SpaceX F9 / Crew Dragon : Crew-2 : 22 April 2021 - DISCUSSION

[eeergo]

So first off, I ain't no SpaceX amazing people so my post was nothing to do with defending SpaceX but more about getting people to drop the easy one-liners and instead back up their opinion (right or wrong) with their arguments and let everyone debate, support or refute.

Your post quote here is substantive and lays our your position and view and that's great. The other post not so much.

Now back to the cutes - This interaction of 4 chutes is well known and also understood. So long as there was no external damage that caused the slow opening it appears that the delayed opening is not a safety issue. SpaceX brought the chute back to NASA, hoisted it on a crane and did an inspection to check for just this.

The alternative is to go back to 3 chutes where this problem will not happen but then you lose the added bonus of a 4th backup chute which still works even with a delayed opening.

Apologies for the qualifier then, but "FUD" is a well-known derogatory umbrella term in the Musk-verse for anything that is perceived to go against his interests. Glad to know it wasn't used in that vein.

There can be many mitigations tried that do not involve reducing the factor of safety by playing Salomon and deleting a chute from the design (that 4th chute was put in place for a reason, not just as a bout of "the more the merrier"). See the post above this one.

[mn]

Some thoughts.

It is 'understood' that the issue is caused by having 4 chutes which can cause one chute to lag.

If that understanding is correct then everything stands, whenever the problem occurs you are down to 3 (even if the 4th never recovers) and all is well.

The worry is that we may discover in the future that this 'understanding' was wrong and the problem is not due to having 4 chutes but rather with the chute design which can cause one (or more) chutes to fail randomly at any time.

If that hypothetical is true, going back to 3 chutes would be the worst possible solution. You are better off with 4 so if the problem occurs you are down to 3 (or 2 if the problem occurs on 2 chutes on the same bad day).

Now there is another hypothetical possibility: What if having 4 chutes can cause 2 to fail while having 3 chutes does not have that particular failure mode. I guess we can always worry but you can't just switch to 3 because of this worry when the other possibilities are worse with 3 chutes and considered more likely to be correct.

[Lee Jay]

but rather to point out that when something happens that's not as-planned,

Or not as expected, we're accustomed to NASA hunkering down, putting everything on hold, and paralyzing by analyzing.

What a happy surprise that, in the current case, they didn't overreact, but came right out and said that it's been modeled and seen before and no big deal, let's launch the next one.

Which is what they said about seal burn through and foam liberation as well.

Right so instead of a FUD post why don't you impart your wisdom and let us all know what they should be doing?  They have done 30 tests so is it more testing that (in your opinion) should be done? if so how many tests are needed? 40, 50 or maybe 100 tests? Love to know what your magic number is?

Or maybe you think Dragon isn't safe at all. So, in your opinion, should it be grounded? And if so then what? A new magic chute design that doesn't behave this way. Again love to know of your wisdom on this.

FUD posts are easy. Coming up with solutions or real fixes are hard.

I'm not a parachute expert so it's not my job to come up with a fix.  However, just being successful on 30 tests is not sufficient.  Shuttle had 112 flights, many with foam liberation, before the 113th killed 7 people and destroyed an orbiter.  If there's a rationale for the idea that one is far more likely to happen than two at the same time (because of them interacting with each other, for example) and you can show a satisfactory statistical likelihood that it's safe enough to meet your requirements for safety as it is, then fine, show that.  But don't just accept something like, "well we've done 30 tests and it never happened to more than one and we've done a few operational flights and it only happened once, so it must be okay."  That's not okay.  "It hasn't happened before" is not evidence that "it will never happen in the future".

[Vettedrmr]

You imply that those 30 tests were all the same.  They're not.  They test different failure modes, multiple failure modes, etc. through that battery of tests.  IIRC the *final* test was a full up no-failure test to demonstrate normal operations, only because they hadn't done one since the beginning of the test phase.

[Silmfeanor]

I'm not a parachute expert so it's not my job to come up with a fix. 

Who says there should be a fix? The argument is whether this is normal or not. SpaceX might have done more then double the required tests and simulations, fully to NASA's safety boards satisfaction. Why do we think this needs fixing?

  However, just being successful on 30 tests is not sufficient.  Shuttle had 112 flights, many with foam liberation, before the 113th killed 7 people and destroyed an orbiter.

Why wouldn't 30 tests be sufficient? Confidence in the system has a cut-off at some point, and if the confidence was reached by doing 30 tests....do we have any indication by anyone with insight that there should have been/be more tests?

If there's a rationale for the idea that one is far more likely to happen than two at the same time (because of them interacting with each other, for example) and you can show a satisfactory statistical likelihood that it's safe enough to meet your requirements for safety as it is, then fine, show that.

Since NASA and safety boards have accepted SpaceX's solution and testing history, including an occurence of this behaviour during these 30 flights, do we know that this hasnt been shown?

But don't just accept something like, "well we've done 30 tests and it never happened to more than one and we've done a few operational flights and it only happened once, so it must be okay."  That's not okay.  "It hasn't happened before" is not evidence that "it will never happen in the future".

I am sure that the certification board didn't do this, as such it is a straw man.

I mean, let me ask - do we have any indication that there is actually a problem?

[DanClemmensen]

but rather to point out that when something happens that's not as-planned,

Or not as expected, we're accustomed to NASA hunkering down, putting everything on hold, and paralyzing by analyzing.

What a happy surprise that, in the current case, they didn't overreact, but came right out and said that it's been modeled and seen before and no big deal, let's launch the next one.

Which is what they said about seal burn through and foam liberation as well.

Right so instead of a FUD post why don't you impart your wisdom and let us all know what they should be doing?  They have done 30 tests so is it more testing that (in your opinion) should be done? if so how many tests are needed? 40, 50 or maybe 100 tests? Love to know what your magic number is?

Or maybe you think Dragon isn't safe at all. So, in your opinion, should it be grounded? And if so then what? A new magic chute design that doesn't behave this way. Again love to know of your wisdom on this.

FUD posts are easy. Coming up with solutions or real fixes are hard.

I'm not a parachute expert so it's not my job to come up with a fix. 

Since you are not a parachute expert, it's also not your job to decide that any fix is needed. 

[Lee Jay]

but rather to point out that when something happens that's not as-planned,

Or not as expected, we're accustomed to NASA hunkering down, putting everything on hold, and paralyzing by analyzing.

What a happy surprise that, in the current case, they didn't overreact, but came right out and said that it's been modeled and seen before and no big deal, let's launch the next one.

Which is what they said about seal burn through and foam liberation as well.

Right so instead of a FUD post why don't you impart your wisdom and let us all know what they should be doing?  They have done 30 tests so is it more testing that (in your opinion) should be done? if so how many tests are needed? 40, 50 or maybe 100 tests? Love to know what your magic number is?

Or maybe you think Dragon isn't safe at all. So, in your opinion, should it be grounded? And if so then what? A new magic chute design that doesn't behave this way. Again love to know of your wisdom on this.

FUD posts are easy. Coming up with solutions or real fixes are hard.

I'm not a parachute expert so it's not my job to come up with a fix. 

Since you are not a parachute expert, it's also not your job to decide that any fix is needed. 

I'm not deciding a fix is needed.  I am of the opinion that an investigation and thorough review is needed precisely for the reason that Wayne Hale said.

[Vettedrmr]

Why do you think that didn't happen?

[Lee Jay]

Why do you think that didn't happen?

Unless I've done the math wrong, crew-2 landing and crew-3 launch were 47 hours apart.  You really think that's enough time to do an "an investigation and thorough review"?

[DanClemmensen]

but rather to point out that when something happens that's not as-planned,

Or not as expected, we're accustomed to NASA hunkering down, putting everything on hold, and paralyzing by analyzing.

What a happy surprise that, in the current case, they didn't overreact, but came right out and said that it's been modeled and seen before and no big deal, let's launch the next one.

Which is what they said about seal burn through and foam liberation as well.

Right so instead of a FUD post why don't you impart your wisdom and let us all know what they should be doing?  They have done 30 tests so is it more testing that (in your opinion) should be done? if so how many tests are needed? 40, 50 or maybe 100 tests? Love to know what your magic number is?

Or maybe you think Dragon isn't safe at all. So, in your opinion, should it be grounded? And if so then what? A new magic chute design that doesn't behave this way. Again love to know of your wisdom on this.

FUD posts are easy. Coming up with solutions or real fixes are hard.

I'm not a parachute expert so it's not my job to come up with a fix. 

Since you are not a parachute expert, it's also not your job to decide that any fix is needed. 

I'm not deciding a fix is needed.  I am of the opinion that an investigation and thorough review is needed precisely for the reason that Wayne Hale said.

I am absolutely not a parachute expert either, so I don't know how much effort "an investigation and thorough review" should take. I do know that a expert can in many cases do "an investigation and thorough review" in less than a minute when a phenomenon is extremely well understood. In many cases an experienced MD can perform a highly accurate diagnosis of a patient within five seconds of walking into a treatment room. I have no reason to believe that this is not the case here.

[Lee Jay]

I am absolutely not a parachute expert either, so I don't know how much effort "an investigation and thorough review" should take. I do know that a expert can in many cases do "an investigation and thorough review" in less than a minute when a phenomenon is extremely well understood. In many cases an experienced MD can perform a highly accurate diagnosis of a patient within five seconds of walking into a treatment room. I have no reason to believe that this is not the case here.

I've done so many such investigations that I've lost count.  I'm in the middle of one right now.  All of us doing these are experts in the field in question (not rocketry or parachutes).  I've never seen one take less than a month, and I have seen them take as long as 18 months.  I'd say the average is in the 3-6 months range.  In most cases, root cause can be found in a few days, but it can take a while to conclusively prove that determination and another while to develop and implement a successful mitigation plan.

[Herb Schaltegger]

I'm not a parachute expert so it's not my job to come up with a fix.  However, just being successful on 30 tests is not sufficient.

Apparently you’re not an expert in irony either.

You’re admittedly not an expert, yet claim that 30 tests is not sufficient. These two statements are logically incompatible. The fact that you may not see it that way does not change that fact.

[thirtyone]

I think the difference between this and shuttle "out of family" incidents may be subtle, but substantial. I was very interested in parachute design for some time while both Starliner and Crew Dragon were having parachute issues, and ended up diving quite deep into various bits of their parachute design. My impression was that lead-lag on these parachutes is not only an understood phenomenon in the field but something that was encountered, documented, investigated, and understood for this design. It sounds like they launched again because they had criteria already set in place which covered the scenario and everything was within expectation (investigate if the slow inflation was due to allowable and expected aerodynamic conditions and not due to parachute damage). Not a "this is the worst bunching we've ever seen, the parachute lines nearly broke apart, and we have to launch anyway."

I guess what I'm thinking is this may be analogous to insisting on an investigation into a phenomenon that has already been fully investigated in advance. It sounds like the only people this effect worried is the public and not the actual engineers.

I should mention that there *are* things that have happened to Crew Dragon in the past that were clearly problematic (NOT within what engineering expected) and I believe went through a thorough investigation. They were just not as public (and frankly should have been more worrisome, if they hadn't been thoroughly investigated). In Crew-1 or Demo-2 (can't quite remember) there were heat tiles near some of the attach points that had ablated more than expected and ate into safety margins. I don't think it came out until immediately before the following launch, but they had apparently gone through a full review on the issue and a change was made to reduce ablation. For scale, it is my opinion that excess ablation of the heatshield beyond determined safety margins is a considerably more severe of an issue than a parachute inflating slowly that had already been documented as a safe and expected condition.

One of the purposes of me mentioning this is that the only reason everyone is so worried about this on a public forum (and why NASA probably even brought it up) was that the effect was incredibly obvious on the livestream. That does not mean it is actually incredibly unsafe condition. The excess heatshield ablation was not visible on a livestream yet was considerably more dangerous, and NASA did not publicly mention it until prior to the following launch.

[Lee Jay]

I'm not a parachute expert so it's not my job to come up with a fix.  However, just being successful on 30 tests is not sufficient.

Apparently you’re not an expert in irony either.

You’re admittedly not an expert, yet claim that 30 tests is not sufficient. These two statements are logically incompatible.

No they aren't.  The point is, showing that something (anything) works properly 30 times in a row is not sufficient to determine that it will work properly every time.  I gave an example - Shuttle foam liberation caused less-than-catastrophic damage 112 times, and catastrophic damage on the 113th time.  This isn't about parachutes, it's about the idea that running a small number of tests is sufficient to demonstrate safety - it's not.  It's a part of the process that could be used to demonstrate safety.

[thirtyone]

I can probably find more in some previous posts, but these sorts of effects are described in quite a few papers on modern parachute design. Almost all of the public ones focus primarily on Orion, which has very similar issues due to a high count of large parachutes. In fact I believe the need for four parachutes directly comes from the fact that you can have slow (lagging) parachutes and fast (leading) parachutes along with the loads imparted by Crew Dragon. The variability in loading results in a need to add a parachute because it is occasionally possible that 1/3 parachutes inflates slowly and the variability in drag due to those atmospheric conditions can result in some edge cases with marginal safety limits. So they added a fourth, with similar lead-lag effects (can occasionally have 1/4 inflate slowly), but with overall better margins.

Random related article that was interesting:
https://ntrs.nasa.gov/api/citations/20130011421/downloads/20130011421.pdf

Lots more related ones mostly published under AIAA

Oh and FYI, lead-lag as described by that engineer is not the same as lead-lag in parachutes. They are literally talking about leading and lagging (fast and slow) parachutes. In feedback loops/electronics (which might be why he wanted to call it that) it's much harder to explain but it's a technique to help keep feedback loops stable.

[thirtyone]

I'm not a parachute expert so it's not my job to come up with a fix.  However, just being successful on 30 tests is not sufficient.

Apparently you’re not an expert in irony either.

You’re admittedly not an expert, yet claim that 30 tests is not sufficient. These two statements are logically incompatible.

No they aren't.  The point is, showing that something (anything) works properly 30 times in a row is not sufficient to determine that it will work properly every time.  I gave an example - Shuttle foam liberation caused less-than-catastrophic damage 112 times, and catastrophic damage on the 113th time.  This isn't about parachutes, it's about the idea that running a small number of tests is sufficient to demonstrate safety - it's not.  It's a part of the process that could be used to demonstrate safety.

Admittedly a bit off topic, but I agree, and I wish we get to the day where we *can* run the test 200 times before putting people on it. Personally I think it's not even sufficiently safe to rely on processes (there is literally no way you can mathematically link a process to a safety reduction - it's really still a guess that doing it one way can improve safety), even though they can improve your odds that your safety factors what you think it is. Best way to know your failure rate in 200 launches is to launch 200 times.

For what it's worth, at least those 30 times weren't with humans at risk. Wish they did more of course. In fact I wish they could've launched Crew Dragon 200 times with dummies before putting humans on board, but I know that's not practical.

[abaddon]

I'm not a parachute expert so it's not my job to come up with a fix.  However, just being successful on 30 tests is not sufficient.

Apparently you’re not an expert in irony either.

You’re admittedly not an expert, yet claim that 30 tests is not sufficient. These two statements are logically incompatible.

No they aren't.  The point is, showing that something (anything) works properly 30 times in a row is not sufficient to determine that it will work properly every time.  I gave an example - Shuttle foam liberation caused less-than-catastrophic damage 112 times, and catastrophic damage on the 113th time.  This isn't about parachutes, it's about the idea that running a small number of tests is sufficient to demonstrate safety - it's not.  It's a part of the process that could be used to demonstrate safety.

....and yet I'm sure you're all on board the Starliner in-flight abort which was demonstrated a total of zero times.  Oh, right, they relied on the (tested once) pad abort and computer modeling.  You have heard of computer modeling, haven't you?  I understand that parachute validation, not being done before the invention of computers, does in fact use computer modeling as well as physical tests to validate the model.

Herb was right, and you're wrong.  You're not educated to weigh in on whether 30 tests were sufficient or not.  Please give it a rest.

[alugobi]

They went ahead and launched the next crew.  Concern over.

[abaddon]

I'm not a parachute expert so it's not my job to come up with a fix.  However, just being successful on 30 tests is not sufficient.

Apparently you’re not an expert in irony either.

You’re admittedly not an expert, yet claim that 30 tests is not sufficient. These two statements are logically incompatible.

No they aren't.  The point is, showing that something (anything) works properly 30 times in a row is not sufficient to determine that it will work properly every time.  I gave an example - Shuttle foam liberation caused less-than-catastrophic damage 112 times, and catastrophic damage on the 113th time.  This isn't about parachutes, it's about the idea that running a small number of tests is sufficient to demonstrate safety - it's not.  It's a part of the process that could be used to demonstrate safety.

Admittedly a bit off topic, but I agree, and I wish we get to the day where we *can* run the test 200 times before putting people on it.

Ironically, given all the of the lamenting by some of these same folks about how unsafe Starship is going to be, Starship is likely going to be the first vehicle that will approach that amount of testing before flight with humans on board.  Starship is firmly off-topic for this thread, so I will leave it at that.

[kevinof]

Right so let's test Starliner and Orion also with 200 launches before we are happy to have crew on them.

Would never work and that's why they do a limited (if 30 is limited) tests and model the hell out of it.

They way I look at it is who has the most to lose (apart from the crew) in the event of an accident? It would be Spacex and the public and politicians would crush them.

Doubt they would ever fly if they were not 100% confident in their vehicle.

I'm not a parachute expert so it's not my job to come up with a fix.  However, just being successful on 30 tests is not sufficient.

Apparently you’re not an expert in irony either.

You’re admittedly not an expert, yet claim that 30 tests is not sufficient. These two statements are logically incompatible.

No they aren't.  The point is, showing that something (anything) works properly 30 times in a row is not sufficient to determine that it will work properly every time.  I gave an example - Shuttle foam liberation caused less-than-catastrophic damage 112 times, and catastrophic damage on the 113th time.  This isn't about parachutes, it's about the idea that running a small number of tests is sufficient to demonstrate safety - it's not.  It's a part of the process that could be used to demonstrate safety.

Admittedly a bit off topic, but I agree, and I wish we get to the day where we *can* run the test 200 times before putting people on it. Personally I think it's not even sufficiently safe to rely on processes (there is literally no way you can mathematically link a process to a safety reduction - it's really still a guess that doing it one way can improve safety), even though they can improve your odds that your safety factors what you think it is. Best way to know your failure rate in 200 launches is to launch 200 times.

For what it's worth, at least those 30 times weren't with humans at risk. Wish they did more of course. In fact I wish they could've launched Crew Dragon 200 times with dummies before putting humans on board, but I know that's not practical.