Identification and response prediction of degrading structures

All structures degrade when acted upon by cyclic forces associated with earthquakes, high winds, and sea waves. Identification and prediction of degradation is thus a problem of considerable practical significance in the field of engineering mechanics. Under cyclic excitation, degradation manifests itself in the evolution of the associated hysteresis loops. The lack of a theory of hysteretic evolution is thus a major barrier to successful design of structures against degradation.; In the past thirty years, cyclic performance testing of structural joints and subassemblies around the world has generated a substantial amount of experimental data on load-displacement traces. In the same period, generalization of the Bouc-Wen differential model of hysteresis permits curve-fitting of practically any hysteretic trace with a suitable choice of its 13 control parameters. Using system identification techniques, it appears highly feasible to put the generalized differential model of hysteresis and the extensive database of experimental hysteretic traces together to advance the methodology for degrading structures. A fundamental objective of this research project is to do just that.; Two principal tasks in connection with hysteretic evolution are addressed in this dissertation. First, a robust identification algorithm will be devised to generate hysteretic models of a deteriorating structure from its experimental load-displacement traces. This algorithm is based upon the generalized differential model of hysteresis and the latest theory of differential evolution, streamlined through global sensitivity analysis. It can account for strength degradation, stiffness degradation, and pinching characteristics in the evolution of hysteretic traces, whereby earlier studies in identification of hysteresis are extended. Second, it will be shown that a hysteretic model obtained by identification can be used for predicting the response of the same structure when driven by other cyclic loads. In addition, the requirements for accurate prediction of system response will be elaborated. Prediction of degradation through system identification is a brute-force approach that offers a close representation of reality. At this time, there is not any method based upon the fundamental postulates of mechanics that can predict the response of a real-life degrading structure well beyond its elastic region.; In the absence of a proper understanding of the mechanisms of degradation, the possibility of identifying and predicting system degradation is indeed intellectually challenging. The research reported herein will establish the basis of a methodology for predicting the performance of real-life degrading structures well beyond their linear ranges. This dissertation will also contribute to the development of a fundamental theory of degradation in the long term.