Probabilistic Considerations for Growth and Detection of Fatigue
Cracks and Impact-Generated Delaminations in Aircraft Structures
Jan D. Achenbach
McCormick School of Engineering and Applied
Science
Northwestern University
Evanston, IL 60208 USA
for presentation at
2013 Far East Forum on Nondestructive
Evaluation/Testing: New Technology & Applications
June 18-21, 2013,
Jinan, China
Diagnostic techniques provide the input
for prognostics. Modeling of constitutive properties, supported by experimental
results, provides damage growth laws which in turn provide information on
damage evolution and remaining life. Depending on its magnitude, the resulting
statement of failure probability may either result in a recommendation for
repair or replacement of a structural component, or for an additional cycle in
the diagnostics/prognostics loop of the structural health management system.
In
this lecture, I will devote particular attention to the probabilistic aspects
of diagnostics and prognostics. Probabilistic considerations play a dominant
role in the four stages of the diagnostics and prognostics of fatigue damage in
metals. Considerable attention has been given to the evolution and detection of
pre-crack fatigue damage and probabilistic aspects of subsequent macrocrack
formation (Stage 1). For Stage 2 (macrocrack growth and detection), Paris’s law
for crack growth under cyclic loading conditions can be useful, particularly if
placed in a probabilistic context. By introducing the probability of detection
concept, various probabilities related to the existence, after N cycles, of a
crack larger than critical size, some based on a Bayesian approach, can be
determined in Stage 3 for the purposes of prognostication. An example is given
for fatigue cracks emanating from rivet holes in a lap joint. An analogous
probabilistic approach will be presented for the generation of delaminations in
laminated fiber-epoxy composites by repeated impact loads.
Significant
progress in QNDE has been achieved. On the other hand, SHM has not yet broken
through in a big way. SHM systems are often not yet affordably maintainable
with near-zero false alarm rates. Huge benefits can, however, be achieved if
SHM can justify reduced design margins, longer life spans and reduced service
interruptions.
|