Expansive soils are characterized as having high amount of clay minerals such as smectite, which lead to swelling during wet seasons by absorbing water and shrinking during dry seasons owing to moisture loss by evapotranspiration. The soil volumetric changes due to moisture fluctuations cause extensive damage to civil engineering structures, namely pavements, retaining walls, low rise buildings and canals founded on such soils. The primary swelling portion of the swell curve has been studied in significant details in previous studies. However, there is a dearth of literature concerning the secondary swelling phenomenon in expansive clays, which has also been observed in experimental studies. While it may be argued that the magnitude of secondary swelling is significantly less as compared to primary swelling, the characterization of the rate of secondary swelling is relevant for fully characterizing the swell potential of the soil. The rate of secondary swelling has been used to predict the long-term swelling of expansive soils. Conventional laboratory swell tests may take over a month for specimens to demonstrate secondary swelling behavior. A centrifuge based method has been recently developed at The University of Texas at Austin to achieve this objective in multiple specimens, and within less than a day. The effects of soil fabric, soil type, relative compaction, molding water content, gravitational gradient, and infiltrating fluid, on the rate of secondary swelling, are thoroughly investigated in this thesis. Four different expansive clays found widely in and around Texas, namely – Eagle Ford Clay, Tan Taylor Clay, Black Taylor Clay and Houston Black Clay, have been used in the study. Based on this extensive experimental evaluation, it may be concluded that secondary swelling behavior could be explained by flow processes associated with the bimodal pore size distribution in expansive clays. The rate of secondary swelling was found to increase with increasing molding water content and increasing compaction dry unit weight. The experimental results revealed that clays with a flocculated structure (compacted dry of optimum) demonstrate rapid primary swelling but exhibit less swelling in the secondary region, as compared to clays with a dispersed structure (compacted wet of optimum). The slope of secondary swelling showed a decline with increasing gravitational gradient. The rate of secondary swelling showed evidence of upward trend with an increase in the plasticity index and clay fraction of the soil. It was observed that soils which exhibit higher primary swelling also demonstrate higher secondary swelling. / text

Expansive clays are located world wide and cause billions of dollars in damage each year. Currently, the expansion is usually estimated using correlations instead of direct testing as direct testing is expensive and often takes over a month to complete. The purpose of this study was to determine if centrifuge technology could be used to characterize expansive clays through direct testing. Testing was performed in an centrifuge permeameter on compacted specimens of Eagle Ford clay. A framework was developed to analyze effective stresses in centrifuge samples and methods were proposed to determine the swell-stress curve of a soil from centrifuge tests. Standard free swell test were also performed for comparison. The swell-stress curve determined by centrifuge testing was found to match with the curve found from free swell tests after correcting for differences in testing procedures. The centrifuge tests were found to be repeatable and required several days for testing rather than weeks. / text

Drainage methods for liquefaction remediation have been in use since the 1970's and have traditionally included stone columns, gravel drains, and more recently prefabricated vertical drains. The traditional drainage techniques such as stone columns and gravel drains rely upon a combination of drainage and densification to mitigate liquefaction and thus, the improvement observed as a result of these techniques cannot be ascribed solely to drainage. Therefore, uncertainty exists as to the effectiveness of pure drainage, and there is some hesitancy among engineers to use newer drainage methods such as prefabricated vertical drains, which rely primarily on drainage rather than the combination of drainage and densification. Additionally, the design methods for prefabricated vertical drains are based on the design methods developed for stone columns and gravel drains even though the primary mechanisms for remediation are not the same. The objectives of this research are to use physical and numerical models to assess the effectiveness of drainage as a liquefaction remediation technique and to identify the controlling behavioral mechanisms that most influence the performance of sites treated with prefabricated vertical drains. In the first part of this research, a suite of three large-scale dynamic centrifuge tests of untreated and drain-treated sloping soil profiles was performed. Acceleration, pore pressure, and deformation data was used to evaluate the effectiveness of drainage in reducing liquefaction-induced lateral deformations. The results showed that the drains reduced the generated peak excess pore pressures and expedited the dissipated of pore water pressures both during and after shaking. The influence of the drains on the excess pore pressure response was found to be sensitive to the characteristics of the input motion. The drainage resulted in a 30 to 60% reduction in the horizontal deformations and a 20 to 60% reduction in the vertical settlements. In the second part of this research, the data and insights gained from the centrifuge tests was used to develop numerical models that can be used to investigate the factors that most influence the performance of untreated and drain-treated lateral spread sites. Finite element modeling was performed using the OpenSees platform. Three types of numerical models were developed - 2D infinite slope unit cell models of the area of influence around a single drain, 3D infinite slope unit cell models of the area of influence around a single drain, and a full 2D plane strain model of the centrifuge tests that included both the untreated and drain-treated slopes as well as the centrifuge container. There was a fairly good match between the experimental and simulated excess pore pressures. The unit cell models predicted larger horizontal deformations than were observed in the centrifuge tests because of the infinite slope geometry. Issues were identified with the constitutive model used to represent the liquefiable sand. These issues included a coefficient of volumetric compressibility that was too low and a sensitivity to low level accelerations when the stress path is near the failure surface. In the final part of this research, the simulated and experimental data was used to examine the relationship between the generated excess pore water pressures and the resulting horizontal deformations. It was found that the deformations are directly influenced by both the excess pore pressures and the intensity of shaking. There is an excess pore pressure threshold above which deformations begin to become significant. The horizontal deformations correlate well to the integral of the average excess pore pressure ratio-time history above this threshold. They also correlate well to the Arias intensity and cumulative absolute velocity intensity measures. / text

Settlement-Induced Bending Moments (SIBM) are an important design condition that must be considered whenever battered piles are placed in settling soils. The objective of this research was to investigate various parameters which can affect SIBM in battered piles within a levee embankment. The results from the current study were compared and verified against those obtained from centrifuge testing and alternative numerical simulations. A series of centrifuge testing as well as finite difference numerical simulations in Fast Langrangian Analysis of Continua (FLAC) were conducted. Different parameters which may affect the bending moments were investigated including pile connection fixity, batter, and stiffness of the pile as well as the magnitude of settlement. The simulations show that these parameters can have large impacts on the magnitude and location of the bending moments. Findings of this research can be used to validate or identify the need for adjustment of the current modeling/design approach.

Soil improvement methods are used to minimize the consequences of liquefaction by changing the characteristics and/or response of a liquefiable soil deposit. When considering sites with previous development, the options for soil improvement are limited. Traditional methods, such as compaction and vibratory techniques, are difficult to employ because of adverse effects on adjacent structures. One potential method for soil improvement against soil liquefaction in developed sites is accelerated drainage through in situ vertical drains. Vertical drains expedite the dissipation of excess pore water pressures by reducing the length of the pore water drainage path. For more than thirty years, vertical gravel drains or stone columns have been employed to ensure the excess pore water pressure ratio remains below a prescribed maximum value. In recent years, the use of prefabricated vertical drains (PVDs) has increased because the drains can be installed with less site disruption than with traditional soil improvement methods. To date, little-to-no field or experimental verification is available regarding the seismic performance of sites treated with PVDs. The effectiveness of PVDs for liquefaction remediation was evaluated via small-scale centrifuge testing and full-scale field testing. A small-scale centrifuge test was performed on an untreated soil deposit and on a soil deposit treated with small-scale vertical drains. Compared to the untreated condition, the presence of the small-scale vertical drains provided numerous benefits including smaller magnitudes of excess pore water pressure generation and buildup, smaller induced cyclic shear strains, reduced times for pore pressure dissipation, and smaller permanent horizontal and vertical displacements. In addition, full-scale in situ field experiments were performed in an untreated soil deposit and in a soil deposit treated with full-scale PVDs using a vibrating mandrel as the dynamic source. In the untreated test area, the maximum induced excess pore pressure ratio reached about 0.95. In the treated test area, the vibratory installation of the first few drains generated significant excess pore pressures; however, significant excess pore pressures were not generated during the vibratory installation of additional drains because of the presence of the adjacent drains. Additionally, the vibratory installation of the drains caused significant settlement and significantly altered the shear wave velocity of the sand. Dynamic shaking after installation of all of the drains induced small accelerations, small cyclic shear strains, and negligible excess pore water pressures in the soil. The results of the field experiment indicate that the prefabricated vertical drains were effective at dissipating excess pore water pressures during shaking and densifying the site. / text

This study uses dynamic centrifuge tests and three-dimensional (3D), nonlinear finite-difference analyses to: (1) evaluate the effect of soil-cement grid reinforcement on the seismic response of a deep soft soil profile, and (2) to examine the dynamic response of structures supported by shallow foundations on soft clay reinforced by soil-cement grids. The soil profile consisted of a 23-m-thick layer of lightly over-consolidated clay, underlain and overlain by thin layers of dense sand. Centrifuge models had two separate zones for a total of four different configurations: a zone without reinforcement, a zone with a "embedded" soil-cement grid which penetrated the lower dense sand layer and had a unit cell area replacement ratio Ar = 24%, a zone with an embedded grid with Ar = 33%, and a zone with a "floating" grid in the upper half of the clay layer with Ar = 33%. Models were subjected to a series of shaking events with peak base accelerations ranging from 0.005 to 0.54g. The results of centrifuge tests indicated that the soil-cement grid significantly stiffened the site compared to the site with no reinforcement, resulting in stronger accelerations at the ground surface for the input motions used in this study. The response of soil-cement grid reinforced soft soil depends on the area replacement ratio, depth of improvement and ground motion characteristics. The recorded responses of the structures and reinforced soil profiles were used to define the dynamic moment-rotation-settlement responses of the shallow foundations across the range of imposed shaking intensities. The results from centrifuge tests indicated that the soil-cement grids were effective at controlling foundation settlements for most cases; onset of more significant foundation settlements did develop for the weakest soil-cement grid configuration under the stronger shaking intensities which produced a rocking response of the structure and caused extensive crushing of the soil-cement near the edges of the shallow foundations. Results from dynamic centrifuge tests and numerical simulations were used to develop alternative analysis methods for predicting the demands imposed on the soil-cement grids by the inertial loads from the overlying structures and the kinematic loading from the soil profile's dynamic response. / Ph. D.