Testing performance of pore pressure models implemented in one-dimensional site response analysis program against centrifuge test data measured in mildly sloping ground

A reliable prediction of the excess pore pressure build-up is essential in performing an effective site response analysis. In this study, centrifuge test measurements on mildly sloping ground subjected to ramped sinewaves are utilized to test the performances of three pore pressure models implemented in a one-dimensional (1D) site response analysis program conditioned on accumulated strain, energy, and stress, respectively. Three shear wave velocity (Vs) and relative density (Dr) profiles are utilized to account for the uncertainties in the soil properties. The pore pressure outputs calculated with the strain-based model is shown to be highly sensitive to Vs. Agreeable predictions are obtained near the surface, whereas the pore pressures at depths of 3 and 4 m are underestimated. The energy-based model significantly underestimates the pore pressure for all cases. The stress-based model is revealed to be less sensitive to Vs, whereas it is considerably dependent on Dr. Exceptional fit is achieved with a cone penetration test based Dr profile, whereas pore pressure is overestimated using the empirical liquefaction triggering chart to determine the cyclic strength. Further comparisons of the acceleration time histories illustrate the use of both the strain and energy-based models provide slightly higher and better estimates of the surface ground motion compared with that calculated with the stress-based model. Considering the sensitivity of the outputs, it is recommended to account for the uncertainties of soil properties and also to use both strain and stress-based models in performing 1D effective stress site response analyses. •Three pore pressure model implemented in 1D site response analysis program is tested.•Strain-based model provides reliable estimates near the surface.•Energy-based model produces poor fit with the recordings.•Stress-based model yields best estimate when Dr estimated from CPT is used.•It is recommended to use both strain and stress based models to account for uncertainty.