A Coupling Method for Hybrid CFD-CAA Simulations using a Dual Mesh Approach Matthias Tautz München, 21. November 2013
Content 1. Motivation Problem Description 2. Analytical Approach 3. Experiments 4. Aeroacoustic Simulation Hybrid Method CFD-Simulation Acoustic Source Distribution Results 5. Conclusion 2
Simplified VHAC Outlet Goemetry Real Outlet Geometry Simplified Model Plate in Channel * Broadband Noise Tonal Frequencies UU Plate Length ll 00 [mmmm] Tonal Frequency ff [HHHH] 20,8 3268 24,4 3084 Analytical Experimental Numerical 28,4 2997 32,2 2828 36,3 2670 40,4 2520 *S. Guerina, E. Thomy, M.C.M. Wright: Aeroacoustics of automotive vents, Journal of Sound and Vibration, 285, 2005, S. 859 875 3
Parker-Mode Theory Kochs Analytical Approach dd 0 /2 Assumptions 2-Dimensial Case Potential Flow Plate Thickness Neglectable UU dd 0 /2 ll 0 Plate σσ 0 Quantitites Plate Length ll 0 Channel Diameter dd 0 Phase Change σσ 0 Inlet Velocity UU W. Koch: Resonant Acoustic Frequencies Of Flat Plate Cascades, Journal of Sound and Vibration 88(2), 1983, S.233-242 4
Parker-Mode Diagram αα-mode ββ-mode γγ-mode δδ-mode UU = 0 Influence of UU neglectable for small Mach-Numbers 5
Parker-Mode Diagram αα-mode ββ-mode γγ-mode δδ-mode UU = 0 Influence of UU neglectable for small Mach-Numbers ββ-mode at plate in channel configuration 6
Experiments - Microphones 5 1 4 0 6 2 7 3 10 D 5 D 2 D 7
Experiments - Wall Pressure Sensors 4 0 1 3 5 2 8
Hybrid CFD-CAA Approach Fluid Dynamic Simulation Star-CCM+ Navier Stokes Equations Finite Volume Approach Coupling Coupled Reader Source Term Calculation Finite Element Approach Acoustic Simulation CFS++ (TU Wien) Lighthill-Analogy Finite Element Approach 2 cc ρρ 0 vv ii vv jj 0 dd xx ii xx jj ωω 2 ρρ tt 2 2 + cc ρρ 0 dd xx ii xx ii 2 = cc ρρ 0 vv ii vv jj 0 dd xx ii xx jj Dual Mesh Creation Energy Conserving Interpolation 9
Fluid Dynamics Simulation - Setup Property Turbulence Model Value LES-Wale Number of timesteps 10000 Inlet Profile from RANS Time step Simulation Time 1e-5 s 0.1 s y-plus (plate) 0.6 Courant Number 0.5 Channel Re-Number 42756 Hexahedra Cells 4250648 10
Time Resolution and Y+ Property Turbulence Model Value LES-Wale Number of timesteps 10000 Time step Simulation Time 1e-5 s 0.1 s y-plus (plate) 0.6 Courant Number 0.5 Channel Re-Number 42756 Hexahedra Cells 4250648 11
Karman Vortex Street Development of Karman-Vortex Karman Vortex street is known to main acoustic source region, for example B.T. Tan, M.C. Thompson, K. Hourigan: Sources of acoustic resonance generated by flowaround a long rectangular plate in a duct, Journal of Fluids and Structures, 18, 2003, S.729 740 12
Velocity Profiles Hot-wire Anemometer Velocity Profile Measurements 2 D 3 cm downstream plate 0.5 D Outlet Profile 0.5 D Profile 2 D Profile 13
Dual Mesh Creation CFD Mesh Hexahedrons 4369324 Dual Mesh Tetrahedrons 4 Wedges 4320 Hexahedrons 4304429 Prisms 1778 Polyhedrons 79400 Sum 4310535 Dual FE-Mesh Tetrahedrons 28 Wedges 10104 Hexahedrons 4304381 Sum 4314513 Cell center based dual mesh creation Splitting cells into valid FE-Cells 14
First Results Far Field Resolving Wave Length of 3000 Hz at least 20 times (5.7 mm) Created in ICEM-CFD around 1 mio cells Perfectly Matched Layer (PML) avoids Reflection upstream and downstream Ideas for changing Simulation.. 15
Possible Solutions CFD Inlet Profile from RANS Transient artificial turbulence Inlet Profile Changing Fluid Simulation Domain longer inlet channel finer mesh more realistic downstream region Or changes to CAA 16
Acoustic Sources Lighthill based Computation by Coupled Reader on dual FE-Mesh ωω 2 ρρ tt 2 2 + cc ρρ 0 dd = xx ii xx ii Computation by Star-CCM+ FV-Mesh 2 cc ρρ 0 vv ii vv jj 0 dd xx ii xx jj 2 cc ρρ 00 vv ii vv jj 0 dd xx ii xx jj cc 2 0 ωω 22 ρρ 00 vv ii vv jj dd xx ii xx jj 17
Acoustic Sources Pressure Based (incompressible) ωω 2 ρρ tt 2 2 + cc ρρ 0 dd = xx ii xx ii cc 0 2 xx ii pp ff xx ii dd cc 0 2 xx ii pp ff xx ii dd cc 0 2 ωω 22 pp ff xx ii 22 dd ωω 2 pp ff tt 2 dd 18
Acoustic Sources Splitting Sources cc 0 2 ωω 22 pp ff xx ii 22 dd = Stationary Sources obtained by temporal average Fluctuations cc 0 2 ωω 22 pp ff xx ii 22 dd cc 0 2 ωω 22 pp ff xx ii 22 dd - cc 2 0 ωω 22 pp ff xx22 dd ii + Used for CAA-Simulation 19
Acoustic Mesh Creation Source Region Resolving Flow Features Far Field Resolving Wave Length of 3000 Hz at least 20 times Far field capturing all microphones Perfectly Matched Layer (PML) avoids Reflections Mortar Interface between regions Acoustic Mesh Nodes Source Region 358980 Upstream PML 47420 Farfield 519792 Farfield PML 279079 Sum 1205271 20
Thank You For Attantion! and now Ideas, Comments, Questions? 21