Engineering Behavior And Slope Stability Analyses For Residual And Diluvia Soil Under Different Acting Forms Of Water

In Midwest of China, there is a large area of residual and diluvia soil covering the mountainous region with complicated geography and geological condition. When constructing high level roads, a number of high and abrupt slopes will come up. The stability of these slopes is primarily threatened by the factor of water. Considering its theoretical and practical significance, this work devotes into the test investigation for the engineering properties of residual and diluvia soil under various acting forms of water, and the slope stability analysis based on the test results, along with the system reliability based optimization design for retaining structure-anchored anti-sliding shaft system usually used in these slopes. The paper is supported by the National Natural Science Foundation of China (Grant No.50878082), Construction Technology Project in Western China of Transportation Ministry (200631880237) and Key Program of Natural Science Foundation of Hunan Province (09JJ3104). The main work and achievements are as follows:(1) The variation of the shearing strength of residual and diluvia soil against different water contents was obtained and the corresponding constitute model was established through the strength tests results. Similarly, the relationships of the strength index of this soil with the cycle times of dry-wet cycle, the duration of soaking, and the water content were observed via laboratory tests. An upgraded BP neural network prediction model for strength and water content of residual and diluvia soil was constructed.(2) An assessment method for slope stability applying Adaptive Neuro-Fuzzy Inference System (ANFIS) was developed by taking advantage of its dynamic nonlinear analysis capacity to handle the complicated nonlinear relationship of the influence factors for residual and diluvia soil slope stability. The results were used to compare with that by BP neural network and indicated superiority in the prediction accuracy. The prediction capacity for slope stability of the ANFIS model constructed herein was verified by engineering practice case.(3) Morgenstern-Price approach in three dimensions (3D M-P) was developed from the most accurate two dimensional one. Base on3D M-P, Response Surface method and Support Vector Machine method were used to analyze the reliability variation of three dimensional slopes in various application ways of water, associated with the test data of engineering properties for residual and diluvia soil on the action of water. ANFIS was first applied to conduct reliability assessment for3D slope in this paper.(4) For residual and diluvia soil, test investigation of long term strength behavior was performed. The variation of shearing strength index against different values of water content over time was found. The relation of shearing strength index over time was formulated. Factor of safety of slope was computed by using improved Bishop Method based on test results, together with reliability analyses by using ANFIS.(5) Considering three dimensional characteristic of slide, a qusi-3D back analysis technique was proposed to calculate shearing strength index of slipping surface with weight-average value of slide bottom area. This technique was derived from usually used residual thrust approach, with two parameters, as the area of each2D cutaway section and the residual thrust of the last slice for this section. This technique is able to back analyze the shearing strength index c and φ at the perspective of the whole slide.(6) A calculation method for system reliability of anchored anti-sliding shaft was included in this work. The anchored anti-sliding shaft was simulated as a series connection, with such four failure modes as tension failure and pulling out of anchor, the shearing failures of both the anchored segment and loaded segment for the anti-sliding shaft. Parameters like the cohesion c, internal friction angle φ and axial compression strength of concrete were extracted to form the random design vector. Then allowing for the reliability as one constraint, an optimization design procedure with ten design variables was programmed using MATLAB, and was used in an engineering practice to give a optimization design for the residual and diluvia soil slope in K6+525-K6+610section of ShuiDong Road in Guiyang city.