An accurate prognosis of fatigue failure and crack growth is beyond current state of the art. Current research is almost exclusively focused on finding the ?correct? physical damage variable. In contrast, the aim of this work is to shift this focus to abstract, slowly evolving, dynamical fatigue damage variables?generalized fatigue damage coordinates?that obey some simple mathematical form of a governing differential equation. Carefully designed experiments will show what these observable coordinates are and how they evolve in time using new multivariate and nonlinear time series analyses (i.e., smooth orthogonal decomposition and phase space warping). Based on this evidence, rational fatigue models, which govern these abstract damage variables and have needed mathematical structure to provide experimentally observed dynamical characteristics, will be developed. Experimental characterization of fatigue and corresponding model identification will be carried out using a new experimental apparatus allowing fatigue damage accumulation in various loading environments, with natural dynamic coupling between the damage evolution and structural dynamics. From the fundamental standpoint, this work will lead to improved understanding of fatigue damage accumulation dynamics, its interaction with structural dynamics and loads, and it will provide tools for the development and/or verification predictive damage evolution models. These tools will be instrumental in developing true prognostic ability for structural health monitoring (SHM) and condition based maintenance (CBM) technologies.
The experimental data analysis tools and models developed in this study will have a major impact on a wide variety of SHM and CBM applications. The same methodology can also be applied to characterize other slowly evolving hidden processes, such as physiologic fatigue, driving nonstationary forces in geophysical and oceanographical processes, disease progression, etc. This work also supports collaborations with international research groups as well as with one national laboratory. One Ph.D. level graduate and several undergraduate students will directly benefit from the proposed activities. The interdisciplinary nature of the proposed effort will enrich the learning experience of students by exposing them to modern nonlinear and multivariate data analysis techniques and their applications in mechanical engineering and other fields. A wider set of students will also acquire a better appreciation for the use of these tools in the dynamical characterization of hierarchical systems due to the planned integration of research into education activities. The development of a new undergraduate course on SHM is expected to enhance the undergraduate mechanical engineering program at the University of Rhode Island by exposing students to a new career field.
