| The stability of reservoir slope is one of common problems in water conservancy and hydropower engineering.In the daily operation of reservoir,the seepage field change of reservoir slope caused by the drawup and drawdown of water level and the degradation of soil parameters are the key factors affecting the stability of reservoir slope.In addition,the change of slope angle,the range of water drawup and drawdown,speeds and time steps are also the potential factors inducing landslide.Under the influence of load history,precipitation and natural factors and so on,the soil parameters of reservoir slope are uncertain and have a significant influence on reliability of reservoir slope.Therefore,under the variation of water level,the Seep/W and Slope/W of Geo-studio finite element software are respectively used to simulate the seepage field and stability of reservoir slope at different speeds and time steps in this paper,and the influence of slope angle change on stability of reservoir slope is further studied.In addition,in order to consider the uncertainty of soil parameters,the Monte Carlo Simulation method is used to carry out reliability analysis of reservoir slope,in order to improve the computational efficiency,the surrogated model of response surface method,convolutional neural network and random forest are calibrated to calculate the influence of the range of water drawup and drawdown,speeds and time steps on the failure probability of reservoir slope.Furthermore,the degradation of soil parameters is considered to study the influence of it coupled with spatial variability on failure probability of reservoir slope.Finally,an orthogonal test is designed to analyze the sensitivity of soil parameters by using range analysis method.Its main research results are as follows:(1)Under the condition of water drawdown,the stability of dam upstream slope firstly decreases and then increases with the increase of time steps,and the upstream slope stability corresponding to the most critical time step gradually decreases with the increase of speed,while the downstream slope stability increases with the increase of time step.Under the condition of water drawup,with the increase of time step,the stability of dam upstream slope increases firstly and then decreases,while the stability of downstream slope decreases gradually.When the upstream and downstream slope angles change,the slope stability decreases significantly on the side with the increase of slope angle,while the stability of the other side is mainly affected by the variation of the dam body phreatic surfaces.With the increase of water level drawdown speed,the failure probability of dam upstream slope increases gradually,while the failure probability of dam downstream slope is basically unchanged.When the water level drawdown,the failure probability of dam upstream slope is low,and it is not affected by the speeds,while the failure probability of downstream slope increases 30.11%when the range of water level rises 8 m.(2)Considering the uncertainty of soil parameters,under different drawdown speeds and time steps,the response surface method and convolutional neural network model were established respectively to predict the safety factor of bank slope.The test results of the surrogated model show that the test effect of response surface method is unstable when considering the multi-dimensional variables such as water level drawdown speed,time steps and soil random parameters,and the accuracy of it in some time steps does not meet the application requirements,while the determination coefficient R~2of convolutional neural network surrogated model is 0.99,and it can be used to analyze the transient reliability analysis of bank slope.Example analysis shows that when the water level drawdown speed increases from 0.5 m/d to 3 m/d,the minimum safety factor of bank slope decreases from 1.136 to 1.051,and the corresponding failure probability increases from 2.1%to 10.29%.With the increase of time step,the failure probability of bank slope increases firstly and then decreases under different water level drawdown speeds.The results of soil parameter sensitivity analysis show that the overall soil parameter sensitivity increases firstly and then decreases with the increase of time steps at 0.5 m/d,1.8 m/d and 1 m/d,while the soil parameter sensitivity decreases gradually at 3 m/d.(3)Considering the vertical one-dimensional random field of soil parameters,the failure probability of bank slope increases with the increase of vertical fluctuation scale,but the increasing amplitude gradually decreases,indicating that the smaller the vertical fluctuation scale is,the stronger the spatial variability of soil parameters is,and vice versa.The influence of speed on the failure probability of bank slope decreases with the gradual decrease of soil spatial variability.When considering the spatial variability coupled with degradation of soil parameters,the failure probability of bank slope increases with the decrease of spatial variability when soil parameter degraded degree is low.The influence of speed on the failure probability of bank slope is greatly affected by spatial variability.However,when degradation of soil parameters reaches a certain threshold value,the failure probability of bank slope decreases with the degradation of spatial variability.The effect of speed on the failure probability of bank slope is almost not affected by the spatial variability.Through the combination of random forest and orthogonal experiments,the sensitivity of cohesion and friction angle on the bank slope stability is high,but the sensitivity of permeability coefficient is low.(4)Based on the modeling calculation of Sangjian Reservoir,the variation of dam phreatic surfaces,the variation of upstream and downstream slope stability on dam,the variation of maximum failure probability with speed,and the variation of slope stability on upstream and downstream slope under soil parameter degradation are obtained.Compared with the above research contents,similar conclusions are obtained by calculating actual working conditions. |
PDF Full Text Request