See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/303185622

A Comparative Study of Wireless Accelerometer Sensors for Water Pipeline

Leakage

Conference Paper · October 2015

CITATIONS READS

2 11,670

3 authors:

Mohd Ismifaizul Mohd Ismail Rudzidatul Akmam Dziyauddin

Universiti Teknologi Malaysia Universiti Teknologi Malaysia
22 PUBLICATIONS   168 CITATIONS    83 PUBLICATIONS   381 CITATIONS   

SEE PROFILE SEE PROFILE

N.A. Ahmad
Universiti Teknologi Malaysia
49 PUBLICATIONS   480 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Master Project View project

Power Allocation and Resource Allocation in Non Orthogonal Multiple Access (NOMA) View project

All content following this page was uploaded by Mohd Ismifaizul Mohd Ismail on 15 May 2016.

The user has requested enhancement of the downloaded file.

 A Comparative Study of Wireless Accelerometer
Sensors for Water Pipeline Leakage

Rudzidatul Akmam Dziyauddin, MIM Ismail Noor Azurati Ahmad @ Salleh

UTM Razak School of Engineering and Advanced School of Advance Informatics System
Technology Universiti Teknologi Malaysia
Universiti Teknologi Malaysia Kuala Lumpur, Malaysia
Kuala Lumpur, Malaysia [email protected]
[email protected], [email protected]

Abstract—An acoustic leak detection is found only feasible for broadly categorised into three: (i) a software-based, (ii) a
the metal pipes and has a limitation for the plastic pipeline. biological, and (iii) a hardware-based. Previous studies have
This paper investigates a vibration technique using demonstrated that the hardware-based method has more
accelerometer sensors, namely, MPU6050 and ADXL335 for advantages than the others, which is the scope of this paper.
an Acrylonitrille Butadience Styrene (ABS) pipe. We have
The hardware-based technique detects the leak from the
built our water pipeline testbed to perform the experimental
work. The Arduino controller board is used to process the exterior of the water pipeline using either visual observation
vibration data from the accelerometer sensors and transmit or appropriate equipment. The main concern regarding the
(Tx) the data over wireless ZigBee networks every ten hardware-based is high cost involved. Since the world has
milliseconds. The water pipeline is examined based on three emerged to sensors technology in which can offer low cost,
cases, (i) no pipe leakage, (ii) a l mm, and (iii) a 3mm size of using sensors can be the answer to detect the leaking pipes
leaking hole. In our experiments, the water pressure is nearly real-time.
constant at 1 kgf/cm2, which indicated the pressure level for the An acoustic method is one of the recent hardware-based
water distribution of domestic. However, the distance between techniques to detect and locate the leak based on the sound
the sensor and the leaking point is varied between 0.5 m up to
of noise generated from the leaking water pipeline. This
2.0 m. The results have shown that the ADXL335 is able to
distinguish the leakage and no leakage cases across all the method has demonstrated on the metal pipeline, but it has a
ranges. Nevertheless, the ADXL335 is failed to detect the limitation for the plastic one. Due to viscoelastic, the plastic
severity of the leaking hole sizes ( 1-mm and 3-mm) except at pipe is made of a soft material and is more elastic; it will
the shortest distance of 0.5 m. In contrast, the MPU6050 can absorb the sound noise generated from the leak that results
detect all the three cases at the distance of 0.5 m and 1.5 m. to the weakening of the sound wave [2]. Hence, an acoustic
The variations in terms of frequencies for all the three cases technique is not doable to detect the leaking from the plastic
are observed despite not that significant for the MPU6050. The pipeline. Furthermore, the disadvantages of the acoustic
results show that the MPU6050 is more likely can detect the method are the requirement of the contact location and the
severity of the leaking, particularly between small leaks and
quiet leak is hard to be identified. This motivated us to
big leaks, but not for the ADXL335. This is because the
ADXL335 demonstrated similar frequencies for the 1-mm and consider another hardware-based technique called vibration
3-mm leakage across the ranges used in this study. using wireless accelerometer sensors that is likely feasible
for the plastic water pipeline.
Keywords- water pipeline leakage, vibration data, The objective of this study is to investigate and analyse
accelerometer sensor, MPU6050, ADXL335, FFT two types of accelerometer sensors in detecting the
conditions of the ABS pipeline over ZigBee network (i.e.
I. INTRODUCTION wireless). In this paper, 6 DOF sensors – MPU6050 and
A robust water pipeline system is vital to support ADXL335 are used, and the sensors are examined based on
human’s daily life activities. This can be ranging from three states of the pipe: i) no leak (normal), ii) a 1 mm size,
domestic to manufacturing purposes. To produce a and iii) a 3 mm sizes of leaking holes. We have built a water
sustainable water supply is still a big issue in most pipeline testbed to conduct our experimental work, and the
developing countries. A primary reason is due to the data is analysed offline based on frequency domain using
condition of the ageing water pipeline. Nevertheless, the Fast Fourier transform (FFT).
leak may also occur with regards to corrosion, excessive The system design and architecture is explained in
pressure resulting from operational error and rapid closing Section II. The water pipeline monitoring system is
or opening the pipe valves. Thus, many engineering described in Section III. The experimental setting is
researchers[1]have invented a number of different explained in Section IV. The results and discussion are
techniques in detecting the leaking water pipelines. discussed in Section V. Section VI concludes this paper.
Water pipeline leakage detection methods can be
 Both sensor are giving different method output data,
measurement range, Analogue to Digital Converter (ADC),
II. WATER PIPELINE TESTBED sensor sensitivity and zero g output.
The purpose of a development of water pipeline testbed
system is to attach the accelerometer sensors on the ABS TABLE 1: Comparison performance of accelerometer sensor between
pipes and to examine whether those sensors based on MPU6050 and ADXL335
vibration can distinct the states of normal and abnormal MPU6050 ADXL335
pipes. A water pipeline testbed system has been designed
Data output X, Digital (1024 byte FIFO
and developed in a soil laboratory at the civil facility on the Analogue
Kuala Lumpur campus at Universiti Teknologi Malaysia Y and Z-axis buffer)
(UTM) in Malaysia. Figure 1 illustrates the system Measurement
architecture of the water pipeline monitoring system used in Up to ±16g ±3g
range
our work. The hardware design and setup of the wireless
Analogue to
water pipeline system using MPU6050 are detailed in [3].
Our water piping testbed was designed using an inch Digital 16-bit (internal sensor 8 to 10-bit (microcontroller
diameter of ABS pipe with a length of approximately 10 Converter converter) converter)
meters. The testbed water pipeline has a water pump to (ADC)
provide water flow and pressure in the pipeline, and the end Sensitivity Up to 16, 384 LSB/g 300 mV/g
of the pipeline is connected to two ball valves. The water is
Zero output
circulated in the test bed system on command using a water
pump that support a pressure of 0.6 to 1.2 kgf/cm2. The and ±50 mg 1.35 - 1.65V
function of the valve is to control the water pressure and ±80 mg 1.2 - 1.8V
also the water flow in the pipeline system. Two acceleration
sensors, namely, MPU6050 and ADXL335, as illustrated in
Figure 2, are timely attached at the ABS pipes to collect the
vibration data via ZigBee network. Due to sustainability, the
water system recycled the water intake during the
experiments conducted. The vibration data are analyzed
using the signal analysis method on detecting the leaks and
determining the sizes of the leaks. Several snapshots of our
water pipeline testbed are depicted in Figure 3.

Figure 3: Real Simulation Pipe system for testing method [3]

III. MECHANICS FLUID

Water molecules averagely travelled in the main
direction of water flow, as depicted in Figure 4. However,
many molecules collided against the pipe wall. According to
the first law of thermodynamics, certain amount of the
Figure 1: Real Simulation Pipe system for testing method [3] kinetic energy produced by the molecules will dissipate to
heat as the turbulent eddies, but most of the kinetic energy
will convert into potential energy in the form of pressure
[4]. The pipe, in turn, deforms converting potential to
kinetic (during deformation) and back to potential
deformation complete. The elasticity of the pipeline material
applies a restoring force. The vibration in the pipeline
resulted from the energy conversion cycle is proportional to
(a) (b) the average water flow rate within the pipeline[5].
Figure 2: (a) MPU6050, and (b) ADXL335 Therefore, the vibration of the pipes is said in parallel with
the direction of the water flows. When the triple axes
accelerometer sensors are used to measure the vibration, the
Table 1 indicates that comparison performance of the
x-axis is often a preferable choice in collecting the vibration
accelerometer sensors between MPU6050 and ADXL335.
data subject to the position of x-axis is horizontal with the V. RESULTS AND DICUSSION
ABS pipes [6]. Figures 5(a) to 5(d) show the Fast Fourier Transform
(FFT) analysis for two kind of accelerometer sensors,
namely, MPU6050 and ADXL335 across various ranges.
At the range of 0.5m, both sensors demonstrated that the
frequencies of the leakage are smaller than that of the non
leakage, as can be seen in Figure 5(a). Because the sensor is
very closed to the leaking point, the frequencies can be
distinguished between the three cases (no leak, a 1mm and 3
Figure 4: Microscopic view of the water flow in a pipeline [6] mm of leaking holes). Nevertheless, the MPU6050 achieved
greater magnitude than the ADXL for each case, which
indicated the MPU6050 offers high sensitivity.
IV. EXPERIMENTAL SETTING Figure 5(b) shows approximately at 1.0 m, ADXL335
To investigate a water pipeline monitoring based on the has a significant difference in terms of frequency between
vibration data, three cases are considered; (i) No pipe leak, the normal pipe, 25.08Hz, and the abnormal pipe, 24.28Hz
(ii) a 1-mm, and (iii) a 3-mm sizes of leaking holes. Table 2 and 24.0Hz for the 1-mm and 3-mm of leaking holes. On the
summarizes the system parameters used in our testbed. A other hand, the MPU6050 has difficulties to detect the
high pressure of ABS pipe, which is an inch of diameter and variation of the frequencies across the three cases. As seen
approximately a 10-meter long, is used to develop our water in Figure 5 (c), at the distance of 1.5 m, both sensors show
pipeline testbed. Because the water pressure at the the frequencies disparity between the normal and abnormal
distribution to the domestic is 1.0 kgf/cm2, this pressure level of the pipe conditions. However, the ADXL335 is unable to
is selected in this paper for all the three cases. The distinct the sizes of the leaking between 1 mm and 3 mm as
accelerometer sensors employed are initially 6DOF sensor – they have almost the same requencies, 24.0 Hz and
MPU6050 and followed by ADXL335. The distance
between the leaking point and the vibration sensor is varied
from the range of 0.5 m to 2 m for every case. The x-axis of
the sensor is placed in parallel with the water pipeline and
thus only x-axis is the preference of our measurement. The
vibration data is collected at an interval of 10 ms and is
transmitted to a laptop over ZigBee. The x-axis data of each
sensor is collected for three times over a duration of 3
minutes.
The x-axis data is bandpass filtered and analysed offline
using Fast Fourier transform (FFT). The purpose of the
analysis is to determine the frequencies of the vibration
signal produced due to various conditions of the water
pipeline, normal (no leakage) and abnormal (leakage), and
they are compared between the two sensors (MPU6050 and
ADXL335). As for the abnormal condition, the frequency
parameter is studied to determine the sizes of the leaking
holes that typically show the severity of the leakage.

TABLE 2: Experiment setting procedure

Parameter Values
Type of Pipe ABS
Pipe length (m) 10.0
Type of Vibration Sensor MPU6050, ADXL335
Distance between pipe leaking 0.5
and sensor (m) 1.0
1.5
2
Duration of data collection (s) 120
Time sampling (ms) 10
Water pressure (kgf/cm2 ) 1.0
Size of leaking hole i. No Leak
ii. 1-mm leak
iii. 3-mm leak
 Figure 6: Frequency (Hz) versus Distance(meter) between leak point and
sensor

VI. CONCLUSION
This paper presented the frequencies of two sensors
namely, MPU6050 and ADXL 335 across a number of
ranges at the water pressure of 1.0 kgf/cm2. The distance
between the sensor and the leaking point was varied from
0.5 to 2.0 m, and the vibration data was collected at x-axis
for both sensors with the consideration of the x-axis is
parallel with the water flow within the pipeline. Both
accelerometer sensors are able to distinguish the normal (no
Figure 5: FFT Analysis for vibration data on pressure 1.0kgf/cm2 with
Range (a) 0.5m (b) 1.0m (c) 1.5m and (d) 2.0m between leak point and leak) and abnormal (leak) conditions of the pipe.
sensor Nevertheless, the variation in terms of frequency for the
conditions of the pipeline is significant for the ADXL335
compared with the MPU6050. At the shorter distance of 0.5
24.8 Hz. In contrast, the MPU6050 is able to detect the m, the ADXL335 can detect all the three cases whilst the
severity of the leaking since the variation of the frequencies MPU6050 can detect all the cases at the distance of 0.5 m
is seen in Figure 5(c). At the distance of 2.0 m, the and also 1.5 m. For other distances, the sensors seemed hard
ADXL335 is still able to distinguish the normal and to detect the severity of the leaking. However, it is more
abnormal pipes, but not for the MPU6050. Because of likely the MPU6050 is able to detect the small and big
similar frequencies of the 1-mm and 3-mm of leaking, both leakage if more than one axis is taken into consideration.
sensors failed to detect the sizes of the leaking, as shown in This is due to some small variations of frequencies can be
Figure 5(d). seen for the 1-mm and 3-mm sizes of leaking for a number
Figure 6 summarised our findings from both of ranges compared with the ADXL335. Future work is to
sensors, and the results of frequencies against the ranges are enhance the analysis for all the three axes (x-, y- and z-axis)
plotted. The ADXL335 considerably shows a constant and also for the MMA sensor.
frequency of 25 Hz for the normal pipe across the ranges
chosen in this study. In addition, a frequency gap of 1 Hz is
clearly seen between the normal and abnormal pipe states. REFERENCES
Nevertheless, the ADXL335 is unable to distinct the sizes of [1] M. F. Ghazali, "Leak Detection Using Instantaneous Frequency
the leaking holes. In contrast, the MPU6050 does not show Analysis," in Department of Mechanical Engineering. vol. PHD
a consistent trend for the normal pipe cases across different Sheffield: University of Sheffield, 2012, p. 199.
ranges including the closest range. However, we could [2] C. P. Liou, "Pipeline Leak Detection by Impulse Response
conclude for the leaking case the MPU6050 demonstrated a Extraction," Journal of Fluids Engineering, vol. 120, pp. 833-838,
detection frequency of 31.4 Hz. There are some small 1998.
variations of frequencies between the sizes of the leaking [3] R. A. D. a. N. A. A. S. MIM Ismail, "Water Pipeline Monitoring
holes for the MPU6050, as presented in Figure 6. Due to System using Vibration Sensor," in IEEE Conference on Wireless
this, it may be feasible for the MPU6050 to distinct a small Sensors (ICWiSe) 2014 Kuala Lumpur, Malaysia, 2014.
and big holes on the pipeline if occur. [4] J. D. B. Robert P. Evans, Alan G. Stephens, "Flow rate
measurements using flow-induced pipe vibration," Transactions on
the ASME, vol. 126, 2004.
[5] A. L. Prashun, Fundamentals of fluid mechanics: Prentice Hall,
1980.
[6] N. G. C. a. A. .A.C.O, "Design and Implementation of Pipeline
Monitoring System Using Acceleration-Based Wireless Sensor
Network," The International Journal Of Engineering And Science
(IJES), vol. 3, pp. 49-58, 2014.
[7] N. A. M. Ir. Dr. Hasnul Mohammad Salleh, "NON-REVENUE
WATER, IMPACT TO THE SERVICE, ENVIRONMENT AND
FINANCIAL ".
[8] D. N. C. Nocole Metje, David Cheneler, Michele Ward and Andrew
M. Thopmas, "Smart Pipes – Instrumented Water Pipes, Can This Be
Made a Reality," Sensors – ISSN 1424-8220, 2011.
[9] T. H. Bee, "Detecting Leakage in Underground Buried Pipe," in
Department of Mechanical Engineering vol. Master Singapore:
National University of Singapore 2010, p. 60.
[10] V. L. Streeter, Fluid Mechanics. New York: McGraw–Hill Inc. ,
1966.
[11] L. J. Clancy, Aerodynamics.
[12] H. Chanson, Hydraulics of Open Channel Flow: An Introduction.,
2004.
[13] R. Mulley, Flow of Industrial Fluids: Theory and Equations.,
illustrated ed.: Taylor & Francis, 2004.
[14] H. C. H. D.J.Wang, Jeremy Wang, Z.D.Tsai, "Water Induced
Vibration in the Nsrrc," in Proceeding of 2005 Paticle Accelerator
Conference Knoxwille, May 2005.
[15] K. W. Hironori Kakuta, and Yosuke Kurihara, "Development of
Vibration Sensor with Wide Frequency Range Based on Condenser
Microphone -Estimation System for Flow Rate in Water Pipes-,"
World Academy of Science, Engineering and Technology, vol. 70,
2012.
[16] H. P. Olesen, "Detection of pressure variations in thin walled tubes
by vibration measurements," in Pipe vibration and pressure
detection Naerum, Denmark: Bruel & Kjaer, pp. 3-9.

View publication stats