Multi-Objective Reliability-Based Optimization of Life-Cycle Maintenance for Bridges under Cost Uncertainty

Analyzing performance of structural systems of bridges is necessary for optimizing maintenance cost including repair and replacement. Due to increasing number of deteriorating bridges reaching critical conditions because of age and lack of proper maintenance, advanced analytical methods are not only beneficial, but essential. The outcome of such methods leads to optimal decision-making regarding maintenance of the structure during its life-cycle. As a continuation of previous related work, which were mainly focused on replace strategies, this research pursues an approach for optimizing repair and replace in bridge maintenance in terms of schedule and type of action considering structural performance, cost uncertainty, system requirements, structure specifications, time constraints, as well as available budget. The aim of research described in this paper is to analyze condition state spectrum to determine appropriate action policies and strategies pertaining to each condition state and develop a multi-objective probabilistic method. The outcome reaches a set of efficient and effective solutions in order to minimize the cost for bridge management using a time-variant performance indicator-system reliability index. The outcome of this reliability-based optimization is a cost-action decision-making framework for the maintenance scheduling during the superstructure lifecycle. The advantage of the proposed methodology is incorporating comprehensive bridge maintenance actions into a probabilistic cost model and performing no-preference mathematical analysis. The proposed approach enables users to obtain reasonable scheduling options in order to avoid difficulties in maintenance decision-making regarding structural performance by focusing on failure and deterioration of the bridge structural components. To investigate the accuracy and validate the approach, a numerical experiment for an existing bridge structure is developed and the result is compared with similar efficient relevant methods.