A distributed-mass model with end-compliance effects for simulation of building pounding

Abstract

Pounding damage in buildings has been observed after almost every urban earthquake. Pounding occurs when the out-of-phase vibrations of buildings causes them to close the at-rest separation and collide together. The relative motion can be caused by the differing dynamic properties of the structures, foundation flexibility effects or the spatial variation of ground motion. The principal objective of this doctoral research is to develop a numerical force model suitable for simulation of building pounding. The model has to be reasonably accurate in predicting both displacement and acceleration responses due to pounding. The first part of the study consists of the experimental and numerical evaluation of numerical pounding models available in the literature. Shake table investigation of pounding between steel portal frames and impact tests between reinforced concrete slabs were carried out, along with numerical simulations. It is shown that the existing numerical models have many limitations which make them unsuitable for general purpose application in pounding simulation. Therefore, a damped Sears impact model is developed and verified in the second part. The proposed model is based on the Sears impact model. It was observed that the original model cannot include the effects of (i) damping and attenuation of impact-induced stress waves, (ii) higher longitudinal modes of vibration of floors and (iii) the storey-stiffness of the buildings. The damped Sears model overcomes these limitations by analysing the stress propagation as superposition of viscously damped longitudinal modes of vibration of the floors. The storey-stiffness is incorporated by substituting the corresponding floor displacement of a lumped-mass model of the building in place of the translation mode of freely moving bars. For validation of the model, the results from numerical simulation were compared with the experiment response from pounding between suspended RC slabs and between steel beams in three configurations: (i) impact between pendulums, (ii) impact between one-storey frames, and (iii) impact between a one-storey frame and a two-storey frame. Other existing models failed to predict both impact-induced acceleration and the transfer of momentum between colliding masses with any consistency. In contrast, the damped Sears model gave reasonable simulation of these quantities.