SpaceX F9 : Starlink 6 (v1.0 L5) : Mar. 18, 2020 - Discussion

[Vettedrmr]

Here's what is concerning me:

Starlink L3: Hard landing, cause not publicly revealed
Starlink L4: Missed landing due to wrong wind data loaded into flight plan configuration
CRS-20: Fully successful mission
Starlink L5: Engine failure on ascent, payload successfully delivered, failed landing attempt.

The starlink flights are using the most used boosters, finding the limits of re-use.

BUT, EM seemed surprised at the engine failure.  He never said what the issue was with L3's hard landing (nor is he required to), but the wind data mistake was a big error, and something in their engine reliability analysis missed whatever failed on L5's ascent.

So, I expect SpaceX to take a breath on their Starlink launches and see what they need to change in their processes. Customer launches are probably either unaffected or minimally affected by this latest anomaly, so I'm not worried about SAOCOM, DM-2, etc.  But starlink may pause until they figure out what needs to change.

Have a good one,
Mike

[Lars-J]

Here's what is concerning me:
...
BUT, EM seemed surprised at the engine failure.

Stop right there. That’s some mighty impressive analysis of a short tweet. You are basing your entire post one one word there, and it falls apart.

[gaballard]

Here's what is concerning me:

Starlink L3: Hard landing, cause not publicly revealed
Starlink L4: Missed landing due to wrong wind data loaded into flight plan configuration
CRS-20: Fully successful mission
Starlink L5: Engine failure on ascent, payload successfully delivered, failed landing attempt.

The starlink flights are using the most used boosters, finding the limits of re-use.

BUT, EM seemed surprised at the engine failure.  He never said what the issue was with L3's hard landing (nor is he required to), but the wind data mistake was a big error, and something in their engine reliability analysis missed whatever failed on L5's ascent.

So, I expect SpaceX to take a breath on their Starlink launches and see what they need to change in their processes. Customer launches are probably either unaffected or minimally affected by this latest anomaly, so I'm not worried about SAOCOM, DM-2, etc.  But starlink may pause until they figure out what needs to change.

Have a good one,
Mike

The answer to all your questions is in your own post... "The starlink flights are using the most used boosters, finding the limits of re-use."

And the wind data issue wasn't a "big error". Everything actually went exactly as planned in that situation... the booster realized something was off and sacrificed itself to save the droneship. A "big error" would have been the booster not knowing something was off and punching a hole through the deck (which has happened before).

[marsbase]

So, I expect SpaceX to take a breath on their Starlink launches and see what they need to change in their processes.

Nope.  Every one of the missions you listed delivered their payloads successfully.  And SpaceX learned from each one.  SpaceX learns when they fly.  They don't learn when they don't fly.

[1]

Here's what is concerning me:
...
BUT, EM seemed surprised at the engine failure.

Stop right there. That’s some mighty impressive analysis of a short tweet. You are basing your entire post one one word there, and it falls apart.

And I don't know which tweet that might be, but it certainly wasn't this one:

https://mobile.twitter.com/elonmusk/status/1240277125023719424?p=v

[meekGee]

During the abort, they got the signature of the faulty engine anomaly.

From Musk's tweet, I kinda infer it was the same engine (or else how would it be related) but that they haven't yet compared the in-flight anomaly signature to the abort one.

 

[sferrin]

Here's what is concerning me:
...
BUT, EM seemed surprised at the engine failure.

Stop right there. That’s some mighty impressive analysis of a short tweet. You are basing your entire post one one word there, and it falls apart.

And I don't know which tweet that might be, but it certainly wasn't this one:

https://mobile.twitter.com/elonmusk/status/1240277125023719424?p=v

How does, "so today isn’t a big surprise."  equate to surprised?

[CJ]

I'm stuck at home, so I thought it'd be nice to inflict some wild, unfounded, hyperbolic speculation on y'all;

The consensus seems to be that they had two problems in flight; an early shutdown of one engine prior to MECO, and then an engine problem during the entry burn. If so, I'm thinking that they are directly related;
they are the same engine.

Here's my theory; one engine was shut down on ascent due to sensor readings (which could have been caused by a physical failure). However, the recovery software went ahead and tried a relight, because the failed engine was one of the 3 needed for the entry burn. It failed again (hence the yaw). This indicates (IMHO) that the failure was physical, as the recovery GNC would IMHO be unlikely to shut down an engine based on sensor readings (LOV either way, so why do it?).

My guess is that the reduced entry burn, coupled perhaps with physical damage, made the entry too much for the F9 to survive, hence no "transonic"callout; they lost contact before that. 

So, my further guess is that this engine is one of the two outboard ones used in the 3 engine entry burn. Further, the failure was physical, not sensor. If I had to guess further (and I do) I'd say a suspect is a turbopump suddenly deciding to run hardware rich.

 

[Vettedrmr]

https://twitter.com/elonmusk/status/1240262636547100672

Musk didn't mention anything about pushing the life limits of the engines: he says instead "thorough investigation needed before next mission."

A, that's not "one word".  B, while these rockets are long life leaders, SpaceX's inspections of their engines either missed something on one of the Merlins, or their analysis was closer to the edge than they realized.  if their QA process wasn't completed for some reason, then it's a good wakeup call for their QA teams (and QA teams aren't infallible).  If it wasn't that, then if they can figure out what the failure was they can dial in their inspection criteria.

As far as the wind correction error "not a big error", it was big enough that they lost a booster.  And we don't know if the water landing was due to the booster identifying a failure mode or that was where the erroneous landing location was.  In either case, they lost the booster.

[freddo411]

I'm stuck at home, so I thought it'd be nice to inflict some wild, unfounded, hyperbolic speculation on y'all;

The consensus seems to be that they had two problems in flight; an early shutdown of one engine prior to MECO, and then an engine problem during the entry burn. If so, I'm thinking that they are directly related;
they are the same engine.

Here's my theory; one engine was shut down on ascent due to sensor readings (which could have been caused by a physical failure). However, the recovery software went ahead and tried a relight, because the failed engine was one of the 3 needed for the entry burn. It failed again (hence the yaw). This indicates (IMHO) that the failure was physical, as the recovery GNC would IMHO be unlikely to shut down an engine based on sensor readings (LOV either way, so why do it?).

My guess is that the reduced entry burn, coupled perhaps with physical damage, made the entry too much for the F9 to survive, hence no "transonic"callout; they lost contact before that. 

So, my further guess is that this engine is one of the two outboard ones used in the 3 engine entry burn. Further, the failure was physical, not sensor. If I had to guess further (and I do) I'd say a suspect is a turbopump suddenly deciding to run hardware rich.

 

If we assume your theory is correct, it would be odd that the software would not gracefully handle an engine failure on ascent by using different engines for the descent (unless of course it was the center engine).

[cwr]

I'm stuck at home, so I thought it'd be nice to inflict some wild, unfounded, hyperbolic speculation on y'all;

The consensus seems to be that they had two problems in flight; an early shutdown of one engine prior to MECO, and then an engine problem during the entry burn. If so, I'm thinking that they are directly related;
they are the same engine.

Here's my theory; one engine was shut down on ascent due to sensor readings (which could have been caused by a physical failure). However, the recovery software went ahead and tried a relight, because the failed engine was one of the 3 needed for the entry burn. It failed again (hence the yaw). This indicates (IMHO) that the failure was physical, as the recovery GNC would IMHO be unlikely to shut down an engine based on sensor readings (LOV either way, so why do it?).

My guess is that the reduced entry burn, coupled perhaps with physical damage, made the entry too much for the F9 to survive, hence no "transonic"callout; they lost contact before that. 

So, my further guess is that this engine is one of the two outboard ones used in the 3 engine entry burn. Further, the failure was physical, not sensor. If I had to guess further (and I do) I'd say a suspect is a turbopump suddenly deciding to run hardware rich.

 

If we assume your theory is correct, it would be odd that the software would not gracefully handle an engine failure on ascent by using different engines for the descent (unless of course it was the center engine).

As I recollect only 3 of the 9 M1Ds on the booster are plumbed with TEA/TEB for inflight ignition.
Those 3 are the only candidates for the boostback/entry/landing burns.

Carl

[TheRadicalModerate]

I'm stuck at home, so I thought it'd be nice to inflict some wild, unfounded, hyperbolic speculation on y'all;

The consensus seems to be that they had two problems in flight; an early shutdown of one engine prior to MECO, and then an engine problem during the entry burn. If so, I'm thinking that they are directly related;
they are the same engine.

Here's my theory; one engine was shut down on ascent due to sensor readings (which could have been caused by a physical failure). However, the recovery software went ahead and tried a relight, because the failed engine was one of the 3 needed for the entry burn. It failed again (hence the yaw). This indicates (IMHO) that the failure was physical, as the recovery GNC would IMHO be unlikely to shut down an engine based on sensor readings (LOV either way, so why do it?).

My guess is that the reduced entry burn, coupled perhaps with physical damage, made the entry too much for the F9 to survive, hence no "transonic"callout; they lost contact before that. 

So, my further guess is that this engine is one of the two outboard ones used in the 3 engine entry burn. Further, the failure was physical, not sensor. If I had to guess further (and I do) I'd say a suspect is a turbopump suddenly deciding to run hardware rich.

I can't imagine the software doing a relight on an engine that failed--especially since it appears to have had chunks coming off of it (check the video a couple of frames before the visible flash of the failure).  Instead, it would just re-map to use three other in-line engines--unless of course it was the center engine that failed, and the video evidence doesn't support that.  (Note:  there are four different sets of three engines in the octaweb that'll give you a 3-engine in-line burn.  Since there are four grid fins, there are plenty of alternate axes of symmetry.)

This is one of the main reasons why I think that the reentry burn control problems visible in the video are likely aerodynamic in nature.  Something blew up and damaged an aerosurface.  It could have been the engine heat shielding or the skirt.  That was fine on ascent because the failure occurred in near-vacuum.  But on reentry, it caused attitude control problems, which likely got worse as the dynamic pressure increased after the reentry burn, and eventually became uncontrollable.

Edit:  Hadn't read Carl's TEA/TEB comment.  Makes sense.

[macpacheco]

I'm stuck at home, so I thought it'd be nice to inflict some wild, unfounded, hyperbolic speculation on y'all;

The consensus seems to be that they had two problems in flight; an early shutdown of one engine prior to MECO, and then an engine problem during the entry burn. If so, I'm thinking that they are directly related;
they are the same engine.

Here's my theory; one engine was shut down on ascent due to sensor readings (which could have been caused by a physical failure). However, the recovery software went ahead and tried a relight, because the failed engine was one of the 3 needed for the entry burn. It failed again (hence the yaw). This indicates (IMHO) that the failure was physical, as the recovery GNC would IMHO be unlikely to shut down an engine based on sensor readings (LOV either way, so why do it?).

My guess is that the reduced entry burn, coupled perhaps with physical damage, made the entry too much for the F9 to survive, hence no "transonic"callout; they lost contact before that. 

So, my further guess is that this engine is one of the two outboard ones used in the 3 engine entry burn. Further, the failure was physical, not sensor. If I had to guess further (and I do) I'd say a suspect is a turbopump suddenly deciding to run hardware rich.

I can't imagine the software doing a relight on an engine that failed--especially since it appears to have had chunks coming off of it (check the video a couple of frames before the visible flash of the failure).  Instead, it would just re-map to use three other in-line engines--unless of course it was the center engine that failed, and the video evidence doesn't support that.  (Note:  there are four different sets of three engines in the octaweb that'll give you a 3-engine in-line burn.  Since there are four grid fins, there are plenty of alternate axes of symmetry.)

This is one of the main reasons why I think that the reentry burn control problems visible in the video are likely aerodynamic in nature.  Something blew up and damaged an aerosurface.  It could have been the engine heat shielding or the skirt.  That was fine on ascent because the failure occurred in near-vacuum.  But on reentry, it caused attitude control problems, which likely got worse as the dynamic pressure increased after the reentry burn, and eventually became uncontrollable.

Edit:  Hadn't read Carl's TEA/TEB comment.  Makes sense.

Only 3 engines have the TEA/TEB for restart. The choices are only the center engine, the 2 side engines or all three. But it can't change which ones will be restarted because the TEA/TEB is hard plumbed into only those 3.

Ground start is accomplished with a ground source for TEA/TEB that covers all engines. TEA/TEB shots on the booster are only used for inflight restarts.

[marsbase]

Only 3 engines have the TEA/TEB for restart. The choices are only the center engine, the 2 side engines or all three. But it can't change which ones will be restarted because the TEA/TEB is hard plumbed into only those 3.

How do the non-plumbed engines start on the launch pad?

[intelati]

Only 3 engines have the TEA/TEB for restart. The choices are only the center engine, the 2 side engines or all three. But it can't change which ones will be restarted because the TEA/TEB is hard plumbed into only those 3.

How do the non-plumbed engines start on the launch pad?

Ground plumbed TEA/TEB. (I believe)

[edkyle99]

Looked like different engine failures to me on the on-board video, one on ascent and one during the entry burn, one on the "left" side and one on the "right" side.  I suspect both are related to whatever happened during the ascent failure.  I also thought it interesting that the engines seemed to be gimbaling just before the first failure and that this might have been around the time of throttle-down.

 - Ed Kyle

[shooter6947]

Looked like different engine failures to me on the on-board video, one on ascent and one during the entry burn, one on the "left" side and one on the "right" side.  I suspect both are related to whatever happened during the ascent failure.

Could the issue on the entry burn been one of fuel starvation if the first failure had led to a longer first-stage burn to make up the lost dV?

[AndrewRG10]

Looked like different engine failures to me on the on-board video, one on ascent and one during the entry burn, one on the "left" side and one on the "right" side.  I suspect both are related to whatever happened during the ascent failure.

Could the issue on the entry burn been one of fuel starvation if the first failure had led to a longer first-stage burn to make up the lost dV?

No. The MECO time was exactly the same as L4 which had the same profile, it did this by not throttling down as much and therefore used practically the same amount of fuel. It wasn't so low on fuel that the landing burn didn't start or re-entry burn failed. Landing burn didn't happen (from what we know) because there wasn't much of a working vehicle come landing burn.

[OneSpeed]

Looked like different engine failures to me on the on-board video, one on ascent and one during the entry burn, one on the "left" side and one on the "right" side.  I suspect both are related to whatever happened during the ascent failure.

Assuming the S1 raceway is orientated at +Z, and that F9 rolls 180° to starboard after liftoff until the raceway is ventral, then the camera is mounted near the -Y orientation.

If the S1 camera is a little left of the -Y orientation, then the engines that re-ignite for the entry burn are probably 4, 9 and 8. If those numbers are right, then you can usually see engine 4 ignition about 4 seconds after engine 9, and shutdown about 1 second before it. This was the case with all of the Starlink missions up to F4, but as far as I can tell, not F5. As per the attached images, I'm not seeing the second ignition for F5. This leads me to suspect that the engine that failed on ascent was engine 4, and that the entry burn was sub-optimal as a result. It is one thing to burn longer on ascent while the atmosphere is thinning, but a two engine entry burn would put the booster in denser atmosphere at higher velocity than intended.

When you say that the entry burn failure was 'on the right', is there an image or video that shows that?

I also thought it interesting that the engines seemed to be gimbaling just before the first failure and that this might have been around the time of throttle-down.

The 'premature shutdown' was at T+142 seconds, and Starlink throttle down usually occurs at about T+149 seconds.

Edit:

No. The MECO time was exactly the same as L4 which had the same profile, it did this by not throttling down as much and therefore used practically the same amount of fuel. It wasn't so low on fuel that the landing burn didn't start or re-entry burn failed. Landing burn didn't happen (from what we know) because there wasn't much of a working vehicle come landing burn.

L5 burnt about a second longer as well. Velocity at separation was slightly higher, so it burnt slightly more propellant overall.

[ugordan]

I also thought it interesting that the engines seemed to be gimbaling just before the first failure and that this might have been around the time of throttle-down.

I believe this is standard behavior in preparation for MECO, nulling out angle of attack to minimize drag-induced torque for stage separation. Prior to that it can fly a noticeable angle of attack, depending on mission trajectory. Some say the angle of attack increase after max-Q is trajectory shaping for gravity losses, others say it also produces aerodynamic lift, but it's not at all uncommon for a F9 launch.

edited for typo.