Water leak location based on improved dual-tree complex wavelet transform with soft thresholding de-noising
Introduction
Water is an irreplaceable natural resource that crucial for the survival of human being and other living organisms. Pipes are used extensively to transport water to consumers in residential areas, industrial areas, commercial areas and so on. Unfortunately, ageing pipes are prone to corrosion and leakages which cause an abundant amount of water lost in transit. The water lost known as non-revenue water (NRW) and it is present in a substantial amount which is typically 20 to 30% of water production [1]. Referring to the statistics presented by the National Water Services Commission [2], Malaysia’s NRW stood at 35.3% which is equivalent to 5.929 billion liters daily. NRW is mainly contributed by leakages [3]. And yet, the prime movers of leakages are due to aged and faulty pipes, corrosion, manufacturing errors and other external factors.
Leakages bring great menaces that can give effect to human’s lives and environment such as economic and natural resources loss [4]. They also lead to soil erosion, damage to foundations of properties and roads [5]. When a leak occurs, there is a pressure drop in the pipe system so the contaminants may intrude the pipe and causes public health issue [6]. The critical aftermath brought up by leakages is the unnecessary waste of water. This may account for water demand along with ever-increasing populations and urbanizations [7]. To this end, water leakage control has drawn attention and growing interest world widely with an aim to conserve water resource.
Under such circumstances, a great effort has been devoted to developing effective and reliable water leak detection method such as mass or volume balance method, negative pressure wave, fiber optic leak detection, acoustic method etc. Acoustic emission (AE) test is a non-destructive testing technique used for structural health monitoring. According to the ASTM E1316 Standard, AE defined as a phenomenon whereby transient waves generated due to the rapid release of energy within a material. When there are discontinuities in the material, it will emit wave which can be captured by the sensor. Then the signals are processed in a computer to perform leak detection and localization via a series of algorithmic procedures.
Detection of water leaks using acoustic method is a promising approach because it can provide a faster and real-time response [8] but it is limited to metallic pipelines only. Its performance is greatly reduced in plastic pipelines because the signals will experience larger attenuation in the plastic pipes. To date, recent works [9], [10], [11] which concerning hydrophone-based method show satisfactory leak-detection results in plastic pipes. Another demerit of AE method is due to the high occurrence of false alarm or missed detection. The signals are easily polluted by noise and this leads to inaccurately leak localization [12]. To resolve this problem, most of the pipeline operators will perform leak detection during night time because background noise from traffic and water usage is lowest.
Fig. 1 illustrates the concept of leak localization. The waves generated by leak noise will travel to both sensor 1 (S1) and sensor 2 (S2). The travelling time for both waves, t1 and t2 are different and leak can be located by means of computing the time difference of arrival (TDOA). Cross-correlation is a typical method used to compute TDOA between two signals by measuring signal similarity. There is a distinct peak at the time shift when two signals have a maximum similarity. Thereafter, the leak location can be calculated through Eq. (1) in combination with the knowledge of wave velocity.where d1 is leak location relative to the first sensor; L is the distance between two sensors; v is wave propagation velocity and t is a time delay.
From equation stated, it is clearly shown that accuracy of leak localization is highly related to two unknown values which are wave velocity, v and time delay, t. Similarly, these issues had been addressed in previous studies [13], [14], [15], [16], [17]:
- i.
Traditional localization method assumes the wave propagation velocity is constant and ignores the fact that leakage-induced wave exhibits dispersive behavior. Hence, an accurate localization method that taking account of the dispersion phenomenon should be considered.
- ii.
In actual cases, the signals are easily polluted by noise which makes leakage-induced signals imperceptible and leads to imprecise time delay computation. The conventional filtering method is inapplicable to eliminate noise from wide-band acoustic signals. Therefore, a workable noise reduction method should be established.
In order to optimize acoustic leak detection practices, two aforementioned problems have to be solved. Hence, this paper introduces an improved noise reduction method which is derived from DTCWT and soft thresholding with the aim to increase leak positioning accuracy. In the meanwhile, the location scheme is based on frequency-varying velocity by taking consider of wave dispersion phenomenon.
Section snippets
Wave propagation characteristics
Conventional correlation-based localization method omits the dispersion behavior of wave. So, the velocity has been known to be constant which can be sourced from a reference table or calculated through equation [18], [19]. This assumption is questionable because wave velocity varies with transmitting medium and pipe properties [20].
As of the propagating distance increase, the wave will experience attenuation because its amplitude and energy reduced [12]. Wave attenuation can cause the
Methods
In the view of the problem that the de-noising performance of traditional wavelet de-noising is greatly affected by DWT, so this paper puts forward an improved de-noising method by combining DTCWT and soft thresholding.
Experimental Setup
An experiment was carried out on a galvanized iron pipeline with 46 m long, 80.4 mm inner diameter and 4 mm thickness. A 5 mm leak hole is simulated by threading using a screw. Fig. 8 shows the mounting positions of 4 AE sensors. The sensors were mounted as indicated position in order to investigate the effect of U-bend on the leak localization results. In the presence of bend, cross-correlation may obtain a false result because the signals might encounter reflection at the bend [22]. The
Simulation results
In order to validate the superiorities of DTCWT over DWT, a signal with frequency components (10 Hz, 25 Hz, 50 Hz and 80 Hz) and mixed with Gauss white noise was constructed as Fig. 10 below. Then, DWT and DTCWT decomposition were performed on the signal and the results are shown in Fig. 11.
DTCWT can perfectly distribute the frequency components of the signal in different bandwidths without aliasing. Per contra, DWT creates spurious frequencies which are unrelated to the signal due to frequency
Conclusion
In summary, the DTCWT with Soft Thresholding De-noising method is presented. The signal is decomposed and band-passed via DTCWT. Then, soft thresholding method is employed to eliminate or reduce noises that are unrelated to leakage event. The test rig results proved that the method is able to accentuate the cross-correlation peak by suppressing additional peaks. Besides, the results also affirmed that the application of frequency-dependent velocity is more reliable instead of a constant speed.
CRediT authorship contribution statement
Ling Ling Ting: Conceptualization, Methodology, Software, Validation, Writing - original draft. Jing Yuen Tey: Conceptualization, Methodology, Writing - review & editing. Andy Chit Tan: Investigation, Writing - review & editing. Yeong Jin King: Supervision. Faidz Abd Rahman: Supervision, Funding acquisition.
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgments
This work was supported by the Malaysian Ministry of Higher Education (TRGS/1/2016/UTAR/01/2/3). The authors would like to acknowledge the final support of Universiti Tunku Abdul Rahman for provided facilities.
References (43)
- et al.
Experimental investigation into vibro-acoustic emission signal processing techniques to quantify leak flow rate in plastic water distribution pipes
Appl Acoust
(2017) - et al.
Analysis of leakage in high pressure pipe using acoustic emission method
Appl Acoust
(2013) - et al.
Contaminant intrusion through leaks in water distribution system: experimental analysis
Proc Eng
(2015) - et al.
Vibration characteristic analysis of buried pipes using the wave propagation approach
Appl Acoust
(2005) - et al.
Leak detection in water-filled plastic pipes through the application of tuned wavelet transforms to acoustic emission signals
Appl Acoust
(2010) - et al.
Amplitude distortion of measured leak noise signals caused by instrumentation: Effects on leak detection in water pipes using the cross-correlation method
J Sound Vib
(2019) - et al.
Acoustical characteristics of leak signals in plastic water distribution pipes
Appl Acoust
(1999) - et al.
Improving the shape of the cross-correlation function for leak detection in a plastic water distribution pipe using acoustic signals
Appl Acoust
(2017) - et al.
Integrated leakage detection and localization model for gas pipelines based on the acoustic wave method
J Loss Prev Process Ind
(2014) - et al.
Experimental study on a de-noising system for gas and oil pipelines based on an acoustic leak detection and location method
Int J Press Vessel Pip
(2017)