TY - GEN
T1 - In-pipe acoustic characterization of leak signals in plastic water-filled pipes
T2 - AWWA Annual Conference and Exposition (ACE) 2010
AU - Chatzigeorgiou, Dimitris M.
AU - Kumar, Sumeet
AU - Khalifa, Atia E.
AU - Deshpande, Ajay
AU - Youcef-Toumi, Kamal
AU - Sarma, Sanjay
AU - Ben-Mansour, Rached
PY - 2010
Y1 - 2010
N2 - Acoustic emissions can be sensed to identify and localize leaks in water pipes. Leak noise correlators and listening devices have been reported in literature as successful approaches to leak detection but they have practical limitations in terms of cost, sensitivity, reliability and scalability. A possible efficient solution is the development of an in-pipe traveling leak detection system. It has been reported that in-pipe sensing is more accurate and efficient since the sensing element can be very close to the sound source. Currently inpipe approaches are limited to large leaks and larger diameter pipes. Development of such a system requires clear understanding of acoustic signals generated from leaks and their variation with different pipe loading conditions, leak sizes and surrounding media. This paper discusses the acoustic characterization of leak signals in controlled environments. A lab experimental setup was designed and built in which measurements were taken from inside 4 PVC water pipes using a hydrophone. Signals were collected for a wide range of conditions: different pressures and leak flow rates to better understand the leak signal signatures. Experiments were conducted with different pipe connecting elements like valves and junctions, and different external conditions such as pipes placed in water, air and soil. Signals were collected at various distances from the leak location (upstream and downstream) to understand the opportunities of spatial crosscorrelation techniques in localizing leaks and the effect of sensor position on the signal. Characterizations of different datasets are presented in frequency domain. The implications of acoustic characterization on the design of algorithms for leak detection are discussed and a simple algorithm to decide whether or not a leak is present near the sensor location is proposed, based on signal power calculations. The characterization gives insights, which will pave ways for the development of smart, autonomous pipetraversing systems capable of detecting and localizing leaks on a city level water distribution network.
AB - Acoustic emissions can be sensed to identify and localize leaks in water pipes. Leak noise correlators and listening devices have been reported in literature as successful approaches to leak detection but they have practical limitations in terms of cost, sensitivity, reliability and scalability. A possible efficient solution is the development of an in-pipe traveling leak detection system. It has been reported that in-pipe sensing is more accurate and efficient since the sensing element can be very close to the sound source. Currently inpipe approaches are limited to large leaks and larger diameter pipes. Development of such a system requires clear understanding of acoustic signals generated from leaks and their variation with different pipe loading conditions, leak sizes and surrounding media. This paper discusses the acoustic characterization of leak signals in controlled environments. A lab experimental setup was designed and built in which measurements were taken from inside 4 PVC water pipes using a hydrophone. Signals were collected for a wide range of conditions: different pressures and leak flow rates to better understand the leak signal signatures. Experiments were conducted with different pipe connecting elements like valves and junctions, and different external conditions such as pipes placed in water, air and soil. Signals were collected at various distances from the leak location (upstream and downstream) to understand the opportunities of spatial crosscorrelation techniques in localizing leaks and the effect of sensor position on the signal. Characterizations of different datasets are presented in frequency domain. The implications of acoustic characterization on the design of algorithms for leak detection are discussed and a simple algorithm to decide whether or not a leak is present near the sensor location is proposed, based on signal power calculations. The characterization gives insights, which will pave ways for the development of smart, autonomous pipetraversing systems capable of detecting and localizing leaks on a city level water distribution network.
UR - http://www.scopus.com/inward/record.url?scp=84871429677&partnerID=8YFLogxK
M3 - Conference contribution
AN - SCOPUS:84871429677
SN - 9781617389696
T3 - American Water Works Association Annual Conference and Exposition 2010, ACE 2010, Papers
BT - American Water Works Association Annual Conference and Exposition 2010, ACE 2010, Papers
ER -