AERODYNAMIC NOISE SIMULATION OF PROPELLER FAN BY LARGE EDDY SIMULATION

Transcription

1 日本機械学会環境工学部門 NEE(Numerical Environmental Engineering) 研究会 P.1 AERODYNAMIC NOISE SIMULATION OF PROPELLER FAN BY LARGE EDDY SIMULATION Shingo Hamada ADVANCED TECHNOLOGY R&D CENTER MITSUBISHI ELECTRIC Corp. Chisachi Kato Institute of Industrial Science, The University of Tokyo Seiji Nakashima ADVANCED TECHNOLOGY R&D CENTER MITSUBISHI ELECTRIC Corp. Yoshinobu Yamade Institute of Industrial Science, The University of Tokyo

2 Background P.2 (in door component) Demanded Silent Operation (noise plays a major part) want to increase Cooling performance Flow quantity propeller fan noise 5 th to 6 th Out door component of Air conditioner Ventilator Fan Fig.1 Our products using propeller fans rotational speed Fig.2 Noise increase

3 Problems by now "Rules of thumb" limit P.3 flow direction most suitable design parameter tip of the blade greatly protrude need further N.R. seems difficult clarify the noise source understand the mechanism of noise generation side view bell mouth (used in small pressure head) Fig.3 Half-open type propeller fan get transient flow information need

4 Past Research Examples (RANS) P.4 RANS (Reynolds Averaged Navier-Stokes Simulation) Nallasamy, M., et al. 2002, "Fan Noise Source Diagnostic Test - Computation of Rotor Wake Turbulence Noise," AIAA predicted turbulence intensity sound pressure level of the exhaust noise Gerhard, A., et al, 2001, "Numerical simulation of aeroacoustic sound generated by fans under installation conditions," AIAA simulated but No direct information of unsteady flow RANS ability to explain the phenomenon of unsteady flow is limited Further more

5 Past Research Examples () P.5 Miyazawa et al (C. Kato Laboratory,University of Tokyo) can be predicted NACA0012 showed good agreement in low freq. overestimate in high freq. Basic Airfoil Yamade et al (C. Kato Laboratory,University of Tokyo) same grid resolution turbulent boundary layer qualitative Sound prediction has not yet been achieved ducted axial-flow fan (HITACHI) streamwise vorticies

6 Objectives P.6 1. Predict aerodynamic noise by 2. Discuss the three-dimensional flow structures that primary influence aerodynamic noise generation our further goals (for a short time) predict broadband spectrum quickly and accurately test without prototyping (reducing cost) develop lower-noise new shape fans

7 Test Propeller Fan P.7 flow direction specifications of propeller fan hub blade bell mouth R15 designed flow rate:30[m3/min] density of air:1.205[kg/m3] number of revolutions:800[r/m] tip speed:16.755[m/s] tip clearance 6 Fig.4 Meridional shape

8 Calculation Method P.8 hexahedral 4D upstream part inlet boundary square mesh topology Grid consisted of three parts (overset) inlet & pressure boundary Non-Reflective Boundary Condition (prevent unphysical P fluctuations) 2D 3D fan part Code streamline-upwind finite element method Standard & dynamic smagorinsky modes 2nd order in both time and space Fractional step for pressure equation low Mach-number assumption (M=0.1, 0.2) Curle's equation code name:"front Flow/Blue" downstream part pressure boundary Fig.5 Computational grid

9 al Apperatus P.9 3D flow Propeller Fan Trigger Sensor Fully Anechoic Room Microphone 1 Microphone 2 45º 0.7 m 0.7 m Capitalize Motor for Consistency Flow Traverse Rail Hotwire Probe Chamber Room Rotating Shaft (the technology that Kuroumaru developed in 1982) Fig.6 Aerodynamic noise measurement Fig.7 wake flow measurement (pressure rise, aerodynamic noise were measured to compare the validity) (ensemble averaged velocity field)

10 Static Pressure Rise 80 overall agreement is satisfactory P.10 StaticPressureRise[Pa] % underprediction FlowRate[m 3 /min] Fig.8 Result of static pressure rise

11 Instantaneous Flow Field P.11 perspective view wake behind the blades Fig.9 Instantaneous velocity field by

12 Averaged Flow Field P.12 not clear:weak tip vortex:strong wake qualitatively predicted wake Discrepancy may have insufficient resolution of the grid overall agreement is satisfactory Fig.10 Comparison of time averaged flow field

13 Wake Flow 1. P.13 tip vortex rotation of blades rotation of blades wake 0.4 R/D 0.3 Different quantitatively caught qualitatively 13 0 [m/s] (a) Magnitude of Velocity Vector 13 0 [m/s] Cmag/Ut velocity distributions which is averaged in time and rotational rotation of blades rotation of blades R/D 0.3 good agreement qualitatively quantitatively 7-4 [m/s] (b) Radial Component 7-4 [m/s] Cr/Ut Fig.11 Comparison of velocity component

14 Wake Flow 2. P.14 rotation of blades rotation of blades R/D 0.3 good agreement qualitatively quantitatively 8-1 [m/s] (c) Tangential Component 8-1 [m/s] C u /U t rotation of blades rotation of blades R/D 0.3 Different quantitatively qualitatively 9-6 [m/s] (d) Axial Component 9-6 [m/s] Fig.11 Comparison of velocity component C m /U t

15 Tip Vortex P.15 tip vortex tip vortex control surface strong vortex instantaneous control surface weak vortex ensemble averaged probably unstable when it detaches Fig.12 Velocity field computed by To the contrary strong in the experiment

16 Aerodynamic Noise P.16 SoundPressureLevel[dB/HZ] Hz 80Hz,SSM,M=0.2,NRBC,DSM,M=0.2,NRBC 120Hz quantitatively predicted Frequency[Hz] (broadband spectrum) blade passing frequency Effects of Sub Grid Scale models SoundPressureLevel[dB/HZ] doesn't show any appreciable difference,dsm,m=0.1,nrbc,dsm,m=0.2,nrbc Frequency[Hz] (low-mach-number assumption) except for high freq. Effects of Mach-Number, M Fig.13 Comparison of sound pressure levels

17 Conclusion P calculation showed 15% smaller static pressure rise than the measurement at the designed flow rate. 2. Flow field characteristics qualitatively agreed with the measurement in the wake. 3. As for the tip vortex, the calculation didn t show it as clearly as in the measurements. 4. The tip vortex generated on the blade surface is stable, but it cannot be seen clearly when it arrived to the downstream blade in. 5. The sound frequency of NZ, 2NZ, 3NZ were quantitatively predicted. The 3NZ sound pressure level is almost corresponded to the measurement value. 6. In the effects of Mach-number M, it doesn t show any appreciable difference in the predicted sound pressure except for high frequency.

18 Thank you very much P.18