Date of Award
Doctor of Philosophy (PhD)
Civil and Environmental Engineering
Eric M. Hernandez
The strength and stiffness of structures degrade with time due to a combination of external forces and environmental conditions. A vehicular bridge, an offshore platform, a ship hull, or a wind turbine are examples of structures that for decades must endure cumulative degradation of their mechanical properties due to cyclic loading. Fatigue-induced damage typically starts at the exterior surface of the component unless microscopic or macroscopic imperfections are present in the material's structure. Structural Health Monitoring (SHM) provides a scientific non-destructive framework to estimate the structure's current state and remaining service life. In many model-based structural health monitoring applications, the models are linear, and commonplace is to formulate them based on modal parameters. The research in this dissertation addresses the implications of model uncertainty to system identification and state estimation. Specifically, determining the highest achievable accuracy in the presence of noise in the measurements, unmeasured excitations, and environmental conditions. The main contributions of this dissertation are summarized as follows: i) derivation of exact mathematical expressions to compute the minimum achievable variance of the identified frequencies and damping ratios from noisy vibration measurements due to initial conditions or external forces, and ii) the development of a weighted sensitivity-based finite element model updating framework to a large scale model of a partially instrumented bridge. Additionally, the dissertation explores the robustness of the Kalman filter in structural dynamics for fatigue monitoring applications.
The dissertation presents recent developments in the feasibility of using global acceleration measurements to assess the level of composite action on operational bridge decks with unknown girder-slab connection stiffness. Our efforts focused on the 58N Bridge constructed in 1963 located on Interstate 89 in Richmond, Vermont, United States. The Bridge has a three-span continuous deck with two build-up outer girders spanning a total length of 558 feet (170.08 m). A portion of the bridge deck was monitored with uni-directional accelerometers and dynamic strain sensors distributed at various locations. Intermittently, for over two years, with measured temperatures ranging from 15F to 87F, data was acquired. This data was used to update a finite element model of the deck. The updated model displayed improved prediction capabilities with respect to the original model. Such an updated model can be used as a baseline model for stress analysis.
Number of Pages
Polanco, Nestor Ramon, "Robustness and Accuracy Bounds of Model-based Structural Health Monitoring" (2021). Graduate College Dissertations and Theses. 1409.