Prediction of deterioration in structural systems is associated with large uncertainties. Inspections can reduce these uncertainties and support the planning of measures to ensure the integrity of structural assets, but inspections are costly and should be optimized. Past research and applications of risk-based inspection planning have treated sequentially the questions “when to inspect?” and “where to inspect?” to limit the computational cost of optimizing inspection plans. To optimize inspections in larger structural systems, we develop a methodology that accounts for component interactions and interdependence such as stochastic dependence in deterioration processes at different locations, structural interactions and progressive damage evolution. The methodology involves a hierarchical dynamic Bayesian network to compute the updated system reliability with component inspection results. The optimization utilizes a heuristic for defining inspection strategies at the system level. In particular, component inspections are prioritized based on their value of information (VoI). We investigate heuristics that combine component characteristics which are closely linked to the VoI. We define a Prioritization Index and study its effect for different combinations of component structural importance, uncertainty and correlation. For numerical investigations, the methodology is applied to an idealized steel structure subject to fatigue deterioration.     «

Prediction of deterioration in structural systems is associated with large uncertainties. Inspections can reduce these uncertainties and support the planning of measures to ensure the integrity of structural assets, but inspections are costly and should be optimized. Past research and applications of risk-based inspection planning have treated sequentially the questions “when to inspect?” and “where to inspect?” to limit the computational cost of optimizing inspection plans. To optimize in...     »