Pounding response of buildings under earthquake motions

Structural pounding occurs when two or more structures with insufficient separation distances in between collide under earthquake loading. It occurs as a result of out-of-phase responses of structures due to differences in their dynamic properties. Seismic pounding damage has been observed in several earthquakes such as Mexico City earthquake in 1985 (Rosenblueth and Meli, 1986), Loma Prieta in 1989 (Kasai and Maison, 1997), and more recently Gorkha earthquake in 2015 (Shrestha and Hao, 2018). The extent of seismic pounding damage ranges from minor crushing to catastrophic collapse of the colliding structure, even though recent design codes provide guidance on mandatory separation gaps to prevent pounding; still, in dense metropolitan areas with growing populations and constructors willing to make the maximum use of available lands, structural pounding is becoming a significant matter that needs attention in the design of new buildings. Over the past decades, simulating structural pounding and assessing the resultant pounding forces have attracted the attention of many researchers. Simulating pounding between two structures and the extent of pounding damage started from a simple model that could only estimate the elasticity and plasticity of a damage using the pre- and post-impact velocities. The research gradually evolved to analytical force-based models called contact element models (other names are impact element or gap friction element models) that simulate pounding between structures idealised as lumped masses. These models have been widely used and modified numerous times; however, they suffer from a lot of uncertainties associated with their contact parameters (i.e contact stiffness, coefficient of restitution, and damping ratio) as proper methods/formulas to calculate the parameters for contact between flat surfaces do not exist. Progressively, these force-based models were implemented in commercial software as gap elements to model structural pounding. In this method of modelling, the gap elements were placed between the colliding structures to measure contact forces. Within the gap element approach, contact parameters have to be defined in advance and be inputted into the gap element before the start of the simulation, and yet the uncertainties with selecting suitable contact parameters persist. The selection of contact parameters influences the pounding response of the structures significantly. Therefore, diverse results are obtained from the existing methods of modelling pounding (Khatiwada and Chouw, 2014). This research work intends to develop a methodology that properly simulates building pounding using the Finite Element Method (FEM). In this methodology, pounding between buildings is modelled using a penalty-based method. Using the proposed FE method, there is no need to define/ or assume contact parameters in advance and the reliability of the simulation is based on the material models used in the simulation. Unlike the existing lumped mass contact models with pre-defined contact points and limited locations to extract data, in the FE model the contact surface area, time of contact and the duration of contact can be obtained anywhere on the colliding buildings. Satisfactory data such as force, displacement, velocity, and acceleration-time histories at every microsecond of the seismic motion is easily accessible. To develop this methodology that leads to a more accurate FE model, contact phenomena at material level (i.e concrete and steel) has been studied using direct impact Hopkinson Pressure Bar (HPB) experiments. Direct HPB numerical simulations were validated against the experiments in LS-DYNA. The results of the investigation at the material level were taken onto the structural level and were combined with numerical simulations of shaking table tests validated against shaking table experiments conducted by Garcia et al. (2010). At a later stage, the findings of the experiments and the numerical simulations are combined into a final FE model that can simulate pounding with more accuracy compared to the existing methods of pounding simulation demonstrating the magnitude of the contact forces and the extent of damage in the colliding concrete buildings. The developed approach allows better insights into the structural and material response mechanisms during earthquake events. Also, more reliable contact parameters can be extracted from the developed FE model and be implemented onto the existing pounding models.