S-N Fatigue Reliability Analysis of Highway Bridges

Abstract

This paper presents a method of calculating the expected fatigue failure probability of a structural detail, given the distribution of resistance and load. The resistance data, in terms of cycles to failure, come from previous laboratory tests. The load data come either from stress range histograms recorded on bridges or from loadmeter surveys. The proposed method replaces each histogram by an equivalent stress range and converts the latter into a distribution in terms of number of cycles. The problem is thus cast into the standard format for reliability analysis and allows one to calculate failure probabilities. Application of the method to designs in accordance with the AASHTO Specifications showed that fatigue failure probabilities for redundant load path (RLP) structures are inconsistent and vary greatly from PF = 9.2 × 10−2 for Category B to PF = 9.2 × 10−10 for Category E'. For nonredundant load path (NRLP) structures, they vary from PF = 5.1 × 10−2 for Category A to PF = 2.1 × 10−22 for Category E. It is proposed that the specifications be revised to include: (1) allowable stress ranges for RLP and NRLP structures with uniform failure probabilities; (2) explicit formulation of the specifications in terms of the actual number of single “fatigue trucks,” each causing an equivalent stress range; and (3) continuous definition of allowable stress range versus truck traffic volume. An example illustrates the design of a bridge, not covered by the AASHTO specifications, to a specified failure probability.