H-adaptivity applied to liquefiable soil in nonlinear analysis of soil–pile interaction

H-adaptivity applied to liquefiable soil in nonlinear analysis of soil–pile interaction
复制标题

DOI:
10.1016/j.soildyn.2004.11.014
复制
发表时间:
2005-08
影响因子:
4
通讯作者:
Xiaowei Tang;Tadanobu Sato
Xiaowei Tang;Tadanobu Sato
中科院分区:
工程技术2区
文献类型:
--
作者:
Xiaowei Tang;Tadanobu Sato

文献摘要

相似文献

本文讨论了h-自适应有限元方法在考虑大变形的可液化土桩动力分析中的应用。在地震作用下可液化土中桩的有限元分析中,特别是考虑到液化地基的大变形,桩附近区域的离散化误差变得很大。我们的目的是改进有限元方法的近似值。采用修正的拉格朗日列式和基于运动硬化准则的循环弹塑性模型来处理土体的非线性。采用混合有限元和有限差分方法,结合u-p列式和Biot两相混合理论。为了提高有限元分析的精度和效率,将后验误差估计和h型网格加密相结合的h-自适应格式应用于有限元分析。用外推法对有效应力的计算结果进行局部光顺,并在每个时间增量的最后一步计算有效应力误差的L2范数。以周期荷载作用下的桩-土相互作用体系为算例,采用基于误差估计的分裂加密方法对网格进行加密。
This paper is concerned with application of the h-adaptive finite element method to dynamic analysis of a pile in liquefiable soil considering large deformation. In finite element analysis of pile behavior in liquefiable soil during an earthquake, especially considering large deformation of liquefied ground, error due to discretization in the zone near the pile becomes very large. Our purpose was to refine the approximation of the finite element method. The updated Lagrangian formulation and a cyclic elasto-plastic model based on the kinematic hardening rule were adopted to deal with the nonlinearity of the soil. The mixed finite element and finite difference methods together with the u-p formulation and Biot's two-phase mixture theory were used. To improve the accuracy and increase the efficiency of finite element analysis, an h-adaptive scheme that included a posteriori error estimation and h-version mesh refinement was applied to the analysis. The calculated results of effective stress were smoothed locally by the extrapolation method and smoothed stress was used to calculate the L2norm of the effective stress error in the last step of the calculation of each time increment. The mesh was refined by a fission procedure based on the indication of the error estimate As a numerical example, a soil–pile interaction system loaded cyclically was analyzed by our method.