Reliability Analysis Of Geotechnical Structures Considering Soil Heterogeneity

There exists many uncertainties in geotechnical structures in hydraulic and hydropower engineering. Three primary sources of the uncertainties are inherent uncertainty, measure error and model uncertainty. Among them, soil heterogeneity is one of the most important factors that play a central role on reliability of geotechnical structure. Three problems exist in the reliability analysis of geotechnical structure in the presence of soil heterogeneity. Firstly, soils in different depth always experience various geologic, environmental and chemical effects. Hence, soil parameters customarily exhibit a non-constant trend with depth. This non-constant trend is also known as non-stationarity in reliability analysis. An ignorance of this trend may lead to a misleading result in reliability analysis of geotechnical structures. However, most of the existing reliability analyses adopt stationary random field model to consider the spatial variability of soil parameters. The non-stationary in soil property is not well studied. Secondly, geological uncertainty is another type of soil heterogeneity. It appears in the form of one soil layer embedded in another or the inclusion of pockets of different soil type within a more uniform soil mass. The geological uncertainty has received considerable attention in other areas, such as petroleum engineering, and groundwater and contaminant transportation. The role of the geological uncertainty has been studied in geotechnical engineering. For example, Eisbacher (1971) pointed out that the geological uncertainty may control the distribution and direction of rockslides after surveying twenty-five rockslides in British Columbia. However, attempts to consider the geological uncertainty in geotechnical practice have been limited. To address this bad situation, the first requisite step is the simulation of the geological uncertainty. Yet, existing models for the geological uncertainty simulation in other fields cannot be directly applied to geotechnical engineering. The reason mainly lies in the different sizes between geotechnical structure and structure in other field. Geotechnical structures are always at the scale of meters. They are much smaller than the sturctures in petroleum engineering, or groundwater and contaminant transportation. Hence, a new method should be proposed to simulate the geological uncertainty, which needs to match the size of geotechnical structures. Furthermore, the effect of geological uncertainty on geotechnical structure should be investigated. Thirdly, geotechnical reliability analysis customarily focused on the failure of probability. The failure mode (such as the critical slip surface of a slope) is rarely studied. However, the critical slip surface controls the size and location of the landslide mass, which are highly related with the loss caused by slope failure. Hence, the critical slip surface of a slope should be well considered in reliability analysis.To address the three problems mentioned previously, this thesis aims to propose simulation methods for both the non-stationary spatial variability and the geological uncertainty. On these bases, the influence of these two forms of soil heterogeneity on the reliability of geotechnical structures is analyzed. Meanwhile, the distribution of critical slip surface for a two-dimensional slope in the presence of spatial variability for soil shear strength parameters is also studied. The details and conclusions for the thesis are summarized as follows.(1) The background and significance of the topic, i.e. reliability analysis of geotechnical structures considering soil heterogeneity, is presented. Firstly, the modeling method for uncertainties in soil parameter is briefly reviewed. Unresolved issues in the random field modeling is pointed out. Secondly, the reasearch method and object for reliability analysis of geotechnical structures is concluded. Thirdly, the research status of geological uncertainty in geotechnical area is summarized. The key problems of reliability analysis of geotechnical structure in the presence of geological uncertainty are highlighted. The application of coupled Markov chain model in engineering field is also introduced.(2) Simulation methods for non-stationary random field of shear strength parameter are proposed. The non-constant trend of both undrained shear strength and effective friction angle is validated and quantified by empirical equations and in-situ test data. The trend of statistics for undrained shear strength and effective friction angle is compared. On this basis, random field models are respectively constructed for the two shear strength parameters. This work lays the foundation of reliability analysis of geotechnical structures using total stress and effective stress analysis method when the spatial variability of shear strength parameters with their means varying with depth is considered.(3) A stochastic method to determine the critical slip surfaces of soil slopes considering spatial variability of soil strength parameters is proposed.. The location, size and distribution of critical slip surface in the presence of spatially variable shear strength parmater are investigated. The method provides an effective tool to determine the failure mode of slopes when spatial variability of soil parameters is considered.(4) A probabilistic method is proposed for infinite slope stability analysis considering the variation of soil shear strength parameters with depth. First, the non-stationary random field of spatially varying soils is constructed. Second, the effect of spatial variability of soil parameters on failure probability and critical slip surfaces is investigated. An infinite slope case is adopted to verify the effectiveness of the proposed method. Real landslide cases are collected to verify the result of numerical analysis. The results provide an evidence for the abnormal failure depths of shallow slope in reality, i.e. failures occuring at the middle depth of slopes.(5) A stochastic method is proposed for bearing capacity analysis of strip footing considering the variation of the mean and standard deviation of undrained shear strength parameters with depth. The effect of spatial variability of the undrained shear strength parameters on the ultimate bearing capacity is investigated. The results of bearing capacity associated with stationary and non-stationary random field models are compared. The results show that the stationary random field model will greatly underestimate the reliability of the ultimate bearing capacity. This conclusion is helpful to prefect the shallow foundation design.(6) A practical estimation method for the horizontal transition probability matrix (HTPM) of a coupled Markov chain model is proposed. This method can be used to simulate the geological uncertainty for geotechnical application. Borehole data from various sites are collected. The first-order Markovian property in soil state transitions is tested. The effectiveness of the proposed method is evaluated. The HTPM and vertical transition probability matrix for soil state transition in reality is studied. The proposed method can effectively estimate the HTPM for soil state transition. This work provide a basis for geological simulation for geotechnical structures.(7) A borehole data-based method is proposed to reduce the uncertainty in factor of safety (FS) of slope in the presence of geological uncertainty. The coupled Markov chain model and borehole data is adopted to simulate geological uncertainty. Monte Carlo simulation of slope stability analysis is conducted using a finite element stress-based slope stability analysis method. The proper distance from the outmost borehole to the slope crest (or slope toe) is investigated. Effect of different borehole layout schemes on reducing the uncertainty in FS and critical slip surface of slope is analyzed. The role of borehole location, borehole number associated with a borehole layout scheme is investigated. This work provides a guidance for the design of site investigation scheme.