Hacker News Reader

Comment by skolos | Hacker News Reader

>The Government Accountability Office’s preliminary findings reveal a pattern of problems with turbine blades that pump fuel into rocket engines

>The crack-prone parts are considered a potentially major threat to rocket safety, the industry officials said, and may require redesign of what are commonly called the Falcon 9’s turbopumps.

>A SpaceX spokesman said “we have qualified our engines to be robust” to such cracks but are “modifying the design to avoid them altogether.” The pending changes “will be part of the final design” for the Falcon 9, he added, “in partnership with NASA to qualify engines for manned spaceflight.”

View on HN

Another interesting bit:

>For Boeing, these officials said, GAO investigators—among other items—raised questions about the status of tests to determine the reliability of its parachute systems designed to help returning manned capsules land safely.

>The GAO also has determined that both companies face an uphill struggle to meet NASA’s statistical goal of no more than one projected astronaut fatality in 270 flights, industry officials said.

That's a weirdly specific, yet round number.

I suspect it is not so much round as rounded. You don't really need more than two significant figures for the number.

My "expected value" knowledge has atrophied, but I'm guessing this is just a weird way of saying a success rate of 99.5% or something. 0.995^135 is about 0.5, so you blow up on average half the time in 135 flights, or once every 270 flights.

> no more than one projected astronaut fatality in 270 flights

> success rate of 99.5% or something. 0.995^135 is about 0.5

Method seems pretty good, but unfortunately that math would only work if there was exactly one astronaut per flight. Per Wikipedia, the capacity is up to 7 crew members, so I reckon at full capacity, that would be 1 full vehicle failure with 7 lives lost in 1,890 flights.

My math is a bit rusty too though.

https://en.wikipedia.org/wiki/Dragon_2

Not a rocket buff but I thought the soyuz rocket (R-7) was one of the most proven and reliable rockets there are. In 60 years that family has had 1,859 launches[1].

So statistics aside it's interesting just how high those standards are if you bore them out. Going to look stupid with some maths here; If spacex launch every 2 weeks which I thought was the goal, that's like 72 years to prove that reliability record.

72.69 = (1890 * 2) / 52

Number of 1 week periods divided by weeks in a year.

1: https://en.wikipedia.org/wiki/R-7_(rocket_family)#Summary_of...

Wasn't NASA's record with the space shuttle more like 1 in 2.5 fatalities per flight? 14 fatalities in 135 launches?

Must have been good enough for government work.

with modern material and computation advances it would be reasonable to raise the bar

They're asking for nearly 2 magnitudes better safety than they could provide. Granted the space shuttle was dumb from the beginning and we'd be flying around in our flying cars and vacationing on Mars if they'd stuck with the far superior Saturn launch system. But who could of known that, besides the plethora of scientists who pointed this out?

Except that the options at the time were space shuttle or NOTHING--not space shuttle or Saturn V.

NASA threw in with the only folks willing to fund spaceflight--the military. And the military demanded an orbiter that could do circumpolar orbits--which was totally crazy from an engineering standpoint.

What does it mean for an orbiter to support circumpolar orbits, and why is it crazy?

It means being able to reach more kinds of orbits, which adds requirements. So does reentry capability. Why add those two sets of requirements to the same vehicle, see?

circumpolar doesn't make sense as an orbit. The crazy shuttle design requirement was to return 40 tonnes from a single polar orbit in 1 pass. I.e bring a spy sat back to america without flying over russia.

This means a massive body to hold an object, The earth would move under the craft as it rotated. Without being able to wait for it to rotate around you need wings for the large cross-range capability to glide back to america.

As KSP teaches us you can't put massive wings at the top of a rocket, without even bigger wings at the bottom. But that's got too much drag to get to space and so, the insane launch system that was the shuttle is invented.

They never used this ability for the entire service life of the shuttle.

We already had the Saturn. If you're saying NASA's other choice was nothing, then it was purely political and if they'd stuck to the science the Air Force would have been told to suck it eventually. Instead NASA wasted 30 years of rocket science on a giant turd that was far more expensive than Saturn and killed a person ever 2.5 launches.

> If you're saying NASA's other choice was nothing, then it was purely political and if they'd stuck to the science the Air Force would have been told to suck it eventually.

Yes, it was political. And it wasn't at all impossible that they would just collapse NASA's budget (this was in the recessions of the 1970's), and NASA wouldn't even exist today.

It's supposed to be achievable precisely because the shuttle was such a dumb/difficult/unwieldy concept no matter how it was implemented. SpaceX and Boeing get to have abort capability all the way to orbit and inherently simpler capsules.

To torture the shuttle comparison, they shouldn't get damaged by a piece falling of their launching rockets because they get to sit on top and keep their heat shields protected during launch, and they shouldn't be destroyed if the booster explodes because they have launch escape systems. The only two shuttle failures causing fatalities were kind of inherent to shuttle.

Considering SpaceX plans to launch up to 100 people at a time to Mars with no real abort possible, I think its quite reasonable to ask them for that level of safety on a system with an abort.

I think that's why they stopped doing manned missions - it was too dangerous. This episode of Gimlet's Undone kind of talks about that - https://gimletmedia.com/episode/the-columbia/

> 1 in 2.5 fatalities per flight?

No. How did you reach this number?

> 14 fatalities in 135 launches?

Yes. Though I think it makes more sense to read this as 2 losses in 135 flights (98.5% safety rating), or 14 fatalities among 817 passengers (98.2% safety rating).

(The per-passenger safety rating is slightly less because the only losses carried 7-passenger crews, while about half the flights carried 5 or 6 passenger crews. I didn't include the non-orbital test flights carrying crews of 2, which were arguably more dangerous than orbital flights, and I tried to handle Mir transfers and ISS crew exchanges sensibly.)

Historically, failures in spaceflight result in a loss of all crew (or a loss of no crew, if they have abort capabilities). This suggests an easy way to artificially improve the launches-per-fatality metric would be to reduce the number of crew per flight, so I'd prefer to use numbers for failures-per-launch or fatalities-per-passenger.

That's not the bar NASA is setting for SpaceX. They are stating fatalities per launch.

Reply on news.ycombinator.com