System Reliability Assessment Of Highway Bridge Structures

This post-doctoral report presented here is mainly based on the cooperation project work undertaken by the University of Surrey and sponsored by the Highways Agency on 'Advanced Reliability Assessment of Bridges'. The project was organized in two phases as follows:1. Review and technology transfer study on developments in system reliability methods for offshore structures and bridges2. Development of system reliability methodology using response surface method and non-linear finite element analysisThis report presents the findings of the second phase of this project. Within this phase the system reliability of composite bridges was addressed. Examples of bridge models for single span and continuous composite bridges were identified from the HA database of composite bridges and these formed the basis for the studies presented here.A methodology for system reliability analysis of composite bridges was developed in this project. The methodology combines the use of the response surface method with non-linear finite element analysis. This study aimed to develop the methodology using a representative general-purpose non-linear finite element program, which is widely used in the industry. Initial review study identified the central composite rotatable design (CCRD) method as an appropriate method for efficient experiment design within the context of response surface calibration.One of the key stages of this development was the development of a finite element model for a composite bridge and examining various modelling considerations. The model was developed using 3D beam and shell elements. The failure mode was examined and failure criteria were established. Some key parameters to control convergence of the solution were also examined to ensure that these were set at reasonable level.The FE model was used to perform a sensitivity study the aim of which was to identify the key parameters that need to be considered as probabilistic variables in the system reliability analysis. Three strength variables and three geometric variables were examined. Results of this study identified the three strength variables as the most significant and these were selected for probabilistic representation in the subsequent reliability analysis.A study of the loading model was undertaken to develop an appropriate loading model for the system reliability analysis. Existing traffic data were analyzed to identify the...