Statistical failure analysis of aero gas turbines presents many special problems. Various difficulties must be overcome, including the paucity and form of available data with its heavily censored nature, selecting the right analysis method, the validity of the underlying modelling assumptions and the correct interpretation of the results.
This paper outlines the formal Rolls-Royce approach to quantifying numerical risk for safety and reliability, and the pivotal role played by the Weibull distribution. It reviews the generic forms of the censored failure data obtained from test and service environments. The limitations of the current analysis techniques are reviewed.
USDOD: Mil-Standard-882C. System Safety Program Requirements, January 1993.
2.
O'ConnorP TStatistics in quality and reliability. Lessons from the past, and future opportunitiesReliability Engng and System Safety J., 1991, 34 (1).
3.
KnezevicJModelling system reliability—a way aheadCODERM Newsletter, June 1995, 12 (2), 8–9 (UKMOD).
4.
CharlesLMIL-Handbook-217: Its time to re-think it!Electronic Des., October 1991.
5.
BazovskyIReliability Theory and Practice, 1961, p. 58 (Prentice-Hall, Englewood Cliffs, New Jersey).
6.
ThompsonJ MA method for optimising time between overhaul (T0) to minimise life cycle costs. In International Symposium on Advances in Logistics, Exeter University, December 1995.
7.
AbernethyRAFWAL-TR-83–2079 Weibull Analysis Handbook, November 1983 (Pratt and Witney).
NelsonWApplied Life Data Analysis, 1982 (John Wiley, Chichester).
10.
NelsonWWeibull analysis of reliability data with few or no failuresTrans. ASME, July 1985, 17 (3).
11.
JohnsonLThe Theory and Technique of Variation Research, 1964 (Elsevier, Oxford).
12.
CrockerJWeibull parameter estimation: a comparison of various methods. Rolls-Royce Military Aero Engines Limited, Filton, Bristol, Internal document published December 1996.
