University of Twente Student Theses

Login

Evaluation of the use of deconvolution algorithms for determination of trailing edge noise levels

Klein, H.D.T. (2016) Evaluation of the use of deconvolution algorithms for determination of trailing edge noise levels.

[img] PDF
Restricted to Restricted until 6 January 2022.

8MB
Abstract:This report covers the evaluation of an extension of the CLEAN-SC algorithm to measure trailing edge noise. For wind turbines noise is often a limiting factor in both placement and maximum rotating speed. Hence, the reduction of noise is of great interest for the sustainable energy sector. One of the most dominating mechanisms for wind turbines is the airfoil noise originating from the trailing edge. This noise can be reduced by the use of so called serrations. Multiple experiments have been conducted to confirm the effective working of serrations. Also in 2014 an experiment on the use of serrations was conducted at the NLR. It is difficult to obtain qualitative good measurements of trailing edge noise because of the presence of spurious noise sources. These spurious noise sources often have known locations such as the noise originating from the junctions between the tested airfoil and the endplates in a wind tunnel caused by the boundary layer and the arising of horse shoe vortices. With previously used methods such as the use of microphone array and conventional beamforming it is difficult to distinguish these noise sources and measure them individually. However, in 2015 an extension of the CLEAN-SC algorithm was developed which was able to distinguish the noise sources and measure solely the noise originating from the trailing edge. This method was tested on a single case. To evaluate the algorithm, the acoustic data from the experiment in 2014 is now processed using the extended algorithm. The algorithm proved to be effective in the determination of trailing edge noise. Although the iterative deconvolution based on the empirical point spread function was not able to obtain robust physical results, the first deconvolution step by linear regression based on theoretical PSF was very effective. The new algorithm was able to accurately determine trailing edge noise above 1 kHz. In comparison with the previously used method, the trailing edge noise was lower and more accurate the frequency ranges below 4000 Hz. This was because the old method overestimated the trailing edge noise in that range due to the spurious corner sources and the limited resolution. The leading edge sources that were filtered out turned out to be independent of the use of serrations and the trailing edge noise data showed a good collapse when scaling rules are applied. This confirmed the validity of the algorithm. The algorithm was then applied on the complete set of cases to get to obtain new data on the performance of the serrations. According to the new method the best serrations, namely the flexible serrations, seemed to cause an average noise reduction of 6 dB compared to 4.6 dB according to the previously used method to measure TE noise.
Item Type:Internship Report (Master)
Clients:
NLR/DNW Flevoland, the Netherlands
Faculty:ET: Engineering Technology
Subject:52 mechanical engineering
Programme:Mechanical Engineering MSc (60439)
Link to this item:http://purl.utwente.nl/essays/71998
Export this item as:BibTeX
EndNote
HTML Citation
Reference Manager

 

Repository Staff Only: item control page