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

[eriblo]

>>friction ignition<<

Friction generating heat?  Ignition of what? Graphite requires extraordinary heat to ignite, even in pure oxygen, as NASA demonstrated back in the Sixties. Certainly not at cryogenic temperatures. The COPVs obviously did not burn up as they were found after the explosion. BTW Falcon COPVs have been found in Brazil after re-entry of the second stage and disintegration of the LOX tank. Despite a high-speed descent through the atmosphere followed by ground impact, they incurred only minimal impact damage.

Some of the copvs were found intact.  I don't think i ever heard anyone say that *all* of them were intact.  It only takes one to have a bad day.

In fact it was confirmed by SpaceX that at least one COPV was not intact. COPV #2 in the US LOX tank blew itself to pieces.

What condition was it in? How many pieces? Rupture of the tank due to mechanical disruption of some of the composite material could do something similar. But was it burned? I don't guess SpaceX has released any images.

What condition that COPV is in does not really matter. What matters is that it failed due to friction ignition of pooled LOX/SOX in between the overwrap and the liner. We can determine for a fact that the COPV was at least locally burned because LOX/SOX on itself does not burn. It needs a fuel. And that fuel was present: aluminium liner, carbon wrap, resin matrix. Don't forget that just about anything will burn in a pure oxygen environment.
So, was the COPV burned? Yes. Was the COPV blown to pieces? Very likely given the highly energetic nature of the event and the fact that it failed structurally while pressurized.

Yes, remember that it was mechanical failure of the COPV that destroyed the stage*. This failure was possibly caused by ignition of a small portion of the tank material due to "breaking fibers or friction" in the presence of oxygen. They don't explicitly state that ignition caused this particular failure (most likely unknowable), just that they have seen it happen. The images of the blackened LOX-tank at McGregor are suggestive of CFRP COPVs failures in LOX having fireball potential on their own...

*Just a fire in an unpressurized  COPV might destroy the stage, but it would not have the same millisecond explosive signature.

[Raul]

We know SpaceX have returned to using four COPVs. Does anyone recall when they first moved to three?

I found from webcast video that three COPVs were first at F9-016 with ABS-3A/Eutelsat 115 West B. Previous flights had four vessels.

[vulture4]

It would take significant burning to rupture the COPV. I was just trying to determine whether any of the fragments showed such damage. I also find it surprising that the COPV could normally be so permeable to oxygen. Even a small amount of LOX between the liner and the windings could boil and displace the liner as the COPV warms up. Is it possible that the extremely low temperatures induced mechanical stresses in the COPV, caused liner separation, or caused cracking of the composite?

[Pete]

It would take significant burning to rupture the COPV. ...

Nope.
If you have LOX-soaked burnable material, such as the epoxy of the COPV wrapping, and you manage to ignite even a teensy spot of it, you will have a huge fire in a fraction of a second.
Depending of the exact material and pressure conditions, you might even get an actual detonation.
(for reference, look up "Sprengel explosive" and "Oxyliquit")

[woods170]

It would take significant burning to rupture the COPV. ...

Nope.
If you have LOX-soaked burnable material, such as the epoxy of the COPV wrapping, and you manage to ignite even a teensy spot of it, you will have a huge fire in a fraction of a second.
Depending of the exact material and pressure conditions, you might even get an actual detonation.
(for reference, look up "Sprengel explosive" and "Oxyliquit")

Correct. Some folks here are just not fully aware of just how reactive a (liquid) pure oxygen environment is.

[RDoc]

I wonder if SpaceX has examined the COPVs on the recovered stages for signs of any buckling deformations. They've got quite a few at this point.

[cambrianera]

We know SpaceX have returned to using four COPVs. Does anyone recall when they first moved to three?

I found from webcast video that three COPVs were first at F9-016 with ABS-3A/Eutelsat 115 West B. Previous flights had four vessels.

Use of supercool LOX has an impact also on the mass of He that can be loaded in COPV.
Assuming same pressure, with 35% lower temperature you can load in a vessel 35 % more mass (PV=nRT).
They should have moved to three (second stage) with LOX supercooling.
I remember that F9-016 was very mass constrained (last legless F9); it was not supercool LOX, but maybe SpaceX tweaked pressures or He heating to increase performance.

[CameronD]

I wonder if SpaceX has examined the COPVs on the recovered stages for signs of any buckling deformations. They've got quite a few at this point.

I got the impression, reading through the information we've been drip-fed over time, that they regularly find some small deformations - but hadn't worried too much about it because they still passed testing.

[ziceva]

I got the impression, reading through the information we've been drip-fed over time, that they regularly find some small deformations - but hadn't worried too much about it because they still passed testing.

If this is true, I cannot but think (sadly) that is the exact same attitude of not worrying about O-rings in the cold and foam shedding during launch: something off-nominal that doesn't look bad and is ignored ... Luckily not the same aftermath ...
Hopefully not the case here ...

[ChrisWilson68]

I got the impression, reading through the information we've been drip-fed over time, that they regularly find some small deformations - but hadn't worried too much about it because they still passed testing.

If this is true, I cannot but think (sadly) that is the exact same attitude of not worrying about O-rings in the cold and foam shedding during launch: something off-nominal that doesn't look bad and is ignored ... Luckily not the same aftermath ...
Hopefully not the case here ...

This is not at all like the O rings and it's unfair to SpaceX to make the comparison.  With the O rings, they had a specific defined limit on temperature and they explicitly decided to ignore that limit over the objections of some engineers involved.  There's no evidence there was ever any rule about small deformations being bad that they decided to override.

In other words, there's no evidence small deformations were ever considered off-nominal in the first place.

[Lee Jay]

I got the impression, reading through the information we've been drip-fed over time, that they regularly find some small deformations - but hadn't worried too much about it because they still passed testing.

If this is true, I cannot but think (sadly) that is the exact same attitude of not worrying about O-rings in the cold and foam shedding during launch: something off-nominal that doesn't look bad and is ignored ... Luckily not the same aftermath ...
Hopefully not the case here ...

This is not at all like the O rings and it's unfair to SpaceX to make the comparison.  With the O rings, they had a specific defined limit on temperature and they explicitly decided to ignore that limit over the objections of some engineers involved.  There's no evidence there was ever any rule about small deformations being bad that they decided to override.

Maybe there should have been.

In other words, there's no evidence small deformations were ever considered off-nominal in the first place.

If true, that could easily be the direct cause of the loss of payload, vehicle and pad.

[Basto]

I got the impression, reading through the information we've been drip-fed over time, that they regularly find some small deformations - but hadn't worried too much about it because they still passed testing.

If this is true, I cannot but think (sadly) that is the exact same attitude of not worrying about O-rings in the cold and foam shedding during launch: something off-nominal that doesn't look bad and is ignored ... Luckily not the same aftermath ...
Hopefully not the case here ...

This is not at all like the O rings and it's unfair to SpaceX to make the comparison.  With the O rings, they had a specific defined limit on temperature and they explicitly decided to ignore that limit over the objections of some engineers involved.  There's no evidence there was ever any rule about small deformations being bad that they decided to override.

Maybe there should have been.

In other words, there's no evidence small deformations were ever considered off-nominal in the first place.

If true, that could easily be the direct cause of the loss of payload, vehicle and pad.

Hindsight is always 20/20.

[CameronD]

In other words, there's no evidence small deformations were ever considered off-nominal in the first place.

If true, that could easily be the direct cause of the loss of payload, vehicle and pad.

No, the direct cause has been said to be formation of SOX in places it wasn't expected to be.  Hence their intended solutions to allow RTF as posted upthread (warmer helium temps, etc.).

By itself, that's no reason small deformations should now be considered a major issue if they weren't before...even with 20/20 hindsight.

[Lee Jay]

In other words, there's no evidence small deformations were ever considered off-nominal in the first place.

If true, that could easily be the direct cause of the loss of payload, vehicle and pad.

No, the direct cause has been said to be formation of SOX in places it wasn't expected to be.

Yeah...where the liner is buckling.  Loading it warmer is a bandaid (that should work at the cost of some performance) with the actual solution making the COPV liners not buckle.

Unless your system is designed to buckle (like the crumple zone on a car) it never should and any buckling, either large scale or local, should be considered off nominal.

[ChrisWilson68]

Look, obviously they missed something.  It's easy to say, with the benefit of hindsight, that they should have figured out that this could happen.  It really doesn't matter whether what they missed was that buckling could happen, or the effects of the buckling, or some other unexpected condition in conjunction with the buckling.

They were doing something nobody had done before and they didn't think of this failure mode.  It's a failure mode nobody had ever seen before.

I don't know about anybody else, but I personally am not very impressed with armchair engineers criticising the folks at SpaceX based on hindsight.

If they noticed something unexpected, such as buckling, it's reasonable to do an analysis and see if there is any failure mode they can think of that would be caused by that buckling.  If not, then it's not justified to insist on fixing the root cause of the buckling if they don't have any reason to believe the buckling is an issue.  If they had been correct about buckling not being a danger, it would be a waste of resources that could be better spent improving safety in some other way.  Worse, by introducing new complexity to try to avoid the buckling they could inadvertently introduce new failure modes.

From the information we have, it certainly sounds like buckling was only one of the ingredients necessary to cause the failure.  Particular temperatures and other conditions were other necessary ingredients.  If they can be certain of that, then making sure the other conditions don't happen is a perfectly reasonable fix.  There's no reason to denigrate it as a band-aid.  Maybe in the long term it's more advantageous to make another change to remove the cause of the buckling to allow more freedom to other conditions.  Maybe not.  But not knowing the details, it's wrong of any of us to conclude that the "real" fix is to stop the buckling from happening.

[Lee Jay]

If they noticed something unexpected, such as buckling, it's reasonable to do an analysis and see if there is any failure mode they can think of that would be caused by that buckling.

No it's not.

Bucking is a very serious failure.  It's very hard to predict so it's reasonable to be surprised that it's happening, but if you find that it's happening - especially in a flight critical component - it's not reasonable to continue to fly with the knowledge that it's happening despite your expectations that it wouldn't be.

If you find the spar in your airplane has some buckling during testing or test flying, you wouldn't fly that plane again until you had changed the design to eliminate the buckling and replaced the one you were flying.  The results of buckling are too hard to predict to assume that even detailed analyses could predict the results.  It's a non-linear phenomenon and your original detailed analysis didn't predict that it would even happen.

I've seen buckling during high cycle fatigue testing and in single cycle proof loading.  It never ends well.  And it always ends with a design modification.

[CameronD]

......

If they noticed something unexpected, such as buckling, it's reasonable to do an analysis and see if there is any failure mode they can think of that would be caused by that buckling.  If not, then it's not justified to insist on fixing the root cause of the buckling if they don't have any reason to believe the buckling is an issue.  If they had been correct about buckling not being a danger, it would be a waste of resources that could be better spent improving safety in some other way.  Worse, by introducing new complexity to try to avoid the buckling they could inadvertently introduce new failure modes.

From the information we have, it certainly sounds like buckling was only one of the ingredients necessary to cause the failure.  Particular temperatures and other conditions were other necessary ingredients.  If they can be certain of that, then making sure the other conditions don't happen is a perfectly reasonable fix.  There's no reason to denigrate it as a band-aid.  Maybe in the long term it's more advantageous to make another change to remove the cause of the buckling to allow more freedom to other conditions.  Maybe not.  But not knowing the details, it's wrong of any of us to conclude that the "real" fix is to stop the buckling from happening.

As someone who works around the aerospace industry and understands how cost-prohibitive some fixes (especially unnecessary ones) can be... I couldn't have said it better myself.

[meekGee]

If they noticed something unexpected, such as buckling, it's reasonable to do an analysis and see if there is any failure mode they can think of that would be caused by that buckling.

No it's not.

Bucking is a very serious failure.  It's very hard to predict so it's reasonable to be surprised that it's happening, but if you find that it's happening - especially in a flight critical component - it's not reasonable to continue to fly with the knowledge that it's happening despite your expectations that it wouldn't be.

If you find the spar in your airplane has some buckling during testing or test flying, you wouldn't fly that plane again until you had changed the design to eliminate the buckling and replaced the one you were flying.  The results of buckling are too hard to predict to assume that even detailed analyses could predict the results.  It's a non-linear phenomenon and your original detailed analysis didn't predict that it would even happen.

I've seen buckling during high cycle fatigue testing and in single cycle proof loading.  It never ends well.  And it always ends with a design modification.

It's been established already that "buckling" was NOT the simple structural buckling that you describe, that causes failure by developing into macro deformation.

Just consider that the tanks are loaded close to their structural limit.  If the buckling you describe occured, they'd fail right away.  All of them.  Every time.  With no need to any ignition.

They're using "buckling" to describe some micro deformation that was maybe even known about, that enabled failure through a mechanism that nobody considered.

[mn]

When did SpaceX say they saw buckling prior to this failure? I must have missed an update because I don't see that anywhere?

[deltaV]

I've seen buckling during high cycle fatigue testing and in single cycle proof loading.  It never ends well.  And it always ends with a design modification.

This reminds me of the bent gusset plates on the I35W bridge that collapsed a few years ago (https://en.wikipedia.org/wiki/I-35W_Mississippi_River_bridge#Collapse). If that bowing were investigated it could easily have lead to the discovery that the gussets plates were designed too thin, which was the cause of the accident (according to the NTSB). However this is only relevant if buckling had occurred before and SpaceX knew or should have known about it. AFAIK whether that's the case is not publicly known.