Efficient hysteresis loop analysis based structural health monitoring of civil structures.
Thesis DisciplineMechanical Engineering
Degree GrantorUniversity of Canterbury
Degree NameDoctor of Philosophy
Real time or rapid structural health monitoring (SHM) enables immediate post-event assessment of the damage state and health condition of civil structures, particularly for critical infrastructure, such as a hospital, fire station, power plant and bridge, which are needed most after an earthquake. They thus offer reliable information for deciding if the structure can continue to be used, and enable more optimum recovery planning and costs. Hence, such methods provide benefits in terms of assessment and recovery from major earthquake events. This thesis examines structural health monitoring from a unique perspective. A model-free hysteresis loop analysis (HLA) is developed for damage identification and structural health monitoring of civil structures subjected to earthquake excitations. The HLA method developed within this research is based on the hypothesis testing and statistical analysis of the reconstructed hysteresis loops to identify the physical parameters that are directly related to structural damage and condition. This approach avoids constraints to a single or fixed model, and is implicitly based on fundamental underlying structural mechanics. It thus provides a novel and computationally efficient means to accurately detect, localize and quantify structural damage for different types of hysteretic structures during or immediately after an earthquake. Importantly, this thesis provides significant experimental validations on the performance of the method to different types of dynamic response, particularly for highly nonlinear hysteretic behaviours. In addition, the effectiveness of the HLA method is also compared to any of the vibration-based and model-based methods using calibrated numerical models. Finally, real data form a base-isolated building is also used to validate the method and its utility. The SHM results on both reinforced concrete (RC) frame experimental structures indicate the HLA method is capable of detecting and assessing the damage location and severity accurately for realistic highly nonlinear RC structures by tracking the evolution of the elastic stiffness of significant half cycles with full dynamic response. More importantly, the method also offers the ability to identify stiffness degradation and damage that may not be evident by external visual appearance. In addition, the extraction of the effective linear stiffness using a multiresolution wavelet analysis (MRA) provides a further useful tool to characterize structural deterioration for more common small events. Comparison of the SHM results between the model-free HLA method and the model-based adaptive LMS filters show that the HLA algorithm is more effective that the model-based method in identifying RC structures with highly nonlinear and variable pinching and/or rocking behaviours. The results also highlight the need for model-based methods to have an appropriate model that can capture the observed response, in order to yield accurate results, even in small events where the structure remain linear. Overall, this thesis develops an efficient HLA-based SHM method to ensure a rapid assessment of the structural damage and safety during or immediately after an earthquake. Significant validation is implemented against both experimental and simulation data of realistic RC structures, as well as real data from a base-isolated building, which thus clearly demonstrate the advantages of this approach over other traditional SHM methods.