Calibration of non-linear effective stress code for seismic analysis of excess pore pressures and liquefaction in the free field

The paper presents numerical predictions of excess pore pressure, liquefaction and settlement response of four centrifuge model tests of 6 m uniform deposits of saturated clean Ottawa sand, placed by dry pluviation and having a relative density ranging from 38% to 66%. The deposits were subjected to 1D uniform base shaking consisting of 10–15 cycles of peak acceleration ranging from 0.04 to 0.12 g. All predictions were conducted with the nonlinear effective stress numerical code Dmod2000. Significant effort was spent in calibrating Dmod2000 by matching the pore pressure and settlement measurements of the first shaking (S1) of a series of shakings conducted in centrifuge Experiment 3. This resulted in very good predictions of both pore pressures and settlement measured in this shaking S1. The exercise showed the importance for realistic simulations of having the correct soil compressibility and permeability. This calibrated version of Dmod2000 was used for a good pore pressure prediction of the preshaken deposit in the same Experiment 3 (S36), by modifying only one parameter in the undrained pore pressure model; and also well predicted pore pressure responses in Tests FFV3 and PFV1, without any change in the parameters of Dmod2000 except for use of the new input motions (Type B predictions). The experimental and numerical results showed that both cyclic shear stress/strains and upward water flow determine together the pore pressure buildup and liquefaction phenomena. The soil response is partially drained rather than undrained, and pore pressure dissipation does take place during shaking both before and after liquefaction occurs.