SpaceX Starship/Super Heavy Engineering General Thread 4

[Twark_Main]

That would applied to manned aerospace.

That saying is incorrect for unmanned aerospace.

Wrong.

Launch failures are more than lost hardware. It's an expensive PR failure too. That leads to lost contracts, policymaking blowback, etc.

Ignoring major failures in favor of adopting a "You Only Launch Once" attitude is penny wise and pound foolish.

Finding out whether something works in the face of failures is more important.

Exactly. The hold-down abort should work in the face of engine failures. They should not not work.

You know how Elon says to call him if there's "anything" that can be done to improve reliability? The day they ever try this (which they won't), I have a call to make...

It would be helpful if someone would come up with a valid technical/math reason to abort one one booster engine failure.  The only argument in that line I've got is common mode (non-independent) failures, which I've addressed.

It improves your odds of mission success. SpaceX prioritizes reliability and safety above all else. Yes, really.

[Twark_Main]

Airlines will often do ferry flights after an engine failure, flying out with 1/2 or 3/4 engines (no passengers).  This is from the ramp, not V1.

While engine out ferry flights are a thing, that's only done on 3 or 4 engine aircraft--with essential crew only, no passengers, minimum fuel, a specially-qualified crew, etc.  And given the extra risk etc. you're probably only going to do it if there's any reasonable way of fixing or replacing the engine on site.  Note that at least the 747 had a way of externally carrying an extra engine to an aircraft in a remote location.
Nobody flies an engine out ferry with a twin.  Leaving aside the no-redundancy factor and associated much higher risk, I'm not even sure you could within realistic runway widths/lengths; especially on modern twins with underwing engines there's too much asymmetric thrust and at low speed you'd have no rudder authority to assist the nosewheel.  If the engine on a twin goes out, you're fixing it wherever it is.

See also

https://www.law.cornell.edu/cfr/text/14/91.611

[meekGee]

Ok, I'll play devil's advocate.

What if there were 330 smaller mini-raptors at the base of the rocket?

Lifting off with 329/330 would be ok, right?

And clearly lifting off with 8/9 is not ok.

So where's the cross-over point?

[kevinstout]

It wont ever be a technical requirment.  It has to be about the mission.

you would need to have a mission so timing crictical,  and so low in value,  to even attempt this. 

something like, absolute last chance of launching a tanker flight before a deepspace mission. And if you dont launch,   the whole mission over mutiple launches would fail. 

and even then,  we are talking about the time it would take to swap out a engine,  or an entire core.  so,  like hours or a few days max.   

its really hard to picutre something that important,  running that late, where a few hours saved is worth more than the launch and payload itself.

[JayWee]

Ok, I'll play devil's advocate.

What if there were 330 smaller mini-raptors at the base of the rocket?

Lifting off with 329/330 would be ok, right?

And clearly lifting off with 8/9 is not ok.

So where's the cross over point?

I remember one lecture by a shuttle designer who complained about this. That the shuttle was designed for airplane-like 4 redundancy precisely to allow taking off with a failed system. But that NASA took the "no failures allowed" approach leading to many launch aborts.


I do believe it was part of the MIT "Aircraft system engineering" course
youtube.com/playlist?list=PL35721A60B7B57386
but don't have the time to find the quote now.


In my opinion it depends on the stage of the development. Right now? No way. 10 years in the future where you are launching few times a day? Probably yes.

[Jim]

Common mode errors are relatively easy to find, because the sample size is very large, especially with SpaceX's large production lots and the McGregor testing facility

No, it is still too small of sample size for statistical analysis.

[Lee Jay]

I get the desire to launch with 32/33 engines, but it's tempting fate.  What if the problem is common to more than one engine?

Airlines will often do ferry flights after an engine failure, flying out with 1/2 or 3/4 engines (no passengers).  This is from the ramp, not V1.

No they don't.  While there are rumors of it happening once or twice, this is certainly not regular practice.  An engine is flown or trucked in to replace the damaged engine before another flight happens.

What does happen is ferry flight when there's an engine with some slight problem that wouldn't allow a passenger flight to occur.  But the engine still functions.  If it fails before V1, the flight is aborted, as always.

[InterestedEngineer]

I get the desire to launch with 32/33 engines, but it's tempting fate.  What if the problem is common to more than one engine?

Airlines will often do ferry flights after an engine failure, flying out with 1/2 or 3/4 engines (no passengers).  This is from the ramp, not V1.

No they don't.  While there are rumors of it happening once or twice, this is certainly not regular practice.  An engine is flown or trucked in to replace the damaged engine before another flight happens.

What does happen is ferry flight when there's an engine with some slight problem that wouldn't allow a passenger flight to occur.  But the engine still functions.  If it fails before V1, the flight is aborted, as always.

There's even an FAA regulation allowing it:

https://www.law.cornell.edu/cfr/text/14/91.611

It's also still apples/oranges.  The statistics of 3/4 is completely different than 32/33, by orders of magnitude.

[Lee Jay]

I get the desire to launch with 32/33 engines, but it's tempting fate.  What if the problem is common to more than one engine?

Airlines will often do ferry flights after an engine failure, flying out with 1/2 or 3/4 engines (no passengers).  This is from the ramp, not V1.

No they don't.  While there are rumors of it happening once or twice, this is certainly not regular practice.  An engine is flown or trucked in to replace the damaged engine before another flight happens.

What does happen is ferry flight when there's an engine with some slight problem that wouldn't allow a passenger flight to occur.  But the engine still functions.  If it fails before V1, the flight is aborted, as always.

There's even an FAA regulation allowing it:

https://www.law.cornell.edu/cfr/text/14/91.611

It's also still apples/oranges.  The statistics of 3/4 is completely different than 32/33, by orders of magnitude.

I was talking about a twin since we're mostly there these days.  I didn't see that you also mentioned 3/4.  The rule doesn't allow it for a twin, which you mentioned first.

"...conduct a ferry flight of a four-engine airplane or a turbine-engine-powered airplane equipped with three engines"

[ppb]

Ok, I'll play devil's advocate.

What if there were 330 smaller mini-raptors at the base of the rocket?

Lifting off with 329/330 would be ok, right?

And clearly lifting off with 8/9 is not ok.

So where's the cross over point?

As previous posters have pointed out, maybe down the line after thousands of  launches have been made and engine failures were statistically shown —IN THE ALL UP SYSTEM ENVIRONMENT—to be independent, then a safe number of failures pre-commit could be safely tolerated. The CAPS are to emphasize that this is the part of launch that is untested at this point in the development cycle and probably only partially understood. A failure of one engine for early launches may be a clue that something misunderstood or non-understood is happening in the maelstrom of 33 engines starting up and the dynamic system (acoustic, thermal, structural, electrical, fluid, etc.) response and cross-coupling. As systems become more complex, it takes just the right test case to reveal a problem, and sometimes only a subtle change, say a a few milliseconds in startup time of a particular group of engines will reveal the problem. Even after thousands of launches, it’s unlikely SpaceX will have stopped tweaking things, so I think I just talked myself into believing it will never be OK to release the clamps with a known bad engine.

[CorvusCorax]

A big difference is also "known bad engine" vs "unexpectedly failed engine"

Assume during a post flight inspection it is noticed that an engine has sustained damage during reentry. Launching the tanker is time critical due to orbital mechanics, so SpaceX might chose to launch on 32 engines and deliberately not attempt to start the bad one.

I could see that happening in the future. The risk is negligible, if the issue with the engine is understood and theres enough margin.

But if an engine UNEXPECTEDLY fails in the ignition sequence, it HAS to be a launch abort.

[matthewkantar]

I would want to be sure the mission is safe if an engine shuts down a tenth of a second after release. If I am confident the vehicle can deal with it, why would I eat the expense of a scrub with only one balky engine? If two engines go down before release a scrub seems more reasonable.

[DanClemmensen]

It wont ever be a technical requirment.  It has to be about the mission.

you would need to have a mission so timing crictical,  and so low in value,  to even attempt this. 

something like, absolute last chance of launching a tanker flight before a deepspace mission. And if you dont launch,   the whole mission over mutiple launches would fail. 

and even then,  we are talking about the time it would take to swap out a engine,  or an entire core.  so,  like hours or a few days max.   

its really hard to picutre something that important,  running that late, where a few hours saved is worth more than the launch and payload itself.

Or, if you are writing a bad suspense novel, the mission is to save some crew stranded by an earlier in-space disaster. The plot just writes itself, and it's pretty bad.

[Robotbeat]

I would want to be sure the mission is safe if an engine shuts down a tenth of a second after release. If I am confident the vehicle can deal with it, why would I eat the expense of a scrub with only one balky engine? If two engines go down before release a scrub seems more reasonable.

Because again, scrubs are cheap. You want to make sure the mission overall succeeds, and that you don’t lose either the booster or the upper stage either. That probability is maximized by scrubbing if there’s a problem with one of the engines.

SpaceX wants to reach higher reliability than has ever been achieved by rocket vehicles.

[DanClemmensen]

I would want to be sure the mission is safe if an engine shuts down a tenth of a second after release. If I am confident the vehicle can deal with it, why would I eat the expense of a scrub with only one balky engine? If two engines go down before release a scrub seems more reasonable.

Because again, scrubs are cheap. You want to make sure the mission overall succeeds, and that you don’t lose either the booster or the upper stage either. That probability is maximized by scrubbing if there’s a problem with one of the engines.

SpaceX wants to reach higher reliability than has ever been achieved by rocket vehicles.

OK, you are right, given some reasonable assumptions. If a tower will eventually be able to launch three times a day, it follows that a stacking operation is quite quick and that a de-press/destack will also be quick. This means they can completely replace the SH with a backup SH, and depending on the mission they can be ready to launch the SS one day later, and probably not even interfere with other launches except in special circumstances.  I had been thinking about the generic launcher, not SH. SLS in particular has nt backup and has a minimum cycle time of 48 hours and cycle times that can be two weeks or more.

[pathfinder_01]

I would want to be sure the mission is safe if an engine shuts down a tenth of a second after release. If I am confident the vehicle can deal with it, why would I eat the expense of a scrub with only one balky engine? If two engines go down before release a scrub seems more reasonable.

The other reason is that this is unnecessary risk. If doing a tanker mission you could have multiple launch sites with flights to the depot so there might not be much delay at all for the scrub. If caring people or cargo leaving without all engines puts you into more risk. Why risk it? Most satellites can wait and you would schedule some extra time or flights if you were filling a tanker.

[LouScheffer]

I'd like to give a shout-out to the folks who designed the thrust puck and thrust dome.  I'm sure many readers here also watch the various hydraulic press channels

,

where various seeming strong stuff (such as sledgehammer heads) are deformed and crushed in the press, and fail much worse at cryogenic temperatures.  And the hydraulic press used in this example tops out at about 1.36 MN, less than the force of *one* Raptor engine (about 1.8 MN).  And the thrust puck and dome have to transfer the forces of *33* of these engines to a relatively thin and fragile tank bottom and side walls.  It's amazing they can do this at all, much less under tight mass constraints.  That's great engineering.

[JayWee]

It's amazing they can do this at all, much less under tight mass constraints.  That's great engineering.

Amazing video! I do remember Elon mentioning that the 304L steel might not be that bad at cryo.

[AstroWare]

Moved from Stoke thread...

F9 upper stage reuse would have been viable if they chose that path. Fully reusable Falcon Heavy’s reuse penalty would’ve been small enough to LEO that it probably would’ve been pretty practical.

The only thing truly radical feature-wise about Starship is the chopsticks.

The other stuff is super aggressive like Raptor or the raw scale, but not what I’d call radical. It’s largely demonstrated on Shuttle, Soviet engines, Atlas, F9, and DC-X.

At least the chopsticks have a pretty straightforward and proven fallback plan (Legs). Kinda like fairing catching - give it a best effort, but if it isn't working then change. SpaceX is good at that.

I'm kinda hoping that future crewed starships have the SN8 type stubby landing legs even after chopsticks work just in case...

[Nomadd]

I would want to be sure the mission is safe if an engine shuts down a tenth of a second after release. If I am confident the vehicle can deal with it, why would I eat the expense of a scrub with only one balky engine? If two engines go down before release a scrub seems more reasonable.

The other reason is that this is unnecessary risk. If doing a tanker mission you could have multiple launch sites with flights to the depot so there might not be much delay at all for the scrub. If caring people or cargo leaving without all engines puts you into more risk. Why risk it? Most satellites can wait and you would schedule some extra time or flights if you were filling a tanker.

Nothing is without risk. You don't determine launch because of "safe or not safe". You compare the odds against the alternative. If they decide that an abort and recycle has greater odds of causing grief than launching with a single bad engine, they'll launch.