Maintaining condition and safety of deteriorating bridges by probabilistic models and optimization

The deterioration of highway bridges over their lifetime is expected. Often the performance of these bridges falls below accepted limits. However, if highway bridges are properly inspected and maintained, they will be able to survive over long time periods with satisfactory performances. Therefore, structural inspection and maintenance of highway bridges have been active fields of research. The main goal is to properly manage limited resources for maintaining satisfactory structural performances. Currently, studies indicate that the deterioration of structures is nonlinear, and many structures having similar design and construction deteriorate with different rates. The problem is more complex, if maintenance is applied. To treat this type of problems, probabilitistic nonlinear deterioration models for structures are needed. Probabilistic models for time-based and performance-based (condition-based or safety-based) maintenance strategies are developed in this thesis. Maintenance strategies with individual cyclic and multiple cyclic actions are considered and combined. A Monte Carlo Simulation computer program called NLCADS is developed for this purpose. Realistic data consisting of bridge condition and safety, and individual maintenance actions are used. Maintenance actions are divided into two groups: preventive and essential maintenance. Maintenance strategies imposed of different actions (individual or multiple cyclic actions) are proposed and studied. The profiles of condition, safety, and maintenance cost ace computed. It is found that the dispersion of cost at a certain time can be very high. The interaction between different maintenances types helps improve the condition and the safety, and also reduce the cumulative maintenance cost. Optimization to minimize total expected cumulative maintenance cost under multiple cyclic actions is performed. Proper maintenance application times and proper combinations of maintenance actions are used in considering an optimum strategy. The optimum maintenance strategy can be selected based on different criteria. It is found that in most cases, both the total expected maintenance cost and the percentiles of cumulative maintenance cost are minimized simultaneously. Finally, the applicability of individual and multiple maintenance actions is illustrated on an existing bridge in Colorado. It is found that the developed methodology in this thesis can be successfully applied to highway bridges.