Almeida, Fabricio C. L.
Brennan, Michael J.
Joseph, Phillip F.
Paschoalini, Amarildo T.
Número total de Autores: 5
Afiliação do(s) autor(es):
 State Univ Sao Paulo UNESP, Dept Biosyst Engn, Tupa Campus, Av Rua Domingos da Costa Lopes, BR-17602496 Tupa - Brazil
 State Univ Sao Paulo UNESP, Dept Mech Engn, Ilha Solteira Campus, Av Brasil Ctr 56, BR-15385000 Ilha Solteira - Brazil
 Univ Southampton, Fac Engn & Environm, ISVR, Univ Rd, Southampton SO17 1BJ, Hants - England
 Chinese Acad Sci, Key Lab Noise & Vibrat Res, Inst Acoust, Beijing 100190 - Peoples R China
Número total de Afiliações: 4
Tipo de documento:
Journal of Sound and Vibration;
APR 28 2018.
Citações Web of Science:
In the use of acoustic correlation methods for water leak detection, sensors are placed at pipe access points either side of a suspected leak, and the peak in the cross-correlation function of the measured signals gives the time difference (delay) between the arrival times of the leak noise at the sensors. Combining this information with the speed at which the leak noise propagates along the pipe, gives an estimate for the location of the leak with respect to one of the measurement positions. It is possible for the structural dynamics of the pipe system to corrupt the time delay estimate, which results in the leak being incorrectly located. In this paper, data from test-rigs in the United Kingdom and Canada are used to demonstrate this phenomenon, and analytical models of resonators are coupled with a pipe model to replicate the experimental results. The model is then used to investigate which of the two commonly used correlation algorithms, the Basic Cross-Correlation (BCC) function or the Phase Transform (PHAT), is more robust to the undesirable structural dynamics of the pipe system. It is found that time delay estimation is highly sensitive to the frequency bandwidth over which the analysis is conducted. Moreover, it is found that the PHAT is particularly sensitive to the presence of resonances and can give an incorrect time delay estimate, whereas the BCC function is found to be much more robust, giving a consistently accurate time delay estimate for a range of dynamic conditions. (C) 2017 Elsevier Ltd. All rights reserved. (AU)