EVALUATION OF STRUCTURAL POUNDING MODELS USING SHAKE-TABLE EXPERIMENTS AND FE MODELS OF BRIDGES

Pounding is a phenomenon that can significantly affect the seismic performance of simply supported bridges by increasing the displacement and rotation demand of the superstructure. This phenomenon has been observed repeatedly after severe earthquakes, such as those in Kobe (1995), Wenchuan (2008), and Chile (2010), causing varying levels of damage or even collapse. Pounding on bridges has been studied both experimentally and numerically. From a numerical point of view, many macro-models (spring-type / gap element) have been developed to characterize structural pounding during earthquakes, namely, the linear elastic model, Kelvin-Voight model, Hertz-damp model, etc. Among those models, in the last decade, the model proposed by Muthukumar (2003) has been extensively used by the bridge engineering community to simulate deck-abutment pounding for its simplicity and implementation in finite element software such as OpenSees. However, there is significant uncertainty in the input parameters, and the accuracy of this model has not been thoroughly compared with experimental results. Thus, this study presents the results of a shake-table experimental campaign aimed at quantifying the impact force developed during deck-abutment pounding to evaluate the input parameters used in pounding models. The experiment features a scaled concrete bridge deck model (1/25) supported on four elastomeric bearing pads and an element representing the abutment. Three skew angles (0, 15, and 30°) and three additional deck weight conditions (0, 50, and 100%) are considered. A uniaxial shake-table is used to conduct sine-sweep tests and pulses with different frequencies to obtain the dynamic properties of the structure and the impact forces, respectively. Finally, based on the experimental results and numerical models developed in OpenSees, an initial evaluation of the pounding model proposed by Muthukumar is performed. This study helps improve our understanding of pounding and the accuracy of numerical models by comparing them to experimental data.