NASA's Artemis Program Updates and Discussion Thread 6

[Vultur]

I guess it's possible that SpaceX could privately fly crews by 2031, but the consequences of killing a crew on a Starship that NASA wouldn't have certified are too terrible to contemplate, especially once they've gone public. 

While this isn't an argument to be cavalier with human life, I'm actually not convinced that there would necessarily be massive consequences. People do die doing dangerous, flashy private things, and there isn't necessarily massive public backlash. The Everest guide industry hasn't been shut down because of deaths. I'd argue that Virgin Galactic's fatal crash delayed them more because of losing the airframe than because of the consequences of a fatality.

NASA astronauts, OTOH, that would be a completely different and far worse thing.

[thespacecow]

CBS 60 Minutes coverage of Artemis II has a segment quoting Jim Bridenstine's anti-Starship congressional testimony, Jared Isaacman replied the following on X: https://x.com/NASAAdmin/status/2018177327050457109

America’s return to the lunar surface is a complex endeavor involving international partners, five+ prime contractors, hundreds of subcontractors, and tens of thousands of skilled workers. Beyond the landers, there is a lot that must converge to meet our timelines, and we will do everything within the power of the world’s most accomplished space agency to meet our obligations..with time to spare.

The architecture both HLS providers are working on enables America to repeatedly and affordably return to the Moon, in line with the President’s National Space Policy, build a lunar base, and continue the journey on to Mars. To throw away game-changing capabilities like reusable heavy-lift launch vehicles and on-orbit propellant transfer would be to surrender the high ground to our adversaries, alongside the economic opportunity, scientific discovery, and national security advantages it brings.

[Vultur]

I like that wording. Both Blue and SpaceX have refueling in their architectures; it's not some weird crazy SpaceX-only thing.

(And for good reason. Refueling enables so much.)

[TheRadicalModerate]

While this isn't an argument to be cavalier with human life, I'm actually not convinced that there would necessarily be massive consequences.

Well, there's a ringing endorsement.  How unconvinced are you about the necessity of massive consequences?  pMassive < 10%?  < 50%? 

When you answer, imagine you have the fiduciary burden of being the CEO of a $1.5T public company.  You should also imagine that half the country hates you, and you have non-trivial political enemies.

Slightly more on-topic:  What's the rush?  Do you really think there's going to be a huge demand for missions with more than four crew?  How price-sensitive are those missions going to be?  I can see larger crews needing to go to Mars, but that's at least the mid-thirties, likely longer.  Until then, F9/D2 is going to be flying and you might as well use it.  Nobody wants to design their LEO space station to handle Starship's mass and moments until they have the scale to handle at least 20 crew, and that's not what's getting launched first for CLD.

Rushing crewed Starship launches and EDLs has sizable risks and small rewards.  The rewards will grow, but not in the timeframe we're talking about.  And it's just insane to rush ahead when you have a system that'll be much safer, albeit a bit more expensive.

[yg1968]

I agree but I think that [CMPS and cislunar commercial crew] go hand in hand. I think that SpaceX will have a commercial lunar alternative to SLS and Orion, once that their crewed Starship for Mars is ready. So starting CMPS and/or Mars-HDL program as soon as possible should help in that respect.

Even if crew-certifying Starship for launch and EDL wasn't going to take a long time,¹ the crew modules used for cislunar commercial crew need to support something like 50 crew-days, while any crewed Mars mission likely needs >6500 crew-days.

Similar arguments can be made for massive increases in mission life requirements for every aspect of a Mars Starship.  The two have almost nothing to do with one another.

__________
¹My bet on the date for Starship launch/EDL crew-certification by NASA:  2033.  I guess it's possible that SpaceX could privately fly crews by 2031, but the consequences of killing a crew on a Starship that NASA wouldn't have certified are too terrible to contemplate, especially once they've gone public.  Elon may not know that, but I'll bet Gwynne does.

SLS cannot be cancelled before Artemis V, so NASA certification of lunar (and Mars) Starship by 2033 would be OK.

[yg1968]

CMPS will have to wait for another day.  And I wouldn't be surprised if NASA decides to go straight to the martian equivalent of HDL.

That is possible and I am OK with either Mars-HDL or CMPS. But I think that Isaacman should still try to start one of these programs this fiscal year.

I would much prefer that Isaacman dealt with a cislunar commercial crew program before setting up Mars programs.

I agree but I think that they go hand in hand. I think that SpaceX will have a commercial lunar alternative to SLS and Orion, once that their crewed Starship for Mars is ready. So starting CMPS and/or Mars-HDL program as soon as possible should help in that respect.

But you can reasonably use Crew Dragon as an interim solution for Earth->LEO and LEO->Earth of an SLS/Orion replacement, even before the EDL crewed Starship is available, so waiting for it keeps NASA wasting money on the inadequate SLS/Orion/Gateway based missions to the Moon.

It's not clear that a NASA CMPS program would cause SpaceX to accelerate development of a Crewed Mars architecture. It might even distort the SpaceX architecture and slow development if NASA's architectural concept differs.

My understanding is that CLPS has very few requirements. Presumably that would also be the case for CMPS.

Using F9 with Starship is a good idea but neither SpaceX and Congress seem very interested in it. I am not against it but I don't see it happening.

[Vultur]

While this isn't an argument to be cavalier with human life, I'm actually not convinced that there would necessarily be massive consequences.

Well, there's a ringing endorsement.  How unconvinced are you about the necessity of massive consequences?  pMassive < 10%?  < 50%? 

When you answer, imagine you have the fiduciary burden of being the CEO of a $1.5T public company.  You should also imagine that half the country hates you, and you have non-trivial political enemies

I'm not saying being careless is good. But I don't think NASA certification is meaningful for Non-NASA-astronaut flights. A good flight history is, but once that's built up, NASA's approval doesn't make it any safer.

One key question is how many flights you expect Starship to have had by 2031. If you take Musk's 'once an hour in the next 3 years, 4 at most' comments literally, it should have had something like 10,000+. I think that's completely insane, and in fact flatly impossible even if they got infinite money from an IPO - but I do think it's very likely that once they have multiple pads Starship launch cadence will scale faster than any rocket in history.

And it's just insane to rush ahead when you have a system that'll be much safer, albeit a bit more expensive.

Well, "much safer" is the question, isn't it? If Starship meets anything like its intended goals, it'll at some point be far safer than F9/D2, even though that's almost certainly the safest human spaceflight system ever.

The question (assuming Starship doesn't badly fail in some way) is how many flights it takes to get to that point.

NASA isn't necessarily a particularly good measure of safety.

[TheRadicalModerate]

I'm not saying being careless is good. But I don't think NASA certification is meaningful for Non-NASA-astronaut flights. A good flight history is, but once that's built up, NASA's approval doesn't make it any safer.

Is there some specific reason you think that the NASA certification criteria are too conservative?  I can see the paperwork requirements being more onerous than necessary, but that's what lawyers are for.  Are the actual engineering requirements wrong?

Note that a long flight history alone isn't what they're looking for.  They're looking for evidence of proper safety engineering, which relies not only on the empirical data for the reliability and failure modes for key parts and subsystems, but on understanding the implications of the failures and how they cascade through the system.  The first can be acquired through flight history.  But the second requires overcoming various failures of imagination, to produce a PRA that reflects the insights gained through failures in the flight history.

Again, this is gonna be different for Starship than it is for capsule-based systems, because all of the critical maneuvers and abort modes are going to be wildly different from the capsule systems.  Flight history is how you're going to flesh out those failures of imagination, but there's some judgment call that needs to be made that enough of them are included in the model that its results can be trusted.

[Vultur]

I'm not saying being careless is good. But I don't think NASA certification is meaningful for Non-NASA-astronaut flights. A good flight history is, but once that's built up, NASA's approval doesn't make it any safer.

Is there some specific reason you think that the NASA certification criteria are too conservative?  I can see the paperwork requirements being more onerous than necessary, but that's what lawyers are for.  Are the actual engineering requirements wrong?

Note that a long flight history alone isn't what they're looking for.

I wouldn't say "too conservative"; I'd say "designed for a low flight rate world". It's too conservative on some things, but too willing to allow other things.

The bolded is basically my issue ... A sufficient volume of real world data should be more valuable than PRAs.

In a world where no system has a high number of flights, you have to rely on PRAs. But I am not at all sure that this is a better method than flying a few hundred times before putting humans on board, if the latter is possible.

All that should need to be demonstrated is that the risk is equal or lower to other comparable HSF systems. (Dragon/F9 doesn't go to the Moon, so it's not really comparable. Being safer than Orion/SLS ... Shouldn't be difficult at all.)

I would also argue that this argument means maybe Starliner should just be dropped. It can't build up a sufficient flight history to be as safe as F9/Dragon in the limited time before the ISS is retired.  Unlike Orion/SLS vs F9/Dragon, the capability is the same.

[DanClemmensen]

Note that a long flight history alone isn't what they're looking for.  They're looking for evidence of proper safety engineering, which relies not only on the empirical data for the reliability and failure modes for key parts and subsystems, but on understanding the implications of the failures and how they cascade through the system.  The first can be acquired through flight history.  But the second requires overcoming various failures of imagination, to produce a PRA that reflects the insights gained through failures in the flight history.

But (presumably) rigorous and exhaustive analysis is also not enough unless the system is actually flown multiple times. Compare Starliner and SLS/Orion to D2/F9.

[catdlr]

Scott Manley
@DJSnM
·
2m
I see people asking why SLS is launching once every 3 years while the more complex shuttle that it was based on flew more often.
SLS Is still in development, that upper stage is an interim stage that needs replaced.
The funding may be a huge part of NASA’s budget, but it’s a lot less than the kind of money needed to fund development and flight operations when the contractors don’t have much incentive to go fast, in fact I expect some contractors make more by dragging out development.

https://twitter.com/DJSnM/status/2018795488611545481

[Coastal Ron]

Scott Manley
@DJSnM
·
2m
I see people asking why SLS is launching once every 3 years while the more complex shuttle that it was based on flew more often.
SLS Is still in development, that upper stage is an interim stage that needs replaced.
The funding may be a huge part of NASA’s budget, but it’s a lot less than the kind of money needed to fund development and flight operations when the contractors don’t have much incentive to go fast, in fact I expect some contractors make more by dragging out development.

Apparently Scott Manley has never worked in the government contracting world because otherwise instead of saying "in fact I expect some..." he would have just said "contractors make more by dragging out development." And this is because the SLS contract that Boeing has with NASA is COST PLUS, so NASA (i.e. taxpayers) are paying for all of Boeings overhead, and Boeing adds a profit on top of that.

I'm sure Scott Manley was trying to be "neutral" with this statement, but instead he looks like an apologist for Boeing.

Remember that when the SLS was created that Congress mandated that the SLS have "operational capability for the core elements not later than December 31, 2016". Then there has been the long list of dates that Boeing promised for when the SLS would be ready, none of which were close to being right. And did Boeing or its sub-contractors suffer because of the delays? No, they profited, both with contract rewards from NASA (i.e. U.S. Taxpayers) and from the profit they get from their Cost Plus contract.

And here we have the 2nd flight unit having the exact same problems the first unit did. Did Boeing learn nothing from the first flight? 

[TheRadicalModerate]

Note that a long flight history alone isn't what they're looking for.

I wouldn't say "too conservative"; I'd say "designed for a low flight rate world". It's too conservative on some things, but too willing to allow other things.

The bolded is basically my issue ... A sufficient volume of real world data should be more valuable than PRAs.

In a world where no system has a high number of flights, you have to rely on PRAs. But I am not at all sure that this is a better method than flying a few hundred times before putting humans on board, if the latter is possible.

With just a plain ol' binomial sample, to get pLOC = 1:270 (equivalent to pSuccessGoodEnoughNotToKillTheCrew = 99.63%) to fall into a confidence interval with a 90% confidence level requires about 800 consecutive successful flights.  Throw in 2 failures along the way, and now you need 1680 flights.

So if Starship suddenly goes to fairly high cadence (say 200/yr) and maintains a perfect safety record, you're still into early 2030 before you can empirically certify.  And if there are a couple of failures sprinkled in there, you're out to mid-2034.

I don't think you can have a trustworthy PRA for Starship with less than 100 flights with some interesting failures.  But you can get there a lot faster than via the empirical route.

Of course, one of the outcomes of a trustworthy PRA is that you don't have an adequately safe system.  Frankly, I think that's a moderately likely outcome.

All that should need to be demonstrated is that the risk is equal or lower to other comparable HSF systems. (Dragon/F9 doesn't go to the Moon, so it's not really comparable. Being safer than Orion/SLS ... Shouldn't be difficult at all.)

Well, now we come to the heart of the issue.  The whole "SLS/Orion uses lots of legacy hardware, so it's cheaper" line is obviously BS, but it is true that all that legacy hardware, arranged in a system that's pretty much like all the other legacy systems, means that the PRA suffers from very few failures of imagination, and therefore is trustworthy.

Does this guarantee that the SLS/Orion PRA is actually correct?  No, of course not.  But it provides the best evidence we have of a safe system, short of flying hundreds of SLS/Orion missions.  But Starship must fly hundreds of missions, even if we're going to eventually rely on a PRA.


An obvious, and completely legitimate, question to this argument is, "So how can HLS get by with a single landing test?"

I suspect (but don't know) that the answer to this is that the HLS Starship, because it neither launches nor EDLs with crew on it, is reduced to a purely in-space module, which removes all the weird flight modes that are required for successful Starship launch and EDL.

The big node in the PRA failure tree is that HLS Starship can abort a landing and get back to its rendezvous point pretty easily, so the actual pLOC can be quite low, even if pLOM is quite high.  That said, that's yet another reason why I think that the Option A HLS must have waist thrusters, if only to make the landing system redundant, and to afford more time and altitude to get the Raptors started for an abort.

[TheRadicalModerate]

Note that a long flight history alone isn't what they're looking for.  They're looking for evidence of proper safety engineering, which relies not only on the empirical data for the reliability and failure modes for key parts and subsystems, but on understanding the implications of the failures and how they cascade through the system.  The first can be acquired through flight history.  But the second requires overcoming various failures of imagination, to produce a PRA that reflects the insights gained through failures in the flight history.

But (presumably) rigorous and exhaustive analysis is also not enough unless the system is actually flown multiple times. Compare Starliner and SLS/Orion to D2/F9.

Yup.  But 100 flights with a PRA almost certainly beats 800-1700 flights without a PRA, with the proviso that the PRA must be able to generate a number that's simultaneously trustworthy and acceptable.

[Jim]

And here we have the 2nd flight unit having the exact same problems the first unit did. Did Boeing learn nothing from the first flight? 

It is not all Boeing.  SLS is not a launch service.  It is a vehicle with a NASA program office that manages and directs the engineering.
Also, Boeing or Marshall doesn't operate the vehicle or manage the GSE for SLS.  That is KSC and its contractor.  There are many to share the blame.

[TheRadicalModerate]

Apparently Scott Manley has never worked in the government contracting world because otherwise instead of saying "in fact I expect some..." he would have just said "contractors make more by dragging out development." And this is because the SLS contract that Boeing has with NASA is COST PLUS, so NASA (i.e. taxpayers) are paying for all of Boeings overhead, and Boeing adds a profit on top of that.

To be fair, most of the SLS and Orion contracts are either cost-plus-incentive or cost-plus-award contracts.  So if the contract administrators were doing their job properly, somebody would've jerked real hard on the reins quite a while ago, and the contractors would have had to perform to make their profit.

It's hard not to catch a whiff of corruption when you start sniffing at this.  Of course, the corruption was kinda the point, a fact the contract administrators are well aware of.

[Vultur]

Note that a long flight history alone isn't what they're looking for.

I wouldn't say "too conservative"; I'd say "designed for a low flight rate world". It's too conservative on some things, but too willing to allow other things.

The bolded is basically my issue ... A sufficient volume of real world data should be more valuable than PRAs.

In a world where no system has a high number of flights, you have to rely on PRAs. But I am not at all sure that this is a better method than flying a few hundred times before putting humans on board, if the latter is possible.

With just a plain ol' binomial sample, to get pLOC = 1:270 (equivalent to pSuccessGoodEnoughNotToKillTheCrew = 99.63%) to fall into a confidence interval with a 90% confidence level requires about 800 consecutive successful flights.  Throw in 2 failures along the way, and now you need 1680 flights.

So if Starship suddenly goes to fairly high cadence (say 200/yr) and maintains a perfect safety record, you're still into early 2030 before you can empirically certify.  And if there are a couple of failures sprinkled in there, you're out to mid-2034.

Well, two points:
 1) this is for the Moon, so 1:270 isn't the proper comparison. Safer than Orion/SLS is the proper comparison. That's worse than 1:100 even officially, and the real number is almost certainly worse than the official numbers. (1/year flight rate was supposed to be required for safety... That's not happening.)

2) I wasn't claiming early 2030, we were previously discussing 2033 for NASA.

2031 for non-NASA, but I'd think that for a private mission to the Moon 5% LOC (comparable to early Antarctic exploration) would be reasonable to accept.

Well, now we come to the heart of the issue.  The whole "SLS/Orion uses lots of legacy hardware, so it's cheaper" line is obviously BS, but it is true that all that legacy hardware, arranged in a system that's pretty much like all the other legacy systems, means that the PRA suffers from very few failures of imagination, and therefore is trustworthy.

I am ... Unconvinced. Artemis I heat shield issues were unexpected.

Is it better understood than Starship? Absolutely. Is that better understanding enough to make up for the inherent dangers of solids + super low flight rate + capsule that's stretching the size limits of capsule? I am very skeptical.

An obvious, and completely legitimate, question to this argument is, "So how can HLS get by with a single landing test?"

I find it hard to avoid the political / mandated side to this. SLS using solids is ok because it has to use solids. Putting humans around the Moon on the first test of thr ECLSS is ok because we don't have the budget and schedule for anything else. One test for a moon lander is ok because we don't have the budget and schedule for anything else.

[DanClemmensen]

An obvious, and completely legitimate, question to this argument is, "So how can HLS get by with a single landing test?"

I find it hard to avoid the political / mandated side to this. SLS using solids is ok because it has to use solids. Putting humans around the Moon on the first test of thr ECLSS is ok because we don't have the budget and schedule for anything else. One test for a moon lander is ok because we don't have the budget and schedule for anything else.

Certainly true for SLS/Orion, but less true for HLS. Additional uncrewed HLS test missions would be about $500 million each, and could be structured as cargo missions. This was infeasible before Starship was selected for HLS, and NASA declined to add such missions after the contract was awarded.

[Jim]

To be fair, most of the SLS and Orion contracts are either cost-plus-incentive or cost-plus-award contracts.  So if the contract administrators were doing their job properly, somebody would've jerked real hard on the reins quite a while ago, and the contractors would have had to perform to make their profit.

It's hard not to catch a whiff of corruption when you start sniffing at this.  Of course, the corruption was kinda the point, a fact the contract administrators are well aware of.

what is the saying; don't assume maleficence when incompetence is more likely

[hektor]

ESM is firm fixed price