UC San Diego

Large servo-hydraulic shaking table systems are essential tools in experimental earthquake engineering. They provide effective ways to subject structural components, substructures, or entire structural systems to dynamic excitations similar to those induced by real earthquakes. A typical shake table system includes mechanical, hydraulic, and electronic components. The main objective of this study is to develop a comprehensive mechanics- based virtual model for the large NEES-UCSD shake table under bare and loaded table conditions. The shake table model developed in this study includes a virtual replica of the actual controller, four servovalve models, two single-ended actuators, two effective accumulators, a two- dimensional mechanical subsystem model, and linear/ nonlinear specimens modelled using the finite element analysis framework OpenSees. OpenSees is integrated to the rest of the simulation model in Matlab-Simulink® using a client-server technique developed within this work. Test- simulation correlation studies show that the virtual system model developed is capable of reproducing the nonlinear dynamic response behaviour of the NEES-UCSD shake table. An extensive set of shake table tests using harmonic and earthquake acceleration records as reference/ commanded signals were performed on the NEES-UCSD table to assess its signal reproduction fidelity after tuning the table controller and using an iterative time-history matching technique. These tests were designed to quantify the effects of the amplitude of the signal used for tuning the table on the signal reproduction fidelity. It was found that the level of fidelity in signal reproduction achieved for a specific amplitude of the commanded signal under the corresponding optimum tuning of the table cannot be maintained when reproducing the same signal at different amplitudes. This is a clear indication that shake tables are highly nonlinear systems and the current state-of-the-art controller and tuning techniques fall short of compensating accurately for these inherent system nonlinearities. The mechanics-based virtual system model developed is extremely useful for: (i) understanding the underlying coupled nonlinear dynamics of a large shake table system; (ii) investigating the most significant sources of signal distortion; (iii) offline tuning of the actual table either by using only the virtual replica of the existing controller or by combining it with simulation -based iterative time history matching techniques; (iv) investigating shake table - linear/nonlinear specimen interaction problem; and (v) future more advanced control algorithm developments