Off-campus UNL users: To download campus access dissertations, please use the following link to log into our proxy server with your NU ID and password. When you are done browsing please remember to return to this page and log out.
Non-UNL users: Please talk to your librarian about requesting this dissertation through interlibrary loan.
Fatigue Prognosis in Bridges Using Long Term Monitoring Data
To account for uncertainties introduced by existing fatigue evaluation methods, probabilistic reliability-based methodologies based on long-term sensor monitoring data are gaining traction as a feasible alternative for steel bridge fatigue assessment. Despite this development, certain challenges have been observed to impede bridge fatigue assessment. Access to fatigue-prone places for sensor installation and data monitoring can be difficult or impossible at times. Cost, maintenance, and power requirements may be limited in situations where these regions are conveniently accessible, and sensors are widely used. While the use of monitoring data has been acknowledged to reduce load-related uncertainties, selecting an appropriate probability distribution to appropriately fit the equivalent stress range extracted from the monitoring data has been identified as a major challenge in completing fatigue assessment. If an improper distribution is applied, the accuracy of the anticipated fatigue remaining useful life (RUL) can be compromised. This research focuses on the development of a reliability analysis methodology for assessing fatigue in steel bridges using Kernel Density Estimation (KDE). Using KDE removes the obstacles involved with choosing an appropriate distribution, making it a realistic and effective way of estimating RUL. To test this method, a reliability study was conducted on a representative, modeled structure subjected to loads with a normal distribution using an established method from the literature and the proposed methodology based on KDE, with the assumption that the load distribution was unknown. Results from both methods were similar. The proposed reliability methodology based on KDE was then applied to perform fatigue assessment on an operational riveted railroad bridge using measured and estimated responses obtained from a 2-month monitoring period. Estimated responses were determined using data-driven Singular Value Decomposition (SVD) method. KDE was used to estimate second moment representations of the equivalent stress range extracted from monitoring data, and reliability indices were predicted across the time of interest. The results demonstrated the reliability methodology based on KDE's potential as a reliable fatigue prognosis tool for steel bridges.
Akintunde, Emmanuel O, "Fatigue Prognosis in Bridges Using Long Term Monitoring Data" (2022). ETD collection for University of Nebraska - Lincoln. AAI29322992.