Monte-Carlo Simulation of Failure Phenomena using Particle Discretization Kenji OGUNI Atsushi WAKAI Muneo HORI Earthquake Research Institute, University of Tokyo
Motivation Deformation Failure x a PDF Solution for Homogeneous body PDF u 2 (x a ) u 2 (x a ) u 1 (x a ) u 1 (x a ) Phenomenon Behavior of Homogeneous body Phenomenon Behavior of Homogeneous body Accurate solution for homogeneous body with expensive discretization Utter Significance Almost Meaningless
Things to to Discuss on Analysis of of Failure Behavior Effect of local heterogeneity Behavior of ideally homogeneous body What really happens (extensive, expensive analysis on ideally homogeneous body?) Convergence in local sense needed? (Failure phenomena do not converge in local sense) Methods with wide variety of failure patterns depending on local heterogeneity Number of DOF (Which could be called mesh dependence ) Fine Mesh High Accuracy does not always hold Proper order of discretization depending on the scale of local heterogeneity
Objectives Numerical analysis on failure behavior of bodies with local heterogeneity See the difference between the failure behavior of ideally homogeneous body and that of locally heterogeneous bodies Examine the applicability of Particle Discretization Scheme (FEM-β) to analysis of failure behavior of bodies with local heterogeneity
Example Problem uniform tension 0.2 2.0 4.0 s Young s modulus 1.0 Poisson s ratio 0.25 Disp. B.C. 0.1 (vertical)
Ideally Homogeneous Body 0.30 0.20 0.10 0.00-0.30-0.20-0.10 0.00 0.10 0.20 Mesh: very fine, incremental crack growth Direction: based on the energy release rate for virtual extension of the crack -0.10-0.20-0.30 Kamaya and Totsuka, Corrosion Science, 2002 The only one pair of crack path is obtained
Body with Local Heterogeneity --- What we expect 0.30 0.20 Different crack paths depending on local heterogeneity? 0.10 0.00-0.30-0.20-0.10 0.00 0.10 0.20 Do they converge to homogeneous solution? -0.10-0.20-0.30
Stochastic Treatment Brute force: Monte-Carlo simulation using models with different distribution of material properties (stiffness, strength etc.) but Meshless related methods: sophisticated discretization requires relatively high computational cost Adaptive mesh: re-mesh at each step costs a lot We need a method with i) less computational cost ii) simple treatment of failure iii) no change in configuration
Easy Treatment of of Failure in in FEM-b stiffness matrix of FEM-β [k [k [k 11 21 31 ] ] ] [k [k [k 12 22 32 ] ] ] [k [k [k 13 23 33 ] ] ] Spring properties are rigorously determined with material properties; E and ν for indirect interaction for direct interaction Ω 1 x 1 direct [k12] = [k12 ] + [k indirect 12 ] x 2 Ω 2 Ω 3 x 3 Appropriately change the components of stiffness matrix/spring constants, according to a suitable failure criterion
Monte-Carlo Simulation of of Crack Propagation in in Locally Heterogeneous Body uniform tension 0.2 2.0 4.0 1000 models with different mesh geometry s Young s modulus 1.0 Poisson s ratio 0.25 Disp. B.C. 0.1 (vertical) 1000 models with different distribution of material strength
Example of of Crack Path Let s have a look at animations
Example of of Crack Path
Example of of Crack Path
Example of of Crack Path
Source of of the Difference 0.1 0.1 0.1 Y 0.0 Y 0.0 Y 0.0-0.1-0.1-0.1-0.1 0.0 0.1 X -0.1 0.0 0.1 X -0.1 0.0 0.1 X
Qualitative Comparison between Ideally Homogeneous and Locally Heterogeneous bodies Ideally Homogeneous Body 0.30 Locally Heterogeneous Bodies 0.30 0.20 0.20 0.10 0.10 0.00-0.30-0.20-0.10 0.00 0.10 0.20-0.10 0.00-0.30-0.20-0.10 0.00 0.10 0.20 0.30-0.10-0.20-0.20-0.30-0.30 Significant variance in crack paths due to local heterogeneity
Quantitative discussion We are going to to look at at PDF of of crack paths 0.4 0.2 0.6
Probability Density Function of of Crack Path Please click it.
0.6 0.5 0.0125 at x=0.0 Convergence of of PDF PDF 0.4 0.3 0.2 0.025 0.1 0 0.6 0.05-0.2-0.1 0 0.1 0.2 Crack location on each check lines 0.5 at x=0.1 0.4 0.0125 PDF 0.3 Y 0.1 0.016 0.014 0.012 0.01 0.008 0 0.006 0.004 0.002 0 0.2 0.1 0 0.6 0.025 0.05-0.2-0.1 0 0.1 0.2 Crack location on each check lines -0.1 0.5 at x=0.2 0.4 0.0125-0.2-0.1 0 0.1 0.2 X x=0.0 x=0.1 x=0.2 PDF 0.3 0.2 0.1 0.025 0.05 0-0.2-0.1 0 0.1 0.2 Crack location on each check lines
Summary Importance of local heterogeneity in analysis of failure phenomena Monte-Carlo simulation of crack propagation in heterogeneous bodies with different distribution of material strength Easy treatment of failure is needed --- FEM-β Wide variety of crack paths, PDF for crack paths
Gains and Losses of of FEM-b What we got rigorous formulation + easy treatment of failure Simple treatment of failure like DEM(Strength of Material) No change in Geometry/Configuration Particle physics type simulation suitable for parallel, massive computation Easy treatment of local heterogeneity What we sacrifice fracture mechanics, local convergence Candidate for crack path is pre-determined when a mesh is made (Crack surface=cavity) Blunt Crack Solution does not converge to the exact solution for the problem of crack growth in ideally homogeneous body