Flysch is an inter-layer structure with bedding features, which is alternately constituted by sandstone, mudstone, shale and other rocks. Apparent spatial variability is one of the material properties. This paper summarizes the sources of uncertainty of geotechnical engineering, evolution of slope stability analysis method, source of uncertain data and Monte Carlo algorithm, and proposes the simulation method of layered slope uncertainty. Through the analysis of the joint structure of the flysch, we put forward two kinds of simulation methods: explicit joint and implicit joint and make the comparison between the deterministic analysis and the uncertainty analysis. Based on the establishment of the random field of the layered slope, we compare the general slope reinforcement method, including the stability and deformation of gravity retaining wall, bolt, bolt retaining wall and soil nail support. The gravity retaining wall can notably restrain the deformation of the slope, but overestimate the safety of the slope after reinforcement, which is destructible. The control effect of soil nail support, short bolt and short bolt retaining wall on the deformation of the slope is pretty general, but it can provide a relatively secure slope safety, and the slope is more prone to the integral slip. The reinforcing effect of the long bolt and long bolt retaining wall is the worst, which can neither restrain deformation nor provide a reliable security guarantee. For the uncertainty in geotechnical engineering, using structural member to allow more rock or soil to enter working state can effectively reduce the potential risk.
Slope instability, earthquake, and volcanic eruption are the three major geological hazards in the world (Zhang Y X, 2008). The slope includes the soil slope and the rock slope. Rock mass is often caused by the internal development of weak interlayer intercalation or large faults, joints and cracks (Niu Yi-Wei et al., 2017). It is difficult to simulate the layer rock slope accurately. However, due to the different sedimentary conditions, stress history, weathering and other geological effects, natural soil and rock have significant spatial variability (Chen Zhao-hui et al., 2013). Two aspects reflect this spatial variability. One is physical uncertainty, and the other is cognitive uncertainty. The cognitive uncertainty also includes statistical uncertainty and model uncertainty. The physical uncertainty is inherent in the physical quantity of the rock and soil material itself, which cannot be eliminated but can be desalinated by the quality control. Model uncertainty refers to the difference caused by different constitutive models used in the numerical simulation of geotechnical engineering. Such as Mohr-Coulomb principle and Drucker-Prager principle. Statistical uncertainty should be focused on, because for a specific construction site, during construction period and operation period which are usually much less than the time when a geological change takes place, the parameters and spatial distribution of rock and soil materials in the region will not transform distinctly. In other words, if the researchers can fully investigate the situation of the field, the uncertainty of the geotechnical engineering will be almost eliminated. However, the resources and energy are finite that can be invested in the field survey, which means the short of data will restrict the accuracy of the numerical simulation. Meanwhile, the spatial distribution of the parameters also requires sufficient survey data. In summary, the study of the spatial variability of geotechnical engineering is essential to find a balance between the complexity of the survey data and the reliability of the simulation results. Conditional probability is the tool for stochastic analysis.