心墙堆石坝渗透稳定可靠性分析的随机响应面法

将大规模渗流有限元计算与随机响应面法相结合，对双江口心墙堆石坝进行渗透稳定可靠性分析。在基于随机响应面法的可靠度分析框架内，堆石坝稳定渗流有限元计算过程和可靠度分析过程分开独立进行，通过对心墙渗透坡降较大区域的节点建立统一的渗透稳定功能函数，采用渗流有限元分析方法和随机响应面法，计算出该区域每个节点处的渗透破坏失效概率，并将最大失效概率作为心墙的失效概率。最后，分析了心墙渗透系数、覆盖层渗透系数、上游水位与心墙具有最大失效概率节点处渗透坡降的相关关系，以及心墙渗透系数和上游水位的变异性对心墙渗透破坏失效概率的影响。计算结果表明，随机响应面法3阶Hermite展开就能够保证良好的计算精度，且计算耗时较小；双江口堆石坝心墙具有最大失效概率节点处的渗透坡降与上游水位密切相关，而与心墙本身的渗透系数呈弱负相关关系，与覆盖层渗透系数的相关性不显著；随着上游水位变异性的增大，心墙失效概率急剧增大，而这种效应对于心墙渗透系数并不明显。研究成果为随机响应面法在实际工程中的应用奠定了一定的基础。

Reliability analysis of seepage stability of core-wall rockfill dam based on stochastic response surface method

The finite element method and stochastic response surface method are combined to analyze the reliability of seepage stability in Shuangjiangkou core-wall rockfill dam project.The seepage flow analysis of the core-wall rockfill dam and the reliability analysis can be conducted separately within the framework of reliability analysis based on the stochastic response surface method.The nodes with the larger hydraulic gradient are selected to establish seepage stability function using the finite element method and stochastic response surface method.The failure probabilities of seepage failure at each node are then calculated and the maximum failure probability is taken as the failure probability of the core-wall.The relationship between the hydraulic gradient of the nodes with maximum failure probability and the hydraulic conductivity of the core-wall and the alluvial deposits,the relationship between the hydraulic gradient of the nodes with maximum failure probability and the upper water level,and the effects of the hydraulic conductivity of core and the variation of upper water level on failure probability of seepage failure of the core-wall are analyzed.The results show that third order Hermite expansion is able to ensure good precision with acceptable time consumption.The seepage at maximum failure probability node in Shuangjiangkou core-wall dam is closely related to upper water level,but has a weak negative correlation with the hydraulic conductivity of the core-wall and is less significant with the hydraulic conductivity of alluvial deposits.On the other hand,as the coefficient of variation of the upper water level increasing,the failure probability increases drastically,but this effect is less significant with regard to the hydraulic conductivity of the core-wall.The results provide a further evidence for readily application of the stochastic response surface method to practical engineering.