With the increase in amount of unaccounted-for water within water distribution network all over the world, method to solve the loss of sacred resource is of interest to many countries.
Therefore, some research groups have been intensively studied in this area to reduce the amount of unaccounted-for water by means of detecting and localizing leak and burst in the form of computer simulation, laboratory experiment to continuous monitoring of water distribution system. This paper aims to review a variety of leak detection and localization techniques using pressure transient behavior of pipeline system characteristics in water distribution network as leak and burst can pose impact on transient behavior of pipeline.
Most of the WDS suffer from the loss of water due to leakage in distribution pipes. As a consequence, leak detection techniques for WDS are being familiarized into the water industry at an increasing rate. Research in the area of leak detection has abided by several avenues.Â Â They have performed numerous laboratory experiments and field experiments to clarify mystery of this field.Â Leak detection in pipe networks has been carried out using a wide range of techniques such as visual inspection, transient analysis and analysis acoustic signal. Nonetheless, these methods have problems with precision and performance in detection and are expensive or impractical to apply.
Common Leak Detection Techniques (17)
Get your grade
or your money back
using our Essay Writing Service!
Detecting leak and burst within water distribution system has been a subject of theoretical and numerical study as well as experimental and practical interest for more than a score. Different researchers analyze different characteristics of hydraulic pipeline system and employ a variety of methods. Hence, it would be beneficial to evaluate and review the strength and weak points of individual method so that one may craft create benchmark among them.
Leak and burst detection impose a wide variety of techniques from visual inspection through commercial leak detection techniques including acoustic leak detection method. Analyzing transient behavior of system to detect leak has been intensively brought as a welled-like research area.
Seeing that Acoustic Leak Detection Method is commercially adopted to verify the suspect leak and to pinpoint the location of leak by listening sounds on the pavement or soil above the water pipes, it is one of the most popular method in this field. Unfortunately, although it can accurately pinpoint the location of leak, it is only effective for metallic pipes and interferences from road traffic and other sources have great impact on the performance of it. Moreover, this method require denser network of sensors to detect, having excessive signal attenuation makes it infeasible to monitor continuously.
In contrast to this, analyzing transient characteristic of pipeline system possesses immense benefit of being able to monitor continuously which has been proved by PIPENET (Stoianov, 2007) and WaterWise project.
Leak Detection Using Pressure transient signal (2,4)
Leak can weaken transient since the wave reflected from the leak has negative effect on transient. So the size of leak can be guessed through transient's behavior. The larger the leak the weaker the transient will be.
Leak Detection in Time Domain
(Dalius Misiunas M. F., 2005) proposed a method to detect and locate medium and large size burst in real life water distribution network using CUSUM change detection test. He attempted to detect Burst-induced transient wave in pressure profile using CUSUM change detection test by continuously monitoring time pressure history which is pre-filtered using Adaptive RLS filter. CUSUM indicates changes in pressure values that are greater than pre-defined threshold. Burst location is evaluated with three objective functions
by using detected values (transient wave arrival time & wave magnitude) from CUSUM test and calibrating wave travel time & coefficients.
This method is validated on real water distribution network and proved to work well for burst size of 0.99% of pipe cross sectional area.
However, as of well-known factor, there is a tradeoff between pre-defined threshold value and performance of CUSUM test. Hence, the overall performance of this burst detection and localization method is highly dependent on the selected values of CUSUM test's parameters (drift value v & threshold h). Furthermore, CUSUM change detection test is vulnerable to flow control operations which has similar effect on transient as burst (Seshan Srirangarajan, 2010).
Always on Time
Marked to Standard
WaterWise project has been attempting to build continuous monitoring of water distribution system in Singapore. They installed sensor node for pressure, flow and water quality measurement around downtown of Singapore to monitor and detect leak and burst within the system ().
Main Assumption of their technique is system characteristics of pipeline and wave speed.
Step: Time synchronized raw pressure signals from sensors installed in real life urban WDS are analyzed and de-noised by means of wavelet de-noising of up to level 7.
Classify the resulting signal to check temporal consistency among them and extract features.
Localize burst using graph based approach which contains two parts of searching; global search among existing nodes and local search around the burst candidate node.
Since this method is not sensitive to weak signal, it is unlikely to detect small leak which impose small effect on pressure transient.
Several factors can posed localization error such as (1) inaccurate wave speed estimation
(2) system characteristics are not perfectly known
(3) time synchronization error
All in all, having a better understanding of system may yield better performance. With better model calibration, the system could perform well to detect medium and large leak.
The betterment of wavelet transform is that signal energy only resides in large-amplitude wavelet coefficient whereas noise is usually distributed uniformly.
Leak of small size would not be detected with this method and it may perform better with data from calibrated system model.
Wavelet Transform and cross-correlation method (1,29)
Goal: To detect and localize leak, to de-noise and detect leak using wavelet
Step: Exploit Wavelet transform method to de-noise acquired signal from two vibration sensors
Introduce optimal maximum likelihood based on cross-correlation technique to determine arrival time difference between two sensors which in turn is used to localize the leak.
Although wavelet transform has the ability to de-noise signal, trade-off between effective number of level of decompositions to de-noise and the amount of information scarified.
Moreover, propagation speed of leak signal, which is largely accounted on system characteristics of pipe, needs to be accurate in order to localize the leak perfectly.
Step: Filter high frequency and white noise from the signal.
Evaluate pressure transient with no leak, compare it with measured signal and take the difference to be noise.
Apply wavelet decomposition and soft threshold function to further de-noise and then pertain optimization program until satisfactory result is achieved.
Inverse Transient Method (6,13,14,19,26)
Goal: To detect leak in pipeline network
Cons: Ability to detect leak at nodal place only.
Influence by the estimation of system characteristics model
Periodic Leak Diagnosis (26) (Dalius Misiunas M. F., 2006) (Dalius Misiunas M. F., 2006) (Dalius Misiunas M. F., 2006)
Improvement on ITA and LRM with no model-related imprecision as it applies monitoring of transient response periodically and compare them to identify leak instead of building and applying numerical model of pipeline which yields many errors due to lack of information about characteristics of the pipeline
Goal: To detect and locate even small leak by applying periodic leak diagnosis as part of monitoring system installed on pipelines
Step: Install transient generating devices and pressure measurement devices on pipeline, generate and measure transient response of system
Assume initial state of pipeline to be intact and leak-free and regards transient response of it as reference.
Check subsequent response against reference to see if there is leak/leaks in pipeline
To locate, LRM is applied. If there is leak, part of generated transient will be reflected by leak.
Derive location of leak using arrival time difference of reference and measured wave.
Pros: Small leak can successfully be detected and located. The method is validated in real life water transmission pipeline with single dead-end pipe.
Cons: Highly improbable or impractical to apply this technique in large real world WDS.
Timing window for initial transient reference model
Leak Detection in Frequency Domain (10,11,12)
Inverse Resonance Technique
Goal: To detect Leak
Step:Â Â Â Â Â Â Frequency Response Diagram and Transfer Matrix Function
Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Construct FRD of the system model
Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Determine the Frequency response function of model
This Essay is
a Student's Work
This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.Examples of our work
Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Apply Inverse Resonance Technique using SCE optimization algorithm to minimize the squared difference of measured and modeled calibrated values.
Main and Cons:Â Highly influence by modeling of pipeline system which is used for FRD or transfer function (necessary to determine FRD accurately to obtain the good result.)
Pros:Â Â Â Â Â Â In contract to inverse transient method of liggett, this method can successfully detect leak at any location on pipeline not just at node and analyzing frequency component of signal makes it much faster to accomplish the optimization.
-Â Â Â Â Â Â Â Â Â Â Resonance Peak Sequencing MethodÂ (11)
Goal: To detect and localize leak
Step:Â Â Â Â Â Â Extract FRD using frequency sweeping technique(Chaudhry 1987) and construct lookup table with sequence coding
Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â In order to localize leak, coincide rank of resonance peak with lookup table entry as rank sequence changes in accordance with leak location
Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â The size of leak is inversely proportional to the magnitude of resonance peak.
Pros:Â Â Â Â Â Â Fast and efficient method to detect leak and accurate location of leak within single pipe system
Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Further improve as a combination of IRM and RPS
Cons:Â Â Â Â Â Only work well under controlled situation and require intensive research to apply for real life application.
Impulse Response Method (21)
For Rapid closure procedure, hydraulic head depends on closure method and system behaviors.
Goal 1: Separating effect of valve closure procedure is to detect leak.
Goal 2: Analysis of pressure signal during transient event to test the reliability of pipe system
Step: Analyze the pressure signal in frequency by deriving the momentum and continuity equation of hydraulic head and flow rate to obtain the transfer function of the single pipe system which in turn is used to evaluate the pressure signal variation during transient event in time domain using Impulse Response Method.
The main advantage of this frequency analysis of pressure transient is small leak and burst become differentiable.
Unfortunately, this method has problems in catching exact arrival time of transient.
Frequency Response Method (27)
Goal: To detect and localize leak
Step: Calculate the point and field matrix for system on account of analyzing steady oscillatory flow in pipes
Determine point and field matrix of leak
Employ overall transfer matrix function to detect and localize leak
Pros: Enable to detect leak in many different types of situations (single or multiple leaks on different pipe systems)
Entail the measurements to be done at a single location
Cons: Demand accurate pipe friction factor
Wave speed need to be known to localize leak correctly
Further improvement need to be done for closed-loop network
All of the leak detection methods seem to work well under certain controlled conditions. Unfortunately, none of them is capable of detecting leak in all situations.
The fidelity of each method largely depends on underlying situation.