Probabilistic fatigue life prediction employing an advanced crack propagation model

Abstract

Steel railway bridges are exposed to repeated train loads that often result in fatigue failure. As critical nodes of a transportation network, railway bridges should be designed and maintained to prevent fatigue failure for a predetermined period of time. However, as this would mandate accurate prediction of the fatigue life of railway bridges, the various sources of uncertainty associated with aging bridges, train loads, and the environment considerably complicate this task. In addition, for accurate fatigue life prediction, it is necessary to introduce an appropriate model for crack propagation. Although various studies have been conducted to predict the probabilistic fatigue life of steel railway bridges, many of them are based on a relatively simple model of crack propagation that assumes zero minimum stress and constant loading amplitude, thereby limiting the application scope and reducing the result accuracy. Thus, in this study, a new probabilistic method employing an advanced crack propagation model is proposed for the fatigue life prediction of steel railway bridges. A series of formulations are derived from an advanced crack propagation model such that probabilistic fatigue failure probability can be calculated through reliability analysis using the derived formulations as limit-state functions. Additionally, the formulations are derived such that the model can handle variable amplitude loading, which is the most common loading pattern for railway bridges. To demonstrate the proposed method, it is applied to a numeric al example of a steel railway bridge, and the effects of an advanced crack propagation model and various loading amplitudes on bridge fatigue life is discussed through a parametric study.