Re: Man-Rating Atlas V

Will McLean wrote:
lou@xxxxxxxxxxx wrote:

Overall, if you apply any realistic factor for 'unexpected' failures,
the better expendables are likely safer alternatives. Of course this
is hard to quantify - if you *could* quantify it, they would not be
unexpected.


Atlas V hasn't flown nearly enough flights to say whether it has a
reliability of 2% or 5%. And the version they'd use to launch the CEV
hasn't flown at all. Estimates of its reliability are also educated
guessework.

I agree that these are all guesswork (after all, as someone once
observed, you get arguments and persecution over religion and politics,
not arithmetic). The question is the size of the errors. Suppose
Atlas has overestimated the reliability by a factor of 2, so in fact it
kills the crew 1 in 500 times, and does not meet the NASA spec. This
is certainly possible, and I think likely. However, it is even more
likely (in my opinion, of course) that NASA has overestimated the
reliability of the stick by a factor of 5 more more. This would result
in a fine record, killing the crew only 1 in 400 times, but it would
not be as good as the Atlas.

Now of course the size of the overestimates is unknown (otherwise we'd
include them in the estimates). But to me the Atlas changes "feel"
minor - better monitoring, new trajectory, redundant electronics, and
so on. Also, the proof of at least some structural margins by higher
performance unmanned launches is comforting. The stick changes seem
more intermediate to major - new engines, new solid configuration, and
so on. And the closer you get to a new vehicle, the less trustworthy
the reliability estimates become (For completely new vehicles the
estimates are terrible - Apollo overestimated by 50, the shuttle by
1000.) This of course is the whole point of making the stick "Shuttle
derived". But the reliability of "derived" is much less than "flight
proven" - see the Contour and Mars Observer missions for some excellent
examples. As the stick design progresses it seems less and derived,
and I fear NASA is continuing to use the reliability of "assemble
proven pieces" (when the reliability estimates are more or less OK)
when the the vehicle is approaching "major redesign", when the
reliability figures should be suspected to be serious over-estimates.

Overall, I suspect NASA is spending a few billion dollars, and a few
years, chasing a safety benefit which is illusory. Furthermore, it
seems much safer, schedule-wise, to modify an existing vehicle rather
than develop a new one - a process prone to serious delay. This has
safety implications as well - suppose the stick is delayed enough so we
need 10 more shuttle flights, as opposed to using a modified Atlas.
This is a (roughly) 5-10% crew loss penalty - one that would not be
made up for 100-200 flights of the new expendible, even assuming it is
twice as reliable as the Atlas.

Lou Scheffer

.

Relevant Pages

  • Re: NASA Picks In-Line (Maybe)
    ... > Lockmart's estimate for the failure probability of the Atlas V Heavy is ... Lockmart estimated the safety of one version of the ... including the Atlas V Heavy. ... >>> reliability was in fact one of the program goals. ...
    (sci.space.policy)
  • Re: What is the best brand of engine.
    ... have come to the conclusion that my Atlas engines seems to be the ... smoothest running, ... My Bachman is ... Durability and reliability are pretty good with every maker, and every maker is very good about repairing or replacing defective units. ...
    (rec.models.railroad)
  • Re: Space Access Update #112 9/19/05
    ... giving us a proven reliability rate in that stage of 99.54%. ... If we went with Atlas V, then we'd have to use solid strap-ons to up the performance, and if a single solid is a bad idea on the Stick, then using multiple solids on Atlas V is also a bad idea. ... As for SpaceX and it's Falcon 9....well, once they've got some Falcon 1 flights under their belt they might by taken more seriously in the EELV market. ...
    (sci.space.policy)
  • Re: 2006 Toyota Avalon
    ... If you actually think that a Toyota 20% or 30% more reliable than ... replaces his vehicle, with another new vehicle, in three to four years. ... It is a matter of reliability (needing>to fix ... MBs and other European cars are no longer known for high ...
    (alt.autos.toyota)
  • Re: Briefing on SRB based CEV at NPS with Scott Horowitz
    ... > mind the actual reliability numbers. ... about where the guidance system for the integrated vehicle would come ... that we've spent huge sums of money designing launch vehicles to ... NLS, OSP, blah blah blah) and trying yet again is not a good idea. ...
    (sci.space.policy)