Soil aging refers to the increase in strength and stiffness that is exhibited over time after it is disturbed. It is common in granular soils, such as sands, occurring over periods from hours to years. There have been relatively numerous laboratory studies on sand aging phenomena. However the majority of these studies were conducted on relatively clean sand (fines content <5%) and were performed under isotropic condition. In nature, granular soils with fines content > 5% are not uncommon. This research is an attempt to gain further insight and understanding of mechanical aging on silty sand by conducting laboratory studies mostly under K0 condition, which better reflects the field condition, at both macro-scale (triaxial test) and micro-scale (fabric test). As many factors (e.g. plasticity of fines, fines content, grain size composition, angularity and shape) affect silty sand behaviour and not all those factors could be investigated during the study period, this study focused on mechanical aging of non-plastic silty sand with 15% fines content. Triaxial tests have been conducted in this study in order to observe creep behavior under different density, initial fabric, and consolidation stress paths (K0 and isotropic). The tests were conducted at low effective confining stress stresses i.e. ’3= 30 – 120 kPa as this is relevant to many geotechnical aging problems (e.g. time effects on freshly deposited or disturbed soils such as in the case of hydraulic fills, mine tailings, and post-liquefaction state of soil behaviour following earthquakes). Creep induced aging effects on undrained shear behaviour at small-strain (<0.1% of shear strain), were investigated, as this strain range is most common in geotechnical structures under gravity-induced working loads. Aging effects on one way cyclic behaviour were also studied. Some new key findings from these tests are as follows: (1) Creep following K0 consolidation indicated that the soil tends to expand radially over time, resulting in a tendency of increasing horizontal stress with time even at low stress. (2) Following K0 consolidation, density appears to have more significant effect on creep compared to initial shear stress ratio and mean effective stress; as demonstrated by loose samples (low stress ratio and mean effectives stress) which exhibited greater creep compared to those of dense sample (higher stress ratio and mean effective stress) (3) For loose soils, there is a trade-off between high confining stresses driving aging and collapsing pore space. Generally higher confining stress was found to increase creep tendency thus enhancing aging, however there was also found to be a certain confining pressure where the aging effects became less due to local structure collapse. (4) Initial fabric plays an important role on creep development, thus aging. For instance, dense dry pluviated samples developed larger axial strain over time but also gained less increase in stiffness compared to dense moist tamped samples. This suggests the importance of specimen preparation for laboratory testing that replicates the field scenarios e.g. natural deposition and associated fabric; (5) Dense K0 consolidated samples produce more increase in stiffness with time than corresponding isotropically consolidated samples. Hence, as the K0 condition generally reflects the level-ground free field stress condition better, it is important to test under K0 if the degree of stiffness gain is important; (6) The number of cycles to trigger cyclic softening and liquefaction for one way cyclic loading increases with the aging duration. In addition there is tendency that the aging effect is more pronounced at lower cyclic stress ratios. Fabric tests under K0 consolidation with similar variables as the triaxial tests were also performed. Some new insights and contributions have been obtained as follows: (1) Moist tamped samples, have particles that are more clustered together and structured than dry pluviated samples; (2) In terms of particle orientation, a change in the degree of orientation for both sand particles and ‘fines’ under constant loading was observed with time. The dominant (i.e. most) rotated particles (sand or “fines’) depends on the initial fabric and density; (3) Over time, under constant loading, growth of micro voids was observed for dense samples while those of loose samples contracted; (4) A new parameter, variance to mean void ratio of void distance, was introduced as a measure of the degree of interlocking during aging. The variance to mean ratio of void distance for moist tamped samples tends to decrease whereas those of dry pluviated samples tends to increase with time. An increase in variance and variance to mean ratio for dry pluviated samples indicates that particles are more clustered together with time; (5) Original work on spatial void distance for the numerical analysis of creep induced aging based on Kang et al. (2012) was conducted (note: the model’s boundary condition allows lateral expansion, which is not the same as the fabric tests conducted). The analysis showed that mean void size in dense soil tends to increase with time under constant load while for loose sample it tends to decrease. However the particles also clustered together more – increasing structure. (6) A microstructural study of “undisturbed samples”, obtained by gel-push sampling, of clean sand (fines content = 4%) and silty sand (fines content = 30%), was conducted to investigate anisotropy of natural fabric of granular soils. The results show that dry pluviation reflects the field condition more, in terms of natural deposition, than moist tamping. In addition, spatial void distance qualitatively indicated the undisturbed samples are relatively “very young”, even in terms of engineering time, as indicated by similar variance to mean ratio and kurtosis with those of 1 hour and 1 week reconstituted samples. This research has shown that there was a relation between changes in the microstructure over time and changes in macro mechanical properties of non-plastic silty sand. Further improvement in theoretical modeling (e.g. numerical modeling of creep on polydisperse granular material) and experimental aspects (e.g. examining different grain size composition and angularity, different fines content, the influence of the shape of sand and fines and use of the photo-elastic method) will allow a better understanding of the sand aging phenomenon in silty sand.

abstract: One of the most economical and viable methods of soil improvement is dynamic compaction. It is a simple process that uses the potential energy of a weight (8 tonne to 36 tonne) dropped from a height of about 1 m to 30 m, depending on the project requirement, on to the soil to be compacted hence densifying it. However, dynamic compaction can only be applied on soil deposits where the degree of saturation is low and the permeability of the soil mass is high to allow for good drainage. Using dynamic compaction on saturated soil is unsuitable because upon application of the energy, a part of the energy is transferred to the pore water. The technique also does not work very well on soils having a large content of fines because of the absence of good drainage. The current research aims to develop a new technology using biogenic gas production to desaturate saturated soils and extend the use of dynamic compaction as a ground improvement technique to saturated soils with higher fines content. To evaluate the feasibility of this technology an experimental program has been performed. Soil columns with varying soil types have been saturated with substrate solution, resulting in the formation of nitrogen gas and the change in soils volume and saturation have been recorded. Cyclic triaxial tests have been performed to evaluate the change in volume and saturation under elevated pressure conditions and evaluate the response of the desaturated soil specimens to dynamic loading. The experimental results showed that soil specimens treated with MIDP under low confinement conditions undergo substantial volume expansion. The amount of expansion is seen to be a factor of their pore size, which is directly related to their grain size. The smaller the grain size, smaller is the pore size and hence greater the volume expansion. Under higher confining pressure conditions, the expansion during gas formation is suppressed. However, no conclusive result about the effect of the desaturation of the soil using biogenic gas on its compactibility could be obtained from the cyclic triaxial tests. / Dissertation/Thesis / Data sheets / Masters Thesis Civil, Environmental and Sustainable Engineering 2018

Deformational properties of soil, in terms of modulus and damping, exert a great influence on seismic response of soil sites. However, these properties for sands containing some portion of fines particles have not been systematically addressed. In addition, simultaneous modelling of the modulus and damping behaviour of soils during cyclic loading is desirable. This study presents an experimental and computational investigation into the deformational properties of sands containing fines content in the context of site response analysis. The experimental investigation is carried on sandy soils sourced from Christchurch, New Zealand using a dynamic triaxial apparatus while the computational aspect is based on the framework of total-stress one-dimensional (1D) cyclic behaviour of soil. The experimental investigation focused on a systematic study on the deformational behaviour of sand with different amounts of fines content (particle diameter ≤ 75µm) under drained conditions. The silty sands were prepared by mixing clean sand with three different percentages of fines content. A series of bender element tests at small-strain range and stress-controlled dynamic triaxial tests at medium to high-strain ranges were conducted on samples of clean sand and silty sand. This allowed measurements of linear and nonlinear deformational properties of the same specimen for a wide strain range. The testing program was designed to quantify the effects of void ratio and fines content on the low-strain stiffness of the silty sand as well as on the nonlinear stress-strain relationship and corresponding shear modulus and damping properties as a function of cyclic shear strains. Shear wave velocity, Vs, and maximum shear modulus, Gmax, of silty sand was shown to be significantly smaller than the respective values for clean sands measured at the same void ratio, e, or same relative density, Dr. However, the test results showed that the difference in the level of nonlinearity between clean sand and silty sands was small. For loose samples prepared at an identical relative density, the behaviour of clean sand was slightly less nonlinear as compared to sandy soils with higher fines content. This difference in the nonlinear behaviour of clean sand and sandy soils was negligible for dense soils. Furthermore, no systematic influence of fines content on the material damping curve was observed for sands with fines content FC = 0 to 30%. In order to normalize the effects of fines on moduli of sands, equivalent granular void ratio, e*, was employed. This was done through quantifying the participation of fines content in the force transfer chain of the sand matrix. As such, a unified framework for modelling of the variability of shear wave velocity, Vs, (or shear modulus, Gmax) with void ratio was achieved for clean sands and sands with fines, irrespective of their fines content. Furthermore, modelling of the cyclic stress-strain behaviour based on this experimental program was investigated. The modelling effort focused on developing a simple constitutive model which simultaneously models the soil modulus and damping relationships with shear strains observed in laboratory tests. The backbone curve of the cyclic model was adopted based on a modified version of Kondner and Zelasko (MKZ) hyperbolic function, with a curvature coefficient, a. In order to simulate the hysteretic cycles, the conventional Masing rules (Pyke 1979) were revised. The parameter n, in the Masing’s criteria was assumed to be a function of material damping, h, measured in the laboratory. As such the modulus and damping produced by the numerical model could match the stress-strain behaviour observed in the laboratory over the course of this study. It was shown that the Masing parameter n, is strain-dependent and generally takes values of n ≤ 2. The model was then verified through element test simulations under different cyclic loadings. It was shown that the model could accurately simulate the modulus and the damping simultaneously. The model was then incorporated within the OpenSees computational platform and was used to scrutinize the effects of damping on one-dimensional seismic site response analysis. For this purpose, several strong motion stations which recorded the Canterbury earthquake sequence were selected. The soil profiles were modelled as semi-infinite horizontally layered deposits overlying a uniform half-space subjected to vertically propagating shear waves. The advantages and limitations of the nonlinear model in terms of simulating soil nonlinearity and associated material damping were further scrutinized. It was shown that generally, the conventional Masing criteria unconservatively may underestimate some response parameters such as spectral accelerations. This was shown to be due to larger hysteretic damping modelled by using conventional Masing criteria. In addition, maximum shear strains within the soil profiles were also computed smaller in comparison to the values calculated by the proposed model. Further analyses were performed to study the simulation of backbone curve beyond the strain ranges addressed in the experimental phase of this study. A key issue that was identified was that relying only on the modulus reduction curves to simulate the stress-strain behaviour of soil may not capture the actual soil strength at larger strains. Hence, strength properties of the soil layer should also be incorporated to accurately simulate the backbone curve.

To investigate the liquefaction mitigation capability of Rammed Aggregate Piers® (RAP) in silty sand, blast liquefaction testing was performed at a soil profile treated with a full-scale RAP group relative to an untreated soil profile. The RAP group consisted of 16 piers in a 4x4 arrangement at 2 m center-to-center spacing extending to a depth of 9.5 m. Blasting around the untreated area induced liquefaction (ru ≈1.0) from 3 m to 11 m depth, producing several large sand boils, and causing settlement of 10 cm. In contrast, installation of the RAP group reduced excess pore water pressure (ru ≈0.75), eliminated sand ejecta, and reduced average settlement to between 2 to 5 cm when subjected to the same blast charges. Although the liquefaction-induced settlement in the untreated area could be accurately estimated using the CPT-based settlement approach proposed by Zhang et al. (2002), settlement in the RAP treated area was significantly overestimated with the same approach even after considering RAP treatment-induced densification. Analyses indicate that settlement after RAP treatment could be successfully estimated from elastic compression of the sand and RAP acting as a composite material. The composite reinforced soil mass, surrounded by liquefied soil, transferred load to the base of the RAP group inducing settlement in the non-liquefied sand below the group. This test program identifies a mechanism that explains how settlement was reduced for the RAP group despite the elevated ru values in the silty sands that are often difficult to improve with vibratory methods.

Pile caps are used in foundation design to aid multiple single piles to act as a pile group to resist lateral forces that may cause overturning moments. The pile cap and pile group resist these forces by pile-soil-pile interaction, base and side friction along the pile cap-backfill interface, and passive earth resistance. Passive earth resistance has been neglected in design due to a limited amount of full-scale testing. This research presents the results of a combination of hydraulic actuator and eccentric-mass shaker full-scale testing of a pile cap with loose silty sand backfill to quantify the contribution of the passive earth resistance to the lateral force resistance. The test cap is 1.12 m tall and 5.18 x 3.05 m in plan view, connecting 12 steel pipe piles (324mm O.D) placed in a 4 x 3 pattern with center-to-center spacing of 4.4 and 3.3 pile-diameters in the long and short dimensions, respectively. The hydraulic actuator applied a static load to the system (backfill + pile group) while the eccentric-mass shaker introduced cyclic and dynamic loading to the system. The passive earth resistance accounted for approximately 22% of the total system resistance, with piles contributing approximately 78%. Furthermore, the results produce general correlations between cyclic and dynamic effects on degradation of the backfill provided by the testing and soil characteristics obtained, including target (static) displacement, dynamic displacement amplitude, stiffness, and damping. The dynamic displacement amplitudes during the eccentric mass shaker tests typically ranged between .4 and 2 mm for frequencies between 5 and 9.5 Hz representing behavior under reloading conditions rather than virgin loading conditions. Generally, the presence of the loose silty sand backfill nearly doubled the dynamic stiffness of the pile cap. The stiffness of the backfill and pile cap combined was typically between 100 and 200 kN/mm for frequencies between 4 and 8 Hz, while the stiffness for the backfill alone was typically a decreasing trend between 100 and 40 kN/mm for the same frequency range. The overall isolated loose silty sand damping ratio shows a general increasing trend with values from 32% to 55% for frequencies 3 and 8 Hz.

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.

The static cone penetration tests are quite extensively used for carrying out in-situ geotechnical investigations both for onshore and offshore sites especially where the soil mass is expected to comprise of either soft to medium stiff clays or loose to medium dense sands. The wide use of the cone penetration tests (CPT) in geotechnical engineering has resulted in a great demand for developing necessary correlations between the cone penetration resistance and different engineering properties of soils. The successful interpretation of the cone penetration test data depends mainly on the various empirical correlations which are often derived with the help of a controlled testing in calibration chambers. The calibration chambers have been deployed in various sizes (diameter varying from 0.55 m to 2.10 m) by a number of researchers. It is quite an expensive and time consuming exercise to carry out controlled tests in a large size calibration chamber. The task becomes even much more difficult when a sample comprising of either silt or clay has to be prepared. As a result, most of the reported cone penetration tests in calibration chambers are mainly performed in a sandy material. Taking into account the various difficulties associated with performing tests in large calibration chambers, in the present study, it is attempted to make use of a miniature static cone penetrometer having a diameter of 19.5 mm. This cone was gradually penetrated at a uniform rate in a triaxial cell in which a soil sample of a given material was prepared; the diameter of the cone was intentionally chosen smaller so that the ratio of the diameter of the cell to that of the cone becomes a little larger. Two different diameters of the cells, namely, 91 mm and 140 mm, were used to explore the effect of the ratio of chamber (cell) size to that of the cone size. In addition, the rate of penetration rate was also varied from 0.6 mm/minute to 6.0 mm/minute (the maximum possible rate for the chosen triaxial machine with the larger cell) to examine the effect of the rate of the penetration of the miniature cone on the tip resistance. By using the chosen experimental setup, a large number of static miniature cone penetrometer tests were carried out on four different materials, namely, (i) clean sand, (ii) sand with 15% silt, (iii) sand with 25% silt, and (iv) sand with 15% fly ash. The cone tip resistance for each material was obtained for a wide range of three different relative densities. The effective vertical pressure (σv) for the tests on different samples was varied in between 100 kPa and 300 kPa. The variations of the tip resistance with axial deformation in all the cases were monitored so as to find the magnitude of the ultimate tip resistance. In contrast to the standard cone, the diameter of the piston shaft was intentionally kept a little smaller than that of the cone itself so as to restrict the development of the piston resistance. For each cell (chamber) size, two different sizes of the pistons were used to assess the resistance offered by the penetration of the piston shaft itself. It was noted that the resistance offered by the chosen piston shaft is not very substantial as compared to that of the cone tip itself. Most of the experimental observations noted from the present experiments were similar to those made by the penetration of the standard size cone in a large calibration chamber. The ultimate tip resistance of the cone was found to increase invariably with an increase in the magnitude of σv. An increase in the relative density of the soil mass leads to an increase in the value of qcu. For the same range of relative densities, an addition of fly ash in the sample of sand, leads to a considerable reduction in the magnitude of qcu. Even with the addition of 25% silt, the values of qcu were found to become generally lower as compared to clean sand and sand added with 15% silt. An employment of a larger ratio of the diameter of the cell to that of the miniature cone leads to an increased magnitude of qcu. An increase in the penetration rate from 0.6 mm/min to 6.0 mm/min, was found to cause a little increase in the magnitude of qcu especially for sand added with fly ash and silt. The effect of the penetration rate on the results was found to increase continuously with a reduction in the rate of penetration. At higher penetration rates, in a range closer to those normally employed in the field (20 mm/sec), it is expected that the rate of penetration of the cone will not have any substantial effects on the magnitude of qcu for clean sands. The magnitude of qcu obtained in this thesis at different values of σv for all the cases with the use of the miniature cone were compared with the two widely used correlations in literature. It is found that except for dense sands, in most of the cases, the present experimental data lie generally in between the two correlation curves from literature; for dense sands the measured values of qcu were found to be significantly lower than the chosen correlation curves. It was noted that with the use of the miniature cone penetrated in a given sample prepared in a triaxial cell, it is possible to obtain reasonably an accurate estimate of the tip resistance of the standard cone especially for loose to medium dense states of all the materials. Further, from the analysis of all the tests results, it was noted that approximately a linear correlation between qcu/σv and soil friction angle (φ) for different chosen materials exists provided the dependency of the φ on the stress level is taken into account. As compared to the standard cone penetrometer which is usually employed in the field, the miniature cone used in this study is expected to provide a little conservative estimate, of the tip resistance of the standard static cone penetrometer with reference to the different materials used in this study on account of the facts that (i) there is a reduced area behind the cone, (ii) the ratio of the diameter of the calibration chamber (cell) to that of cone is not very high, (iii) the chosen size of the cone is smaller than the standard cone, and (iv) the chosen penetration rate is much smaller than the standard rate of penetration. Further, in the case of clean sand, an attempt has also been made in this thesis, with the help of a number of direct shear tests at different stress levels, to generate an expression correlating peak friction angle, critical state friction angle, relative density of sand and vertical effective stress. A correlation has been generated with the help of which, the value of peak dilatancy angle can be obtained from the known values of peak friction angle and critical state friction angle. In confirmation with the available information in literature, this exercise on clean sand has clearly indicated that a decrease in the magnitude of vertical effective stress leads to an increase in the values of both peak friction angles and peak dilatancy angles.

Ante la falta de materiales que cuentan con las propiedades mecánicas necesarias para ser usados como relleno estructural en algunas zonas en donde se realizan proyectos, surge la preocupación de mejorar las propiedades mecánicas del limo arenoso. En investigaciones anteriores adicionaron fibras de acero, fibras naturales y aditivos químicos para mejorar las propiedades mecánicas de los suelos. Por otro lado, el deterioro ambiental que actualmente viene sufriendo nuestra sociedad a causa del consumo masivo de productos descartables como el plástico, producto que se descompone en una media de 450 años, y que está compuesto principalmente con PET (Tereftalato de polietileno). Este material al ser arrojado al medio ambiente viene afectando a los animales acuáticos y al agua. Sin embargo, están surgiendo emprendimientos para darles un uso alternativo. La presente tesis se centra en la problemática de mejorar las propiedades mecánicas de la arena limosa de las inmediaciones de los Pantanos de Villa, con la inserción del PET reciclado en porcentajes de 1% al5%. Para desarrollar la hipótesis se realizaron ensayos de granulometría del suelo, PET, y en las mezclas de suelo más PET con porcentajes de 1%, al 5%; límite de Atterberg, gravedad especifica al suelo y al PET. También, se realizaron ensayos de proctor estándar y corte directo para todas las mezclas. Finalmente, se realizaron ensayos triaxiales CD de 3.5 mm al suelo puro y al porcentaje de la mezcla que mejor comportamiento mecánico obtuvo. / Due to the lack of materials that has the necessary mechanical properties to be used as a structural filler in some zones where projects are carried out, the concern arises to improve the mechanical properties of sandy silt. In previous investigations they added steel fibers, natural fibers and chemical additives to improve the mechanical properties of soils. On the other hand, the environmental deterioration that our society is currently suffering due to the massive consumption of disposable products such as plastic, a product that decomposes in an average of 450 years, which is mainly composed of PET (polyethylene terephthalate). This material, when released into the environment, has been affecting aquatic animals and water. However, ventures are emerging to give them an alternative use. This thesis focuses on the problem of improving the mechanical properties of the silty sand in the vicinity of the Pantanos de Villa, with the insertion of recycled PET in percentages of 1% to 5%. To carry out the development of the hypothesis, they were made soil granulometry tests, PET, and in the mixtures of soil plus PET with percentages of1% to 5%; Atterberg limit, specific gravity the soil and PET. Also, Proctor Standard and Direct Cut tests will be carried out for all mixtures. Finally, 3.5 mm Triaxial CD tests were performed to the pure soil and to the percentage of the mixture that obtained the best mechanically performance. / Tesis

Several hydraulic loadings impose earthen levees to time-dependent variably saturated seepage conditions. The main objective of this study is to improve the analysis of levees under transient seepage with the use of unsaturated soil mechanics. An extensive set of laboratory testing, field monitoring and numerical modeling are performed to analyze a silty sand setback levee located near Seattle, WA. In-situ data obtained from field monitoring are used to monitor suction and effective stress within the levee’s embankment and foundation over the past two years. Soil samples taken from the site are used to perform index, water retention, and unsaturated multi-stage triaxial tests in the laboratory. A finite element model of transient seepage under saturated-unsaturated conditions is then developed and calibrated to reasonably match the field data. The results highlight the need to consider unsaturated soil mechanics along with climatic variables and soil-atmosphere interaction when analyzing levees under transient seepage conditions.

Les matériaux constitutifs des digues sont souvent des sables silteux dont le comportement mécanique peut manifester une « instabilité » à des états de contraintes situés bien avant la limite de plasticité de Mohr-Coulomb. Ce mécanisme de rupture pourrait expliquer plusieurs brèches dont l'occurrence n'a pas été attribuée aux mécanismes classiques de rupture de digues telle que l'érosion interne. L'objectif de cette thèse est de mieux comprendre, au regard de ce phénomène d'instabilité, le comportement mécanique d'un sable silteux provenant de la zone adjacente à une brèche de digue du Rhône lors de la crue de 2003, à l'aide d'essais triaxiaux.Les résultats sur le sable propre montrent que le matériau étudié est très sensible au phénomène d'instabilité. Ce dernier peut être prédit par une courbe d'instabilité reliant, au déclenchement de l'instabilité, le rapport des contraintes au paramètre d'état. L'analyse de ces résultats suggère qu'une forte contractance du matériau est la condition nécessaire pour que l'instabilité du sable lâche ait lieu et qu'une direction appropriée du vecteur d'incrément de contraintes est la condition suffisante.Les résultats sur les mélanges constitués du sable avec des fines montrent qu'un départ de fines, causé par exemple une érosion de type suffusion, favorise l'occurrence de l'instabilité dans les digues. Par ailleurs, ces résultats révèlent que l'indice des vides équivalent est un paramètre pertinent dans l'analyse du comportement du sable silteux. Le calcul de l'indice des vides équivalent nécessite la détermination du paramètre b pour lequel une nouvelle formule a été proposée. / The constitutive materials of dikes are often silty sands whose mechanical behavior can become unstable at stress states located well under the Mohr-Coulomb plasticity limit criterion. This failure mechanism could explain several dike breaches whose occurrence has not been assigned to conventional mechanisms of dike failure such as internal erosion. The objective of this thesis is to better understand, with respect to the instability phenomenon, the mechanical behavior of a silty sand which is collected in the area adjacent of a breach of the Rhone embankment dike during the flood of 2003, by means of triaxial tests.The test results obtained on clean sand show that this material is very sensitive to the instability phenomenon. The occurrence of the latter can be predicted by an instability curve relating, at the onset of instability, the stress ratio to the state parameter. The analysis of these results suggests that a strong contractiveness of the material is the necessary condition for instability of loose sand and an appropriate direction of stress increment vector is the sufficient condition.The test results obtained on mixtures of sand with fines show that a removal of fine particles caused by an erosion, such as suffusion for example, favors the occurrence of instability in the dikes. Moreover, these results indicate that the equivalent void ratio is a relevant parameter in analyzing the behavior of silty sand. The calculation of the equivalent void ratio requires the determination of the parameter b for which a new formula has been proposed.