The problems of estimating a fatigue service life with a low probability of failure

✍ Scribed by Sam C. Saunders

• Publisher: Elsevier Science
• Year: 1976
• Tongue: English
• Weight: 954 KB
• Volume: 8
• Category: Article
• ISSN: 0013-7944
• DOI: 10.1016/0013-7944(76)90086-2

✦ Synopsis

An extensive collection of actual fatigue life data of certain qualified structural airframe components is used to show the inability of the standard life distributions to predict early failures. The difference between the fitted distribution and the real data occurs in the tails of the distribution. This disparity will be amplified in any calculation of the time to first failure in a large fleet of such components. A new family of distributions, called &normal and facilitating the comparison of log-life with the cumulative normal dis~ibution, is in~~uced. It is derived from considemtioff of the principles of fracture mechanics in crack advance from the time of initiation until ultimate size is reached. The family contains three parameters: location, scale and Aexure. For the type of distribution exemplified by the data, the flexure parameter cannot be estimated by the usual statistical estimation techniques and must be calculated from the frequency of initial flaws. When the flexure is known, maximum liielihood equations are derived, the solution of which yields estimates of the location and scale parameters even when the data is truncated.