Soil-rock Mixture Slope Deformation Behavior Study Based On Structuredness

In recent years, great concern has been focused on the geo-material involved eluvium, colluvium and deluvium, namely soil/rock mixture (SRM),during the process of slope (landslide) remediation, hydropower station and foundation engineering cgnstruction under the background of the infrastructure further development. SRM is characterized by the complex composition, irregular structure and difficult sampling, thus lead to great difficulty in corresponding research. Accordingly, the existing theory of geotechnical mechanics is incapable in providing precise description and generalization of this special geo-material.The SRM distributed most commonly in the debris landslide of Three Gorges Reservoir Area (TGRA) is the colluvial-deluvial soil-rock mixture (CDSRM), which refers to the SRM formed from the long-term geological history of colluviation and deluviation. Generally, this category of debris landslide tend to give rise to serious risks of human life and property security, associated with the characteristics of great scale, complex inducing factors, unpredictable failure and complex sliding condition. Therefore, it is significant to combine SRM characteristic studies and the debris landslide stability analysis. In fact, as the TGRA started to impound in 2003, and the highest water level reaches to 175m at present, debris landslides in TGRA generally deformed obviously, which is affected by both water level fluctuation and concentrated rainfall. Therefore, it is of great importance to carry out reasonable stability evaluation, deformation behavior analysis, and provide effective prevention and remediation strategies to release the damage to people’s safety and property.The most common strategy when dealing with CDSRM slope (landslide) is homogenizing CDSRM without considering the structural characteristics, and utilizing the geotechnical properties of the matrix soil instead of the overall properties. This simplification ignores the contribution of the structuredness of SRM thus produces great error in computational analysis. The dissertation selected a representative debris landslide in TGRA, Baishuihe landslide, as the research object, aimed at study the deformation behavior of landslide with a point cut of the structuredness. The structuredness can be described from two aspects, mesostructuredness and macro structuredness. Mesostructuredness study is discussed by the statistical analysis to the corresponding quantified mesostructural and morphological indexes of the ingredient blocks, based on digital outcrop image process (Chapter II). With regard to the macrostructuredness study, it is quantified discussed by the auto-correlation structure based on anisotropic variogram analysis for the hydraulic properties of CDSRM obtained from surface nuclear magnetic resonance (SNMR) technology (Chapter V). On the basis of mesostructuredness, stochastic CDSRM surrogated models can be generated (Chapter Ⅲ) for numerical tri-axial compression tests. The numerical experimental results, associated with back analysis and analytical analysis, can be used to determine the shear strength of the CDSRM in representative landslide (Chapter IV). On the basis of macrostructuredness, the influence of different auto-correlation structure of geotechnical properties to the seepage process, factor of safety and deformation behavior of a hypothetic CDSRM slope is discussed utilizing a non-invasive random FEM method. In addition, the computation model for the representative landslide is generated, and conducts seepage-stress coupling analysis under one hydraulic year of actual outer condition (water level fluctuation and rainfall), considering the spatially variable hydraulic conductivity. The simulated results associated with the actual monitored data, is analyzed for deformation behavior discussion.The innovations of this dissertation can be stated as follows.1. Based on the observation of block distribution characteristics, questioned the applicability of commonly adopted block/matrix threshold (B/M), and carried out a series of numerical mechanical and permeability experiments for actual CDSRM outcrop specimens with various B/Ms to select the most adoptable B/M. The B/M can be used in generating stochastic CDSRM surrogate models on the basis of the mesostructural statistics. Meanwhile, the scale effect of mechanical properties is discussed to select a rational specimen dimension.2. Quantificationally analyze the macrostructuredness using anisotropic variogram of geotechnical properties for the first time. The spatial hydraulic properties by mean of SNMR data interpretation is used as the spatial correlated variables, for correlation distance and maximum correlation orientation analysis obtained by anisotropic variograms.3. Propose a non-intrusive random-FEM method, which combines the sequential Gaussian condition simulation results to the FEM analysis. This method has the ability of considering geotechnical properties random field with different spatial correlation structures and the corresponding influence on the slope stability and deformation behavior are analyzed.The conclusion can be drawn as follows.1. Most research of the dissertation takes Baishuihe landslide as the example, thus the engineering geological characteristic of this representative debris landslide was introduced according to the extensive engineering geological survey. As investigated, the distributed CDSRM has an uneven thickness from 7.75 m to 38.5 m. Most of the blocks in the landslide have the size of 20-200mm, and partially reach to 300-500 mm. The observations indicate that the block dimension does not increase with the studying dimension, which differences from the self-similarity assumption for melange.2. Digital image process has been utilized in the mesostructural and morphological quantitive characterization based on the indexes of a mesostructuredness quantitive evaluation system. As for the mesostructural indexes, results indicated that the probabilistic model of maximum observed dimension (MOD) and aspect ratio (AR) are all in line with lognormal distribution, while the OA obeys normal distribution. And the areal block proportion (ABP) had a significant linear correlation with the fitting parameters for the probability density function of MOD. Otherwise, the morphological quantitive indexes have been investigated utilizing Fourier series. Specifically, the form index and angularity index had been calculated to explore the characterizations. It is observed that more than 75% of the blocks have the Fourier form index (FFI) less than 1.2, while 92% of the blocks have the Fourier angularity index (FAI) less than 0.05. The result indicates that the majority of the blocks are sub-rounded with considerable small angularity.3. Stochastic CDSRM surrogate models based on mesostructuredness are tended to be generated for numerical experiments. Two preliminary issues, B/M and scale effect, however, should be solved. With regard to the B/M determination, numerical mechanical and hydraulic experiments on the CDSRM outcrop simulated specimens with various pre-defined B/Ms have been performed to exploring the best adoptable threshold. The obtained results of geotechnical parameters (unconfined compression strength, Young’s modulus, shear strength and hydraulic conductivity) obtained from different thresholds, associated with grain size distribution, has been analyzed. It reveals that 5 mm is the most rational threshold for CDSRM through comprehensive comparison.With regard to the scale effect, the stochastic large models with different block proportions (from 20% to 70%) have been generated according to the mesostructuredness and constant 5 mm B/M, from which the specimens with different dimension were sampled. Unconfined compression tests were performed to these specimens with different sizes and slenderness ratios to investigate the impact of the size and slenderness effect on the strength, deformability and failure process. The results indicates that the unconfined compression strength varies in a logarithmic correlation with the specimen size and tends to be asymptotic when side length of the specimen reaches up to 1500 mm. The ubiquitously distributed relative larger blocks makes the specimen with higher BPs more of homogenous, thus exhibits strength reduction with specimen sizes. While the small amount of larger blocks in specimens with lower BPs increases the tortuosity and results in strength enhancement when the size increases. Regardless of the platen effect and micro-defects in specimens, the UCS and E values decrease with the slenderness of specimens as the slender specimens provide more possible propagating path for failure zones. The results follow the exponential and the power regression functions, respectively. This trend is much profound for the higher BPs, and tends to be asymptotic when slenderness ratio reaches to 2. Based on the simulated observations that the strength tend to be asymptotic when side length of the specimen reaches up to about 1500 mm, and practical experience, the recommended specimen dimension for thick CDSRM specimen is 1000 mm×2000 mm.4. Back analysis, analytical analysis and numerical experiments have been conducted in determining the shear strength of CDSRM, exemplifies by the CDSRM in Baishuihe landslide. The first method determined the optimal shear strength combination of the sliding zone of Baishuihe landslide by comparing the monitored GPS displacement with the numerical simulation results within a certain period of time using unweighed pair-group method of analysis (UPGMA). The second method determines the shear strength of sliding mass and sliding zone according to the proposed experimental estimation model on the basis of extensively laboratory and field test. The third method performed numerical tri-axial compression tests on the stochastic CDSRM specimens in terms of stochastic CDSRM surrogate model generation system.60 sets of numerical tests had been performed for each material, sliding zone and sliding mass, and 120 sets of shear strength combination has been obtained totally. Taken the mean value of c and φ for comparison with the results of the above two method, the simulated shear strength tended to have greater φ and smaller c. Based on the stress state analysis of the specimen under tri-axial compression, it is considered that the friction between block and matrix embodied in the remarkably increased internal friction angle under large loading condition. Based on the comprehensive comparison of the obtained results, the recommendation shear strength for the CDSRM is proposed.5. Despite of the seemingly chaotic structure of CDSRM, the mechanical and seepage properties have characterization of a certain macrostructuredness, which can be illustrated by spatial correlation (variability). The dissertation conducted field-scale anisotropic variogram analysis for the seepage parameters (water content w and saturated hydraulic conductivity Ks) interpreting form the SNMR technology. With regard to the horizontal analysis, the hydraulic conductivity exhibited the characterization of geometric anisotropy with the highest correlation distance (508.37m) at the orientation of EN9°, and the lowest correlation distance (286.75m) at the orthometric NW9°. Meanwhile, the water content exhibited the characterization of zonal anisotropy with the highest correlation distance (262.98m) at the orientation of NW45°, and the lowest correlation distance (113.82m) at the North orientation. The field-scale anisotropy of the properties has great relation with the experienced engineering geological action and deformation behavior of the landslide. Few attempts were focus on the vertical analysis. Kriging interpolation had been performed to estimate the distribution of hydraulic conductivity on the basis of the analyzed variogram and anisotropic correlation distance. The interpolation result provided us with a visualized actual hydrogeological condition for the study field and remedies the insufficiency of measured data.6. For the purpose of studying the influence of spatially variable geotechnical properties to the seepage process, slope stability and deformation behavior, a hypothetic SRM slope is selected as the corresponding research object. A non-invasive random FEM method based on conditional stochastic simulation (sequential Gaussian simulation) method was proposed to generate the spatial variable random field for c, φ and Ks, according the appointed correlation structures. There are 7 correlation structures considered, including the isotropic structures with different correlation distance and anisotropic structures with different maximum correlation orientations.50 realizations have been simulated and calculated for each correlation structure scenarios.The calculated seepage field for different correlation structures of Ks indicated that the anisotropic spatially Ks produces higher negative pore-water pressure, and more tortuous pore-water pressure profile. As for the isotropic scenario, a higher correlation distance produces larger bound of negative pore-water pressure and the maximum difference of the upper and lower bounds of the groundwater table (D), which implies a higher degree of variation. The anisotropic tend to produce larger pore-water pressure bound, especially the cases that maximum correlation orientation vertical to the section line. According to the mean factor of safety (Fs) obtained by different parameters with the same coefficient of variation, spatially variable φ tends to produce lower Fs than other variable parameters and deterministic analysis, which implies that design approaches using mean φ values tend to be risky by ignoring the influence of spatial variation. Spatially variable c value, comparatively speaking, has negligible impact on the mean Fs. Besides, the spatially variable shear strength has more significant impact on the critical slip surfaces than spatially variable hydraulic conductivity. For the isotropic scenario, the critical slip surfaces variation bound enlarges with the increase of correlation distance of shear strength. For the anisotropic scenario, the cases that with the maximum correlation orientation horizontally and parallel to slope surface have the greatest critical slip surfaces variation bound.7. The geological-seepage-mechanical simulation model is generated subsequently based on aforementioned outcomes. The spatially variable hydraulic conductivity of the CDSRM was determined using the Ordinary Kriging interpolation method on the basis of known SNMR tested points. Seepage-stress filed coupling simulation results of the actual working condition of one hydrological year of rainfall and water level fluctuation indicated that Baishuihe landslide can be regarded as a retrogressive landslide, and the main controlling factor is rainfall. The groundwater table rises gradually as the combination action of rainfall and water level, which reduces the shear strength of the landslide mass beneath the water table as pore-water pressure. Therefore, the deformation occurs at the middle platform of the landslide at the beginning. With the further rise of water table, even when the whole landslide slide mass beneath the water table, significant deformation is produced at the rear area as the reduced shear strength and steeper slope profile. The reliability of the simulation results has also been verified through comparing with the monitoring data and macro-scopical deformation investigation.The dissertation aims at systematically study the slope deformation behavior based on the structuredness of CDSRM through comprehensive technologies, exemplified by a specific case. This research outline can be popularized to the study of the same category of landslide, thus has good engineering reference significance.