Acoustics Laboratory 1 at the Center for Noise and Vibration Control in ME, KAIST Supervisor: Prof. Jeong-Guon Ih (e-mail: J.G.Ih@kaist.ac.kr) Lab members: (as of March 2015) Ph.D. Students: 6 (1 part-time student) MS Students: 5 (1 foreign student) Graduates: (as of March 2015) Ph.D.: 26 MS: 54 Major research area Inverse techniques in vibro-acoustics: Rendering, measurement, identification Design of machinery with low noise & vibration: Wave propagation concept Perceptual design of machine source and radiated field: Auditory, haptic feelings NoViC: 6 regular staffs + 5 participating staffs from ME & AE Prof. Y Park (ANC, AVC, 3D sound, Vehicle control): Director Prof. KJ Kim (Time series, modal analysis, damping, isolation) Prof. YH Kim (Signal processing, 3D sound, holography) Prof. Emeritus: CW Lee (Rotor dynamics, dynamics), YS Park (Structural analysis)
Some samples of current research topics 2 Generation of virtual pencil-writing feeling Discrepancy of writing feeling Artificial sliding resistance Digital pen w/ vibrating stylus Glassy panel Vibration field control for Haptic sensation Actuators on boundary Control of vibration field For haptic feeling Ff () t f N (t) ICT device Dynamic characteristics of pencil in writing DY Kim JH Woo Acoustic source localization Estimation of 2D temperature distribution Localization method by double 3D intensity array [deg] Estimated azimuth angle w/ twisted double array 25 24 23 22 21 20 19 18 17 16 15 500 700 900 1100 1300 1500 Frequency [Hz] Target temp. field Measurement of TOF of sound Estimated temp. field IJ Jung TK Kim
CAV workshop, Penn State Univ.; 05/05/2015 Jeong-Guon Ih, YH Heo
Difficulties of source ID in a duct Inverse estimation of a fluid borne source: ID based on the measured acoustic pressure at up/down stream Major problems of in duct source ID 1)Insufficient spatial resolution 2)High order modes <Real source> <Observations> 4 L p (db) 100 90 80 3) Measured pressure corrupted by flow noise 4) Doppler shift due to the rotation of source <Pressure spectra> Sound pressure + Flow noise Sound pressure L p (db) 80 60 40 20 0 <Doppler shift due to source rotation> L p (db) 100 80 60 Rotating spk @ 600 RPM 70 Air blower, U f = 7 m/s 60 0 0.3 0.6 0.9 1.2 1.5 1.84 kr -20 400 450 500 550 600 Frequency (Hz) 40 20 400 450 500 550 600 Frequency (Hz)
Transverse modes in a circular duct M f (=U f /c) = Mach number of flow, U f = flow speed, c = speed of sound 5 R ω = angular frequency, k mn = Transverse wave number Transverse waves: J ( k r) e i mn m mn i ( 1) jm i = 1 for CCW i = 2 for CW <Cross modes> Axial wave number: M k k (1 M ) k k 2 2 2 f f mn zmn, 2 1 M f <Wave number domain (kr = 5.5)> 1 2 k (1 M ) k 2 2 2 f mn k (1 M ) k 2 2 2 f mn 2 1
Doppler shift due to rotation of source Doppler shift by the rotation of source in a duct (CCW direc.) Ex. Doppler shift w/ (m,n) = [(1,0), ( 1,0)] propagating modes Without source rotation W/ source rotation 6 <CCW modes> <CW modes> e e Axial direction e Ω e Ω <Sound pressure (Pa)> Circumferential waves in the same direction with fluid machine rotates & travels faster * 3 rd row, w/o rotation for comparison Ex. Doppler shift with (m,n) = [(5,0), ( 5,0)], evanescent modes CW modes CCW modes Evanescent Propagating Evanescent ω e -4Ω ω e -3Ω ω e -2Ω ω e -Ω ω e ω e +Ω ω e +2Ω ω e +3Ω ω e +4Ω <Angular frequency ω (rad)> <CCW modes> <CW modes> e 5Ω e 5Ω
Inverse estimation of source parameters 7 1) Sound propagation in stationary frame p G c c si i+ i+ m i p e T s (i 1,2) 2) Inverse identification of modal amplitude i+ i+ T H -1 H si p e s i i i m c c = G G I G p (i 1,2) 3) Change of the modal amplitudes for the rotating frame (forward wave only) <Sound pressure (Pa)> CW modes CCW modes Evanescent Propagating Evanescent ω e -4Ω ω e -3Ω ω e -2Ω ω e -Ω ω e ω e +Ω ω e +2Ω ω e +3Ω ω e +4Ω <Angular frequency ω (rad)> c ( mω ) c ( ) 1 1 smn, rmn, c ( mω ) c ( ) 2 2 smn, rmn, 4) Estimation of source parameters in the rotating frame r 1 2 1 2 1 2 1 2 T p MpTp MpTp MeTe MeT e cp cp ce c e r 0 r 1 2 1 2 1 2 1 2 T v MpLpTp MpLpTp MeLeTe MeLeT e cp cp ce c e r 0 *The measured pressures are s1 s2 used as an input: p & p m m 1 for counter-clockwise, 2 for clockwise rotation + for right-going, - for left-going wave M = cross-sectional mode matrix, T = axial transmission matrix c p,e = propagating and evanescent modal amplitudes λ = regularization parameter Subscript s for the stationary reference, and r for the rotating reference frame
Stationary source: Ex. air blow hole Use of ref. mic. technique to suppress the turbulent flow noise 8 <Measurement configuration>... Anechoic termination <Sensor numbering> 2 3 4 5... 1 16... 13 Air blower Flush mounted array Air blower Mic. #5 under strong turbulent flow 15 db difference in low freq. range Mic. #13 no flow No difference btw. L p & L p w/ ref. L p (db) 90 80 70 60 <Flow noise suppression in L p > 50 40 L p #5 w/ ref. L p #5 w/o ref. L p #13 w/ ref. L p #13 w/o ref. 100 200 300 400 500 600 700 800 Frequency (Hz)
Stationary source: Ex. air blow hole Source ID of air blower (f = 644 Hz, kr = 1.5) Accurate estimation of location and strength w/ proper regularization 9 Removal of unphysical peak in the duct center w/ regularization <Pseudo inversion only> < Tikhonov regul. + GCV > 100 98 96 94 92 98 96 94 92 90
Rotating source: Ex. 1. rotating speaker Comparison of source ID of stationary and rotating spk. Observation of Doppler shift when the source rotates Δf= rotation speed in sec 70 60 <Stationary source> 10 <Measurement system with flushmounted microphone arrays> L p (db) 50 40 30 20 10 0 200 400 600 800 1000 1200 Frequency (Hz) 90 80 Doppler shift <Rotating source> L p (db) 70 60 50 40 30 400 450 500 550 600 Frequency (Hz)
Rotating source: Ex. 1. rotating speaker Tikhonov regularization + quasi optimality cond. 11 <Stationary frame> <Rotating frame> * Frequency in Hz Rotation speed in RPM <f e = 500, Ω = 0 > <f e = 500, Ω = 300> <f e = 500, Ω = 600> <f e = 1000, Ω = 0> < f e = 1000, Ω = 300> <f e = 1000, Ω = 600>
Difference btw. two reference frames Rotating source observed in stationary frame 12 ω s = {ω -7Ω ω -6Ω ω -5Ω ω -4Ω ω -3Ω ω -2Ω ω - Ω} Main noise radiator of a rotating source cannot be identified in stationary frame ω s = {ω} Increase of frequency ω s = {ω + 7Ω ω + 6Ω ω + 5Ω ω + 4Ω ω + 3Ω ω + 2Ω ω + Ω} Rotating source observed in rotating frame Identification of main noise radiator Possibly, simulation of sound field change due to the variation of design parameters
Rotating source: Ex. 2. axial fan An axial fan in operation (5 th BPF, Ω = 2400 RPM) Use of a Laser pickup for a synchronized measurement <Pickup for the synchronization> Laser system Laser system (V) 0.5 0.4 0.3 0.2 0.1 0 Before treatment after treatment 0.0195 0.02 0.0205 0.021 Time (s) 20 db suppression of noise in broadband Tonal noise kept unchanged 100 <Flow noise suppression with the aid of rotation pickup signal> 13 <Measurement system with flush-mounted mic arrays> 80 L p (db) 60 40 20 20 db noise reduction 0 0 1 2 3 4 5 6 7 8 9 10 Order Blade of blade passing passing frequency frequency
Rotating source: Ex. 2. axial fan Identified main noise radiator: no. propagating modes = 1 Junction of tip clearance & leading edge no. evanescent modes = 26/56 <Pressure, no. modes=27> <Pressure, no. modes=57> 14 (ex. Use of load method) <Velocity, no. modes=27> <Velocity, no. modes=57>
Summary & Conclusion 15 Source identification of the rotating fluid machines in a wide duct Consideration of the effects of cross modes and flow Far field mic. or synchronizing pickup as a reference signal Inclusion of Doppler shift in the modelling and identification Reasonably accurate estimation of location and strength of the fluid machine Stationary air blower w/ flow Rotating loudspeaker w/o flow Rotating axial fan w/ flow 98 96 94 92 90