High-Order Accurate Numerical Simulations of Flow around a - PowerPoint PPT Presentation

Introduction Base Phantom Yaw LSB Conclusion High-Order Accurate Numerical Simulations of Flow around a Projectile using PyFR Jin Seok Park 1 1 Agency For Defense Development, South Korea PyFR Symposium 2020 1/39

Introduction Base Phantom Yaw LSB Conclusion Outline Introduction 1 Subsonic Flow around Projectile Base 2 Flow around Cylindrical Body at High Angle of Attack 3 Laminar Separation Bubble on Low-Reynolds Number Aerofoil 4 Conclusion 5 2/39

Introduction Base Phantom Yaw LSB Conclusion Agency for Defense Development The Center for the Development of Defense Science and Technology at South Korea 3/39

Introduction Base Phantom Yaw LSB Conclusion Agency for Defense Development Analyse aerodynamics around missile via various methods Semi-empirical methods Computational Fluid Dynamics Wind-tunnel test Flight test Wind-tunnel Facilities 4/39

Introduction Base Phantom Yaw LSB Conclusion GPU Computing GPU Supercomputer @ ADD 1280 Intel Xeon Gold CPU (@3.5Ghz) - 140 DP-TFLOPS/s 40 NVIDIA Tesla P100 GPU (@3.2Ghz) - 212 DP-TFLOPS/s 5/39

Introduction Base Phantom Yaw LSB Conclusion GPU Computing Weak scaling on ADD Supercomputer 2 Normalised Run-time 1 . 5 1 1 Billion DOFs at 94.3TFLOP/s, 55.7% Peak FLOP/s on 0 . 5 0.1M CUDA cores 0 0 5 10 15 20 25 30 35 Number of GPU 6/39

Introduction Base Phantom Yaw LSB Conclusion GPU Computing Strong scaling on ADD Supercomputer 8 6 Speed-up 4 6.26x faster, 84.9TFLOP/s, 50.1% Peak FLOP/s, 12% Loading/GPU 2 5 10 15 20 25 30 Number of GPU 7/39

Introduction Base Phantom Yaw LSB Conclusion Missile Aerodynamics Linearized and slender-body aerodynamics at low angle of attack (except transonic flow) Highly non-linear flow regime Asymmetric body vortex and interactions Strong shock interactions Body and fin-shed vortex modeling(J. B. Doyle et al., AIAA 2015-2587) Classification of Missile Aerodynamics (F. G. Moore) 8/39

Introduction Base Phantom Yaw LSB Conclusion Motivation Conventional CFD methods predicts steady-state aerodynamics well. 2nd-order accurate spatial discretization RANS turbulent models Highly non-linear flow regime High-Order accurate methods can potentially resolve detailed flow structure 9/39

Introduction Base Phantom Yaw LSB Conclusion Overview Methodology Results Subsonic Flow around Projectile Base Introduction 1 Subsonic Flow around Projectile Base 2 Overview Methodology Results Flow around Cylindrical Body at High Angle of Attack 3 Laminar Separation Bubble on Low-Reynolds Number Aerofoil 4 Conclusion 5 10/39

Introduction Base Phantom Yaw LSB Conclusion Overview Methodology Results Flow around Projectile Base Highly-Separated flow at the base 1 1 M. Van Dyke. An Album of Fluid Motion, 1982 11/39

Introduction Base Phantom Yaw LSB Conclusion Overview Methodology Results Flow around Projectile Base RANS simulation cannot resolve detailed flow strucute around the end of base. Inaccurate prediction of base drag 12/39

Introduction Base Phantom Yaw LSB Conclusion Overview Methodology Results Subsonic Flow around Projectile Base Reliable Experimental Data Base pressure and base velocity are measured 2 Re D = u ∞ d/ν = 9 . 6 × 10 4 , M = 0 . 1 Wall-resolved ILES 3:1 Elliptic nose and cylinder body l = 400 mm ,d = 70 mm 2 A. Mariotti, G. Buresti, Experimental Investigation on the Influence of Boundary Layer Thickness on the Base Pressure and Near-Wake Flow Features of an Axisymmetric Blunt-Based Body. Exp Fluids. 2013 13/39

Introduction Base Phantom Yaw LSB Conclusion Overview Methodology Results Subsonic Flow around Projectile Base Previous numerical study 3 VMS-LES @ Re D = 9 . 6 × 10 4 DNS @ Re D = 1500 Relation between base pressure and the size of recirculation 3 A. Mariotti, G. Buresti, M. V. Salvetti, Connection between Base Drag, Separating Boundary layer Characteristics and Wake Mean Recirculation Length of an Axisymmetric Blunt-Based Body, J. Fluids and Structures. 2015 14/39

Introduction Base Phantom Yaw LSB Conclusion Overview Methodology Results Methodology 5th-order accurate FR scheme, Wall-resolved ILES Long-period averaging over 150 T c (= d/u ∞ ) Quadratically curved tetrahedral meshes 0.32M tet elements 15/39

Introduction Base Phantom Yaw LSB Conclusion Overview Methodology Results Animation 16/39

Introduction Base Phantom Yaw LSB Conclusion Overview Methodology Results Pressure Distribution at Base Time-averaged Pressure coefficients at base Cp r/d< 0 . 4 = 0 . 1606 17/39

Introduction Base Phantom Yaw LSB Conclusion Overview Methodology Results Recirculation Region Time-averaged pressure contour with streamline at wake region Recirculation region l r /d = 1 . 20 18/39

Introduction Base Phantom Yaw LSB Conclusion Overview Methodology Results Flow around Cylindrical Body at High Angle of Attack Introduction 1 Subsonic Flow around Projectile Base 2 Flow around Cylindrical Body at High Angle of Attack 3 Overview Methodology Results Laminar Separation Bubble on Low-Reynolds Number Aerofoil 4 Conclusion 5 19/39

Introduction Base Phantom Yaw LSB Conclusion Overview Methodology Results Phantom Yaw Asymmetric shedding of body vortices at high angle of attack 4 Strongest in the subsonic regime Phantom Yaw 4 R. M. Cummings et al., Computational Challenges in High Angle of Attack Flow Prediction, 2013 20/39

Introduction Base Phantom Yaw LSB Conclusion Overview Methodology Results Flow around Cylindrical Body at High Angle of Attack Experimental and numerical studies to measure force 5 Re D = 200 , 000 Ogive-Cylinder body with fineness ratio 2.5 and 3.5 nose. 5 E.S. Lee et al., Experimental Reproduction and Numerical Analysis of the Side Force on an Ogive Forebody at High Angle of Attack, KSCFE, 2013 21/39

Introduction Base Phantom Yaw LSB Conclusion Overview Methodology Results Flow around Cylindrical Body at High Angle of Attack Peak of the side force occurs at 50 degree angle of attack Convective instability at 46 degree angle of attack Global instability at higher angle of attack Three stable states. 22/39

Introduction Base Phantom Yaw LSB Conclusion Overview Methodology Results Methodology Re D = u ∞ d/ν = 200 , 000 , M = 0 . 1 50 degree angle of attack, Every 60 degree of roll angle 4th-order accurate FR scheme, Wall-resolved ILES Long-period averaging over 50 T c (= d/u ∞ ) Ogive-Cylinder with fineness ratio 3.5 nose and fineness ratio 4.0 body Quadratically curved tetrahedral meshes 0.59M tet elements 23/39

Introduction Base Phantom Yaw LSB Conclusion Overview Methodology Results Methodology Re D = u ∞ d/ν = 200 , 000 , M = 0 . 1 50 degree angle of attack, Every 60 degree of roll angle 4th-order accurate FR scheme, Wall-resolved ILES Long-period averaging over 50 T c (= d/u ∞ ) Ogive-Cylinder with fineness ratio 3.5 nose and fineness ratio 4.0 body Quadratically curved tetrahedral meshes 0.59M tet elements 23/39

Introduction Base Phantom Yaw LSB Conclusion Overview Methodology Results Animation 24/39

Introduction Base Phantom Yaw LSB Conclusion Overview Methodology Results Normal and Side Force Constant normal force along the bank angle Changes in side force Three stable states : Cy = 0 , 0 . 25 , 0 . 5 25/39

Introduction Base Phantom Yaw LSB Conclusion Overview Methodology Results Sectional Distribution of Side Force Large side force occurs at ogive nose. Three stable states of sectional distribution φ = 0 , 60 , 120 ◦ φ = 240 ◦ φ = 180 , 300 ◦ 26/39

Introduction Base Phantom Yaw LSB Conclusion Overview Methodology Results Q-Criterion Time-averaged Q-Crierion φ = 0 , 60 , 120 ◦ : Slightly inclined vortex structures after mid nose φ = 240 ◦ : Close and Inclined vortex structures φ = 180 , 300 ◦ : Almost symmetric vortex Figure: φ = 60 ◦ Figure: φ = 0 ◦ Figure: φ = 240 ◦ Figure: φ = 300 ◦ 27/39

Introduction Base Phantom Yaw LSB Conclusion Overview Methodology Results Laminar Separation Bubble on Low-Reynolds Number Aerofoil Introduction 1 Subsonic Flow around Projectile Base 2 Flow around Cylindrical Body at High Angle of Attack 3 Laminar Separation Bubble on Low-Reynolds Number Aerofoil 4 Overview Methodology Results Conclusion 5 28/39

Introduction Base Phantom Yaw LSB Conclusion Overview Methodology Results Laminar Separation Bubble on Low-Reynolds Number Aerofoil Benchmark test case at Korean Society for Aeronautical and Space Sciences (KSAS) 6 KSAS EFD (Experimental Fluid Dynamics)-CFD Session Experimental data (KARI) Re=27,500 M = 0 . 48 , various angle of attacks 6 Y. J. Lee and J. S. Park, Implicit Large Eddy Simulation on Flow over Low Reynolds Number Airfoil using PyFR, KSAS 2018 29/39

Introduction Base Phantom Yaw LSB Conclusion Overview Methodology Results Laminar Separation Bubble on Low-Reynolds Number Aerofoil Non-linear aerodynamics due to Laminar separation bubble Abrupt change of transition point around 4 ∼ 5 ◦ AoA 30/39

Introduction Base Phantom Yaw LSB Conclusion Overview Methodology Results Methodology 4th-order accurate FR scheme, Wall-resolved ILES Quadratically curved hexahedral meshes 0.13M hexahedral elements 31/39