Reliability Analysis Of Loess High Slope

The 98km long main channel of Baojixia division project goes through Baoji city,Guozhen,Caijiapo etc industrial center and densely populated areas in Shaanxi Province along the edge of the slope, and border upon the Longhai Railway,XiBao middle road,XiBao highway, the stability of the loess high slope is extremely important. During the project running more than 30 years, although the people did a lot of theoretical research and reinforcement to the stability of the loess high slope, but there were still lots of breakage to the slope, the reason for this, on the one hand the slope have complex geological conditions. On the other hand in the design, not in line with the actual conditions that using limit equilibrium method which deal the uncertainties affecting the slope stability with certain theory. Therefore, using the reliability analysis method which based on probability theory and mathematical statistics principles, to evaluate the stability of the loess high slope with failure probability, which offer more reliable foundation to the governing of the loess high slope. This paper introduced the present situation of the reliability theory and three kinds of normal using method, by investigating the loess high slope of Baojixia division project on the spot, concluding the failure form of high slope of Baojixia division project and the main factors affecting the stability of the slope. Analyzed the stability of two slope sections (a steady and a failed) by using Monte Carlo method and the estimated probability moments point method, form the functional function with Swedish circle and bishop's simplified method, when the water content are 22% and 30%. Analyzed and confirmed the sensitivity of stability of the loess high slope to the random variables by choosing different parameters as random variables. The main conclusions are as follow:1)By investigating on the spot, summed up various failure forms of the slope in Baojiaxia division work and the main factors affecting the stability of the slope, the main failure forms are: landslide, partly collapse, drops on the slope surface and crushed area at the toe of the slope; the main affecting factor are: slope section, rain, gully, breakage of the beast wall of the trench, cutting of the upper part of the slope, drainage ditch on the slope surface, afforestation on the slope surface, beast wall restoring, toe failure, planting on the slope surface, man-made hole in the slope surface and so on, and analyzed the reasons for the slope failure combined with calculations.2) The water content and slope section are the most important factors affecting the stability of the loess high slope. Combined with the climate of Baojixia irrigation area and the time when the slope section 1 collapsed (November 1984), it is deduced that the slide happened when the slope was in a higher water content state, meanwhile nail down the slope cutting and the laying of the drainage ditch on the slope surface are important method to governing of the high slope.3) When the water content, the slope section, the functional function and the choosing of the random variables are all certain, the comparison result of the calculated reliability index of Monte Carlo and the probability estimates calculated moments method shows: the difference between the results of the two methods are very small (0.032) when the random variables are choosing rationally. But if the random variables are choosing irrationally, the results of the probability estimates calculated moments method are discrete obviously, the difference with the results of the Monte Carlo method is 0.77. The probability estimates calculated moments method is of much higher efficiency than Monte Carlo method, so it is recommended to use probability estimates calculated moments method in premise of choosing rational random variables.4) The comparison result of reliability index when choosing different random variables shows: the calculated result of reliability index is very sensitive to the random variables choosing, the discrete of the calculated result of reliability index increasing when taking cφγ,cφ,cγ,c,φγ,φ,γas random variables, the difference of takingγas random variable is 6.653 times to taking cφγas random variable. The susceptibility of the astringency of the calculated result of reliability index to choosing c,φ,γas random variables is decreasing.