's - does this pass anywhere near the common bulkhead? Could the flow of helium have induced some kind of flutter in the plumbing a la organ pipe and could this have lead to failure and subsequent damage near the common bulkhead? Could the helium reservoir causing the damage been the charging supply rather than the COPV's? Was this sort of failure forseen and would the telemetry have detected it? Just some thoughts - I have no knowledge so just wondering.

They are getting there - I suspect there will be a RTF this year if they can dot the i's and cross the t's to everyones satisfaction. After all, unless they have changed something they have done this many times before. Regards

[Maine1]

Does the LoX tank vent to atmosphere during filling?  If there was an over pressurization
would we see a change in the vapor plume, if even for a video frame or two before the 'explosion'?

If there was an over pressurization, if there was a change in the vapor plume, it was not visible in any video frame prior on the USLR video.

Trust me on this.  If not, download the 1,300 odd stabilized frames and look for yourself, or look at the prior delta analysis provided in thread one.

While there may have plausibly been a change in the vapor plume, it's not detectable in the USLR video.  I'd have sucked on that like honey plated gold if it were there. 

The only video change I have seen, is a statistically insignificant but intriguing change in the last few videos that I could not testify in court are real.

Actually I have looked at a lot of your video analysis, and agree.  What I am getting at is that the initial fireball was a result of shrapnel from the helium system, rather than an over pressurized tank failing.

[docmordrid]

What are the odds of them pointing a tiger team at developing an autogenous pressurization system for the F9/FH LOX tanks? Just to retire the cLOX v He issue once and for all. Implement as "v1.2.1"

[glennfish]

Actually I have looked at a lot of your video analysis, and agree.  What I am getting at is that the initial fireball was a result of shrapnel from the helium system, rather than an over pressurized tank failing.

Thanks for your vote of confidence.    Image analysis is usually a neglected sport unless you're trying find who shot whom on an over pixelated surveillance video.

Re the shrapnel issue, if you go back to my debris analysis in thread 1, you'll see multiple trajectory analysis posts... The effort was to find where the explosion was.  The conclusion basically was... lots of places... at least from the 16 ms video source.  SpaceX saying helium system failure of some kind is cool, but there's no data that confines where that system is, both within the F9 and within the TE.   To me, that's a clarification of, "well, the He pressure dropped somewhere for some reason, and we'll get back to you." 

The "detonation" as I call it event has a geometry that can be quantified, but that geometry doesn't point to a "where".  It points to a sorta maybe possibly here or there.  Sometimes inside the F9, sometimes without... your mileage may vary.


Re the base suggestion that helium system failure resulted in shrapnel?  Absolutely it would, see my comments in thread 2.  The big exciting unanswered question is, was it in the F9 or not?  No data in the public domain provides evidence to support any answer to that question.

My humble opinion?  Debris trajectory data suggests, not in the F9. 

What that means?  F9 helium systems are both in and outside, and my trajectory analyses say, outside.

How and why?  Cripes, that answer is above my pay grade. 

[biosehnsucht]

Some have argued if the He system failed spectacularly enough to cause this resulting high speed failure of the second stage, COPVs must have ruptured violently. But I also understand that the He system / COPVs shouldn't have been fully charged yet, so perhaps a COPV failure, while still potentially catastrophic, might have taken long enough to occur to be easily detected in telemetry as rising pressure (and possibly even mitigated partially by LOX venting, though potentially still leading to failure just in a slower fashion)... and thus a COPV failure would seem unlikely to be directly the cause, as they perhaps weren't under significant pressure yet?

How are COPVs loaded with He? Do you have to force feed / expose the system to a high pressure of He into the whole He system and the COPVs fill until they're in equilibrium with the pressure you've chosen, or do you gradually raise the pressure of the entire He system / pump in He raising the pressure via the pump until you reach capacity? A bad analogy being a fuel cell car vs regular gas car, with the fuel cell car the hydrogen station has to pressurize the tank with high pressure hydrogen, then afterwards the fueling system must 'recharge' the pressure in the hydrogen station before the next tank can be filled. A regular gas car merely pumps it in at regular pressure until it's full, instead of giving it many PSI of pressure all at once.

I'm wondering if the supply lines to the COPV to pressurize them (which might include the same lines used to provide pressure to the LOX tank and whatever else uses He from the COPVs, or might be dedicated to just filling/draining He) had a failure somewhere along the line (such as the previously suggested fitting failure). The wording of the SpaceX update is vague enough that one could even make an argument it might have been part of the He filling system which could include equipment on the TEL, the umbilicals, or even the mating point between the umbilicals and the stage, in addition to the plumbing / COPVs inside the stage.