Computer simulations are used to predict responses of a building or a bridge to extreme loading caused by earthquake, blast or hurricane. When subjected to these intense loads, damage is most often concentrated into small local zones within the structure. Computer Simulation of local damage and incorporating it with structural system is difficult because simulations of structures operate over large dimensions (more than 10-20 feet) whereas local damage occurs over small dimensions (a few inches). This research project will create a new method that will resolve this scale difference, enabling accurate simulation of structures under extreme loads. The research involves expertise from structural engineering as well as from computational mechanics. The results and software products resulting from this research will improve the safety and economy of the country's infrastructure of buildings and bridges. The involvement of underrepresented groups and the transfer of new knowledge to students will result in broader impacts.
Localized damage in structural members occurs when cross-sectional or material loss of strength (i.e. softening) concentrates damage in a small region of the member. Accurate modeling of local damage is critical, because extreme limit states in structures (e.g. collapse) are highly sensitive to softening behavior that typically accompanies local damage. Current approaches to simulate local damage in structures are either extremely difficult in complex scenarios, or are pathologically mesh-sensitive, leading to inaccurate simulations in either case. This research seeks to bridge this scale gap between structural and local scales through use of a 'nonlocal' approach, by explicitly embedding cross-sectional and material length scales within constitutive and element formulations. The research will include theoretical development of nonlocal approaches for beam-column elements, to produce mesh-independent, accurate and robust simulations of local damage in reinforced concrete and steel members. This development will be informed by finite element simulation as well existing experimental data. This will be followed by numerical implementation and validation.
