Sensitivity Analysis On Loess Slope Reliability

Three common methods of calculating slope reliability are Monte Carlo Method, Rosenblueth Method and First-Order Second-momenet Method. In the case of slope style and its parameters having been decided, the results calculated by using these three methods have some specified difference. At present, slope projects have no uniform standard for reliability accounting. Using different measures can cause some specified obsession to reasonable evaluation of slope reliability. Aiming at solving this problem, the paper take high loess slope to be the object of study, choosing simplified Bishop Method to strike up state function and making systemic analysis on the result that variational intension parameter,coefficient of variation of the parameter and slope style bring to the three methods of calculating slope reliability, the conclusions are following as:(1) In the slope whose stability coefficient is greater than 1.0, the disabled probability attained from Monte Carlo Method is greater than the ones from Rosenblueth Method and First-Order Second-momenet Method,the greatest difference can reach to 8%, yet, the disabled probability from First-Order Second-momenet Method has less difference with the one from Rosenblueth Method, when the stability coefficient comes about 1.3, the first difference is the greatest, with the stability coefficient increasing to a value which is greater than 1.5, the depressed extent of the increscent disabled probability is infinitesimal.(2) The stability coefficient is under the control of the average value of parametric which fartherly controls the disabled probability and reliability index, however, the coefficient of variation of the parameter has great effect on the disabled probability, withδc andδφincreasing, the difference between the disabled probability obtained from Monte Carlo Method and the one from First-Order Second-momenet Method is growing, thus not monotonously growing, afterδc andδφarriving to the peak value, that is betweenδc=0.2-0.35 andδφ=0.1～0.25, the difference falls on the contrary, the difference between Rosenblueth Method and First-Order Second-momenet Method is monotonously growing withδφincreasing,δc has great effect on the Monte Carlo Method, thus little effect on the Rosenblueth Method and First-Order Second-momenet Method. The average angle of internal friction and the coefficient of variationδφhave more sensitivity on the three methods than the average adhesion and the coefficient of variationδc.(3) This three methods of calculating reliability have lower sensitivity on the changement of slope height and slope ratio, the difference between these methods fluctuates less when the stability coefficient change between 1 and 1.5, as the stability coefficient increasing, the depressed extent of disabled probability received by Monte Carlo Method is smaller than by Rosenblueth Method and First-Order Second-momenet Method.(4) Through synthesizing the analysis of the differences resulting from intensity parameter,coefficient of variation and slope style by using the three methods, this paper finally brings up object disabled probability and reliability index of Rosenblueth Method and First-Order Second-momenet Method in highway slope design. Therefore, it commends to use Rosenblueth Method to calculate reliability in the project.