Performance prediction and decision analysis in bridge management

This study presents performance prediction and decision analysis methodologies applied in the management of bridge infrastructure systems. The study aims to improve the effectiveness of four models essential to the network-level maintenance decision process, namely; (i) condition assessment, (ii) deterioration, (iii) condition prediction, and (iv) performance. An application of the Decision Analysis model for the evaluation of treatment strategies of large multiple-span bridges is also investigated in the study.; The study focuses primarily on methodological enhancements specific to a type of infrastructure management system previously developed for evaluating the maintenance needs of bridges located on the New York State Thruway's highway network. Probabilistic methods are applied in the study to model the uncertainty of functional performance and to evaluate risk in the bridge maintenance planning process. The presented methodologies feature: (i) an explicit definition of initial condition state, (ii) time-dependent, transition probability functions derived from historical condition data, (iii) reliability-based performance models, and (iv) a model that incorporates multiple-uncertainty and risk into the rehabilitation decision processes of independent large bridges. Demonstration of the new methodologies and analysis of the effects on bridge decision processes are presented in case studies of selected bridges located on the New York State Thruway.; Significant contributions to bridge infrastructure management resulting from this study include: (i) a characterization of bridge condition state with respect to the beginning of a planning horizon, (ii) a methodology for objectively evaluating bridge deterioration processes and estimating transition probability, (iii) condition prediction algorithms that can accommodate initial condition state and parametric distribution functions of transition time, (iv) a framework for defining effective performance measures, and (v) an alterative methodology for evaluating rehabilitation decisions of large-size bridges.; Important conclusions based on the findings of this study are as follows: (1) A precise definition of initial condition state is essential to the accuracy of prediction models. (2) Cumulative distribution functions of transition time data enable the development of more accurate deterioration models. (3) Reliability-based measures improve the characterization of functional performance in bridge maintenance decision processes. (4) Decision Analysis models enhance the programming of long-term treatment strategies for large bridges.