Aeroacoustic Testing of Wind Turbine Airfoils: February 20, 2004 - February 19, 2008

Abstract

Another research effort undertaken in support of the U.S. wind turbine industry involves a series of aeroacoustic field tests conducted at the NWTC. Using well documented, consistently applied test procedures, noise spectra were measured for eight small wind turbine configurations. Test results provide valuable information to manufacturers as well as potential users of these turbines.

Determination of local flow conditions and preliminary investigation on the validity of the PNoise code for large-size wind turbine

“…However, the helicopter blade profile had a small camber and thickness when compared to the NREL/Somers "S" Series airfoils and other airfoils designed for large wind turbines. Devenport [26], on the other hand, reported experimental airfoil self-noise measurements for NACA 0012 and other airfoils of significant camber (B1-18; S831), at Reynolds numbers up to 3.2 million which, at first, seemed a suitable source from which to try and expand the PNoise validation range. This experimental dataset was recorded at the Virginia Tech (VT) Aeroacoustic Tunnel and involved four different airfoil profiles: NACA 0012, RISO B1-18, DU-96 and S831.…”

“…Among those, the NACA 0012 and the S831 were selected for an out-of-initial-scope validation study. During preliminary analysis, Devenport's (or "VT") data [26] displayed some flow overlap with the original BPM data [3] around the low Reynolds number of 620,000. However, in this low range, where the BPM TE noise model had been thoroughly validated against experimental data, SPL differences as high as 10 dB could be noticed close to the 1000 Hz frequency band.…”

“…However, in this low range, where the BPM TE noise model had been thoroughly validated against experimental data, SPL differences as high as 10 dB could be noticed close to the 1000 Hz frequency band. In spite of that significant initial difference, the authors [26] considered the results to be "in good agreement in the 1000-4000 Hz range" and attributed the differences to wind tunnel test facilities and methods to compute the airfoil self-noise from raw data.…”

“…Despite the initial concern, the VT aeroacoustic spectra [26] remain the single publicly available set for the NACA 0012 and cambered airfoils at Reynolds numbers above the original validation range and so the authors decided to set up a working plan with progressive steps: The first study concerned the NACA 0012 airfoil in AOA ranging from 0° to 4°, in untripped Reynolds number flow of 3.2 × 10 6 . The differences found between VT spectra calculated from data at 0° and PNoise calculations at 0° were significant, ranging from 6 to 12 dB excess noise prediction by the code in the 500 Hz-5 kHz range, indicating poor numerical and geometrical validation as may be seen in Fig.…”

“…A closer review of the Devenport experimental [26] geometry setup was then accomplished [8] resulting in the revision of the directivity angles and adjustment of the distance from the TE to the center of the microphone array. The geometry of the wind tunnel with the microphone array setup is depicted in detail at [26], p. 30, allowing for the calculation of the directivity angle, e , with relation to the center of the microphone array, which resulted in 103.2°, instead of 90° used as default input in the previous modified-BPM model calculation. Also, from the same drawing the distance from the TE to the center of the array was found to be 3.08 m, as opposed to 3.00 m. Unfortunately, those adjustments responded with yet another increase in the difference among the code OASPL prediction and the measurements, as shown in Table 4.…”