The capability to correctly predict the liquefaction occurrence with numerical models is fundamental for the adequate design or verification of strategic structures and infrastructures under seismic conditions. Earthquake records provided by vertical liquefaction arrays constitute a powerful tool to verify the goodness and limitations of constitutive models. In this context, the recorded motions of the 1995 Kobe earthquake carried out by the instrumented vertical array located in Port Island (Japan) were used to assess the performance of several constitutive models able to predict the occurrence of liquefaction in the last three decades. This paper tries to retrace the history of the Port Island vertical array and the main contributions derived from the interpretation and modelling of the seismic records of the 1995 Kobe earthquake. Then, the study focuses on the on the simulation of the seismic response of the instrumented soil column of the array by using some of the most popular numerical models according to a loosely coupled and fully coupled approach for effective stress analysis. Also total stress analyses are performed for comparison. Special attention is devoted to model the dynamic behaviour of the man-made gravelly deposit that liquefied during the seismic event, and the related model calibration, which is based on the results of in-situ and laboratory tests. The results enrich the discussion on the numerical reproducibility of the seismic behaviour of gravelly soils which are largely adopted in man-made geo-structures, such as reclaimed ports and embankment dams.

The Great Hanshin Earthquake has been a decisive moment in the evolution of modern seismic design concepts. The devastating consequences of this very strong seismic event gave rise to new ideas and approaches, aiming for enhanced seismic resilience of modern infrastructure. One of the most dramatic cases from the Kobe Earthquake is the collapse of 16 segments of the elevated Hansin Expressway. The specific case study still inspires generations of engineers worldwide. This paper offers an overview of the different studies performed by the authors, with increasing level of complexity, over the past 25 years. Starting with an early study in 1999, the dramatic failure was successfully reproduced, employing nonlinear beam elements, and modelling the pilegroup foundation with elastic swaying and rocking impedances to explore the effects of soil-structure interaction (SSI). This early study illustrated in an emphatic way that the effects of SSI are not always beneficial. Almost 20 years later, the case study was revisited employing modern numerical tools, and presented during the 59th Rankine Lecture of Prof. George Gazetas. The structural members (reinforced concrete pier and piles) were modelled with the Concrete Damaged Plasticity (CDP) model, while the soil hysteretic response under cyclic loading with a Kinematic Hardening model. Geometric nonlinearities (sliding – detachment of the foundation from the surrounding soil) were captured employing frictional-tensionless interface elements. The revised work focuses on two locations of the bridge, one where the pier is supported by piles and one where the pier is founded on an embedded footing. Counter-intuitively, the less stiff/strong foundation has a beneficial effect, preventing the dramatic collapse and highlighting the potential of allowing for strongly nonlinear SSI. The paper discusses the evolution of the numerical tools over the past 25 years and their contribution in deeper understanding of complex SSI effects. Both in the early and in the recent study, their evaluation is based on postearthquake observations from Kobe, highlighting the significant benefits of well-documented case histories.

Following the Kobe earthquake (1995), researchers have conducted several full-scale field tests using controlled blasting in order to determine the performance of critical infrastructure due to liquefaction and lateral spreading. This study summarizes the lessons learned from the field testing conducted to date and the advancement of the geotechnical earthquake engineering practice for the path moving forward. In all of the large field-scale studies discussed herein, controlled blasting methods were used to trigger liquefaction in various deep foundations and ground improvement test programs. Blast charges are installed in a pattern at different depths with various charge weights depending on the project's objective. The detonation of buried explosive charges generates various stress waves, which propagate from the blast source and dynamically load the soil. When multiple blast charges are used with some delay time in between each charge, the soil mass is repeatedly stressed, which leads to the collapse of the soil matrix, and residual excess pore pressure is gradually or rapidly developed (depending on the specifics of the blast program) leading to liquefaction. Soil and structural response were captured during and following the induced liquefaction. Further, opportunities with this testing technique have been discussed for future studies where the understanding of several unexplored earthquake geotechnical engineering issues could be improved.

Quay walls in Kobe Port are severely damaged in 1995 Kobe Earthquake Disaster. The fact that a very severe shaking was observed at the site revealed the difficulties and limitations of conventional seismic design process. Thus, the advanced dynamic FEM method has been implemented in the seismic design practice in Japan. However, in the process of implementation of dynamic FEM analysis in seismic design practice, various kind of effort have been made. In this paper, these efforts after Kobe earthquake are reviewed from the viewpoint of V&V (Verification and Validation) procedure. The achievement and remaining future tasks on the practical use of FEM after Kobe earthquake are summarized.

ISO23469 (International Organization for Standardization, 2005) provides guidelines for specifying seismic actions in the design of geotechnical works. This international standard is founded on past damage histories such as those during the 1995 Kobe earthquake. The central idea of ISO23469 was to introduce a performance-based approach. Now it is 18 years since its publication and the contents were reviewed by the Seismic Design Standard Committee, JSCE. It was recognized that this international standard can be further improved by covering such issues as (1) consideration of beyond-design-basis events, (2) restorability, (3) fault displacements and (4) effective use of strong motion data. This article briefly summarizes the author’s personal opinion on the future of ISO23469.