Newsgroups: comp.risks
X-issue: 13.28
Date: Sun, 15 Mar 92 20:11:54 EST
From: henry@zoo.toronto.edu (Henry Spencer)
Subject: X-15 reliability experience

On reading the Proceedings of the X-15 First Flight 30th Anniversary
Celebration (NASA CP-3105, Jan 1991), I ran across a section of some relevance
to Risks.  Insertions in [] are mine.

	In 1962, a very comprehensive, but little known, study was
	initiated by Bob Nagle at AFFTC to quantify the benefits of
	having a pilot and redundant-emergency systems [this seems to
	be essentially a buzzword for "redundant systems"] on a research
	vehicle.  Each individual malfunction or abnormal event that
	occurred after B-52 [X-15 launch aircraft] takeoff for the
	first 47 free flights of the X-15 was analyzed.  The outcome
	of each event was forecast for three hypothetical models;
	one with only the pilot but no redundant-emergency systems,
	one with only the redundant-emergency systems but with no pilot,
	and one with neither the pilot nor redundant-emergency systems
	(i.e. single-string [buzzword for no redundancy], unmanned).

	[The bar chart of results shows an expected failure rate of over
	50% for the "neither" configuration, with many of the failures
	destroying aircraft.  Adding just a pilot or just redundant systems
	produces only small improvements.  Adding both takes the failure
	rate down to near zero and eliminates aircraft losses.]

	[Referring to the graph.]  The unmanned, single-string system
	would have had 11 additional aborts and resulted in the loss of
	15 X-15s.  [The actual program built only three!]  Not surprising
	is the fact that the pilot is of little value in a system
	without redundant-emergency systems.  He must have some alternate
	course available in order to be effective.  The redundant-emergency
	systems were also found to be of little value in an unmanned
	system primarily because the fault detection and switchover logic
	must presuppose the type of failure or event.  For example, few
	designers would have built in a capability to handle an
	inadvertent nose gear extension at Mach 4.5.

	[That last refers to something that actually happened to an X-15.
	Landing gear is normally designed to be extended at a maximum
	of a few hundred MPH.  Having gear extend at 3000+ MPH is a
	horrifying prospect, but the X-15 was landed safely with minor
	damage to the aircraft and the pilot unhurt.]

	Of more than academic interest was a parallel, but independent,
	study conducted by Boeing on the first 60 flights of their
	BOMARC missile, an unmanned, single-string, ramjet-powered
	interceptor.  The authors collaborated on the ground rules for
	the study but not on the actual analysis.  The similarity of
	the results [a virtually identical bar chart] is striking,
	especially when considering that the X-15 study was projecting
	from a piloted, redundant design to an unpiloted, nonredundant
	design, and the BOMARC study was the reverse...

("X-15 Contributions to the X-30", Robert G. Hoey, pp 103-121.)

Henry Spencer at U of Toronto Zoology    henry@zoo.toronto.edu    utzoo!henry