APPLYING VIRTUAL SOUND BARRIER AT A ROOM OPENING FOR TRANSFORMER NOISE CONTROL Jiancheng Tao, Suping Wang and Xiaojun Qiu Key laboratory of Modern Acoustics and Institute of Acoustics, Nanjing University, Nanjing 210093, China e-mail: jctao@nju.edu.cn A virtual sound barrier is applied for active control of noise from 180 MVA transformers in an enclosure. The system consists of 15 loudspeakers, 15 error microphones and a multi channel adaptive feedforward active control controller. The loudspeakers are placed with 45-75 cm intervals at a 5.4 square meters opening of the enclosure, the microphones are 1 m away in front of the loudspeakers facing noise radiation directions, and the multi channel controller adopts a specific transformer noise control algorithm with an internally synthesized reference signal. Measurement results show that the average noise reduction at the error microphones is 12.7 db for 100 Hz, 19.9 db for 200 Hz and 22.2 db for 300 Hz, and the noise reduction still remains 5.4 db for 300 Hz in the far field points 7 m away, which can match the noise reduction performance of the closed window onsite. The advantages of the proposed system are natural ventilation and lighting, effective saving in space, and easy maintenance for the transformers. 1. Introduction Transformer noise is dominated by a harmonic series based on twice the line frequency [1]. Due to the heat dissipation requirement and space limitation, it is usually not economic to apply traditional passive treatments to control the low-frequency transformer noise. Initial experiments [2 4] have proven that active noise control (ANC) technique is an effective solution for transformer noise control which satisfying the maintenance and ventilation requirements. With the development of digital processing chips, the application of multi-channel ANC systems attracts more and more attentions [5]. For transformers in free field, the spherical harmonic expansion of the primary filed and genetic algorithms were proposed to optimize the configurations of the control sources and error sensors [6], and it was validated that an active control system with only eight loudspeakers and ten error microphones is promising for a 630 kva transformer (dimension: 1.6 1.0 1.8 m). Eight near-field acoustic error sensing strategies were also compared and the result is that most of them can be regarded as minimizing the sum of the weighted mean active intensity in the normal direction [7]. Based on the on-site measurement of a 160 MVA transformer (dimension: 4.6 4.0 4.0 m), it was shown that the ANC system with 80 control sources and 96 error sensors can provide 21 db, 16 db and 16 db noise reduction at 100 Hz, 200 Hz and 300 Hz respectively [8]. ICSV22, Florence (Italy) 12-16 July 2015 1
To implement a fully-coupled ANC system with nearly 100 channels is still difficult nowadays considering the processing requirement. Decentralizing the multi-channel system is an effective way to reduce the system complexity but at the cost of lower system stability [9 12]. A more practical solution is to reduce the covering area of the ANC system. More and more transformers are now placed in rooms due to the safety and environment reasons in cities, so ANC systems can be used to constitute a virtual sound barrier (VSB) at the room opening [13], where the noise from indoor transformers radiates outwards. The active control for the noise through the building opening has already been investigated. A reflecting surface was proposed with sixteen independent ANC systems adopting Filtered-x LMS algorithm [14], and it was found that more than 10 db noise reduction was achieved at the error sensors from 200 Hz to 700 Hz. The concept of Active Acoustic Shielding (AAS) cell [15] was also proposed to define the feedforward ANC system composed by one reference microphone and one collocated control speaker. Four AAS elements were installed on a 250 mm 250 mm window with the interval of 125 mm to control noise from outside, and it is found the system is effective for different primary sources from 500 Hz to 2000 Hz. The control targets in these studies are traffic noises rather than low-frequency harmonics from transformers. There are also some other investigations on VSB systems [16 18], but these are all three-dimensional systems rather than the planar VSB at a building opening. Previous work by the authors [19] showed that a centralized 15-channel planar VSB can provide 16 db and 7.7 db noise reduction at 100 Hz and 200 Hz in the far field when it was installed on a 5 m 2 opening in the laboratory. The primary noise was a monopole source in a room and the distance between the control loudspeakers nearly equals a half of the wavelength corresponding to 200 Hz. The system drawback is that the channel scalability is poor, as the computation load of the fully coupled Filtered-x LMS algorithm increases significantly with the channel number. In this paper, a 15 channel planar VSB adopting the specific transformer algorithm [20, 21] and internally synthesized reference signal [22] is implemented at the room opening for 180 MVA transformers, and its noise control performance is investigated. The basic concept of VSB systems, the implementation of the planar VSB system and the experimental results will be presented. 2. The planar virtual sound barrier A sketch map of the planar virtual sound barrier is presented in Fig. 1, where the system is composed by a planar control source array, a collocated error microphone array near the room opening and a real-time processing controller. Figure 1. System configuration of the planar VSB. ICSV22, Florence, Italy, 12-16 July 2015 2
Assume the noise from the indoor transformers only radiates from the opening, it is hoped that the noise radiated outward can be reduced by minimizing the sound pressure at the error microphones. 3. Experimental setup A 15-channel VSB was installed in a transformer station in Shantou, China on December 17, 2014. To obtain good channel scalability, the perturbation method [20, 21] based waveform synthesis algorithm was adopted because it requires less memory and less computation load compared to the normally used FXLMS algorithm. Besides, an internally synthesized signal [22] was used as the reference signal to save the cost and avoid the inference from the outer environment and control sources. 3.1 Experiment environment A photo of the transformer station consisting two 180 MVA transformers is presented in Fig. 2. The transformers are completely covered by the acoustic shield in response to the complaints from the residents living in the apartments nearby. For natural lighting and ventilation reasons, the acoustic shield is not completely built with solid steel panel. As shown in Fig.2, there are two rows of double-layer glass windows and some blinder structures near the ground. Figure 2. A photo of the transformer station with two 180 MVA indoor transformers. 3.2 Implementation of the VSB system The front view of the acoustic shield is shown in Fig. 3 with the width of 35 m and the height of 20.3 m. The lower row of windows locates at the area with a height between 2.7 m and 5.4 m and is above the blinder structures for heat dissipation. The 15-channel planar VSB system is applied to replace the pre-existing glass window at the position marked by red solid block. The covering area of the VSB is 2 m (W) 2.7 m (H), and control sources are arranged at the opening with 5 rows and 3 columns in the vertical and horizontal directions. The vertical interval is 45.5 cm, 48.5 cm, 64.5 cm and 64.5 cm respectively from top to bottom, and the horizontal interval is 58.5 cm. Considering ICSV22, Florence, Italy, 12-16 July 2015 3
the actual size of the loudspeaker, 82% of the opening area is there used for natural ventilation and lighting. The error microphones are placed 1.0 m in front of the corresponding control sources. The controller and the accessory equipments, including the 16-channel signal conditioner for microphones, the power amplifiers for control sources, are installed in a cabinet. The detailed experimental setup of the planar VSB system can be found from Fig. 4. Figure 3. The position of VSB in the acoustic shield. (a) (b) (c) Figure 4. Experimental setup of the planar VSB system (a) the overview (b) the transducer arrays (c) the cabinet 3.3 Performance evaluation To evaluate the noise reduction performance of the VSB system, both the sound pressure level at the error microphones and in the far field are measured at positions shown in Fig. 5. Five vertical ICSV22, Florence, Italy, 12-16 July 2015 4
planes with the same area and height as the VSB are chosen as the evaluation plane, and the horizontal distances to the acoustic shield is 1.0 m (the plane of error microphones), 2.0 m, 3.0 m, 5.0 m and 7.0 m. For each evaluation plane, the sound pressure levels at twelve evenly distributed evaluation points are averaged. The insertion loss of the pre-existing window is obtained by measuring the difference of the averaged sound pressure when the window is closed or opened. The noise reduction of the VSB system can also be calculated by comparing the sound pressure level difference when the window is removed and when the VSB system is turned on or off. Figure 5. Side view of the positions of the evaluation points 4. Results and discussions The measured sound pressure level at a randomly chosen position inside the room is shown in Fig. 6, where it can be found that the noise peaks mainly occur at the harmonic frequencies of 50 Hz. The highest noise peak exists at 300 Hz and the sound pressure at the target frequencies 100 Hz, 200 Hz and 300 Hz is at least 10 db higher above the background. Figure 6. Sound pressure level at a random chosen position inside the room The insertion loss of the window at 100 Hz, 200 Hz and 300 Hz at different distances is presented in Fig. 7. When the distance is fixed, the insertion loss become larger with the increase of the ICSV22, Florence, Italy, 12-16 July 2015 5
frequency in general, indicating that the investigated frequencies lie in the mass control region of the window as a passive wall. When the distance is fixed at 3.0 m and 7.0 m, the insertion loss at 100 Hz is a little larger than that at 200 Hz and the difference is less than 0.9 db. These exceptions may be caused by the measurement error or the variation of the environmental noise. It can also be found from Fig. 7 that the insertion loss decreases with the distance for a given frequency. This means that the sound energy radiated from the investigated window is the main contributor of the sound energy only in near field region. Its contribution percentage decreases when the distance is far from the window where all other windows and blinder structures also contribute to the total sound energy. The effective distance of the window in terms of the sound contribution from the window is about 3~5 m at 100 Hz and 200 Hz as the insertion loss equals zero, and it is larger than 7.0 m at 300 Hz. Within this distance, a reduction of the sound transmission from this window will reduce the total sound pressure level correspondingly. Figure 7. Insertion loss of the double-layer glass window The noise reduction at the error microphones when the VSB system is applied are listed in Table. 1. The averaged noise reduction is 12.7 db, 19.9 db and 22.2 db at 100 Hz, 200 Hz and 300 Hz respectively. The reason for the largest noise reduction at 300 Hz is that the sound pressure level at 300 Hz is at least 20 db higher than those at 100 Hz and 200 Hz in the primary field as shown in Fig. 6. Negative noise reduction occurs at error microphones with No. 1 and No. 13 at 100 Hz because the 100 Hz component of the primary noise is already quite low there. Table 1. Noise reduction at error microphones. No. Noise reduction (db) 100 Hz 200 Hz 300 Hz 1 1 21.9 16.7 2 11 23 6.2 3 20.4 20.4 22 4 10.9 23.9 23.8 5 14.9 22.2 20.9 6 22.8 24.5 26.7 7 15.3 16.4 29.9 ICSV22, Florence, Italy, 12-16 July 2015 6
8 14.6 15.5 32.9 9 17.2 21.7 36.5 10 11.1 11.2 14.4 11 17.5 18.4 29.7 12 17.6 24.4 22.8 13 2 6.6 12.1 14 10.2 22.7 12.9 15 10.1 25 26 Averaged 12.7 19.9 22.2 A comparison between the noise reduction of the VSB system and the insertion loss of the window is presented in Fig. 8. For the 100 Hz and 200 Hz noise components, the VSB system performs better in the near field and its noise reductions decreases with distance of the evaluation plane. When the evaluation plane is 3~5 m away, both the VSB system and the window have little effect on the noise for 100 Hz and 200 Hz. The noise reduction of the VSB system is almost the same as the insertion loss of the window at 300 Hz, and the value is still over 5 db when the evaluation plane is 7 m away. 5. Conclusions Figure 8. Measured noise reduction of the VSB A 15-channel planar virtual sound barrier system was implemented in a transformer station with two 180 MVA indoor transformers, where a specific adaptive algorithm for active control of transformer noise with internally synthesized reference signal were adopted to control the noise radiation through a 5.4 m 2 opening at 100 Hz, 200 Hz and 300 Hz. Experimental results show that when the error microphone array is placed 1.0 meter away from the control source array, the averaged noise reduction is 12.7 db, 19.9 db and 22.2 db at 100 Hz, 200 Hz and 300 Hz. The noise redution performance of the VSB system in the far field is alomost the same as that of a closed double-layer glass window, especially at 300 Hz where the maxium noise peak occurs. ICSV22, Florence, Italy, 12-16 July 2015 7
ACKNOWLEDGEMENTS This research was supported under National Science Foundation of China (11474163). REFERENCES 1 Qiu, X., Zhang, L. and Tao, J. Progress in research on active control of transformer noise, Proceedings of the 41 th International Congress on Noise Control Engineering, New York, USA, 19 22 August, (2012). 2 Conover, W.B. Fighting noise with noise, Noise Control, 2, 78 82, (1956). 3 Hesselmann, N. Investigation of noise reduction on a 100 kva transformer tank by means of active methods, Applied Acoustics, 11, 27 34, (1978). 4 Ross, C. F. Experiments on the active control of transformer noise, Journal of Sound and Vibration, 61, 473 480, (1978). 5 Angevine, O. L. Active acoustic attenuation of electric transformer noise, Proceedings of the 10 th International Congress on Noise Control Engineering, Amsterdam, Netherlands, 6 8 Oct., (1981). 6 Martin, V. and Roure, A. Active noise control of acoustic sources using spherical harmonics expansion and a genetic algorithm: simulation and experiment, Journal of Sound and Vibration, 212(3), 511 523, (1998). 7 Qiu, X., Hansen, C. H. and Li, X. A comparison of near field acoustic error sensing strategies for the active control of harmonic free field sound radiation, Journal of Sound and Vibration, 215, 81 103, (1998). 8 Li, X. Physical Systems for the Active Control of Transformer Noise, PhD Thesis, Adelaide University, (2000). 9 Leboucher, E., Micheau, P., Berry, A. and ĽEspérance, A. A stability analysis of a decentralized adaptive feedback active control system of sinusoidal sound in free space, Journal of the Acoustical Society of American, 111(1), 189 99, (2002). 10 Micheau, P., Leboucher, E., and Berry, A. Technical Report, Implementation of decentralized active control of power transformer noise, Université de Sherbrooke, Canada, (2004). 11 Zhang, L., Tao, J. and Qiu, X. Performance analysis of decentralized multi-channel feedback systems for active noise control in free space, Applied Acoustics, 74(1), 181 188, (2013). 12 Cordioli, J. A., Hansen, C. H., Li, X. and Qiu, X. Numerical evaluation of a decentralised feedforward active control system for electrical transformer noise, International Journal of Acoustics and Vibration, 7(2), 100 104, (2002). 13 Tao, J., Wang, S., Qiu, X., Han, N. and Zhang, L. Virtual sound barrier for indoor transformers, Proceedings of the 43 th International Congress on Noise Control Engineering, Melbourne, Australia, 16 19 Nov., (2014). 14 Ise, S. The boundary surface control principle and its applications, IEICE Transactions on Fundamentals of Electronics, E88-A(7), 1656 1664, (2005). 15 Murao, T. and Nishimura, M. Basic Study on Active Acoustic Shielding, Journal of Environment and Engineering, 7(1), 76 91, (2012). 16 Qiu, X., Li, N. and Chen, G. Feasibility study of developing practival virtal sound barrier system, Proceedings of 12 th International Congress on Sound and Vibration, Lisbon, Portugal, 11 14 July, (2005). 17 Zou, H., Qiu, X., Lu, J. and Niu, F. A preliminary experimental study on virtual sound barrier system, Journal of Sound and Vibration, 307, 379 385, (2007). 18 Epain, N. and Friot, E. Active control of sound inside a sphere via control of the acoustic pressure at the boundary surface, Journal of Sound and Vibration, 299, 587 604, (2007). 19 Wang, S., Tao, J., and Qiu, X. Active control of transformer noise radiated outside a three-dimensional building with one opening, Proceedings of the 21 st International Congress on Sound and Vibration, Beijing, China, 13 17 July, (2014). 20 Qiu, X. and Hansen, C. H. An algorithm for active control of transformer noise with on-line cancellation path modelling based on perturbation method, Journal of Sound and Vibration, 240(4), 647 665, (2001). 21 Qiu, X., Li, X., Ai, Y. and Hansen, C. H. A waveform synthesis algorithm for active control of transformer noise: implementation, Applied Acoustics, 63(5), 467 479, (2002). 22 Zhang, L., Tao, J. and Qiu, X. Active control of transformer noise with an internally synthesized reference signal, Journal of Sound and Vibration, 331(15), 3466 3475, (2012). ICSV22, Florence, Italy, 12-16 July 2015 8