Experiences Extending the CFD Solver of the PDE Framework Peano T. Neckel, M. Lieb, R. Sangl TUM, Department of Informatics, Chair of Scientific Computing in Computer Science P. Schoeffel, F. Weyermann Gesellschaft für Anlagen- und Reaktorsicherheit (GRS)
Outline The PDE Framework Peano Thermohydraulic Simulations Goals & Approach Numerical Experiments Outlook
The PDE Framework Peano Cartesian grids (recursive adaptivity, full grid hierarchy) Low memory requirements Space-Filling curves + stack data structures high cache-hit rates (>98%) flexible insertion/deletion of data (grid changes) source: T. Weinzierl Shared/distributed mem. parallelisation Software engineering aspects CFD component Incompressible flow (FEM, IDO) Explicit + implicit time-integration schemes (FE, RK4, BE, (adaptive) TR)
The PDE Framework Peano SFC & Stacks ordering of cells along a Peano curve stacks as non-persistent data structure adaptivity & generating systems multi-level cell-oriented operator evaluation no separation of grid points and DoF (1 package/data type) high spatial and time locality of data access well suited for grid changes (insertion/deletion) 4
Thermohydraulic Simulations Goals & Approach Typical approach: 1D FVM for overall pipe setup 3D compressible simulations for special sections Idea (Peano usage): incompressible flow for special sections couple overall pipe setup and 3D simulations
Thermohydraulic Simulations Goals & Approach Incompressible Navier-Stokes Equations Discretisation low-order FEM (Q1Q0, etc.)
Thermohydraulic Simulations Goals & Approach Incompressible Navier-Stokes Equations Discretisation low-order FEM (Q1Q0, etc.) Boussinesq approximation energy conservation: temperature T Buoyancy term in momentum equation SE aspect: alternative data packages / solver
Validation of Thermal Heat Transfer 2D Rayleigh-Bernard
Validation of Thermal Heat Transfer II 2D flat plate in parallel flow Pr = 7 Re = O(1e5)
Validation of Thermal Heat Transfer II
Validation of Thermal Heat Transfer III 3D natural convection Pr = 7, Re=11063, Ra=20000
Cold Leg Scenario
Cold Leg Scenario
Cold Leg Scenario II
Outlook Thorough analysis of Cold Leg Turbulent effects (LES or turbulence model) Two-phase flow SE: automatic inclusion of features for other data packages / solvers
Thanks for your attention!
The PDE Framework Peano Regular (lexicographic) Adaptive (spacetree) http://www5.in.tum.de/peano/
Div-free Elements Driven Cavity Re=1000
Div-free Elements Flow around a Cylinder # DoF Re = 20 Re = 100 c d c l c d,max c l,max St 88,857 5.68 0.0151 3.225 0.94 0.299 ref. 5.58 0.0107 3.230 1.00 0.298
CFD Extensions joint work with Janos Benk, Bernhard Gatzhammer, Miriam Mehl, Kristof Unterweger, and Tobias Weinzierl Moving geometries Update of data + grid (regular + adaptive) Divergence correction
The PDE Framework Peano Cartesian grids (arbitrary dimensions) Plug-in concept for applications Space-filling curves, spacetrees, and stack data structures Strictly element-wise access Low memory demands Dynamical load balancing Moving geometries, dynamical adaptivity, geometric multigrid Software Engineering automatic tests, continuous integration, OO, design patterns,... CFD component Incompressible flow (FEM, IDO) Explicit + implicit time-integration schemes (FE, RK4, BE, (adaptive) TR) http://www5.in.tum.de/peano/
Backup I Low memory requirements (FEM + adap.): 2500 2000 1500 1000 bytes 2D bytes 3D 500 0 FE RK TR adap. sundance
Numerical Results FEM Q1Q0
Numerical Results - Performance 2D FEM 2 1,8 1,6 1,4 1,2 1 0,8 0,6 0,4 0,2 0 level 6 12,676 level 7 116,061 level 8 1,051,253 ratio adaptive vs. regular 3D FEM Overhead adaptive vs. regular < 3% 2D IDO Overhead Peano vs. Aoki (regular): 1.3 4.4