| Multiple 300m high core earth-rock dams are under construction or to be built in China,such as the Lianghekou Dam in the Yalong River Basin,the Shuangjiangkou Dam in the Dadu River Basin,and the Rumei Dam in the Lancang River Basin.These dams are all controlled reservoirs on the cascades of the basin where they are located,and if safety issues arise,they will have catastrophic consequences.From a historical perspective,other than overtopping,seepage failure has always been the main cause of failure and accidents in earth-rock dams.However,until now,research on the risk analysis of seepage failure in earth-rock dams is particularly limited.At the same time,there is an increasing consensus that the traditional deterministic single safety factor method for quantifying dam safety margins may not provide a conservative measure of dam safety,as it highly relies on subjective and conservative deterministic input parameters without properly considering the inherent spatial variability of soil properties.This paper takes the Nuozhadu high core earth-rock dam as a research case,closely focusing on the probability analysis and research of the impact of considering hydraulic parameters(saturated permeability coefficient ks and soil-water characteristic curve fitting parameters a and n)uncertainty on the seepage behavior and seepage stability of the dam.The research results can provide a theoretical basis for the control of seepage disasters and risk assessment of high core wall earth-rock dams in China.The main research content of this article is as follows:(1)The deterministic seepage numerical simulation of the Nuozhadu high core earth-rock dam was conducted under the saturated-only flow mode and the saturated-unsaturated flow mode,with a focus on comparing the pore water pressure distribution and flow velocity distribution of the dam core wall calculated under the two flow modes,It was verified that the traditional saturated-only flow model that neglects the flow in the unsaturated zone may seriously affect the position of the phreatic line and the flow velocity distribution,leading to misleading results.(2)Based on the Hermite stochastic polynomial expansion method and the Latin hypercube sampling collocation method,a surrogate model that can replace time-consuming finite element calculation is established.Under the framework of the random variable model,the influence of variability and correlation of hydraulic parameters on the distribution of pore water pressure in the dam core wall and the flow rate per unit width through the model is analyzed.The results indicate that the uncertainty of the distribution of pore water pressure in the core wall of the dam is relatively weakly affected by the variation of COVks values,but there is always a larger COVks,which results in a smaller mean and larger standard deviation of pore water pressure at the core wall nodes calculated statistically;The mean flow rate per unit width obtained through the probabilistic framework of Monte Carlo simulation is basically not affected by the coefficient of variation and correlation coefficient values of hydraulic parameters,while the coefficient of variation of the flow rate is only significantly affected by the variability of the saturated permeability coefficient ks.Additionally,it is proved that coupling the surrogate model within the Monte Carlo simulation framework can effectively reduce the calculation cost.(3)The hydraulic parameters(ks,a,n)are modeled as a cross-correlation lognormal distribution random field through the Cholesky matrix decomposition technology.Based on the random finite element method,the effects of the heterogeneity,autocorrelation,and crosscorrelation of hydraulic parameters on the pore water pressure distribution of the core wall of the dam and the flow rate per unit width through the model are systematically studied.The sensitivity analysis results show that the pore water pressure distribution of the dam core wall and the flow rate through the model are significantly affected by the variability of the saturated permeability coefficient ks and the autocorrelation distance.In addition,it is also found that due to the difference in the uncertainty characterization principle between the random variable model and the random field model,the probability analysis results of the dam seepage behavior under the two models are also quite different.(4)Based on the maximum velocity and the critical velocity,the calculation method of seepage stability safety factor of high core earth-rock dam is established.And the random variable model and random field model are used to systematically study the influence of the variability and correlation of hydraulic parameters on the reliability of dam seepage stability.Sensitivity analysis results indicate that under the random variable model,the SWCC fitting parameter n plays a dominant role in reliability analysis,while under the random field model,ks and autocorrelation distance are dominant.Additionally,this paper emphasizes and proves that considering the correlation between the saturated permeability coefficient field and the critical hydraulic gradient field has a significant impact on evaluating the reliability of dam seepage stability.However,this subtle but important potential characteristic has not been taken seriously in previous studies. |
