I think with the latest launch of the Atlas V, this brings us to 31 out of 32 missions with 100% success or 96.875% payload to orbit.
Quote from: BrightLight on 08/30/2012 06:23 pmI think with the latest launch of the Atlas V, this brings us to 31 out of 32 missions with 100% success or 96.875% payload to orbit.Or, considering that the customer considered it a success (it was a useful orbit), we can say 100% success, with an asterisk, just like how we say 133 out of 135 Shuttle flights were successful, even though STS-51F was an abort-to-orbit (engine out, so although it was a useful orbit and they did the rest of the mission, there was an underperformance of the launch system). STS-93, just like the Atlas V in question, also had a premature shutdown and ended up in a lower but still useful orbit, and just like Atlas V, the mission continued. But people RARELY call STS-51F and STS-93 actual mission failures, so the same standard should be applied to Atlas V.Either:Shuttle has a 98.5% success rate (133 successful, 2 flight failures) and Atlas V an 100% success rate (32 successful, 0 flight failures).ORShuttle has a 97% success rate (131 successful, 4 failures) and Atlas V has a 97% success rate (31 successful, 1 failure).Have to use the same standard. I prefer to use the same standard people usually apply to Shuttle, which is to say Atlas V has a 100% mission success rate (according to the customer).
Quote from: Robotbeat on 08/30/2012 06:49 pmQuote from: BrightLight on 08/30/2012 06:23 pmI think with the latest launch of the Atlas V, this brings us to 31 out of 32 missions with 100% success or 96.875% payload to orbit.Or, considering that the customer considered it a success (it was a useful orbit), we can say 100% success, with an asterisk, just like how we say 133 out of 135 Shuttle flights were successful, even though STS-51F was an abort-to-orbit (engine out, so although it was a useful orbit and they did the rest of the mission, there was an underperformance of the launch system). STS-93, just like the Atlas V in question, also had a premature shutdown and ended up in a lower but still useful orbit, and just like Atlas V, the mission continued. But people RARELY call STS-51F and STS-93 actual mission failures, so the same standard should be applied to Atlas V.Either:Shuttle has a 98.5% success rate (133 successful, 2 flight failures) and Atlas V an 100% success rate (32 successful, 0 flight failures).ORShuttle has a 97% success rate (131 successful, 4 failures) and Atlas V has a 97% success rate (31 successful, 1 failure).Have to use the same standard. I prefer to use the same standard people usually apply to Shuttle, which is to say Atlas V has a 100% mission success rate (according to the customer).In a statistical evaluation I try to be conservative thus, the 96.9 value. However, one can also easily claim LOM 0% (as you did and thus 32 out of 32 completed launches and payloads to useful orbits 100%. In the end, getting the payload up in one piece and in a productive orbit is what counts.
But I still don't think it's fair to count that one early shutdown against Atlas V as a full failure, since people don't do that for Shuttle.
Quote from: Robotbeat on 08/30/2012 08:10 pmBut I still don't think it's fair to count that one early shutdown against Atlas V as a full failure, since people don't do that for Shuttle.I do, in both cases. Fail is fail. Success is success. I measure launch vehicle hardware performance strictly as fail/succeed, because it either works or it doesn't.Centaur's RL-10 leaked propellant during coast - due to a faulty valve - and the stage ran out of propellant during its second burn, falling short of its planned final velocity. That's a vehicle failure, regardless of the mission outcome. (But FWIW, I've heard that the customer was not happy.)Others may keep track of mission fail/success, which is a different measurement. - Ed Kyle
Quote from: edkyle99 on 08/30/2012 08:44 pmQuote from: Robotbeat on 08/30/2012 08:10 pmBut I still don't think it's fair to count that one early shutdown against Atlas V as a full failure, since people don't do that for Shuttle.I do, in both cases. Fail is fail. Success is success. I measure launch vehicle hardware performance strictly as fail/succeed, because it either works or it doesn't.Centaur's RL-10 leaked propellant during coast - due to a faulty valve - and the stage ran out of propellant during its second burn, falling short of its planned final velocity. That's a vehicle failure, regardless of the mission outcome. (But FWIW, I've heard that the customer was not happy.)Others may keep track of mission fail/success, which is a different measurement. - Ed KyleBy this standard, no one would ever use redundant engines, because it's still considered a failure... In the data storage industry, that's a little weird... you don't count single drive failures (in a RAID 5 or 6 or 1 or 10, etc) as system failures even if there's a performance degradation, because your data still is intact and usable. But I digress.I can respect your view, Ed, because it's consistent, even if I think it's a little weird to count failures or underperformances as full failures when the spacecraft is intact and usable for its original purpose.
Quote from: Robotbeat on 08/30/2012 09:03 pmQuote from: edkyle99 on 08/30/2012 08:44 pmQuote from: Robotbeat on 08/30/2012 08:10 pmBut I still don't think it's fair to count that one early shutdown against Atlas V as a full failure, since people don't do that for Shuttle.I do, in both cases. Fail is fail. Success is success. I measure launch vehicle hardware performance strictly as fail/succeed, because it either works or it doesn't.Centaur's RL-10 leaked propellant during coast - due to a faulty valve - and the stage ran out of propellant during its second burn, falling short of its planned final velocity. That's a vehicle failure, regardless of the mission outcome. (But FWIW, I've heard that the customer was not happy.)Others may keep track of mission fail/success, which is a different measurement. - Ed KyleBy this standard, no one would ever use redundant engines, because it's still considered a failure... In the data storage industry, that's a little weird... you don't count single drive failures (in a RAID 5 or 6 or 1 or 10, etc) as system failures even if there's a performance degradation, because your data still is intact and usable. But I digress.I can respect your view, Ed, because it's consistent, even if I think it's a little weird to count failures or underperformances as full failures when the spacecraft is intact and usable for its original purpose.In verification and validation (V&V) there can be system performance (provider requirements)and mission performance (customer requirements) metrics. Very often they are combined in some set of requirements that the provider and customer agree upon. From a systems perspective, pass/fail is used to determine if the system i.e. the AV LV has performed its job, so far that's 31 out of 32 launches (97%). In contrast, the mission requirements might be more subjective as in "did the payload get to a usable orbit" and it would appear to be 32 out of 32 launches (100%). I didn't know the customer was not happy with NRO L-30 placement.
Counting historical failure rates is at best a rough way of determining a system's safety. For one thing once a failure happens the system is changed to minimize or eliminate the risk on future launches. This was very visible in the SSP with the SRB joint, and ET redesign after the losses. Actually there were a lot of other potential failure modes that got fixed on both those RTF efforts. SpaceX tweaked the Falcon 1 design until they got it right. So its not very meaningful to add all the launches of a rocket together in one statistical basket. The last launch of a rocket or program is going to be safer then the first.
Its also assuming that all payloads and orbits are the same. Higher orbits require longer burn times, more engine starts, and sometimes more stages. There are a few recent Proton launches which would have been successful if they were launched into LEO. The failures happened in the Breeze-M or Block-DM stages.