The aim of the research is to investigate the effect of increased axle loading on saturated and unsaturated railway foundation materials for heavy haul applications. The research methodology comprises of a literature review to identify the lacuna in the scientific knowledge, finite element modelling for characterisation of railway cyclic loading, development of a cyclic triaxial apparatus for laboratory testing and experimental work, followed by analysis, interpretation and discussion of results and lastly the formulation of conclusions and recommendations. The axle loading of interest start with a base load of 20 tonnes per axle for general freight followed by increased axle loading of 26, 30, 32.5 and 40 tonnes per axle for heavy haul. The test materials used in the experimental work are representative of the subballast and subgrade layers in a railway substructure. As a reproduction of the climatic conditions in the field and the loading from passing trains, experimental testing was carried out on saturated samples under undrained conditions and unsaturated samples under constant water content. Unsaturated samples were prepared to matric suctions of 50, 100 and 225 kPa via axis translation. Monotonic and cyclic tests were conducted to investigate the behaviour of railway foundation materials. Critical state theory for saturated and unsaturated soils is used as a method of analysis in establishing the failure criterion and the failure envelope. Various parameters, such as stress states, strains, resilient modulus, pore water pressure and matric suction are also utilised in investigating trends and behaviours. Based on the monotonic test results, the shear strength of unsaturated samples was found to be greater than that of saturated samples, attributed mainly to strain hardening caused by the unsaturated soil conditions, with the presence of a peak deviator stress when plotted on the stress-strain graph. However, unsaturated samples were also found to be prone to load-collapse during monotonic shear, even when the water content and confining stress remained constant, which resulted in brittle behaviour with the sudden rupture and formation of multiple bifurcation shear bands and slip planes. Based on the cyclic tests on saturated materials, it was discovered that increased axle loading can result in phase-transition in soil behaviour, based on the stress states in the soil relative to the critical state line plotted in the effective stress space. Stress states below the critical state line resulted in a no-phase transition with dilation behaviour. Stress states on the critical state line resulted in a single-phase transition from dilation to contraction. Stress states above the critical state line resulted in a double-phase transition from dilation to contraction behaviour and then strain-softening. It is therefore concluded that increased axle loading can only be sustained by materials which presented dilation and no phase-transition in soil behaviour, which occurred at axle loading of 20 and 26 tonnes per axle for the subballast and subgrade materials. Based on the cyclic tests on unsaturated materials, it was established that increased axle loading did not cause material failure for all load axle cases and materials. The stress states of all tests plotted well below the failure envelope in the net stress space, which is indicative of resilient and elastic behaviour. Increased axle loading instead resulted in decreased permanent strain, until the critical level of repeated deviator stress of 32.5 tonnes per axle was found, where the permanent strain increased. It is therefore concluded that, as a result of the increased shear strength from the strain hardening property of unsaturated materials, an increased axle loading of 32.5 tonnes per axle can be safely sustained by the tested materials provided the matric suction in the soil is greater than 50 kPa. / Thesis (PhD)--University of Pretoria, 2020. / Civil Engineering / PhD / Unrestricted

Soil is the most widely used material in the construction of various civil infrastructure. Various types of soils are extensively used in its natural or compacted form in the construction of dams, canals, road and railway subgrades, and waste containment structures such as soil covers and liners. These infrastructure and foundation soils are exposed to the influence of environmental factors. In the permafrost and seasonally frozen regions, soils can be in different states (e.g., saturated or unsaturated, frozen or thawed, or combinations of them) due to the variations in moisture content and temperature. The soil-water characteristic curve (SWCC), which is the relationship between soil water content and suction, is used in the interpretation and prediction of unsaturated soils behavior. Similarly, the soil-freezing characteristic curve (SFCC), which is the relationship between unfrozen water content and subzero temperature, is used in the prediction and interpretation of frozen soils behavior. In this thesis, the SWCC and SFCC of two Canadian soils (i.e. Toronto silty clay (TSC) and Toronto lean clay (TLC)) were extensively investigated for better understanding the fundamental relationship between SWCC and SFCC. The soil resilient modulus (MR) is a key material property used in the rational design of pavements. Experimental investigations were undertaken to determine the MR of five Canadian soils (i.e., TSC, TLC, Kincardine lean clay (KLC), Ottawa Leda clay (OLC), and Indian Head till (IHT)), considering the influence of moisture and temperature, with the aid of an advanced triaxial testing equipment. Two simple models were proposed for estimating the MR of frozen soils, in this thesis. In addition, an artificial neural network (ANN) model was developed for estimating the MR of the five Canadian soils considering various influencing factors. The conclusions from the various studies in this thesis are succinctly summarized below. (1) Four expressions (i.e. power relationship, exponential relationship, van Genuchten equation, and Fredlund and Xing equation) that are widely used for representing the SFCC were selected for providing comparisons between the measured and fitted SFCCs for different soils. The results suggest that the exponential relationship and van Genuchten equation are suitable for sandy soils. The power relationship reasonably fits the SFCC for soils with different particle sizes, but not for saline silts. The Fredlund and Xing equation is flexible and provides good fits for all the soils. (2) The SFCC and SWCC of TSC and TLC were experimentally determined, analyzed, and compared. Many factors influence the reliable measurement of SFCC, which include sensors’ resolution and stability, sensor calibration for each soil, and thermodynamic equilibrium condition. The hysteresis of SFCC for the two soils is mainly attributed to the supercooling of pore water. The quantitative dissimilarity in the measured SFCC and SWCC may be attributed to specimen structure variations during compaction and saturation, and during freezing / thawing processes, and cracks formation due to sensors insertion. In addition, some fundamental differences may exist between the drying / wetting and freezing / thawing processes, resulting in dissimilarity. (3) Two novel models were proposed for the estimation of MR of frozen soils. The semi-empirical model extends the mechanics of unsaturated soils and employs SFCC for prediction. Several coarse- and fine-grained saturated soils were used to validate this model. The empirical hyperbolic model was proposed considering that the frozen MR versus subzero temperature relationship resembles hyperbola. This model was validated on coarse- and fine-grained soils under saturated / unsaturated conditions. The hyperbolic model has wider application since it can be used for both saturated and unsaturated frozen soils. Both the models are simple and promising. (4) The MR of five Canadian soils subjected to wetting and freezing was determined by using the GDS ELDyn triaxial testing system. A freezing system was established for controlling the desired testing temperatures within the soil specimens. The results suggest: (i) The effect of subzero temperature on the MR is significant. (ii) For TLC, KLC, OLC, and IHT, the frozen MR versus subzero temperature relationship of the saturated specimen typically has steeper slope than specimen at the optimum water content, for the temperature range from 0 to -5 °C. (iii) The effect of stress levels on the frozen MR depends on soil type, water content, and subzero temperature. Lastly, (iv) Loading frequency does not show a significant influence on the frozen MR. (5) The MR of the five Canadian soils was determined considering wetting and freeze-thaw (F-T) conditions. The results suggest: (i) The F-T cycles result in weak soil structure due to reduction in suction, particles movement, loss of cohesion, and formation of cracks / channels. (ii) The critical numbers of F-T cycles were determined as 1, 1, 2, and 1 for TLC, KLC, OLC, and IHT at the optimum water content, respectively. (iii) The percentage of reduction in MR after the critical number of F-T cycles was strongly related to the plasticity index for specimens tested at the optimum water content. (iv) The wetting process results in the decrease in suction and enlargement of soil pores. Consequently, relatively low MR values were measured at high water contents, and the effect of F-T cycles becomes insignificant. Finally, (v) The effect of stress levels on the MR was dependent on the initial water content of the specimen and soil type.

The movements of salt and water, particularly in fairly dry soils or compacted shales, are important processes in both agronomic studies and hydraulics of deep ground water which is important in the location of oil.

Field compaction control is arguably the most common yet critical quality control procedure in geotechnical engineering. Since the early 1930s, the systematic process for performing quality control of compacted soils has often been performed by measuring the in-place dry unit weight (or density) and as-compacted soil moisture content after placement in a fill. However, the current practice overlooks several facts resulting from comparing soil prepared and compacted in the laboratory to soils placed and compacted in the field. These issues include comparing the compaction energy in the lab versus what is applied in the field, and the behavior of saturated soils in the laboratory to the performance of unsaturated soils in the field. To address some of these gaps, this study presents a new saturation-based framework for compaction quality control. The aim of this new framework is to reduce the uncertainties and assumptions of the compaction control process and provide practicing engineers with further insight into the key engineering attributes of compacted soils. The proposed saturation-based approach compares a degree of saturation difference to a normalized dry unit weight ratio, making saturation upon compaction the controlling diagnostic variable and the focus of the monitoring effort. In essence, the optimal compaction conditions will be referenced to a characteristic saturation state near 80%. Compared to the conventional quality control system for field compaction, the saturation-based approach is developed with the same field and reference data collected for most earth fill projects. The results of this approach enhance the engineering judgment required to match the laboratory reference values to the field conditions. For illustration purposes, the proposed saturation-based framework is applied to compaction control data of a large earth dam and compared against the conventional method side-by-side. The proposed framework builds on the unique physical features of the "family of curves" and expands the ability of the user to select the compaction criterion using that relationship to produce project design properties. Overall, the proposed approach enhances the knowledge of the physical behavior of compacted soils and provides a more comprehensive understanding of the long-term performance of compacted fills.

The shear stress in an unsaturated soil increases rapidly with limited shear strain to a peak value and then drops gradually with a further increase in the shear strain until a residual value is reached. In other words, there is a significant strain-softening behavior under large shear deformation. A variety of geotechnical structures (e.g., slopes, foundations, retaining walls and piles) associated with unsaturated soils typically undergo a large progressive deformation prior to reaching failure conditions due to the influence of environmental factors (e.g., rainfall infiltration and wetting-drying cycles). As a result, the shear strength of soils in sliding zones typically reduces from a peak to a residual value with the progressive development of large shear deformation, while the shear strength of soils in other zones are still at the peak level. In other words, in many scenarios the strain-softening behavior of unsaturated soils can significantly influence the mechanical behavior of geo-structures. Therefore, a thorough understanding of the shear strength behavior of unsaturated soils during strain-softening is required to reliably interpret the mechanical behavior of geo-structures that undergo large shear deformation. Significant advances have been made during the last thirty years to understand and model the strain-softening behavior of unsaturated soils. Most of these studies however focus on the strain-softening behavior within a relatively small shear deformation due to the limitations of the experimental apparatuses. Only limited experimental studies under large shear deformation were reported based on the modified suction-controlled ring shear apparatus. Therefore, more investigations are still required to provide a comprehensive understanding of the shear strength behavior of unsaturated soils during strain-softening under large shear deformation. Studies presented in this thesis are directed towards investigating the shear strength behavior of unsaturated soils during strain-softening and its application in geotechnical engineering practice. The following studies have been conducted: (i) A state-of-the-art review of the strain-softening behavior of unsaturated soils published in the literature during the past three decades is summarized. The physical mechanisms and modelling methods of the strain-softening behavior and the peak, critical and residual shear strength of unsaturated soils are investigated. (ii) A disturbed state concept model is proposed to predict the variation of shear stress in unsaturated soils during strain-softening process under drained condition. Five sets of experimental data gathered from the literature on unsaturated soils varying from coarse- to fine-grained soils are used to verify the proposed model. The proposed model can provide reasonable predictions for the strain-softening stress-strain relationships of various types of unsaturated soils. The model is simple in concept and all the required parameters can be obtained from conventional saturated and unsaturated shearing tests and pressure plate tests. (iii) Two sets of suction-controlled multistage ring shear tests are conducted on unsaturated SP-SM soil and Indian Head till (IHT), respectively. The variation of the shear stress, void ratio, and water content of specimens during shearing (the shear displacement reaches 100 mm) under multi levels of net normal stress and matric suction are described and discussed. The influence of matric suction and net normal stress on the residual shear strength envelops of unsaturated soils are critically discussed. (iv) A model for predicting the residual shear strength for a wide range of unsaturated soils comprising coarse- to fine-grained soils is developed in terms of two stress state variables (i.e., the net normal stress and matric suction) by using the soil water characteristic curve as a tool. The model is formulated and validated based on experimental data in a series of suction-controlled ring shear tests using the axis-translation technique, including the two sets of tests (SP-SM and IHT) conducted in this research and another three sets of tests (SM, SC-SM and CH) gathered from the literature. The fitting parameters are related to the plasticity index (Iₚ); thus, only four basic parameters (i.e., cᵣ', φᵣ', Sᵣ and Iₚ) are included in this approach. (v) A series of slope stability analyses of a landslide in unsaturated condition are conducted using Geoslope software based on the peak and residual shear strength parameters. The analyses results highlight the role of residual shear strength in the slope stability of unsaturated soils. In summary, the mechanical behavior of unsaturated soils under large shear deformation is comprehensively investigated in this thesis. The experimental results of the suction-controlled ring shear tests reported in this research contribute towards understanding the fundamental shear strength behavior of unsaturated soils during strain-softening under large shear deformation. The models proposed in this research provide simple tools to predict the shear strength of unsaturated soils under different levels of shear deformation.

An investigation into the transient impacts of flood loadings on earthen embankments was conducted. Two embankments were instrumented and monitored over a period of four years. One of these embankments was a levee located along the Mississippi River just north of Cairo, Illinois. The other embankment was part of a catchment basin at the Engineer Research and Development Center located in Vicksburg, MS. Tensiometer and porous block sensors were used to monitor the pore water pressures in the embankments. It was found that when measuring the field soil water retention, tensiometers were more responsive than porous block sensors at low suctions; although, at shallower depths, the tensiometer performance was limited during periods of extended drying. It was shown that during the start of flooding, pore water pressures in the embankment soils were near −10 kPa at depths less than 2 m, which was greater than the normally assumed hydrostatic conditions. An investigation into flood hydrographs collected from across the United States showed that flood durations could be hundreds of days long. These hydrographs were collected over a period of 10 years. It was found that the recorded peak flood stage exceeded the major flood stage 11% of the time. An uncouple transient seepage model of a 2015 Mississippi River flood event that occurred at the Cairo levee showed that an uncoupled model could simulate the field measurements; however, the material properties that resulted in the most accurate simulation differed from those measured in the laboratory. Soil water retention characteristics of the embankment soils were assessed, and it was found that laboratory measured soil water retention curves could be used to bracket field measurements. Slope stability analyses were performed as a proxy to assessing the progression of the wetting front in the levees. Accounting for the increase in shear strength due to the presence of matric suction resulted in minimal impacts to stability factors of safety for levee embankments during flood loadings. The results of this investigation will help to improve the reliability of transient seepage analyses and provides guidance for future embankment monitoring investigations. / Doctor of Philosophy / An investigation into the movement of flood water through flood control embankments was conducted. Typically, analysis of this phenomenon is performed independent of the effects of time. For this investigation, the impacts of time were considered. When considering the effects of time dependent loadings, an initial distribution of water pressures must be considered. Typical assumptions regarding these distributions were investigated using four years of sensor measurements from two embankments. These measurements were also used to investigate appropriate material properties when considering saturated and unsaturated soil properties necessary for these analyses. Results show that typical assumptions may not be appropriate regarding initial water pressure distributions. Additionally, recommendations for assigning material properties were provided and it was found that these types of analyses can simulate flood loadings, but a range of material properties must be explored to understand the full range of performance. The impact of these results will lead to better predictions of embankment performance during flood loadings.

Este trabalho apresenta resultados de provas de carga em placa com monitoramento do teor de umidade e da sucção matricial do solo, realizadas a 1,5 m de profundidade em um solo não saturado e colapsível, considerado representativo da região centro-oeste do estado de São Paulo. Foram executados ensaios com e sem inundação da cava, do tipo lento (SML), rápido (QML) e misto (lento até um determinado estágio, passando a rápido a partir de então). O acompanhamento do teor de umidade e da sucção durante os ensaios foi realizado através de pesagem de amostras, com a utilização da técnica de reflectometria no domínio do tempo (TDR) e com o auxílio de tensiômetros. A partir dos dados obtidos, procedeu-se a uma análise da influência da sucção na resposta do sistema solo-placa. Foi observado que o aumento da sucção provoca um crescimento substancial da capacidade de carga do solo. Comparando-se ensaios inundados com não inundados, observou-se que o colapso do solo mostra-se mais proeminente com o aumento da sucção. Com respeito à técnica de reflectometria no domínio do tempo, a mesma demonstrou um desempenho satisfatório para medidas de teor de umidade in situ. / This work presents results of plate load tests with measurements of soil moisture content and matric suction, carried out at a 1.5 m depth in an unsaturated collapsible soil, considered as representative of the middle-west region of the state of São Paulo, Brazil. The tests were performed with and without the wetting of the pit. Three test methods were used: slow maintained load (SML), quick maintained load (QML), and mixed tests (SML up to a certain stage, switched to QML from there on). Soil moisture content and suction were measured through gravimetric determination on samples, the time domain reflectometry technique (TDR), and tensiometers. The gathered data allowed an assessment of the influence of the soil suction over the soil-plate system response. It was observed that the increase in suction causes a substantial increase in soil bearing capacity. Comparisons between wet and non-wet tests showed that soil collapse is higher when suction rises. The time domain reflectometry technique showed a promising performance with respect to in situ water content measurements.

Este trabalho apresenta os resultados de provas de carga em placa com monitoramento da sucção e do teor de umidade, realizadas nas profundidades de 1,5, 4,0, 6,0 e 8,0 m no Campo Experimental de Fundações da Escola de Engenharia de São Carlos, cujo perfil geotécnico é representativo da região centro-oeste do Estado de São Paulo. Para cada profundidade também se realizou ensaios com pré-inundação do terreno, representando a condição de sucção de matricial nula. Constatou-se que a profundidade e a sucção matricial média no solo sob a placa influem significativamente na capacidade de carga no sistema solo-placa. As curvas tensão x recalque obtidas não se caracterizam ruptura nítida nem indicam o modelo de ruptura física, exigindo assim, a adoção de critérios de ruptura convencional. Devido à semelhança dessas curvas com as curvas e x log\'sigma\' de ensaios de adensamento e, por isso, utilizou-se o método de PACHECO SILVA (1970) como um critério ruptura convencional, comparando-se com valores da tensão de pré-adensamento obtidos em laboratório. Também se utilizou o recalque de 25 mm para obter a correspondente ruptura convencional. / This work presents the results of plate load tests with measurements of soil matric suction and moisture content, carried out in the depths of 1,5 , 4,0 , 6,0 and 8,0 m at the Experimental Field of Foundations of School of Engineering of São Carlos, whose soil profile is representative of the middle-west region of the State of São Paulo. For each depth tests with previous flooding of the land, representing the condition of null matric suction. Were also carried out It was verified that the depth and the average matric suction in the soil under the plate influence significantly in the load capacity of the soil-plate system. The stress settlement curve obtained does not characterize clear rupture nor indicate the model of physical rupture, demanding the adoption of approaches of conventional rupture. Due to the similarity of those curves with the e x log\'sigma\' curves from consolidation tests and, to put that, PACHECO SILVA (1970) method was used as an approach to conventional rupture, being compared with values of the preconsolidation stress obtained in laboratory. It was also used settlement of 25 mm to obtain the corresponding conventional rupture.

Apresenta-se um método para a previsão da curva tensão-recalque a ser obtida em provas de carga em placa circular, instalada a diferentes profundidades e com vários diâmetros, para diversos níveis de sucção em solo não saturado. Também é proposto um procedimento para correção dos recalques medidos em ensaio com carregamento do tipo rápido, para a obtenção dos correspondentes recalques estabilizados de um ensaio lento. A previsão realizada apresenta resultados bastante próximos dos obtidos em provas de carga, verificando-se assim a aplicabilidade do método proposto para reproduzir o comportamento do sistema solo-placa neste tipo de solo. Além disso, este trabalho apresenta um estudo do módulo de deformabilidade a partir de ensaios de campo e de laboratório, para o solo do campo experimental de fundações da EESC/USP, aplicando conceitos da mecânica dos solos não saturados. Em campo, o módulo de deformabilidade foi obtido por meio de provas de carga em placa circular com e sem inundação prévia do solo. Em laboratório, o módulo foi obtido a partir de ensaios de compressão confinada e triaxiais, ambos com sucção controlada. Observou-se que a sucção matricial tem grande influência no módulo de deformabilidade do solo. / This thesis presents a method for prediction the stress-settlement curve to be obtained in circular plate load tests, at different depths and with several diameters, to differents suction levels on unsaturated soil. A procedure is also proposed for correction the settlement measured with a quick maintained load (QML) tests to obtain the stabilized settlement in a slow maintained load (SML) tests. This prediction presents quite results of the real measured stress-settlement curve through load tests, thus verifying the applicability of the proposed method to reproduce the behavior the soil-plate system in this soil. Besides, this work presents a study of the deformation modulus using field and laboratory tests, to the soil of the foundations experimental field of EESC/USP, applying concepts of the unsaturated soil mechanics. In field, the deformation modulus was obtained by means of circular plate load tests with and without previous soaking of the soil. In laboratory, the modulus was obtained using confined compression and triaxial tests, both with controlled suction. It was observed that the matric suction has great influence in the soil deformation modulus.

Este trabalho teve como objetivo utilizar um tensiômetro de alta capacidade (TAC) para realizar medidas diretas de sucção em ensaios de resistência ao cisalhamento de um solo compactado. Foram realizados ensaios de resistência à compressão triaxial do tipo consolidado drenado na condição saturada, ensaios consolidados não drenados na condição não saturada e ensaios de resistência à compressão simples. Os ensaios de compressão simples foram realizados em uma série de repetibilidade e outras duas para a verificação da influência da sucção na resistência de um solo compactado, fazendo-se variar a massa específica aparente seca do solo e o teor de umidade. Em todos os ensaios de compressão simples o TAC registrou aumento da sucção até cerca de 3% de deformação axial, quando a partir daí a sucção tendeu a estabilização. Os valores de resistência à compressão simples e de sucção tenderam a crescer com o aumento da massa específica aparente seca do solo e com a redução do teor de umidade. A resistência foi correlacionada com a sucção e com a massa específica aparente seca para os ensaios com teor de umidade constante e com a sucção e o teor de umidade para os ensaios com massa específica aparente seca constante. Os ângulos \'fi\'b que relacionam o aumento da resistência com a sucção variaram dependendo da forma de determinação e foram maiores quando determinados diretamente a partir da resistência não drenada do que quando determinados a partir da proposta de Fredlund et al. (1978). / The purpose this study was to measure the suction with high capacity tensiometer (HCT), placed at the bottom of the sample, during shear strength tests on a compacted soil. Three kinds of shear strength tests were performed on the sample in different conditions of saturation: triaxial consolidated drained in saturated condition, triaxial consolidated undrained with constant water content and unconfined compression in unsaturated condition. To assess the development of the suction and its influence on the shear strength of a compacted soil, the unconfined compression tests were performed in a series of repeatability and the others with variation of the dry density and water content of the sample. In all unconfined compression tests, the HCT recorded increase of suction until 3% axial deformation, approximately, from where the suction tended to stabilized. The values of compressive strength and suction increased with higher dry densities and lower water contents. Firstly, the shear strength was correlated with suction and the dry density, for constant water content tests. In sequence, it was correlated with suction and the water content for constant dry density tests. The \'fi\'b angle, that relates the increase of the strength with suction, varied depending of determination method. Its values were higher when directly determined from the undrained strength than when determined from the model presented by Fredlund et al. (1978).