The present study proposes a probabilistic performance-based procedure for assessing the mean annual frequency of pounding corresponding to a given value of the separation distance between adjacent buildings.

The methodology is consistent with and can be easily incorporated into modern performance-based earthquake engineering frameworks. It considers the uncertainty affecting both the seismic input (i.e., site hazard and record-to-record variability) and the parameters used to describe the structural systems of interest (i.e., material properties, geometry, and damping properties). The seismic input is modeled as a nonstationary random process. The seismic pounding risk is computed from the solution of a first-passage reliability problem. While the approach proposed is general, the methodology presented here is specialized to linear elastic systems subjected to Gaussian loading.

Validation studies of the methodology are undertaken by considering simple single-degree-of-freedom systems and more complex multi-degree-of-freedom systems. The computed results are compared with purely numerical simulation results to demonstrate the capability of the proposed procedure to provide reliable estimates of the safety level corresponding to given values of the separation distance.