ESS-based probabilistic fatigue life assessment of steel bridges: Methodology, numerical simulation and application

Abstract

The selection of an appropriate S-N curve for a specified type of welded joints is an essential part for the fatigue reliability assessment of steel bridges by means of a stress-life approach. However, this is still a challenging issue in the real engineering applications due to the complexity of the structural components and the lack of the available numbers and types of S-N curves. In this study, a generalized fatigue reliability model based on equivalent structural stress (ESS) is proposed, which has the merit of taking no account of the selection of the S-N curve. An ESS-based probabilistic S-N formulation is first derived by introducing a probability index to the expression of the S-N curve. Then, the fatigue reliability model, which incorporates the random variables of the ESS range, the number of cycles and material properties, is deduced by combining the Miner’s rule. Numerical case studies and sensitivity analyses on the effects of the ESS range, the number of cycles and the growth in traffic volume are performed to evaluate the practicability and the effectiveness of the proposed fatigue reliability model. Finally, the proposed method is applied to an instrumented steel arch bridge with the aid of field dynamic strain data. The fatigue failure probability and the fatigue reliability index versus fatigue life are achieved through random samples generated by means of the Monte Carlo (MC) simulation on the basis of the joint probability density function (PDF) of random variables. The results show that the proposed method provides a viable way of resolving the difficulty in selecting a suitable category for the evaluation of the probabilistic fatigue life of the welded details in steel bridges. In addition, it can be found that both the PDF of stress spectrum and the growth of traffic volume are sensitive to the failure probability and reliability index obtained by the proposed fatigue reliability model.