Location

La Jolla, CA

Start Date

1981 12:00 AM

Description

A methodology is evaluated to predict the probability of specimen failure with subsequent fatigue, after a short surface crack has been detected in Al 2219-T851 alloy. Cracks are detected and tracked to failure using optical microscopy. Predictions of remaining lifetime distributions are made with a Monte Carlo procedure in conjunction with growth laws which model the effect of grains of differing size, shape and crystallographic orientation in the crack path on propagation rate. Because the surface of the alloy cyclically hardens, the average rate of crack growth is less for cracks formed later during fatigue. The predictive methodology successfully describes this phenomenon, as well as predicts the probability of early failure arising from the statistical nature of the growth process, for failure probabilities substantially smaller than conveniently measureable in the laboratory.

Book Title

Proceedings of the ARPA/AFML Review of Progress in Quantitative NDE

Chapter

3. Retirement For Cause

Pages

24-31

Language

en

File Format

application/pdf

Share

COinS
 
Jan 1st, 12:00 AM

Remaining Fatigue Lifetime Prediction for Retirement-for-Cause in Metals

La Jolla, CA

A methodology is evaluated to predict the probability of specimen failure with subsequent fatigue, after a short surface crack has been detected in Al 2219-T851 alloy. Cracks are detected and tracked to failure using optical microscopy. Predictions of remaining lifetime distributions are made with a Monte Carlo procedure in conjunction with growth laws which model the effect of grains of differing size, shape and crystallographic orientation in the crack path on propagation rate. Because the surface of the alloy cyclically hardens, the average rate of crack growth is less for cracks formed later during fatigue. The predictive methodology successfully describes this phenomenon, as well as predicts the probability of early failure arising from the statistical nature of the growth process, for failure probabilities substantially smaller than conveniently measureable in the laboratory.