No abstract

Several methods have been proposed for the analysis of building pounding. In previous studies some of these models have been considered in isolation, while others have been studied as a specific configuration of buildings. Although the pounding force predicted by each model varies significantly, results from the evaluation of these models have been inconclusive. This leads to uncertainties in the practical applications of pounding force models. This study addresses the limitations in numerical and experimental simulations of building pounding. The Sears impact model is suggested to overcome some of these limitations. It is concluded that contact stiffness and coefficient of restitution are the main reason for the diverse numerical results, while differences observed in the experiments could be due to sampling rate and instrumental constraints.

Pounding damage in major earthquakes has been observed frequently in the form of aesthetic, minor or major structural cracks and collapse of buildings. These observations have attracted many numerical and experimental studies that led to analytical models for simulating seismic pounding. This study considers pounding between two steel portal frames without a seismic gap. The first frame has a constant natural period while the second frame has variable stiffness and mass values. Five different ground motions are applied to eight combinations of adjacent frames using a shake table. Numerical simulations for the same configurations are carried out with five pounding force models, viz. linear viscoelastic model, modified linear viscoelastic model, nonlinear viscoelastic model, Hertzdamp model and modified Hertzdamp model. The contact element stiffness and coefficient of restitution for numerical models are determined experimentally. The amplification of maximum displacement of the first frame predicted by the numerical simulations is compared with the shake table results. It was found that the Hertzdamp model always overestimated the responses while the other four models also frequently overestimated the amplifications. The predictions from the four models were not significantly different. Since the linear viscoelastic model requires substantially less computation, compared with the other models this model is more suitable for numerical modelling of pounding responses. However, more study is required to refine the numerical models before building pounding can be modelled with enough confidence.

Abstract. Seismic pounding affects some buildings and bridge spans with insufficient expan

No abstract