[SystemSafety] USAF Nuclear Accidents prior to 1967

[Peter Bernard Ladkin]

The Guardian today has an article on an accident to a US B-52 bomber in North Carolina in 1961. The 
aircraft, suffering a mid-air break-up, released two nuclear weapons, which were armed. One of the 
bombs was, according to a book by Ralph Lappe, "equipped with six interlocking safety mechanisms, 
all of which had to be triggered in sequence to explode the bomb. ...Air Force experts....found that 
five of the six interlocks had been set off by the fall! Only a single switch prevented the 24 
megaton bomb from detonating..."

This quote is contained in a short memo by Parker F Jones, an analyst at Sandia Labs, written in 
October 1969. He deprecates Lappe's general account but says that on this point he is correct; 
emphasises the vulnerability embodied by the switch, its type and function (it does not appear to 
have been adequately assessed for reliability in an accident scenario) and concludes that this type 
of bomb "did not provide adequate safety for the airborne alert role in the B-52." and footnotes 
that the "same conclusion should be drawn about present-day SAC bombs."

This is all contained in an article in The Guardian at 
http://www.theguardian.com/world/2013/sep/20/usaf-atomic-bomb-north-carolina-1961  Jones's memo is 
presented at 
http://www.theguardian.com/world/interactive/2013/sep/20/goldsboro-revisited-declassified-document

This is due to Eric Schlosser, who is about to publish a book called Command and Control. Schlosser 
has visited facilities, and so on, and gave an interview to The Guardian at 
http://www.theguardian.com/books/2013/sep/21/eric-schlosser-books-interview
Apparently, he made an FOIA request for all the incidents in the 10 years to 1967, and received 245 
pages of them.

Scott Sagan made similar inquiries in his 1993 book The Limits of Safety, for which he is justly 
famous. I didn't find the incident in Scott's book, so asked him if he knew about it. Scott's thesis 
in that book was testing Charles Perrow's Normal Accidents theory against the 
high-reliability-organisation theory of La Porte and colleagues.

The NA hypothesis is that tightly-coupled interactively-complex systems are unavoidably vulnerable 
to accidents which occur while everything is operating "as designed". The HRO theory says that there 
are certain characteristics of complex organisations which have proven to have had high reliability. 
One example of such an organisation is USN peacetime carrier operations (launching and retrieval of 
aircraft); another is Pacific Gas and Electric's nuclear power plant operations (which was a bit of 
a surprise to us who lived through part of the Diablo Canyon controversy).

USAF has obviously not had an accident in which a nuclear weapon has been accidently detonated. The 
question therefore was whether USAF SAC exhibited the characteristics of a La Porte HRO. Sagan 
argued that such accidents had been avoided through happenstance, and that the history rather 
supported the NA theory. It seems from the advance commentary that Schlosser's book will make a 
similar case.

PBL

-- 
Prof. Peter Bernard Ladkin, Faculty of Technology, University of Bielefeld, 33594 Bielefeld, Germany
Tel+msg +49 (0)521 880 7319  www.rvs.uni-bielefeld.de