| Popis: |
In this study, inspection planning of deteriorating bridges is optimized to determine the inspection application times and methods based on various objectives. These objectives can be formulated by considering the probabilistic structural performance and service life after inspection and maintenance. Probabilistic structural performance and service life prediction are generally based on the probability of failure (or reliability). However, there are difficulties associated with optimizing inspection planning when a low probability of failure is estimated. In this study, we address inspection planning using extrapolation approaches to efficiently compute a low probability of failure. The inspection planning method proposed in this study determines the inspection application times for a given inspection method. We investigated the applicability of direct Monte Carlo simulation (MCS), subset simulation, and two extrapolation approaches (i.e., kernel density estimation (KDE) and KDE combined with generalized Pareto distribution (GPD)) for inspection planning. The probability of failure for optimum inspection planning was based on the damage detection-based state function and extended service life-based state function. These state functions were formulated by considering damage propagation, damage detection by inspections, and service life extensions by maintenance. Illustrative applications to general examples and an existing bridge are provided to investigate the effects of approaches for computing the failure probability on the accuracy and variation of the optimum inspection application times. Finally, the most appropriate approach for optimum inspection planning was determined considering the accuracy and reliability of the solution, computational efficiency, and the applicability of the probabilistic optimization process. The presented investigations revealed that KDE is more appropriate than MCS and the combination of KDE and GPD for optimum inspection planning. |
