The 2016 M_w7.8 Kaikōura earthquake caused widespread liquefaction in the port of Wellington, New Zealand (CentrePort). Several challenges and limitations in the engineering evaluation of soil liquefaction for CentrePort reclamations have been identified when utilizing semi-empirical procedures as they comprise mixtures of gravel, sand, and silt. This paper presents results from 1D effective stress analysis (ESA) performed on representative soil profiles of the gravelly reclamations to gain insights into the liquefaction response in addition to observations from the Kaikōura earthquake. Key steps in the analyses are first presented, including calibration of appropriate constitutive relationships. Results from analyses are presented to illustrate the effects of dynamic interactions within liquefying reclamations including how the response evolves from localized liquefaction at some depths to thick liquefied zones in the fill with increasing intensity of the earthquake excitation. Comparisons are also made between conventional simplified methods and ESA through damage index calculations.

To validate the generalized scaling law (GSL) from both macroscopic and microscopic scales, and investigate the validity of Rocha's assumption under both drained and undrained condition, the monotonic drained and undrained triaxial tests of granular assembly were performed by using discrete element method (DEM) simulations. The tests were performed with initial confining pressures of 50 kPa, 100 kPa, 200 kPa and 500 kPa, where the specimens of 500 kPa were considered as the prototype and the others treated as the model. The results of the drained triaxial tests show that, if a normalized strain is introduced, the stress-strain relationship and the evolution of microscopic parameter a_c could be normalized before the peak, which implies that the Rocha's assumption holds true at both macroscopic and microscopic scales before the peak. However, the normalized strain is smaller than that recommended by the Type II GSL. The undrained test results indicate that the stress-strain curve could be normalized before phase transformation state, while could not be normalized due to the difference of initial mechanical average coordination number between the model and the prototype after phase transformation state.

This article presents a method for simulating hollow torsional shear tests using the discrete element method. The undrained and stress conditions observed in laboratory experiments were reproduced with the proposed method. The specimens with anisotropic stress conditions (K0≠1.0) exhibited lower liquefaction strength than that of isotropically consolidated specimens (K0=1.0). The lower coordination number observed in anisotropic states could contribute to the reduction in liquefaction strength. The different morphologies of contact density might also potentially account for differences in macroscopic behaviors, such as the liquefaction strength.

The residual or post-liquefaction shear strength of liquefiable soils and the driving shear stresses are the main factors determining whether a soil mass will experience flow failure and large deformations due to earthquake or non-earthquake loadings. Despite its importance, determining the residual shear strength remains challenging, if not controversial. One method to determine the residual shear strength is by laboratory testing of reconstituted soil samples and to correct for soil sample disturbance using the framework of Critical State Soil Mechanics (CSSM). There are several key requirements before a Critical State (CS) framework can be used for sands: (1) the definition or existence of a CS for sands, (2) the relationship between void ratio e and mean stress p for the Isotropic Consolidation Line (ICL) and the Critical State Line (CSL) for sands, (3) the parallelism of the ICL and CSL, (4) the normalization of the undrained CS or residual shear strength qcs by the effective consolidation or initial stress, and (5) the non-uniqueness of the ICL for sands. This paper presents a new procedure to determine the undrained liquefied or CS shear strength of sands based on the linearity between e and p for both the ICL and CSL. A growing amount of experimental data supports such linearity, particularly at high confining stresses, but has been hitherto ignored because of the "deeply ingrained" use of e vs. log(p) relationship in soil mechanics. The proposed new normalized procedure does not require parallelism between the ICL and CSL. The evaluation is supported by and validated against experimental data.

In this study, an attempt is made to predict the shear strain and excess pore water pressure ratio (r_u) response of liquefiable sands using deep learning (DL) models. The DL model, such as long short-term memory (LSTM), is considered for predicting the response. The inputs of the models are basic soil properties, including applied shear stress time history, existing initial vertical effective stress, relative density, coefficient of uniformity, and mean grain size of the soil. The databases considered for the DL model training and testing are the cyclic laboratory test data of Ottawa F-65 and Nevada sands. For r_u model training, suitable inputs are considered based on the available domain knowledge on the generation of excess pore water pressure during liquefaction in sands. Shear strain is also an essential input parameter to predict the excess pore water pressure ratio during liquefaction. Therefore, an additional shear strain model is developed to predict the shear strain time history for the testing datasets using the same training datasets. Then this shear strain model is used along with the r_u model to predict the excess pore water pressure response for separate testing datasets. The predicted responses of shear strain and excess pore water pressure ratio agree well with the actual responses for testing datasets and perform well in terms of the evaluation metric. However, the models are trained using the limited datasets of specific soil types and their performance has to be tested for various global soil types under different loading conditions (e.g., transient loading). The study concluded with recommendations for improving the DL models for the cyclic response of other sands for seismic stability assessments.

With the development of high-performance computing finite element method (HPC-FEM), it is becoming possible to perform 3D seismic response analysis using large-scale analytical models that faithfully reproduce the geometry of the ground and buildings. On the other hand, 2D analyses or 3D models with simplified geometry have been the mainstream in seismic design in Japan due to the difficulty of applying HPC technology with sequential calculation codes and creating 3D ground meshes with complex geometry. For the purpose of using HPC-FEM in practice, we have developed an HPC-FEM program that incorporates constitutive laws for soil and liquefaction, which are commonly used in seismic design in Japan. This program extends the large-scale nonlinear finite element method program STRIKE (Ichimura et al., 2022). This extended program showed high scalability on the supercomputer

Fugaku and enabled the automatic creation of meshes with tetrahedral elements for complex geometries (Fujita et al., 2018). We did the seismic response analysis of a 1.4 × 1.3 km urban basin geometry. We used 12,288 CPU cores on Fugaku to perform the analysis of a highly detailed FEM model with 230 million DOF and 5,000 time steps. The analysis took approximately 48 hours. The results were compared with those of previous studies conducted in the same area with a voxel model of simplified ground geometry. The results of both analyses were consistent, verifying the developed program. In addition, the response near the formation boundary was reasonable compared to the results of the voxel model. These findings suggest potential superiority and usefulness of the developed 3D HPC-FEM program.