Journal of Applied Fluid Mechanics, Vol. 12, No. 2, pp. 485-494, 2019.

Available online at www.jafmonline.net, ISSN 1735-3572, EISSN 1735-3645.

DOI: 10.29252/jafm.12.02.28612

Numerical Analysis on Pipeline Leakage Characteristics

for Incompressible Flow
Y. Zeng and R. Luo
National Metrology Centre, 1 Science Park Drive, Singapore, Singapore, 118221, Singapore

†Corresponding Author Email: zeng_yan@nmc.a-star.edu.sg

(Received November 22, 2017; accepted September 28, 2018)

ABSTRACT

Systematic Computational Fluid Dynamics (CFD) simulations on incompressible water pipe flow with
leakage were conducted in the present study. The aim is to provide the understanding of how different
parameters, including the leakage pipe diameter, inlet mass flow rate, and main pipe length, affect the flow
phenomena at the vicinity of the leakage location. The present CFD data show that the leakage pipe diameter
has dominant effect on the leak mass quantity, pressure change at the vicinity of leak location, total pressure
drop and pressure gradient along the main pipe. The effects of both inlet mass flow rate and the main pipe
length on leak mass quantity are comparably important. Due to existence of the leakage pipe, larger velocity
but lower pressure at upstream, and lower velocity but larger pressure at downstream occur at the vicinity of
leakage, which causes adverse pressure at this region. The pressure change resulted from the adverse pressure
increases approximately linear with the leak size ratio (ratio of leakage pipe diameter to main pipe diameter)
when it is smaller than approximately 40%, at which the maximum pressure change at the leak location
occurs. When the leak size ratio is smaller than approximately 5%, the pressure change at the leak location is
seen to be approximately zero, implying negligible pressure difference at the two boundary points of leakage
pipe. There is sudden change in the pressure gradient along the flow direction at the leak location, which
results from a local pressure increase there. When farther away from the leakage, the magnitude of the
maximum pressure gradient along the flow direction is reduced due to attenuation of leakage effect. The
present study proves that CFD analysis could be an effective and less-costly way to investigate pipe flow with
leakage, so as to provide scientific understanding of the physics on pipe flows with leakage.

Keywords: Incompressible flow; Pipeline with leakage; Numerical analysis; Leakage characteristics.

NOMENCLATURE

A* leak area ratio min inlet mass flow rate

d diameter of the main pipe mL leak mass flow rate
dL diameter of the discharge pipe for mL* leak quantity ratio
leakage u velocity in x direction
dL* leak size ratio v velocity in y direction
DPT pressure drop along the main pipe w velocity in z direction
DPL pressure change at the vicinity of leak x spanwise direction
location xL leak location
L the length of the main pipe y vertical direction
PL@up pressure at upstream of leak location z flow direction
PL@down pressure at upstream of leak location
Pout@main pressure at the outlet of the main pipe ρ density of water
P@Z1 pressure at z1= 0.05m μ dynamic viscosity of water
P@Z2 pressure at z2= 1.50m

1. INTRODUCTION pipeline rupture cannot be avoided which leads to

leaks in pipeline, and thus risks associated with
Pipeline transportation is an efficient and economic accidental releases of transported product are still
means to convey the transport of gas, oil and water. high (da Silva et al., 2005; Costa, et al., 2001).
However, due to corrosion and pressure surges, the
F Y. Zeng / JAFM, Vol. 12, No. 2, pp. 485-494, 2019.

Fluid leakage may cause serious pollution, injuries, leak detector has been verified via hardware
and fatalities. In deepwater operation which is an implementation (Zhang et al., 2013). Based on
important part of producing oil and gas supplies, vibration sensors and generalized cross-correlation
leakage from subsea pipelines has been a serious techniques, a new method to detect and locate
problem, because of its potential threat to leakage in water pipes was proposed by Choi et al.
surrounding marine environments, significant (2017). With the adoption of a modified maximum-
monetary loss from the delay of hydrocarbon likelihood (ML) prefilter with a regularization
production, and difficulty in detection and remedial factor, this method outperforms the conventional
processes (Kam, 2010). The problem of leakage is techniques by the verification from field
also serious for water pipelines including fresh measurements.
water supply as well as waste water treatment
which may cause not only waste of water resources With the development of high performance
but also waste water contamination to the computing, Computational Fluid Dynamics (CFD)
environment (Liggett and Chen, 1994; Vitkovsky et simulations have been applied to investigate
al., 2001; Kim et al., 2008). Delay in leak detection pipeline flows. Olivares (2009) conducted two- and
of water pipeline may lead to environmental health three-dimensional simulations to investigate
disasters. turbulence flow and acoustics induced in heating
water pipes. It was found that eddies induced by
Delay in leakage detection and repair of a failed leakage cause pipe acoustics. Ben-Mansour et al.
water pipe may also cause significant water loss and (2012) conducted numerical simulations on small
serious damage to infrastructure near the failure. leaks in water pipelines with three different
The amount of non-revenue water has been reported rectangular shapes of leakage: 1mm × 1mm, 2mm ×
to be 15.2-35.1% of drinking water supply in 2mm, 1mm × 10mm. The results showed that the
United States (http://nepis.epa.gov/), 6.5-24.6% in presence of a leak causes measureable differences
Europe (http://www.ndew.de/), and 4.3-27.0% in in the magnitude and frequency of the pressure
Korea (http://me.go.kr/). These losses can be signal spectrum in the range of 220-500Hz. Barbosa
classified into unbilled public usage, apparent losses et al. (2012), numerically studied three-phase flow
which include unauthorized consumption and (heavy oil, water and gas) in vertical pipeline using
metering inaccuracies, and real losses through ANSYS-CFX. The results showed that the effect of
overflows at storage tanks and burst leaks in small leak on the velocity and temperature
distribution pipelines which were caused by bad distribution along the pipe can be negligible. de
connections, pipe corrosion, and physical damages Sousa et al. (2013) studied hydrodynamic of two-
(Choi et al., 2017). Therefore, early detection of phase (oil-water) flow in vertical pipe with a
leakage in water pipeline is of great importance to leakage by using ANSYS-CFX. It was concluded
reduce water production costs and to protect the that volumetric fraction of phases and fluid mixture
safety of public (Puust et al., 2010; Ishido and velocity affect pressure drop and mass flow rate at
Takahashi, 2014; Martini et al., 2015). the leak hole. Shehadeh and Shahata (2013) carried
out CFD analysis to study incompressible pipeline
Many leak detection and localization techniques flow with leakage under different rupture diameters
and methods have been developed and investigated, and fluid flow properties. The numerical results
which can be generally classified into three showed that leaks at high Reynolds number may be
categories, biological, hardware, and software detected more easily than at lower Reynolds
methods (Murvay and Silea, 2012; Shehadeh and number. de Vasconcellos Araújo et al. (2014)
Shahata, 2013) based on the technical nature. conducted simulations to investigate hydrodynamic
Biological methods, referred to as non-technical characteristics in oil pipe. The velocity field for two
methods, use trained animals such as smell- leaks at different locations were reported. The
sensitive dogs and pigs to find out the leak locations results revealed that the leak closer to the inlet
along the pipeline. Hardware methods mainly rely would generate higher pressure drop. CFD analysis
on the usage of special sensors or devices to detect was carried out by Jujuly (2016) to study steady and
leak locations. Further classification can be made transient pipe flow with leakage for four different
depending on the type of sensors or devices used for fluids: methane, nitrogen, water, and crude oil.
detection (Shehadeh and Shahata, 2013). Software Comparisons among the four different flow media
methods use computer programs to monitor the and three different leak sizes were conducted.
evolution of pipeline parameters such as pressure, Water flow behavior in the pipe with leakage, as
temperature, flow rate and so on, so as to infer well as induced acoustics, were investigated at
whether leak occurs or not. different leak locations to give insights into flow
Among all the leak detection methods, the most characteristics near the leak region. A recent study
popular approach for water leakage is to install by Zeng and Luo (2017) numerically investigated
acoustic/vibration sensors or pressure transducers the water flow phenomena in a short pipe with
(Choi et al., 2017). With the utilization of two leakage. The results showed that the relationship
sensors, the leak can be located by estimating the between the pressure change at the upstream and
time difference through correlation of the signals downstream of the leak location and leak mass flow
received. The accuracy of the time difference rate is approximately linear, with dominant effect of
estimation schemes based on spectral transform can the inlet mass flow rate and negligible effect of the
be improved by the short-time Fourier transform pipe length. All these numerical studies have
(STFT) and wavelet transform (Lay-Ekuakille et verified that CFD simulation could be an effective
al., 2009; Ge et al., 2009), and the correlation-based and efficient way to provide theoretical basis to

486
F Y. Zeng / JAFM, Vol. 12, No. 2, pp. 485-494, 2019.

understand the pipeline flows with leakage. Table 1 Pipe sizes and flow boundary conditions
Among plenty of published references on pipeline Pout
Boundary xL d min L dL
flows with leakage, there is short of systematic @main
conditions (m) (mm) (kg/s) (m) (mm)
investigations to explore the underline mechanism (Pa)
governing the leakage which is still not yet clear. In 0
the present study, a series of systematic numerical 0.635
simulations were therefore carried out to investigate
incompressible pipe flow under different conditions 0.953
including the inlet mass flow rate, leakage pipe 1.27
diameter, and main pipe length. Moreover, a wide Mass flow
0.13 2.23 1.588
range of leak size ratio (ratio of the leakage pipe inlet
0.73 12.7 81 & & 1.91
diameter to main pipe diameter) was included in & Pressure
0.32 3.35
this study to cover extensive real-field problems in outlet 2.54
engineering applications. The flow characteristics at 3.2
the vicinity of the leakage in pipe flows were 5.2
captured and analyzed. The simulation results could
provide reference for the actual cases, and deepen 7.2
the understanding of the flow phenomena and 10
leakage characteristics for incompressible pipeline Note that dL = 0 represents no-leak case.
flows.
Three-dimensional governing equations for steady
2. MODEL CONFIGURATION AND state, incompressible flow in Cartesian coordinate
COMPUTATIONAL METHOD are expressed as:

A straight pipe with water as working fluid is u v w

  0 (1)
considered in the present study, while the leakage is x y z
performed by a discharge pipe to atmosphere. All
the dimensions and boundary conditions were set  u u u  P
the same as the experimental work (Molina-  u v w   2u
 x y z  x
Espinosa et al., 2013), which has been illustrated in
our previous study (Zeng and Luo, 2017). The fluid
 v v v  P
domain of the main and discharge pipes is shown in  u v w   2v (2)
Fig. 1, with the origin of the coordinate at the center  x y z  y
of the main pipe inlet surface.
 w w w  P
In this study, two pipe lengths and ten discharge pipe  u v w   2w
diameters at two inlet mass flow rates have been taken  x y z  z
into account, while the diameter of the main pipe,
CFD software ANSYS FLUENT (2013) was
leakage location, and the pressure at outlet remain
utilized to solve steady three dimensional (3D)
unchanged. The length of the discharge pipe is
Navier–Stokes equations to capture the flow
longer than 5 times of its diameter to make the flow
phenomena. In the present study, the Reynolds
fully developed so as to make use of ambient
number (Re) ranges over 1.2 × 107 ~ 3.2 × 107
pressure condition at its outlet. The dimensions and
which is much larger than critical value for the
boundary conditions are listed in Table 1.
turbulent pipe flow. Therefore, the flow is turbulent
and the k-ε turbulence model was adopted due to its
applicability for fully developed flow with fast
convergence. The k-ε turbulence model, which is
numerically robust and stable, can provide fairly
reasonable result and has been widely used for
industrial applications including pipe flow with
leakage (Barbosa et al. 2012; de Sousa et al. 2013;
Jujuly, 2016). SIMPLE algorithm was employed for
pressure-velocity coupling. Second order
discretization scheme was employed for flow
variables, turbulent kinetic energy and turbulent
dissipation rate.
Note that in actual conditions, small leak may often
happen which could be difficult to be detected and
located. Therefore, small step of variation in small
leak size ratio (below 3.2mm) is considered to
capture the slight change of small leak mass
quantity, which could be a reference for the actual
cases.
Fig. 1. 3D and 2D configurations for the main
and discharge pipes.

487
F Y. Zeng / JAFM, Vol. 12, No. 2, pp. 485-494, 2019.

Table 2 Mesh elements and mesh quality

Mesh details Set I Set II Set III
Number of elements 356,553 479,962 619,332
Minimum mesh size close to the leak tube (m) 0.0001656 0.00016580 0.0001476
Maximum mesh size (m) 0.001567 0.0013832 0.001321
Maximum aspect ratio 15.5621 14.9335 15.8503
Maximum skewness 0.7403 0.7243 0.7447
Minimum orthogonal quality 0.2023 0.2037 0.2001
Notes:
1. Skewness from 0 to 1, where the values close to 1 correspond to low quality.
2. Orthogonal quality from 0 to 1, where the values close to 0 correspond to low quality.

3. RESULTS AND DISCUSSION and 10-6 for velocities. A typical example for the
residual change with iteration steps is shown in Fig.
3.1 Grid-Independent Study 4. For all the simulations cases, after around 2000 –
4000 iterations, all the residuals already reduce to
To conduct mesh-independent study, three sets of the set values, and the converged results are
tetrahedral mesh were generated by ANSYS obtained.
meshing for the case of inlet mass flow rate
0.13kg/s, main pipe length 3.35m, and discharge
pipe diameter 5.2mm. The mesh topology near the
leakage pipe region is shown in Fig. 2. It is seen
that finer mesh was generated at the leak pipe and
the region close to the leak pipe, as well as coarser
mesh along the main pipe with smooth transition.
The mesh elements and mesh quality are listed in
Table 2. The velocity magnitude variation and static
pressure variation curves along the line which is
3mm above the leakage along the main pipe under
different sets of mesh are presented in Fig. 3. Based
on mass conservation, it is expected that velocity
magnitude nearly keeps constant before and after
the leak location while the value is decreased after
leakage due to water flowing out through the
leakage pipe, as shown in Fig. 3(a). A sudden
change is found for the static pressure curves at the
leak location (Fig. 3(b)). The velocity variation and
static pressure variation curves obtained from
different sets of mesh nearly overlap, which
indicates the coarsest mesh can produce the grid-
independent results. However, to make the
simulation more accurate, the finest mesh (Set III
with 619,332 mesh elements) was selected to Fig. 3. (a) Velocity variation and (b) static
conduct the simulations. For other cases with pressure variation curves along the line which is
different dimensions, the same mesh resolution was 3mm above the leakage along the pipe under three
applied. sets of mesh for the case of inlet mass flow rate
0.13kg/s, main pipe length 3.35m, and discharge
pipe diameter 5.2mm.

Fig. 2. Mesh Topology at the leakage pipe region.

In the present study, the convergence criteria were Fig. 4. The trend of residual .vs. iteration for the
set as the residuals of 10-3 for continuity, turbulent case of inlet mass flow rate 0.13kg/s, main pipe
kinetic energy as well as turbulent dissipation rate, length 3.35m, and discharge pipe diameter 5.2mm.

488
F Y. Zeng / JAFM, Vol. 12, No. 2, pp. 485-494, 2019.

Table 3 Comparison of present CFD and the reference experimental data

(Molina-Espinosa et al., 2013)
DP (KPa)
Conditions mL * = mL/min
= P@Z1 - P@Z2
CFD 1.76 CFD 10.51%
L=3.35m dL=3.2e-3m min=0.13kg/s EXP 1.35 EXP 8%
Dif (%) 30.37% Dif (%) 31.38%
CFD 6.5 CFD 16.34%
L=2.23m dL=5.2e-3m min=0.32kg/s EXP 6 EXP 17%
Dif (%) 8.33% Dif (%) 3.88%
CFD 5.31 CFD 26.34%
L=2.23m dL=7.2e-3m min=0.32kg/s EXP 4.6 EXP 31%
Dif (%) 15.43% Dif (%) 15.03%
CFD 4.18 CFD 38.03%
L=2.23m dL=1.0e-2m min=0.32kg/s EXP 3.5 EXP 49%
Dif (%) 19.43% Dif (%) 22.39%
Note that Dif (%) = |(CFD-EXP)|/EXP *100%

3.2 Model Validation larger velocity and lower pressure at the upstream
of the leakage location. At the downstream of the
Several cases in the experimental work (Molina- leakage location, there is a lower velocity region
Espinosa et al., 2013) were selected to validate the which causes higher pressure there. Thus, adverse
present CFD modelling. The pressure drop at the pressure occurs at the vicinity of leakage which
two locations (z1=0.05m and z2=1.5m from the results in a sharp change in the pressure variation
inlet) and the leak mass quantity between the curve along the flow direction.
present simulation and the reference experimental
data (Molina-Espinosa et al., 2013) are compared
and listed in Table 3, which has been shown in the
previous study (Zeng and Luo, 2017). The deviation
between CFD and experimental data is within the
range of 3.88% - 31.38%, with the smallest
difference for middle leakage mass quantity ratio of
17% and the largest difference for small leak mass
quantity of 8%, which is reasonable because it is (a)
more difficult to capture small leak mass quantity
by both experiments and simulations due to
uncertainty in experiments and computation error in
simulations.
3.3 Flow Patterns
Figure 5 shows the flow and pressure contours in
the vicinity of the leakage at the center plane along
the flow direction under the condition of inlet mass (b)
flow rate of 0.32 kg/s, main pipe length of 3.35m Fig. 5. (a) Velocity magnitude contour and (b)
and discharge pipe diameter of 5.2mm. Lower pressure contour at Centre Plane (X=0) in the
pressure at the upstream but higher pressure at the vicinity of leak-location under the conditions of
downstream of the leak can be observed, which is inlet mass flow rate 0.32kg/s, main pipe length
due to lower flow velocity downstream resulted 3.35m, and discharge pipe diameter 5.2mm.
from leak. The trend is consistent with the reference
simulation results (Ben-Mansour et al., 2012) and The flow velocity close to the leakage pipe varies
experimental data (Molina-Espinosa et al., 2013). the most as shown in Fig. 7(b), showing the most
significant leakage effect there. With the plane
To further understand the flow characteristics in the
moving upward and farther away from the leakage,
vicinity of the leakage, 3D Streamline and velocity
the velocity variation is reduced, which confirms
vector plots at this region are shown in Fig. 6 and
that the leakage effect is diminishing.
Fig. 7 respectively, for the same case with inlet
mass flow rate of 0.32 kg/s, main pipe length of To show the pressure change at the vicinity of the
3.35m, and discharge pipe diameter of 5.2mm. leak location more clearly, Fig. 8 depicts the static
pressure contours at three different XZ planes:
Consistent with Fig. 5(a), the results of 3D
Plane XZ1 with y = -5.35mm (1mm above the
streamline (Fig. 6) and velocity vector plot (Fig.
leakage pipe), Plane XZ2 with y = -3.35mm (3mm
(7)a) at the vicinity of leakage clearly show that part
above the leakage pipe), and Plane XZ3 with y = 0
of water smoothly flows through the new outlet
(centerline plane and 6.35mm above the leakage
with smaller diameter (leakage pipe), leading to

489
F Y. Zeng / JAFM, Vol. 12, No. 2, pp. 485-494, 2019.

pipe) for the same case under those conditions mesh quality can be enhanced, the leak mass quantity
indicated in Figs 5-7. Similar to the leakage effect can be predicted at even a smaller leak size ratio.
on the flow velocity (see Figs. 7(b) – 7(d)), the
pressure change near the leakage pipe is more
significant and gradually reduces when the plane is
moving upwards to the centerline of the main pipe,
indicating that the effect of the leakage decreases. If
the plane continuously moves upward, the effect of
the leakage will reduce to be negligible eventually
(data not shown).

Fig. 6. 3D streamline at the vicinity of leakage pipe.

The variation curves of the static pressure along

the centerline of the above three XZ planes are
presented in Fig. 9. A sudden change in the static
pressure can be observed, which is caused by the
adverse pressure occurring at the upstream and
downstream of leak location. The pressure
change decreases with an increase in the distance
between the plane and the leakage pipe,
confirming that the leakage effect is reducing.
The results are consistent with the results shown
in Figs. 5-8. Compared with the reference
simulation data (Ben-Mansour et al., 2012), the
pressure change along the centerline of the Plane
XZ3 (y = 0 mm) still exists in the present study,
which could be due to smaller diameter of the
main pipe and thus larger domain that can be
affected by the leakage effect.
3.4 Leakage Quantity Ratio
Leakage mass quantity ratio (mL*) at different
conditions are summarized in Table 4.
As expected, leakage mass quantity ratio increases
with increasing the leak size (or area) ratio when the
other conditions remain unchanged, indicating the key
effect of the leak size (or area) ratio. The effects of
both the inlet mass flow rate and main pipe length are
comparable. For the same inlet mass flow rate, at the
leak size ratio larger than 7.5% (corresponding to the Fig. 7. Velocity vector plots at different planes (a)
area ratio 0.56%), larger leak mass quantity ratio are Plane YZ (x = 0) and Plane XY (z = 0.73mm), (b)
observed at longer main pipe due to larger driven Plane XZ1 (y = -5.35mm), (c) Plane XZ2 (y = -
force, which will be explained later. For the same 335mm), and (d) Plane XZ3 (y = 0), in the vicinity
main pipe length, higher inlet mass flow rate causes of leakage under the conditions of inlet mass flow
larger driven force and larger leakage mass quantity, rate 0.32kg/s, main pipe length 3.35m, and
but the leak quantity ratio may be reduced based on discharge pipe diameter 5.2mm.
larger denominator. It is worth mentioning that for the
leak size ratio below 5% (corresponding area ratio 3.5 Pressure change
0.25%), the leak mass quantity may not be predicted
accurately by present simulations, which could be due Figures 10 shows the effect of leak size ratio on total
to the poor mesh quality under this condition. If the pressure drop along the main pipe (DPT) at different It is

490
F Y. Zeng / JAFM, Vol. 12, No. 2, pp. 485-494, 2019.

Table 4 Leakage quantity ratio at different conditions

mL * mL *
dL* (%) A* (%) min = 0.13 (kg/s) min = 0.32 (kg/s)
L=3.35m L=2.23m L=3.35m L=2.23m
0 0.00% 0 0 0 0
5.00% 0.25% 0.071% 0.077% 0.090% 0.125%
7.50% 0.56% 0.422% 0.270% 0.417% 0.333%
10.00% 1.00% 0.980% 0.688% 0.990% 0.863%
12.50% 1.56% 1.858% 1.317% 1.767% 1.540%
15.00% 2.26% 2.900% 2.265% 2.757% 2.070%
20.00% 4.00% 5.162% 3.911% 5.571% 4.008%
25.20% 6.35% 10.510% 8.000% 9.640% 7.000%
40.94% 16.76% 23.100% 18.520% 21.130% 16.340%
56.69% 32.14% 36.320% 30.060% 33.750% 26.340%
78.74% 62.00% 53.380% 44.150% 47.880% 38.030%

observed that for the lower inlet mass flow rate of the above three XZ planes is shown in Fig. 12.
0.13kg/s, the main pipe length approximately has Compared with the static pressure variation
negligible effect on DPL with the two curves nearly depicted in Fig. 7, the magnitude of pressure
overlapping. For the larger inlet mass flow rate of 0.32 gradient at the leak location is much larger, which
kg/s, the main pipe length effect is more significant, indicates the most rapid change in the rate of
especially for the leak size ratio larger than 15% the pressure there. The sudden change in the
(corresponding to the area ratio 2.25% and the leak pressure gradient at the leak location is because the
mass quantity ratio 2 ~ 2.8%). The pressure change at local pressure increases there, representing the
the vicinity of leak location can be neglected for leak typical characteristic in leak flows. It inlet mass flow
size ratio smaller than 5%, corresponding to area ratio rates and pipe lengths. Consistent trends show that
of 0.25%, and leak quantity ratio of approximately with an increase in the leak size ratio with other
0.07 ~ 0.13%. conditions unchanged, total pressure drop is reduced
which could be due to more mass of water flows
through larger discharge pipe at larger leak size,
leading to lower velocity, higher pressure at the outlet
of the main pipe, and thus lower total pressure drop.
Moreover, with higher inlet mass flow rate but other
conditions remain the same, larger driven force is
needed to push larger quantity of water, the inlet
pressure is increased which causes larger pressure
drop along the main pipe. Similarly, with longer main
pipe but other conditions remain the same, larger
driven force is needed to overcome larger resistance
along the main pipe, the inlet pressure and pressure
drop along the main pipe are both increased.

Fig. 9. Static pressure variation curves along

centerlines at different planes (XZ1, XZ2, and XZ3)
under the conditions of inlet mass flow rate
0.32kg/s, pipe length 3.35m, and discharge pipe
diameter 5.2mm.
Fig. 8. Static pressure contours at different planes
The pressure change between upstream and downstream
(XZ1, XZ2, XZ3) in the vicinity of leak-location
of leakage location (DPL), at different inlet mass flow
under the conditions of inlet mass flow rate
rate and pipe length are summarized relative to the leak
0.32kg/s, main pipe length 3.35m, and discharge
size ratio and leak quantity ratio, as shown in Fig. 11.
pipe diameter 5.2mm.
The upstream and downstream of leakage locations are
set at the two points located 1mm above the leakage pipe
3.6 Pressure Gradient and half-diameter of the leakage pipe away from the
The pressure gradient along the three centerlines of leakage pipe center, as marked in Fig. 2(b).

491
F Y. Zeng / JAFM, Vol. 12, No. 2, pp. 485-494, 2019.

Table 5 Peak pressure gradient at different conditions

min = 0.13 (kg/s) min = 0.32 (kg/s)
Planes
dL* (%) A* (%) mL* dL* (%) A* (%) mL*
XZ1
40% 16% 7-10% 25% 6.25% 18-23%
(y= -5.35mm) 1mm above the leakage pipe.
XZ2
50% 25% 25% 30% 9% 12%
(y=-3.35mm) 3mm above the leakage pipe.
XZ3
No observation of peak pressure gradient
(y=0mm) 5.35mm above the leakage pipe

leak size (or area) ratio, corresponding to higher leak

mass quantity ratio. There no observation of the peak
pressure gradient at Plane XZ3 (centerline plane),
confirming that the leakage effect becomes more
trivial with the plane being farther away from the
leakage pipe. It can be summarized that the pressure
gradient along the main pipe is much more obvious to
be observed due to its larger magnitude, and thus it
would be much more favorable to detect and locate
the leakage if the pressure gradient can be measured
in experiments.

Fig. 10. Total pressure drop along the main

pipe .vs. leak size ratio at different inlet mass flow
rates and main pipe lengths.

Lower pressure at upstream and higher pressure at

downstream of the leak lead to positive magnitude of the
pressure change at the vicinity of leak location, consistent
with the experimental results results (Molina-Espinosa et
al., 2013). The increase in pressure after leak is promoted
by decreasing the downstream flow rate and kinetic
energy due to leak. It is seen that the pressure change at
the vicinity of leak location increases nearly linearly with
the leak size ratio when it is below 40%, while above that
it remains nearly unchanged for the lower inlet mass flow
rate and decreases for the larger inlet mass flow rate,
indicating the peak pressure change at the vicinity of leak
location, corresponding to the area ratio 16% and leak
mass quantity ratio of approximately 20%, consistent
with the data shown in Table 4.
is clearly seen that the effect of leakage decreases
when the plane is farther away from the leakage pipe,
consistent with the previous results shown in Figs. 7-
9. The magnitude of the maximum pressure gradient Fig. 11. Pressure change at the vicinity of leak
decreases from 2344 KPa/s at Plane XZ1 (y = -5.35 location at different inlet mass flow rates and main
mm), to 632.8 KPa/s at Plane XZ2 (y = -3.35 mm), pipe lengths .vs. (a) leak size ratio, and (b) leakage
and finally 166.7 KPa/s at Plane XZ3 (y = 0 mm). quantity ratio.
Compared with the reference simulation data (Ben-
Mansour et al., 2012), much larger magnitude is
found in the present study, which could be probably 4. CONCLUSIONS
due to smaller diameter of the main pipe and thus
more significant leakage effect.
Systematic numerical simulations on incompressible
To further figure out the leakage effect, the water pipe flow with leakage were performed in the
magnitudes of the maximum pressure gradient along present study. Different parameters including the inlet
the three centerlines of the above three XZ planes at mass flow rate, leakage pipe diameter, and pipe length
different leak size (or area) ratios and leakage mass were considered. The effects of these parameters on
quantity ratios are summarized in Table 5. the leak mass quantity, total pressure drop along the
main pipe, pressure change at the vicinity of leak
Table 5 shows the leak size (or area) ratio for the location, and pressure gradient along the main pipe
maximum pressure gradient at above three XZ planes. were investigated and analyzed. The leakage pipe
It is seen that with the plane moving farther away diameter (represented as leak size ratio or area ratio)
above the leakage pipe (from XZ1 to XZ2), the shows dominant effect, which is reasonable and
maximum pressure gradient will be reached at larger

492
F Y. Zeng / JAFM, Vol. 12, No. 2, pp. 485-494, 2019.

expected. Compared with the main pipe length, the phenomena.

simulation results show that the inlet mass flow rate
has more significant impact. Larger velocity but lower REFERENCES
pressure at upstream, and lower velocity but larger
pressure at downstream take place at the vicinity of ANSYS Fluent 15.0 Theory Guide (2013).
leakage, which causes adverse pressure with a sharp
variation in pressure at this region. The pressure Barbosa, L.M.C., A.G.B. de Lima and S.R. Farias
change at the vicinity of leak location increases Neto (2012). Non-isothermal Transient Three-
approximately linearly with the leak size ratio when phase Flow (Heavy Oil, Water and Gas) in
the ratio is smaller than approximately 40%, Vertical Pipeline: the Effect of Leakage.
corresponding to the leak area ratio 16% and leak International Journal of Modeling and
mass quantity ratio 16-23%. The pressure change at Simulation for the Petroleum Industry 6 (2),
the vicinity of leak location is observed to be 23-31.
negligible, when the leak mass quantity ratio smaller
than approximately 0.07 ~ 0.13%, corresponding to Ben-Mansour, R., M. Habib, A. Khalifa, K.
the leak size ratio 5% and area ratio 0.25%. Note that Youchef-Toumi and D. Chatzigeorgiou (2012).
the pressure change at the vicinity of leak location Computational fluid dynamic simulation of
actually occurs over a definite distance but not at the small leaks in water pipelines for direct leak
point of leakage location. pressure transduction. Computers & Fluids 57,
110-123.
Choi, J., J. Shin, C. Song, S. Han and D. Park
(2017). Leak detection and location of water
pipes using vibration sensors and modified ML
prefilter. Sensor 17, 2104.
Costa D., I. Stoianov, D. Bulter, C. Marksimovic,
N. Graham and H. Ramos (2001, September).
Leak detection in a pipeline systems by inverse
transient analysis: from theory to practice. In
Proceedings of International Conference on
Computing and Control for the Water Industry,
Leiceter. U.K.
da Silva, H.V., C. Morooka., I. Guiherme, T. da
Fonseca and J. Mendes (2005). Leak detection
in petroleum pipelines using a fuzzy system.
Journal of Petroleum Science and Engineering
49, 223-238.
de Sousa1, J.V.N., C.H. Sodré, A.G.B. de Lima and
S.R. de Farias Neto (2013). Numerical
Analysis of Heavy Oil-Water Flow and Leak
Detection in Vertical Pipeline. Advances in
Chemical Engineering and Science 3, 9-15.
de Vasconcellos Araújo, M., S.R. de Farias Neto,
A.G.B. de Lima and F.D.T. de Luna (2014).
Fig. 12. Pressure gradient variation curves along Hydrodynamic Study of Oil Leakage in
centerlines at different XZ planes under conditions Pipeline via CFD. Advances in Mechanical
of the inlet mass flow rate 0.32kg/s, main pipe Engineering 6, 170-178.
length 3.35m, and discharge pipe diameter 5.2mm. Ge, C., H. Yang, H. Ye and G. Wang (2009). A fast
leak locating method based on wavelet
Due to an increase in the local pressure at the leak transform. Tsinghua Science and Technology
location, there is a sudden change in the pressure 14, 551-555.
gradient along the flow direction and the magnitude is
much larger than that of the pressure change there, Ishido, Y. and S. Takahashi (2014). A new indicator
representing the typical characteristic of the most for real-time leak detection in water
rapid change in the rate of the pressure change there in distribution networks: design and simulation
leak flows. From the summary of the maximum validation. Procedia Engineering 89, 411-417.
pressure gradient along flow direction, it is observed Jujuly, M.M. (2016). Computational Fluid
that the effect of the inlet mass flow rate is dominant Dynamics (CFD) based approach to
compared with that of the main pipe length. It is consequence assessment of accidental release
expected that the magnitude of the maximum pressure of hydrocarbon during storage and
gradient reduces with farther away from the leakage transportation. Master Thesis. Memorial
pipe, due to diminishment of the leakage effect. The University of Newfoundland, Newfoundland
present study verifies that simulation is an effective and Labrador, Canada.
and economic tool to investigate pipe flow leakage, so
as to help understand the physics of such flow Kam, S. (2010). Mechanistic modeling of pipeline

493
F Y. Zeng / JAFM, Vol. 12, No. 2, pp. 485-494, 2019.

leak detection at fixed inlet rate. Journal of Uppsala, Sweden.

Petroleum Science and Engineering 70, 145-
156. Puust, R., Z. Kapelan, D.A. Savic and T. Koppel
(2010). A review of methods for leakage
Kim, Y., K. Miyazaki and H. Tsukamoto (2008). management in pipe networks. Urban Water
Leak detection in pipe using transient flow and Journal 7, 25-45.
genetic algorithm. Journal of Mechanical
Science and Technology 22, 1930-1936. Shehadeh, M. and A. Shahata (2013). Modelling the
effect of incompressible leakage patterns on
Lay-Ekuakille, A., G. Vendramin, A. Trotta and P. rupture area in pipeline. CFD Letters 5 (4),
Vanderbemden (2009). STFT-based spectral 132-142.
analysis of urban waterworks leakage
detection. In Proceedings of the XIX IMEKO Vitkovsky, J., A. Simpson and M. Lambert (2001).
World Congress Fundamental and Applied Leak detection and calibration using transient
Metrology, Lisbon, Portugal, 6-11 September. genetic algorithms. Journal of Water
Resources Planning and Management-ASCE
Liggett, J. and L. Chen (1994). Inverse transient 126 (4), 262-265
analysis in pipe networks. Journal of
Hydraulic Engineering 120 (8), 934-955. Zeng, Y. and R. Luo (2017, May). Numerical
Analysis of Incompressible Flow Leakage in
Molina-Espinosa, L., O. Cazarez-Candia and C. Short Pipes. In Proceedings of International
Verde-Rodarte (2013). Modeling of Pipeline Technology Conference. Berlin.
incompressible flow in short pipes with leaks. Germany
Journal of Petroleum Science and Engineering
109, 38–44. Zhang, L., Y. Wu, L. Guo and P. Cai (2013).
Design and implementation of leak acoustic
Olivares, P.A.V. (2009). Acoustic Wave signal correlator for water pipelines.
Propagation and Modeling Turbulent Water Information Technology Journal 12, 2195-
Flows with Acoustics for District Heating 2200.
Pipes. PhD Thesis, Uppsala University,

494