Energy Based System Identification Using Quick-Release Experiments

Abstract

This study presents a time-domain system identification method which is used to identify nonlinear hysteretic properties of isolation bearings installed in seismically isolated bridge structures from quick release experiments. The method utilizes a computational model of the bridge and an optimization technique to minimize an objective function which ensures the energy balance within the structural system at any given instant of time. It is shown that in the proposed method that the objective function is an algebraic quantity which does not require the solution of nonlinear second-order differential equations of motion. Hence, computational effort is minimal. INTRODUCTION Full-scale testing of bridge structures has gained interest during the last two decades in order to better understand their overall behavior under in-situ conditions. Dynamic testing methods such as ambient vibration and forced vibration (e.g. quick-release) testing are considered to be the most viable experimental methods. These experiments can be used to develop reliable nonlinear as well as equivalent linear models that are representative of the type of bridges tested. However, this can be achieved if certain parameters that govern the behavior of the structural system are accurately extracted from the experimental recordings. The suitability and accuracy of the technique used to extract these parameters, which is generally referred to as system identification, is highly dependent on the type of structural system. The dynamic response of seismically isolated structures is controlled by the nonlinear hysteretic properties of the isolators. Most of the research involving seismic isolation has been conducted under controlled-laboratory conditions and on the individual isolation bearings only. Several experimental studies have been undertaken to investigate the overall system behavior on scaled model bridges (Kelly, 1985; Tsopelas, 1994). However, it is generally accepted that in-situ conditions may be different and field verification of the entire structural system is needed. Among all the forced vibration experiments conducted on bridges, only a limited number of full- scale experiments were on bridges with nonlinear isolation bearings (Lam, 1990; Kakinuma et al., 1994; Hasegawa et al., 1994; Gilani et al., 1995; Wendichansky et al., 1998). In their study, Wendichansky et al. (1998) conducted numerous ambient vibration and quick-release experiments on two three-span slab-on-girder highway bridges located on Rte 400 in Erie County, New York before and after the replacement of the steel bearings with isolation bearings. The bearings on the Northbound bridge consisted of laminated neoprene bearings. At the abutments of the Southbound bridge, the steel bearings were replaced by laminated lead-rubber bearings and the same type of bearings without the lead core were used over the piers (fig. 1). Experiments were conducted by pulling (transversely) and releasing the superstructure above the bearings at either both piers simultaneously or at one pier only to excite various modes of vibration. Wendichansky et al. (1998) proposed a dual time-frequency domain system identification method which was used to identify the structural dynamic characteristics of the overall structural system. The nonlinear bearing properties were determined using static test data and based on the laboratory