Journal of Loss Prevention in the Process Industries 41 (2016) 97e106

Contents lists available at ScienceDirect

Journal of Loss Prevention in the Process Industries

journal homepage: www.elsevier.com/locate/jlp

A review on different pipeline fault detection methods

Shantanu Datta, Shibayan Sarkar*
Department of Mechanical Engineering, ISM, Dhanbad, India

a r t i c l e i n f o a b s t r a c t

Article history: Pipeline faults like leakage and blockage always create problem for engineers. Detection of exact fault
Received 15 August 2015 quantity and its location is necessary for smooth functioning of a plant or industry and safety of the
Received in revised form environment. In this paper brief discussion is made on various pipeline fault detection methods viz.
12 March 2016
Vibration analysis, Pulse echo methodology, Acoustic techniques, Negative pressure wave based leak
Accepted 12 March 2016
detection system, Support Vector Machine (SVM) based pipeline leakage detection, Interferometric fibre
Available online 15 March 2016
sensor based leak detection, Filter Diagonalization Method (FDM), etc. In this paper merit and demerits
of all methods are discussed. It is found that these methods have been applied for specific fluids like oil,
Keywords:
Pipeline
gas and water, for different layout patterns like straight and zigzag, for various lengths of pipeline like
Fault detection methods short and long and also depending on various operating conditions. Therefore, a comparison among all
Overview methods has been done based on their applicability. Among all fault detection methods, Acoustic
Leakage and blockage reflectometry is found most suitable because of its proficiency to identify blockages and leakage in pipe
Safety as small as 1% of its diameter. Moreover this method is economical and applicable for straight, zigzag and
long, short length pipes for low, medium and high density fluid.
© 2016 Elsevier Ltd. All rights reserved.

1. Introduction
Pipes are widely used for transportation or conveying fluids
from one place to another, generally found in domestic areas, in-
dustries and for sewage. The pipeline networks are extended for
several kilometres and consist of number of pipe segments that are
connected by joints. These pipelines are subjected to many in-
cidences such as traffic and surface loads that cause the overstress
on the pipes and joints, which may lead to leak and burst in
pipeline (Mansour et al., 2012). Moreover, bad workmanship,
destructive cause, pitting corrosion and water hammer may also
lead to pipeline leaks (Sun, 2012; Lazhar et al., 2013). Incidence of a
leak in pipeline causes a sudden decrease in the pressure (Silva
et al., 1996) that would cause inefficiency in terms of delivering
time and the volume flow rate of fluid being transferred. Some-
times, impurities and clogging also disturb the usual flow in piping
system. Blockages begin in the form of a small growth in the wall
roughness that increases with time due to physical or chemical
processes and can finally encroache a significant part in the internal
cross sectional area of pipe (Duan et al., 2014). Generally, fluid flow
rate through pipeline may get reduced due to presence of various
* Corresponding author.
E-mail address: shibayan.sarkar@gmail.com (S. Sarkar).
pipe faults i.e. blockage, leakage etc. Due to insufficient flow of
water or other fluid causes the failure of pipe network. If the
pipeline faults are not quickly detected and repaired, that may
result to product loss and other serious damages (Sandberg et al.,
1989).
In this paper some selective processes of pipe line fault detec-
tion have been discussed and compared corresponding to their
applicability. At present, the general methods for pipeline leakage
detection range from manual inspection by trained linesmen to
advanced satellite imaging. Usually, two types of methods can be
noticed for the detection of pipeline failure such as hardware-based
and software-based. In hardware-based methods, special sensing
devices are used to detect pipe leakage. Hardware-based methods
can be further classified depending on the type of sensors and
equipments used for fault detection such as acoustic monitoring,
optical fibre, cable sensor, vibration analysis etc. Similarly, the
methods based on softwares depend on various software programs
(Jin et al., 2014). Different soft computing techniques have also been
effectively used for fault detection in pipe networks (Laurentys
et al., 2011, Zadeh, 1965, Zidani et al., 2008 & Poulakis et al.,
2003). Moreover, in some methods, hardware and software both
are used simultaneously. A detailed classification of fault detection
methods for blockage and leakage is given in Fig 1.
The aim of the paper is to focus towards the process and mea-
surement principles that are used for the detection of pipe line

http://dx.doi.org/10.1016/j.jlp.2016.03.010
0950-4230/© 2016 Elsevier Ltd. All rights reserved.
98 S. Datta, S. Sarkar / Journal of Loss Prevention in the Process Industries 41 (2016) 97e106

Fig. 1. Flowchart of different fault detection methods.

faults. The initial section of the paper will provide an overview of
the various methods that are used to detect blockage and leakage in
pipeline and finally, a comparative discussion will be provided
based on their applicability.
2. Determination methods for fault detection
Some of the important methods applied for accurate detection
of pipeline blockage are discussed below:
2.1. Blockage detection techniques
2.1.1. Vibration analysis
Lile et al. (Lile et al., 2012) used vibration analysis method to
describe blockage effects in circular pipe where Fast Fourier
Transform (FFT) graph has been presented to describe the corre-
lation of blockage levels to vibration signal. In case of fluid flows
through an obstacle, the flow cross section area reduces and as per
continuity equation fluid velocity increases. As per Bernoulli's
principle, pressure of the fluid then decreases. Due to this fluctu-
ating pressure and high velocity, prominent vibration response is
observed in the pipeline (Kim et al., 2007). The vibration
 S. Datta, S. Sarkar / Journal of Loss Prevention in the Process Industries 41 (2016) 97e106 99

Fig. 2. Accelerometers mounted on pipe surface (Lile et al., 2012).

parameters are then measured using accelerometer as shown in Fig

2. Relationship between various frequency responses correspond- 2s cot kLb
j¼ (2)
ing to different accelerometer locations (point A, B and C) with s2 þ 1
various blockage thicknesses inside the pipe are depicted in Fig 3. It The general solution for area ratio (s) can be represented as
is observed that frequencies at point ‘B’ are maximum as compared
pffiffiffiffiffiffiffiffiffiffiffiffi
to point ‘A ‘and ‘C’, due to the presence of blockage at ‘B’. Variation ∧ð1 þ j2 Þ þ 1±2 ∧ð1þj2 Þ
of area and pressure changes the vibration pattern. After comparing s2 ¼  (3)
1  ∧ð1 þ j2
the frequencies at various points of pipe, it is observed that at the
place of maximum blockage thickness, velocity is maximum and Finally governing equation for blockage length can be expressed
frequency varies from 7 to 9 kHz. as
"  #
2.1.2. Pulse echo methodology 1 s2  1 j np
Lb ¼ cot1 þ ; n ¼ 0; 1; 2; ::: (4)
Pulse echo methodology is a fast and efficient way of deter- k 2s k
mining the presence of a blockage inside a pipeline (Duan et al.,
2015). This method is offered for obtaining both the length and Depending on the frequency (u ¼ kc), infinite number of solu-
the equivalent cross-sectional area of a blockage where a single tions can be obtained for Lb. Finally, relation between area ratio and
microphone is used to capture the incident and reflected pulse. The length of the blockage can be obtained through equation (Bhuiyan
power reflection ratio (i.e ratio of square of amplitude for incident et al., 2016) and (Chen and Zhao, 2006).
and reflected waves) and phase change of the reflected signal with
respect to the incident signal are combined to produce character-
2.1.3. Acoustic reflectometry
istics of a blockage. Further, each pulse is allowed to filter out re-
In acoustic reflectometry technique a pulse of sound is injected
flections from the pipeline ends.
by an acoustic pulse generator into a pipeline. As a result, reflection
Equation (Altschuler and Pignotti, 1995) through Equation (Chen
is produced; it passes through an amplifier, drives a loudspeaker
and Zhao, 2006) indicates the relationship of different parameters
and travels along the length of the pipe. This pulse is measured by a
of wave that is propagating in a pipe that contains a blockage at
microphone that is installed at the end. Wang et al. (Wang et al.,
certain length of pipe (Lb), having area ratio (s), wave number of air
2009) performed this experiment in a 63 mm diameter and 16 m
(k ¼ u=c), power reflection ratio (∧),u is radian frequency and c
long PVC pipeline that consisted of several sections, which were
denotes speed of sound.
connected by bends. The recorded signals of 800 Hz (Fig 4) were

2 
2  analysed using matched filters to overcome the problem of back-
∧ ¼ s2  1 s2 þ 1 þ 4s2 cot2 kLb (1) ground noise (Fig 5). Matched filtering generates series of peaks
which indicates the change in cross sectional area inside the
Where, phase difference (j) can be expressed as

Fig. 3. Frequency response at three accelerometer location (Lile et al., 2012). Fig. 4. Reflections of 800 Hz waves.
100 S. Datta, S. Sarkar / Journal of Loss Prevention in the Process Industries 41 (2016) 97e106
Fig. 5. Matched filtering analysis of 800 Hz waves.
pipeline i.e. presence of blockage (Fig 6). Acoustic reflectometry
method is also used for detecting leakage in a pipe line
(Papadopoulou et al., 2008).
2.1.4. Transient wave blockage interaction and blockage detection
Duan et al. (Duan et al., 2014) demonstrated another method to
detect blockage in pipes by observing the changes of system reso-
nance frequencies. The analytical method is used to explain the
blockage-induced frequency shifts in water pipelines. To explain
the frequency shift phenomenon induced by an extended blockage,
a wave perturbation analysis has been conducted. Numerical and
experimental tests have been carried out for the validation of
analytical results which shows the efficiency and accuracy of the
method.
2.1.5. Stochastic successive linear estimator (SLE)
Massari et al. (Massari et al., 2014) proposed a stochastic tool
which can determine the presence of partial blockages in pipelines.
Due to partial blockage, estimation of diameter distribution shows
unbiasness. SLE provides the best unbiased estimation among them
Fig. 6. Presence of Blockage in a pipe (Wang et al., 2009).
to understand the position and size of the blockage. The main
advantage of the approach is its accountability of complex geom-
etries induced by blockages, even with the existence of many
structure errors of transient simulation.
Apart from the above mentioned method, Impedance method is
also a useful way to detect both blockage and leakage in branched
pipeline (Kim, 2014). Impedance method converts various formu-
lated transient responses into time domain and then calibrate
different results related to both blockage and leakage detection. The
radioisotope technology (Robins, 2005) can be used to determine
blockage in pipe rapidly. For locating blockages in a pipeline,
transmission of gamma-ray is an efficient method and can
enumerate the quantity of deposition in a pipe easily. The ther-
modynamic solideliquid equilibrium model is another approach
for calculating the precipitation of wax in oil pipeline (Chen and
Zhao, 2006). For the estimation of precipitated solid at different
temperatures in a pipeline, a model has been prepared based on a
theory related to molecular thermodynamics. The predicted results
from the model are compared with experimental data and the re-
sults obtained are quite satisfactory. Another study was also per-
formed to detect partial blockage in pipeline where damping of
fluid transient was used (Wang et al., 2005). In this study, the
magnitude of damping rate showed the blockage size and ratios of
various damping rates, & the location of the blockage inside a pipe.
2.2. Leakage detection techniques
Some of important the methods applied for accurate detection
of pipeline leakage is discussed as follows.
2.2.1. Negative pressure wave based leak detection system
Among the considerable leak detection and location methods,
the Negative Pressure Wave (NPW) based method has been widely
used for locating leaks in long distance liquid pipelines in station-
ary operation mode (Sun and Chang, 2014). Depending on the
pipeline model, this methodology uses different mathematical
expressions for attenuation of the negative pressure wave travel-
ling along the pipeline and is used for Smallest Detectable Leakage
Flow Rate (SDLFR) calculation (Chuanhu et al., 2008). It is found
that speed of wave in pipeline, outlet and inlet pressure, instrument
precision, and noise etc. influences the SDLFR computation.
2.2.2. Fibre sensor based leak detection
Distributed fibre optic sensors are use to detect pipe line fault
due to its higher sensitivity and wide range of safety applications
(Huang et al., 2007). Here, carrier technique generated by broad-
band phase is used. This method detects phase signal induced by
the acoustic wave. Finally using null frequency of the output
spectrum, the location of leakage is calculated (Kurmer et al., 1991).
From the experimental result, the 1.2 mm single mode fibre (SMF)
with Kevlar jacket has been used that has high tensile ability. The
advantage of this system is its in-line configuration i.e. assembling
the fibre along with the pipelines in order to get real time field data.
The results of this method proved its efficiency in detecting the leak
location. But it has certain limitations, like soundproof material has
to be added to protect the sensing fibre from the environmental
noise which is fitted in the pipeline in order to heighten the signal
to noise ratio. Due to its wide applicability, fibre optic sensors are
used for thermal monitoring in a pipe for the detection of leakage in
gas and oil pipelines (Bhuiyan et al., 2016).
2.2.3. SVM-based pipeline leakage detection
Qu et al. (Qu et al., 2010) proposed that Support Vector Machine
can be used efficiently to detect the defect in the pipe line. SVM can
identify the anomalous measures along with a pipeline effectively.
 S. Datta, S. Sarkar / Journal of Loss Prevention in the Process Industries 41 (2016) 97e106
The sensors, in an optical cable, sense the signals of vibration using
MacheZehnder optical fibre interferometer principle in a pipeline.
Optical cable consists of three fibre out of which one fibre is used
for the transportation of signals and remaining two are for sensing
purpose. The optical signal are transformed into electric signals in a
photodiode where signals are transmitted. With the help of
amplifying and filtering circuit, these signals are then processed.
These signals are processed through an Analogue to Digital con-
verter during transmission to a computer. Meanwhile, the optical
fibres sense various stress and strains and consequently, phase
change of light waves occur. Thus, leakage along a pipeline can be
detected using the distributed optical fibre sensor through the real
l
Hi  Hi
(5)
time vibration signals.
2.2.4. Piezoelectric AE sensor
In order to define the shortest wave paths generated form a
leakage in a pipeline the Piezoelectric AE sensor methodology
combines the pipeline geometric boundaries using the local coor-
dinate system (Ozevin and Harding, 2012; Grabec, 1978). It con-
siders the distance of leak sources to the AE sensors as the shortest
wave path (Ozevin., 2011). Leak in a pressurised pipe line causes
turbulent flow at its location and as a result solid particles or gas
bubbles impact on the pipeline material. The energy generated due
to impact causes propagating elastic wave. This wave can be
detected by the sensors fitted on the pipeline. It is observed that,
the acoustic wave is less attenuative (in dB/m) in a gas filled
pipeline, comparative to the liquid filled pipe. Therefore, AE
method is more reliably applicable to gas pipe. As attenuation is
related to frequency so low frequency AE sensors i.e. less than
20 kHz can be used in such cases.
2.2.5. Filter diagonalization method (FDM)
As discussed in Section 2.1.1 along with blockage detection Fast
Fourier transform (FFT) can also be used for leak detection but has
some limitation such as FFT can be applied for a pipeline having few
bends, pipeline should not be buried and should be free from traffic
noise. Ekuakile et al. (Ekuakile et al., 2009) compared filter diago-
nalization method (FDM) or an alternative method for leak detec-
tion with FFT based spectrum analysis method. FDM is applicable in
case of different complications in a zigzag pipeline, underground
pipe lines or in noisy environment. In this study FDM is used in an
experimental setup to study the behaviour of spectral analysis for
leak detection. Further a comparison is made with the results of
leak detection using FDM and FFT technique on same condition.
Results show that FDM can detect leak sooner than FFT technique.
2.2.6. Harmonic wavelet analysis
Hu et al. (Hu et al., 2011) explored harmonic wavelet method for
small leakage detection in a long length pipe line. In the long dis-
tance pipeline, small leak detection and remote monitoring is an
important issue. Signals from small leak and its precise detection
are typically complicated in noise background and may cause false
alarm or miss alarm. Harmonic wavelet method minimises the
above problem. During sudden pressure drops in side a pipeline,
the pressure waves are acquired from upstream and downstream
stations and time difference is analysed which shows the leakage
location. Pipeline pressure signal is decomposed onto each wavelet
level by harmonic wavelet transform, and the results are shown in
terms of, time-frequency contour map, the time frequency mesh
map and time frequency profile plot.
2.2.7. Genetic algorithm (GA) in combination with the inverse
transient method
Generally a most important doubt to develop a computer model
is the interior condition of pipes in the network, mainly when they
101
are old. Different researches show that genetic algorithm is a pro-
ficient tool for leak detection and prediction in a pipe network in
different conditions(Murphy et al., 1993; Simpson et al., 1994). But
Genetic Algorithm technique in combination with the inverse
transient method (ITM) is a better way for detecting leaks and
friction factors in water pipe lines (Vitkovsky et al., 2000). Ac-
cording to Liggett and Chen (Liggett and Chen, 1994) ITM minimizes
the difference between calculated Hydraulic Grade Line (HGL) and
measured values of HGL. GA is generally formulated for maximizing
the fitness of a string as follows.
n
fitness ¼ 
i¼1
Where n is the number of data points, Hil is measured HGL, Hi is
modelled HGL.
A pipe network is considered that it consists of 11 pipes and
seven nodes. Chromosomes have been encoded in the GA for the
pipe network. Two new GA operators for crossover and mutation
have been introduced. In the GA method, an initial population of
solutions of leaks and friction factors are arbitrarily generated. Here
ITM, using the Genetic Algorithm technique, is efficient to find out
leakage locations and magnitudes and at the same time finding the
friction factors for various transient data record lengths.
2.2.8. Computational fluid dynamic simulation
Mansour et al.(Mansour et al., 2012) investigated the influence
of size and shape of the leakage on the water pressure distribution
in a pipe with the help of FLUENT CFD package. The calculations
were performed within the flow domain of 0.1 m pipe diameter and
2 m length. Thereafter, Fluent CFD simulated 3-D turbulent flow
inside the pipe which utilizes standard keε model (Fig. 7). In Fig 7,
the location of leak is placed at top of pipeline surface (X ¼ 0,Y ¼ D/
2, Z ¼ 0) and flow is occurring in Z direction (Z ¼ 1 to Z ¼ 1 i.e.
2 m). Due to presence of leak, variation of pressure is occurring. This
variation of pressure along the pipe centreline is shown in Fig 8. But
because of small leak flow rate, this difference is difficult to notice
(Fig 8), which appears as almost linear. Moreover, pressure gradient
(i.e. vP=vZ) in flow direction can also fluctuate due to the presence
of leakage in a pipe. Fig 9 shows the variation of pressure gradient
along pipe centreline. Generally pressure gradient is almost con-
stant along the pipe cross section unless it is close to any leak. Only
close to the leak hole, pressure gradient fluctuation can be observed
alsong the centreline.
Apart from the above methods, several procedures are also
applicable for successful implementation of fault detection tech-
niques for a pipe. Lee et al. (Lee et al., 2005) proposed a technique
for leak detection using frequency response diagram (FRD) in a
pipeline. The nature of resonance peaks in frequency response di-
agram indicates the presence of pipeline fault or leakage. For
explanation of resonance peaks and its nature, different analytical
expressions have been used. For locating single as well as multiple
leaks, this frequency pattern can be used effectively. Grey Relational
Analysis is other effective method which can analyse and evaluate
all the factors that influences pipeline leakage (Jing and Hong,
2012). GRA follows Grey System Theory and establishes a predic-
tion model for fault detection by multiple linear regression analysis
using factors related to leakage (Liu et al., 2004; Deng, 1986). As per
prediction accuracy, GRA used in water distribution network is
quite satisfactorily. For detection of leakage in buried water pipe-
line, ground penetrating radar (GPR) can be used as a reliable and
effective method (Ayala-Cabrera et al., 2011; Lai et al., 2016;
Nakhkash and Mahmood- Zadeh, 2004; Stampolidis et al., 2003;
Takahashi and Sato, 2006). To get enhanced recognition of
102 S. Datta, S. Sarkar / Journal of Loss Prevention in the Process Industries 41 (2016) 97e106

Fig. 7. Physical model and coordinate system (leak centre at x ¼ 0, y ¼ D/2, z ¼ 0) (Mansour et al., 2012).

Fig. 8. Variation of pressure along the pipe centreline (Mansour et al., 2012).

Fig. 9. Pressure gradient variation along pipe centreline (Mansour et al., 2012).

pipeline leakage Fuzzy system can be used satisfactorily (da Silva
et al., 2005). Primarily the system is trained offline with leakage
transient data sets. Various parameters such as flow rate, pressure
variation etc. are used as input variable. Variables related to pipe
line failure diagnoses such as alarm of leakage, measurement of
false alarm etc. are considered as output variable. Fuzzy system is a
suitable and proficient tool which reduces error during measure-
ment with minimal cost. For locating leaks in a water pipelines,
artificial neuro fuzzy model can be employed (Wachla et al., 2015).
Using subtractive clustering method (Chiu, 1994) and anfis (Jang,
1993) algorithm accuracy of leakage location and detection has
been increased, where at the same time increases complexity.
 S. Datta, S. Sarkar / Journal of Loss Prevention in the Process Industries 41 (2016) 97e106
Elucidation of Magnetic Flux Leakage signals and inversion tech-
niques for fault detection and delineation of oil and gas pipeline are
discussed by several researchers (Joshi et al., 2006; Mandache and
Clapham, 2003; Altschuler and Pignotti, 1995; Saha and
Mukhopadhyay, 2010). But to reduce error and viability ,a three-
stage algorithm has been proposed where large volume of MFL
signals are compressed (Kathirmani et al., 2012). The proposed al-
gorithm reduces the data loss but also increases computational
complexity. For handling a large amount of data, Mulholland et al.
(Mulholland et al., 2014) proposed a linear programming technique
using sparse flow measurement, for identification of leak in a water
pipe network. Verde et al.(Verde et al., 2007) proposed a process to
identify the presence of leak or pipe line fault. Using transient and
static behaviour of fluid, detection of leak in a pressurized pipeline
has been implemented. The method consists of nonlinear model of
family F using parameters related to leakage in a pipeline and is
extended to locate more leaks in a pipe. This technique minimises
the error during the calculation and reduces research interval. Us-
ing Ad-hoc wireless sensors in a pipe network is an effortless and
efficient technique consumes less power to detect any leaks in the
underground pipes. Without any physical connection, the receivers
receive the signals from sensors and communicate to central unit
that identify the leakage in pipe(Trinchero and Stefanelli, 2009; Sun
and Akyildiz, 2010; Kadi et al., 2013; Sun et al., 2011). Lei et al. (Lei
et al., 2012) used characteristics of pressure fluctuation and damp
wave originated from reflection of leak to detect pipeline leakage.
Presence of leakage in a pipe line increases reflected waves and
generate discontinuation in signals. During measurement discon-
tinuation of signals is observed. To show the correlation among
pipeline leakage with magnitude of reflected wave, Method of
Characteristics approach has been used. The outcome of the
method is found more reliable.
3. Discussion of different methods
As mentioned in previous sections, there are many methods and
devices for the detection of faults in the pipeline. Overviews of
these methods are listed below:
i. In frequency response diagram analysis, it is observed that
with the increase of blockage thickness frequency increases.
ii. Choice of frequencies in pulse eco methodology for PVC pipe
is considered in such a manner so that upper frequency limit
is kept less than the frequency corresponding to fast high
order mode terminates.
iii. Using acoustic reflectometry method, the pulse of sound of
low frequency signals is injected into a long pipe line in oil
and gas industry to measure the reflections in order to detect
blockage or leakage both.
iv. Software based method like Stochastic successive linear
estimator is used to get both the size and the location of
blockage.
v. Process of leakage detection in a pipe is done by the change
of amplitude and attenuation of negative pressure wave
along pipeline generated due to leakage.
vi. Support vector machine based leakage detection is appli-
cable to both oil and gas pipelines, which has good
precession.
vii. For real time damage detection in pressurised pipeline,
acoustic emission process is applicable. AE sensors give good
result in gas pipe line comparative to the liquid.
viii. FDM algorithm can be used for zigzag pipeline with bends.
ix. Errors for FDM are comparatively less than FFT.
x. Frequency response diagram is used for leakage detection.
But main drawback of the method is that if two leaks are
103
located on symmetric position of pipeline, they will be
sensed as a single leak. Moreover, if any leak is located at
middle point it will not be sensed by the method.
xi. Software based method like Non linear based model can be
used to detect multiple leaks in a pipeline using transient
and steady state condition of fluid in leak condition.
Moreover, merits and demerits of these above mentioned
methods are also discussed in Table 1. Finally applicability of these
methods and their comparisons are depicted in Table 2.
4. Conclusion
In this paper an attempt has been made to discuss on several
pipeline fault detection methods, which help for smooth func-
tioning of a plant or industry and to keep safety of the environment.
Generally, for pipeline fault detection, different software based and
hardware based methods are used. It is observed that rapid spec-
trum analysis can be used for blockage detection. Whereas to detect
very small size of leak, attenuation of negative pressure is useful.
For getting good precision value in long length oil pipeline Support
Vector Machine (SVM) based leak detection process is efficiently
used. In addition, Fibre sensors are effective for detecting both
larger and smaller leaks of gas and liquid pipe line both. But for
zigzag pipelines FDM algorithm approach is generally preferred.
Similarly AE sensors, FRD approach, Method of characteristics
(MOC) etc. are also used for leak detection. Computational Fluid
Dynamic is applicable for simulation of leak in a pipe. For low
computational cost Fuzzy system can be used.
Among all fault detection methods, Acoustic reflectometry is
found most suitable due to following reasons: (i) This methodology
is economical, proficient to identify blockages and holes in pipe as
small as 1% of its diameter (Tables 1 and 2) (ii) Methodology is
applicable for pipe layouts such as straight, zigzag and long, short
length pipes (Table 2) (iii) Testing frequency is 100e800 Hz which
is sufficient to detect blockage (iv) For low density fluid i.e. gas,
accuracy of acoustic reflectometry is good but this approach can
also be useful for high and medium density fluid (i.e. water, oil)
pipeline. As per preference and apllicability, Acoustic reflectometry
is followed by another effective method viz. Impedance method
because of the following reasons: (i) Methodology is applicable for
leakage and blockage both in branched pipeline (Table 2). (ii) This
method has satisfactory calibration for leak quantity though it
shows uncertainty for location of blockage and leakage within
pipeline (Table 1). Further, it is observed that Vibration analysis is
another suitable option because of easy fault detection. Similarly
radioisotope technology is other good option to detect blockage in
oil and gas pipelines (Table 2). But safety handling is required to
avoid different hazards (Table 1). Otherwise different fault detec-
tion techniques such as Magnetic flux analysis for leakage detection
and sensor based fault detection techniques are also efficient ap-
proaches but the processes are costly. However, cost of different
other methods could not be found in literature, therefore could not
be included in the comparison analysis in Table 2.
Moreover, in view of the discussions made in previous sections,
following aspects are needed to be improved for better operation of
pipelines in industries: (i) Standardization of fault (blockage and
leakage) detection techniques for pipeline is required to be made
and proposed method should be economically viable. (ii) Method
should cover different types of flowing fluid. (iii) Method should be
useful for different layout i.e. zigzag as well as straight, short and
long distance pipeline. (iv) To improve the accuracy of the predic-
tion about location of various faults in a pipeline, more field data
are required to be used in soft computing techniques which may be
more effective to produce good result and cost effective.
104 S. Datta, S. Sarkar / Journal of Loss Prevention in the Process Industries 41 (2016) 97e106

Table 1
Merits and demerits of different fault detection techniques.

Techniques/Approach Merits Demerits References

Vibration analysis Satisfactory performance and easy to detect 1. Good contact between accelerometer and pipe Lile et al., 2012; Kim et al., 2007.
blockage in pipe by using the relationship is necessary.
between size of blockage and vibration signal. 2. Accuracy depends on the closeness of the fault
to different measuring points.
Pulse eco methodology Methodology is fast and efficient, somewhat Quite difficulty for short blockages. Duan et al., 2015
huge number of blockages can be fruitfully
detected.
Acoustic reflectometry Methodology is economical, proficient to Fault detection is challenging due to presence of Wang et al., 2009; Papadopoulou et al., 2008.
technique identify blockages and holes in pipe as small high level of noise through the pipe and requires
as 1% of its diameter. signal processing technique.
Transient wave blockage Approach is precise and efficient. Difficult to detect multiple extended blockages. Duan et al., 2014.
interaction and
blockage detection
Stochastic Successive Approach can simply applicable for complex Error related to size, length and location of Massari et al., 2014.
linear estimator geometries due to presence of blockage in blockage in a pipe may present.
pipe.
Impedence method Applicable for blockage and leakage both in a Prediction to Locate blockage and leakage is Kim, 2014
pipe. little bit difficult.
Radioisotope technology The measurements related to pipeline fault is Skilled operator is require during measurements Robins, 2005
very fast. otherwise hazardous environment will be
created.
Damping of fluid Response given by the technique is fast and Approach is not suitable for older pipeline. Wang et al., 2005.
transient precise.
Thermodynamic solid Quality of predicting results are good and Methodology is not appropriate for wide range Chen and Zhao, 2006.
eliquid equilibrium satisfactory specially for crude oil. of fluid having different characteristics.
model
Negative pressure wave Simple performance, low cost, satisfactory Methodology is not appropriate for short Sun and Chang, 2014; Chuanhu et al., 2008.
based leak detection leak location and detection. distance transportation pipelines.
Fibre sensor based leak Methodology is more practical and can allow In the absence of sound proof material, the Huang et al., 2007; Kurmer et al., 1991;
detection for spreading the sensing fibre along the approach is inefficient. Bhuiyan et al., 2016.
pipelines easily.
SVM-based pipeline Good performance and precision for For good accuracy large amount of field datasets Qu et al., 2010;
leakage detection identification and locating the leakage in are required.
pipelines.
Piezoelectric AE sensor Online monitoring is effortless and able to In complex pipeline network leakage detection Ozevin and Harding, 2012; Grabec, 1978;
locate time dependent failure (leakage) in using AE method is difficult. Ozevin., 2011.
pipeline.
Filter diagonalization Wide range of applicability and can detect FDM produces an error during the locating of Ekuakile et al., 2009.
method leakage in noisy environment also. leakage but less than FFT.
Harmonic wavelet Approach is trustworthy, safe and can easily Process is so long. Hu et al., 2011.
analysis detect small leakage in a pipeline accurately.
Genetic algorithm (GA) in The GA method does not assure to get the best This method is able to bind its search in the Vitkovsky et al., 2000; Liggett and Chen, 1994;
combination with the possible solution. parameter space and also calibrate the friction Murphy et al., 1993; Simpson et al., 1994.
inverse transient factors in water distribution systems efficiently.
method
Computational fluid Simulation results are good. Prediction of leakage in a pipeline is difficult. Mansour et al., 2012.
dynamic simulation
Frequency Response For locating single as well as multiple leaks, Detection of leaks in pipe exactly at the mid- Lee et al., 2005.
Diagram. this approach is well used. point is too much difficult.
GRA Prediction of leakage in pipe network is Create difficulty in large diameter pipe. Jing and Hong, 2012.
realistic.
Ground penetrating radar Approach is reliable and applicable for 1. Leakage detection is difficult for drainage Ayala-Cabrera et al., 2011; Lai et al., 2016;
(GPR) pressurized water supply pipes. pipes having reinforcement bars in concrete Nakhkash and Mahmood- Zadeh, 2004;
pavement. Stampolidis et al., 2003; Takahashi and Sato,
2. Skilled operator is needed. 2006
3. Costly.
Fuzzy system Low level of false alarms with minimal cost. Procedure is complex. da Silva et al., 2005.
Artificial neuro fuzzy Accuracy of fault detection is good. Procedure is complex Wachla et al., 2015.
model
Magnetic Flux Leakage Data loss has been reduced. Computational complexity Kathirmani et al., 2012.
detection
Sparse flow Large amount of data can be handled. Procedure is complex Mulholland et al., 2014.
measurement
Transient behaviour of During calculation, this methodology reduces On-line implementation is not possible. Verde et al., 2007.
fluid error and research interval.
Ad-hoc wireless sensors Easy maintenance and low power Limited communication range. Trinchero and Stefanelli, 2009; Sun and
consumption. Akyildiz, 2010; Kadi et al., 2013; Sun et al.,
2011.
Transient pressure Approach is simple and efficient. Methodology is not appropriate for noisy Lei et al., 2012
oscillation environment.
 S. Datta, S. Sarkar / Journal of Loss Prevention in the Process Industries 41 (2016) 97e106 105

Table 2
Applicability of different methods and their comparisons.

Types of fluid Pipe layout Length Methodology to detect fault[group rank] (Reference)

Straight Zigzag Network Branched Long Short(<1 km) Blockage Leakage

(i) (ii) (iii) (iv) (v) (vi) (vii) (viii) (ix)

High Yes e e e e Yes Vibration acceleration response tested e

density on the specimen having 1000 mm
(r  1000kg=m3 , length. (Lile et al., 2012)
e.g. Water) Yes e e e Yes e Pulse echo methodology. (Duan et al., e
2015)
Yes e e e e Yes Transient wave blockage interaction e
and blockage detection. (Duan et al.,
2014)
Yes e e e e Yes Stochastic Successive linear estimator. e
(Massari et al., 2014)
e e e Yes e Yes Impedence method. (Kim, 2014) Impedence method. (Kim, 2014)
Yes e e e Yes e Radioisotope technology. (Robins, e
2005)
Yes e e e e Yes Damping of fluid transient. (Wang e
et al., 2005)
e Yes e e e Yes e Spectral analysis by FDM method. (Ekuakile et al.,
2009)
e e Yes e e Yes e Genetic algorithm (GA) in combination with the
inverse transient method. (Vitkovsky et al., 2000)
Yes e e e e Yes e CFD simulation with averaged power spectral
density and Fast Fourier Transmission of pressure
fluctuation. (Mansour et al., 2012)
Yes e e e Yes e e Frequency Response Diagram. (Lee et al., 2005)
e e Yes e Yes e e GRA. (Jing and Hong, 2012)
e e Yes e e Yes e Transient behaviour of fluid. (Verde et al., 2007)
Yes e e e e Yes e Transient pressure oscillation. (Lei et al., 2012)
e Yes e e Yes e e Sparse flow measurement. (Mulholland et al., 2014)
Yes e e e e Yes e Ground penetrating radar.(Lai et al., 2016)
e e Yes e Yes e e Artificial neuro fuzzy model.(Wachla et al., 2015)
e e Yes e e Yes e Ad-hoc wireless sensors.(Trinchero and Stefanelli,
2009; Sun and Akyildiz, 2010; Kadi et al., 2013; Sun
et al., 2011)
Medium Yes e e e Yes e Thermodynamic solideliquid e
density equilibrium model. (Chen and Zhao,
(Oil) 2006)
Yes e e e Yes e e NPW based Method. (Sun and Chang, 2014)
Yes e e e Yes e e SVM technique. (Qu et al., 2010)
Yes e e e Yes e e Harmonic wavelet analysis. (Hu et al., 2011)
Yes e e e Yes e Radioisotope technology. (Robins, e
2005)
Yes e e e Yes e e Magnetic Flux Leakage detection.(Kathirmani et al.,
2012)
Yes e e e Yes e e Interferometric fibre sensor based leak detection.
(Huang et al., 2007; Kurmer et al., 1991; Bhuiyan
et al., 2016)
Low density Yes Yes e e Yes Yes Acoustic reflectometry technique. Acoustic reflectometry technique. (Wang et al.,
(e.g. Gas) (Wang et al., 2009) 2009)
Yes e e e Yes e Radioisotope technology.(Robins, e
2005)
Yes e e e e Yes e SVM technique. (Qu et al., 2010)
Yes e e e Yes e e Interferometric fibre sensor based leak detection.
(Huang et al., 2007; Kurmer et al., 1991; Bhuiyan
et al., 2016)
e e Yes e e Yes e Piezoelectric sensors. (Ozevin and Harding, 2012)
Yes e e e e Yes e Fuzzy system. (da Silva et al., 2005)
Yes e e e Yes e e Harmonic wavelet analysis. (Hu et al., 2011)
Yes e e e Yes e e Magnetic Flux Leakage detection.(Kathirmani et al.,
2012)

Acknowledgement ∧ Power reflection ratio

K Wave number in air
The authors gratefully acknowledge the Department of Me- C Speed of sound
chanical Engineering, ISM Dhanbad, for constructive analysis and Hi Modelled hydraulic grade line
support. Lb Blockage at certain length of pipe [m]
j Phase difference
List of Symbols u Radian frequency
Hil Measured Hydraulic grade line
s Area ratio in a pipe having blockage
106 S. Datta, S. Sarkar / Journal of Loss Prevention in the Process Industries 41 (2016) 97e106

References Liu, Si-feng, Dang, Yao-guo, Fang, Zhi-geng, 2004. Grey System Theory and its
Application, 5. Beijing Science Press, pp. 50e59.
Mandache, C., Clapham, L., 2003. A model for magnetic flux leakage signal predic-
Altschuler, E., Pignotti, A., 1995. .Nonlinear model of flaw detection in steel pipes by
tions. J. Phys. D. Appl. Phys. 36, 2427e2731.
magnetic flux leakage. NDT E Int 28 (1), 35e40.
rez-García, R., 2011. Location of buried Mansour, R. Ben, Habib, M.A., Khalifa, A., Toumi Youcef, K., Chatzigeorgiou, D., 2012.
Ayala-Cabrera, D., Herrera, M., Izquierdo, J., Pe
.Computational fluid dynamic simulation of small leaks in water pipelines for
plastic pipes using multi-agent support based on GPR images. J. Appl. Geophys
direct leak pressure transduction. Comput. Fluids 57, 110e123.
75 (4), 679e686.
Massari, C., Yeh, T.C.J., Ferrante, M., Brunone, B., Meniconi, S., 2014. A stochastic tool
Bhuiyan, Samad Md Abdus, Hossain Alamgir, Mohammad, Alam, M. Jahrul, 2016.
for determining the presence of partial blockages in viscoelastic pipelines: first
A computational model of thermal monitoring at a leakage in pipelines. Int. J.
experimental results. Procedia Eng. 70, 1112e1120.
Heat Mass Transf. 92, ,330e338.
Mulholland, Michael, Purdon, Andrew, Latifi Abderrazak, M.,
Chen, Wuhua, Zhao, Zongchang, 2006. Thermodynamic modelling of wax precipi-
Brouckaert, Christopher, 2014. Leak identification in a water distribution
tation in Crude oils. Chin. J. Chem. Eng. 14, 685e689.
network using sparse flow measurements. Comput. Chem. Eng. 66, ,252e258.
Chiu, S., 1994. .Fuzzy model identification based on cluster estimation. J. Intelligent
Murphy, L.J., Simpson, A.R., Dandy, G.C., 1993. Design of a network using genetic
FuzzySystems 2 (3).
algorithms. Water 20, 40e42.
Chuanhu, Ge, Guizeng, Wang, Hao, Ye, 2008. Analysis of the smallest detectable
Nakhkash, M., Mahmood-Zadeh, M.R., 2004. Water leak detection using ground
leakage flow rate of negative pressure wave-based leak detection systems for
penetrating radar. In: Proceedings of the 10th International Conference on
liquid pipelines. Comput. Chem. Eng. 32, 1669e1680.
Ground Penetrating Radar, pp. 525e528.
da Silva, H.V., Morooka Celso, K., Guilherme Ivan, R., Fonseca Tiago, C., Mendes
Ozevin, Didem, Harding, James, 2012. Novel leak localization in pressurized pipeline
Jose, R.P., 2005. Leak detection in petroleum pipelines using a fuzzy system.
networks using acoustic emission and geometric connectivity. Int. J. Press.
J. Petroleum Sci. Eng. 49, 223e238.
Vessels Pip. 92, 63e69.
Deng, Ju-long, 1986. Grey Prediction and Decision. Wuhan University of Technology
Ozevin, D., 2011. .Geometry based acoustic source location for spaced frames. Int. J.
Press.
Struct. Health Monit. 10 (5), 503e510.
Duan, Wenbo, Kirby, Ray, Prisutova, Jevgenija, Horoshenkov Kirill, V., 2015. On the
Papadopoulou, K.A., Shamout, M.N., Lennox, B., Mackay, D., Taylor, A.R., Turner, J.T.,
use of power reflection ratio and phase change to determine the geometry of a
Wang, X., 2008. An evaluation of acoustic reflectometry for leakage and
blockage in a pipe. Appl. Acoust. 87, 190e197.
blockage detection. IMechE 222, 959e966.
Duan, H.F., Lee, P.J., Ghidaoui, M.S., Tuck, J., 2014. Transient wave-blockage inter-
Poulakis, Z., Valougeorgis, D., Papadimitriou, C., 2003. Leakage detection in water
action and extended blockage detection in elastic water pipelines. J. Fluids
pipe networks using a Bayesian probabilistic framework. Probabilistic Eng.
Struct. 46, 2e16.
Mech. 18, 315e327.
Ekuakile, Aime Lay, Vendramin, Giuseppe, Trotta, Amerigo, 2009. Spectral analysis
Qu, Zhigang, Feng, Hao, Zeng, Zhoumo, Zhuge, Jingchang, Jin, Shijiu, 2010. A SVM-
of leak detection in a zigzag pipeline: a filter diagonalization method-based
based pipeline leakage detection and pre-warning system. Measurement 43,
algorithm application. Measuremnt 2, 358e367.
513e519.
Grabec, I., 1978. Application of correlation techniques for localization of acoustic
Robins, Lee, 2005. On-line diagnostics techniques in the oil, Gas, and chemical In-
emission sources. Ultrasonics 16 (3), 111e115.
dustry. In: 3rd Middle East Non-destructive Testing Conference & Exhibition.
Hu, Jinqiu, Zhang, Laibin, Liang, Wei, 2011. Detection of small leakage from long
Saha, S., Mukhopadhyay, S., 2010. .Empirical structure for characterizing metal loss
transportation pipeline with complex noise. J. Loss Prev. Process Industries 24,
defects from radial magnetic flux leakage signal. NDT E Int 43,507e43,512.
449e457.
Sandberg, C., Holmes, J., McCoy, K., Koppitsch, H., 1989. The application of a
Huang, Shih-Chu, Lin, Wuu-Wen, Tsai, Meng-Tsan, Chen, Mao-Hsiung, 2007. Fiber
continuous leak detection system to pipelines and associated equipment. IEEE
optic in line distributed sensor for detection and localization of the pipeline
Trans. Industries Appl. 25, 906e909.
leaks. Sensors Actuators A 135, 570e579.
Silva, A Re1naldo, Buiatti, M. Claudio, Cruz, L.Sandra, Pereira, A.F.R.J.O.A.O., 1996.
Jang, J., 1993. Anfis:Adaptive-network based fuzzy inference systems. IEEE Trans.
Pressure wave behaviour and leak detection in pipelines. Comput. Chem. Engg
Syst. Man Cybern. 23 (3), 665e685.
20, 491e496.
Jin, Hao, Zhang, Laibin, Liang, Wei, Ding, Qikun, 2014. Integrated leakage detection
Simpson, A.R., Dandy, G.C., Murphy, L.J., 1994. Genetic algorithms compared to other
and localization model for gas pipelines based on the acoustic wave method.
techniques for pipe optimization. J. Water Resour. Plan. Manag. 120 (4),
J. Loss Prev. Process Industries 27, 74e88.
423e443.
Jing, Kang, Hong, Zhi Zou, 2012. .Time prediction model for pipeline leakage based
Stampolidis, A., Soupios, P., Vallianatos, F., Tsokas, G.N., 2003. Detection of leaks in
on Grey relational analysis. Phys. Procedia 25, 2019e2024.
buried plastic water distribution pipes in urban placesda case study. In: Pro-
Joshi, A., Udpa, L., Udpa, S., Tamburrino, A., 2006. Adaptive wavelets for charac-
ceedings of the 2nd International Workshop on Advanced Ground Penetrating
terizing magnetic flux leakage signals from pipeline inspection. IEEE Trans.
Radar, Delft, The Netherlands, pp. 120e124.
Magn. 42 (10), 3168e3170.
Sun, Liang, Chang, Ning, 2014. Integrated-signal-based leak location method for
Kadi, AL Tariq, Tuwaijri, AL Ziyad, Omran, AL Abdullah, 2013. Wireless sensor net-
liquid pipelines. J. Loss Prev. Process Industries 32, 311e318.
works for leakage detection in underground pipelines: a survey paper. Procedia
Sun, Liang, 2012. Mathematical modeling of the flow in a pipeline with a leak. Math.
Comput. Sci. 21, 491e498.
Comput. Simul. 82, 2253e2267.
Kathirmani, S., Tangirala, A.K., Saha, S., Mukhopadhyay, S., 2012. Online data
Sun, Z., Akyildiz, I.F., 2010. Magnetic induction communications for wireless un-
compression of MFL signals for pipeline inspection. NDT&E Int. 50, 1e9.
derground sensor networks. IEEE Trans. Antenna Propag. 58 (7), 2426e2435.
Kim, Y., Simpson, A.R., Lambert, M.F., 2007. The effect of orifices and blockages on
Sun, Z., Wang, P., Vuran, M.C., Al-Rodhaan, M.A., Al-Dhehaan, A.M., Akyildiz, I.F.,
unsteady pipe flows. World Environ. Water Resour. Congr. 1e10. http://
2011. MISE-PIPE: magnetic induction-based wireless sensor networks for un-
dx.doi.org/10.1061/40927(243)477.
derground pipeline monitoring. Ad Hoc Netw. 9, 218e227.
Kim, S., 2014. Inverse Transient Analysis for a branched pipeline system with
Takahashi, K., Sato, M., 2006. A novel inversion of cross-hole radar data for location
leakage and blockage using impedance method. Procedia Eng. 89, 1350e1357.
of buried pipes. In: Proceedings of the 11th International Conference on Ground
Kurmer, J.P., Kingsley, S.A., Laudo, J.S., Kark, S.J., 1991. Distributed fiber optic acoustic
Penetrating Radar.
sensor for leak detection. Distributed Multiplexed Fiber Opt. Sensors, Proc. SPIE
Trinchero, D., Stefanelli, R., 2009. Microwave architectures for wireless mobile
1586, 117e128.
monitoring networks inside water distribution conduits. Micr. Theory Tech.
Lai, W.L. Wallace, Chang, K.W. Ray, Sham, F.C. Janet, Pang, K., 2016. Perturbation
IEEE 12, 3298e3306.
mapping of water leak in buried water pipes via laboratory validation experi-
Verde, Cristina, Visairo, Nancy, Gentil, Sylviane, 2007. .Two leaks isolation in a
ments with high-frequency ground penetrating radar (GPR). Tunn. Undergr.
pipeline by transient response. Adv. Water Resour. 30, 1711e1721.
Space Technol. 52, 157e167.
Vitkovsky, P. John, Simpson, R. Angus, Lambert, F. Martin, 2000. Leak detection and
Laurentys, C.A., Bomfim, C.H.M., Menezes, B.R., Caminhas, W.M., 2011. Design of a
calibration using transients and Genetic Algorithms. J. Water Resour. Plan. And
pipeline leakage detection using expert system: a novel approach. Appl. Soft
Manag. 126, 262e265.
Comput. 11, 1057e1066.
Wachla, D., Przystalka, P., Moczulski, W., 2015. A method of leakage location in
Lazhar, Ayed, Taieb, Hadj Lamjed, Taieb, Hadj Ezzeddine, 2013. Two leaks detection
water distribution networks using artificial neuro-fuzzy system. IFAC
in viscoelastic pipeline systems by means of transient. J. Loss Prev. Process
1216e1223.
Industries 26, 1341e1351.
Wang, Jian Xiao, Lambert, F Martin, Simpson, R Angus, 2005. Detection and location
Lee, J. Pedro, Vitkovsky, P. Jhon, Lambert, F. Martin, Simpson, R. Angus, Liggett, A.
of a partial blockage in a pipeline using damping of fluid transients. J. Of Water
James, 2005. Leak location using the pattern of the frequency response diagram
Resour. Plan. And Manag. 131, 244e249.
in pipelines: a numerical study. J. Sound Vib. 284, 1051e1073.
Wang, X., Lennox, B., Turner, J., Lewis, K., Ding, Z., 2009. Blockage detection in long
Lei, Xin Guo, Kai-lin, Yang, Fu-tian, L.I., Tao, Wang, Hui, Fu, 2012. Analysis of first
lengths of pipeline using a new acoustic method. In: 16th International
transient pressure oscillation for leak detection in a single pipeline.
Congress on Sound and Vibration, Krako  w, Poland, pp. 1e8.
J. Hydrodynamics 24 (3), 363e370.
Zadeh, L., 1965. Fuzzy sets. Inf. Control 8 (8), 338e353.
Liggett, J.A., Chen, L.C., 1994. Inverse transient analysis in pipe networks. J. Hydraulic
Zidani, F., Diallo, D., Benbouzid, M., El, Hachemi, Nait-Said, R., 2008. A fuzzy-based
Eng. 120 (8), 934e955.
approach for the diagnosis of fault modes in a voltage-fed PW Minverter in-
Lile, N.L.T., Jaafar, M.H.M., Roslan, M.R., Azmi Muhamad, M.S., 2012. Blockage
duction motor drive. IEEE Trans. Industrial Electron. 55, 586e593.
detection in circular pipe using vibration analysis. Int. J. Adv. Sci. Eng. Inf.
Technol. 2, 54e57.