Evaluation of network-level bridge management system

A network level bridge management system (BMS) is a decision support tool that supplies analyses and summarizes data, uses mathematical models to make prediction and recommendations, and provides the means by which alternative policies and program may be efficiently considered (Federal 1993). It is intended to optimally distribute the available bridge funding resources and to provide the best overall level of service, user cost savings, and safety throughout the network. An optimization model is a component of a BMS that finds a policy that minimizes the long-term cost for all bridge elements in the network.; In an effort to establish a network level bridge management system, almost every facet of current bridge management practice must change. The network level BMS modifies the existing procedures for collecting, processing, and updating data. Most important, it changes the procedures for predicting deterioration, identifying alternative actions, and recommending the optimal actions. The implementation of a BMS, and in particular the optimization model, is characterized by uncertainties about the cost and transition probability parameters, which leads to uncertainties about the performance of the overall network level BMS.; The newness and complexity of the network level models and the scrutiny of their results in matters of policy and decision-making create a need for sensitivity evaluation of the results. This dissertation describes procedures for establishing the cost and transition probability data, as well as the development of a set of methods that can be used to evaluate the performance of the optimization model. The methods and associated tool enable the investigation of the sensitivity of the network level results to variations in the cost and transition probability.; In addition, this research evaluates the quality of the cost and transition probability data, to identify recommended MR&R policies that can be optimized according to alternative project scenarios, to evaluate the elements' steady-state condition, and to study the behavior of the long-term discounted cost used as optimization criteria.