Study on the natural gas pipeline safety monitoring technique and the time-frequency signal analysis method

Accepted Manuscript Study on the natural gas pipeline safety monitoring technique and the time-frequency signal analysis method Zhigang Qu, Yanfen Wang, Huanhuan Yue, Yang An, Liqun Wu, Weibin Zhou, Huayang Wang, Zhichao Su, Jian Li, Yu Zhang, Likun Wang, Xiliang Yang, Yuchen Cai, Daxian Yan PII:

S0950-4230(17)30175-4

DOI:

10.1016/j.jlp.2017.02.016

Reference:

JLPP 3425

To appear in:

Journal of Loss Prevention in the Process Industries

Received Date: 10 October 2016 Revised Date:

29 January 2017

Accepted Date: 20 February 2017

Please cite this article as: Qu, Z., Wang, Y., Yue, H., An, Y., Wu, L., Zhou, W., Wang, H., Su, Z., Li, J., Zhang, Y., Wang, L., Yang, X., Cai, Y., Yan, D., Study on the natural gas pipeline safety monitoring technique and the time-frequency signal analysis method, Journal of Loss Prevention in the Process Industries (2017), doi: 10.1016/j.jlp.2017.02.016. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

Study on the natural gas pipeline safety monitoring technique and the time-frequency signal analysis method Zhigang Qu1 , Yanfen Wang1, Huanhuan Yue1, Yang An1, Liqun Wu1, Weibin Zhou1, Huayang Wang1, ＊

Zhichao Su2, Jian Li2, Yu Zhang2, Likun Wang3, Xiliang Yang3, Yuchen Cai1 and Daxian Yan1

＊

1 College of Electronic Information and Automation, Tianjin University of Science & Technology, Tianjin, 300222, China

3 Pipeline R&D Center, CNCP, Langfang, Hebei, 065000, PR China E-mail: [email protected]

Abstract

RI PT

2 State Key Laboratory of Precision Measuring Technology & Instruments, Tianjin University, Tianjin 300072, PR China

Hydrate plugging and leakage in natural gas pipelines have been big issues for the industry with the

SC

globally increasing demand for natural gas. A natural gas pipeline safety monitoring technique is studied in this paper, which is based on acoustic excitation. In this technique hydrate plugging and leakage can be monitored online at multiple locations and distinguished by "energy-pattern" method based on the wavelet packet analysis. The position of a hydrate plugging or leakage can be achieved by

M AN U

correlation algorithm. The results of experiments and modeling work show that the technique can locate hydrate plugging or leakage at multiple positions with good accuracy and distinguish them effectively.

Key words: natural gas pipeline; hydrate plugging; leakage; monitoring; wavelet packet

1.Introduction

With the globally increasing demand for energy, natural gas, which is a clean and environmentally

TE D

friendly fossil fuel, plays an increasingly important role all over the world. Among fossil fuels, natural gas is the one which emits the least carbon dioxide per unit energy (Esen and Oral, 2016; Macknick et al., 2013). The growth rate of annual global demand of natural gas is expected to be over 10 percent from 2007 to 2035 (Girgin and Krausmann, 2016; Furuoka, 2016). Pipelines are widely used for transportation or conveying most of the natural gas resources, consequently, the flow assurance is ever

EP

more important (Lasich et al., 2014). Hydrate blockage and leakage can lead to serious environmental pollution and high economic costs, which are of big issues for the industry. Detection and localization of such faults are necessary for smooth functioning of the industry and safety of the environment (Datta

AC C

and Sarkar, 2016).

For the past few years, a series of methods and research work have been developed for natural gas

pipeline hydrate and leakage detection respectively. Lung and Doige developed a time averaging transient testing technique for measuring the acoustic properties of piping systems and mufflers (Lung and Doige, 1983), but this method cannot deliver accurate work for locating and analyzing the internal characteristics of pipeline. Hasan reported a transient analysis solution to locate and characterize the plugs in gas wells (Hasan, et al., 1996), however, the localization accuracy depends on the dimensions of the deposit. Papadopoulou and Wang have reported an acoustic pipeline blockage and leakage detection method (Papadopoulou et al., 2008; Wang et al., 2009, 2012), respectively. However, these papers do not cover online hydrate monitoring and relevant experimental work, furthermore the water deposit in the experiments should not be regarded as hydrate. Yang has developed a hydrate early warning system (Yang et al., 2012), which is able to examine hydrate plugging offline. However, the system cannot monitor hydrate plugging online and is not suitable for

ACCEPTED MANUSCRIPT leakage detection (Yang et al., 2013). Yuan presented the modeling work for a partial blockage detection method for natural gas pipelines (Yuan et al., 2012, 2013), however, no experimental results were reported. Stouffs and Giot reported leakage detection method based on the mass balance (Stouffs and Giot, 1993). However, this method cannot detect tiny leakages. Buerck has reported a method using distributed fiber optic (Buerck et al., 2001), however, this technique requires huge construction work and the false alarm rate can be high. Zhang S Q and Zhang Y presented the NPW (Negative Pressure

RI PT

Wave) method (Zhang et al., 2006; Zhang et al., 2014), which is especially suitable to detect large and sudden leakage for oil pipelines.

At present there is no suitable method to monitor both hydrate plugging and leakage in a natural gas pipeline. In order to solve this problem, a natural gas pipeline safety monitoring technique is studied in this paper, based on acoustic excitation, which is able to monitor both hydrate plugging and leakage at

SC

multiple locations online. Since both hydrate plugging and leakage can cause reflected waves, this paper reports on an "energy-pattern" method based on wavelet packet analysis, which can distinguish the reflection signals of the two cases. In this paper the system principle including measurement principle, relevant acoustic theories and modeling work are introduced in section 2 and 3 respectively.

M AN U

And in section 4 the reflection signal analysis method based on wavelet packet is introduced and the experimental results are presented in section 5.

2. Measurement principle

The schematic configuration of the monitoring system (Qu et al., 2016) is shown in Fig. 1. At the start point of the pipeline there is a loudspeaker, which is used to emit the sound waves into the pipeline. A microphone is used to detect the reflected wave caused by a hydrate plugging or leakage.

TE D

The Data Acquisition (DAQ) card is not only used to collect the signals for further processing, but also

AC C

EP

to generate the driving signal which is then amplified by a power amplifier to drive the loudspeaker.

Fig. 1. Measurement principle.

The location of leakage x can be calculated by the following equation:

x=

ct , 2

(1)

where c is the sound velocity in the pipeline, t is the time that the reflected wave takes traveling back to the microphone and can be calculated by cross correlation algorithm. y , the location of hydrate plugging, can be achieved in the same way.

3. Acoustic theories and detection models 3.1 Acoustic theories When an acoustic wave propagates along a pipeline, the propagation equation of wave in the damping medium is indicated as

ACCEPTED MANUSCRIPT p = A0e −αx ,

(2)

where p is the acoustic pressure, Pa; A0 is the amplitude at the initial moment, Pa, and

α is the

attenuation coefficient of medium and can be expressed as (Kinsler et al., 2000) :

where

η ′ is

1 η ′ω ω2 ω2 1 1 + ⋅η ′′ + ⋅ χ( − ) . 3 rc 2 ρ 0 2 ρ 0 c 2 ρ0 c3 Cv C p

(3)

RI PT

α = αη ′ + αη ′′ + α χ =

the shear viscous coefficient of medium, Pa ⋅ s ; η ′′ is the volume-change viscous

coefficient of medium, Pa ⋅ s ;

χ

is the heat transfer coefficient, W / (m2 ⋅ K ) ; Cv is the constant

SC

volume specific heat and C p is the constant-pressure specific heat, kJ /( kg ⋅ K ) ; ω = 2πf is the angular frequency, f is the frequency of acoustic wave, Hz; r is the inner radius of pipeline, m; 3

is the density of medium, kg / m ;

c is the speed of acoustic wave in the pipeline, m / s .

M AN U

ρ0

When the acoustic wave comes across a blockage during the propagation in a pipeline, then the change of pressure can be indicated as below：

j ( ωt − k p 1 x )

pr = Apr e

j (ωt + k p 1 x )

pt = Apt e

j ( ωt − k p 2 x )

TE D

pi = Api e

,

(4)

,

(5)

,

(6)

pi = p r + pt ,

(7)

where Api is the complex pressure amplitude of the incident wave ,

Apr is that of the reflected wave

Apt is that of the transmitted wave. Since the sound speed is different, the wave numbers k p1 in

medium A and

k p 2 in medium B are different. R is the reflection coefficient, which is led by the

EP

and

AC C

acoustic impedance (Kinsler et al., 2000; Melling, 1972) changes, is given by equation (8).

zA

and

zB

medium and

 z − zB   Z =  A  zA + zB 

2

.

(8)

z = ρc ,

are the acoustic impedance of medium A and B, respectively.

(9)

ρ

is the density of the

c is the sound speed in medium.

Considering when a valve is suddenly opened in a pipe, there is an abrupt change in acoustic impedance of the media around the leakage (Layekuakille et al., 2010), which leads to a reflection wave back. Thus, the leakage can also be localized by calculating the traveling time of the reflected wave.

3.2 Detection models According to the theories in section 3.1, a series of models have been built to understand the

ACCEPTED MANUSCRIPT whole measurement process. All the models have been built based on carbon steel pipes (US Standard ASTMA 539-90a) with 100mm internal diameter under the normal situation (298.15K, sealed away from the atmosphere with 0.4MPa internal pressure and 2.45m/s internal gas flow rate.). An acoustic attenuation model is shown in Fig.2, in which a single cycle 800Hz acoustic signal travels in a 5Km straight pipe. In this model the x-axis is the distance from the start point of the pipe and the y-axis is the normalized sound pressure (the amplitude of the incident wave is set to 1). The sound pressure is measured at the points along the pipe and is found to decreases exponentially with

RI PT

increase in distance. The model shows that the power of the incident wave needs to be increased accordingly for the same microphone when a longer monitoring range is expected, which means a

M AN U

SC

higher output power from the loudspeaker is required.

Fig. 2. Acoustic signal attenuation model

A hydrate detection model has been studied, which is based on a 2Km long pipe with two hydrate deposits as shown in Fig.3. Two hydrate deposits are placed at 993.5m (Hydrate deposit A, 0.03m

AC C

EP

start point of the pipe.

TE D

thickness*1m long) and 1303.5m (Hydrate deposit B, 0.05m thickness*1m long) respectively from the

Fig. 3. Pipe layout for hydrate detection model

The localization results for the two hydrate deposits can be found in Table 1, which indicates that the method is able to locate hydrate deposits at multiple locations with good accuracy in a 2Km long range pipe. Table 1 Localization results of the hydrate deposit detection model Actual position (m)

Localization result in the model (m)

Hydrate deposit A

993.5

993.168

Hydrate deposit B

1303.5

1304.336

ACCEPTED MANUSCRIPT The reflection signals of the model are shown in Fig.4, in which the reflection signal A and B are caused by hydrate deposit A and B respectively. In this model the x-axis is the distance from the

SC

RI PT

start point of the pipe and the y-axis is the normalized sound pressure.

Fig. 4. Reflection signals for hydrate detection model

M AN U

4. Reflection signal analysis method

Both the hydrate plugging and leakage can cause reflected waves, hence it is important to distinguish the reason for the reflected waves so that people can take measures properly in time.

4.1 Wavelet packet analysis The

reflection

signals

are

typically

non-stationary

signals,

for

which

the

traditional Fourier transform are not suitable. Wavelet transform possesses excellent characteristic of

TE D

time-frequency localization and provides an effective measure for analyzing non-stationary signals (Paul et al., 2013). Compared to the wavelet transform, the wavelet packet transform is a technique to decompose a signal repeatedly into successive low and high frequency components using a recursive filter-decimation operation (Peng et al, 2012). For this study in order to discover the time-frequency

EP

characters of detected signals in a continuous frequency band, in which the signals change actively, wavelet packet is selected as it can cover both higher frequency and lower frequency (Daubechies, 1992) offering more application advantages. The original signal f (t ) can be constructed by the sum of

2 j components (Daubechies and

AC C

Grossmann, 2006; Coifman and Wickerhauser, 2010) as: 2j

f (t ) = ∑ f ji (t ) ,

(10)

i =1

where f ji (t ) is the wavelet packet component signal that can be expressed by a linear combination of wavelet packet functions as follows: ∞

∑c

f ji (t ) = where

k = −∞

ψ ij , k (t ) ,

i j ,k

(11)

cij ,k and ψ ij ,k (t ) are defined as wavelet packet coefficients and wavelet packet functions, i, j

and k are integers and defined as the modulation, scale and translation parameter. The wavelet packet coefficients can be obtained from:

c ij , k =

∫

∞

−∞

f (t )ψ ij , k (t ) dt ,

(12)

ACCEPTED MANUSCRIPT The wavelet packet function is defined as:

ψ ij , k (t ) = 2 j / 2ψ i (2 j t − k ) .

(13)

ϕ (t ) ∈ L2 ( R ) is the orthogonal scaling function, ϕ jk (t ) = 2 j / 2 ϕ ( 2 j t − k )

as: µ0

ϕ (t )

and orthogonal wavelet function

are illustrated

= φ (t ) , µ1 = ψ (t ) . µn (t ) the wavelet packets about the orthonormal scaling functions can

be expressed as:

(15)

SC

 µ 2 n (t ) = 2 ∑ h(k ) µ n (2t − k )  k ∈Z ,  = − µ ( t ) 2 g ( k ) µ ( 2 t k ) ∑ n  2 n +1 k ∈Z 

h(k ) and g (k ) are the conjugate filter coefficient, h(k ) is the coefficients of lowpass filter

M AN U

where

ψ (t )

RI PT

Orthogonal scaling function

(14)

k and g (k ) is the highpass filter and g (k ) = (−1) h(1 − k ) . The two coefficients are in an orthogonal

relationship. The function cluster

{µn (t )}(n ∈ Z + )

the following orthogonal quality:

of wavelet packet possesses translational orthogonality and

kl

,k,l ∈ Z

(16)

TE D

µ n ( t − k ), µ n ( t − l ) = δ δ kl is a Kronecker delta:

Then

EP

 δ kl = 0 , k ≠ l   δ kl = 1 , k = l

{un (t )}n∈Z

constructs the orthogonal basis in

(17)

L2 ( R) .

AC C

According to equation (16), the algorithm of wavelet packet decomposition is thus obtained, that is,

{d } and {d

j , 2 n +1

j,2n

l

l

} can be obtained from {d }: j +1, n

l

 d l j , 2 n = ∑ ak − 2l d kj +1,n  k ,  j , 2 n +1 j +1, n d = b d ∑ l k − 2 l k  k 

(18)

1 where ak = 1 h~ (k ) , bk = g~ (k ) . 2 2 The algorithm of wavelet packet reconfiguration is obtained in the following relationship: can be obtained from

{d } and {d

j , 2 n +1

j,2n

l

l

}. That is,

{d } j +1, n

l

ACCEPTED MANUSCRIPT d l j +1,n = ∑ (hl −2 k d kj , 2 n + g l −2 k d kj , 2 n +1 ) .

(19)

k

4.2 Reflection signal analysis method based on wavelet packet The "energy-pattern" method based on wavelet packet is employed to distinguish hydrate plugging and leakage. Wavelet packet decomposition is firstly conducted on a reflection signal. Then through the reconfiguration of decomposition coefficients within every frequency band on a certain scale, a new

RI PT

time sequence is constructed on each decomposition node (the node on the signal decomposition

tree). Then time domain analysis is carried out to extract eigenvector showing the time-frequency characters of the reflection signal. Suppose the sampling frequency of the signals is 2 f , and then if decomposition is carried out on the signals,

j

j layer wavelet packet

2 frequency bands of equal width can thus be formed.

SC

f / 2 j . After the decomposition, the coefficient of j layer

The frequency width of every interval is

wavelet packet is C mj ,k , k = 0,1LL 2 j − 1 and m is the location in the wavelet packet space.

M AN U

According to Parseval energy integral equation, −∞

∫ f ( x)

+∞

2

dx = ∫ C j , k 2 ,

(20)

which demonstrates that the coefficients of wavelet packet have energy quantity. Suppose

E j ,k is the signal energy of the frequency band at the node k of j layer, and then

TE D

E j , k = ∑ | C mjk |2 .

(21)

m

And energy

E jk is normalized when E=

∑E k =0

EP

Then,

j,k

(k = 0,1,L,2 j − 1) .

E ′jk = E jk / E ,

(22)

(23)

E ′jk is the energy obtained after normalization.

AC C

where

2 j −1

5. Experiments and results 5.1 Experimental setup

In the experiments, the pipe is made of carbon steel with the length of 21.58m and the internal

diameter is 100mm (see Fig. 5) (Qu et al, 2016). The pipe is sealed away from the atmosphere and the external temperature is about 25℃ (298.15K). There are ten bends in the pipe in order to simulate the real situation, which causes significant acoustic signal attenuation. As shown in Fig. 5(e), in order to simplify the procedures, ice is employed as hydrate because their physical properties, especially acoustic properties, are very similar to each other (Sloan, 1998, 2000). Six controllable holes (A-F) are set on the pipe. Among the holes, four (A, B, C and D) of them are controlled by ball valves and employed for leakage, the other two (E and F) are reserved for air pump to control the internal pressure. The experimental setup is shown in Fig. 5(b) below.

ACCEPTED MANUSCRIPT In order to avoid the acoustic signals external to the pipe, a zero-phase bandpass filter

SC

RI PT

covering the frequency band of the reflection signals has been designed in the system to suppress such noise outside the pipe.

(a)

TE D

M AN U

(b)

(c)

(e) (d)

EP

Fig. 5. (a) Pipe schematic diagram. (b) Experiment setup. (c) Pipe with the pump. (d) Hydrate within the pipe. (e) Detail of the leakage holes.

AC C

5.2 Experimental results

When two hydrates with 50 mm diameter are placed at 14.90m (Hydrate①) and 18.05m (Hydrate②)

in the pipe respectively under the pressure of 0.4MPa, the results are shown in Fig. 6(Qu et al, 2016), in which two reflection signals can be found and localized. The localization results and the actual position of hydrate plugging can be found in Table 2. From the comparison results it can be concluded that the system is able locate multiple hydrate deposits within a pipe with good accuracy.

RI PT

ACCEPTED MANUSCRIPT

Fig. 6. Reflection signals for the two cases: with two hydrates and without hydrate in the pipe.

Fig. 7 shows the results for the two cases, in which one is without leakage and the other one is with two leakages (15mm diameter) at the location A and C (in Fig.5 (a)) under the pressure of 0.4MPa.

SC

Two reflection signals caused by the two leakages can be found and the localization results can be

M AN U

found in Table 2.

TE D

Fig. 7. Reflection signals for the two cases: with two leakages and without leakage in the pipe. Table 2 Localization results in the experiment. Events

EP

Hydrate ① Hydrate ②

Actual position (m)

Localization position (m)

14.90

14.936

18.05

18.087

4.56

4.588

14.43

14.480

AC C

Leakage A

Leakage C

It should be pointed out that the complex internal structures within the pipe, like bends, defects etc.,

can cause the small reflection signals as well, which can be found in Fig. 6 and Fig. 7. However such small reflection signal can be suppressed significantly by signal processing. As shown in Table 2, the experiment results indicate that the technique can detect and locate both hydrate plugging and leakage in a pipe at multiple positions with good accuracy.

5.3 Reflection signal analysis results From the experimental data it is difficult to distinguish hydrate plugging and leakage, however it is important to know so that people can take measures properly in time. A 1x8 vector, consisting of normalized energy of 8 sensitive frequency bands based on equation (23), is employed as the eigenvector of the reflection signals by using "energy-pattern" method. The eigenvector is expressed as

ACCEPTED MANUSCRIPT T = [E 1 , E 2 , E 3 , E 4 , E 5 , E 6 , E 7 , E 8 ] ,

(24)

where E1-E8 are the 8 elements (normalized energy) defined in equation (23) and calculated via equation (20) to (23).

The frequency range selected covers 0 KHz-3.125 KHz, which is then divided into 8 sensitive frequency bands. Typical eigenvectors for hydrate plugging and leakage are shown in Fig.8 (a) and (b)

M AN U

SC

(a) Eigenvector for hydrate deposit

RI PT

respectively, which clearly shows that the eigenvectors for the two cases are different from each other.

(b) Eigenvector for leakage Fig. 8. Eigenvectors for hydrate plugging and leakage

In order to verify the uniformity of the eigenvectors for hydrate plugging and leakage, different diameter of the hydrate (from 3cm-7cm) and different leakage location (A-D in Fig.5 (a)) are chosen in

TE D

the experiment. Furthermore for each case (each diameter for hydrate or each leakage position) 20 samples have been analyzed by the method and the average values are shown in Table 3, in which E1-E8 are the normalized energy defined in equation (23) for the two cases. The results in Table 3 indicate that the eigenvectors for leakage are similar and the same situation can be found for hydrate plugging, although the conditions and locations differ in the experiments. On the other side, the recognition.

EP

eigenvectors are different significantly between the two cases, which is a good foundation for

AC C

Table 3. Eigenvectors for hydrate plugging and leakage. Events

Leakage (15mm)

Hydrate Plugging

Details

E1

E2

E3

E4

E5

E6

E7

E8

A

0.9025

0.0623

0.0009

0.0339

0.0000

0.0001

0.0001

0.0002

B A and D B and C A and B

0.9146 0.9231 0.9315 0.9080

0.0595 0.0413 0.0502 0.0536

0.0015 0.0016 0.0016 0.0007

0.0240 0.0335 0.0161 0.0371

0.0000 0.0000 0.0000 0.0000

0.0001 0.0000 0.0001 0.0000

0.0001 0.0001 0.0002 0.0001

0.0003 0.0004 0.0004 0.0004

D=7cm

0.5875

0.2539

0.0033

0.1544

0.0000

0.0003

0.0002

0.0004

D=6cm D=5cm D=4cm D=3cm

0.6359 0.6194 0.6422 0.6004

0.2397 0.2459 0.2402 0.2415

0.0008 0.0009 0.0007 0.0007

0.1228 0.1330 0.1160 0.1565

0.0000 0.0000 0.0000 0.0000

0.0003 0.0003 0.0004 0.0004

0.0001 0.0001 0.0001 0.0001

0.0004 0.0004 0.0004 0.0004

Fig. 9 below shows the comparison results of the eigenvectors for leakage and hydrate plugging based on Table 3. The first five eigenvectors are the results for leakages and the rest five are the results

ACCEPTED MANUSCRIPT

RI PT

for hydrate plugging.

Fig. 9. Comparison between the eigenvectors of the two cases under different conditions

SC

The following conclusion can be achieved from the above results:

(1) The eigenvectors for leakage are similar with each other and the same conclusion can be achieved for hydrate plugging, although the conditions and locations differ in the experiments. On the other side, the eigenvectors are different significantly between the two cases, which is a good

M AN U

foundation for further recognition. The signal analyzing results of the experimental data prove that the "energy-pattern" method is able to extract eigenvectors for hydrate plugging and leakage effectively. (2) Almost all the energy in the reflection signals for leakage focus in the first frequency interval which is around 90%. The sum of average energy in the second and the fourth frequency intervals is around 10%. The energy in the other frequency intervals can be ignored.

(3) Most of the energy in the reflection signals for hydrate plugging concentrates on the first frequency interval. The normalized energy in the first frequency interval, the second one and the fourth

TE D

one are about 62%, 24% and 14% respectively. The energy of the other frequency intervals can be ignored.

6. Conclusion

As currently hydrate plugging and leakage are of big issues for the industry, a natural gas pipeline

EP

safety monitoring technique is studied in this paper, which is based on acoustic excitation. According to the results of experiments and modeling work, the technique is able to monitor both hydrate plugging and leakage online at multiple positions with good accuracy in a gas pipeline. The "energy-pattern"

AC C

method, based on wavelet packet, is proven by the results of experimental data to be able to extract eigenvectors for hydrate plugging and leakage effectively, which is a good foundation for further pattern recognition. The technique enables people to know the reason of a reflection signal and its location so that measures can be taken in time to avoid possible serious environmental pollution and high economic costs.

Acknowledgments

This work was funded by the National Natural Science Foundation of China (51674176), the Tianjin Research Program of Application Foundation and Advanced Technology (15JCZDJC39200), Tianjin Sci-tech Project (15KPXM01SF015), Foundation of State Key Laboratory of Precision Measuring Technology and Instruments (Tianjin University) (PIL1506).

ACCEPTED MANUSCRIPT

Appendix Symbol

Definition

Unit

p

Sound pressure

Pa

α

Attenuation coefficient

dB/m

η′ η ′′

Pa ⋅ s Pa ⋅ s

Volume-change viscous coefficient

χ

Heat transfer coefficient

Cv

Constant volume specific heat

Cp

Constant-pressure specific heat

ω

Angular frequency

RI PT

Shear viscous coefficient

W / (m2 ⋅ k )

kJ /( kg ⋅ K ) kJ /( kg ⋅ K ) rad/s

Frequency of acoustic wave

r

Inner radius of pipeline

m

ρ0

Density of medium

kg / m 3

c

Speed of sound

pi

Sound pressure of the incident wave

M AN U

SC

f

pr

Sound pressure of the reflected wave

Hz

m/s Pa Pa

pt

Sound pressure of the transmitted wave

Pa

k

Wave numbers

N/A

Reflection coefficient

N/A

Acoustic impedance of medium

N/A

R

TE D

z

References

Esen, V., Oral, B., 2016. Natural gas reserve/production ratio in Russia, Iran, Qatar and Turkmenistan: A political and economic perspective. Energ. Policy. 93, 101-109.

EP

Macknick, J., Meldrum, J., Heath, G., Nettles-Anderson, S., 2013. Life cycle water use for electricity generation: a review and harmonization of literature estimates. Environ. res. lett. 8(8), 1880-1885. Girgin, S., Krausmann, E., 2016. Historical analysis of U.S. onshore hazardous liquid pipeline

AC C

accidents triggered by natural hazards. J. Loss. Prevent. Proc. 40, 578-590. Furuoka, F., 2016. Natural gas consumption and economic development in China and Japan: An empirical examination of the Asian context. Renew. Sust. Energ. Rev. 56, 100-115.

Lasich, M., Mohammadi, A. H., Bolton, K., et al., 2014. Influence of unlike dispersion interactions in modeling methane clathrate hydrates. Fluid. Phase. Equilibr. 381(6), 108-115.

Datta, S., Sarkar, S., 2016. A review on different pipeline fault detection methods. J. Loss. Prevent. Proc. 41, 97–106. Lung, T. Y., Doige, A. G., 1983. A time-averaging transient testing method for acoustic properties of piping systems and mufflers with flow. J. Acoust. Soc. Am. 73(73), 867-876. Hasan, A. R., Kouba, G. E., Wang, X., et al., 1996. Transient Analysis To Locate and Characterize Plugs in Gas Wells. In:SPE Annual Technical Conference and Exhibition. Denver, Colorado. Papadopoulou, K. A., Shamout, N. M., Lennox, B., et al., 2008. An evaluation of acoustic reflectometry for leakage and blockage detection. P. I. Mech. Eng. C-J. Mec. 222(6), 959-966. Wang, X,. Short, G., Dawson, K., 2009. ACOUSTIC REFLECTOMETRY FOR GAS PIPELINES

ACCEPTED MANUSCRIPT -MONITORING FEATURES IN GAS PIPELINES USING ACOUSTEK® PPSA Aberdeen Seminar UK. Wang, X., Lewis, K. M., Papadopoulou, K. A., et al., 2012. Detection of hydrate and other blockages in gas pipelines using acoustic reflectometry. P. I. Mech. Eng. C-J. Mec. 226(7), 1800-1810. Yang, J., Chapoy, A., Mazloum, S., Tohidi, B., 2012. A Novel Technique for Monitoring Hydrate Safety Margin. Spe. Prod. Oper. 27(4), 376-381. Conductivity and Acoustic Velocity. Energ. Fuel. 27(2), 736-742.

RI PT

Yang, J., Tohidi, B., 2013. Determination of Hydrate Inhibitor Concentrations by Measuring Electrical Yuan, Z., Deng, Z., Jiang, M., et al., 2015. A modeling and analytical solution for transient flow in natural gas pipelines with extended partial blockage. J. Nat. Gas. Sci. Eng. 22, 41-149.

Yuan, Z., Deng, Z., Jiang, M., et al., 2015. Extended partial blockage detection in a gas pipeline based on Tikhonov regularization. J. Nat. Gas. Sci. Eng. 27, 130-137. term. J. Loss. Prevent Proc. 6(5), 307-312.

SC

Stouffs, P., Giot, M., 1993. Pipeline leak detection based on mass balance: Importance of the packing Buerck, J., Roth, S., Kraemer, K., et al., 2001. Application of a fiber-optic NIR-EFA sensor system for in situ monitoring of aromatic hydrocarbons in contaminated groundwater. J. Hazard. Mater.

M AN U

83(1-2), 11-28.

Zhang, S. Q., Jin, S. J., Yang, F. L., et al., 2006. Crucial technologies of oil-transporting pipe leak detection and location based on wavelet and chaos. Meas. Sci. Technol. 17(3), 572-577. Zhang, Y., Chen, S., Li, J., 2014. Leak detection monitoring system of long distance oil pipeline based on dynamic pressure transmitter. Measurement. 49, 382–389.

Qu, Z. G.., Wang, H. Y., An, Y., et al., 2016. Online monitoring method of hydrate agglomeration in natural gas pipelines based on acoustic active excitation. Measurement. 92, 11–18.

TE D

Kinsler, L. E., Frey, A. R., Coppens, A. B., Sanders, J. V., 2000. Fundamentals of Acoustics 4th Edition. John Wiley and Sons Inc., New York, pp. 126-127. Melling, T. H., 1972. The acoustic impendance of perforates at medium and high sound pressure levels. J. Sound. Vib. 29(1), 1-65.

Layekuakille, A., Vergallo, P., Trotta, A., 2010. Impedance Method for Leak Detection in Zigzag

EP

Pipelines. Meas. Sci. Rev. 10, 209-213.

Paul, R., Sengupta, A., Pathak, R. R., 2013. Wavelet based denoising technique for liquid level system. Measurement. 46(6), 1979-1994.

AC C

Peng, X. L., Hao, H., Li, Z. X., 2012. Application of wavelet packet transform in subsea pipeline bedding condition assessment. Eng. Struct. 39(6), 50-65.

Daubechies, I., 1992. Ten Lectures On Wavelets. Comput. Phys. 6(3), 1671-1671. Daubechies, I., Grossmann, A., 2006. Frames in the bargmann space of entire functions. Commun. Pur. Appl. Math. 41(2), 151-164.

Coifman, R. R., Wickerhauser, M. V., 2010. Entropy-Based Algorithms for Best Basis Selection. IEEE. T. Inform. Theory. 38(2), 713-718. Sloan, E. D., 1998. Gas Hydrates:  Review of Physical/Chemical Properties. Energ. Fuel. 12(2), 191-196. Sloan, E.D, 2000. Hydrate engineering. Society of Petroleum Engineers Inc., Texas, pp. 63-61.

ACCEPTED MANUSCRIPT

Highlights



The method is able to monitor and locate both hydrate plugging and leakage

A time-frequency signal analysis method based on wavelet packet has been proposed and proven to be able to distinguish hydrate plugging and leakage

EP

TE D

M AN U

SC

effectively.

AC C



RI PT

online at multiple locations in a gas pipeline.