首页> 外文学位 >Atomistic-to-continuum coupling via a spacetime discontinuous Galerkin method.
【24h】

Atomistic-to-continuum coupling via a spacetime discontinuous Galerkin method.

机译:通过时空不连续Galerkin方法进行原子到连续体耦合。

获取原文
获取原文并翻译 | 示例

摘要

Accurate simulations across multiple spatial and temporal scales are required for design and analysis purposes in an expanding range of applications. Beyond addressing the high computational cost intrinsic to resolving multiple scales, simulations that span from continuum to atomistic scales must also resolve discrepancies between distinct physical models. This thesis presents a concurrent, atomistic-to-continuum (AtC) coupling method that is based on the spacetime discontinuous Galerkin (SDG) finite element method. The proposed AtC method couples an SDG method for continuum elastodynamics to an atomistic discontinuous Galerkin (ADG) finite element method in time for molecular dynamics. The formulation couples nonoverlapping continuum and atomistic models across a sharp interface in a mathematically consistent fashion by weakly enforcing momentum balance and kinematic compatibility across the interface. Mechanical signals from the atomistic region are fully resolved in the continuum through the use of h-adaptive causal meshing for the SDG solution.;Two- and three-field formulations are presented for the SDG continuum formulation. Both are unconditionally stable and dissipate energy. Convergence studies show the relative performance of each formulation. A special spacetime meshing procedure constructs unstructured, causal grids that support an adaptive patch-by-patch solution process with O (N) computational complexity. An error indicator based on energy dissipation drives a dynamic h-adaptive solution process that controls the total numerical dissipation. A weakened form of the ADG method balances energy to within the accuracy of the numerical integration of the interatomic forces within the atomistic domain. Especially when combined with high-order elements in time, the weakened ADG method can reduce the dissipation within the atomistic part of the model to levels consistent with the machine precision. The h-adaptive solution process in the continuum domain ensures that mechanical signals from the atomistic region are fully resolved in the continuum domain. Further, optimization of an atomic scale configuration parameter virtually eliminates spurious reflections at the coupling interface.;The SDG and ADG methods are both implicit procedures. This is not an issue in the continuum part of the model where the causal solution scheme delivers O (N) scaling. However, due to the non-local interaction between atoms, most molecular dynamics simulations use explicit integrators. An iterative solution scheme developed for the implicit ADG and coupled SDG--ADG models is shown to scale linearly with problem size and to outperform the popular Velocity Verlet integrator for this class of problems.;Atomistic-to-continuum coupling at finite temperature will require a thermomechanical continuum model. The length and time scales under consideration at an AtC coupling interface require a hyperbolic heat conduction model for the continuum. An SDG model for hyperbolic conduction based on the Maxwell-Cattaneo-Vernotte (MCV) model is developed as a first step towards a full thermomechanical model. The MCV model generates a finite signal speed, enabling the use of a causal solution process. Numerical results demonstrate the differences between the MCV and Fourier models of heat conduction.
机译:为了在不断扩大的应用范围内进行设计和分析,需要跨多个时空尺度进行准确的仿真。除了解决解决多尺度固有的高计算成本外,从连续尺度到原子尺度的模拟还必须解决不同物理模型之间的差异。本文提出了一种基于时空不连续伽勒金(SDG)有限元方法的并发原子连续体(AtC)耦合方法。提出的AtC方法将用于连续弹性动力学的SDG方法与用于分子动力学的原子间断Galerkin(ADG)有限元方法相结合。该公式通过在整个尖锐的界面上弱地加强动量平衡和整个运动学上的运动相容性,以数学上一致的方式将不重叠的连续体模型和原子模型耦合在一起。通过对SDG解决方案使用h自适应因果网格划分,可以连续地完全分解来自原子区域的机械信号。;为SDG连续体公式提供了两场和三场公式。两者都是无条件稳定并耗散能量。收敛性研究显示了每种配方的相对性能。特殊的时空网格划分程序会构建非结构化的因果网格,这些网格支持具有O(N)计算复杂度的自适应逐块解决方案过程。基于能量耗散的误差指示器驱动动态h自适应求解过程,该过程控制总的数值耗散。 ADG方法的弱化形式使能量平衡到原子域内原子间力的数值积分精度之内。特别是当与高阶元素及时组合时,弱化的ADG方法可以将模型的原子部分内的耗散降低到与机器精度一致的水平。连续域中的h自适应求解过程可确保来自原子区域的机械信号在连续域中得到完全解析。此外,优化原子尺度配置参数实际上消除了耦合界面处的杂散反射。SDG和ADG方法都是隐式过程。在因果解决方案提供O(N)缩放的模型的连续部分中,这不是问题。但是,由于原子之间的非局部相互作用,大多数分子动力学模拟都使用显式积分器。针对隐式ADG和耦合的SDG-ADG模型开发的迭代解决方案显示出与问题大小呈线性比例关系,并且对于此类问题,其性能优于流行的速度Verlet积分器。热机械连续模型。在AtC耦合接口处考虑的长度和时间尺度需要连续体的双曲线热传导模型。开发基于Maxwell-Cattaneo-Vernotte(MCV)模型的双曲线传导的SDG模型是迈向完整热力学模型的第一步。 MCV模型生成有限的信号速度,从而可以使用因果求解过程。数值结果证明了MCV和傅里叶热传导模型之间的差异。

著录项

  • 作者

    Miller, Scott T.;

  • 作者单位

    University of Illinois at Urbana-Champaign.;

  • 授予单位 University of Illinois at Urbana-Champaign.;
  • 学科 Engineering Mechanical.
  • 学位 Ph.D.
  • 年度 2009
  • 页码 158 p.
  • 总页数 158
  • 原文格式 PDF
  • 正文语种 eng
  • 中图分类
  • 关键词

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号