Abort options for Starship and Starship/SuperHeavy

[InterestedEngineer]

To match motorcycles you can have 30 fatalities per 100 million miles travelled. So around 4000 orbits. Or 250 days in orbit continuously.

Or a fatality every 10 days.

That seems achievable.

The correct metric is fatalities per trip started, not per mile.

All y'all posting random numbers need to do as nearly every science teacher and engineering teacher I've had has said:  State your units.

1:500 trips is good to start with, but for long term 1:10,000 trips is a reasonable expectation.

I’m not sure you’re the arbiter of correct metrics. But I was just making a point that let’s not use literally the safest form of travel as our aim point. Let’s use sailing, or motorcycles, or those giant dune buggies that carry like 40 tourists that flip over… something more bold.

I entirely agree that airlines as a baseline is silly.

That guy proposing such things or those unable to do a simple google search is approaching troll levels.

[MichaelBlackbourn]

Wingsuits are 1 in 500. Which is open to the public and there is no outcry to ban them as far as I know.

[Lee Jay]

Wingsuits are 1 in 500. Which is open to the public and there is no outcry to ban them as far as I know.

Seriously? Comparing participating in an extreme sport to being a passenger on a transportation service?

The 737MAX was grounded worldwide after two accidents, 4 months apart, while flying over 8,000 flights per day.  So, an accident rate of around 1 per 500,000 was so high as to ground the entire fleet everywhere on the planet.

[tyrred]

So there are still no viable abort options for Starship? Cool. It's going to be a very exhilarating ride for whoever can afford it. Just remember to bring a towel.

[markbike528cbx]

So there are still no viable abort options for Starship? Cool. It's going to be a very exhilarating ride for whoever can afford it. Just remember to bring a towel.

I’d also recommend Joo Janta 200 Super-Chromatic Peril Sensitive Sunglasses.
“Zaphod . . . took out his Peril-Sensitive Sunglasses again. They were completely black.”
—From The Restaurant at the End of the Universe[src].  https://hitchhikers.fandom.com/wiki/Joo_Janta_200_Super-Chromatic_Peril_Sensitive_Sunglasses

[Eka]

When will Starship fly a million times? It’ll take longer than that to prove 1 in a million reliability, but at least it becomes numerically possible.

Well, just the Mars self sustaining colony effort plans on 1000+ Starship flights to and from Mars per synod when it gets into full swing. SpaceX plans on making 100 Starships a year just to cover the colonization efforts. Starship is designed to be the low cost reliable truck or bus that gets the job done.

If Moon colonization gets into full swing, I can easily see it needing more flights per year. Then there are LEO space stations for manufacturing, R&D, and tourism.  Also point to point flights for cargo, then later passengers, will rack up a lot of flights fast. Starship will get the reliability, unless rocket engine technology advancements totally change the viable shape for ground to orbit craft.

If it takes giving each flip and burn engine it's own header tanks, it will be done to get the reliability needed.

On a side note: I used commercial Concorde flights every time they fit my schedule. They were less than half the time gate to gate as any other passenger aircraft were. On a trans Atlantic flight that meant 4+ hours less cooped up in a plane. I expect if point to point Starships flew routine daily routes, they would get most long haul business travel. Having an engineer or executive down for a day or two flying costs a lot of money.

[chopsticks]

Where did you come up with the 6 billion and why do think it would have a lifetime less than 7 years?

well established up thread.  You want me to do the Google Search for you?

Sigh. Okay, I will

https://web.archive.org/web/20210301220753/https://christopherrcooper.com/how-much-does-spacex-dragon-cost/

https://letmegooglethat.com/?q=https%3A%2F%2Fwww.google.com%2Fsearch%3Fq%3Dhow%2Bmuch%2Bdoes%2Bthe%2Borion%2Bcapsule%2Bprogram%2Bcost%26oq%3Dhow%2Bmuch%2Bdoes%2Bthe%2Borion%2Bcapsule%2Bprogram%2Bcost%26gs_lcrp%3DEgZjaHJvbWUyBggAEEUYOTIHCAEQIRigATIHCAIQIRigAdIBCDY0NTdqMWo5qAIAsAIA%26sourceid%3Dchrome%26ie%3DUTF-8


Starliner is at about $1.5B.  So it's a range of $1.5B - $20B.

Man rated capsules are insanely expensive to develop.

Google search irrelevant.

SpaceX already have Dragon. Making something similar would almost certainly not be nearly as difficult or as expensive as designing something from scratch. And your idea of 7 years of obsolescence is simply a guess.

Besides, no one said it was easy or necessarily very cheap. Human lives are at stake here. Designing a system with no conceivable way of getting astronauts out of harm's way just seems a bit reckless and betting on success.

All this being said, I feel like I'm talking to a brick wall here because I don't think SpaceX is going to design anything like this - it's really more of a what-if thought exercise at this point.

[chopsticks]

If it takes giving each flip and burn engine it's own header tanks, it will be done to get the reliability needed.

That's an interesting idea..

[joek]

Executive Summary of the "survive the most common crash" viewpoint ...

Thanks for putting that together. Might be a hard sell for some, but an option worth considering as adds at least adds an arrow to the quiver.

[InterestedEngineer]

Google search irrelevant.

SpaceX already have Dragon. Making something similar would almost certainly not be nearly as difficult or as expensive as designing something from scratch. And your idea of 7 years of obsolescence is simply a guess.

Besides, no one said it was easy or necessarily very cheap. Human lives are at stake here. Designing a system with no conceivable way of getting astronauts out of harm's way just seems a bit reckless and betting on success.

All this being said, I feel like I'm talking to a brick wall here because I don't think SpaceX is going to design anything like this - it's really more of a what-if thought exercise at this point.

I believe there was a requirement to get to 20+ passengers, so Dragon itself won't work.

The hypergolic fuels in an an enclosed space (the rest of the Starship cargo section) is a non starter.  So it can'tt be Dragon.

It's solid rockets or bust.   Like the ejection seats of fighter jets or of Orion capsule.

7 years is an optimistic estimate of the development time of a new human rated capsule.   In that time Starship will have completed well over 200 consecutive landings in a row and worked out the bugs, which makes the new capsule less reliable than well tested hardware.

[joek]

The correct metric is trips with fatalities per trip started, not per mile.
...
1:500 trips is good to start with, but for long term 1:10,000 trips is a reasonable expectation.

That puts us about where CCP is... loss of crew (LOC) 1:500 for ascent; 1:500 reentry; 1:270 entire mission (including on orbit). Note use of LOC: crews/event; not individuals/event.

[joek]

...
The hypergolic fuels in an an enclosed space (the rest of the Starship cargo section) is a non starter.  So it can'tt be Dragon.

It's solid rockets or bust.   Like the ejection seats of fighter jets or of Orion capsule.

Would generally agree, but given this is a single-use solution, might be options?
1. SRB's as you suggest.
2. Hypergols in separate sealed system/area. Unlike current Dragon and CST-100 conops, nominal flights should not require any use of those hypergols, and thus the extensive sniffing for leakage after reentry. There should always be zero traces unless the system has been used, which in this case is once in an off-nominal situation. How to contain/isolate from the crew area if it is used another issue.
3. Raptor rapid/emergency start. We hear about the damage that can result without pre-chill etc. But what if in this type of off-nominal case, that damage was acceptable (obviously without the engines going boom on startup)? The entire ship (engines included) is likely to be scrap when the dust settles, so in this scenario they are single use / disposable.

[Kazioo]

Not sure how realistic this is, but watching the first integrated launch of Starship it occurred to me that maybe an abort option for 9 engine Starship (the second stage) could also be a software-only feature without added complexity and risk of more hardware - to save crew only in cases when the booster is slowly malfunctioning and doesn't explode immediately, but with enough data to tell that orbit won't be possible. Just let second stage detach early and try to land. And if S2 doing it completely on its own is not possible, then at least do it in scenarios where booster still has some control and is still capable of participating in the abort process.

Currently it seems (?) that even serious performance issues without RUD mean you have to accept upcoming disaster and that feels wrong. What if more and more engines on the booster start failing? There should be a procedure to try saving the ship without trying to get into orbit. Obviously, if it's too far away there may not be enough propellant to return the ship to the launch site, but even in that case trying to land on water is better than doing nothing (it worked for Falcon 9 once, not sure if this is feasible for Starship).

These vehicles are supposed to be relatively cheap to make, so it shouldn't be a problem for SpaceX to test several different scenarios.

Perfect is the enemy of good. Improving safety even by 10% is worth trying, and let's not forget about the psychological aspect. Everyone would feel better after some abort tests and that's important too.

[Lee Jay]

Not sure how realistic this is, but watching the first integrated launch of Starship it occurred to me that maybe an abort option for 9 engine Starship (the second stage) could also be a software-only feature without added complexity and risk of more hardware - to save crew only in cases when the booster is slowly malfunctioning and doesn't explode immediately, but with enough data to tell that orbit won't be possible. Just let second stage detach early and try to land. And if S2 doing it completely on its own is not possible, then at least do it in scenarios where booster still has some control and is still capable of participating in the abort process.

Currently it seems (?) that even serious performance issues without RUD mean you have to accept upcoming disaster and that feels wrong. What if more and more engines on the booster start failing? There should be a procedure to try saving the ship without trying to get into orbit. Obviously, if it's too far away there may not be enough propellant to return the ship to the launch site, but even in that case trying to land on water is better than doing nothing (it worked for Falcon 9 once, not sure if this is feasible for Starship).

These vehicles are supposed to be relatively cheap to make, so it shouldn't be a problem for SpaceX to test several different scenarios.

Perfect is the enemy of good. Improving safety even by 10% is worth trying, and let's not forget about the psychological aspect. Everyone would feel better after some abort tests and that's important too.

I think what you're describing was similar to "fast sep" or "RTLS" in the STS era, the idea being to get your fully fueled (or nearly so) crew-carrying craft away from a failing booster, fly around to burn off fuel in the process of returning to launch site.  Since liquid rockets test to fail in a slower fashion than solids, that could probably work provided that you can get a clean separation after convincing the booster to stop thrusting.  It seems to me that the approach to this would be for the booster to recognize this situation, cut thrust and initiate sep.  One problem here is there is no separation system on this rocket, instead relying on the booster to initiate a pitch or yaw to initiate separation.  If the booster if failing, I'm not sure that could be trusted to work.  I'm also unsure how safe it is to ignite the upper stage with it still attached or in close proximity to the booster.

None of this helps if the problem is on the ship, during ascent or during the flip and landing maneuver.

[Barley]

a software-only feature without added complexity and risk of more hardware

Let me guess, you're a manager that used to be a mechanical engineer?

[Greg Hullender]

Just let second stage detach early and try to land.

If you look back at the very first post in this (rather long) thread, you'll see that, as presently designed, the second stage cannot do that.

As shown in this thread:
https://forum.nasaspaceflight.com/index.php?topic=56619.0

Starship is aerodynamically unstable when at an AoA below 20° - as such loss of thrust during ascent would likely cause the vehicle to tumble and crash or break apart.

There is a window of time between liftoff and the point where aerodynamics don't matter anymore. During that window, the second stage will be destroyed if it separates and is not under thrust. Of course, it could always fire its engines to pull away from the booster, but, as currently designed, Starship cannot do this kind of "hot staging."

Elsewhere in this thread, there's a very interesting discussion about how to modify the vehicle to make hot-staging possible, but it requires a good bit more than a software change.

[joek]

...
None of this helps if the problem is on the ship, during ascent or during the flip and landing maneuver.

Therein might be some of the disconnect in these debates? With traditional architectures (e.g., Soyuz, Apollo, Dragon, CST-100) we have a clear delineation between the spacecraft and the rocket (stages) and their roles-responsibilities. Launch and reentry safety responsibility has always belonged primarily with the spacecraft. With SS/SH the division is not as clear, so maybe start there.

We could as easily be having a similar conversation about an SSTO architecture: the "ship" might fail, what is our LAS/LES plan? That leads to a ship-in-a-ship-in-a-ship LAS/LES architecture which never ends. Don't think we want to go there.

[Eka]

2. Hypergols in separate sealed system/area. Unlike current Dragon and CST-100 conops, nominal flights should not require any use of those hypergols, and thus the extensive sniffing for leakage after reentry. There should always be zero traces unless the system has been used, which in this case is once in an off-nominal situation. How to contain/isolate from the crew area if it is used another issue.

No seal, joint, valve, tube, burst disk, etc is perfect. Expose them to lots of vibrations from a launch or 20, and things will leak. Hope you haven't just made your main burn to transit to the Moon or Mars when it starts to leak. Also no hypergol resupply on the Moon and Mars.

Your best bet is super high reliability with redundancy in case of failure. Yet it needs to be remembered that every additional system adds weight, takes up space, and adds complexity. This is why later test flights fired up all three center Raptors when only two were needed for the flip and burn. SpaceX is right now testing what do they need to do as a minimum to get a highly reliable system that will do the tasks needed.

[mmeijeri]

I believe there was a requirement to get to 20+ passengers, so Dragon itself won't work.

The hypergolic fuels in an an enclosed space (the rest of the Starship cargo section) is a non starter.  So it can'tt be Dragon.

It's solid rockets or bust.   Like the ejection seats of fighter jets or of Orion capsule.

A simpler short-term solution that has been suggested before would be to use Dragon + F9 to launch crew separately and to dock with Starship in orbit. But in the long run that isn't practical as it would require keeping F9 in service.

[drzerg]

What is worst case scenario for the starship stack on ascend?

There is no such thing as a worst case, it can always be worse.  A hole in and collapse of a StarShip (second stage) methane tank is worse than a Super Heavy (first stage) failure you give as a worse case.

Airliner analogy to this is wing could fall off. Is it deadly? yes if you are not f15. Is in frequent? No.
If you add probability to equation worst case start to appear as the severity*probability entity