Physics-based seismic input for engineering design and risk assessment

The use of physics-based numerical simulations (PBS) of ground motion is investigated and proposed as a viable alternative, or complementary tool, to GMPE relations. The aim is to supply engineers and analysts with a seismic input that is as realistic and complete as possible, suitable for both structural analyses and seismic hazard and risk assessment. In fact ground motion is described by means of complete seismograms inclusive of the effects related to the source, path and site conditions. The PBS technique employed in this work is part of the NDSHA framework The results of the PBS procedure have been compared to the corresponding assessment obtained via different GMPE relations. The comparison aimed at verifying whether the numerical simulations of ground motion are consistent with observed events, employing the GMPE relations as a proxy. A large database (86400 events) of physics-based numerically simulated events has been created considering 3 different seismic sources (faults) and different causes of variability that may affect the ground motion. The comparison has been performed in terms of four parameters which are commonly used to quantify the level of ground motion and of engineering significance: Spectral acceleration, Arias intensity, CAV, and significant duration. Moreover, the response of nonlinear SDOF systems has been considered and the comparison with the GMPEs has been performed in terms of maximum inelastic displacement and equivalent number of cycles. The comparisons have shown a good agreement between the results of PBS and the observed data, assessed via GMPE relations and suggest that, on average, no relevant bias is present and results of PBS are acceptable for engineering purposes. The results of PBS procedure have also been compared to the recorded data of a real seismic event. The aim of the study was not to the recreate the actual event occurred in Aquila but rather to verify whether the simulation technique employed is able to, to some extent and from an engineering perspective, predict a possible earthquake event associated to a certain seismic source and magnitude. Records from 6 stations, located at various distances and characterized by different soils, were considered and 300 physics-based events have been simulated modelling the source via the Extended Source model (ES). The comparison has been performed in terms of response spectra and various peak and integral parameters of engineering significance. The response of nonlinear (EPP) SDOF systems has also been considered. The comparisons have shown a good agreement between the results of PBS and the recorded data for all the considered stations, despite the simplified model employed for the soil and the complex geological conditions of the area. An improvement of the standard PSHA procedure has been proposed and discussed. GMPE are dropped in favor of PBS. The methodology, called physics-based PSHA (PBPSHA), aims at combining the strengths of both the NDSHA and the PSHA procedure: accounting for many physics-related effects that greatly influence the ground motion (PBS) and accounting for the uncertainties (PSHA), providing the seismic input as seismograms suitable for both structural analysis and seismic hazard and risk assessment. A method for the selection of ground motions, called "Direct Method", is proposed and implemented for many definitions of IM, e.g. structure-specific IMs for which no GMPE is available. The formulation proposed for PBPSHA is extended to the "Multi-site" case, leading to a formulation for physics-based MSPSHA (PBMSPSHA). Finally a validation is provided for this formulation.