Industrial Use of EsDs ETP4HPC Workshop 22 June 2017 Frankfurt DLR CFD Solver TAU & Flucs for external Aerodynamic

Transcription

1 Industrial Use of EsDs ETP4HPC Workshop 22 June 2017 Frankfurt DLR CFD Solver TAU & Flucs for external Aerodynamic Thomas Gerhold Institute of Aerodynamics and Flow Technology German Aerospace Center (DLR)

2 DLR TAU-Code t = T (Re-) Deformation Partitioning CPU-k CPU-j CPU-i Adaptation Data extraction direct (in memory) data access between modules MPI Transition prediction python interface for script based work flow Parallel data management Primary grid Solution t = T + t Preprocessor Dual grid Solver Unstructured 2 nd Order Finite Volume compressible CFD solver Used in European aircraft industry, research organization and academia Since more than 15 years > 100 daily/frequently users today Used e.g. for A380 & A350 wing design

3 Numerical Simulation of Aircraft Aerodynamics Status Cruise A380 Performance W/T-based scaling Pre-Flight prediction flight test 1% L/Dmax Source: Airbus Mach CFD-based scaling CFD & wind tunnel based performance prediction are comparable

4

5 ACARE 2020 / Flightpath ACARE: Advisory Council for Aeronautics Research in Europe Europe s Vision for Aviation Maintaining Global Leadership & Serving Society s Needs Goals (relative to typical aircraft in 2000) CO 2 emissions reduced by 75% NOx emissions reduced by 90% 65% reduction in perceived aircraft noise Consequence Heavy demands on future product performance Step changes in aircraft technology required New design principles mandatory

6 Numerical Simulation Key Enabler for Future Aircraft Design Future aircraft Design may be driven by unconventional layouts Flight characteristics may be dominated by non-linear effects High-fidelity methods indispensible for design & assessment of step changing aircraft Reliable insight to new aircraft technologies Comprehensive sensitivity analysis with risk & uncertainty management Best overall aircraft performance through integrated aerodynamics / structures / systems design Consistent and harmonized aerodynamic and aero-elastic data across flight envelope Further improvement of simulation capability necessary

7 Numerical Simulation for Aircraft Design Current Status Computational Fluid Dynamics (CFD) has significantly evolved over the last 30 years Mature tool for configurations at their design point in flight envelope Complementary to wind tunnel testing and flight tests Key tool for aeronautical research and aircraft technology development Total potential not yet fully exploited: full flight envelope, all relevant disciplines, multidisciplinary optimization

8 Numerical Simulation for Aircraft Design Vision: Digital Aircraft

9 HPC for Aircraft Design Challenge: consider algorithmic and parallel efficiency and trade off Collaboration of experts in application domain and parallel computing necessary to extract or even generate data parallelism in algorithms + algorithmic efficiency parallel efficiency Related research and prototypic Implementations done in the recent years

10 DLR CFD Code Strategy Next Generation Solver Flucs Activity started mid of 2012, supported by several prototype codes

11 Flucs what is new New code design started from scratch C++ 11 / Python Finite Volume & HO Discontinous Galerkin & zonal coupling Focus on algorithmic efficiency: use of strong implicit solvers Focus on HPC GASPI / MPI + OpenMP (2 level parallelization) Sub-partitions per node/socket are assigned to threads, threadsynchronization avoided wherever possible Implementation supports overlap of communication & computation

12 Flucs what is new Airbus accepted Flucs as a basis for their next generation CFD capability further development needed EIS planned ~2020/21

13 HPC for Aircraft Design Status TAU TAU: pure MPI Parallelization (typical grid sizes between Mpoints) Ivy Bridge 2x12 cores 89% 80% parallel efficiency 95% (vs 24 cores) 65% 73% 48% scalability limit ~6.5K points/core 31% 78% 4W multigrid vs. SG gain ~factor % 50% 68%

14 HPC for Aircraft Design Status TAU TAU: pure MPI Parallelization, effect of interconnect (RK 31 mio points) Cray Aries interconnect FDR fat tree interconnect Aries Benchmark by M. Pütz, Cray Europe

15 TAU CFD-Proxy Xeon Phi EXA2CT Christian Simmendinger F6 100 Full Model 31 mio mesh points, 3 V Multigrid 1345 mesh points / mesh points / mesh points / Strong Scaling linear comm_free mpi_bsync_se_gs mpi_early_recv_se_gs mpi_async_se_gs gaspi_bsync_gs gaspi_async_gs mpi_fence_bsync_se_gs mpi_fence_async_se_gs mpi_pscw_bsync_se_gs mpi_pscw_async_se_gs Exa2ct Slide 15-6/29/201

16 Flucs Parallel Efficiency Strong Scaling Scenario 1.9 Mio C²A²S²E Cluster % 2 mio. points Flucs Note: single grid results: no multi grid technique yet implemented (each with two 12-core IVB EP)

17 Summary / Outlook Flucs Flucs will become TAU successor for industrial applications Still important functionality to be implemented: strong implicit solvers, multigrid method, further turbulence models, etc. This is ongoing. Increased scalability demonstrated vs. legacy pure MPI parallelization Further work for improved HPC performance planned, e.g. SIMD usage EIS planned for 2020/21

18 We have interest to test Flucs on EsDs Possibly NDA for Flucs solver necessary Flucs is a python extension use of dynamic libraries needed