ACOUSTIC WAVE ATTENUATION THROUGH BURIED

WATER MAINS

Abu Hena Muntakim, Ashutosh Sutra Dhar & Amgad Hussein

Department of Civil Engineering, Memorial University of Newfoundland, St. Johns, NL, Canada

ABSTRACT
This paper describes a laboratory facility developed at Memorial University of Newfoundland for the study of acoustic
wave propagation from leak noise through buried pipes and the results of acoustic emission testing in the new laboratory
facility. The facility is designed and developed to investigate the propagation of leak noise through water mains. The
developed test facility is used to investigate leak noise attenuation through an in-air and a buried pipe. The buried pipe is
backfilled with crushed stone. The laboratory study indicates that the leak noise attenuates during propagation of
acoustic wave. The attenuation is higher for the high frequency waves. The attenuation also increases with the increase
of flow rate through the pipe. Between the in-air pipe and the pipe buried in crushed stone, the attenuation is higher for
the buried pipe. For a distance of 0.6 m, the amplitude ratio (the ratio of the amplitude at two locations at a distance of
0.6 m and 1.2 m from the source, A0/A1) is 0.987 for in-air pipe and 0.93 for pipe buried in crushed stone at a frequency
of 2000 Hz and an inflow rate of 6.7 gallons per minute (25.36 litres per minute). An attenuation parameter is calculated
based on the amplitude ratio to be 0.032 dB/m for in-air pipe and 0.12 dB/m for the buried pipe. Using the experimental
results, a finite element modelling technique is developed for simulating the acoustic wave propagation and determining
the effective distance of sensor from the leak source for successful identification of the leak noise.

RSUM.
Cet article dcrit un laboratoire mis au point lUniversit Memorial de Terre-Neuve pour ltude de la propagation de
londe acoustique du bruit de la fuite par lintermdiaire de tuyaux enfouis et les rsultats de lmission acoustique stable
dans les nouvelles installations de laboratoire. Une installation dessai de laboratoire est conue et dveloppe pour
tudier la propagation du bruit de la fuite par lintermdiaire de canalisations deau. Linstallation dessais avancs est
utilise pour tudier lattnuation de bruit de fuite grce une teneur de lair et un tuyau enterr. La conduite enfouie est
remblay avec la pierre concasse. Ltude de laboratoire indique que le bruit de fuite attnue au cours de la
propagation dune onde acoustique. Lattnuation est plus leve pour les ondes haute frquence. Lattnuation
augmente galement avec laugmentation du dbit travers le tuyau. Entre le tuyau dans lair et le tuyau enterr dans la
pierre concasse, lattnuation est plus leve pour le tuyaux enfouis. Sur une distance de 0,6 m, lamplitude (le ratio de
lamplitude deux endroits une distance de 0,6 m et 1,2 m de la source, A0/A1) est 0.987 pour tuyau dair et 0,93 pour
tuyaux enfoui dans la pierre concasse une frquence de 2 000 Hz et un taux dafflux de 6,7 gallons par minute (25,36
litres / minute). Un paramtre attnuation est calcul selon le rapport damplitude 0,032 dB/m pour le tube dans lair et
0,12 dB/m pour le tuyaux enfouis. partir des rsultats exprimentaux, une technique de modlisation par lments finis
a t dvelopp pour simuler la propagation des ondes acoustiques et dterminer la distance effective du capteur de la
source de la fuite pour une identification russie du bruit de fuite.

1. INTRODUCTION complex and expensive. Literature on full-scale testing on

acoustic wave propagation is therefore limited. Juliano et
Acoustic is often used for inspecting, testing, evaluating al. (2013) tested a 304.8 m long buried steel water pipe
material and elements for discontinuities in continuum. for leak detection using acoustic emission technique.
Acoustic emission method is one of the popular methods They used a 305 mm diameter, welded steel pipeline.
used for detecting leak in water mains. In this technique, Vertical access tubes were used to gain access to 17
acoustic noise from leak is propagated through pipe that different locations along the pipe. In their tests, a range of
is received at two access points to the pipe. Successful leak rates from 15.2 mL/s to 16.6 mL/s was successfully
leak detection using this method depends on the detected when the maximum sensor distance was 65.5 m.
propagation characteristics of the acoustic noise. Noise However, for a sensor distance of 78 m or above, the
can be attenuated while propagating through the pipeline. acoustic emission method was not successful, potentially
Thus, if the sensors are not located close enough, the due to the attenuation of the sound wave. The attenuation
leaks may remain undetected. Understanding the characteristics were however not investigated.
propagation behaviour of acoustic wave is therefore very Pollock and Hsu (1982), Lee and Lee (2006) and
important for determining the distance over which the Thenikl et al. (2012) studied attenuation characteristics of
method would be successful. acoustic wave propagation during leak detection using
In order to develop a better understanding of the amplitude ratio. In the method of Thenikl et al. (2012),
wave propagation through buried pipe, an experimental noise is recorded at two different locations along the
investigation under controlled conditions is required. A full length of the pipe and ratio of the averaged amplitude was
scale pipe test would be ideal for the experimental taken to determine the exponential law of attenuation. The
investigation. Full-scale experiments are however attenuation characteristic is given by Equation 1.
1 1 test bed and pipe bed to facilitate draining of accumulated
= [1] water.
2 2

Here, A1 and A2 are the recorded amplitudes at the

respective distances of d1 and d2 from the noise source
and is the attenuation parameter. Thenikl et al. (2012)
used cross-correlation method to find the distances to
leak from two points where the acoustic sensors were
placed. The amplitudes measured at these points are
then used to determine . They calculated the value of
as 0.6 dB/m for a 100 mm diameter pipe. This parameter
is then used to determine the leak location for other two
cases. In those cases, the diameter of the steel pipe (a) Cross-section
without insulation were 100 mm and 300 mm and the AE
sensor distance were 52 m and 46 m, respectively.
For a better understanding of attenuation of leak
noise, a new laboratory facility was developed and the
experiments were conducted in the facility to study leak
noise propagation through water pipe. Experiments were
carried out to investigate the attenuation characteristics of
leak noise in fluid filled pipe placed in an open air and a
buried condition. Detailed test procedure and tests are (b) Longitudinal section
provided in Muntakim et al. (2017 a). This paper includes Figure 1. Schematic views of the test bed
description of the test procedure and the summary of the
test results presented in Muntakim et al. (2017 a). The
test results are used to develop Finite Element (FE)
modelling technique for acoustic wave propagation and
attenuation of leak noise. Finite element modelling is then
extended for the determination of the distance up to which
the acoustic emission leak detection method can
successfully be used.
Test bed
2. DESIGN OF LABORATORY FACILITY

A new test bed has been designed and developed to Gravel

investigate acoustic wave propagation through buried
pipes. The test setup is capable of maintaining a
continuous water flow in a buried pipe and measure the
flow of the water.

2.1 Test Bed Figure 2. Test bed for buried pipe testing

The test bed has been designed to house a typical water 2.2 Backfill Material
main of 152 mm diameter. The length of the test bed was
to accommodate a 3 m long pipe. About 0.3 m clear Gravel is used to backfill the pipe in the test bed. Gravel is
spacing was provided to accommodate fittings at two generally used for backfilling the pipes in the field,
ends of the pipe segment. Considering the clear spacing, particularly in the province of Newfoundland and Labrador
the length of the tank was chosen to be 365.75 cm in Canada. A preliminary test was conducted earlier using
(Figure 1(b)). The width of the soil layer surrounding the a different facility where sand was used besides the pipe
pipe segment was chosen to be approximately two times (Muntakim et al. 2016). However, the sand particles were
of the diameter of the pipe. The width and depth of the found to be washed away by the water escaping from
test bed is thus 762 mm (Figure 1a). The test bed is made leak. To assess the particle size distribution of the gravel,
of aluminium, which is a cost effective material. sieve analysis of the backfill material was conducted
Aluminium is also less likely to have corrosion. The wall according to ASTM C136 standard test method. The
thickness of the bed is designed to carry the lateral load results are presented in Figure 3. Figure 3 reveals that
from the soil-pipe system. The thickness of the tank wall the gravel particle dimeters range from about 0.12 mm to
is 4.76 mm. Horizontal stiffeners are used at the top of the 40 mm.
facility.
Water escaping from the pipe through the leak during 2.3 Pressure and Flow Measuring Facility
tests is expected to be accumulated in the test bed. A
facility is designed to drain out the accumulated water. A A multipurpose pressure gauge is attached at both ends
fountain pump is placed inside the tank between wall of of the test pipe to measure the inflow water pressure and
outflow water pressure. The acoustic emission measurement points are at
Inline flowmeters are attached at the inlet and the 600 mm (point L2) and 1200 mm (point L3) from the leak
outlet of the pipe to measure the inflow and outflow rate of in the flow direction (Figure 5). Two acoustic sensors from
water flow. The flowmeters have capacity of 1-10 GPM Physical Acoustics with a frequency bandwidth of 1 to 30
(3.8-38 Liter per min). kHz and resonance frequency of 20 kHz are used in this
research. High sensitivity and low-noise input capabilities
make this sensor suitable for recording acoustic signals
120
from leaks in water main.
100 The data acquisition system consists of a Data
acquisition (DAQ) module and a personal computer
Percent finer

80 equipped with LabVIEW Signalexpress software. The NI

9218 is connected with personal computer using USB
60 cable. The sensors are connected to the system using NI
9982D screw terminal block.
40

20

0
100 10
Perticle size, mm
Figure 3. Particle size distribution

2.4 Water Supply

A 200 litre capacity water tank house a water pump that

supply continuous water flow in the pipe within the test
bed. The pump has the maximum flowrate capacity of
3180 GPH (200.6 litres per hour) and the power capacity
of 1/3 HP (248.6 watt). The water tank is filled and is
connected to a city water line for continuous supply of
water. There is a float switch to turn off the pump
automatically if the water level goes to a certain low level
in the tank.

2.5 Test Setup

Figure 4. Physical connections of the test pipe on test bed
The laboratory setup includes the test bed that
accomodates a test pipe. The test setup includes a
prepared pipe sample, two acoustic sensors, a data
acquisition system and a facility to maintain continuous
flow of water through the pipe.
A pipe sample is prepared that consists of a 152.5
mm diameter ductile iron pipe segment of approximately 3
m long (Figure 4). The pipe segment are capped using
two steel plates welded at the ends. Two nipples with 12.7
mm diameter and 76 mm length are connected at the
ends to facilitate water circulation and the attachment of a
flow control mechanism. An artificial leak of 4.75 mm Figure 5. Acoustic signal measurement points
diameter was created on the pipe wall at approximately
700 mm from one of the ends of the pipe. 3. LABORATORY TESTING
The pipe sample is placed in the test bed and
connected to the facility developed to maintain water flow The test program was designed to investigate the effect of
through the pipe in a loop (Figure 4). Water from the tank pipe surroundings on the attenuation behaviour of
enter to the pipe from one end which is collected from the acoustic leak noise through buried water mains. Tests
other end back to the water tank. During the test, the were conducted for a test pipe in the air and the pipe
water flow rates and pressures are measured using inline buried in crushed stone. Detail test procedure and
flow-meter and pressure gauge, respectively. The analysis of results are available in Muntakim et al. (2017
difference between the inflow water and outflow water, a). Some of the results are included here to the extent
measured by the flowmeters, provide the water loss relevant for the development of FE model presented in
through leak. Two acoustic sensors are placed at two this study.
known locations to measure acoustic signals. The laboratory investigation program included
recording of time domain acoustic emission data at the
two sensors during the tests. Tests were conducted at indicating that the inflow rate affect the propagation
different flow rates of water through the pipe. The behavior of the acoustic wave. The rate of increase of
recorded data are analysed using MATLAB software. The attenuation is higher for buried pipe than in-air pipe.
MATLAB program converts the data to sound wave and
1
store in the PC. The sound wave is then analysed to
obtain the frequency spectrum. From the frequency
spectrum, the ratio of the amplitude, A1, of the signals 0.99
measured at location L3 to the amplitude, A0, of the
signals measured at the location L2 is calculated for each 0.98
frequency within the range of dominant frequencies.

A1/A0
Figure 6 shows the amplitude ratio (A1/A0) for different
0.97 Inflow 6.8 GPM, Leak 1.2 GPM
inflow rates for in-air pipe. In Figure 6, the ratio of A1/A0
Inflow 7.4 GPM, Leak 1.2 GPM
is lower for higher inflow rate of water, indicating that the Inflow 8.2 GPM, Leak 1.0 GPM
attenuation of leak noise is higher for higher inflow water 0.96 Inflow 8.8 GPM, Leak 4.8 GPM
rates. To study the rate of change of attenuation with flow Inflow 9.7 GPM, Leak 4.5 GPM
rates, the amplitude ratios are plotted against the inflow Inflow 6.8 GPM, Leak 6.0 GPM
rates of water in Figure 7. Figure 7 shows the amplitude 0.95
500 1000 1500 2000 2500 3000
ratio (A1/A0) for three different frequencies. In Figure 7, Frequency (Hz)
the amplitude ratio is almost independent on the Figure 6. Amplitude ratio for different inflow rate for in-air
frequencies considered and decreases with the increase pipe
of inflow rate. The rate of decrease increases beyond an
1
inflow rate of 8.4 gallons per minute (31.8 litres per Frequency 500 Hz
minute) of flow rate. At a flowrate of 6.8 gallons per Frequency 1000 Hz
minute (25.74 litres per min), the amplitude ratio is about Frequency 3000 Hz
0.995
0.993. At a flowrate of 9.7 gallons per minute (36.71 litres
per min), the amplitude ratio is 0.981. The amplitude ratio
A1/A0

change per gallons-per-minute is 4.13e-3. Based on the 0.99

amplitude ratios, the attenuation parameter () is
calculated to be 0.012dB/m and 0.032dB/m, respectively.
These parameters are much less than the parameter 0.985
calculated in Thenikl et al. (2012).
Similar results were obtained for the buried pipes
(Muntakim et al. 2017 a). However, the rate of decrease 0.98
was higher beyond 7.8 gallons per minute (29.5 litres per 6 7 8 9 10
Inflow rate (GPM)
minute) of flow rate. Based on the amplitude ratios, the
Figure 7. Amplitude ratio for different frequencies for in-air
attenuation parameter () is for buried pipes were
pipe
calculated to be 0.045 dB/m and 0.12dB/m, respectively,
which are again lower than the parameter calculated in
4. FINITE ELEMENT MODELLING
Thenikl et al. (2012).
The surrounding soil appears to cause higher
Numerical simulation of leak noise propagation through
damping on sound energy, resulting in the higher
the pipe wall is essentially performed by simulation of
attenuation of the acoustic wave for the buried pipe.
wave propagation through fluid filled pipeline as an
Figure 8 plots the frequency spectrum of the acoustic
acoustic medium. Acoustic waves are longitudinal waves
wave for the in-air and buried pipes measured at
which propagate by means of adiabatic compression and
corresponding locations. The frequency spectrum trend
decompression. In ABAQUS, acoustic medium can be
are similar for location L2 for both case (Figure 8a). Since
used to model the sound propagation problems (Dassault
the location L2 is very close to the source of the wave
Systemes 2011). Generally, acoustic medium is assumed
(leak), attenuation was negligible for the both cases. As a
as elastic medium, generally like as fluid. In acoustic
result, the frequency spectrum were similar. However, for
medium, stresses are purely hydrostatic. The pressure is
sensor at location L3, the amplitudes for the buried pipe
proportional to volumetric strain. The volumetric drag
are lower than those for in-air pipe except at the
coefficient has effect on the equilibrium equation of
frequencies of 4500 Hz to 5500 (Figure 8b). The
acoustic medium. The equilibrium equation for small
amplitudes are particularly less at frequencies above
motions of a compressible, adiabatic fluid is taken to be:
6000 Hz.
The amplitude ratio is compared for in-air pipe and
buried pipe for different inflow rate at different frequencies + (, ) + (, ) = 0 [2]

in Figure 9. The change in amplitude ratio is calculated to
-3 -3
be 4.13e per gallons per min (1.1e per litres per min) of Here, p is the excess pressure in the fluid, x is the spatial
inflow rate for in-air pipe while the change in amplitude position of the fluid particle, is the fluid particle
-3 -3
ratio is 15.92e per gallons per min (4.2e per litres per velocity, is the fluid particle acceleration, f = the
min) of inflow rate for buried pipe. Figure 9 reveals that
attenuation becomes higher for higher inflow rate, density of fluid, = the volumetric drag, i is independent
field variables such as temperature, humidity of air or An incident wave is considered on one end and
salinity of water. propagation of this wave through the conduit is analyzed
at some reference points. A total wave formulation is used
for a nonlinear acoustic medium.
100
Pipe placed in the open air The propagation of axisymmetric acoustic wave-
Pipe buried in the crushed stone modes in the fluid-filled pipe in the air is simulated using
80 FE analysis. Damping behaviour of ductile iron and water
are considered in this simulation. The effect of the air is
not considered for simplification of the model.
Magnitude

60
ABAQUS/Explicit (version 6.11) is used in the
analysis with default viscosity parameters in the software.
40 The default viscosity parameters include linear bulk
viscosity parameter of 0.06 and quadratic bulk viscosity
20 parameter of 1.2. Pipe and water are modeled using
AC3D8R element from ABAQUS library. AC3D8R is an
8-node linear brick element with reduced integration and
0 hourglass control.
0 2000 4000 6000 8000 10000
Frequency (Hz) The sound wave speed of 1200 m/s is used based on
(a) Sound frequency comparison at location L2 the results of field investigation (Muntakim et al. 2017 b).
The incident wave frequency of 3000 Hz is considered.
140
Pipe placed in the open air Based on the speed and the frequency, the wavelength is
120 Pipe buried in the crushed stone calculated to be 0.4 m. Element length is thus chosen as
0.04 m to provide 10 elements in one wavelength. The
100 dilatational wave velocity speed is calculated as, Cd =
(K/p) = 4145 m/s. The stable time increment is taken as
Magnitude

80 -7
5E second which is less than the value of characteristic
length divided by dilatational wave velocity (ABAQUS
60
2005). Time step is chosen long enough to separate the
40 transient and stable wave propagation. The time step
chosen is 0.01 second. A pipe with 0.15 m diameter and
20 1.5 m of length is modelled. A mesh sensitivity analysis
was performed to identify the optimum mesh size. If the
0 element number increase from 100 to 250 (corresponding
0 2000 4000 6000 8000 10000
Frequency (Hz) to the mesh size of 0.004 m to 0.0015 m), the amplitude
b) Sound frequency comparison at location L3 ratio was found to change by only 0.7%, which is
Figure 8. Sound frequency comparison for pipe placed in insignificant. Therefore, 100 element in one wavelength is
the open air and pipe buried in the crushed stone chosen for the analysis. The corresponding mesh size is
0.004 m. Total number of elements for the model is
1
24000. The FE is shown in Figure 10. The pipe and water
was modelled as acoustic medium. The density and bulk
3 9
modulus are taken as 1000 Kg/m and 2.15e Pa,
0.98 3 9
Buried pipe respectively, for water and 7100 Kg/m and 122e Pa,
respectively, for the pipe.
In-air pipe
0.96 Finite element modelling of acoustic wave
A1/A0

propagation in water pipeline is generally performed by

0.94 In-air pipe, Frequency 500 Hz incident plane wave. Graf et al., (2014) described that at
In-air pipe, Frequency 1000 Hz low frequencies the fundamental mode of wave is the
In-air pipe, Frequency 3000 Hz
0.92 Buried pipe, Frequency 2000 Hz
plane wave. For the particular problem considered here,
Buried pipe, Frequency 3000 Hz plane wave is applied at one end of the pipeline. The
Buried pipe, Frequency 10000 Hz plane incident wave is applied at the pipe cross section
0.9
6 6.5 7 7.5 8 8.5 9 9.5 10 with varying frequency ranging from 500 to 3000 Hz to
Inflow rate (GPM)
simulate the attenuation behaviour. In Figure 11, the
incident plane is shown by purple color. The opposite
Figure 9. Comparison of amplitude ratio for in-air pipe and surface of the incident surface is assigned as non-
buried pipe for different frequencies reflecting boundary. The amplitude of incident wave of
0.04 is used as recorded in the laboratory experiments.
The constitutive behaviour of the acoustic medium The water filled pipe is modeled as acoustic medium
(usually a fluid) described in ABAQUS 6.11 as to simulate the leak noise propagation. The material
parameters required for water and pipe (as acoustic

= (, ) [3] medium) are density and bulk modulus. Typical values of

these parameters for water and ductile iron are used as
Here, Kf is the bulk modulus of the fluid. stated above. Modeling of attenuation of wave requires
volumetric drag coefficient. However, no accepted value Using the parameters discussed above, the acoustic
of this parameter is published in the literature. A study is pressure are calculated at two reference points, which are
therefore conducted to identify suitable parameters 600 mm and 1200 mm from the incident plane (similar to
applicable for the test condition analysed. those measured during the laboratory test). The two
recorded signals have time lag due to travel time required
for the acoustic wave to travel from one point to the other.
The recorded signals are then analyzed using MATLAB
code. The time lag was obvious in the recorded signals,
as the signal amplitudes initiated at two different times at
these points. The time-lag was then calculated as the
difference of the time for signal initiations. Both signals
are then initialized to start at t=0 sec, for the frequency
spectrum analysis. Fourier analysis have been performed
to obtain the frequency spectrum. The amplitudes are
obtained from FFT analysis and used to calculate the
amplitude ratio for those two reference points. These
a) Mesh of the pipe results are compared with those from the experiment in
Figure 12. As seen in the figure, at 500 Hz of frequency,
the amplitude ratio is 0.998 from the FE simulation and
0.986 from the experiment. At 3000 Hz frequency, the
amplitude ratio is 0.991 from the FE simulation and 0.986
from the experiment. The inflow rate was 8.8 gallons per
min (33.3 litres per min) and the outflow was 4.8 gallons
per min (18.2 litres per min) and the leak rate was 4
gallons per min (15.1 litres per min) for the experiment
considered here.
1

0.99

(b) Three dimensional view

Figure 10. Finite element mesh 0.98 Case a
A1/A0

Case b
Experimental Result
0.97

0.96

0.95
500 1000 1500 2000 2500 3000
Frequency (Hz)
Figure 12. Amplitude ratio for comparison with assumed
volumetric drag

In Figure 12, the amplitude ratios from FE analysis

are higher than those from the experiment. Thus, a higher
volumetric drag is required for the water to simulate the
Figure 11. Surface of incident wave application (purple experimental results. The volumetric drag value is then
4
color) doubled to 26400 Ns/m for the water. The volumetric
drag for the pipe is taken as 0 for case a and half of the
4
Chatoorgoon and Zhou (1995) used a volumetric value of volumetric drag of water (i.e., 13200 Ns/m ) for
4
drag of 13200 Ns/m for water for a study using ABAQUS. case b, as before.
Based on this value, the volumetric drag of 13200 Ns/m
4 Figure 13 illustrates the results after increasing the
for water is chosen in this study. To study the effect of the volumetric drag of water and pipe (first revision). The FE
volumetric drag of pipe, two different values are first and experimental amplitude ratios approaches closer with
chosen for two cases for a parametric study. In case a, a higher volumetric drag of water. At 500 Hz frequency,
the volumetric drag for pipe is chosen as 0 (as the amplitude ratio is 0.992 from FE simulation whereas
Chatoorgoon and Zhou, 1995). In case b, volumetric drag the ratio is 0.986 from the experiment. At 3000 Hz
of pipe is chosen as half of the volumetric drag of water frequency, the amplitude ratio is 0.988 from the FE
4
(i.e. 6600 Ns/m ). Volumetric drag for steel is expected to simulation and 0.986 from the experiment. Further
be less than the volumetric drag of water. increase in the volumetric drag is then used to improve
the performance of the FE simulation.
 The volumetric drag value of water is then increased conducted for 250 m long pipe to simulate acoustic wave
4
to three times of the value of 13200 Ns/m (used in attenuation over the length. A frequency of 500 Hz is
Chatoorgoon and Zhou, 1995). The value of volumetric considered.
4
drag for water is chosen to be 39600 Ns/m . The Due to meshing constrain of this large model, 10
volumetric drag for the pipe are used as in the previous elements per wave length was chosen. The element
cases. length along the length of the pipe is 0.25 m. The element
size in radial direction is the same as those used in Figure
10(a). The time step of 0.2 second is used, and time
1
increment is calculated using similar procedure,
discussed above.
0.99 The distance of the sensors for effective leak
detection (effective distance) would, however, depend on
0.98
the amplitude of leak noise at source. During laboratory
tests, the amplitude of acoustic wave was measured to be
A1/A0

Case a
Case b
0.04. A set of noise source amplitude is used to identify
0.97 the effect of source signal amplitude on the effective
Experimental Result
distance. The noise source magnitudes of 0.04, 0.06 and
0.96 0.12 are considered.
Figure 15 shows the attenuation of acoustic wave
with distance. The ambient noise level is shown in the
0.95 figure in red line denoting the level under or at which the
500 1000 1500 2000 2500 3000
Frequency (Hz) sensor cannot identify leak noise signals. This level
Figure 13. Amplitude ratio with first revision of volumetric represents the cut-off level. When the magnitude of noise
drag source (potentially leak noise) is 0.04, the noise can be
identified up to a distance of 140 m from the source.
Figure 14 illustrated the results after increasing the When the magnitude of the noise source is 0.06, the
volumetric drag value for water. In Figure 14, at 500 Hz noise can be identified up to 170 m. For leak noise with
frequency, the amplitude ratio is 0.986 from FE magnitude of 0.12, noise magnitude remains above the
simulation, which is the same as the value obtained from ambient noise up to a distance of 210 m. Thus, the
the experiment (i.e. 0.986). At 3000 Hz frequency, the acoustic method can be effectively used up to a sensor
amplitude ratio is 0.985 from FE simulation and 0.986 distance of 140 m, 170 m and 210 m for leak noise with
from experiment, which are very close to each other. magnitudes of 0.04, 0.06 and 0.12, respectively.
1
Case a
0.14
Case b Ambient Noise Level
0.99 Experimental Result Noise source of magnitude 0.06
0.12
Noise source of magnitude 0.12
0.1 Noise source of magnitude 0.04
Sound Amplitude

0.98
A1/A0

0.08
0.97
0.06

0.96 0.04

0.02
0.95
500 1000 1500 2000 2500 3000
Frequency (Hz) 0
0 50 100 150 200
Figure 14. Amplitude ratio from experiment and simulation Distance from source (m)
Figure 15. Sound wave attenuation over length for
In the three different assumptions above, the different noise source magnitude
volumetric drag value for the pipe is taken as zero and
half of the value of water in different case. Figures 12, 13 5. CONCLUSION
and 14, reveal that the volumetric drag of pipe has
relatively small effect on the amplitude ratio while A laboratory test facility has been developed to
volumetric drag of water significantly controls the acoustic investigate the propagation of acoustic wave through
wave propagation. Based on these results, the volumetric water mains. The developed test facility is used to
4
drag of water is taken as 40000 Ns/m . Using this investigate leak noise attenuation through an in-air and a
parameter, a study has been conducted to identify the buried pipe. The buried pipe is backfilled with crushed
maximum sensor distance over which the acoustic stone. The laboratory study indicates that the leak noise
sensors should be placed to identify the leak noise. It is to attenuates during propagation of acoustic wave. The
be noted that sensor may miss the leak noise if it is attenuation increases with the increase of flow rate
located too far from the source (leak). FE analysis is through pipe. Between the in-air pipe and the pipe buried
in crushed stone, the attenuation is higher for the buried Muntakim, A. H., Dhar, A. S., & Dey, R. (2017 b).
pipe. For a distance of 0.6 m, the ratio (A1/A0) is 0.987 for Interpretation of Acoustic Field Data for Leak
in-air pipe and 0.93 for pipe buried in crushed stone at a Detection in Ductile Iron and Copper Water-
frequency 2000 Hz and inflow rate of 6.7 gallons per Distribution Pipes. Journal of Pipeline Systems
minute (25.36 litres per minute). An attenuation parameter Engineering and Practice, 05017001.
() is calculated to be 0.032 dB/m for in-air pipe and 0.12 Pollock, A. A., & Hsu, S. Y. (1982). Leak detection using
dB/m for buried pipe at the inflow rate of 6.7 gallons per acoustic emisson. Journal of Acoustic Emission, 1,
minute (25.36 litres per minute). 237-243.
A FE modelling technique is developed to model the Thenikl, T., Vallen, H., & Aljets, D. (2012). Location of
acoustic wave propagation observed in the laboratory Leaks in Liquid Filled Pipelines under Operation. 30th
test. The FE model is then used to calculate the sensor European Conference on Acoustic Emission Testing &
distance from leak source for effective identification of 7th International Conference on Acoustic Emission
leak noise. For leak noise with magnitudes of 0.04, 0.06
and 0.12 at the source, the effective sensor distances are
found to be 140 m, 170 m and 210 m, respectively.
Research is currently underway to develop FE modelling
technique for buried pipe.

6. ACKNOWLEDGEMENT

The authors gratefully acknowledge the financial supports

for this research provided by Research and Development
Corporation of Newfoundland and Labrador, Natural
Sciences and Engineering Research Council of Canada
through its Collaborative Research and Development
Grants, the City of Mount Pearl in the province of
Newfoundland and Labrador and Civil Engineering
Department at Memorial University of Newfoundland.

7. REFERENCES

ABAQUS (2005) Lecture on ABAQUS/Explicit: Advanced

Topics, Overview of ABAQUS/Explicit, ABAQUS Inc.
2005
Chatoorgoon, V., & Zhou, R. (1995). Studies on acoustic
modelling techniques for CANDU reactors. 16, Annual
conference of the Canadian Nuclear Society,
Saskatoon, SK (Canada) 4-7 June 199
Dassault Systemes (2011) Analysis user manual.
ABAQUS 6.11, Dassault Systemes Simulia Corp.,
Providence, RI, USA.
Graf, T., Gisler, T., Sollberger, P., & Schlli, O. (2014).
Acoustic Wave Propagation in Water Filled Buried
Polyethylene Pipes. In Comsol Conference,
Cambridge.
Juliano, T. M., Meegoda, J. N., & Watts, D. J., (2013).
Acoustic emission leak detection on a metal pipeline
buried in sandy soil, Journal of Pipeline Systems
Engineering and Practice, 4(3), 149-155.
Lee, M. R., & Lee, J. H. (2006). A study on characteristics
of leak signals of pipeline using acoustic emission
technique. In Solid State Phenomena, Trans Tech
Publications, Vol. 110, pp. 79-88.
Muntakim, A.H., and Dhar, A. S. (2016). Laboratory
investigation of acoustic noise from leaks in ductile
iron water pipe. Annual Conference of the Canadian
Society for Civil Engineering, June 1-4, 2016, London,
ON, Canada
Muntakim, A. H., Dhar, A. S., & Thodi, P. (2017 a).
Experimental Investigation of Acoustic Wave
Attenuation through Ductile Iron Water Pipes. Under
review.