Extraction of dynamic pore water pressure generation behavior via inverse analysis in seismic site response modeling

Seismic site response analysis is commonly used to predict ground response due to local soil effects. Numerous site response analysis models have been developed to reproduce a measured response, but the accuracy of such models is subject to the lengthy calibration of pre-defined constitutive parameters to match such response. Inverse analysis approaches are used to calibrate 1-D seismic site response models using field measurements. Self-learning Simulations (Self Sim) has been previously introduced to integrate site response analysis with field measurements to extract the underlying soil behavior whereby the functional form of the material constitutive model is unknown. This development was made assuming total stress behavior of the soil. In this paper we describe the extension of SelfSim to extract not only the constitutive behavior of the soil, but also of the pore water pressure generation behavior of the soil. This is important where vertical array readings, such as from Kobe, recorded pore water pressure changes during seismic shaking. A synthetic recording is used to demonstrate this new development.