基于随机比例边界有限元法的高混凝土坝地震响应不确定量化分析

UNCERTAINTY QUANTIFICATION ANALYSIS IN SEISMIC ANALYSIS OF HIGH CONCRETE DAMS UPON STOCHASTIC SCALED BOUNDARY FINITE ELEMENT METHOD

  • 摘要: 高混凝土坝在地震作用下的动力响应分析一直是工程抗震领域的热点,该分析过程中存在诸多随机性。本工作从改善网格精度和提高计算效率的角度出发,提出一种新颖的基于随机比例边界有限元法(SSBFEM)的高混凝土坝抗震不确定量化分析方法。该方法采用八叉树网格技术离散高混凝土坝-地基-库水系统,并构建了随机比例边界有限元法,分析随机参数对系统动力响应的影响。提出的计算方法通过商业软件ABAQUS的用户自定义单元集成八叉树网格和随机比例边界有限元法,为基于蒙特卡罗模拟(MCs)的不确定量化分析提供高精度样本。针对蒙特卡罗模拟计算耗时的特点,提出一种模型降阶(MOR)组合径向基函数(RBF)的快速算法。通过线性组合构造的低阶子空间,快速获得MCs所需的解,有效地提升了传统MCs的计算效率。最后,通过三个不同的工程实例,验证了所提出的不确定量化方法准确性和有效性。

     

    Abstract: The dynamic response analysis of high concrete dams under earthquake actions has always been a hot spot in the field of engineering seismic resistance. The analysis process exists many randomness. This work starts from the point of view of improving mesh accuracy and of increasing computation efficiency, a novel method for uncertainty quantification (UQ) in the seismic analysis of high concrete dams based on stochastic scaled boundary finite element method (SSBFEM) is proposed. In this method, octree mesh technique is used to discretize a high concrete dam-foundation-reservoir system. The SSBFEM is constructed to analyze the influence of random parameters on the dynamic response of the system. Additionally, the method proposed can integrate octree mesh and SSBFEM on commercial software ABAQUS user defined element (UEL), and it provides high-precision samples for UQ based on Monte Carlo simulation (MCs). For the time-consuming nature of MCs, a fast algorithm based on model order reduction (MOR) combined radial basis function (RBF) is proposed. Through the linear combination of the constructed low-order subspaces, the solution required for MCs is quickly obtained, and the computational efficiency of traditional MCs is improved. Finally, the accuracy and effectiveness of the UQ method proposed are verified by three different engineering examples.

     

/

返回文章
返回