Centrifuge testing of soil-pile-bridge systems was conducted using the NEES (Network of Earthquake Engineering Simulation) geotechnical centrifuge at UC Davis. This testing was a part of a multi-university and multi-disciplinary collaborative research utilizing NEES with goal of investigating the effects of Soil-Foundation-Structure- Interaction (SFSI) while demonstrating NEES research collaboration. The centrifuge experiments complement the 1-g shake table and field experiments conducted at other universities. The data from the centrifuge experiments was compared and combined with the data from other universities to provide integrated analytical models for SFSI problems of soil-pile-bridge systems. This dissertation presents results of these experiments, including collaborations, comparisons with other experiments and numerical simulations, and end-to-end usage of data. Although many aspects of the collaboration exercise were successful, one conclusion of this part of the work was that significant discrepancies between simulations and experiments may be caused by soil-container shaker interaction in the experiments.
Some aspects of the interaction between the shaker and the specimen were accounted for by implementing in the OpenSees finite element simulations a novel method for simulating the excitation of the shaking table as a dynamic force in the actuator (flexible-actuator-prescribed force approach) instead of the conventional approach of specifying the excitation as a prescribed- displacement of the shaking table. Other aspects of the interaction were accounted for by including a more accurate model of the model container, bearing, and reaction mass of the system. Initial attempts to include the servo-hydraulic control system in the simulations were attempted.
Based on a systematic series of simulations of the site response of the centrifuge model that included different approximations of the centrifuge-shaker system, it was concluded that the sensitivity of simulation results to uncertainties in modeling parameters depends on how the aspects of soil-container-shaker interaction are accounted for. This raises a fundamental and very general question: How can we assess the significance of a discrepancy between a simulation and an experimental result? Although this dissertation does not provide a general answer to this fundamental question, it does show that for centrifuge shaking table experiments, the significance of errors in the simulations cannot be rigorously assessed without accounting for test specimen-actuation system interaction.
Cite this work
Researchers should cite this work as follows: