numerical and experimental investigation of the flow ... · • croaker et al. 2016 derived a...
TRANSCRIPT
Numerical and Experimental Investigation of the Flow-Induced Noise
of a Wall Mounted Airfoil
Paul Croaker, Danielle Moreau, Manuj Awasthi, Mahmoud Karimi, Con Doolan, Nicole Kessissoglou
School of Mechanical and Manufacturing Engineering UNSW Australia
• Motivation for current work
• Previous work on wall mounted airfoil flow-induced noise
– Experimental measurements
– Hybrid RANS-BEM technique
• Current experimental and numerical investigation
– Experimental and numerical flow data
– Numerical prediction of flow-induced noise
• Conclusions
Overview
• Physically interesting case study
– Trailing edge noise
– Tip vortex noise
– Turbulence – leading edge interaction noise
• Moreau et al. 2016 demonstrated that the tip vortex noise scales with M 7.5
– Low to moderate Reynolds numbers and low Mach number flows
– Expect scaling to be closer to M 6 (surface dipoles) or M 5 (sharp edge)
– Noise generating mechanism clearly complex and interesting
Wall Mounted Airfoil Study
• Croaker et al. 2016 derived a hybrid RANS-BEM technique to predict flow-induced noise produced by bodies in flow
– Successfully predicted the sound generated by the wall mounted airfoil
– Lack of experimental flow data made it difficult to fully validate the model
• Devised an experimental and numerical study into the flow around and noise generated by a finite wall mounted airfoil
– Gain a deeper understanding into the tip noise mechanism
– Generate validation data for the RANS-BEM technique
Wall Mounted Airfoil Study
• Reynolds Averaged Navier Stokes
– Computational aeroacoustic methods based on LES are time demanding
– LES techniques have a significant data storage and processing burden
– RANS based approaches are computationally efficient
RANS-BEM Motivation
LES
RANS
• Boundary Element Method
– Scattering from simple geometries using analytical Green’s function
– Extended by multiple scattering concept to include finite aerofoils
– BEM required to consider complex geometries or thick section aerofoil or hydrofoils
Hybrid RANS-BEM – Motivation
S831 Profile
• Reynolds Averaged Navier Stokes
– Only provides a statistical representation of turbulence
– Lacks knowledge of phase relationship between flow noise sources
• Boundary Element Method
– Requires the incident field on the body due to the flow noise sources
– Predicts the acoustic field scattered by the body
– Phase relationship between flow noise sources vital in accurate calculation of scattered field
Hybrid RANS-BEM – Challenges
• Reynolds Averaged Navier Stokes
– Only provides a statistical representation of turbulence
– Lacks knowledge of phase relationship between flow noise sources
• Boundary Element Method
– Requires the incident field on the body due to the flow noise sources
– Predicts the acoustic field scattered by the body
– Phase relationship between flow noise sources vital in accurate calculation of scattered field
Hybrid RANS-BEM – Challenges
• RANS-based Statistical Noise Sources
– Accounts for phase relationship between flow noise sources through two point space-time correlations
– The space-time correlations are derived from the RANS-based turbulence statistics
Phase Relationship from RANS
RANS-BEM Solution Procedure
RANS-BEM Solution Procedure
Source, b
Far-field point,
,
RANS-BEM Solution Procedure
Far-field point,
,
Source, c
RANS-BEM Solution Procedure
, , ∗ ,
, Φ , , , ,
RANS-BEM Solution Procedure
ls, ωs, us
RANS-BEM Solution Procedure
• Velocity cross-spectra approximated from RANS turbulence statistics
– Requires extensive validation and empirical constant tuning
– Experimental and high-fidelity numerical flow and noise data required to achieve this
• Scattered sound field calculated for each flow noise source separately
• Total far-field acoustic power spectral density obtained by combining individual source contributions based on velocity cross-spectra
Finite Wall Mounted Foil
Moreau, DJ, Doolan, CJ, Alexander, WN, Meyers, TW and Devenport, WJ 2016, ‘Wall-mounted finite airfoil-noise production and prediction’, AIAA Journal, vol. 54, 5, pp. 1637–1651.
Flow Structures
Flow Structures - Tip
Characteristic Scales - Tip
z/s = 0.95
z/s = 0.975
Characteristic Scales - Tip
z/s = 0.9875
z/s = 1.0
Characteristic Scales - Tip
z/s = 1.0125
z/s = 1.025
Far-Field Acoustic Power Spectral Density
Moreau, DJ, Doolan, CJ, Alexander, WN, Meyers, TW and Devenport, WJ 2016, ‘Wall-mounted finite airfoil-noise production and prediction’, AIAA Journal, vol. 54, 5, pp. 1637–1651.
Source Contribution Regions
Trailing and Leading Edge PSD
Trailing edge Leading edge
Junction and Tip PSD
Junction Tip
Conclusions of RANS-BEM work
• Hybrid RANS-BEM technique to predict flow-induced noise from wall mounted airfoil
• Velocity cross-spectra approximated from RANS turbulence statistics
• BEM used to calculate the scattered sound field
• Good comparison of results with experiment
Conclusions of RANS-BEM work
• Lack of available hydrodynamic validation data
• Need an extensive experimental and high-fidelity numerical dataset to
– Validate RANS-BEM technique
– Inform RANS-BEM model parameters
Current Numerical and Experimental Investigations
• Finite wall-mounted airfoil studied
– NACA0012 section airfoil
– Chord of 0.2m
– Span of 0.2m
– Trailing edge thickness of 3mm
• Flow velocity of 20 m/s
– Reynolds number of 274,000
– Mach number of 0.06
Experimental Investigation at UNSW
• Perform flow and noise calculations on finite wall-mounted airfoils in the UNSW Anechoic Wind Tunnel
Experimental Investigation at UNSW
Experimental Investigation at UNSW
• Single hot wire measurements taken in the near wake of the airfoil at various spanwise locations
– Mean flow and turbulence statistics
– Spectral content of fluctuating velocities
Experimental Investigation at UNSW
• Acoustic measurements are currently not possible due to excessive background noise levels
– New fan and motor are being installed
– Outlet ductwork modifications including muffler are being installed
– Facility will be ready for acoustic measurements from June 2017
High-fidelity Numerical Simulations
High-fidelity Numerical Simulations
• Structured mesh with approximately 11 million cells
– y+ ~ 1
– Spanwise and streamwise resolution less refined
• Numerical discretisation schemes
– Second order backward differencing in time
– Second order central differencing in space, with 10% blending of second order upwind differencing
• Pressure and velocity coupling achieved using the PISO algorithm
Flow Structures
Tip Vortex
Transition to Turbulence
Trailing Edge Turbulence
Horseshoe Vortex Rollup
Laminar Flow over Leading Edge
Mean flow statistics at 50% span
Velocity Autospectral Density at 50% span
y = 0 mm y = 2 mm
Mean flow statistics at 75% span
Velocity Autospectral Density at 75% span
y = 0 mm y = 2 mm
Mean flow statistics at 97.5% span
Velocity Autospectral Density at 97.5% span
y = 0 mm y = 2 mm
High-fidelity flow noise prediction
• Flow-induced noise sources extracted from LES data based on Lighthill’s acoustic analogy
• Pressure wave propagation from flow noise sources to the airfoil predicted
– Uses a near-field propagation algorithm which regularises the singular integrals
• Boundary element model used to predict the scattered and total acoustic fields
– Acoustic pressure predicted 1m perpendicular to airfoil surface at a height equal to the span
Boundary element model
Far-field pressure
Far-field pressure
Blunt trailing edge vortex shedding
Low frequency noise caused by laminar to turbulence transition
Conclusions
• Experimental and numerical investigation of flow around a wall mounted airfoil
– Experimental results exhibit a slight asymmetry which suggests airfoil is not aligned perfectly with 0 degrees
– Numerical results show reasonable results directly behind the airfoil, however a more rapid drop-off is observed transversely through the wake
• High-fidelity prediction of the flow induced noise
– Remains to be validated with experimental measurements
• Grid refinement study underway to improve grid resolution in streamwise and spanwise directions