Techniques In Overcoming Physical And Temporal Inaccuracies In Hybrid Seismic Simulations

Abstract

Experimental testing remains the most reliable tool to help understanding the response of civil engineering structures due to earthquake excitations despite significant advancements in structural analysis software. The hybrid simulation method addresses the challenges found in the shake table and the quasi-static testing methods by enabling dynamic tests on large scale structures at reduced speed, yet still obtaining the full dynamic response of the tested structure. Further efficiency in hybrid simulation is achieved through incorporating the substructuring concept, by only physically testing critical regions of a structure while numerically modelling the rest. A recent advance in the field is the development of the realtime hybrid simulation method. This method overcomes limitations in the conventional hybrid simulation, making it suitable for rate-dependent structures. One of the major challenges in real-time hybrid simulation is actuator delay, which imposes negative numerical damping and leads to inaccurate test results and potentially test instability. This study proposes an intuitive delay compensation procedure to correct the response in real-time, utilising energy balance between those resulted from external and internal forces during a simulation. A series of numerically simulated real-time hybrid simulations with delay is presented to demonstrate the effectiveness of the procedure in the presence of small to moderate delay. This study also develops a Kalman filter algorithm to be used in conjunction with the proposed delay compensation algorithm. An extensive parametric analysis through numerical simulations show that the algorithms improves the simulation accuracies and extend the stability limit of hybrid simulations in the presence of delay much further than the stability limit of the proposed delay compensation alone. This study adopts nonlinear co-ordinate transformation algorithm for multi-axial testing to perform moment-axial load compatible in-plane tests on wall structures, and bidirectional tests on simple columns. The co-ordinate transformation procedure accounts for global and local actuator co-ordinates, geometric nonlinearity, as well as changes in the test setup geometry during testing. The multi-axial test capability leads to a supplementary investigation on the effect of different displacement paths in bidirectional quasi-static and hybrid simulations on a rocking concrete column. The experiments demonstrate that for the same displacement amplitude in the quasi-static tests, out-of-phase displacement patterns produces lower force envelopes and energy dissipations compared to in-phase displacement patterns. In hybrid simulations, “staggering” displacement tracking strategies result in higher displacement amplitudes and hysteretic energy dissipations compared to “direct” tracking strategy. The last contribution from this study is an experimental validation of the substructuring concept in hybrid simulations, where there is behaviour incompatibility between the physical substructure and the full prototype structure. This is demonstrated through a hybrid simulation using a squat wall as the physical substructure, with intrinsically shear-dominant behaviour, to replicate the response of a flexure-dominant wall. The experiments show good agreements in terms of the hysteretic envelopes and the maximum force and displacement amplitudes between the squat and the flexure-dominant wall, as well as a good agreement between the experimental results and numerical models.