首页> 外文会议>AIAA Spacecraft Structures Conference;AIAA SciTech forum >Structural Analysis Methodology for Space Deployable Structures using a High Performance Parallel Nonlinear Finite Element Solver
【24h】

Structural Analysis Methodology for Space Deployable Structures using a High Performance Parallel Nonlinear Finite Element Solver

机译:使用高性能并行非线性有限元求解器的空间可展开结构的结构分析方法

获取原文

摘要

The Sierra Solid Mechanics computational platform is evaluated for simulating the deployment of large space structures. Sierra is a suite of nonlinear finite element, multiphysics codes developed by Sandia National Laboratories to exploit parallel computing environments. Sierra includes implicit and explicit nonlinear finite element solvers with frictional contact, geometrically nonlinear large motions, and material nonlinearity. This paper examines the functionality, performance and scalability of Sierra using a set of deployable space structures benchmark problems that embody the key phenomena in the structural concepts of interest. The focus is on comparing the implicit and explicit solvers for these benchmark problems. In the absence of contact, the implicit solver is found to be as much as two orders of magnitude faster than the explicit solver, depending on the number of processors. This is due to the low velocities inherent in these benchmarks. Under contact, the efficiency of the implicit solver is shown to depend on the contact velocities and on the stiffness of the contacting parts. For sufficiently low rates, and with proper selection of a key augmented lagrangian scale factor, the implicit solver is found to be preferred over the explicit solver during contact as well. These results motivate a proposed hybrid approach for large scale simulations that alternates longer periods of implicit solution with shorter periods of explicit solution when necessary.
机译:对Sierra固体力学计算平台进行了评估,以模拟大型空间结构的部署。 Sierra是由Sandia国家实验室开发的一组非线性有限元,多物理场代码,用于利用并行计算环境。 Sierra包括具有摩擦接触,几何非线性大运动和材料非线性的隐式和显式非线性有限元求解器。本文使用一组可部署的空间结构基准测试来检验Sierra的功能,性能和可伸缩性,这些基准测试问题体现了感兴趣的结构概念中的关键现象。重点是比较这些基准问题的隐式和显式求解器。在没有接触的情况下,根据处理器的数量,隐式求解器的速度比显式求解器快两个数量级。这是由于这些基准固有的低速。在接触下,隐式求解器的效率显示为取决于接触速度和接触零件的刚度。为足够低的速率,并且与密钥增强拉格朗日比例因子的适当选择,隐式求解器找到的联系人以及期间要优于明确的解算器。这些结果激励了提出的用于大规模仿真的混合方法,该方法在必要时将较长的隐式解决方案周期与较短的显式解决方案周期进行交替。

著录项

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