Discussion of NASA data review delaying SpaceX launches

[LouScheffer]

We are just debating whether delaying the launch is logically correct. (Because we like to debate lots of things that make almost no difference in the big picture)

If it make no difference

and allows people to cover their asses, then it is logically correct.

Except delaying the Starlink launch doesn't even accomplish this.  Assume Psyche fails, and the Starlink launch could have foretold this  -  the situation where ass-covering is needed.   Then NASA managers will get hauled before Congress, and asked why they delayed the launch of Starlink beyond Psyche, thus ruling out the possibility of even a blatant error (like a second stage failure) being found.

Wrong.  The premise is that the previous launch can’t be reviewed without delaying Psyche.  There is no way to review the previous.

Wrong.  There are errors that can be reviewed in time, such as a second stage failure.

[whitelancer64]

We are just debating whether delaying the launch is logically correct. (Because we like to debate lots of things that make almost no difference in the big picture)

If it make no difference and allows people to cover their asses, then it is logically correct.

Except delaying the Starlink launch doesn't even accomplish this.  Assume Psyche fails, and the Starlink launch could have foretold this  -  the situation where ass-covering is needed.   Then NASA managers will get hauled before Congress, and asked why they delayed the launch of Starlink beyond Psyche, thus ruling out the possibility of even a blatant error (like a second stage failure) being found.

NASA didn't decide to delay the Starlink launch, SpaceX did.

[Jim]

We are just debating whether delaying the launch is logically correct. (Because we like to debate lots of things that make almost no difference in the big picture)

If it make no difference

and allows people to cover their asses, then it is logically correct.

Except delaying the Starlink launch doesn't even accomplish this.  Assume Psyche fails, and the Starlink launch could have foretold this  -  the situation where ass-covering is needed.   Then NASA managers will get hauled before Congress, and asked why they delayed the launch of Starlink beyond Psyche, thus ruling out the possibility of even a blatant error (like a second stage failure) being found.

Wrong.  The premise is that the previous launch can’t be reviewed without delaying Psyche.  There is no way to review the previous.

Wrong.  There are errors that can be reviewed in time, such as a second stage failure.

wrong. That is not the type of issue that a data review is for.  That is an accident/failure investigation and all launches are stopped.

[meekGee]

How is this getting so complicated.

Psyche's timing is fixed.
Starlink can launch either a day before or be delayed and launch after. That's the obly decision to make.

If it launches before, there's no time for a data review.
If it lahnches after, there's no data at all.
Some (major) failures don't require a data review to become apparent.

Therefore:
Delaying the launch of Starlink deprives Psyche of some data, in the relatively rare case of a major fault with Starlink.

So for the sake of some ass covering, a little bit of risk was added to the Psyche mission.

On the bright side, risk to the Starlink mission was reduced by the same small amount.

[LouScheffer]

How is this getting so complicated.

Psyche's timing is fixed.
Starlink can launch either a day before or be delayed and launch after. That's the obly decision to make.

If it launches before, there's no time for a data review.
If it lahnches after, there's no data at all.
Some (major) failures don't require a data review to become apparent.

Therefore:
Delaying the launch of Starlink deprives Psyche of some data, in the relatively rare case of a major fault with Starlink.

So for the sake of some ass covering, a little bit of risk was added to the Psyche mission.

On the bright side, risk to the Starlink mission was reduced by the same small amount.

Agreed.  NASA's end goal is not to do a data review of all previous launches.  NASA's end goal is is increase the likelihood of success on future launches.  Data review is a means to this end, as are accident reviews and stand-downs due to previous launch failures.    Data reviews are for subtle bugs and stand-down for major problems.  Both are there to increase the likelihood of success on later flights.

Incidentally, I think the approach used covered some asses and uncovered others.  If Psyche failed, and it could have been foretold by Starlink, then there will be recriminations.  Middle managers can say "NASA has this longstanding rule, which historically has made excellent sense.  Modern developments have made this rule slightly counterproductive.  But modifying this rule, or obtaining a waiver, is time-consuming and difficult, and deemed not worth the effort for what we thought was a small gain."  But the managers that set the rules will be asked why the rules were not adapted to modern flight rates, even though everyone could see them coming for years. 

[DanClemmensen]

Incidentally, I think the approach used covered some asses and uncovered others.  If Psyche failed, and it could have been foretold by Starlink, then there will be recriminations.  Middle managers can say "NASA has this longstanding rule, which historically has made excellent sense.  Modern developments have made this rule slightly counterproductive.  But modifying this rule, or obtaining a waiver, is time-consuming and difficult, and deemed not worth the effort for what we thought was a small gain."  But the managers that set the rules will be asked why the rules were not adapted to modern flight rates, even though everyone could see them coming for years.

The rule has not changed because there is no incentive within NASA to change it. Those folks are busy, and until this year the rule had no real negative effect. SpaceX needs to add an additional fee to the launch of any mission that includes this rule. The charge will cover the opportunity costs of the potential lost launch slots. This will give NASA an incentive to evaluate the cost  of the rule versus the benefit.The fee should increase automatically as launch cadence goes up.

[abaddon]

Those folks are busy, and until this year the rule had no real negative effect.

Still had no real negative effect.  The amount of crying over a Starlink flight being held until after Psyche launched is absurd.  Launches get held all the time, SpaceX has delayed Starlink launches when a crewed launch is about to go up for example.  Psyche had priority because of its limited launch window.  Starlinks can fly whenever.

[DanClemmensen]

Those folks are busy, and until this year the rule had no real negative effect.

Still had no real negative effect.  The amount of crying over a Starlink flight being held until after Psyche launched is absurd.  Launches get held all the time, SpaceX has delayed Starlink launches when a crewed launch is about to go up for example.  Psyche had priority because of its limited launch window.  Starlinks can fly whenever.

As I understand the rule, the no-launch window affects all F9 launches at all sites. The launch cadence is still accelerating, and will reach 130/yr by the end of 2024 by the most conservative projection. Even if we have not saturated the launch slots quite yet, we are getting close, and delays go up sharply as the system nears saturation. In most systems I know of (e.g., packet switching delays, or your daily commuting route), a system at 50% utilization has no allocation problems (delay) and a system that nears 100% has very serious problems.  Theoretically the "problem" curve is a hyperbola asymptotic to infinity at 100% utilization. The "knee" of the curve depends on the details.

[mn]

Those folks are busy, and until this year the rule had no real negative effect.

Still had no real negative effect.  The amount of crying over a Starlink flight being held until after Psyche launched is absurd.  Launches get held all the time, SpaceX has delayed Starlink launches when a crewed launch is about to go up for example.  Psyche had priority because of its limited launch window.  Starlinks can fly whenever.

The discussion is was never about the slight delay of 6-22.

The discussion is about the 'absurdity' (to use your language) of the 'logic' claiming that the delay is good for the Psyche mission.

When a decision is made and a reason is given and people realize that the reason is illogical it drives them crazy, so we get a thread. (The impact of the decision is completely secondary to the discussion)

Edit: On 2nd thought I removed last paragraph.

[LouScheffer]

Those folks are busy, and until this year the rule had no real negative effect.

Still had no real negative effect.  The amount of crying over a Starlink flight being held until after Psyche launched is absurd.  Launches get held all the time, SpaceX has delayed Starlink launches when a crewed launch is about to go up for example.  Psyche had priority because of its limited launch window.  Starlinks can fly whenever.

I don't think the crying is for the Starlink delay.  The crying is for Psyche, a billion dollar, one of a kind, exceedingly hard to replace mission,  taking an unnecessary risk.  And NASA, for pursuing a course that actively mandates taking that risk.  Fortunately it worked, but we may not be so lucky going forward. 

[RedLineTrain]

Those folks are busy, and until this year the rule had no real negative effect.

Still had no real negative effect.  The amount of crying over a Starlink flight being held until after Psyche launched is absurd.  Launches get held all the time, SpaceX has delayed Starlink launches when a crewed launch is about to go up for example.  Psyche had priority because of its limited launch window.  Starlinks can fly whenever.

I don't think the crying is for the Starlink delay.  The crying is for Psyche, a billion dollar, one of a kind, exceedingly hard to replace mission,  taking an unnecessary risk.  And NASA, for pursuing a course that actively mandates taking that risk.  Fortunately it worked, but we may not be so lucky going forward.

I think the larger question begged is whether LSP is earning its keep.  Why does NASA need to review data when SpaceX has already done so, on a rocket that has been exceedingly well characterized?  The value of a re-review sure seems to be diminishing quickly.

[JayWee]

Btw, does "review of previous flight" include the previous flight of the reused boosters?

[Barley]

the launch vehicles are not static.  These are not the same design as a year ago.  Every upperstage is new.

The solution to this is version control.  You should not assume that chronological order of the launch implies version number of the hardware.

For example if Psyche is using upper stage serial number 1002 you can launch serial numbers 1001 and 1003 and use those as your baseline.  (Assuming 1001-1003 are identically configured, if they are not rearrange things so they are.)

Similarly for boosters and launch pad procedures.

[abaddon]

The crying is for Psyche, a billion dollar, one of a kind, exceedingly hard to replace mission,  taking an unnecessary risk. 

Funny.  But ridiculous.

I'm out.  Everyone else, have fun.

[steveleach]

the launch vehicles are not static.  These are not the same design as a year ago.  Every upperstage is new.

The solution to this is version control.  You should not assume that chronological order of the launch implies version number of the hardware.

For example if Psyche is using upper stage serial number 1002 you can launch serial numbers 1001 and 1003 and use those as your baseline.  (Assuming 1001-1003 are identically configured, if they are not rearrange things so they are.)

Similarly for boosters and launch pad procedures.

Coming up with a solution isn't the issue, tbh; recognising that there's a problem is.

[mn]

The crying is for Psyche, a billion dollar, one of a kind, exceedingly hard to replace mission,  taking an unnecessary risk. 

Funny.  But ridiculous.

I'm out.  Everyone else, have fun.

Instead of leaving, why don't you try to explain...

[Jim]

I think the larger question begged is whether LSP is earning its keep.  Why does NASA need to review data when SpaceX has already done so, on a rocket that has been exceedingly well characterized?  The value of a re-review sure seems to be diminishing quickly.

how do you know that SpaceX already has?  It isn't for "characterization"

[Jim]

the launch vehicles are not static.  These are not the same design as a year ago.  Every upperstage is new.

The solution to this is version control.  You should not assume that chronological order of the launch implies version number of the hardware.

For example if Psyche is using upper stage serial number 1002 you can launch serial numbers 1001 and 1003 and use those as your baseline.  (Assuming 1001-1003 are identically configured, if they are not rearrange things so they are.)

Similarly for boosters and launch pad procedures.

They still aren't identical

[Robotbeat]

[Edit to add context from Psyche mission thread - SpaceX originally delayed the F9 Starlink 6-22 launch because there would not have been enough time for NASA to analyse the launch data prior to the following FH Psyche launch]

I find it unfortunate that they have to analyse every prior Falcon launch. Of course I understand why (if there was an indicator of a possible issue that wasn’t analysed and anything ever happened there’d be hell to pay).

But they have way more data on Falcon than any other launch vehicle NASA uses. They have also analysed all prior FH launches. It saddens me that data from 71 prior launches this year wouldn’t be enough and it would have to be 72 if Starlink 6-22 launches.

Yeah, it’s not rational because delaying the Starlink 6-22 launch provides zero improvement to launch risk (in fact it increases it) but it does solve schedule risk due to procedures not always lining up 1:1 to the most Pareto-optimal risk analysis. And it just delayed for a few days so who cares, problem solved.

Alright let’s put this to bed. Thread closed, problem solved LOL

[Barley]

the launch vehicles are not static.  These are not the same design as a year ago.  Every upperstage is new.

The solution to this is version control.  You should not assume that chronological order of the launch implies version number of the hardware.

For example if Psyche is using upper stage serial number 1002 you can launch serial numbers 1001 and 1003 and use those as your baseline.  (Assuming 1001-1003 are identically configured, if they are not rearrange things so they are.)

Similarly for boosters and launch pad procedures.

They still aren't identical

The chronological predecessor is also not identical.