The geotextile tube provides a cost-effective alternative for constructing containment structures, while tailings and coal ash disposal are relatively new fields of application. However, the performance of multilayered stacked geotextile tubes is still poorly understood. A coal ash levee, which was constructed in two stages (namely primary levee and upper levee) using multilayered stacked geotextile tubes, failed as a result of backward erosion piping (BEP). To promote an understanding of the performance of multilayered stacked geotextile tubes, in this study, a postfailure investigation was carried out by multiscale modeling with the transient fully coupled discrete-element method and computational fluid dynamics (DEM-CFD) approach. The geotextile tubes at the downslope toe deformed significantly due to BEP and subsequently triggered a retrogressive sliding deformation to the overburden geotextile tubes of the primary levee but not of the upper levee. This suggested that the multilayered stacked geotextile tubes exhibited a strong resilience against a global failure. The critical slip surface was more inclined to a block failure instead of a circular failure. A sensitivity analysis indicated that increasing the interface friction coefficient between geotextile tubes, & micro;g-g, can significantly enhance the stability of the levee against sliding and reduce the likelihood of significant deformation.

Rockfill materials are widely employed in dam construction, with their deformation and strength critically dependent on particle-size distribution. Various gradation scaling techniques are adopted to accommodate the laboratory apparatus size by converting the grading of realistic rockfill materials into a reasonable range. However, these gradation transformations can introduce significant disparity in the mechanical behavior compared to the original prototype. In this study, the discrete element method (DEM) is used to compare the mechanical properties of rockfill material using different gradation scaling techniques such as equivalent substitution, scalping, and similar grading methods. One-dimensional compression and drained triaxial tests are conducted while the macroscopic and microscopic properties are analyzed. The findings reveal that gradation scaling techniques significantly affect the compressibility and strength of rockfill materials, as indicated by the variations in the microscopic parameters.

Laboratory blasting tests on granite were conducted under varying confining pressure conditions. Damage data across blast crater cross-sections in various directions were obtained using 3D laser scanning and image processing techniques and systematically analyzed based on fractal theory to characterize damage morphology and spatial distribution. The directional anisotropy of blast-induced rock damage was investigated, along with the spatial heterogeneity evolution of damage cross-sections in the direction of applied confinement. The research results indicate that confining pressure significantly regulates the spatial distribution of explosive damage in rock. Under no confining pressure, damage propagation is primarily governed by the intrinsic heterogeneity of the rock, resulting in an approximately isotropic distribution of blast-induced damage, as indicated by a low coefficient of variation of 0.159. Under asymmetric biaxial confining pressure, the cross-sectional damage exhibits a stress-direction-dependent deflection and a redistribution of damage complexity. This phenomenon is herein defined as the “inductive deflection” effect. Lower equal biaxial confining pressure enhances the dominant role of the rock’s heterogeneous structure in the damage evolution path, resulting in differences in the degree of damage across different directional cross-sections. Higher equal biaxial confining pressure suppresses the expansion of damage along the free surface and redirects damage development toward greater depths, resulting in a 21.4 D reflects the regulatory effect of confining pressure on the “propagation mode” of the damage, while the intercept A is directly related to the “spatial occupancy rate or coverage” of the damage. Both are crucial parameters for describing the effects of rock blasting and for assessing the regulatory role of confining pressure on explosive damage.

Particle size and shape significantly influence the crushing behaviors of rockfill materials. In this study, single-particle crushing tests were conducted on Baihetan rockfill particles with four size groups. Particle shape was quantified using parameters derived from 3D scanning. The results show that both particle size and shape significantly affect the crushing behaviors. The particle crushing modes are categorized into splitting, chipping, and explosive. Larger particles and particles with shapes closer to ellipsoids and cuboids are more prone to explosive failure, whereas smaller particles and particles with reduced corner sharpness tend to experience major splitting without prior localized fragmentation. The crushing force, crushing energy, and their variability all increase with particle size, and the effect of shape on the crushing behaviors also increases with particle size. The applicability of the Weibull model in describing the size effects on the crushing strength was evaluated. The particle shape effect was incorporated into the calculation of crushing strength using a shape factor determined by an artificial neural networks algorithm. A probabilistic model was developed to predict the crushing strength distribution based on the particle size and shape parameters.

This paper presents a time-dependent reliability analysis method for concrete-faced rockfill dams (CFRDs) by integrating multiple failure modes and multi-source monitoring data via Bayesian networks. Initially, two sub-Bayesian networks are constructed to fuse dam parameters, two related failure modes, and three types of monitoring data. Subsequently, the prior failure probabilities of the dam system for each period are calculated through the time-variant response relationships among network nodes. These response relationships introduce a time-variant term to quantify the effects of water level and creep. Finally, various types of monitoring data are utilized to update parameter distribution, resulting in the posterior failure probabilities. The proposed method is applied to 233-meter-high Shuibuya CFRD. The results indicate that Bayesian networks offer a more comprehensive and reliable assessment. Water level induces periodic variations in system reliability, while creep drives the long-term trend by increasing slabs' failure probabilities. The failure probabilities of dam system increase over the initial ten years and stabilize as creep converges. The slabs’ failure probabilities vary from location. Seepage failure probability is primarily dominated by the most critical slab. Utilizing multi-source monitoring data can reduce uncertainties, mitigate the interference of localized abnormal data, and identify potential failure locations. This approach supports enhanced dam safety management.

Creep, once considered an inherent characteristic of granular materials, is primarily governed by time and the current stress state. However, recent studies indicate that creep development is also influenced by the loading history. To better reveal the creep revolution law of the rockfill under the influence of loading history such as historical stress rates, creep tests were conducted under oedometric loading. Alternative loading-creep steps, different stress increment sizes, and various precreep stress rates were considered. Independent of other factors, the development of the creep rate was governed by the recent precreep stress rate (the prior stress rate defined in this study). When the prior stress rate was higher than a threshold value, the relationship between the creep rate and time was double logarithmic linear; thus the creep strain-time relationship tended to converge on a power law (referred to as the creep baseline herein). However, when the prior stress rate was lower than the threshold value, the initial creep rate was lower than that of the creep baseline and did not decrease until several minutes after the start of the creep. The development of the creep rate with time in the initial stage can be generalized as a straight horizontal line, suggesting that the rate remains almost unchanged for a certain time, until the straight horizontal line approached the creep baseline. The inheritance and hysteresis of different strain rates in the initial stage of subsequent creep resulted in differences in the creep magnitude and time development process of the creep rate. The above findings are constructive for predicting the deformation of deep layers of rockfill, such as embankments, with more accuracy, especially for that with some large-sized rigid-structure buildings on its surface.

Seepage -induced suffusion involves the migration of fine particles within a soil matrix. Seepage flow is affected by the soil permeability anisotropy of anisotropic soil fabric; however, suffusion anisotropy is unclear because of the limited function of existing permeameters. In recent studies, the effect of seepage direction has been investigated under only low hydraulic gradients because the control of seepage direction relies merely on gravity. In this study, a new, large -sized permeameter is developed with which suffusion tests can be conducted along horizontal or vertical seepage directions under high hydraulic gradients. Correspondingly, the permeameter can accommodate a specimen of 540 x 500 x 470 or 540 x 540 x 440 mm3 (length x width x height). The seepage direction is switched by changing the boundary conditions of the specimen with detachable perforated plates that allow pressurized water originating from different inlets to flow along horizontal or vertical directions. Two repeated pairs of tests were performed on a gap -graded clayey gravel to investigate the suffusion anisotropy of saturated clayey gravel. The results show that the maximum relative deviations of measurements for initial hydraulic conductivity, initiation, and failure hydraulic gradients are less than 3.5 %, demonstrating satisfactory reliability. The ratio of the initial horizontal hydraulic conductivity to vertical hydraulic conductivity for the test soil is 13.87, indicating a significantly anisotropic fabric induced by compaction. The ratios of horizontal initiation and failure hydraulic gradients to vertical initiation and failure hydraulic gradients are 0.52 and 0.59, respectively. This implies that suffusion anisotropy should not be neglected for evaluating the internal instability of anisotropic soils.

Understanding the in-situ behavior of rockfill materials through laboratory tests is challenging due to the influence of sample size. In this study, the discrete element method (DEM) is utilized to investigate the effects of sample size and boundary condition on the compressibility of rockfill materials at both macroscopic and microscopic scales. The results reveal that rockfill compressibility increases with sample size when rigid boundaries are applied, but no significant size effect is observed for periodic boundaries. Besides, the one-dimensional compression behavior of different initial packing varies with sample size under rigid boundaries, with the variance decreasing as size increases; however, this effect is negligible under periodic boundaries. Additionally, both the distribution uniformity of contact number and fabric anisotropy increase with increasing sample size under rigid boundary conditions. At a microscopic level, it can be observed that the sample size effect of granular materials is correlated to the coordination number per unit volume CN/(1 + e) for the considered particle shapes.

Under various stress paths, the deformation characteristics represented great differences. In this paper, a series of cyclic triaxial tests have been conducted with Fujian standard sand. By comparing the constant deviatoric (CDS) and constant axial stress paths (CAS), the influence mechanism of the cyclic amplitude of the deviatoric stress was discussed. The test results showed that the stress path significantly influenced the volumetric and shear strains. The increasing and decreasing trend in the volumetric strain (epsilon v) was consistent with the spherical stress (lnp). Compared with the two stress paths, the slope of the epsilon v-lnp curve during the loading and unloading stages was larger under the CAS path. In the CDS path, qc almost did not affect the cumulative volumetric strain, and in the CAS path, the effect was obvious. The shear strain curve was in accordance with the direction of the stress path. As the cyclic number increased, the shear strain gradually accumulated. The shear strain accumulation under the CAS path was larger. The shear strain largely depended on the relative position between the critical state line (CSL) and the stress state of the soil during cyclic loading and unloading. In practical engineering, the soil will experience various stress paths. For example, in slope or earth-rock dam engineering, where the water level rises and falls repeatedly, the soil often goes through the stress path of constant deviational stress with the cyclic increase and decrease in the spherical stress. In foundation pit engineering, the soil often experiences the stress path of the constant axial stress (CAS) with cyclic loading and unloading of the lateral stress. The stress path greatly influences the deformation and strength of soil. Therefore, the previous two stress paths are compared in this paper to discuss the influence of the cyclic amplitude of deviatoric stress. Under three different consolidation states, the cyclic amplitude of the deviatoric stress significantly influenced the volumetric and shear strains. The shear strain largely depended on the relative position between the critical state line (CSL) and the stress state of the soil during cyclic loading and unloading. Therefore, in practical engineering, if the stress path in the experiment differs from the actual value, the influence of the stress path should be properly considered. The results should be modified according to the degree of influence of each stress condition.

Variables that can be considered in experimental investigations of internal instability of earth dam materials are often bounded by practicality and feasibility constraints. To overcome these shortcomings, a total of 164 previous published experimental results were compiled to study statistically the relative significance of vertical effective stress, grain size ratio (D15/d85), fines content, coefficient of uniformity (Cu), and specimen length in the mechanism of internal instability. The present statistical results agreed with previous experimental findings on the combined dominant roles played by fines content and stress in the internal instability. However, the stress reduction factor was found to be inversely correlated with the vertical effective stress, which was not reported previously. The scale effect of specimen size has significant positive correlations with both the stress reduction factor and critical hydraulic gradient, whereas the influences of Cu and D15/d85 were minimal, in contrast to previous experimental findings. An improved theoretical hydromechanical envelope for predicting internal instability of cohesionless soil in the space of stress-gradient-stress reduction factor was proposed in this study. The unstable envelope moved with increasing applied hydraulic gradient until the state of material touching the envelope.

堆石料的压实密度是反映其工程力学特性的重要指标.提出了一种基于降维映射的颗粒堆积算法,仅有一个模型参数即可在给定堆石料级配的情况下实时预测当前级配下的最大干密度.与试验结果对比发现,该算法能够较好预测给定的连续、间断级配堆石料最大干密度,为堆石料级配的高效优化设计提供了新思路.采用该算法模拟了堆石料细料截断和缩尺对堆石料压实密度的影响,结果表明:本算法可以较好地根据缩尺后的结果预测原型级配的堆石料压实密度,但堆石料压实密度的缩尺效应预测存在细料截断误差,误差与细料含量呈正相关;堆石料缩尺后粗料部分骨架的孔隙尺寸降低,有降低压实密度的趋势,而缩尺时增加的细料含量对密度的影响并无统一的规律,而是与粗料和细料的具体粒径分布相关.

The suffusion involves selective erosion and gradual migration of fine particles through the voids of soil skeleton formed by coarse particles under seepage flow. As a result, redistribution of soil skeleton stress and deformation of soil may be induced. In this study, a series of suffusion tests are carried out using the triaxial erosion apparatus with measurable local pore pressure. The effects of the initial fine particle content and initial relative density on the suffusion of a gap-graded cohesionless soil are investigated. According to the spatial-temporal evolution of local hydraulic gradients along seepage path, the evolution process of the suffusion is revealed. Test results show that both the initiation and the failure hydraulic gradients of the gap-graded cohesionless soil increase with the increase of the fine particle content and relative density. The cumulative loss of fine particles decreases significantly with the increase of the relative density. When the relative density increases to a certain value under isotropic stress condition, the gap-graded cohesionless soil will change from an unstable state of seepage to a stable one. Additionally, the internal manifestation of suffusion initiation of soil is the mutation and uneven distribution of local hydraulic gradient along seepage path. The suffusion will cause the loss of fine particles, as well as the increase of the void ratio. Under the isotropic stress condition, the volume shrinkage is induced.

Internal erosion may affect soil hydraulic properties, inducing localized deterioration and even failure of hydraulic geo-structures. Soil is always subjected to stress in engineering practice; however, the effect of stress on the internal erosion-induced change in soil hydraulic properties along seepage path has not been thoroughly investigated in the literature. In this study, downward seepage-induced internal erosion tests on gap-graded cohesionless soil were conducted using a triaxial permeameter with measurable local hydraulic gradients and deformations. Effects of confining stress on global and local responses of internal erosion were investigated regarding global and local hydraulic gradients and soil deformations as well as eroded fine particles. The results show that the development of seepage velocity during internal erosion exhibits a two-stage linear progression with increasing applied hydraulic gradient under lower confining stress, while it exhibits a single-stage linear progression under higher confining stress. As confining stress increases, both initiation and failure hydraulic gradients are significantly increased, and the onset of the nonuniform distribution of local hydraulic gradients is delayed. Higher confining stress leads to a lower cumulative loss of fine particles and a smaller volume shrinkage, concurrently decreasing the degree of the nonuniform distribution of local hydraulic gradients. Spatial nonuniformity of local hydraulic conductivity and deformations indicates that specimens become more heterogeneous due to the migration of fine particles. Furthermore, although specimens subjected to confining stress exhibit volume shrinkage during internal erosion, the post-erosion void ratio of specimens increases due to the dominant effect of fine particles loss on soil pores.

The geotextile tube provides a cost-effective alternative for constructing containment structures, while coal ash disposal are relatively new fields of application, the performance of multi-layered stacked geotextile tubes for ash disposal are still poorly understood. A coal ash levee, which was constructed in two stages (namely primary levee and upper levee) using multi-layered stacked geotextile tubes, failed as the result of backward erosion piping (BEP). To put forward understanding the performance of multi-layered stacked geotextile tubes, a post-failure investigation was carried out by multiscale modeling with the transient fully coupled discrete element method and computational fluid dynamic (DEM-CFD) approach. The flexible geotextile tubes showed good adjustment and self-healing qualities in response to large deformations. The seepage field - ash foundation - geotextile tubes interactions were fully presented. The geotextile tubes at the downslope toe deformed significantly due to the BEP, and subsequently triggered a retrogressive sliding deformation to the overburden geotextile tubes of the primary levee, but not the upper levee. The critical slip surface was more inclined to a block failure instead of a circular shape. The overburden geotextile tubes, in turn, affected the stress distribution at the ash foundation, and subsequently affected the development of BEP. The indirect interaction between the seepage field and the stacked geotextile tubes was governed by their direct mutual interaction with the ash foundation.

This paper dedicates to exploring the requirement of a representative volume element (REV) on sample size in the distinct element method (DEM) simulations. Two groups of differently sized dense samples are prepared, respectively, in rigid wall and periodic cells for drained plane strain compression simulations. The results show that, under given boundary conditions, the qualitative consistency of the confinement-dependent shear dilation with experiments, i.e., a higher confining pressure leading to less shear dilation, is observed from adequately large samples, but the qualitative deviation from overly small samples. In the samples of adequate size, the dilation mostly localizes within a shear band, and decrease in the shear band thickness with increasing confining pressure largely results in the confinement-dependent decrease in dilation. In contrast, the overly small samples incapable of completely accommodating a shear band dilate uniformly, and consequently, the uncertainty in the shear band probably brings about the contradictory results. From the standpoint of REV, the criterion that a DEM sample needs to satisfy is that its dimensions need to be “much larger” than the thicknesses of shear bands. Additionally, the formation of shear bands in periodically bounded samples provides strong evidence against the argument of “uniform” deformation patterns.

Prediction of dam behavior based on monitoring data is important for dam safety and emergency management. It is crucial to analyze and predict the seepage field. Different from the mechanism-based physical models, machine learning models predict directly from data with high accuracy. However, current prediction models are generally based on environmental variables and single measurement point time series. Sometimes point-by-point modeling is used to obtain multi-point prediction values. In order to improve the prediction accuracy and efficiency of the seepage field, a novel multi-target prediction model (MPM) is proposed in which two deep learning methods are integrated into one frame. The MPM model can capture causal temporal features between environmental variables and target values, as well as latent correlation features between different measurement points at each moment. The features of these two parts are put into fully connected layers to establish the mapping relationship between the comprehensive feature vector and the multi-target outputs. Finally, the model is trained for prediction in the framework of a feed-forward neural network using standard back propagation. The MPM model can not only describe the variation pattern of measurement values with the change of load and time, but also reflect the spatial distribution relationship of measurement values. The effectiveness and accuracy of the MPM model are verified by two cases. The proposed MPM model is commonly applicable in prediction of other types of physical fields in dam safety besides the seepage field.

Measuring the pore-water pressure within a soil mass is critical for studying its hydraulic properties. Due to the granular structure characteristics of soil, the fine particles within soil are prone to migrate with the movement of water flow. The migration of fine particles may change the distribution of the pore-water pressure, thereby altering the local structure and hydraulic properties of soil. In this study, a new method with measurable local pore pressure is introduced to capture the localized response of pore structure of soil induced by the migration of fine particles. The new method meets the sealing requirements of the latex membrane with inserted pipes under stress. Thus, the pore-water pressure transducers can be arranged along the axial and radial directions of the soil specimens to measure the pore pressure at various positions, thereby helping to study the local hydraulic properties. The global and local evolution characteristics of the soil specimens during suffusion are studied using new the method in the suffusion tests. The results show that the new method has a satisfactory response. The spatial and temporal evolution of local hydraulic conductivity reveals that the migration of fine particles along seepage path exhibits significant localization and non-uniformity.

Abstract Multistage constant stress rates loading-creep tests were carried out on air-dried slate rockfill. The influences of the loading history on the creep deformation of rockfill were investigated, including the prior loading stress rate, the stress increment size of prior loading, and the history of early loading and early creep, in which 'prior loading' was defined as the loading just before the creep in the current loading-creep step, while 'early loading' or 'early creep' was that before prior loading. Regardless of the size of the prior loading stress increment, as long as there was a loading-creep process before this step of loading-creep, the time development process of the subsequent creep strain and strain rate was related to the creep stress and the prior stress rate before creep. The creep process had no obvious relationship with the early stress rate in the last step, as well as the earlier loading history before the last creep. The subsequent creep strain was positively correlated with the creep stress and prior stress rate. When the prior stress rate before creep was high, the subsequent creep rate and time were linear in the double logarithmic coordinate. The creep rate and time accorded with a power-law relationship which appeared to be the baseline for creep development. Whatever the prior loading rate, the creep rate time progression in the middle and later stages of the subsequent creep conformed to this baseline. Besides, when the prior stress rate was lower than a certain value, the initial strain rate of subsequent creep was significantly lower than the baseline, which decreases less until several minutes after the start of creep. The development law of creep rate at the initial stage could be generalized as a straight horizontal line in double logarithmic coordinates, which is kept almost unchanged for a certain time. The strain rate would not resume the power-law relationship with time until it approached the base creep rate. The inheritance and hysteresis of different prior strain rates to the initial stage of subsequent creep due to prior stress rate with the discrepancy, result in differences in the creep magnitude and time development process of creep rate. Moreover, the prior low strain rate both depressed the initial rate of subsequent creep and pushed back the initiation time of subsequent creep according to the power law.