Abstract:

The increased need for experimental verification of the seismic performance of conventional and novel structural systems has resulted in highly sophisticated dynamic test procedures. Hybrid simulation, including pseudo-dynamic testing of experimental substructures, offers an efficient method for assessment of dynamic and rate-dependent behavior of large-scale structural systems subjected to earthquake excitation. Compared to earthquake simulations using shake tables, hybrid simulation may have significant advantages in terms of cost, scale, geometry, and required physical mass of structures and components that can be tested. However, recent hybrid simulations have been limited to simplified structural models with only a few degrees of freedom. This is primarily due to the fact that hybrid simulation is a relatively new test method that is still being improved through research. Currently, the major challenges for using hybrid simulation in large and complex structural systems are the lack of robust simulation algorithms, and the sensitivity of the results to experimental errors in the presence of high-frequency modes. The main motivation for this research is to develop reliable test procedures that can be easily applied to fast and real-time hybrid simulations of large and complex structural systems. It is also attempted to develop test procedures that are effective for geographically distributed hybrid simulations.

In this dissertation, recent developments to improve the accuracy and stability of hybrid simulation are described using the state-of-the-art pseudo-dynamic hybrid simulation system at the Structural Engineering and Earthquake Simulation Laboratory, University at Buffalo. In particular, delay compensation procedures are examined, and new methods are proposed. These methods are based on the correction of tracking errors in force measurement signal, and using the numerical integration procedure for prediction and compensation of command displacement signal. A new online procedure is proposed for estimation of delay during the simulation, and is shown to have better performance compared to existing online delay estimation methods. Furthermore, two numerical integration procedures are introduced for hybrid simulation, which are shown to improve the stability and accuracy properties of the simulation. The proposed integration algorithms use experimental measurements to iterate within implicit scheme and also take advantage of a new approach to estimate the tangent stiffness matrix of experimental substructures. For assessment of the reliability of hybrid simulation results, energy-based error monitors are proposed to examine the severity of experimental and numerical errors. These measures are then used to demonstrate the improved accuracy offered by new algorithms proposed here through analytical and numerical studies, and numerical and experimental simulations.