Hybrid simulation with distributed substructures including overlapping domains
Cortes-Delgado, Maria D.
MetadataShow full item record
Hybrid simulation can be an efficient tool for structural collapse evaluation by testing experimentally key parts of structures where mechanisms leading to collapse are expected to initiate, while the remaining structural system can be modeled numerically. Present substructuring techniques require simplifying assumptions in order to reduce the controlled boundary degrees of freedom and to minimize the number of actuators in the experimental setup of framed structures. This simplification involves assuming points of inflections at the boundaries, typically at the mid-length of beams and columns, which no longer requires application of moments and rotations and the boundary by actuators. Past studies have shown that the inflection point assumptions provide only a slight variation in drift through nonlinear numerical simulations with moderate yielding. The effects of substructuring techniques on the structural response of a prototype to collapse are examined here through numerical studies. The numerical studies apply the conventional substructuring approach, as well as new innovative approaches with overlapping domains between the substructures. In this approach, the boundaries of the experimental substructure are extended to overlap with the numerical model with the purpose of better applying the full boundary conditions including moments while minimizing the number of required actuators. A full-scale four-story steel moment frame tested to collapse on Japan's E-Defense earthquake simulator is examined in this dissertation and considered as the prototype. A key objective is to repeat these experiments using hybrid simulation with overlapping substructuring techniques. The same sequence of earthquake loading used in the E-Defense test is repeated include simulations with the 1995 Kobe JR Takatori station record applied at scale factors of 0.2, 0.4, 0.6 and 1.0 with the final test resulting in collapse of the structure. Prior to the hybrid simulations, two numerical models based on distributed and concentrated plasticity beam and column elements were examined for the 0.6 Takatori earthquake level. These studies evaluated the adequacy of the numerical models and also the substructuring techniques. Overall the global and local response for the overlapping ( OL ) method performed better compared to the prototype's global and local response than the pinned ( PS ) method considering the concentrated plasticity model. The distributed plasticity model is less sensitive to the PS method than the concentrated plasticity model and when compared with the OL method it provides similar results in terms of the global and local response studied. For the collapse level earthquake, only concentrated plasticity is examined and the results showed that both methods provide similar results. Therefore, the OL method was chosen as the substructuring technique method for a Distributed Hybrid Simulation test ( DHST ) to be conducted at between the University at Buffalo and Kyoto University. The results from the distributed hybrid simulation are compared with shake table tests of the complete structural model to evaluate substructuring assumptions through a detailed analysis including global and local response parameters. For the 0.6 Takatori level, the DHST compared well with the global and local response of the E-Defense. For the collapse level, the DHST achieves the numerical model displacement time history and reaches collapse at the same instance as the E-Defense earthquake simulator tests.