非线性时域问题的显式有限元方法.ppt

非线性时域问题的显式有限元方法 内容提要 非线性时域问题有限元法简介 隐式与显式有限元法比较 显式有限元基本方程 中心差分法 质量矩阵对角化 稳定性和收敛性 显式有限元流程图 在汽车安全设计中的应用 非线性时域问题有限元法简介---Governing Equations Weak Form: Principle of Virtual Power Finite Element Semi-Discrete Equations Implicit Time Integration of FE Semi-Discrete Equations Implicit Time Integration: Observation Explicit Time Integration of FE Semi-Discrete Equations Explicit Time Integration: Discussion 隐式与显式有限元法比较 显式有限元基本要素 Central Difference Time Integration Lumped Mass Matrix Stability Control Central Difference Time Integration Explicit Updated Equation Lumped Mass Matrix Methods for Lumped Mass Matrix Lumped Mass Matrix: 1-D example (truss element) Lumped Mass Matrix: Example Frequency Calculation Discussion on Lumped and Consistent Mass Matrices Stability Definition of Numerical Stability A numerical procedure is stable if small perturbations of initial data result in small changes in the numerical solutions. Stability of Central Difference Method If the time step is less than the critical time step, the method is stable Element Eigenvalue Inequality Element Eigenvalue Inequality (Belytschko, 1981) Global Matrices K and M are assembled from elements The global eigenvalue and element eigenvalue are defined by: Then Element Time Step Element Time Step Nodal Time Step Some Discussions on Time Step The time step is determined by The smallest step from elements and contact definitions in the mesh Or, the smallest for all nodes Options to increase time step (Be careful!) Constant characteristic length Delete some contact nodes Adding mass to the critical node 稳定性和收敛性:  L2 Norm H1 Norm (Hr Norm) Maximum Norm Definitions Completeness: If a set of basis functions NI(x) is complete in Hr, then for any f(x) Hr, it follows that Reproducing Conditions A set of approximating functions NJ(x) reproduce p(x) if when uJ =p(xJ), then Consistency A discrete approximation Lh(u) of a patial differential equation L(u) is consistent if the er