Numerous, globally-distributed tailings storage facilities (TSFs) are located in active seismic regions. Seismic instability of TSFs can result in significant damage to civil infrastructure, serious environmental consequences, and potential fatalities. Assessments of the cyclic resistance of silty tailings are complicated owing to the potential differences in soil fabric relative to natural silts due their anthropogenic nature, limited tailings-specific in-situ test correlations, and their relatively young, depth-varying age. A recently-developed suite of cyclic resistance models for the cyclic resistance ratio, CRR, of intact, natural silts is evaluated for the purpose of predicting the cyclic resistance of fine-grained non-plastic and plastic mine tailings and identifying data gaps for future development of tailings-specific cyclic resistance models. Comparison of the cyclic resistance of intact specimens of silty tailings collected from the literature are made to that estimated using the newly developed cyclic failure models considering their plasticity index, PI, and overconsolidation ratio, OCR. The direct comparison of the observed and model-estimated CRR is facilitated by corrections for stress path and loading frequency. The importance of gradation, particle morphology, tailings ore mineralogy, plasticity, and aging/stress history is explored to identify knowledge gaps on the variables contributing to cyclic resistance of silty tailings.

Recent occurrences of major earthquakes have been associated with a series of foreshock and/or aftershock events (i.e., 2011 Tohoku and 2016 Kumamoto) induced reliquefaction of sand grounds. Influence of these small magnitude events needs to be examined as it may result in increased/decreased reliquefaction potential. Catastrophic damages were observed during these repeated earthquakes around the world, whereas studies on this line of research are limited. In the present study, centrifuge model experiments were performed to examine the influence of foreshocks-mainshock-aftershocks sequence on reliquefaction potential. Five shaking events (two foreshocks followed by a mainshock and two aftershocks) were applied in sequence with sufficient time between two shakings. Acceleration amplitude and shaking duration of the mainshock event is twice that of the foreshock and aftershock events. Results show that complete liquefaction was observed only in the mainshock, whereas significant excess pore pressure was recorded in the foreshock and aftershock events. Beneficial effects of seismic preshaking were reported in the second foreshock and were completely eliminated during the mainshock owing to the complete liquefaction. Further application of aftershocks rebuilt resistance to excess pore pressure generation and regained beneficial effect of seismic preshaking. From these presented results, it was concluded that liquefaction destroys preshaking benefits obtained from foreshocks and the liquefied sand ground possesses low resistance to future reliquefaction.

Repeated episodes of soil liquefaction during the 2010-2011 New Zealand earthquakes have sparked intense interest in studying multiple liquefaction phenomenon. The present study develops a new stacked ring torsional shear apparatus based on a pressure compensation technique for repeated liquefaction tests on saturated sands. The new stacked ring shear apparatus overcomes some limitations of a similar device using dry sands by the University of Tokyo. To effectively impose lateral constraints on the soil specimen, stainless steel stacked rings are installed on a specially designed bearing system, which minimizes circumferential soil-ring friction. On the other hand, reducing vertical friction between the sample and the ring is a key issue to generate uniform stresses within the sample. In this study, the soil sample is fully saturated, with inner and outer chamber pressure applied to reduce direct contact force between the membrane and rings, such that the effective vertical stress will be much more uniformly distributed along the sample height. Our results indicate that approximately 75% of the applied vertical stress was maintained at the bottom of the specimen by using the pressure compensation. A series of reliquefaction and multiple liquefaction tests were conducted on saturated Toyoura sand using the new stacked ring device, which enables a single specimen to undergo cyclic loading, consolidation, and reloading in multiple liquefaction tests. The test apparatus shows great promise to investigate some fundamental physics of multiple liquefaction, including the effects of relative density, strain history, and fabrics on multiple liquefaction behavior.

Ischia is an active volcanic island historically hit by several earthquakes, the last of which occurred on August 21^st, 2017. After this event, seismic microzonation studies for three municipalities of the island were carried out, addressed to map the amplification of ground motion. These studies completely neglected the phenomena of ground instability and permanent deformation related to liquefaction and landslides, although these latter have been frequently triggered in the same area by both hydro-meteoric and seismic events, and pre-disposed by the volcanic origin of soil deposits. For the above reasons, a research project was addressed to fill these gaps, with the endorsement of the National Civil Protection and local authorities. This paper will show the main outcomes of the multi-level interdisciplinary approach adopted for liquefaction assessment. The results of simplified analyses showed a low susceptibility to soil liquefaction, while the advanced coupled analyses predicted significant pore water pressure accumulation along some soil profiles. The procedure adopted may constitute a valid methodological proposal for evaluating liquefaction susceptibility and for a preliminary estimate of its impact on the building heritage, through the determination of suitable damage parameters related to post-seismic settlement.

In the empirical framework of liquefaction assessment, it has been long noted an apparent increase of cyclic resistance ratio (CRR) with increased fines content (FC). However, whether this an increase is caused by increase of liquefaction resistance or a decrease of penetration resistance (SPT-N) is not clear. Therefore, this study evaluates the FC effect on the liquefaction resistance and SPT-N using a Taiwan and an international liquefaction database that includes enriched fines data. The SPT-N based hyperbolic function curves that yield to 50% of liquefaction probability are systematically derived for different FC ranges by two methods. Method 1 is the direct correction of SPT-N due to FC (i.e. reduction) in the hyperbolic model that defines the upper bound curve using the maximum N value. Method 2 is to increase SPT-N to be consistent with a higher liquefaction resistance due to FC, which is typically adopted in the liquefaction analysis. The result indicates that apparent increase of N (Method 2) is not equal to the actual reduction of N due to FC (Method 1) but reflects the change associated with increase of CRR.

The abundant post-earthquake data from the Canterbury, New Zealand (NZ) area is poised for use with machine learning (ML) to further advance our ability to better predict and understand the effects of liquefaction. Liquefaction manifestation is one of the identifiable effects of liquefaction, a nonlinear phenomenon that is still not well understood. ML algorithms are often termed as “black-box” models that have little to no explainability for the resultant predictions, making them difficult for use in practice. With the SHapley Additive exPlanations (SHAP) algorithm wrapper, mathematically backed explanations can be fit to the model to track input feature influences on the final prediction. In this paper, Random Forest (RF) is chosen as the ML model to be utilized as it is a powerful non-parametric classification model, then SHAP is applied to calculate explanations for the predictions at a global and local feature scale. The RF model hyperparameters are optimized with a two-step grid search and a five-fold cross-validation to avoid overfitting. The overall model accuracy is 71% over six ordinal categories predicting the Canterbury Earthquake Sequence measurements from 2010, 2011, and 2016. Insights from the SHAP application onto the RF model include the influences of PGA, GWT depths, and SBTs for each ordinal class prediction. This preliminary exploration using SHAP can pave the way for both reinforcing the performance of current ML models by comparing to previous knowledge and using it as a discovery tool for identifying which research areas are pertinent to unlocking more understanding of liquefaction mechanics.