SpaceX F9 : Starlink v1.0 L19 : CCSFS SLC-40 : 15/16 Feb 2021 (0359 UTC)

[Lee Jay]

Ground cameras were not shown at the time the derived TM shows a large dip (last 15 seconds, check out the archived webcast),

Seems like we do:

so there's no information to claim no shutdown plumes were observed. Let me know if there are any alternative high quality amateur videos that I may have missed and show no change in plume just before MECO.

Hard to tell:



I don't see evidence of an engine shutdown in either video.

[edkyle99]

So how do you get a life-leader boot on a non-life-leader booster?

One possibility is that this booster did more RTLS flights than any other booster, if I'm remembering correctly.  The boost-back burns for these missions are long, so total engine burn time could have accumulated faster on some of these engines.  Just a guess.

 - Ed Kyle

[PreferToLurk]

So how do you get a life-leader boot on a non-life-leader booster?

One possibility is that this booster did more RTLS flights than any other booster, if I'm remembering correctly.  The boost-back burns for these missions are long, so total engine burn time could have accumulated faster on some of these engines.  Just a guess.

 - Ed Kyle

It was only specified to be a life leader "for that design". Total life leader booster must use a different boot design.

[leetdan]

So how do you get a life-leader boot on a non-life-leader booster?

One possibility is that this booster did more RTLS flights than any other booster, if I'm remembering correctly.  The boost-back burns for these missions are long, so total engine burn time could have accumulated faster on some of these engines.  Just a guess.

 - Ed Kyle

https://en.wikipedia.org/wiki/List_of_Falcon_9_first-stage_boosters

This booster had three RTLS boostback burns, more than any other.  The next closest are the supposedly retired Block 5 FH side boosters with two each. Boostback burns are themselves about twice as long as a typical entry burn.  This means 1059 spent more time than any other in this flight regime, so in hindsight an engine boot failure isn't very hard to imagine.

[Tangilinear Interjar]

Falcon 9 B1059.6 landing failure update. A Merlin engine boot (a life leader) developed a hole and sent hot gas to "where it wasn't supposed to be" and shut down during first stage flight. Not enough thrust for landing.

So how do you get a life-leader boot on a non-life-leader booster?

The only boosters with more flights are still around, I believe.  So it would be odd for a boot to be swapped off of one of them.

Maybe all others boosters and engines had their boots replaced before flight 6?

Maybe boots are in one big stockpile (they might have to remove them for refurbishing).  So after refurbishing, you go grab 9 boots from the stockpile, and they happened to get a well-used one?

Curious minds want to know....

You’ve answered your own question, I think. Boots don’t necessarily stay with their original engine or booster. They are (possibly? probably?) considered wear items and subject to replacement if needed. However, learning where that “if needed” point falls on the spectrum is obviously tricky.

It's entirely possible that this boot has flown well over a dozen times. If they are removed after each flight, inspected and refurbished prior to reinstallation they could just be put on the next flight, regardless of which booster that is.

[gongora]

There is unfounded speculation that the engine failed on ascent - based solely on a comment that the booster has engine out capability and that the payload made it to orbit.  Evidently,  pointing out that the mission was a success is an indication of failure?

If the engine didn't fail on ascent then Benji Reed's wording was extremely poor.  Benji also didn't answer a follow-up question about the timing of the shutdown.

[Vettedrmr]

If the engine didn't fail on ascent then Benji Reed's wording was extremely poor.  Benji also didn't answer a follow-up question about the timing of the shutdown.

Agreed.  It clarified one thing nicely (the point of failure), and garbled the rest of it badly.

[eeergo]

Ground cameras were not shown at the time the derived TM shows a large dip (last 15 seconds, check out the archived webcast),

Seems like we do:

so there's no information to claim no shutdown plumes were observed. Let me know if there are any alternative high quality amateur videos that I may have missed and show no change in plume just before MECO.

Hard to tell:

I don't see evidence of an engine shutdown in either video.

I meant to say onboard video rather than ground. The jittering just before MECO made me think at the time it was almost impossible to look for something like that - but rewatching it more carefully I actually think *something* is visible at precisely the timestamp the pre-MECO bucket happens when the camera hasn't started trembling too much yet.

Have a look right after the 2:25->2:26 transition, in the official video counter. The camera jitters for a moment and when it stabilizes there's a sudden increase in downwind plume visibility for 2-3 seconds just after an exhaust stripe brightening, which goes away and doesn't return. The rocket is too high for any kind of atmospheric contrailing effect to be the cause. I'm attaching some screenshots, but it's more apparent in motion. Far from conclusive with the available quality, but it well can be a single-engine shutdown.

The other VLR video also appears to show downwind plume enhancement around 12 seconds before MECO (10:45-10:47 timestamp), but there is too much noise in the zoomed image to attempt to extract much detail at a finer scale, and the picture is wobbling near the top edge so not much plume is visible.

[PreferToLurk]

There is unfounded speculation that the engine failed on ascent - based solely on a comment that the booster has engine out capability and that the payload made it to orbit.  Evidently,  pointing out that the mission was a success is an indication of failure?

If the engine didn't fail on ascent then Benji Reed's wording was extremely poor.  Benji also didn't answer a follow-up question about the timing of the shutdown.

That is interesting.  I did not see the raw presser. I just read the article which makes the logical leap without fully explaining why.  Thanks for the context.

[Joffan]

Video of the presser:


Benji's answer starts at about 53 minutes, direct link to that portion: https://youtube.com/watch?v=9I8lvxoljvw&t=3150

[cscott]

So if the boot was compromised then this would really show up on the way back when the exhaust would be flowing back into the stage.

So maybe they did relight on the way back but the plume impacted the innards and that's where we saw the "flames"

Perhaps an engine was shutdown late in ascent, but that engine is required for landing so it was relit for reentry despite the sensor telling them there's an issue (nothing to lose by trying), and then the engine bay apparently caught fire during the reentry burn.

An alternative which might also explain the confused wording of the presser (a little):

Perhaps the failure was late in ascent, but affected a *non landing engine*.  Thus there would be no reason not to relight the three landing engines for the re-entry burn, etc.

However, the hole in the boot continued to allow hot gases to eat away at the failed engine compartment, eventually leading to a cascading failure of some kind which ultimately doomed the landing. (Perhaps causing one of the landing engines to additionally fail, but also perhaps just aero instability etc.)

This would explain why the initial part of the re-entry burn looked nominal, even though the failure occurred at the tail end of the ascent burn.

[CorvusCorax]

Any thrust deviation on ascent can be compensated for the achieved orbit by burning the other engines longer (you still reach the same altitude and speed, however a little bit more downrange) any such thrust deviation stretches and flattens the trajectory. The booster will invariably also come down further downrange than planned.

F9 has some limited capability to compensate trajectory deviations aerodynamically by altering the angle of attack during descent. But at apex, F9 is going more than 2km/s and on Starlink flights there is no boostback burn which could correct for that. I don't think you need a lot of deviation to overshoot to the point where F9 can no longer reach the droneship.

There were some hints between Benji's lines (words that could be interpreted either way) that makes it sound like the booster failed the landing ultimatively because it was no longer able to reach the droneship, not because of the damages it had incurred.

So while speculation about catastrophic heat damage accumulating during reentry is fair enough, (telemetry did cut out on the stream) there are other factors (plasma radio cutout, etc...) so we cannot rule out that the booster was still alive, just a tiny bit too far downrange to get a landing solution.

On the one other reentry of a booster with ascent engine failure, we saw the stage come back into the atmosphere with quite an extreme angle of attack - maybe in an attempt to still get to the droneship. On L19 it was too dark to see the AoA directly, but the one-sided "fire effect" we saw after the reentry burn might also be explainable by a high AoA entry and lopsided heating. It doesn't necessarily mean that the engine- or engine bay was on fire.

We just don't have enough data to tell.  But SpaceX does

[edkyle99]

Any thrust deviation on ascent can be compensated for the achieved orbit by burning the other engines longer (you still reach the same altitude and speed, however a little bit more downrange) any such thrust deviation stretches and flattens the trajectory. The booster will invariably also come down further downrange than planned.

Couldn't the booster also increase the thrust of its remaining engines?  Since it is already in a throttled-down condition, might it be possible that total burn time in this specific situation (very late shutdown) might not be extended?

 - Ed Kyle

[edzieba]

Stage could not have overshot the droneship too far: impact was still visible from the camera (albeit off screen).

[Craftyatom]

Any thrust deviation on ascent can be compensated for the achieved orbit by burning the other engines longer (you still reach the same altitude and speed, however a little bit more downrange) any such thrust deviation stretches and flattens the trajectory. The booster will invariably also come down further downrange than planned.

F9 has some limited capability to compensate trajectory deviations aerodynamically by altering the angle of attack during descent. But at apex, F9 is going more than 2km/s and on Starlink flights there is no boostback burn which could correct for that. I don't think you need a lot of deviation to overshoot to the point where F9 can no longer reach the droneship.

It is true that you can compensate for a loss of thrust by burning for longer.  However, keep in mind that you can also compensate for a loss of thrust by increasing the thrust of the remaining engines.  Since the Falcon 9 usually throttles its engines down near the end of its first-stage burn to limit g-loading, if an engine fails during this period, other engines can simply not throttle down as much to compensate, resulting in the same amount of thrust and the same g-loading.  In fact, since the throttle-down is about 1/9th, shutting down one engine and keeping the others at full throttle has almost exactly the same effect.

The strange thing is that the telemetry we have doesn't fit with any such hypothesis very well.  On flight 18, MECO was at 158 seconds at 7955kph.  On flight 19, MECO was at 158 seconds at 7929kph.  Both vehicles reached the same speed (within .4%) at the same time (within a second), implying that there shouldn't be a significant difference in their downrange distance at the time, given that they followed roughly similar flight profiles.  And in fact, flight 19 actually produced less thrust in its terminal guidance phase, because it was already running hot.  At T+145 seconds, flight 18 was at 6574kph, while flight 19 was at 6699kph - 2% faster, or almost 1 whole second of thrust extra.  And this wasn't the result of a flatter trajectory, because at MECO flight 19 was actually 1km higher than flight 18.

So in my mind, either SpaceX got supremely lucky and had an engine fail at throttle-down on a flight that was running slightly hot, resulting in MECO coincidentally being at exactly the same time... or the rocket did exactly what it's supposed to do, and throttled its engines down more in order to reach the expected time/speed at MECO.  Not to say it couldn't've impacted the failure - that extra altitude or lower throttle might've contributed to the engine failure - but I'd consider the telemetry we have to be evidence against the possibility of an engine failure on ascent.

In my mind, Benji Reed is doing what the Director of Human Spaceflight Programs should do: explaining that, even if this failure had occurred during the phase of flight important to the crew, redundancy in the system would've allowed them to continue with the mission.

For reference: I measured T+145 seconds as the first frame in each webcast where the mission clock read T+00:02:25.  I measured MECO as the frame during first-stage flight in each webcast where velocity was highest.

Stage could not have overshot the droneship too far: impact was still visible from the camera (albeit off screen).

This does put a limit on how far away it could be - it'd be visible out to at least 10km (possibly more if backlighting clouds), but given that we can see it for more than a split second, it's likely much closer, even given the cloud cover.  And (in some of the strangest math I've ever done) 25 frames pass between the sudden extinction of the light and the first visible reaction of all three birds on the drone ship.  That's 0.83 seconds, which for sound at sea level is 286 meters (assuming instantaneous reaction time).  If we instead consider the birds to be startled by landing burn startup, which might coincide with the start of the faint orange glow, about 10 seconds pass, or ~3.5km (as the crow flies), meaning the booster really did come quite close - probably within 2.5km horizontally.

[CorvusCorax]

Any thrust deviation on ascent can be compensated for the achieved orbit by burning the other engines longer (you still reach the same altitude and speed, however a little bit more downrange) any such thrust deviation stretches and flattens the trajectory. The booster will invariably also come down further downrange than planned.

Couldn't the booster also increase the thrust of its remaining engines?  Since it is already in a throttled-down condition, might it be possible that total burn time in this specific situation (very late shutdown) might not be extended?

 - Ed Kyle

I think in that flight stage it could have, but

1. This compensation takes time, so some trajectory deviation would be unavoidable - probably not enough to prevent landing though.
2. the telemetry plot suggests this was not actually done and the acceleration before MECO1 was reduced compared to the previous starship flight
(we don't know for sure how SpaceX has programmed this. My hunch would be - if anything goes wrong, the prime directive is to safe the primary mission and make orbit - and if this is possible without any change to the thrust of the remaining engines, just by burning a bit longer and eating into propellant margins - this might be the way to go - considering this has to work for any shutdown regardless of cause, and if something is seriously wrong, aggressive up-throtteling of other engines might make things more wrong - never change a running system)

We also have seen that the deceleration during entry was not as big as it was supposed to be - that of course would have thrown the stage further off course (it would have been faster, lower - and as such might have had less altitude left for course corrections)

Stage could not have overshot the droneship too far: impact was still visible from the camera (albeit off screen).

we saw some bright light, but Falcon9's shine is visible for many miles at night over open ocean.  That rocket turns the night to day - could have been 2 km away, or 5, or even 10 - hard to say with the rocket not on screen. Someone did a calculation on the time delay between flash and seagulls-getting-startled, but I don't remember the numbers

[the_other_Doug]

An alternative which might also explain the confused wording of the presser (a little):

Perhaps the failure was late in ascent, but affected a *non landing engine*.  Thus there would be no reason not to relight the three landing engines for the re-entry burn, etc.

However, the hole in the boot continued to allow hot gases to eat away at the failed engine compartment, eventually leading to a cascading failure of some kind which ultimately doomed the landing. (Perhaps causing one of the landing engines to additionally fail, but also perhaps just aero instability etc.)

This would explain why the initial part of the re-entry burn looked nominal, even though the failure occurred at the tail end of the ascent burn.

There is something of a false assumption, here.  Y'all make it sound like earnest young SpaceX engineers are looking at their consoles, seeing that one engine is failing, and so they shut it down smoothly.  Then, figuring they have nothing to lose, they decide to relight the engine they shut down.

In reality (and correct me if I'm wrong, but I'm pretty sure I'm not), all the ground crew does is watch.  The onboard navigation and control computer systems respond to out-of-family sensor readings, shut down engines, etc.  And there is no real way for the ground "controllers" to actually update what the Falcon 9 booster is doing at any time during flight.  They don't tell the booster whether or not to proceed with its next steps, it's all automatic.

So, let's suppose the flight control system had shut down one of the three engines capable of relight.  It doesn't remember that it had shut an engine down; at the start of the entry burn, it just knows to light the engines in the specified sequence.  Then, later, to light the correct engines for the landing burn.  It doesn't have a choice, it's going to do it whether the engines are gonna work properly or not.

What we saw was what looked like a proper entry burn initiation, but it did look to sort of feather well before the end of the 3-engine burn, and there was that steady, heavy stream of bright fluid or gas venting from the center of the image off to the right.

I would guess that one of the relit engines was the one with the hole in its boot and had been shut down when it started to flutter late during ascent.  It was relit, and worked for some of the entry burn, but really seriously began to fail a good 15 seconds prior to the end of the entry burn.  When the entry burn ended, there was active, steady and heavy leakage from the bad engine or its plumbing, which is most likely what killed the booster.

Again, though, no one decided anything, really.  The booster is pretty well autonomous in flight, it will just do what it's programmed to do. 

[Vettedrmr]

I didn't read anything in cscott's post that implied human intervention.  Read it from the POV that the booster is autonomous (which is common knowledge) and it makes just as much sense.  It's easy to have software treat failure modes differently during different flight regions (i.e. shut down an engine during ascent to preserve the payload but attempt to re-light that same engine during entry and landing burns).

[catdlr]

Just wanted to show the damage on the engine boots from a previous flight (See video at 1:20)

SpaceX - All Twice Flown Merlin's - 4K - 07-01-2017

[Steven Pietrobon]

Screen grab of the damaged boots.