240 likes | 319 Views
Réduction de Modèles à l’Issue de la Théorie Cinétique. Francisco CHINESTA LMSP – ENSAM Paris Amine AMMAR Laboratoire de Rhéologie, INPG Grenoble. q 1. q 2. r 1. r 2. r N+1. q N. The different scales. R. Atomistic. Brownian dynamics. Kinetic theory: Fokker-Planck Stochastic.
E N D
Réduction de Modèles à l’Issue de la Théorie Cinétique Francisco CHINESTA LMSP – ENSAM Paris Amine AMMAR Laboratoire de Rhéologie, INPG Grenoble
q1 q2 r1 r2 rN+1 qN The different scales R Atomistic Brownian dynamics • Kinetic theory: • Fokker-Planck • Stochastic
Atomistic The 3 constitutive blocks:
q1 q2 r1 r2 rN+1 qN Brownian dynamics Beads equilibrium usually modeled from a random motion
q1 q2 r1 r2 rN+1 qN • Kinetic theory: • Fokker-Planck • Stochastic The Fokker-Planck formalism
Coming back to the macroscopic scale: Stress evaluation q F q F With F & R collinear:
Solving the deterministic Fokker-Planck equation Two new model reduction approaches
Model Reduction based on the Karhunen-Loève decomposition Continuous: Discretization: Karhunen-Loève:
Application in Computational Rheology Fokker-Planck discretisation Initial reduced approximation basis First assumption: 1 dof ! Fast simulation BUT bad results expected
Enrichment based on the use of the Krylov’s subspaces: an “a priori” strategy IF IF continue The enrichment increases the number of approximation functions BUT the Karhunen-Loève decomposition reduces it
1D 300.000FEM dof ~10dof FENE Model 3D ~10 functions (1D, 2D or 3D)
q1 q2 r1 r2 rN+1 qN • It is time for dreaming! For N springs, the model is defined in a 3N+3+1 dimensional space !! ~ 10 approximation functions are enough
BUT How defining those high-dimensional functions ? Natural answer: with a nodal description 10 nodes = 10 function values 1D
q1 q2 r1 r2 rN+1 1080 ~ presumed number ofelemental particles in the universe !! qN 1D 10 dof 10x10 dof 2D 1080 dof 80D >1000D No function can be defined in a such space from a computational point of view !! F.E.M.
Advanced deterministic approaches of Multidimensional Fokker-Planck equation Separated representation and Tensor product approximation bases q1 q2 q9 Our proposal FEM GRID Computing availability ~109
Example I - Projection:
II - Enrichment: Only 1D interpolations and 1D integrations!
q2 q1
1D/9D q1 q2 q9 809 ~ 1016 FEM dof 80x9 RM dof 2D/10D 1040 FEM dof 100.000 RM dof
Solving the Stochastic representation of the Fokker-Planck equation New efficient solvers
Stochastic approaches … A way for solving the Fokker-Planck equation: (Ottinger & Laso) W : Wiener random process We need tracking a large ensemble of particles and control the statistical noise!
BCF Stochastique: Fokker-Planck: Brownian Configuration Fields
SFS in a simple shear flow Rouge: MDF 1000 ddl / pdt a11 Bleu: BCF 100 BCF 1000 ddl / pdt Vert: Reduced BCF 100 BCF 4 ddl / pdt t The reduced approximation basis is constructed from some snapshots computed on the averaged BFC distributions
Perspectives (réduction de deuxième génération) Séparation de variables ? Base commune pour les différents « configuration fields »?