Citation

Broek D, Rice RC, Steele RK. Transp. Res. Rec. 1978; 694: 2-10.

Copyright

(Copyright © 1978, Transportation Research Board, National Research Council, National Academy of Sciences USA, Publisher SAGE Publishing)

DOI

unavailable

PMID

unavailable

Abstract

The objective of this study was the development of a computational model for the prediction of fatigue-crack propagation in rails under train-service loading. Constant-amplitude fatigue-crack-growth properties were determined for 66 rail steels. The effects of mean stress, temperature, and crack orientation were investigated. Variable-amplitude tests showed almost no load-interaction effects in fatigue crack growth in rail steels. Thus a linear integration scheme could be used for crack-growth prediction. Service-simulation tests were performed on the basis of four measured load spectra. The specimens were subjected to random loading, train-by-train loading, loading that used a sequence associated with 170 trains of six different types, and unit-train loading. The load sequence represented 0.9 million gross metric tons (1 million gross tons) of traffic and was repeated until failure of the specimens. Crack growth in the service-simulation tests could be reproduced by using the computational prediction model within a factor of 2 and within a factor of 1.5 in most cases. Discrepancies between predictions and tests results are partially due to the variability of crack-growth properties of rail steels. The way in which the prediction model can be used in a reliability analysis for failure-rate prediction is discussed. Such an analysis would allow management decisions with regard to the most cost-effective means to reduce failure rates of a given track. This is possible because the relative accuracy of the crack-growth prediction is expected to be better than its absolute accuracy.