Field-based liquefaction evaluation procedures include the stress-based, strain-based, and energybased based approaches. The existence of a volumetric threshold shear strain, γtv, under which there is no development of excess pore pressures, and the unique relationship between pore pressure ratio and cyclic shear strain, γc, make a compelling argument for using a strain-based approach. However, the cyclic strain approach has not yet been standardized for field evaluations. The primary objective of this thesis is to use published databases of 415 shear-wave velocity and 230 Standard Penetration Test liquefaction field case histories to investigate the performance of the cyclic strain approach for the evaluation of initial liquefaction relative to the cyclic stress approach. Additionally, the concept of the γtv is expressed in terms of the peak ground surface acceleration and defined as the threshold amax. Computing (amax)t could provide a fast and simple evaluation for initial liquefaction, where no liquefaction is expected for a minimum computed (amax)t determined from the case histories. The variant of the strain-based procedure proposed herein avoids the direct need for laboratory cyclic testing by employing pore pressure generation models that are functions of cyclic shear strain, number of equivalent cycles, and relative density to predict initial liquefaction. The results from the proposed procedure are compared with those of the stress-based approach to determine which better matches the field observations of the case histories. It was found that the cyclic strain approach resulted in 70% to 77% correct predictions. In contrast, the cyclic stress approach yielded 87% to 90% correct predictions. The reasons why the predictions were not always correct with the cyclic strain approach are due to inherent limitations of the cyclic strain approach. Most significantly, an inherent and potentially fatal limitation of the strain-based procedure is it ignoring the softening of the soil stiffness due to excess pore pressure in representing the earthquake loading in terms of γc and neqγ. / Master of Science

The response of an aggregate pier foundation system during seismic loading was investigated. The factors and phenomena governing the performance of the aggregate pier and the improved ground were identified and clarified. The key factors affecting the performance of the aggregate pier include soil density, stiffness modulus, and drainage capacity. The improved ground is influenced by soil stratification, soil properties, pore pressure dissipation, and earthquake time history. Comprehensive numerical modeling using FLAC were performed. The focus of the study in this research was divided into three parts: the studies of the ground acceleration, the excess pore water pressure ratio and the shear stress in soil matrix generated during seismic loading. Two earthquake time histories scaled to different peak acceleration were used in the numerical modeling: the 1989 Loma Prieta earthquake (pga = 0.45g) and the 1988 Saguenay earthquake (pga = 0.05g). The main results of the simulation showed the following effects of aggregate pier on liquefiable soil deposits: 1) The aggregate pier amplifies the peak horizontal acceleration on the ground surface (amax), 2) The aggregate pier reduces the liquefaction potential up to depth where it is installed, 3) Pore pressures are generally lower for soils reinforced with aggregate pier than unreinforced soils except for very strong earthquake, 4) The maximum shear stresses in soil are much smaller for reinforced soils than unreinforced soils. The excess pore water pressure ratio and the shear stress in the soil matrix calculated by FLAC were generally lower than those predicted by available procedures. / Master of Science

This study examines the use of performance-based engineering in earthquake liquefaction hazard analysis with Cone Penetration Test data (CPT). This work builds upon previous research involving performance-based liquefaction analysis with the Standard Penetration Test (SPT). Two new performance-based liquefaction triggering models are presented herein. The two models used in this liquefaction analysis are modified from the case-history based probabilistic models proposed by Ku et al. (2012) and Boulanger and Idriss (2014). Using these models, a comparison is made between the performance-based method and the conventional pseudo-probabilistic method. This comparison uses the 2014 USGS probabilistic seismic hazard models for both methods. The comparison reveals that, although in most cases both methods predict similar liquefaction hazard using a factor of safety against liquefaction, by comparing the probability of liquefaction, the performance-based method on average will predict a smaller liquefaction hazard.

A macroelement is developed for soil-structure interaction analyses of piles in liquefiable soils, which captures efficiently the fundamental mechanisms of saturated granular soil behavior. The mechanical model comprises a nonlinear Winkler-type model that accounts for soil resistance acting along the circumference of the pile, and a coupled viscous damper that simulates changes in radiation damping with increasing material non-linearity. Three-dimensional (3D) finite element (FE) simulations are conducted for a pile in radially homogeneous soil to identify the critical parameters governing the response. The identified parameters, i.e., hydraulic conductivity, loading rate of dynamic loading, dilation angle and liquefaction potential are then expressed in dimensionless form. Next, the macroelement parameters are calibrated as a function of the soil properties and the effective stress. A semi-empirical approach that accounts for the effects of soil-structure interaction on pore pressure generation in the vicinity of pile is used to detect the onset of liquefaction. The predictions are compared with field data obtained using blast induced liquefaction and centrifuge tests and found to be in good agreement. Finally, the macroelement formulation is extended to account for coupling in both lateral directions. FEM simulations indicate that response assuming no coupling between the two horizontal directions for biaxial loading tends to overestimate the soil resistance and fails to capture features like 'apparent negative stiffness', 'strain hardening' and 'rounded corners'.

Liquefiable soils are common at ports due to the use of hydraulic fills for construction of waterfront facilities. Liquefaction-induced ground failure can result in permanent ground deformations that can cause loss of foundation support and structural damage. This can lead to substantial repair and/or replacement costs and business interruption losses that can have an adverse effect on the port and the surrounding community. Although numerous soil improvement methods exist for remediating a liquefaction-prone site, many of these methods are poorly suited for developed sites because they could damage existing infrastructure and disrupt port operations. An alternative is to use a passive remediation technique. Treating liquefiable soils with colloidal silica gel via permeation grouting has been shown to resist cyclic deformations and is a candidate to be used as a soil stabilizer in passive mitigation. The small-strain dynamic properties are essential to determine the response to seismic loading. The small-to-intermediate strain shear modulus and damping ratio of loose sand treated with colloidal silica gel was investigated and the influence of colloidal silica concentration was determined. The effect of introducing colloidal silica gel into the pore space in the initial phase of treatment results in a 10% to 12% increase in the small-strain shear modulus, depending on colloidal silica concentration. The modulus reduction curve indicates that treatment does not affect the linear threshold shear strain, however the treated samples reduce at a greater rate than the untreated samples in the intermediate-strain range above 0.01% cyclic shear strain. It was observed that the treated sand has slightly higher damping ratio in the small-strain range; however, at cyclic shear strains around 0.003% the trend reverses and the untreated sand begins to have higher damping ratio. Due to the nature of the colloidal silica gelation process, chemical bonds continue to form with time, thus the effect of aging on the dynamic properties is important. A parametric study was performed to investigate the influence of gel time on the increase in small-strain shear modulus. The effect of aging increases the small-strain shear modulus after gelling by 200 to 300% for the 40-minute-gel time samples with a distance from gelation (time after gelation normalized by gel time) of 1000 to 2000; 700% for the 2-hour-gel time sample with a distance from gelation of 1000; and 200 to 400% for the 20-hour-gel time samples with a distance from gelation of 40 to 100. The treatment of all potentially liquefiable soil at port facilities with colloidal silica would be cost prohibitive. Identifying treatment zones that would reduce the lateral pressure and resulting pile bending moments and displacements caused by liquefaction-induced lateral spreading to prevent foundation damage is an economic alternative. Colloidal silica gel treatment zones of varying size and location were evaluated by subjecting a 3-by-3 pile group in gently sloping liquefiable ground to 1-g shaking table tests. The results are compared to an untreated sample. The use of a colloidal silica treatment zone upslope of the pile group results in reduced maximum bending moments and pile displacements in the downslope row of piles when compared to an untreated sample; the presence of the treatment zone had minimal effect on the other rows of piles within the group.

No description available.

Les matériaux granulaires présentent un large spectre de propriétés mécaniques. Développer des modèles constitutifs permettant d'intégrer ces caractéristiques dans le cadre de simulations à l'échelle de l'ouvrage demeure un réel challenge scientifique. A cet égard, les approches multi-échelles constituent aujourd'hui une voie très prometteuse. Elles permettent de faire émerger des propriétés macroscopiques à partir de modèles micromécaniques calibrés à l'échelle microscopique.Parmi les modèles multi-échelles, le modèle H marque une avancée majeure pour la prise en compte des effets de la microstructure dans le comportement des matériaux granulaires. La structure du matériau granulaire est décrite par une distribution d'hexagones orientés dans l'espace. A partir d'opérations d'homogénéisation, les contraintes et les déformations incrémentales sont reliées à l'échelle de la distribution, donnant lieu à un modèle de comportement qui a la capacité à reproduire propriétés mécaniques essentielles des matériaux granulaires.Nous étudions dans un premier temps les propriétés mécaniques de l'assemblage hexagonal de grains, élément de base du modèle H, afin d'identifier les conditions menant à sa déstabilisation. Nous réalisons dans un second temps une étude de sensibilité du modèle constitutif vis-à-vis des paramètres micro-mécaniques et microstructurels. Enfin, nous démontrons les capacités opérationnelles du modèle à partir d'essais triaxiaux non drainés réalisés sur un sable lâche liquéfiable.Dans un second temps, le modèle H est implémenté en tant que loi constitutive dans un code de calcul aux différences finies. Des simulations d'essais biaxiaux non homogènes sont conduites afin d'explorer les capacités du modèle à reproduire les différents modes de rupture observés en laboratoire. L'utilisation du modèle H pour modéliser des essais biaxiaux drainés et non drainés met clairement en évidence l'influence de la microstructure sur la réponse mécanique des matériaux granulaires. Enfin, le modèle H est utilisé dans le cadre d'une simulation hydro-mécanique couplée à l'échelle de l'ouvrage pour modéliser le chargement d'une fondation superficielle et la rupture d'une digue soumise à une crue. / Granular materials generally exhibit a broad spectrum of mechanical properties. Developing constitutive models to integrate these properties in the context of simulations at the structure scale remains a real scientific challenge. In this respect, multi-scale approaches offer very promising solutions as they allow the emergence of macroscopic properties from micromechanical models calibrated on a microscopic scale.Among the multiscale models, the H model marks a major step forward in taking into account the effects of the microstructure in the behavior of granular materials. The structure of the granular material is described by an assembly of hexagons, oriented in space. From homogenization operations, stresses and incremental strains are related to the scale of the assembly, giving rise to a constitutive model that has the ability to reproduce the essential mechanical properties of granular materials.We first study the mechanical properties of the hexagonal grain assembly in order to identify the conditions leading to the triggering of its instability. We then carry out a study of the sensitivity of the constitutive model with respect to micro-mechanical and microstructural parameters. Finally, we demonstrate the operational capacities of the model from triaxial undrained tests carried out on a liquefiable loose sand.In a second step, the H model is implemented as a constitutive law in a finite difference code. Simulations of non-homogeneous biaxial tests are carried out in order to explore the model's capacities to reproduce the different failure modes observed in the laboratory. The use of the H model to model drained and undrained biaxial tests highlights the influence of the microstructure on the mechanical response of granular materials. Finally, model H is used in numerical simulations at the structure scale to model the loading of a shallow foundation and the failure of a levee subjected to a flooding event.

The liquefaction resistance of undisturbed soil samples collected from a selected location in Carbondale, Southern Illinois was evaluated by conducting cyclic triaxial tests. Index property tests were carried out on the sample for identification and classification of the soil. Cyclic triaxial tests were conducted on undisturbed sample after saturation, undisturbed sample at natural water content and remolded samples prepared by compaction in the lab. The results were used to evaluate the effect of saturation and remolding on liquefaction resistance of the local soil. Effect of effective confining pressure on dynamic properties of soil (Young's Modulus and Damping ratio) was also studied. Forty five stress controlled cyclic triaxial tests were performed. Three different values of initial effective confining pressure (5 psi, 10 psi and 15 psi) were used and cyclic stress ratio was varied from 0.1 to 0.5 in order to apply different cyclic shear stresses. The results show that the cyclic load resistance of soil decreases as a result of remolding. Saturated undisturbed samples show increase in resistance to liquefaction with increase in initial confining pressure. Remolded samples were prepared by compaction in the lab keeping unit weight and water content equal to that of undisturbed samples. Remolded samples show increase in liquefaction resistance with increase in confining pressure. Undisturbed samples at natural water content show increase in resistance to develop axial strain with increase in confining pressure. Both the rate of excess pressure development and axial strain development increase significantly as a result of remolding. While investigating the effect of saturation of undisturbed samples on liquefaction resistance of soil, interesting observations were made. The excess pressure buildup rate was faster in case of saturated undisturbed samples compared to that in samples with natural water content. On the other hand, rate of strain development was significantly high in case of sample with natural water content compared to that in saturated sample. Also, results obtained from cyclic triaxial tests on saturated undisturbed samples were compared with results obtained from similar tests on Ottawa Sand (Lama 2014) sample. The comparison shows that the saturated undisturbed soil samples of the selected local soil have very high resistance to liquefaction both in terms of initial liquefaction and development of 2.5% and 5% axial strain. Modulus of Elasticity and damping ratio were studied as important dynamic properties of soil. Young's Modulus was observed to decrease significantly at higher strain levels for all three types of samples. Young's modulus increased with increase in effective confining pressure, the effect of confining pressure being large at low strain level and almost insignificant at higher strain level. Damping ratio was highest in undisturbed sample at natural water content and smallest in remolded sample and damping ratio for saturated undisturbed sample falls in between. The damping ratio did not show any definite correlation with strain and confining pressure at lower strain level. But, for strain higher than 1% double amplitude axial strain, damping ratio significantly decreases with increase in strain. Damping ratio increases with increase in confining pressure as observed at high strain for all samples.

Seismically-induced liquefaction has been the cause of significant damage to infrastructure and is a serious concern in current civil engineering practice. Several methods are available for assessing the risk of liquefaction at a given site, each with its own strengths and limitations. One probabilistic method has been shown to provide more consistent estimates of liquefaction risk and can be tailored to the specific needs of a given project through hazard-targeted (i.e. based on return periods or likelihoods) results. This type of liquefaction assessment is typically called “performance-based,” after the Pacific Earthquake Engineering Research (PEER) Center's performance-based earthquake engineering framework. Unfortunately, performance-based liquefaction assessment is not easily performed and can be difficult for practicing engineers to use on routine projects. Previous research has shown that performance-based methods of liquefaction assessment can be simplified into an approximation procedure. This simplification has successfully been completed for the Cetin et al. (2004) empirical, probabilistic standard penetration test -based liquefaction triggering model. Until now, such a simplification has not been performed for another popular liquefaction triggering model developed by Boulanger and Idriss (2012). As some engineers either wish to use or are required to use the Boulanger and Idriss (2012) model in their liquefaction assessments, there is a need for a simplified performance-based method based on this model to supplement that based on the Cetin et al. (2004) model. This thesis provides the derivation of a simplified performance-based procedure for the assessment of liquefaction triggering using the Boulanger and Idriss (2012) model. A validation study is performed in which 10 cities across the United States are analyzed using both the simplified procedure and the full performance-based procedure. A comparison of the results from these two analyses shows that the simplified procedure provides a reasonable approximation of the full performance-based procedure. This thesis also describes the development of liquefaction loading maps for six states and a spreadsheet that performs the necessary correction calculations for the simplified method.

Stone column treatment is commonly used to mitigate liquefaction hazard in sandy soils. Research and experience indicate that this method is effective for clean sands but that it may not be effective for silts and sands with fines contents greater than 15-20%. An alternative to the stone column method involves supplementing stone column treatment with pre-fabricated vertical wick drains installed prior to the stone columns installation. Although this method is used in practice, there has not been a formal academic study of its effectiveness. This thesis evaluates seven different case histories where wick drains were used and one where wick drains were not used, for comparison purposes. The site locations varied as well as the soil properties and treatment plans. CPT testing was done at 3 sites and SPT testing was performed at the other 5 sites. CPT data were correlated to SPT data to facilitate comparisons. One of the case histories includes a unique study in which three different variations of the stone column treatment were applied at the same site, providing a direct comparison of the effectiveness of each method. A 26% area replacement ratio (Ar) with drains was determined to be more effective overall than a 26% Ar without drains and more effective in increasing low initial blow counts than the 34% Ar without drains. The areas with drains were more likely to exceed the minimum project criteria consistently throughout the site. Significant scatter were observed in the results and probable causes for the scatter are noted. Final blow count coefficients of variation ranged from 28% to 77%. Increased fines contents required increased Ar in order to maintain similar average final blow counts. Site improvements were evaluated separately and collectively. Individual site results were compared to clean sand curves developed by Baez (1995). Sites with average fines contents less than 20% which were improved using drains and an 11-15% Ar treatment were comparable to clean sand sites without drains and with 5-10% Ar. To achieve similar improvement at sites with 40-46% fines necessitated drains and Ar values of 23-26%. Design recommendations are provided.