creator Ramachandran, Rakesh Chandran advisor Raman, Ganesh The knowledge of noise source location on a wind turbine is crucial for de- signing low noise wind turbines. Even though there has been considerable e ort to theoretically predict the noise from wind turbines in the past, very few experimental investigations have been performed to validate these models. Currently, compliance standards for wind turbine noise are based on single microphone measurements at various designated locations around a wind turbine. This only provides amplitude and frequency data of the overall wind turbine noise. However, it is impossible to locate noise sources, and to understand the dominant noise generation mechanism, using single microphone measurements. A sophisticated method, namely, the use of a microphone phased array with beamforming is necessary to locate the noise sources. Even though microphone arrays have been widely used to study aircraft yover and jet noise, it has just recently found application in locating wind turbine noise. Typical microphone arrays that have been used in the past for monitoring aircraft noise and wind turbine noise are very large ranging up to 270 m2 in area with about 148 mi- crophones. The setup is xed for a particular wind turbine and takes a considerable amount of time. In this thesis it is shown that a compact microphone array (with 24 microphones spread over 1.5 m2) is su cient to locate and separate wind turbine noise sources successfully. It is also shown that in order to use a compact micro- phone array, advanced deconvolution based beamforming methods such as DAMAS, CLEAN-SC, LP, and TIDY are necessary. However, before attempting to validate our idea, it is necessary to understand the concept of array resolution and di erent methods to improve it. We start with an in-depth study on the microphone array resolution and the e ect of di erent beamforming algorithms on it. This study has resulted in several interesting results: (i) The resolution limits that govern the resolution of the imaging system in optics also govern the resolution of microphone array. (ii) The resolution is a function of source frequency, diameter of the array, separation distance between sources, and the distance between the array and the sources. (iii) The resolution lost due to the compact microphone array can be recovered by using deconvolution algorithms. (iv) Di erent algorithms are useful for di erent scenarios; CLEAN-SC is most e cient in extracting the dominant noise source from the source map, LP is e cient in locating multiple or extended noise sources, and TIDY is e cient in locating moving sources. (v) LP and DAMAS show superresolution. (vi) The dynamic range of source maps could be improved by eliminating the diagonal elements of CSM or CCM but care should be take as this could also eliminate weak sources. Following this study, we discuss the results from the full scale large wind tur- bine noise measurements. Our study focuses on the ability of a compact microphone array to successfully locate both mechanical and aerodynamic noise sources on the wind turbine. Several interesting results have emerged from this study: (i) A compact microphone array is su cient to perform a detailed study on wind turbine noise if advanced deconvolution methods are applied. (ii) Noise sources on the blade and on the nacelle can clearly be separated. (iii) Noise of the blades is dominated by trailing edge noise which is frequency dependent and is distributed along the length of the blade with the dominant noise source closer to the tip of the blade. (iv) The LP and DAMAS algorithms represent the distributed trailing edge noise source better than CLEAN-SC and conventional beamforming. (v) Additional tonal noise produced dur- ing yawing operation is believed to be radiating from the tower of the wind turbine that acts like a resonator. (vi) Ground re ection is not believed to have a signi cant e ect on noise source location estimates in this study. (vii) The asymmetry in the aerodynamic noise is attributed to the Doppler ampli cation e ect. Finally, the noise measurements made on the small scale wind turbine showed that the compact array was successfully able to separate out the mechanical and aerodynamic noise. The mechanical noise from the nacelle was found to be the dominant noise source below 3000 Hz and the aerodynamic noise from the blades was found to be dominant above 3000 Hz. The asymmetry in the aerodynamic noise was also present in this case. The study on mechanical noise found that gear box was the dominant noise source. It is hoped that the work presented in this thesis will serve as a guide to researchers who intend to use compact microphone arrays with advanced beamforming algorithms in the future and also to those who intend to validate theoretical models of wind turbine noise. Submitted by Erma Thomas (thomase@iit.edu) on 2015-07-20T20:02:10Z No. of bitstreams: 2 Rakesh_Ramachandran_PHD.pdf: 23868353 bytes, checksum: 74329445ceec8f786a244b5bb94ee222 (MD5) First Page.pdf: 233448 bytes, checksum: c8064eb37caef71854086fa843ea3e1f (MD5) Made available in DSpace on 2015-07-20T20:02:10Z (GMT). No. of bitstreams: 2 Rakesh_Ramachandran_PHD.pdf: 23868353 bytes, checksum: 74329445ceec8f786a244b5bb94ee222 (MD5) First Page.pdf: 233448 bytes, checksum: c8064eb37caef71854086fa843ea3e1f (MD5) Previous issue date: 2014-12 Ph.D. in Mechanical and Aerospace Engineering, December 2014 2014 2014-12 http://hdl.handle.net/10560/3460 en Dissertation born digital application/pdf In Copyright http://rightsstatements.org/page/InC/1.0/ Restricted Access MMAE / Mechanical, Materials, and Aerospace Engineering Illinois Institute of Technology Affiliated department