Abstract
This paper presents the probabilistic procedure for estimating fatigue resistance of highway bridge cable to vehicle loads, which takes account of the stochastic properties of both live loads and cable strength. The statistical model of the live loads can be constructed on the basis of the field measurements of traffic loads previously performed in the usual traffic conditions on actual bridges. The ranges and frequencies of cable stress amplitudes are directly estimated from its random wave samples produced by the live loads with the aid of the associated influence lines, and it can be shown that the exponential distribution function is very well fitted to the histogram. Miner's rule is applied to fatigue analysis, in which the fractile of strength distribution can be used to estimate the probability of fatigue failure. Examples are given for the cases of a cable-stayed bridge.
