Journal of Geotechnical and Geoenvironmental Engineering

Risk assessment for large engineered systems involving geotechnical uncertainties—such as dams, coastal works, industrial facilities, and offshore structures—has matured in recent decades. This capability is becoming an ever more pressing concern considering increasing natural hazards and the effects of climate change. This progress and the current state of practice are reviewed by considering four projects. An underlying theme is what we have learned about how geoprofessionals should think about uncertainty, judgment, and risk. The organizing key is Bayesian thinking, which is neither new nor unintuitive to most geoprofessionals. However, this way of thinking and analyzing data is powerful, even when dealing with the sparse data of geotechnical projects. The presentation builds on the past Terzaghi Lectures of Casagrande, Whitman, and Christian. It uses examples of infrastructure systems for which geotechnical considerations were important. Reliability concepts developed in this earlier practice provide a window into the processes of integrating site characterization data, judgment, engineering models, and predictions to achieve tolerable levels of risk. Along the way, arguments are made on the nature of uncertainty and on the benefits of processing data, judgment, and inferences through a consistent lens.

A soil water unit weight of $9.8\mathrm{kN}/{\mathrm{m}}^3$ has been universally considered to quantify soil volumetric phase properties such as void ratio and degree of saturation, but has been found to greatly vary depending on soil type and the volume scale with which it is defined. Recent experimental and theoretical evidence has indicated that the unit weight of soil water can be significantly greater than $9.8\mathrm{kN}/{\mathrm{m}}^3$ for clayey soils when gravimetric water content is less than 30%. A soil water unit weight as high as $18.8\mathrm{kN}/{\mathrm{m}}^3$ is evident for some expansive soils at low water content. The significance of abnormally high water unit weight in quantifying soil phase volumes, saturation, and void ratio was experimentally assessed for various clayey soils and theoretically interpreted as a function of water content and soil type. For clayey soils with low liquid limit, average soil water unit weight can be as high as $12.5\mathrm{kN}/{\mathrm{m}}^3$ for gravimetric water content less than 10%. This leads to an overestimation of liquid-phase saturation and void ratio by as much as 8% if a soil water unit weight of $9.8\mathrm{kN}/{\mathrm{m}}^3$ is used. For clayey soil with a high liquid limit, the average soil water unit weight can be as high as $18.8\mathrm{kN}/{\mathrm{m}}^3$ for water content less than 18%, leading to overestimation of liquid-phase saturation by as much as 36% and void ratio by as much as 20% if $9.8\mathrm{kN}/{\mathrm{m}}^3$ is used. Charts were developed to estimate average soil water unit weight as a function of soil specific surface area and water content, and as a function of liquid limit and water content. The commonly used value of $9.8\mathrm{kN}/{\mathrm{m}}^3$ for water unit weight can lead to significant errors in estimating phase volumes, void ratio, and saturation for clayey soils.

This study presents physical observations and insights into particle migration characteristics throughout the suffusion process. Using a purpose-built coaxial permeameter cell, suffusion experiments were conducted on idealized internally unstable gap-graded granular soils at varying fines content and hydraulic loading conditions. The specimens were prepared with a mixture layer comprising finer and coarser fractions underlying a coarse layer composed of the coarser fraction alone. This enabled the finer fraction within the mixture layer to migrate through the coarse layer with upward seepage flow. The local porosity profile along the specimen was determined using spatial time-domain reflectometry and an inversion algorithm, which enabled the development of a novel field map of the difference in porosity from the initial condition. This field map provided a visual guide of the spatial and temporal variation in porosity and enabled particle migration internally within the specimen to be quantitatively characterized from onset to progression to washout. The limiting onset condition identified from the field map was shown to be comparable to that obtained using conventional approaches, thereby providing strong validation for the application of porosity-based field maps. As suffusion progressed, the height of infiltrating finer particles into the coarse layer increased linearly with time, while the overall rate of particle migration from the mixture layer to the coarse layer evolved in a non-linear manner with the rate of migration increasing as the specimen reached a complete mixture condition, where the finer fraction infiltrated the entire coarse layer. The attainment of a complete mixture condition was dependent on the fabric of the gap-graded soil. Specimens with an underfilled fabric showed a gradual migration process, while specimens with a transitionally underfilled fabric resulted in minimal particle migration followed by a very rapid formation of the complete mixture.

Ground improvement (GI) techniques have shown promise in effective liquefaction mitigation, but the physical mechanisms governing their three-dimensional (3D) response during dynamic loading are not yet fully understood. To evaluate the 3D performance of one GI technique, the rammed aggregate pier (RAP), in-situ site characterization, full-scale field test data, and calibrated baseline constitutive soil model parameters are combined to model the 3D fully coupled hydromechanical response of natural (unreinforced) and improved (reinforced) soil profiles. To the authors’ knowledge, this contribution represents the first 3D model of a columnar-reinforced soil profile being calibrated using full-scale field testing. The field observations from the comprehensive ground improvement testing (GIT) program in New Zealand and insights from two-dimensional (2D) finite-difference analyses using constitutive model parameters calibrated against the field measurements were used. The developed 3D models were subjected to dynamic loads simulating the excitation generated by a vibroseis truck at one of the in-situ test sites of the GIT program, as well as unidirectional and bidirectional earthquake ground motions from the Canterbury Earthquake Sequence events. The 3D simulations showed that the improved soil profiles experienced reduced excess pore pressures and reduced dynamically induced shear strains compared to the natural, unreinforced soil models. The developed 3D finite-element predictions were compared and validated vis-à-vis the field observations of the GIT program. Compared to 2D analyses, 3D analyses provide a more accurate description of actual field conditions, and, for instance, it was observed that multidirectional shaking has a significant effect on liquefaction triggering, particularly for natural soil profiles. Finally, it was shown that soil densification around the installed pier elements and the lateral earth pressure increase within the densified soil and are the primary ground improvement mechanisms contributing to the reduction of dynamically induced shear deformations and excess pore pressure generation during earthquake shaking. It was also found that the permeability and shear stiffness of the installed RAP piers did not have a significant influence on the pore pressure response and shear strains developed along the centerline of the improved area.

The effect of an anisotropic stress state on the monotonic behavior of calcareous sand has been rarely investigated, although calcareous sands in the field mainly experience anisotropic consolidation rather than isotropic consolidation. This study provides a laboratory study of the undrained and drained behaviors of dense calcareous sand consolidated anisotropically under various initial mean effective stresses ($p_0^{\prime }$) and anisotropic stress ratios ($K_c$), where significant effects of $p_0^{\prime }$ and $K_c$ on the strength-deformation-degradation characteristics of calcareous sand can be observed. For undrained tests, higher deviatoric stress can be reached under a lower $p_0^{\prime }$. This could be due to the buildup of negative excess pore-water pressure ($u$) induced by sample dilatancy. As $K_c$ increased, the negative $u$ accumulated more distinctly, whereas the maximum and minimum $u$ decreased. For drained tests, the maximum deviatoric stress decreased as $K_c$ increased, even though more dilatancy occurred. Undrained loading generally resulted in a higher extent of particle breakage due to the buildup of negative $u$ that increased the mean effective stress. A unique critical-state line was observed in both isotropic and anisotropic samples regardless of the loading condition. However, unlike quartz sand, a pseudosteady state that defined the transition from a dilatancy-dominant nonflow response to breakage-dominant stress degradation was observed in calcareous sand before reaching the critical state. Results also reveal the negligible effect of $K_c$ on the peak-state stress ratio at undrained conditions. However, $K_c$ had significant effect on stress ratio at peak or phase-transformation state in experiments under drained conditions, which can be well represented by using the concept of state dependence. Moreover, $K_c$ had an adverse effect on the initial secant modulus.

The damage caused by seismic shaking and liquefaction-induced permanent ground deformation has conventionally been assessed as two separate problems often by different engineers. However, the two problems are inherently linked, since ground shaking causes liquefaction, and liquefaction-induced soil softening affects ground shaking. Modelling both problems within a single numerical model is complex for both the engineer and the software, and most finite element software only have the capabilities to address one of them. To improve the estimates of the seismic performance of buildings on liquefiable soil, a new sub-structuring approach is proposed called the macro-mechanism approach. This approach allows the soil-liquefaction-foundation-structure interaction to be considered in a series of sub-models accounting for the major nonlinear mechanisms of the system at a macro level. The proposed approach was implemented in the open-source finite element software OpenSees and then applied to a case study of a building where significant liquefaction- and shaking-induced damage was observed after the 1999 Mw 7.4 Kocaeli Earthquake. The case study building was also simulated using two different commercial software programs, the finite difference software FLAC, and the finite element software PLAXIS, by two different research teams. A comparison between the results from the macro-mechanism approach compared to full numerical models shows that the macro-mechanism approach can capture the extent of the foundation deformation and provide more realistic estimates of the building damage than full approaches since the FLAC and PLAXIS models consider elastic elements for the building.

Foam, as a group of disconnected gas bubbles wrapped with liquid film, also exists in municipal solid waste (MSW) landfills. In this study, a foam flow model in MSW is established, in which the population balance equation is adopted to calculate the dynamic process of foam generation, bursting, and transport. The status of foam is characterized using the number density, which is defined as the number of bubbles per unit volume of pore. The relative permeability and viscosity of foam are corrected according to its number density. Then, the proposed model is verified through the foam displacement experiment conducted in an MSW column. The foam flow model can accurately simulate the transient change of pressure drop and liquid saturation during foam displacement. The values of coefficients of foam generation and bursting rates are fitted to be suitable for MSW. Finally, the foam flow model is applied to predict the gas pressure development in an MSW landfill. Foam is able to maintain the landfill gas pressure at a high value even for aged MSW with low gas generation rate. The S-shaped curve of gas pressure versus depth suggests that the foam flow model is capable of simulating the phenomenon of high gas pressure accumulated in the deep layer of landfills because it takes into account the nonuniform distribution of foam number density. The advantage of the foam flow model over the gas flow model is discussed with respect to landfill gas pressure prediction.

Current simplified models for predicting liquefaction triggering and manifestation do not account for the mechanisms of liquefaction triggering and surface manifestation in a consistent and sufficient manner. Specifically, an artifact of the way simplified triggering curves have traditionally been developed is that they inherently include some factors that influence surface manifestations, particularly for medium-dense to dense soils. As a result, using the simplified triggering curves in conjunction with the manifestation models results in the double-counting, omission, or general obscuration of distinct factors that influence triggering and manifestation. Accordingly, the objective of this paper is to develop a framework for deriving a true simplified liquefaction triggering model consistent with a defined manifestation model, such that factors influential to triggering and manifestation are handled more rationally and consistently. Operating in conjunction with the $LS{N}_{ish}$ manifestation model, the performance of the true triggering curve is compared with an existing, popular triggering curve using a set of 50 global case histories; the results are favorable for the proposed framework. Furthermore, the ability of the true liquefaction model to be used in conjunction with manifestation models that account for complex free-field soil profile stratigraphies or developed sites is a significant advantage over traditionally developed triggering models.

In this paper, efforts to characterize and compare the full-scale in situ three-dimensional (3D) dynamic response of a low-plasticity silt deposit to multidirectional loading from two different sources, a vibroseis shaker named T-Rex and controlled blasting, are presented. Horizontal vibroseis shaking at a frequency, $f$, of 10 Hz, produced dynamic responses in the silt that ranged from linear-elastic to nonlinear-inelastic, inducing maximum equivalent direct simple shear (DSS) shear strains, ${\gamma }_{\mathrm{DSS},\max }$, up to 0.15% and residual excess pore pressure ratios, $r_{u,r}$, of 14.1%. Blast-induced shear waves with predominant frequencies ranging from 9.6 to 14.6 Hz excited nonlinear-elastic and nonlinear-inelastic responses in the silt deposit, with ${\gamma }_{\mathrm{DSS},\max }$ of 1.14% and maximum $r_{u,r}$ of 61%. Importantly, these responses were observed to be minimally influenced by high frequency compression waves. Multidirectional loading, and excess pore pressure, $u_e$, migration and impedance were identified as the predominant factors for achieving the large $r_{u,r}$ in the silt deposit from these two in situ testing techniques. The cyclic threshold shear strain, ${\gamma }_{tp}$, to trigger $r_{u,r}$ observed from the T-Rex shaking equaled 0.007% to 0.011% and varied with the initial soil stiffness. The two testing techniques demonstrated that the in-situ shear modulus, $G$, reduced to 90% of the maximum shear modulus, $G_{\max }$, at ${\gamma }_{\mathrm{DSS},\max }\approx {\gamma }_{tp}$, whereas by ${\gamma }_{\mathrm{DSS},\max }\approx 1\%$, $G$ further reduced to 10 to 30% of $G_{\max }$ corresponding to $r_{u,r}$ of $\sim 60\%$. Changes in soil fabric were quantified using small-strain shear-wave velocity measurements performed before and/or upon initiation and after each stage of dynamic testing, and were linked to the observed increase and decrease in ${\gamma }_{tp}$ for the shallower and deeper 3D elements, respectively, following T-Rex shaking. The side-by-side comparison of the dynamic responses and soil properties derived from these two distinctly different field-testing techniques validate the use of controlled blasting for quantifying in situ dynamic soil properties and responses.

The paper presents an interpretation of the temperature monitored during the application of artificial ground freezing in the Isarco River Underpass tunnel, within the Brenner Base Tunnel project. The observations gathered in the in situ monitoring provided an opportunity to (1) examine the response of a heterogeneous glacial and river deposit composed of gravelly loose sand and boulders subject to seepage, upon freezing; and (2) set up a coupled (thermohydraulic) theoretical formulation, which accounts for the water retention curve and relative permeability function, able to properly simulate the artificial ground freezing process on a heterogeneous coarse subsoil under seepage. The simulation was able to provide a satisfactory representation of the development of the in situ monitoring data and it revealed the dependence between the freezing process, the retention properties of the subsoil, and the seepage velocity. A sensitivity analysis was also carried out to examine the potential effect of water retention parameters on the frozen wall formation in terms of thickness and closing time. The critical examination of in situ observations and the results of the numerical analyses allowed the definition of a new criteria to define the freezing temperature of the soil based on the water retention curve features and the residual degree of saturation, with the aim to optimize the design of the artificial ground freezing process by maintaining the safety level against waterproofing of the excavation area.

The effects of fines content on the liquefaction behavior and the dynamic response of gravelly soils were evaluated in this study. A total of six models of level-ground profiles of two saturated gravel-sand-fines mixtures were tested in a geotechnical centrifuge. The first soil mixture was composed of 50% gravel, 38% sand, and 12% fines, and the second soil mixture was composed of 50% gravel and 50% sand. Each soil mixture was subjected to three levels of uniform sinusoidal shaking amplitude: $0.19g$, $0.27g$, and $0.40g$. Both models had comparable relative densities, void ratios, and shear-wave velocities, but the model with fines content had lower permeability by an order of magnitude. Acceleration time histories and pore-water pressure ratios were obtained and used to establish shear stress–strain behaviors during the shaking events. Shear modulus and damping are evaluated and analyzed in relation to the number of cycles and pore pressure development. Lateral displacement, surface settlement, and grain-size distribution of the model after each test were examined to understand the effects of fine particle migration and its influence on shaking amplitude. Based on the results of the two soil mixtures, the fines content had a relatively low impact on the cyclic shear strain, the rate of pore pressure development during shaking, the cyclic resistance ratio, and the reduction of shear modulus. However, the fines content has a noticeable impact on soil dilatancy, the rate of pore pressure dissipation during and after shaking, damping, volumetric strains, and soil fabric after shaking.