Development of a reliability-based deflection limit state for steel girder bridges

A new approach is introduced for the prediction of maximum live load and load effect during the service life of a bridge, namely 75 years. Two extreme value models, the generalized extreme value (GEV) and generalized Pareto distribution (GPD), are considered for estimating maximum values for various return periods. The live load database consists of records collected from a weigh-in-motion (WIM) system installed at a New Jersey bridge for the purpose of this research. Future predictions of truck weights are compared to those based on other types of probability distributions.; The stochastic behavior of live load parameters, such as frequency of occurrence, gross vehicle weight, axle weights and spacings, are determined. Multiple presence statistics are also developed based on timestamp records from actual truck traffic. Dominant traffic patterns are identified and further classified with respect to vehicle class and total truck weight.; The proposed methodology is used to predict the maximum 75-year live load deflection using experimental measurements collected from a long-term LVDT-cable system installed at the bridge site and simulated records based on the developed statistics and the Monte-Carlo method. The effect of record length and distribution type are evaluated. Finally, the reliability index for the live load deflection serviceability limit state is computed.