Jim - 5/2/2007 8:00 AM
1982-2006 (don't know when exactly in 2006)
DOD 164 out of 176 = 93.2%
Commercial 105 out of 116 = 88.2%
NASA 110 of 113 = 97.3%
I was wondering about this, so I did a little bit of digging. I found the following document, which lists NASA-managed ELV launches during the CY1987-2005 period.
http://corport.hq.nasa.gov/launch_services/Launch_History.pptIt shows 74 NASA-managed launches during that time-span, with 73 successes. That's 98.6%, which on the face of it is better than pretty good and is clearly better than the average for U.S. ELVs during that span.
But then I noticed that four launch vehicles (Scout, Pegasus, Delta, and Atlas 2A(S)) accounted for 65 of the launches. That's a pretty reliable bunch with or without NASA. During the years that NASA used these four launch vehicles, they performed a grand total of 219 launches for all users with 216 successes (98.86% success). Statistically, NASA should have suffered about one failure in 65 uses of these launchers, which is exactly what happened when one of NASA's 12 Pegasus launches failed.
Where NASA seems to have either squeezed more reliability out of its launch vehicles or to have simply gotten lucky was with its four Atlas I launches and its five single launches on five different launch vehicles during the period (Atlas Centaur, Titan 3 Commercial, Titan 401B, and Titan 23G). All nine of these flights were successful, when statistics suggest that NASA should have expected to lose at least one, and possibly two.
Where NASA seemed the most lucky, or smart, was with Atlas I. NASA used the first Atlas I to launch CRRES in 1990. NASA didn't use Atlas I again until 1994. During the intervening period, three of five Atlas I launches for commercial and DoD customers failed. Was this luck, or did NASA deftly hold back until commercial and DoD launches had worked out the launch vehicle's problems?
NASA may have used this "wait until proven" approach with Pegasus too. It didn't start "managing" (providing technical oversight for) Pegasus launches until 1996. Pegasus had suffered four failures from its inception through 1995. Since 1996, inclusive, Pegasus has only failed once.
Lucky, or smart? Perhaps a bit of both, but I think that NASA should not fool itself into believing that launch vehicles work better simply because NASA managers sit in on the launch review meetings. It is prudent to let launch vehicles prove themselves before putting civil space payloads on them, but that means that the failures go on someone else's record.
- Ed Kyle