Time-variant fatigue reliability assessment of rib-to-deck welded joints using ANN-based methods

Abstract

Environmental corrosion and vehicle load significantly influence the fatigue damage of steel bridges. To efficiently investigate the combined effect of stochastic vehicle load and environmental corrosion on the fatigue damage of welded joints of orthotropic steel decks (OSDs), an artificial neural network (ANN) assisted assessment framework is proposed. A stochastic fatigue vehicle load model and an atmospheric corrosion model are first adopted for probabilistic modeling of stress response of rib-to-deck (RD) welded joints. To reduce the calculation cost of large-scale finite element analyses (FEA) runs in probabilistic simulation, the relation is generated between the equivalent stress range and variables (i.e., vehicle loads and corrosion depth) using ANNs. The time-variant fatigue reliability, therefore, can be calculated using Monte Carlo (MC) simulations by considering the uncertainties of variables in limit-state functions. The efficiency and feasibility of the proposed framework are demonstrated with a steel bridge. Finally, parametric studies investigate the influence of traffic volume and corrosion on the time-variant fatigue reliability of RD welded joints.