Sunteți pe pagina 1din 10

1 Copyright 2012 by ASME

CALCULATION OF DYNAMIC STRESS INTENSITY FACTORS FOR PIPES DURING


CRACK PROPAGATION BY DYNAMIC FINITE ELEMENT ANALYSIS

Masaki Mitsuya
Tokyo Gas Co., Ltd.
Yokohama, Japan
Hiroyuki Motohashi
Tokyo Gas Co., Ltd.
Yokohama, Japan

Noritake Oguchi
Tokyo Gas Co., Ltd.
Tokyo, Japan
Shuji Aihara
The University of Tokyo
Tokyo, Japan

ABSTRACT
A dynamic finite element analysis method was proposed
for calculating the dynamic stress intensity factors for pipes
during crack propagation. The proposed method can directly
calculate the stress intensity factors without the simplification
used in theoretical analyses, and it can consider the effects of
the crack velocity and gas decompression. It was found that the
stress intensity factors of long propagating cracks in pipes
saturated at a certain value in the case of a high crack velocity.
However, although the stress intensity factors for pipes were in
good agreement with those of band plates in the case of a high
crack velocity, the stress intensity factors for pipes were
different from those of band plates in the case of a low crack
velocity. This result could be explained by the effect of bulging
on the stress distribution around a crack tip. The effect of
bulging was more prominent for pipes with smaller diameters.
In contrast, the dynamic stress intensity factors for band plates
were in good agreement with the theoretical values that
consider the dynamic effects and tended to decrease
monotonically with increasing crack velocity. Additionally, the
effects of gas decompression, caused by leakage from opened
cracks, on the stress intensity factors for pipes were
investigated. An explanation of crack deviation, which is
observed in actual pipeline fractures, was provided by
analyzing the ratio of the longitudinal stress to lateral stress.

1 INTRODUCTION
Prevention of brittle fracture propagation is essential for
designing high-pressure gas pipelines. The basic method of
preventing brittle fracture propagation involves estimating the
material resistance by a drop-weight tear test (DWTT). Brittle
fracture propagation in a pipeline can be effectively arrested if
the ratio of the shear area to fracture area obtained in a DWTT
is greater than 85%. This indicator was obtained by noting a
similarity between DWTT and burst test results in terms of the
transition behavior of fractured material for line pipes. When
the ratio of shear area in a DWTT exceeds 85%, the ratio in a
line pipe is approximately 100%. This indicator focuses only
on the fracture toughness of a material; therefore, it cannot
consider the effect of internal pressure on a pipeline.
However, a brittle fracture can be arrested even if the ratio
of shear area in a DWTT is less than 85% in the case of low
internal pressure, because a fracture can be estimated by
comparing fracture toughness to crack driving force. An
evaluation method for brittle fracture in line pipes was
proposed by Maxey et al [1]. The method evaluates the
arrestability of brittle fracture by comparison between the
energy release rate G [2], which is a factor related to crack
extension and is expressed by Eq. (1), and the absorbed energy
R, which is the material resistance obtained by specimen tests
such as DWTT.
G
E
R
m h
t o
2
=
(1)
where o
h
is the nominal hoop stress; R
m
is the middle radius of
Figure 1 Band plate
2h
Crack propagation
Crack
Enforced displacement
x
y
2h
Crack propagation
Crack
Enforced displacement
x
y
Proceedings of the 2012 9th International Pipeline Conference
IPC2012
September 24-28, 2012, Calgary, Alberta, Canada
IPC2012-90205
2 Copyright 2012 by ASME
the pipe, which is the average of the outer radius R
o
and the
inner radius R
i
; and E is Youngs modulus. Eq. (1) corresponds
to an analytical result for a static semi-infinite crack on a band
plate under the enforced displacement, as shown in Figure 1.
The nominal stress on a band plate generated by the enforce
displacement corresponds to the nominal hoop stress on a
pipeline, and the width 2h of the band plate in Figure 1
corresponds to the circumference of a pipeline, tD
m
. That is to
say, diameter of pipe D
m
corresponds to 2h/t. An elastic energy
corresponding to Eq. (1) is released per unit length of crack
growth. Eq. (1) is equivalent to the stress intensity factor K,
expressed by Eq. (2).
m h
R K t o = (2)
However, in actual phenomena, a brittle crack propagates
not statically but dynamically. Under a dynamic condition, the
stress intensity factor decreases with crack velocity because the
crack propagates ahead before stress field, which is a source of
the stress intensity factor, is sufficiently generated. An
analytical result for a semi-infinite crack on a band plate under
a dynamic condition was proposed by Freund [3]. Similarly, an
analytical result for a semi-infinite crack in a pipeline under
constant internal pressure was also proposed by Freund [3].
The analytical results for the band plate and pipeline are
similar; this result is expressed by Eq. (3), which ignores the
minor term.
( ) V A
R
K
I
m h
t o
=
(3)
( )
( ) D c
V
V A
s
d
I 2 2
2
1 v
o

=
(4)
( )
2
2
1 4
s s d
D o o o + = (5)
2
1
|
|
.
|

\
|
=
d
d
c
V
o
(6)
2
1
|
|
.
|

\
|
=
s
s
c
V
o
(7)
( )
2
1 v
=
p
d
E
c
(8)
( ) v +
=
1 2
p
s
E
c
(9)
where A
I
(V) is the universal function, V is the crack velocity, c
d

is the velocity of the longitudinal wave, c
s
is the velocity of the
transverse wave,
p
is the density of the pipe material, and v is
Poissons ratio. Eq. (8) expresses the velocity of the
longitudinal wave under a plane stress condition.
The analytical Eq. (3) implies that the stress intensity
factor is constant, although the stress intensity factor is
generally proportional to the square root of the crack length
and diverges with crack length. Therefore, the stress intensity
factor for a long crack in a pipeline should be calculated by
another method.
Moreover, the effect of bulging is not considered in Eq.
(3). The effect of bulging is important in an internally
pressurized pipe, and this effect is generally considered in
evaluations of fracture initiation. However, Emery et al. [4]
found that the crack tip opening angle of a propagating crack,
which is a crack extension force, is affected by bulging for
ductile fracture propagation. Though the effect of bulging is
expected to not be severe in the case of brittle fracture
propagation, which is not accompanied by large deformation,
the effect must nevertheless be clarified.
Also, the effect of gas decompression, which is caused by
gas leakage from an opened crack in a pipeline, on stress
intensity factor is not considered in Eq. (3). The internal
pressure ahead of the crack tip is independent of gas
decompression because the velocity of brittle fracture
propagation, which is from 365 m/s to 914 m/s [5], is generally
greater than the speed of sound in the internal gas, which is
about 430 m/s in the case of pure methane. However, the
internal pressure behind the crack tip may depend on the gas
decompression due to leakage. Therefore, the effects of bulging
and gas decompression must be considered when calculating
the dynamic stress intensity factors for pipes during crack
propagation to ensure pipeline integrity.
The present study aims to use FEA to obtain the dynamic
stress intensity factor for a pipe during crack propagation. This
paper starts with computational procedures and conditions,
including a method for simulating crack propagation and gas
decompression. The following section shows the calculated
stress intensity factors obtained by FEA. The difference
between band plate and pipe as well as the effects of crack
velocity, pipe diameter, and gas decompression are shown. The
subsequent section discusses the effect of crack velocity and
meandering of a crack, which has been observed in burst tests
and actual damaged pipe. The last section summarizes the main
conclusions drawn from the current study.

2 COMPUTATIONAL PROCEDURE
2.1 Crack Propagation
Crack propagation was simulated by the node release
technique, in which the binding forces on nodes are removed
one by one. It was assumed that the crack propagates in a
straight line with constant velocity. Crack propagation was
simulated in a band plate and a pipe.
The boundary conditions of the FEA are summarized in
Figure 2. Figure 2(a) shows the conditions for a band plate and
Figure 2(b) shows the conditions for a pipe. Both in the band
plate and pipe analyses, a quarter of a whole part was modeled
under consideration of the symmetry as shown in Figure 2. In
the case of analyses for band plates, a uniform enforced
displacement was applied at a right angle to the crack
propagation direction. The value of the enforced displacement
was determined as the nominal stress on the band plate, which
corresponds to the hoop stress of the pipe analyses. Pipes
3 Copyright 2012 by ASME
pressurized by internal gas were simulated and analyzed.
Internal pressure was applied to the internal surface of the pipe
in the radial direction, and axial force caused by internal
pressure was applied to the end of the pipe in the axial
direction. Dimensions of the models are described later.
The band plate was modeled by eight-node plane stress
elements; the pipe was modeled by eight-node thin shell
elements. The material of the pipe was assumed to have linear
elastic properties with a Youngs modulus of 206 GPa, a
Poissons ratio of 0.3, and a pipe material density of 7800
kg/m
3
. Figure 3 shows the division of the FE model. As shown
in this figure, the mesh size of a part where a crack propagated
was 1.0 mm. The crack propagation was simulated using the
node release technique. The restraining forces at the nodes
were gradually released in each step; this simulated energy
absorption by a fracture and generated a precise stress field
ahead of the crack tip. To simulate this behavior, the implicit
dynamic analysis option was used. Abaqus/Standard version
6.9 software was used. The crack propagated 2 mm, which
corresponded to two elements in each step. Each step was
divided into 5 increments. It was confirmed that the analytical
results did not change even when using 20 increments in each
step.
The dynamic stress intensity factor was directly calculated
by stress distribution ahead of the crack tip. Generally, the
stress intensity factor K is described by Eq. (10)
r
K
y
t
o
2
= (10)
where o
y
is the stress component at a right angle to the crack,
and r is a distance from the crack tip in the direction of
propagation. This formula is valid under both static and
dynamic conditions [3]. The base 10 logarithm of Eq. (10) is
expressed as Eq. (11).
Figure 3 Mesh division
1 mm
x
y
C
r
a
c
k

p
r
o
p
a
g
a
t
i
o
n
Symmetry about a plane y
Symmetry about a plane x
Zero z displacement
In
te
rn
a
l p
re
ssu
re
, p
Crack propagation
by node release
Axial load corresponding
to an internal pressure on
the end of pipe
x
y
z
D
i
a
m
e
t
e
r
,

D
m
M
o
d
e
l le
n
g
th
, L
=
2
5
Dm
o
r
3
0
Dm
r
Symmetry about a plane y
Symmetry about a plane x
Zero z displacement
In
te
rn
a
l p
re
ssu
re
, p
Crack propagation
by node release
Axial load corresponding
to an internal pressure on
the end of pipe
x
y
z
D
i
a
m
e
t
e
r
,

D
m
M
o
d
e
l le
n
g
th
, L
=
2
5
Dm
o
r
3
0
Dm
r
Symmetry about a plane y
x
y
r
W
i
d
t
h
,

h
Crack propagation
by node release
Symmetry about a plane x
Model length, L=252h/t
Enforced displacement
Symmetry about a plane y
x
y
r
W
i
d
t
h
,

h
Crack propagation
by node release
Symmetry about a plane x
Model length, L=252h/t
Enforced displacement
Figure 2 Boundary conditions
(a) Band plate (b) Pipe
Figure 4 Stress distributions around crack tip
(Pipe, D
m
= 849.3 mm, Crack length a = 1.6 m)
7.5
8.0
8.5
9.0
9.5
10.0
-3.0 -2.5 -2.0 -1.5 -1.0 -0.5 0.0
log10(2r )
l
o
g
1
0
(
o
y
)
V = 400m/s
V = 600m/s
V = 700m/s
V = 800m/s
V = 1000m/s
V = 1400m/s
V = 2000m/s
r = 0.5 1.5 5 16 mm
7.5
8.0
8.5
9.0
9.5
10.0
-3.0 -2.5 -2.0 -1.5 -1.0 -0.5 0.0
log10(2r )
l
o
g
1
0
(
o
y
)
V = 400m/s
V = 600m/s
V = 700m/s
V = 800m/s
V = 1000m/s
V = 1400m/s
V = 2000m/s
r = 0.5 1.5 5 16 mm
4 Copyright 2012 by ASME
( ) K r
y 10 10 10
log 2 log
2
1
log + = t o .
(11)
Figure 4 shows an example of stress distribution obtained by
FEA. In this study, the dynamic stress intensity factor was
calculated by substituting the local stress obtained by FEA at r
= 3 mm into Eq. (11) after checking that the logarithm of the
stress distribution ahead of the crack tip was on a 1/2 line as
shown in Figure 4. It was confirmed that the dynamic stress
intensity factor obtained by this method was in good agreement
with theoretical results for flat plates [6].

2.2 Gas Decompression
Two cases were assumed for internal gas pressure. One
assumed constant internal pressure, which omitted the effect of
gas decompression caused by leakage from an opened crack.
The other simulated the effect of gas decompression. The
method for simulating gas decompression behavior was
proposed and validated by Misawa [7]. This method is
described here.
One-dimensional flow of a perfect gas in a pipe was
assumed. By assuming gas leakage from an opened crack, mass
conservation can be expressed as
( ) ( )
0 = +
c
c
+
c
c
m
x
uA
t
A
g g
|

(12)
where t is time, x is the axial coordinate of the pipe,
g
is the
gas density, u is the gas velocity, A is the cross-section area of
the pipe, | is the width of the crack opening at a cross-section
of the pipe, and m is the mass flow rate of the escaping gas.
The cross-section area A was assumed to be according to Eq.
(13)
2
2
|
.
|

\
|
+ =
t
|
t
i
R A (13)
where R
i
is internal radius of pipe. This equation is obtained by
an assumption that the radius of cross-sectional area A is
enlarged by the crack opening.
The mass flow rate of the escaping gas under a choking
condition can be expresses as
1
1
1
2

+
|
|
.
|

\
|
+
=

T R
p
m
g
(14)
where p is the internal gas pressure, R
g
is the specific gas
constant, T is the gas temperature, and is the specific heat
ratio. The choking condition can be expressed as
1
1
2

|
|
.
|

\
|

>

atm
p
p
(15)
In the case where the choking condition is not satisfied, which
means the difference between the internal pressure and
atmospheric pressure is smaller than the choking condition, the
mass flow rate of the escaping gas is expressed as
|
|
|
.
|

\
|
|
|
.
|

\
|

|
|
.
|

\
|
|
|
.
|

\
|

=
+

1 2
1
2
p
p
p
p
T R
p
m
atm atm
g
(16)
where p
atm
is the absolute external pressure, which corresponds
to atmospheric pressure.
The equation of motion is expressed as
( ) ( )
x
p
A
x
A u
t
uA
g g
c
c
=
c
c
+
c
c
2

(17)
We assume that the gas flow is isentropic, i.e., there is no heat
exchange between the gas and the pipe wall, and there is no
friction at the pipe wall. Then, for a perfect gas, is expressed
as


1
0
0
|
|
.
|

\
|
=
p
p
g g
(18)
where
g0
and p
0
are the initial density and pressure of the gas,
respectively. Note that T is expressed as a function of p under
an isentropic change condition.
Eqs. (12), (17), and (18) contain three unknown functions:
p, u, and . These functions can be solved simultaneously if
|(x,t) is given. Transitions of the three unknown functions
between each increment of the dynamic FEA as described in
the last section were calculated by solving the equations.
Table 1 Conditions
Shape Dimension Boundary condition
Crack velocity,
V [m/s] *
Maximum crack
length [m]
Band plate
2h/t = 849.3 mm,
t = 14.3 mm
Enforced displacement: o = 0.385 mm
Nominal stress = 59.4 MPa
4002000 4.2
D
m
= 849.3 mm,
t = 14.3 mm
Constant internal pressure: p = 2.0 MPa
Nominal stress = 59.4 MPa
4002000 4.2
D
m
= 310.1 mm,
t = 8.4 mm
Constant internal pressure: p = 3.0 MPa
Nominal stress = 55.4 MPa
4002000 1.6 Pipe
D
m
= 849.3 mm,
t = 14.3 mm
Gas decompression (Variable internal
pressure): p
0
= 2.0 MPa
400, 500 8.5
* Constant crack velocity was assumed in each case.
5 Copyright 2012 by ASME
Eulers method was used for integration. The calculation grid
size in the x direction was assumed to be 10 mm, and the time
step size was assumed to be 10
7
s. Composition of the internal
gas was assumed to be CH
4
:89.6%, C
2
H
6
:5.62%, C
3
H
8
:3.43%,
and C
4
H
10
:1.35%. In this composition, the specific heat ratio is
1.276, and the speed of sound is 411 m/s.
3 CONDITIONS
The FEA conditions for obtaining the dynamic stress
intensity factor are summarized in Table 1. In all cases, it was
assumed that the crack propagates at a constant velocity and in
a straight line.
Basically, the length of the entire model was 25D
m
and the
distance of crack propagation was 5 D
m
. Except for studying
the effect of gas decompression, the length of the entire model
was 30 D
m
, and the distance of crack propagation was 10 D
m
.
The analysis of the band plate was conducted as a basic
case for comparison to pipe. Enforced displacement was
Figure 5 Deformed shape and stress distribution perpendicular to the crack propagation
(Crack length a = 1.6 m, with mirror image)
Crack tip
[Pa]
(a) Band plate, 2h/t = 849.3 mm, V = 400 m/s
(b) Band plate, 2h/t = 849.3 mm, V = 800 m/s
(c) Band plate, 2h/t = 849.3 mm, V = 1400 m/s
(d) Pipe, D
m
= 849.3 mm, V = 400 m/s
(e) Pipe, D
m
= 849.3 mm, V = 800 m/s
(f) Pipe, D
m
= 849.3 mm, V = 1400 m/s
6 Copyright 2012 by ASME
applied at the top of the band plate, as shown in Figure 2(a).
Nominal stress on the band plate generated by the enforced
displacement corresponds to the nominal hoop stress of a pipe
with a middle diameter of 849.3 mm, and the width h of the
band plate in Figure 1 and Figure 2(a) corresponds to the half
the circumference of the pipe. Effects of crack velocity on the
dynamic stress intensity factor were compared between the
band plate and the pipe.
Middle diameters of 849.3 mm and 310.1 mm were
assumed in pipe analyses under constant internal pressure to
study the effects of pipe diameter. Effects of gas decompression
by leakage were studied by focusing on crack velocities of 400
m/s and 500 m/s because the speed of sound in the assumed
internal gas was 411 m/s.
4 RESULTS
4.1 Band Plate
Deformed shapes obtained by FEA are shown in Figure 5.
Examples of stress distribution around crack tips are shown in
Figures 4, which shows the results for pipes, are plots of the
logarithm of the stress distribution around the crack tip on a 1/2
line. Therefore, the stress intensity factor can be calculated by
substituting the local stress into Eq. (11).
Histories of the obtained stress intensity factors for band
plates are shown in Figure 6. As shown in Figure 6, the stress
intensity factor for the band plates remained constant with the
progress of crack propagation. The converged stress intensity
factors were less than
m
R t o
independent of crack velocity.
Figure 7 shows the relationship between the average value of
the stress intensity factor after the propagation of an 849.3 mm
crack, which corresponds to 2h/t, and the crack velocity. The
triangles are the results obtained from the dynamic FEA. For
comparison, the theoretical solution obtained by Eq. (3) is
shown in Figure 7; this solution is represented by a bold gray
line. As shown in this figure, the stress intensity factor for band
plates decreased monotonically with increased crack velocity,
and it was in good agreement with the theoretical solution.

4.2 Pipe under Constant Internal Pressure
In contrast to band plates, the stress intensity factor for
pipes with constant internal pressure exhibited complicated
behavior. Deformed shapes obtained by FEA are shown in
Figure 5(d)-(e). Histories of the obtained stress intensity factors
for pipes are shown in Figure 8. As shown in Figure 8, the
stress intensity factor of pipes did not increase monotonically
0.0
0.5
1.0
1.5
2.0
2.5
3.0
0 1 2 3 4
Crack length, a [m]
K

/

R
m

[
-
]
V = 400 m/s
V = 600 m/s
V = 800 m/s
V = 1000 m/s
V = 1400 m/s
V = 2000 m/s
K = Rm
K = a
Figure 6 Histories of Stress intensity factor for band plates
(2h/t = 849.3 mm)
0.0
0.5
1.0
1.5
2.0
2.5
3.0
400 800 1200 1600 2000
Crack velocity, V [m/s]
K

/

R
m

[
-
]
Band plate (2h/ = 849.3 mm)
K= Rm
Theoretical solution for band plate
Figure 7 Effects of crack velocity on stress intensity factor
for band plate
0.0
0.5
1.0
1.5
2.0
2.5
3.0
0 1 2 3 4
Crack length, a [m]
K

/

R
m

[
-
]
V = 400 m/s
V = 500 m/s
V = 550 m/s
V = 600 m/s
V = 700 m/s
V = 800 m/s
V = 1000 m/s
V = 1400 m/s
V = 2000 m/s
K= Rm
K= a
Figure 8 Histories of Stress intensity factor for pipes
(D
m
= 849.3 mm)
Figure 9 Effects of crack velocity on stress intensity factor
for pipes with different diameter
0.0
0.5
1.0
1.5
2.0
2.5
3.0
400 800 1200 1600 2000
Crack velocity, V [m/s]
K

/

R
m

[
-
]
Pipe (Dm = 849.3 mm, Dm/t = 59.4)
Pipe (Dm = 301.1 mm, Dm/t = 36.9)
K= Rm
Theoretical solution for band plate
7 Copyright 2012 by ASME
and oscillated. With regard to the results of crack velocity
being 550 m/s or greater, the stress intensity factor did not
surpass
m
R t o
and remained nearly constant with the
progress of crack propagation. The average value of the stress
intensity factor after the propagation of an 849.3 mm crack,
which corresponds to D
m
, is shown in Figure 9; these values are
represented by circles.
In the case where the crack velocity was 800 m/s or
greater, the stress intensity factor for pipes tended to decrease
with an increase of crack velocity, and it was in good
agreement with both the theoretical solution and the result for a
band plate. However, in the case where the crack velocity was
600 m/s or less, the stress intensity factor obtained by the FEA
was quite different from both the theoretical solution and the
result for a band plate.
Figure 9 compares the stress intensity factor for different
pipe diameters. The stress intensity factor for pipe whose
diameter D
m
is 849.3 mm is shown by circles and the stress
intensity factor for pipe whose diameter D
m
is 310.1 mm is
shown by diamonds. As shown in this figure, the stress
intensity factor for pipes with a diameter of 849.3 mm was in
good agreement with the theoretical solution for a crack
velocity of 800 m/s or higher. In contrast, the stress intensity
factor for pipes with a diameter of 310.1 mm was in good
agreement with the theoretical solution for a crack velocity of
1400 m/s or higher. That is to say, the difference in the stress
intensity factors between the theoretical solution for a band
plate and the result of the dynamic FEA for a pipe was greater
for a smaller diameter.

4.3 Pipe with Gas Decompression
Figure 10 shows histories of stress intensity factors under
constant internal pressure and under simulated gas
decompression. Figure 10(a) shows the case of a crack
propagating with a velocity of 400 m/s. The stress intensity
factors under simulated monotonic gas decompression
increased with crack propagation as compared to the case of
constant internal pressure. However, Figure 10(b) shows that in
0
20
40
60
80
100
120
140
160
180
0 2 4 6 8
Crack length, a [m]
S
t
r
e
s
s

i
n
t
e
n
s
i
t
y

f
a
c
t
o
r
,

K
[
M
P
a

m
]
0
10
20
30
40
50
60
70
80
90
D
i
f
f
e
r
e
n
c
e

o
f

K

[
M
P
a

m
]
Gas decompression
Constant internal pressure
(Difference)
0
20
40
60
80
100
120
140
160
180
0 2 4 6 8
Crack length, a [m]
S
t
r
e
s
s

i
n
t
e
n
s
i
t
y

f
a
c
t
o
r
,

K
[
M
P
a

m
]
0
10
20
30
40
50
60
70
80
90
D
i
f
f
e
r
e
n
c
e

o
f

K

[
M
P
a

m
]
Gas decompression
Constant internal pressure
(Difference)
0.0
0.5
1.0
1.5
2.0
2.5
-8 -6 -4 -2 0 2
Relative cordinate from crack tip [m]
I
n
t
e
r
n
a
l

p
r
e
s
s
u
r
e

[
M
P
a

(
g
a
u
g
e
)
]
t = 0.005 s
t = 0.010 s
t = 0.015 s
0.0
0.5
1.0
1.5
2.0
2.5
-8 -6 -4 -2 0 2
Relative cordinate from crack tip [m]
I
n
t
e
r
n
a
l

p
r
e
s
s
u
r
e

[
M
P
a

(
g
a
u
g
e
)
]
t = 0.005 s
t = 0.010 s
t = 0.015 s
Figure 10 Effects of gas decompression on stress intensity factor (D
m
= 849.3 mm, p
0
= 2.0 MPa)
Figure 11 Distribution of internal pressure (D
m
= 849.3 mm, p
0
= 2.0 MPa)
(a) Crack velocity V = 400m/s (b) Crack velocity V = 500m/s
(a) Crack velocity V = 400m/s (b) Crack velocity V = 500m/s
Crack tip Crack tip
8 Copyright 2012 by ASME
the case where the crack velocity was 500 m/s, there was no
significant difference observed in the stress intensity factors.
Internal gas distributions during crack propagation are
shown in Figure 10. The horizontal axis in these figures is a
relative coordinate whose origin is at the tip of the propagating
crack. A negative value on this axis indicates a position behind
the crack tip. Note that a surge of internal pressure observed at
the left end of each line corresponds to the x-coordinate being
zero where a symmetric boundary condition is applied. The
surge indicates a stagnation of flow generated by this
symmetric condition.
Figure 11(a) shows the result for the case where the crack
velocity was 400 m/s. The decompressed area caught up with
the crack propagation and expanded because the crack
propagation speed was subsonic. Therefore, the difference
between the stress intensity factor under constant internal
pressure and that under simulated monotonic gas
decompression increased with crack propagation. However, in
the case where the velocity was 500 m/s, Figure 11(b) shows
that the decompressed area ran ahead of the crack propagation.
Therefore, the effect of gas decompression on the stress
intensity factor was hardly observed.
That is to say, the stress intensity factor for a crack velocity
of 400 m/s under constant internal pressure, which is shown in
Figure 11(a), overestimates the stress intensity factor for actual
pipeline. However, the stress intensity factor of a crack with a
velocity of 500 m/s or greater under constant internal pressure
is valid because gas decompression does not affect the stress
distribution around the crack tip.
5 DISCUSSION
5.1 Effect of Bulging
The reason for the observed difference between pipes and
band plates under low crack velocity is discussed here.
Bulging, which is a phenomenon peculiar to pipes, can explain
the difference.
In a static condition, it is well known that bulging
increases loading on the crack tip [8]. In the case of a low crack
velocity, such as 400 m/s, it is reasonable that the stress
intensity factor of pipe was greater than that of band plate as
shown in Figure 9. However, this idea cannot explain why the
stress intensity factor was below that of band plate at crack
velocities of 600 m/s and 700 m/s.
A compressive stress in the circumferential direction
generated from bulging can provide an explanation. As
indicated by black arrow in Figure 5(d), an area whose
circumferential stress is less than the surrounding area was
observed in front of the crack tip in the case where the crack
velocity was 400 m/s. This compressive region has also been
measured in past experiments [9,10] of crack propagation in
ductile materials with a velocity of about 300 m/s. This
compressive stress in the circumferential direction was
generated because this area is under tension in longitudinal
direction due to bulging at opened area.
In the case where the crack velocity is 600 m/s or 700 m/s,
the crack tip is located in this compressed region. Therefore, it
is presumed that the stress intensity factors for pipes with crack
velocities of 600 m/s and 700 m/s were smaller than those of
band plates. Note that actual cracks propagating in pipeline
create a sinusoidal pattern [11]. Therefore, the stress intensity
factor for an actual pipe would be different from the calculated
value.
As shown in Figure 9, the difference from the theoretical
solution for a band plate was greater for a pipe with a diameter
of 301.1 mm compared to an 849.1 mm pipe. This result
indicates that the peculiar deformation of a pipe is more
prominent in a pipe with a diameter of 301.1 mm. That is to
say, the result indicates that the effect of bulging is stronger in a
pipe with a diameter of 301.1 mm. In contrast, the Folias factor,
which represents the effect of bulging under a static condition,
is generally expressed by a monotone increasing function of
a
2
/(Rt) [8], where a is equal to half the crack length. For
dynamic crack propagation, it is difficult to determine a
substantial crack length. Therefore, the relationship between
the Folias factor for a static condition and the result of the
dynamic FEA shown in this study is unclear.

5.2 Change of direction of a Crack
Actual cracks propagating in a pipeline create a sinusoidal
pattern [11]. The reason for this phenomenon is discussed here.
The sign of the non-singular T-stress has been widely used to
determine whether directional stability prevails for straight
cracks [12,13]. However, some studies have indicated the
irrelevance of the T-stress [14,15].
Radon et al. showed that the direction of the crack
propagation in a poly(methyl methacrylate) (PMMA) sheet
depends on the nominal biaxiality ratio caused by propagating
cracks under a biaxial stress condition [16]. The nominal
biaxiality ratio is the ratio of the nominal stress in the x
o
x, nominal
= R o
y, nominal
Figure 12 Crack paths observed by Radon et al. [16] in
experiments on biaxially stressed PMMA sheets
x
y
o
y, nominal
9 Copyright 2012 by ASME
direction to the nominal stress in the y direction. As shown in
Figure 12, a crack tends to deviate if this ratio increases. Using
a combination of this experimental result for a PMMA sheet
and the results of the dynamic FEA described in this study, an
explanation for the turning of a crack observed in an actual
pipeline was given in this section.
The local biaxiality ratio, which is the ratio of the local
stress in the x direction to the local stress in the y direction, was
scrutinized. The relationship between the local biaxiality ratio
and nominal biaxiality ratio was calculated using a 2-
dimensional static FEA for the PMMA sheet. The boundary
conditions are shown in Figure 13. The dimensions were
determined based on an experiment by Radon et al. [16]. Eight-
node plane stress elements were used for modeling. The mesh
size of all the elements was 1 mm. The local stress was the
stress 3 mm ahead of the crack tip.
The relationship obtained by the analysis for the PMMA is
shown in Figure 14. As shown in Figure 12, in a case where the
nominal biaxiality ratio is greater than 1.4, the crack deviates.
Applying this experimental result to the analytical result shown
in Figure 14, the crack deviates in a case where the local
biaxiality ratio is greater than about 1.2.
The local biaxiality ratio obtained by the dynamic FEA,
which simulates crack propagation for band plates and pipes,
are shown in Figure 15(a) and 15(b). In these figures, the
results using realistic crack propagation velocities in brittle
material, which is from 400 m/s to 800 m/s, were analyzed.
The local biaxiality ratio in band plates was less than 1.2 as
shown in Figure 15(a). However, the local biaxiality ratio in
pipes increased to more than 1.2 as shown in Figure 15(b). This
result indicates that a crack propagating in a pipe tends to
deviate compared to band plates. As described earlier, the
difference in the local biaxiality ratio between band plates and
pipes is generated by bulging in the pipes.
Note that although this analysis could explain the deviation
of a propagating crack, it was not able to explain the
meandering observed in actual pipelines, which means the
crack returned to the circumferential direction.
20 mm
95 mm
Symmetry about a plane y
x
y
r = 3 mm
Non-propagating crack
Symmetry about a plane x
Distributed load, o
y, nominal
Local biaxiality ratio,
R = o
x, local
/o
y, local
Distributed load,
o
x, nominal
Nominal biaxiality ratio,
R = o
x, nominal
/o
y, nominal
9
5

m
m
o
x, local
o
y, local
20 mm
95 mm
Symmetry about a plane y
x
y
r = 3 mm
Non-propagating crack
Symmetry about a plane x
Distributed load, o
y, nominal
Local biaxiality ratio,
R = o
x, local
/o
y, local
Distributed load,
o
x, nominal
Nominal biaxiality ratio,
R = o
x, nominal
/o
y, nominal
9
5

m
m
o
x, local
o
y, local
Figure 13 Crack paths observed by Radon et al. [16] in
experiments on biaxially stressed PMMA sheets
0.0
1.0
2.0
3.0
4.0
5.0
0.0 1.0 2.0 3.0 4.0 5.0
Crack length, a [m]
L
o
c
a
l

b
i
a
x
i
a
l
i
t
y

r
a
t
i
o
,

R
'
V = 400 m/s
V = 600 m/s
V = 800 m/s
V = 1000 m/s
0.0
1.0
2.0
3.0
4.0
5.0
0.0 1.0 2.0 3.0 4.0 5.0
Crack length, a [m]
L
o
c
a
l

b
i
a
x
i
a
l
i
t
y

r
a
t
i
o
,

R
'
V = 400 m/s
V = 600 m/s
V = 800 m/s
V = 1000 m/s
Figure 15 Ratio of local stress (at r = 3 mm)
(a) Band plates (2h/t = 849.3 mm) (b) Pipes (D
m
= 849.3 mm)
0.0
0.5
1.0
1.5
2.0
0.0 0.5 1.0 1.5 2.0 2.5
Nominal biaxiality ratio, R
L
o
c
a
l

b
i
a
x
i
a
l
i
t
y

r
a
t
i
o
,

R
'
deviating
Figure 14 Relationship between local biaxiality ratio and
nominal biaxiality for the PMMA sheet
going straight
10 Copyright 2012 by ASME
6 CONCLUDING REMARKS
A dynamic finite element analysis was used for calculating
the dynamic stress intensity factors for pipes and band plates
during crack propagation.
The dynamic stress intensity factors for band plates tended
to decrease monotonically with increasing crack velocity and
were in good agreement with the theoretical values that
considered the dynamic effects.
It was shown that the stress intensity factors for long
propagating cracks in pipes saturated at a certain value in the
case of a high crack velocity. However, although the stress
intensity factors for pipes were in good agreement with those of
band plates in the case of a high crack velocity, the stress
intensity factors for pipes were different from those of band
plates in the case of a low crack velocity. This result could be
explained by the effect of bulging on the stress distribution
around a crack tip. The effect of bulging was more prominent
for pipes with smaller diameters.
The effects of gas decompression, caused by leakage from
an opened crack, on the stress intensity factors were hardly
observed in the case of a high crack velocity.
The deviation of a crack, which is observed in actual
pipeline fractures, was explained by analyzing the ratio of the
longitudinal stress to lateral stress.
ACKNOWLEDGMENTS
The authors thank Ms. Kimiko Keduka of Japan Open
Systems Corporation for her assistance with the FE analyses.
REFERENCES
[1] Maxey, W.A., Kiefner, J.F. and Eiber, R.J., Brittle
Fracture Arrest in Gas Pipelines, Battelle report to
Pipeline Research Committee of American Gas Association
(currently PRCI), PRCI Report number L51436e, July
1975.
[2] Irwin, G.R. and Corten, H.T., Evaluating the Feasibility of
Basing Pipeline Operating Pressure on In-Place
Hydrostatic Test Pressure, Report to Northern Natural Gas
Company and the El Paso Natural Gas Company, 1968.
[3] L.B. Freund, Dynamic Fracture Mechanics (Cambridge
Monographs on Mechanics), Cambridge University Press,
1998.
[4] Emery, A.F., Kobayashi, A.S., Love, W.J., Place, B.W., Lee,
C. and Chao, Y.H., An experimental and analytical
investigation of axial crack propagation in long pipes,
Engineering Fracture Mechanics, Volume 23, Issue 1,
pp.215-226, 1986.
[5] Cosham, A., Jones, D.G., Eiber, R. and Hopkins, P., Don't
drop the drop weight tear test, Proceedings of Pipeline
Technology Conference, Ostend2009-090, 2009.
[6] Ando, T. and Aihara, S., Fundamental Investigation on the
Evaluation of Brittle Crack Arrest Toughness of Steel
Plates, CAMP-ISIJ vol. 21, 1434, 2008. (in Japanese)
[7] Misawa, K., Imai, Y. and Aihara, S., A New Model for
Dynamic Crack Propagation and Arrest in Gas Pipelines,
Proceedings of the 8th International Pipeline Conference,
IPC2010-31475, ASME, 2010.
[8] Kiefner, J.F., Maxey, W.A., Eiber, R.J., and Duffy, A.R.,
Failure stress levels of flaws in pressurized cylinders,
ASTM STP 536, 1973.
[9] Ives, K.D., Shoemaker, A.K. and McCartney, R.F., Pipe
deformation during a running shear fracture in line pipe,
Journal of Engineering Materials and Technology, Volume
96, Issue 4, pp.309-317, ASME, 1974.
[10] Shoemaker, A.K. and McCartney, R.F., Displacement
consideration for a ductile propagating fracture in a line
pipe, Journal of Engineering Materials and Technology,
Volume 96, Issue 4, pp.318-322, ASME, 1974.
[11] Eiber, R.J., Fracture Propagation, Paper I, Fourth
Symposium on Line Pipe Research, AGA, Catalog No.
L30075, 1969.
[12] Marder, M., Instability of a crack in a heated strip,
Physical Review E 49, R51-54, 1994.
[13] Smith, D.J., Ayatollahi, M.R. and Pavier, M.J., The role of
T-stress in brittle fracture for linear elastic materials under
mixed-mode loading, Fatigue and Fracture of Engineering
Materials and Structures, Volume 24, Issue 2, pp.137-150,
2001.
[14] Melin, S., The influence of the T-stress on the directional
stability of cracks, International Journal of Fracture,
Volume 114, Number 3, pp.259-265, 2002.
[15] Yang, B and Ravi-Chandar, K., Crack path instabilities in
a quenched glass plate, Journal of the Mechanics and
Physics of Solids, Volume 49, Issue 1, pp.91-130, 2001.
[16] Radon, J.C., Lever, P.S. and Culver, L.E., Fracture
Toughness of Pmma Under Biaxial Stress, International
Conference of Fracture, Volume 4, No. 3B, pp.1113-1118,
1977.

S-ar putea să vă placă și