See

discussions, stats, and author profiles for this publication at: http://www.researchgate.net/publication/279313987

Experimental study on the probe dynamic

behaviour of feeler pigs in detecting internal
corrosion in oil and gas pipelines
ARTICLE in JOURNAL OF NATURAL GAS SCIENCE AND ENGINEERING SEPTEMBER 2015
Impact Factor: 2.16 DOI: 10.1016/j.jngse.2015.06.033

READS

22

5 AUTHORS, INCLUDING:
Shimin Zhang

Shuhai Liu

22 PUBLICATIONS 4 CITATIONS

China University of Petroleum

SEE PROFILE

62 PUBLICATIONS 356 CITATIONS

SEE PROFILE

Xiaoxiao Zhu
China University of Petroleum
8 PUBLICATIONS 7 CITATIONS
SEE PROFILE

All in-text references underlined in blue are linked to publications on ResearchGate,

letting you access and read them immediately.

Available from: Xiaoxiao Zhu

Retrieved on: 17 November 2015

Journal of Natural Gas Science and Engineering 26 (2015) 229e239

Contents lists available at ScienceDirect

Journal of Natural Gas Science and Engineering

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

Experimental study on the probe dynamic behaviour of feeler pigs in

detecting internal corrosion in oil and gas pipelines
Xiaolong Li, Shimin Zhang*, Shuhai Liu, Xiaoxiao Zhu, Kang Zhang
College of Mechanical and Transportation Engineering, China University of Petroleum e Beijing, Changping, Beijing, 102249, China

a r t i c l e i n f o

a b s t r a c t

Article history:
Received 9 April 2015
Received in revised form
15 June 2015
Accepted 16 June 2015
Available online xxx

Leakages are the major cause of in-service natural gas and oil pipeline accidents, and many factors, such
as corrosion, can lead to leakages. The feeler pig is one of the most typical contact-testing tools in oil and
gas pipelines. In this paper, the probe dynamic behaviour of the feeler pig in detecting internal corrosion
has been investigated using a handmade inspecting system. The dynamic characteristics of the probe are
different in the uphill section and the downhill section, and the trajectory of the probe is asymmetric. A
bouncing phenomenon was found in the outlet region of the corrosion. The experimental results indicated that both the speed and the spring pre-tightening elastic force are closely related to the inspection
precision. The research in this paper provides guidance for studying the inspection precision of the feeler
pig.
2015 Published by Elsevier B.V.

Keywords:
Pipeline
Feeler pig
Dynamic behaviour
Corrosion
Detection precision

1. Introduction
Pipelines function as blood vessels to bring such necessities as
oil and natural gas, and they are considered to be the most favoured
mode of transportation of gas and liquid in large quantities
(Kishawy and Gabbar, 2010; Tolmasquim et al., 2008; Xiaoxiao Zhu
et al., 2014; Esmaeilzadeh et al., 2009; Nguyen et al., 2001a, b). With
the long timelines employed, the problem of pipeline ageing is
becoming more and more serious, and pipelines aging greatly
promotes defects. Among pipeline defects, corrosion is the critical
factor affecting the service life of pipelines (Teixeira et al., 2008;
Cosham et al., 2007; Podgorbunskikh et al., 2008; Choi et al.,
2003). According to statistics by the US National Transportation
Safety Board, 59% of pipeline accidents are caused by corrosion.
According to the former Soviet Union pipeline accident statistics,
during 1981e1987, 1210 accidents happened in the 240,000 km
pipeline. Among these, accidents caused by corrosion were 45.1% of
the total accidents (Tiratsoo, 1992). Therefore, regular pipeline inspection is signicant, not only prolonging pipeline life but also
greatly reducing the loss caused by accidents (Cosham et al., 2007;
Choi et al., 2003; Kim et al., 2003a, b).
At present, many internal inspection technologies can be used to

* Corresponding author.
E-mail address: zsm1976748@126.com (S. Zhang).
http://dx.doi.org/10.1016/j.jngse.2015.06.033
1875-5100/ 2015 Published by Elsevier B.V.

investigate pipeline defects all over the world. Depending on the

different detection principles, the technology can be divided into
two categories: non-destructive testing (NDT) and contact testing
(CT) (Kim et al., 2003a, b; Dai et al., 2012; Carvalho et al., 2006). NDT
includes ultrasonic technology (UT) and magnetic ux leakage
(MFL). UT uses acoustic wave reection to inspect corrosion, so it
needs a homogeneous uid with good acoustic properties to serve
as sonic wave coupling. Because of this limitation, UT is difcult to
use for inspecting gas pipelines and is mostly used in oil pipelines
(Skjelvareid et al., 2013; Park,.1996; Siqueira et al., 2004). The MFL
technology produces a magnetic eld in the pipe wall to sense
leakages inside the pipe as the wall thickness changes. Therefore, it
is difcult to inspect small diameter and thick wall pipelines using
MFL. Similar to UT, the biggest advantage of MFL is high inspection
accuracy, but it also has disadvantages including low spatial resolution and low readability of the MFL pig report generated by
magnetic spots (Gloria et al., 2009; Carvalho et al., 2006;
Camerini et al., 2008). Furthermore, due to the limited size of the
sensor, the MFL pig has difculty going through a curved pipeline.
In addition, owing to the detection principle of MFL, the pipeline
needs to be degaussed when the pig sweeps across the pipeline.
Through collecting analogue signs and converting it into the
angular variation of the inspection arm, the CT registers the
movement process generated by internal defects and then delineates the corrosion features (Dai et al., 2012; Kim et al., 2003a, b;
Camerini et al., 2008). Compared to the NDT, CT performs the direct

230

X. Li et al. / Journal of Natural Gas Science and Engineering 26 (2015) 229e239

L0

List of symbols
ax0
ay0
ax1
ay1
.

The acceleration of the barycenter moving along

horizontal direction in the downhill section
The acceleration of the barycenter moving along
vertical direction in the downhill section
The acceleration of the barycenter moving along
horizontal direction in the uphill section
The acceleration of the barycenter moving along
horizontal direction in the uphill section

an0

Normal acceleration in the downhill section

at0

Tangential acceleration in the downhill section

an1

Normal acceleration in the uphill section

.
.

.
at1

Dh
k
l
n
t0
t1
x0
x1
x
x0

Dx
y1
EkY1
EkY1
F
F0
H

the
L1
the
the

the

Tangential acceleration in the uphill section

The depth of the defect
Stiffness of the spring
The length of the probe
The number of pulses
Time spent in the downhill section
Time spent in the uphill section
The distance of the probe head moving along the
horizontal direction in the downhill section
The distance of the barycenter moving along the
horizontal direction
The initial spring length
The spring length corresponding to a changing angle

Da
The changing spring length corresponding to a
changing angle Da
The distance of the barycenter moving along the
vertical direction
Kinetic energy of the probe along the Y direction
during the bouncing of the probe
Kinetic energy of the probe along the negative Y
direction during the bouncing of the probe
The spring pre-tightening force
The spring force corresponding to a changing angle Da
The distance between the xed bracket and the rotary
table

dimensional measurement of the pipeline internal corrosion by

contact with no practical limits of wall thickness or a need for a
homogeneous uid during the inspection. CT also has an excellent
ability to go through a curved pipeline. Moreover, compared to the
NDT, CT has easy data post-processing and inner surface pipeline
reconstruction. In addition, the cost of the CT is far lower than that
of the MFL and the UT; because of these unique advantages, the CT
is widely used for the internal detection of pipelines.
Based on the difference of contact arms, the product of the CT
can be classied into three types: wheel, arm and probe pigs (Cox
et al., 1936; Smith et al., 1990; Rosenberg et al., 1992). Among
them, the probe pig is the most typical product of contact-testing
and is also called the feeler pig. Compared to wheel and arm pigs,
not only could a feeler pig inspect the diameter, but it could also
detect and quantify internal corrosion and welds. Furthermore, the
detection precision of CT is similar to the UT and MFL pig. Therefore,
the feeler pig has been widely recognized in internal pipeline
corrosion inspection. Fig. 1 shows the detection process of the
feeler pig. Since the 1960s, the pipeline industry has been using and
developing feeler pigs to inspect small diameter production

V
Vx0
Vy0
Vt0

Vx1
Vy1
Vt1

We

a
Da
_

d0
d1
D4
u0
u1

The distance of the xed bracket moving along the

direction of the xed bracket speed in the downhill
section
The distance of the xed bracket moving along the
direction of the xed bracket speed in the uphill
section
Thermal energy generated during the bouncing of the
probe
The length of the defect in the direction of the moving
probe
The velocity of the xed bracket
The average velocity of the probe barycenter along the
horizontal direction in the downhill section
The average velocity of the probe barycenter along the
vertical direction in the downhill section
The average velocity of the probe barycenter along the
tangential direction of the trajectory in the downhill
section
The average velocity of the probe barycenter along the
horizontal direction in the uphill section
The average velocity of the probe barycenter along the
vertical direction in the uphill section
The average velocity of the probe barycenter along the
tangential direction of the trajectory in the uphill
section
Elastic energy generated during the bouncing of the
probe
The initial angle between the xed bracket and the
probe
The changing angle corresponding to the depth Dh
The corresponding radian to rotate angle D4
The average angular acceleration in the downhill
section
The average angular acceleration in the uphill section
The corresponding rotate angle to pulse n
The average angular velocity in the downhill section
The average angular velocity in the uphill section
The distance between the defect and the central axis of
the rotary table

Fig. 1. Schematic diagram of the feeler pig inspection process.

X. Li et al. / Journal of Natural Gas Science and Engineering 26 (2015) 229e239

231

pipelines, such as 800 , using just a few sensors to inspect furnace

renery pipes. Thanks to the excellent results obtained in eld
tests, the concept was extended to large diameter pipelines, such as
12e1600 , 2200 and 2400 (Camerini et al., 2008). Nevertheless, how to
improve the detection precision of the feeler pig is an important
problem. To better study the detection precision improvements, the
process and dynamic behaviour of the detection probe swept across
the surface of corrosion should be carefully studied rst, as shown
in Fig. 1(c).
Currently, many companies are working on product development and eld experiments. A literature survey revealed that few
papers have studied the dynamic behaviour of feeler pigs in
detecting internal corrosion in gas and oil pipeline, especially
with respect to the detection probe (Li et al., 2015; Nguyen et al.,
2001a, b). A feeler pig depends on the probe to detect the
corrosion, and the detection precision is closely related to the
movement process of the probe sweeps across the corrosion. The
movement process and dynamic behaviour of the probe is easily
inuenced by velocity and spring pre-tightening force, but few
papers have studied the specic movement process of the probe
running in different velocities and spring pre-tightening forces.
Consequently, this work mainly studies the inuence of the velocity and spring pre-tightening on the probe movement process;
therefore, an experimental facility was built to record the realtime movement process of the probe. Furthermore, theoretical
analysis of the dynamic behaviour of the probe was performed by
experimental method, which provided a theoretical basis for
improving the detection precision of the feeler pig (Hou et al.,
2014; Huang et al., 2008).
2. Experimental conditions
In this study, a detailed analysis of the detection probe movement process is conducted. To record the whole process, an
experimental device is designed, as shown in Fig. 2. In the experiment, a rotary table is used to simulate the pipeline wall, and the
rotary table rotates around the central axis of the detection probe,
which has an initial angle relative to the xed bracket that will
change constantly when swept across the corrosion surface. An
angle encoder is connected to the detection probe, and the
changing angle during the detection probe sweeping across the
corrosion will be recorded. Through analysing the constantly
changing angle, the dynamic characteristics of the probe can be
obtained.
2.1. Test instruments
As shown in Fig. 2(a), the experimental device mainly includes
three parts: the inspection unit, the data acquisition unit and the
data processing unit. The inspection unit is the main part, which
includes the motor, support platform, rotary table and the angle
inspection device. During the detection process, the probe remains
stationary while the rotary table rotates at different speeds. When
the probe sweeps across the surface of the rotary table, the data
acquisition device can record the constantly changing angle using
the angle encoder. After data collection, the data processing
equipment is used to analyse the data and conclude the law from
the movement process of the probe.
The structure of the angle inspection device, including the xed
bracket, the angle encoder, the spring and the detection probe, is
shown in Fig. 3. The angle encoder, connected to the detection
probe through a coupling, is mounted on the side of the inspection
device. One end of the spring is connected to the xed bracket and
the other end is connected to the detection probe, and this spring is
used to provide the pre-tightening force for the probe. The

Fig. 2. (a) The diagram of the experimental device. (b) The diagram of the inspection
unit.

inspection probe in direct contact with the wall transfers the variable angle value to the angle encoder as it sweeps across the wall
surface.
The schematic diagram of the probe sweeps across the concave
surface is shown in Fig. 4. Through analysing the geometric relationship in Fig. 4, formula (1) can be obtained. If the value of a
could not meet formula (1), the probe is not able to detect the
concave surface completely. The object of study in this paper is a
feeler pig, which applies to pipelines 355 mm in diameter.
Therefore, according to the requirements of the ability of the pig to
go through curved pipelines, the installation height of the probe H
is 61 mm, and, considering the probe strength, the diameter of the
inspection probe is 3 mm. Through analysing the geometrical
relationship of Fig. 4, the relationship between a and Da can be
obtained, as shown in formula (2). In the experiment, the depth of
the corrosion Dh is 6 mm. Therefore, the relationship between a
and Da can be obtained, as shown in Fig. 5. If the depth of the
corrosion is identical, the bigger the value of Da is, the higher the
detection precision will be. Consequently, when the a meets
formula (1), the smaller the value of a is, the higher the detection
precision will be. Therefore, the value of a is 40.3 , and the length
of probe is 80 mm.

232

X. Li et al. / Journal of Natural Gas Science and Engineering 26 (2015) 229e239

_

length of probe l is the radius of the arc PSo . The slope of the line
QM meets formula (3). This paper mainly studies the process of the
probe sweeps across the corrosion, so the shape of the simulated
corrosion should not access the largest area that the probe is able to
detect. In this experiment, the length and depth of the simulated
corrosion are 48 mm and 6 mm, respectively, and the shape of the
corrosion is similar to an arc to ensure that the corrosion is included
in the largest area that the probe is able to detect. Consequently, in
the experiment, the process of the probe sweeps across the
corrosion can be recorded completely.



kQM tan a  90

(3)

2.2. Experimental process

In the experiment, adjusting the velocity of the rotary table and
the spring pre-tightening elastic force can simulate the pig running
at different speeds and different pre-tightening elastic forces. The
experiment adopts the absolute value angle encoder to record the
rotating angle of the rotary table. When the rotary table takes a
turn, the angle encoder records 3600 pulses. The relationship between the number of pulses and the rotating radian is presented in
the following.


D4 360  n=3600
_

z l$D4

(4)
(5)

The specic experimental process involves two steps:

Fig. 3. (a) The diagram of the angle inspection device. (b) Prototype of the angle inspection device.

a arccos

H
H
 arccos
l
H Dh
0

(1)
1

Dh H
1
B
B
C
C
Da arcsin@ qA  arcsin@ qA
2
H$ 1 tan a
H$ 1 tan a2
(2)
As shown in Fig. 6, when the length of the corrosion along the
axial direction S is conrmed, the geometric relationship OMP is the
largest area that the probe is able to detect. O is the centre, and the

(a) Keeping the pre-tightening elastic force constant, the velocity of the probe movement varies from 1 m/s to 8 m/s, and
then the experimental results are analysed.
(b) Keeping the velocity constant, the pre-tightening elastic
force varies from 9 N to 15 N, and then the experimental
results are analysed.
3. Results and discussion
The experimental results are shown in Figs. 7e9. Through analysing these results, the following phenomenon were found.
3.1. The trajectory asymmetry
As shown in Fig. 7, when the probe swept across the concave

Fig. 4. The schematic diagram of the probe sweeps across the concave surface.

X. Li et al. / Journal of Natural Gas Science and Engineering 26 (2015) 229e239

233

Fig. 5. The relationship graph between a and Da. The depth of the corrosion is 6 mm.

Fig. 8. Force analysing diagram of the inspection probe.

Fig. 6. The diagram of the largest area that the probe is able to detect.

surface, the slope of the trajectory in the downhill section was

larger than that in the uphill section. In other words, the
displacement of the probe along the velocity direction in the
downhill section is less than that in the uphill section. In this paper,
this phenomenon was named trajectory asymmetry.
As described earlier, the defect is ideally a concave surface. In
other words, the defect is symmetric. Based on the location difference, it can be divided into two parts: the downhill section and
the uphill section. When the probe sweeps across the concave

Fig. 7. The experimental results diagram of the probe sweeps across the concave
surface. The speed is 2.5 m/s.

surface, it will pass the downhill section and uphill section in turns.
The schematic diagram of the probe sweeps across the concave
surface is shown in Fig. 4. Through analysing the geometric relationship in Fig. 4, formula (6) can be obtained.

Dh HcosDa tan a sinDa  1

(6)

To better explain the experimental result, the mechanical model

should be built rst. From Fig. 8, we know that during the process of
the probe movement, ve forces work together on the probe:
gravity from the probe itself, traction provided by the xed bracket,
the elastic force offered by the spring, friction and the reaction force
from the inner surface of the concave defect. Under the action of the
ve forces, the movement process of the probe sweeps across the
inner surface of the concave defect covers two types of motion:
parallel motion and rotational motion..The former is uniform motion in a straight line with velocity V , and the latter is circular

Fig. 9. The schematic of the probe dynamics analysis.

234

X. Li et al. / Journal of Natural Gas Science and Engineering 26 (2015) 229e239

motion around the xed bracket with angular velocity u. During the
process, the linear motion cannot generate acceleration. Nevertheless, the circular motion is a variable acceleration motion and
.
will become two types of accelerations: normal acceleration an and
.
tangential acceleration at . Because the ve forces working on the
probe are constantly changing, the kinematic parameters are also
variable.
As shown in Fig. 9, during the process of the probe sweeping
across the concave surface, the barycenter of the probe P moves
from P0 to P1. As described earlier, the movement of the probe includes two types of motion: parallel motion and rotational motion.
Based on the different coordinate systems, there are two types of
kinematic parameters to describe the movement process. From the
geometric relationship in Fig. 9, the formulas (7e11) can be
obtained.

x0

Dh
HcosDa tan a sinDa  1

tana  Da=2
tana  Da=2

sections and uphill sections, we will study the processes separately.

(a) The inuence of speed
Fig. 12 shows the process of the probe sweeps across the
downhill section. During the process, the speed of the xed bracket
remained constant, and the probe always kept contact with the
inner face of the concave surface. Based on the geometric relationship and dynamic equation, formulas (12e20) can be obtained.

Vx0

(12)
Vy0

H$V tana  Da=2cosDa tan a sinDa  1

S tana  Da=2  2HcosDa tan a sinDa  1

(13)

(7)

u0

2VDa tana  Da=2

S tana  Da=2  2HcosDa tan a sinDa  1

(14)

VDa tana  Da=2

l
S tana  Da=2  2HcosDa tan a sinDa  1

(15)

L0 S=2 

HcosDa tan a sinDa  1

tana  Da=2

(8)

Vt0

L1 S=2

HcosDa tan a sinDa  1

tana  Da=2

(9)

ax0

t0 S=2V 

t1 S=2V

HcosDa tan a sinDa  1

tana  Da=2V
HcosDa tan a sinDa  1
tana  Da=2V

S$V tana  Da=2  V$HcosDa tan a sinDa  1

S tana  Da=2  2HcosDa tan a sinDa  1

4V 2 H tana  Da=2cosDa tan a sinDa  1

fS tana  Da=2  2HcosDa tan a sinDa  1g2
(16)

(10)
ay0
(11)

Analysing formulas (7e11) shows that when the speed of the

xed bracket is kept constant, the time of the probe spent in the
downhill section is less than that in the uphill section; therefore, it
is inevitable that the displacement of the probe along the velocity
direction in the downhill section is less than that in the uphill
section. This effect also inuences the asymmetric phenomenon.
The asymmetric phenomenon is caused by the inspection device
structure, and it will always exist, regardless of the speed and
spring pre-tightening forces.
3.2. The bouncing phenomenon
Fig. 10 shows that, with increasing speed, a convex trajectory
appears in the exit area. After careful observations, the process can
be divided into three different types with increasing speed. The rst
process includes gures a, b and c. In this process, no convex trajectory exists in the exit area of the concave surface. The second
process includes gures d, e and f. This process shows a convex
trajectory appearing in the exit area, and with increasing speed, the
convex trajectory is larger. The third process includes gures g and
h. During the process, two convex trajectories appear in the exit
area. The above three processes indicate that the probe will
continue to move along the circumferential direction when it
sweeps to the exit area of the concave surface. This phenomenon is
called the bouncing phenomenon. In addition, with increasing
speed, the bouncing amplitude will be larger.
Fig. 11 shows that with constant velocity and increasing spring
pre-tightening force, the bouncing amplitude will decrease.
Many reasons cause the above experimental phenomena, and
analysing the experimental results found that the speed and spring
pre-tightening force play critical roles. The remainder of the paper
will analyse the experimental results from these two aspects.
Because of the different kinematic parameters of the downhill

4V 2 H tana  Da=22 cosDa tan a sinDa  1

fStana  Da=2  2HcosDa tan a sinDa  1g2

(17)

(
an0

)2
p
2VDa tana  Da=2
l
S tana  Da=2  2HcosDa l$tan a sinDa  1
(18)

d0

8DaV 2 tana  Da=22

fStana  Da=2  2HcosDa l$tan a sinDa  1g2
(19)

at0

4DaV 2 tana  Da=22 l

fStana  Da=2  2HcosDa tan a sinDa  1g2

(20)

In the same way, Fig. 13 shows the process of the probe swept
across the uphill sections. Based on the geometric relationship and
dynamic equation, formulas (21e29) can be obtained.

Vx1

SV tana  Da=2 VHcosDa tan a sinDa  1

S tana  Da=2 2HcosDa tan a sinDa  1

(21)

Vy1

H$V tana  Da=2cosDa tan a sin a  1

S tana  Da=2 2HcosDa tan a sinDa  1

(22)

u1

2VDa tana  Da=2

S tana  Da=2 2HcosDa tan a sinDa  1

(23)

Vt1

VDa tana  Da=2

l
S tana  Da=2 2HcosDa tan a sinDa  1

(24)

4V 2 H tana  Da=2cosDa tan a sinDa  1

ax1 
fStana  Da=2 2HcosDa tan a sinDa  1g2
(25)

X. Li et al. / Journal of Natural Gas Science and Engineering 26 (2015) 229e239

235

Fig. 10. The experimental results diagram of the probe sweeps across the concave surface with different speeds. The value of spring pre-tightening elastic force is 12 N.

4V 2 H tana  Da=22 cosDa tan a sinDa  1

ay1 
fStana  Da=2 2HcosDa tan a sinDa  1g2
(26)
(
an1

)2
p
2VDa tana  Da=2
l
S tana  Da=2 2HcosDa tan a sinDa  1

8DaV 2 tana  Da=22

d1 
fStana  Da=2 2HcosDa tan a sinDa  1g2
(28)
4DaV 2 tana  Da=22 l
at1 
fStana  Da=2 2HcosDa tan a sinDa  1g2

(27)

(29)

236

X. Li et al. / Journal of Natural Gas Science and Engineering 26 (2015) 229e239

Fig. 11. The experimental results diagram of the probe sweeps across the concave surface with different spring pre-tightening elastic forces. The speed varies from 5 m/s to 7 m/s.

Fig. 12. The process that the probe swept across the concave surface in the downhill
sections.

Fig. 13. The process that the probe swept across the concave surface in the uphill
sections.

Formulas (21e29) show the relationship between the kinematic

parameter and the speed of the xed bracket V. Because of the
bouncing phenomenon in the exit area of the concave surface, the
dynamic characteristics when the probe reaches the position M0
should be studied, as shown in Fig. 14. Analysing the movement
process of the probe sweep across the downhill sections can obtain
the average velocity of the probe along the Y direction Vy1 , which is
shown in formula (22). When the values of H, a, S and Da are
constant, as the velocity V of the xed bracket increases, the

average velocity of the probe along the Y direction Vy1 will also
increase. Similarly, with increasing velocity V, the average acceleration of the probe along the Y direction ay1 will also increase. Based
on the law of energy conservation, formula (30) can be obtained.
Analysing formula (30) shows that with increasing Vy1 , the value of
the spring deformation will increase and the height of the probe
bouncing will also increase. Consequently, because of the velocity
and acceleration, when the probe sweeps to the exit area of the
concave surface, the probe will continue moving along the
circumferential direction, causing the bounce phenomenon.

X. Li et al. / Journal of Natural Gas Science and Engineering 26 (2015) 229e239

237

Fig. 14. The process of the probe moves in the bouncing area.

1 2
mv
2 y1
1
We kDx2
2
Eky Q We
Eky
1

(30)

In addition, the experimental results show that, with increasing

speed, the bouncing phenomenon will appear twice.
As shown in Fig. 14, when the probe sweeps to the bouncing
area, the endpoint of the probe will move along the dotted lines (HI-J-K-P). When the endpoint of the probe moves from point I to J,
the average speed of the probe along the Y axis will increase. According to the theorem of momentum and conservation of energy,
when the endpoint of the probe collides with the wall of the
pipeline, the kinetic energy of the probe along the negative Y direction Ek will transform into thermal energy Q and the kinetic
energy of the probe along the Y direction Ek, as shown in formula
(31). The kinetic energy Ek along the Y direction generates the
twice bouncing phenomenon, and the larger the probe speed is, the
larger the bouncing amplitude is.

Ek Q Ek

(31)

(b) The inuence of spring pre-tightening force

The experimental results show that, with increasing speed, the
bouncing phenomenon appears, and the area in which the
bouncing phenomenon appears is named the bouncing area.
As described earlier, during the probe sweep across the concave
surface, ve forces work together on the probe. Fig. 15(a) shows
that during the movement process, the length of the spring is
constantly changing, so the spring force is also variable.
Fig. 15 (b, c) show the length variation of the spring in the
concave surface area and in the bouncing area, respectively. According to the geometric relationships, the force of the spring in
different positions was presented as follows.
In the concave surface area:
According to the law of cosines, the triangle OMN (shown in Fig.
15(b)) can be obtained:

x0

q
4ax sinDa=2cosa  Da=2  x2  4a2 sinDa=22
(32)

Fig. 15. (a) The schematic diagram of the changing spring length. (b). The schematic
diagram of the changing spring length in the concave surface area. (c). The schematic
diagram of the changing spring length in the bouncing area.

And then:

Dx x  x0 ; F0 F  kDx

(33)


q
F0 F  k x  x2 4a2 sinDa=22  4ax sinDa=2cosa  Da=2

(34)

238

X. Li et al. / Journal of Natural Gas Science and Engineering 26 (2015) 229e239

In the bouncing area:

q
x0 4ax sinDa=2cosa Da=2 x2 4a2 sinDa=22
(35)
Dx x0  x; F0 F kDx

and the probe dynamic behaviour should be studied rst. Based on

the above research, a reasonable spring pre-tightening force should
be chosen and then a correction algorithm should be proposed to
eliminate the inuence of the bouncing phenomenon on the
detection precision. Consequently, all of the studies in the paper
provide a basis to improve the detection precision.

(36)

q

F0 F k
4ax sinDa=2cosa Da=2 x2 4a2 sinDa=22  x

(37)

Acknowledgements
Formula (34) shows that every changing angle corresponds
toDa a force F0. In the concave surface area, the elastic force is
inversely proportional to the changing angle Da. Therefore, the
spring force, in the downhill section, varies from large to small,
and in the uphill section, from small to large. The force of the
spring in the uphill section is symmetrical to that in the downhill
section.
Similarly, formula (37) shows that when the probe moves to
the bouncing area, the spring force is proportional to the angle
Da. As the changing angle Da increases, the elastic force also
increases.
As the earlier experimental results described, increasing the
spring pre-tightening force decreases the bouncing amplitude.
Formula (37) shows that, in the bouncing area, once the pretightening elastic force is larger, then the force in every position
will be larger. When the speed in the exit area is identical, the
momentum theorem shows that if the pre-tightening force of the
spring is larger, the bouncing time will be shorter. Consequently,
the bouncing amplitude will inevitably be smaller.
4. Conclusions
In this paper, the probe dynamic behaviour of feeler pigs in
detecting internal corrosion has been investigated using a homebuilt inspection system. The probe dynamic characteristics when
the feeler pig sweeps in the uphill section and in the downhill
section are different, and the trajectory of the movement is asymmetric. The asymmetric phenomenon is caused by the detection
device structure, and it will always exist, regardless of the speed
and spring pre-tightening forces. Through changing the speed, a
bouncing phenomenon was found in the outlet region of the
corrosion. When the spring pre-tightening remains constant, as the
speed increases, the bouncing amplitude will increase, and when
the speed reaches a certain point, twice bouncing will appear.
When the speed remains constant, as the spring pre-tightening
force increases, the bouncing amplitude will decrease. Analysing
the experimental results showed that the velocity and spring pretightening force are closely related to the detection precision. To
better understand the relationship between the probe dynamic
behaviour and the detection precision, a theoretical model of the
process that the probe swept across the concave surface was built.
Analysing the theoretical model revealed that the greater the probe
velocity is, the greater the detection error will be. Similarly, the
smaller the spring pre-tightening force is, the greater the detection
error will be.
As described earlier, to improve the feeler pig inspection precision, the process of the probe swept across the concave surface

The work was nancially supported by the Program for New

Century Excellent Talents in University (NCET-13-1028).
References
Camerini, Claudio, von der Weid, Jean Pierre, Freitas, Miguel, et al., 2008. Feeler pig:
a simple way to detect and size internal corrosion. In: International Pipeline
Conference, Calgary, Alberta, Canada, September 29-October. Paper no.64626.
Carvalho, A.A., Rebello, J.M.A., Sagrilo, L.V.S., et al., 2006. MFL signals and artificial
neuralnet works applied to detection and classification of pipe weld defects.
NDT E Int. 39, 661e667.
Choi, H.S., Bomba, J.G., 2003. Acceptance criteria of defects in under sea pipeline
using internal inspection. Ocean. Eng. 30, 1613e1624.
Cosham, A., Hopkins, P., Macdonald, K.A., et al., 2007. Best practice for the assessment of defects in pipelineseCorrosion. EngineeringFailureAnalysis 14,
1245e1265.
Cox, Albert B., Wilmington, Calif, 1936. Caliper. Patent 395974, USA.
Dai, Lisha, Zhu, Xiaoxiao, Zhang, Shimin, 2012. Research development of pipeline
caliper tool technology. Oil Gas Storage Transp. (China) 31, 808e813.
Esmaeilzadeh, F., Mowla, D., Asemani, M., et al., 2009. Mathematical modeling and
simulation of pigging operation in gas and liquid pipelines. J. Pet. Sci. Eng. 69,
100e106.
Gloria, N.B.S., Areiza, M.C.L., Miranda, I.V.J., et al., 2009. Development of a magnetic
sensor for detection and sizing of internal pipeline corrosion defects. NDT & E
Int. 42, 669e677.
Hou, Qingmin, Jiao, Wenling, Ren, Liang, 2014. Experimental study of leakage
detection of natural gas pipeline using FBG based strain sensor and least square
support vector machine. J. Loss Prev. Process Ind. 32, 144e151.
Huang, Yi, Ji, Dawei, 2008. Experimental study on seawater-pipeline internal
corrosion monitoring system. Sensors Actuators B: Chem. 135, 375e380.
Kim, Dong Kyu, Cho, Sung Ho, Park, Seoung Soo, et al., 2003a. Development of the
caliper system for a geometry PIG based on magnetic eld analysis. KSME Int. J.
17, 1835e1843.
Kim, Dong Kyu, Cho, Sung Ho, Park, Seoung Soo, et al., 2003b. Design and implementation of 30 geometry PIG. KSME Int. J. 17, 629e636.
Kishawy, Hossam A., Gabbar, Hossam A., 2010. Review of pipeline integrity management practices. Int. J. Press. Vessels Pip. 87, 373e380.
Li, Xiaolong, Zhang, Shimin, et al., 2015. An experimental evaluation of the probe
dynamics as a probe pig inspects internal convex defects in oil and gas pipelines. Measurement 63, 49e60.
Nguyen, T.T., Yoo, H.R., et al., 2001a. Speed control of pig using bypass ow in
natural gas pipeline. IEEE Int. Symp. Ind. Electron. 12, 863e868.
Nguyen, Tan Tien, Kim, Dong Kyu, Rho, Yong Woo, et al., 2001b. Dynamic modeling
and its analysis for PIG ow through curved section in natural gas pipeline. In:
IEEE International Symposium on Computational Intelligence in Robotics and
Automation, Banff, Alberta, Canada, July 29eAugust 1, pp. 492e497.
Park, Moon Hlo, Kirn, In Sup, Yoont, Young Ku, 1996. Ultrasonic inspection of long
steel pipes using Lamb waves. NDT & E Int. 29, 13e20.
Podgorbunskikh, A.M., Loskutov, V.E., et al., 2008. Improving the quality of diagnostics of gas-main pipelines by using a device for automated control of the
velocity of pig aw detectors. Russ. J. Nondestruct. Test. 44, 334e342.
Jeffrey S. Rosenberg, Kevin W. Lockyear, 1992. Pipeline Caliper Pig. Patent 5088336,
USA.
Siqueira, M.H.S., Gatts, C.E.N., daSilva, R.R., et al., 2004. The use of ultrasonic guided
waves and wavelets analysis in pipe inspection. Ultrasonics 41, 785e797.
Skjelvareid, Martin H., Birkelund, Yngve, Larsen, Yngvar, 2013. Internal pipeline
inspection using virtual source synthetic aperture ultrasound imaging. NDT & E
Int. 54, 151e158.
Smith, Dennis R., Tulsa, Okla, 1990. Bend Detector Pig. Patent 4953412, USA.

X. Li et al. / Journal of Natural Gas Science and Engineering 26 (2015) 229e239

Teixeira, A.P., GuedesSoares, C., Netto, T.A., et al., 2008. Reliability of pipelines with
corrosion defects. Int. J. Press. Vessels Pip. 85, 228e237.
Tiratsoo, J.N.H., 1992. Pipeline Pigging Technology. Gulf Professional Publishing.
Tolmasquim, S.T., Nieckele, A.O., et al., 2008. Design and control of pig operations

239

through pipelines. J. Pet. Sci. Eng. 62, 102e110.

Zhu, Xiaoxiao, Zhang, Shimin, et al., 2014. Experimental study on dynamics of
rotatable bypass-valve in speed control pig in gas pipeline. Measurement 47,
686e692.