Professional Documents
Culture Documents
A Study On Seismic Behavior of Buried Steel Pipelines Crossing Active Faults
A Study On Seismic Behavior of Buried Steel Pipelines Crossing Active Faults
No other uses without permission. Copyright (c) 2012. American Society of Civil Engineers. All rights reserved.
Pipelines 2010: Climbing New Peaks to Infrastructure ReliabilityRenew, Rehab, and Reinvest
2010 ASCE
Pipelines 2010
Downloaded from ascelibrary.org by North Carolina State University on 10/07/12. For personal use only. No other uses without permission. Copyright (c) 2012. American Society of Civil Engineers. All rights reserved.
Pipelines 2010: Climbing New Peaks to Infrastructure ReliabilityRenew, Rehab, and Reinvest
2010 ASCE
account. The most important conducted researches on this issue are as follows
(Ziatohidi 2004).
Newmark studied the pipe response toward fault movements. To do so, he ignored
the deformed curve of pipe in the vicinity of a fault and also the effects of soil-pipe
interaction. These assumptions had led to a fallacy of estimations about pipe
response toward fault movements. Then, by applying some adjustments, other
researchers had improved on the Newmark method (Newmark 1975). In models
presented by Kennedy, the pipeline demonstrated by beam elements and pipe
deformation curve in the fault point of intersection was considered to be uniform
(Newmark 1975). Later, Chiou and Wang considered this curve as a variable and
situations dependant issue separately (Chiou 1994), (Wang 1985).
Takada claimed that the pipe has shell behavior due to fault movement (Takada
2001). Although the accuracy of these studies for estimating the pipe responses
improved progressively, they did not consider soilpipe interaction effects on the
pipe response. Regarding the critical importance of interaction on the pipe response,
the models used in these studies needed to be adjusted; of course, they suggested the
necessity of completion and adjustment of their studies.
MODELING, MATERIAL BEHAVIOR AND LOADING
ABAQUS is powerful finite element software which is able to simulate the
pipeline crossing a fault. These simulations are composed of three main sections.
Two of them demonstrate the fault and its plane, and the third deals with the steel
pipe (Figure 1). Following initial geometrical modeling and meshing corresponding
to fault behavior, and also supposing that in points where there are more stresses, we
should use tinier meshing, several elements were dedicated to the model. The soil
simulated by solid elements and steel pipes was modeled according to its nature and
behavior with shell elements. These elements are able to assign non-linear materials
and geometry to them and they can go through various defined analyses in the
software.
After modeling and allocating the desired elements, we should define failure criteria
for them. A perfect three-line stress-strain curve was utilized for steel used in the
pipes. This curve uses Mises yield surface and Associated Flow Rule for its
performance and the steel type, which derived from API guidelines, was X-65
(Bolvardi 2008).
Drucker-Prager plasticity model is also dedicated to solid elements which represent
soil around the pipe. This model is an approximate version of the Mohr-Columb
model and has an elastoplastic behavior so that during consecutive loadings and
unloadings the limit yield of the model would increase or decrease, respectively. The
Associated Flow Rule for isotope material was implemented in this plasticity model,
as well (Sadrnejad 2000). Of course, based on presented assumptions by the
software, this class of models has three yield surfaces including Linear, Hyperbolic
and General Exponent forms that are used for granular materials and their main
difference is in their yielding form. In this study, we used the linear yield surface of
the Drucker-Prager model (Bolvardi 2008).
Pipelines 2010
Downloaded from ascelibrary.org by North Carolina State University on 10/07/12. For personal use only. No other uses without permission. Copyright (c) 2012. American Society of Civil Engineers. All rights reserved.
Pipelines 2010: Climbing New Peaks to Infrastructure ReliabilityRenew, Rehab, and Reinvest
2010 ASCE
With respect to the static nature of the large ground motions (faulting), the main
applied load to the model is in the form of a static displacement. Weights of pipe and
soil can be calculated and implemented automatically by entering their specific
weights.
In order to develop stability in the model by means of defining boundary conditions,
all degrees of freedom at the bottom of the soil layer are closed in the Initial Step and
in the static analysis, one of the transitive degrees of freedom is kept open and the
amount of desired displacement would occur. Therefore, the free part of the model
will displace solidly and rupture will be occurred. Regarding the static analysis, there
is no need for applying boundary conditions for surrounding elements because they
have no effect on the pipe response (Bolvardi 2008).
SOIL-PIPE INTERACTION
In a few studies conducted for analysis of pipelines, they were modeled in
two-dimensional form. In those models, by considering the stable and steady curve in
deformed pipe and determining friction force between soil and pipe through a trial
and error method on length of pipe and finally applying those, the pipe response to
displacement and effective parameters have been studied. No interactional forces
were applied to the model and results were obtained with improper approximation. In
some recent studies, 3D simulation has been used. In these studies, the interactional
forces were applied in such a way that, in three directions of axial, horizontal and
vertical around the pipe, several springs with non-linear stiffness similar to the soil
properties, were implemented. In these models, in order to consider interactional
forces through calculation of anchored length (the length in which the pipe in both
sides of the fault is in contact with the soil and interaction is running between them),
the equivalent friction forces were applied to interactional forces instead. This
method certainly affects the accuracy of the pipe responses and decreases them to
what extent.
Here, with respect to capabilities of the software we can claim that for the simulation
of soil and pipe interaction, there is no need to use springs and simplification.
There are two methods for contact modeling between two objects. The first one is
using contact elements between surfaces that are potentially capable to come in
contact. For example, if two objects or two sections of an object have the capability
of interaction, we should define Contact Elements in the contact points for specific
analyses. Of course, it is noteworthy that it is possible to use these elements in two
and three dimensional models.
The second method for contact modeling between two objects is through using
Contact Pairs. This method is more accurate and more useful compared to the
previous methods. Here, in the first step, the sections with contact capability are
recognized and a surface is defined separately for each section. The surfaces are
intended to present sections of two objects which are in contact with each other in
such a way that these surfaces could take different geometrical shapes. This kind of
interaction is called Surface to Surface Contact. In this study, considering the shell
elements of the pipe and solid elements of the soil, there are two surfaces for
applying contact: One is the outer surface of pipe and the other one is the inner
Pipelines 2010
Downloaded from ascelibrary.org by North Carolina State University on 10/07/12. For personal use only. No other uses without permission. Copyright (c) 2012. American Society of Civil Engineers. All rights reserved.
Pipelines 2010: Climbing New Peaks to Infrastructure ReliabilityRenew, Rehab, and Reinvest
2010 ASCE
surface of soil outside the pipe. Thus, we define two surfaces to those areas and then
for this case, Normal contact and Fricative contact are assigned simultaneously. The
friction coefficient between pipe and soil for Fricative contact is defined equal to
0.15, based on an estimation of friction between soil and steel. Besides, the Normal
contact is defined through considering the nature of pipe and soil interaction,
supposing that there is no contact between soil and pipe before fault rupture and also
assuming that with regard to soil properties, there is no tensile contact between soil
and pipe at the time of rupture. In the other words, when there is no interaction
between soil and pipe, the Normal contact with Separation ability will be used.
As it was mentioned before, for modeling the fault plane, we should define a suitable
interaction between two parts of soil, as well. Similar to the former one, we will first
recognize two soil sections in the contact point and then we will assign surfaces to
them. Then we should assign a Fricative contact with a coefficient of 0.7 and Hard
Normal contact while assuming No Separation between two soil sections.
Meanwhile, fault rupture would not happen unless the excitation forces on the fault
reach a critical value, we can define a critical shear stress (critical) which halts rupture
before reaching that point.
At the end of this step, the simulation of soil-pipe interaction and fault plane, which
approximately present the real behavior of a fault will terminate. In the simulation,
the pipe response to fault movement was studied more accurately and results of
different parameters effects were derived with proper approximation. Therefore, the
simulation of this software indicates a considerable superiority compared to previous
methods (Bolvardi 2008).
ANALYSIS AND STUDY OF PARAMETERS
Following a suitable simulation, we should investigate the parameters which
influence the pipe response. These parameters, which considerably influence the
performance of the pipe against fault movements, include: pipe diameter, pipe
thickness, pipe burial depth, intersection angle between the pipe axis and fault plane
and pipe anchored length.
With respect to the fact that the pipe mass is negligible compared to its stiffness, we
can say that the inertia forces produced by pipe mass in the pipeline is negligible
compared to the force which is proportional with stiffness of the soil-pipe system;
therefore, the natural period of system is very small. Since the natural period of the
system is very small compared with the time of movement resulting from the fault
activity, utilization of a static analysis for investigating buried pipe behavior against
large ground motions seems to be logical. In studies so far, the dynamic response of
the pipeline is considered very small compared with the pipe deformation resulting
from fault movement. Taking that point into account, in this study we supposed that
the non-linear statical analysis is able to exhibit suitable and satisfactory results. In
other words, we will investigate the effects of large ground motion on pipelines.
Generally, as the result of fault movement, the pipeline will face compressive and
tensile stress depending on fault type and intersection angle between pipe axis and
fault plane. Normal and Strike slip faults usually lead to tensional stress in the
pipeline, while the Reverse faults result in compressive stress in the pipe wall and
Pipelines 2010
Downloaded from ascelibrary.org by North Carolina State University on 10/07/12. For personal use only. No other uses without permission. Copyright (c) 2012. American Society of Civil Engineers. All rights reserved.
Pipelines 2010: Climbing New Peaks to Infrastructure ReliabilityRenew, Rehab, and Reinvest
2010 ASCE
lead to local buckling. In this study, we have done sensitivity analysis on the effect of
parameters on the response of buried pipes against Normal and Strike slip faults.
Pipe diameter. As the Strike slip fault displacement results reveal (Figure 3), the
axial strain has slight changes but the principal strain shows a decrease. This
decrease is caused by the subtractive effect of the pipe flexural stiffness increase on
the soil-pipe interactional force, which leads to a slight increase in the pipe response.
This analysis can also be represented in accordance with the approximate similarity
in the results which are derived from the pipe response to the provocations of the
Normal fault (Figure 4).
Note that in the modeling carried out by Newmark and Kennedy, the effect of pipe
diameter on the pipe response capacity has been estimated as insignificant due to
excluding the flexural stiffness in the modeling process; and the pipe resistance was
obtained proportional to its diameter (Newmark 1975), (Kennedy 1973).
Pipe thickness. Increasing pipe thickness in the crossing of the Strike slip fault has
led to a significant decrease in the principal strain and also a slight increase in the
axial strain. This pipe response results from the subtractive effect of the pipe stiffness
increase, which itself is due to the thickness increase on soil-pipe interactional forces
(Figure 5). With regard to the results shown in the figure, by the flexural stiffness
increase we mean the increase of the pipe response due to the thickness increase. Its
also notable that the effects of the pipes axial forces have been ignored because they
have no influence on the principal strain.
The Normal faults behavior against the pipe thickness increase is similar to that of
the Strike slip fault. Due to an increase in the pipe thickness, the principal and axial
strains have shown a decrease and a slight increase, respectively. It is noteworthy
that the values of this decrease and increase are far less than those of the Strike slip
fault, so that the graphs slope has changed at a lower rate in a smaller numerical
interval (Figure 6). Moreover, this figure shows that in the lesser thicknesses, the
decreasing rate of the principal strain is greater, so considering the changes of strain
and axial force, its effects on the pipe response can be ignored.
Burial depth. Reaction of the buried pipes to the 3D movements of the Normal and
Strike slip faults due to increase of the pipe burial depth reveal that the increase of
the burial depth would lead to a rise of axial force and flexural moment and to an
increase in the principal and axial strains in both fault types, as well (Figures 7&8).
Additionally, results exhibit that as the pipe burial depth increases, the effect of the
soil-pipe interactional force increases, with different slopes and in different intervals
within each of the faults, and therefore, the pipe response decreases.
Intersection angle between the pipe axis and fault plane. Rise of the pipes
intersection angle up to 90 degrees crossing the Strike slip fault shows that the more
the intersection angle approaches 90 degrees, the less the axial force and axial strain
and the more the flexural force will become in the pipe, but there is not any observed
obvious change in the principal strain (Figure 9). This indicates that by increasing the
Pipelines 2010
Downloaded from ascelibrary.org by North Carolina State University on 10/07/12. For personal use only. No other uses without permission. Copyright (c) 2012. American Society of Civil Engineers. All rights reserved.
Pipelines 2010: Climbing New Peaks to Infrastructure ReliabilityRenew, Rehab, and Reinvest
2010 ASCE
Pipelines 2010
Downloaded from ascelibrary.org by North Carolina State University on 10/07/12. For personal use only. No other uses without permission. Copyright (c) 2012. American Society of Civil Engineers. All rights reserved.
Pipelines 2010: Climbing New Peaks to Infrastructure ReliabilityRenew, Rehab, and Reinvest
2010 ASCE
effects of the resistance forces on the pipe which itself results from the change of the
pipe angle effect on the soil-pipe interactional forces (Figure 13).
Comparison of results. To compare the results, some models should be made
which would resemble the desired models of the researchers in this field. Therefore,
considering the current results derived from other researches, it was possible to
compare only three parameters of pipe diameter, burial depth, and the intersection
angle between the pipe axis and the fault plane in the software-made model with the
models studied by Wang and Chiou.
In the first model, the changes of the axial force and principal strain were compared
with the similar results of those researchers studies. It is noteworthy that the major
differences between the software-made model and the models have been studied by
researchers are soil-pipe interaction, so that they either have replaced the
interactional force with the friction force in order to simplify the model or used the
spring with the soil stiffness. Then if there is a difference in the comparison of the
results derived from the software model and those models, it is certainly due to the
precision in the simulation which is closer to the reality.
The axial force in the software-made model has a descending trend while in the
Wang and Chiou models it has a rising trend which, of course, there is an
insignificant difference between their values. And the difference is seen in the force
value interval between software-made model and other studied models, the reason
for that is the more real modeling (Figures 14&15). The same analysis can be applied
for the principal strain with the difference that their value interval is almost identical
(Figure 16).
To compare the pipe burial depth parameter, the same moldings have been
accomplished as above. The figures show constant values of the pipe axial force
against different values of pipe burial depth. It also indicates that the pipe axial force
in the software-made model does not depend on the changes of the pipe burial depth,
contrary to the other studied models (Figures 17&18). The principal strain also has a
trend similar to that of the Chiou model although its inclination is different (Figure
19).
Finally, the parameter of the intersection angle between the pipe axis and the fault
plane is being compared. Investigation of the axial forces of the software-made
model and Wang and Chiou models reveals that there is a decrease in both of them,
but their value intervals are different due to the mentioned reason; and moreover, this
difference is slightly observed in the value intervals of the Wang and Chiou models,
the reason of which is the modeling quality (Figures 20&21). The principal strain in
the software-made model shows a decrease which is in common with the results of
the Chiou model, up to 60 degrees; but more than this degree, the principal strain in
the Chiou model increases, contrary to the software-made model (Figure 22); (Wang
1985), (Chiou 1994).
CONCLUSION
In the end, according to the aforementioned results preserved on different
effective parameters on pipelines, we can assert the summary of this study. First, to
Pipelines 2010
Downloaded from ascelibrary.org by North Carolina State University on 10/07/12. For personal use only. No other uses without permission. Copyright (c) 2012. American Society of Civil Engineers. All rights reserved.
Pipelines 2010: Climbing New Peaks to Infrastructure ReliabilityRenew, Rehab, and Reinvest
2010 ASCE
the point that the economical issues allow, an increase in the pipe diameter would
create no problem and leads to the rise of the pipe response capacity in both type of
faults. Then, using greater pipe thickness will lead to an increase in the pipe stiffness
and, subsequently, the rise of the created resistance force in the pipe and of the pipe
response. Its notable that this increase in the low thickness range is more significant
than that of the greater ones. After that, decreasing the pipe burial depth will lead to
an increase in the pipe response. Although in order to decrease the pipe burial depth
other issues such as the weather should also be taken into consideration, since the
pipe burial depth can be decreased up to a certain depth which more than that may
lead to some other problems such as freezing, burst, etc. due to the inappropriate
coverage over the pipe. Moreover, the more the intersection angle between pipe and
the fault plane approaches 90 degrees, the greater the effect of the created resistance
forces on the interactional forces gets and, consequently, the greater the pipe
response will be. Next, the results derived from the anchored length comparison of
two studied faults indicate that by the smaller diameters, the larger thicknesses, and
the angle closer to 90 degrees, an appropriate anchored length can be selected.
Finally, the most important reason regarding that the results of the Wang and Chiou
studies are more conservative than those of the software-made model is the modeling
quality of the software model and also the hypotheses which make the modeling
closer to the actual conditions.
ACKNOWLEDGEMENT
I would like to express the appreciation to the National Iranian Gas Company for in
part financial support.
REFERENCES
Bolvardi, V. (2008). A Study on Seismic Behavior of Buried Steel Pipelines
Crossing Active Faults. MS. Thesis, Sharif University of Technology, Tehran.
Chiou, Y. J., Chi, S. Y. (1994). A Study on Buried Pipeline Response to Fault
Movement. Journal of pressure Vessel Technology, ASCE, Vol. 116.
Hosseini, M., Tiv, M. (1995). Guideline for the Seismic Design of Oil and Gas
Pipelines System. IIEES, Tehran.
Kennedy, R. P., Chow, A. W. (1973). Fault Movement Effects on Buried Oil
Pipeline. Transportation Engineering Journal, ASCE, 617-633.
Newmark, N. M., Hall, W. J. (1975). Pipeline Design to Resist Large Fault
Displacement. Proc. of the U.S. National Conf. on Earthquake Engineering,
Oakland, 416-425.
Sadrnejad, S. A. (2000). Principles of Soil Plasticity. K. N. Toosi University of
Technology Publication, Tehran, Iran.
Takada, S., Hassani, N. (2001). A New Proposal for Simplified Design of Buried
Steel Pipes Crossing Active Faults. Earthquake Engineering and Structural
Dynamics Journal, Vol. 30, 1243-1257.
Pipelines 2010
(a)
0.16
0.14
0.12
Emax
E11
0.1
0.08
0.06
0.04
0.02
0
40
60
80
100
120
140
P/Pu
0.3
M/Mu
0.002
0.001
0.2
0
0.1
160
Diameter (cm)
-0.001
0
40
60
80
100
120
140
Moment Ratio
0.003
0.4
0.2
0.18
Max. Strain
Downloaded from ascelibrary.org by North Carolina State University on 10/07/12. For personal use only. No other uses without permission. Copyright (c) 2012. American Society of Civil Engineers. All rights reserved.
Pipelines 2010: Climbing New Peaks to Infrastructure ReliabilityRenew, Rehab, and Reinvest
2010 ASCE
-0.002
160
Diameter (cm)
Pipelines 2010
(b)
(a)
0.19
0.005
0
-0.005
-0.01
Emax
-0.015
P/Pu
0.01
Max. Strain
0.0045
E11
60
80
100
120
140
0.0035
0.0025
0.09
0.0015
0.04
-0.02
40
M/Mu
0.14
Moment Ratio
0.015
0.0005
-0.01
160
40
60
80
Diameter (cm)
100
120
-0.0005
160
140
Diameter (cm)
(a)
E11
10
12
14
16
18
0.45
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
0.01
0.008
0.006
P/Pu
0.004
M/Mu
0.002
8
20
Moment R atio
Emax
Max. S train
0.2
0.18
0.16
0.14
0.12
0.1
0.08
0.06
0.04
0.02
0
10
12
14
16
18
20
Thickness (mm)
Thickness (mm)
(b)
Max. Strain
0.003
0.0025
0.002
0.0015
0.001
Emax
0.0005
E11
0
8
10
12
14
16
18
0.45
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
0.01
0.008
0.006
P/Pu
0.002
8
20
0.004
M/Mu
Moment Ratio
0.0035
10
12
Thickness (mm)
14
16
18
20
Thickness (mm)
(b)
Max. S train
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
Emax
0.2
E11
0.15
0.1
0.05
0.014
0.012
0.01
0.008
P/Pu
50
0
50
70
90
110
130
150
170
190
100
150
0.006
M/Mu
0.25
0.004
250
200
210
Figure 7. Effect of various pipe burial depths crossing Strike slip fault,
a) on pipe strain; b) on pipe axial force and flexural moment.
(b)
Emax
E11
0.005
0.32
0.004
0.28
0.003
0.24
0.002
P/Pu
M/Mu
0.2
-0.005
50
70
90
110
130
150
170
190
210
Moment Ratio
(a)
0.01
Max. S train
Downloaded from ascelibrary.org by North Carolina State University on 10/07/12. For personal use only. No other uses without permission. Copyright (c) 2012. American Society of Civil Engineers. All rights reserved.
Pipelines 2010: Climbing New Peaks to Infrastructure ReliabilityRenew, Rehab, and Reinvest
2010 ASCE
0.001
50
100
150
200
250
Pipelines 2010
10
(a)
(b)
3.00E-01
1.50E-01
1.00E-01
5.00E-02
40
60
80
100
120
0.025
M/Mu
0.02
0.015
0.01
0.005
0.05
0
0.00E+00
20
P/Pu
0.25
0.2
0.15
0.1
20
140
40
60
80
100
Moment R atio
Max. S train
E11
2.00E-01
0.03
0.35
0.3
Emax
2.50E-01
0
140
120
Figure 9. Effect of various intersection angles bet. pipe and Strike slip fault,
a) on pipe strain; b) on pipe axial force and flexural moment.
(a)
(b)
Max. S train
0.0025
0.002
0.0015
0.001
Emax
0.0005
E11
0.01
0.2
0.18
-0.01
-0.02
0.16
-0.03
0.14
M/Mu
20
30
40
50
60
70
80
90
-0.04
0.12
0
20
P/Pu
30
100
40
50
60
70
80
90
0.003
-0.05
100
Figure 10. Effect of various intersection angles between pipe and Normal fault,
a) on pipe strain; b) on pipe axial force and flexural moment.
60
50
40
30
20
STR.
10
NOR.
0
40
60
80
100
120
140
70
45
80
70
60
50
40
30
20
STR.
10
NOR.
160
10
11
12
13
14
15
16
17
18
0.64
0.54
0.44
0.34
0.24
ABAQUS
0.84
CHIOU
60
80
110
140
80
100
120
CHIOU
2.04
0.44
0.34
0.24
1.54
1.04
0.54
20
40
60
80
100
120
Diameter (cm)
Diameter (cm)
ABAQUS
2.54
0.64
0.54
0.04
40
50
0.74
0.04
20
NOR.
3.04
0.94
0.14
0.14
0
STR.
20
E m ax.
WANG
25
20
1.04
ABAQUS
30
19
0.94
0.74
35
Thickness (mm)
0.84
40
15
Diameter (cm)
Downloaded from ascelibrary.org by North Carolina State University on 10/07/12. For personal use only. No other uses without permission. Copyright (c) 2012. American Society of Civil Engineers. All rights reserved.
Pipelines 2010: Climbing New Peaks to Infrastructure ReliabilityRenew, Rehab, and Reinvest
2010 ASCE
0.04
0
20
40
60
80
100
120
Diameter (cm)
Pipelines 2010
Figure 16.
Comparison of pipe
max. strain in
ABAQUS and Chiou
model at different
pipe diameters.
11
3.04
1.24
WANG
0.84
0.64
0.44
0.24
ABAQUS
1.04
50
100
150
0.44
50
100
150
200
0.04
0
0.64
0.54
0.44
0.34
0.24
0.14
0.04
60
200
ABAQUS
CHIOU
1.04
2.04
0.84
0.64
0.44
80
100
CHIOU
1.54
1.04
0.24
0.54
0.04
40
150
2.54
ABAQUS
1.24
E m ax.
WANG
20
100
1.44
ABAQUS
50
1.04
0.74
1.04
0.54
200
Figure 17.
Comparison of pipe
axial force in
ABAQUS and Wang
model at different
pipe burial depths.
0.84
1.54
0.24
0.94
CHIOU
2.04
0.64
0.04
0.04
ABAQUS
2.54
CHIOU
0.84
E max.
ABAQUS
1.04
1.24
Downloaded from ascelibrary.org by North Carolina State University on 10/07/12. For personal use only. No other uses without permission. Copyright (c) 2012. American Society of Civil Engineers. All rights reserved.
Pipelines 2010: Climbing New Peaks to Infrastructure ReliabilityRenew, Rehab, and Reinvest
2010 ASCE
20
40
60
80
100
0.04
0
20
40
60
80
100
Figure 20.
Comparison of pipe
axial force in
ABAQUS and Wang
model at different
intersection angles.
Pipelines 2010
12