Author Topic: Launch failures - industry impacts  (Read 11007 times)

Offline pippin

  • Regular
  • Senior Member
  • *****
  • Posts: 2575
  • Liked: 312
  • Likes Given: 45
RE: Launch failures - industry impacts
« Reply #20 on: 09/07/2007 05:37 pm »
Quote
edkyle99 - 7/9/2007  7:23 PM

Quote
mr.columbus - 7/9/2007  2:31 AM

Quote
edkyle99 - 6/9/2007  9:50 PM

I hate to mention it, but these failure things even out over time.  Arianespace seems to be flying high right now, since it hasn't had a failure since 2002.  Good for Arianespace, but with 19 consecutive successes (if I'm counting right), Ariane 5 is now starting to race against the statistics of launch vehicle failure.  Ariane 5 could fly without error for years more before the next failure, but it is still a race that no launch vehicle has ever won.

- Ed Kyle

Ed, the Long March rocket family has had 59 consecutive successful rocket launches (since 1996). The chance of another failure of Ariane 5 is remote, of course there is a chance at every launch, however I would not be suprised if we see another 50 launches before another failure. Ariane is a very reliable rocket.

Ariane 5 has had 4 failures in 33 flights.  That is a 12% realized failure rate, with enough samples to be able to state, with 90% confidence, that Ariane 5 is no more than 95% reliable.  A 95% maximum reliability level means that, on average, at least one out of every 20 launches will fail.  Ariane 5 could very well extend its string of successes well beyond 20, or 50, etc., but the odds are against it.  There is less than an 8% chance that Ariane 5 will reach 50 consecutive successes, assuming 95% reliability, and less than a 0.2% chance if 88% reliability is assumed.

 - Ed Kyle

You are mixing ex post and ex ante analysis.
Fact 1: you don't know the long term success rate yet.
Fact 2: assuming your assumption (95% success rate) is correct (see also: Fact 4), that means the chance a flight will fail is 5% for each flight. Realized success rates in the past have no influence of the chances for failure.
Fact 3: Since you only deduct your success rate from the success/failure sample a long success streak could also mean, your initially assumed success rate was too low. You can't know whether the failures you did see were a cluster not consistent with the underlying success rate.
Fact 4: Statistically speaking, a confidence level of 90% is equal to "I don't know". Useful confidence levels start at 95% at least (and here you need twice the sample, first).
Fact 5: That was all only on statistics and did not include fundamental aspects like things that lead to failure in the past getting fixed or more experience being gained with the LV.

Offline meiza

  • Expert
  • Senior Member
  • *****
  • Posts: 3067
  • Where Be Dragons
  • Liked: 5
  • Likes Given: 3
Re: Launch failures - industry impacts
« Reply #21 on: 09/07/2007 05:53 pm »
I think it's easier to contrast it this way:
by pure statistics, a launch vehicle with a long success streak is no more likely to crash than a vehicle that has just crashed on the previous flight, when they have similar reliability. Ie the "crash odds" don't "pile up", since the samples are independent.

Offline sammie

  • Veteran
  • Full Member
  • ****
  • Posts: 553
  • Liked: 1
  • Likes Given: 0
Re: Launch failures - industry impacts
« Reply #22 on: 09/07/2007 05:59 pm »
Can anyone elaborate on McDew's comment on the near failure of Fregat during the Globestar Mission, it's the first time I heard of it.

There is going to be backlog of large ComSats in the near future. I can imagine that the Immarsat-4 that switched from Atlas 5 to Proton recently will be changing back, as it had an option on this ride.

"The dreams ain't broken downhere, they're just walking with a limp"

Offline anik

  • Global Moderator
  • Senior Member
  • *****
  • Posts: 7776
  • Liked: 955
  • Likes Given: 368
Re: Launch failures - industry impacts
« Reply #23 on: 09/07/2007 06:45 pm »
Quote
sammie - 7/9/2007  9:59 PM

Can anyone elaborate on McDew's comment on the near failure of Fregat during the Globestar Mission, it's the first time I heard of it

There was problem with sensor (or telemetry), which showed zero gas pressure in tanks of pressurization system after performing of the first burn by Fregat upper stage... The second and third (deorbit) burns were performed successfully...

Offline edkyle99

  • Expert
  • Senior Member
  • *****
  • Posts: 15391
    • Space Launch Report
  • Liked: 8566
  • Likes Given: 1356
RE: Launch failures - industry impacts
« Reply #24 on: 09/07/2007 08:21 pm »
Quote
pippin - 7/9/2007  12:37 PM
Fact 1: you don't know the long term success rate yet.
Nor will I until the last Ariane 5 has flown, thus the need to provide an estimated reliability.
Quote
Fact 2: assuming your assumption (95% success rate) is correct (see also: Fact 4), that means the chance a flight will fail is 5% for each flight.
Right
Quote
Realized success rates in the past have no influence of the chances for failure.
'Gotta disagree with this assertion.  The past does provide lessons for the future.  NASA and DoD both use reliability estimating methods that initially are based on "a comparison to historical data from previous launches of vehicles developed and launched in similar circumstances".  
http://www.faa.gov/about/office_org/headquarters_offices/ast/media/FAA_AST_Guide_to_Reliability_Analysis_v1.pdf
Quote
Fact 3: Since you only deduct your success rate from the success/failure sample a long success streak could also mean, your initially assumed success rate was too low. You can't know whether the failures you did see were a cluster not consistent with the underlying success rate.
Historically, launch vehicle demonstrated reliability increases during the first dozen or so flights before settling near the final value.  That is why I guessed at a high number (95%) rather than using the current 88% demonstrated reliability.  
Quote
Fact 4: Statistically speaking, a confidence level of 90% is equal to "I don't know". Useful confidence levels start at 95% at least (and here you need twice the sample, first).
Fact 5: That was all only on statistics and did not include fundamental aspects like things that lead to failure in the past getting fixed or more experience being gained with the LV.
[/quote]
Launch vehicle reliability doesn't provide large samples on an individual launch vehicle basis, so the confidence level has to be lower than with other conventional statistical situations.  There is uncertainty, yes, but it isn't the same as "I don't know", IMO.  

Large sample sizes are available for space launch vehicles in general.  There have been more than 4,800 orbital launch attempts during the space age, about 91% of which succeeded.  Since 1980, there have been 2573 orbital attempts and 2428 successes - a 0.944 success rate.  (As much as we wish it weren't true, launch vehicle reliability has not changed much since 1980.)  A 95% confidence level requirement would produce a plus or minus 1% confidence interval, roughly, around that 94.4% demonstrated number.  

- Ed Kyle

Offline HIPAR

  • Member
  • Full Member
  • ****
  • Posts: 585
  • NE Pa (USA)
  • Liked: 1
  • Likes Given: 0
Re: Launch failures - industry impacts
« Reply #25 on: 09/07/2007 08:55 pm »
Are all Ariane rockets identical?  I would think new design weaknesses are discovered for each launch.  Perhaps the software is improved.  These kinds of things lead to incremental improvements which present difficulties with a reliability analysis because the basic reliability of the vehicle improves with each launch experience.

---  CHAS

Offline meiza

  • Expert
  • Senior Member
  • *****
  • Posts: 3067
  • Where Be Dragons
  • Liked: 5
  • Likes Given: 3
Re: Launch failures - industry impacts
« Reply #26 on: 09/07/2007 09:07 pm »
Ed, the misconception is that a vehicle that has just had a failure is somehow more reliable than one that has not had one, because the one that has not had would have a problem like this: "Ariane 5 could very well extend its string of successes well beyond 20, or 50, etc., but the odds are against it." Every launch starts a new string so to speak. A succesful backlog of X launches does not mean that the future string of Y succesfull launches is dependent on probabilities calculated with X+Y as input, which would favor launchers with small X: no, you calculate with Y and Y only.

Offline pippin

  • Regular
  • Senior Member
  • *****
  • Posts: 2575
  • Liked: 312
  • Likes Given: 45
RE: Launch failures - industry impacts
« Reply #27 on: 09/07/2007 09:22 pm »
Quote
edkyle99 - 7/9/2007  10:21 PM

Quote
pippin - 7/9/2007  12:37 PM

 Realized success rates in the past have no influence of the chances for failure.
'Gotta disagree with this assertion.  The past does provide lessons for the future.  NASA and DoD both use reliability estimating methods that initially are based on "a comparison to historical data from previous launches of vehicles developed and launched in similar circumstances".  
http://www.faa.gov/about/office_org/headquarters_offices/ast/media/FAA_AST_Guide_to_Reliability_Analysis_v1.pdf

- Ed Kyle

As meiza stated: the one thing has nothing to do with the other. Of course, you use historical data to FIND OUT about the reliability of a system but that does not mean, that this history INFLUENCES the chances of future flights.

The UNDERLYING reliability of course has an effect on the chances for a flight and that is what you want to assess using analysis of historical data. But this is not part of the sample, it's the "real" reliability while all your statistical measures are only ESTIMATIONS of the reliability.

Oh, and just the fact that you don't have a better measure does not make 90% confidence a good one. For example, there is no chance to judge about a 99% success rate with 95% confidence level because there simply have not been enough launches until today. So: yes it is pretty much equal to "don't know".

Offline McDew

  • Regular
  • Full Member
  • **
  • Posts: 270
  • Liked: 110
  • Likes Given: 51
Re: Launch failures - industry impacts
« Reply #28 on: 09/07/2007 10:16 pm »
Quote
anik - 7/9/2007  2:45 PM

Quote
sammie - 7/9/2007  9:59 PM

Can anyone elaborate on McDew's comment on the near failure of Fregat during the Globestar Mission, it's the first time I heard of it

There was problem with sensor (or telemetry), which showed zero gas pressure in tanks of pressurization system after performing of the first burn by Fregat upper stage... The second and third (deorbit) burns were performed successfully...

Anik, Not sure if your comments are implying that the pressure was fine and it was just a sensor error.  My sources informed me that the "actual" pressure for second burn was only about 25% of the mininum specification for operation of the engine.  The customer was greatful that they dodged a bullet and acknowledged that it should have been a failure.  Information identifying that an anomaly occurred on the mission become public about a month ago and was linked to the continued delay of the Fregat stage for next Globalstar mission.  Of course Starsem declared the mission a complete success (which it was), but has not publicly acknowledged any details on the near failure.

Offline @RD170@

  • Member
  • Member
  • Posts: 65
  • Liked: 0
  • Likes Given: 0
Re: Launch failures - industry impacts
« Reply #29 on: 09/08/2007 12:06 am »
I think reliability depends of the moment of a LV is living.

Take some examples:

Arianne 5 failed at the begining of development, a power plant problem ( vulcain I ), this was issolated and changed by Vulcain II
And another problem with upper stage. And solved again. So the actual Ariane isnīt the same than nowadays.

So reability depends of several things like: invest money in security systems and check health of LV, to take the best brains of business like Glusko, Korolev or V. Braunn in his times.

So reability isnīt only a cuestion of numbers. Is a cuestion of confidence.

In what car you trust a german car o a vietnam car ŋ?.

Offline anik

  • Global Moderator
  • Senior Member
  • *****
  • Posts: 7776
  • Liked: 955
  • Likes Given: 368
Re: Launch failures - industry impacts
« Reply #30 on: 09/08/2007 07:42 am »
Quote
McDew - 8/9/2007  2:16 AM

Not sure if your comments are implying that the pressure was fine and it was just a sensor error

Yes, that is what I meant... I was told that it was problem with sensor or telemetry...

Quote
McDew - 8/9/2007  2:16 AM

My sources informed me that the "actual" pressure for second burn was only about 25% of the mininum specification for operation of the engine

Hmm... I heard this too, but hoped for the best and thought that it is mistake... :( Thanks for confirmation...

Quote
McDew - 8/9/2007  2:16 AM

Information identifying that an anomaly occurred on the mission become public about a month ago and was linked to the continued delay of the Fregat stage for next Globalstar mission

I was told that the postponing of this launch (from July to September and then to October) is related to explosion of chemical battery during testing of Fregat upper stage in NPO Lavochkin...

Offline McDew

  • Regular
  • Full Member
  • **
  • Posts: 270
  • Liked: 110
  • Likes Given: 51
Re: Launch failures - industry impacts
« Reply #31 on: 09/08/2007 12:09 pm »
Quote
anik - 8/9/2007  3:42 AM

Quote
McDew - 8/9/2007  2:16 AM

Information identifying that an anomaly occurred on the mission become public about a month ago and was linked to the continued delay of the Fregat stage for next Globalstar mission

I was told that the postponing of this launch (from July to September and then to October) is related to explosion of chemical battery during testing of Fregat upper stage in NPO Lavochkin...

I heard they were both causing delays.  Not sure which problem is currently driving the schedule.  Batteries can be a pain to recertify!!!  It's understandable to focus the blame of schedule delays on a battery explosion "during ground testing" and avoid discussion of a "flight anomaly".

Offline Seer

  • Member
  • Full Member
  • **
  • Posts: 251
  • Liked: 6
  • Likes Given: 1
Re: Launch failures - industry impacts
« Reply #32 on: 09/11/2007 12:46 am »
Spacex have just announced a launch order for falcon 9. Does anyone know whether this was connected to the Proton failure?

Offline McDew

  • Regular
  • Full Member
  • **
  • Posts: 270
  • Liked: 110
  • Likes Given: 51
Re: Launch failures - industry impacts
« Reply #33 on: 09/11/2007 01:50 am »
Quote
Seer - 10/9/2007  8:46 PM

Spacex have just announced a launch order for falcon 9. Does anyone know whether this was connected to the Proton failure?

Nothing to do with Proton.  Spacecraft is small, only about 2-3 tons.  Customer has funding problems (ie. no $'s) and could not afford either Land Launch or Soyuz.  Cheapest thing on the market is the unproven Falcon 9.

Tags:
 

Advertisement NovaTech
Advertisement Northrop Grumman
Advertisement
Advertisement Margaritaville Beach Resort South Padre Island
Advertisement Brady Kenniston
Advertisement NextSpaceflight
Advertisement Nathan Barker Photography
0