Stochastic Back Analysis Of Material Parameters And Reliability Analysis Of Tailings Dam Using A Bayesian Approach

With the rapid development of China’s social economy,a large number of mines are exploited on a large scale every year,and the number of tailing ponds has also increased significantly.Tailings ponds are stacked high-potential energy buildings,occupy a large area,have a huge impact on the ecological environment,and are susceptible to disasters such as debris flow caused by natural environments such as rainfall and earthquakes.In particular,the overhead warehouse poses a major safety hazard to the lives and property of downstream people.Although the factors and mechanisms that induce dam failure in tailings ponds are usually very complex,they are all closely related to the uncertainty and spatial variability of the parameters of accumulated tailings and dam construction materials.Although the current parameter inversion and tailing dam stability analysis methods are relatively mature in geotechnical engineering,the tailing dam parameter inversion and stability reliability analysis do not systematically explain the inherent uncert.ainty and spatial variability of the tailings material parameters Impact,which has led to the wrong design of the tailings pond reinforcement.To solve the above problems,the main work and conclusions of this article are as follows:(1)The effects of the prior information and the likelihood function on the stochastic back analysis of tailings material parameters are briefly described,and a stochastic back analysis method of tailings material parameters based on the analysis of random finite element and displacement monitoring data is developed.Relying on the Daheishan tailings dam project,using 2 representative displacement monitoring points for stochastic parameter back analysis,it is found that the location of the monitoring point has a greater impact on the parameter stochastic back analysis.It is basically unchanged,and the uncertainty of the material parameters closer to the monitoring point is significantly reduced.(2)A sequential stochastic back analysis method of tailings material parameters based on Bayesian theory is proposed,and the calculation process is given.Relying on the Daheishan tailings dam and Jiangxi Yongping tailings dam project,the tailings parameter infiltration is monitored using 9 water level monitoring values Sequential stochastic back analysis of coefficients.The coefficient of variation of the permeability coefficients of coarse sand,fine sand and silt of Daheishan tailings dam obtained from the stochastic back analysis of water level observations in April 2016 has been significantly reduced compared with that in March.The parameter posterior probability distribution of water level observation stochastic back analysis inferred from month to month is different.The results of the study can provide theoretical and technical reference for the integration of multi-source monitoring data to guide the optimal arrangement of actual engineering monitoring points.(3)A method for characterizing the spatial variability of tailings material parameters based on the random field theory was developed.Relying on the Daheishan tailings dam project,the spatial variability and parameters of the tailings material parameters were investigated using the Karhunen-Loeve series expansion and the non-invasive random finite element method The effect of cross-correlation on the reliability of the tailings dam.The research shows that cohesion and internal friction angle are cross-correlations that have an important effect on the reliability of the tailings dam.Ignoring the negative correlation between cohesion and internal friction angle will This leads to the reliability evaluation results of the more conservative tailings dam.