Liquefaction and site response analysis using the PDMY03 effective stress model

Conventional one-dimensional site response analyses often use simplified nonlinear effective stress models to assess porewater pressure generation and liquefaction potential in soft soils. This study evaluates the performance of the coupled effective stress model PDMY03 using a dataset of four case histories and high-quality centrifuge tests where liquefaction was observed. Consistent with previous research, early liquefaction during shaking induces significant shear strains and discrepancies between nonlinear total and effective stress analyses. Comparisons of recorded and predicted surface spectra, piezometer measurements, and subsurface accelerations indicate that the model generally accurately predicts liquefaction onset, with some exceptions. In particular, excess porewater pressure ratios exceeding ru 0.8 are associated with shear strain levels on the order of 3 - 6%, which lead to significant divergence between nonlinear total stress and effective stress predictions. Under these conditions, the results show short-period deamplification and long-period amplification, depending on the timing of pore pressure generation. In contrast, when ru < 0.6, both approaches yield similar spectral responses and ground motion characteristics. In particular, surface waves and lateral spreading are challenging because one-dimensional shear wave propagation does not capture their effects. Quantitatively, PDMY03 predicted porewater pressure ratios within +-0.15 ru of measured values in three of four cases, and spectral accelerations within +-20% for periods between 0.1 and 2.0 s. Unlike previous validations of the PDMY03 model that focused on laboratory element tests, this study provides a multi-case systematic evaluation integrating field downhole array records and controlled centrifuge experiments. The primary limitation in predicting liquefaction and surface response is the accurate characterization of the dynamic soil properties. In cases of poor model performance, minor adjustments to these properties have improved predictions; specifically, variations in shear wave velocity and stiffness within approximately +-15 – 40% resulted in significantly improved agreement with measured response spectra, pore pressure evolution, and acceleration profiles. Overall, PDMY03 provides reliable predictions when porewater pressure evolution and shear stiffness are well-characterized; its applicability is limited under conditions dominated by lateral spreading and multidimensional wave propagation, highlighting the need for complementary numerical approaches in such cases.