Beruflich Dokumente
Kultur Dokumente
DOI 10.1007/s10518-013-9430-y
ORIGINAL RESEARCH PAPER
Received: 27 May 2012 / Accepted: 27 January 2013 / Published online: 14 February 2013
Springer Science+Business Media Dordrecht 2013
Abstract A buried pipe extends over long distances and passes through soils with different
properties. In the event of an earthquake, the same pipe experiences a variable ground motion
along its length. At bends, geometrically a more complicated problem exists where seismic
waves propagating in a certain direction affect pipe before and after bend differently. Studying
these different effects is the subject of this paper. Two variants for modeling of pipe, a beam
model and a beam-shell hybrid model are examined. The surrounding soil is modeled with
the conventional springs in both models. A suitable boundary condition is introduced at the
ends of the system to simulate the far field. Effects of angle of incidence in the horizontal
and vertical planes, angle of pipe bend, soil type, diameter to thickness ratio, and burial
depth ratio on pipe strains at bend are examined thoroughly. It is concluded that extensional
strains are highest at bends and these strains increase with the angle of incidence with the
vertical axis. The pipe strains attain their peaks when pipe bend is around 135 and exceed
the elastic limit in certain cases especially in stiffer soils, but remain below the rupture limit.
Then equations for predicting the seismic response of the buried pipe at bend are developed
using the analytical data calculated above and regression analysis. It is shown that these
semi-analytical equations predict the response with very good accuracy saving much time
and effort.
Keywords Buried pipe Bend Wave propagation Hybrid model Numerical analysis
Semi-analytical equation
123
1374
1 Introduction
Categorizing piping systems among lifelines demonstrates importance of their appropriate
behaviour in maintaining societies safety and health. Because of its extension over a vast
region where it might cross active faults or liquefiable soils, a piping system is prone to a larger
risk compared to other facilities occupying small areas. Failure potential in earthquakes in
such a problem is quantified using permanent and transitional ground deformations. While the
permanent ground deformations generally originate from faulting, liquefaction, and sliding,
the transitional deformations are produced by seismic waves propagating in the soil medium
(Dash and Jain 2008).
Incidence of seismic waves on buried pipes can result in failures mainly as excessive
tension and compression, and local and general buckling. As the seismic waves propagate in
ground, every two particles of soil vibrate out of phase. Due to this phenomenon and interaction of soil and pipe at their interface, bending and longitudinal strains develop along pipe.
Investigation on the seismic behavior of buried pipes dates back at least to 1967 when Newmark and his colleagues presented an approximate analysis method under wave propagation
(Newmark 1968; American Society of Civil Engineers 1984; Berrones and Liu 2003). Afterwards, various analytical and numerical procedures have been developed and tested by different researchers. Within the closed form solutions derived, only a few are for bends and the
rest are specific to straight pipes. As of the analytical methods, Newmark ignored the effects
of the inertia force and pipe-soil interaction and assumed equal strains for ground and pipe.
Sakurai and Takahashi studied the role of inertia force on dynamic response of straight
pipes and concluded that it was unimportant compared to other factors (Sakurai and Takahashi
1969; ORourke and Liu 1999). Hindy and Novak (1979) modeled pipe as a continuous system
consisting of lumped masses resting on springs and calculated the response of such a system.
Other studies on buried straight pipes were those of Shinozuka and Koike; El Hmadi and
ORourke in which the response equation proposed by Takahashi and Sakurai was modified
by waving the assumption of equal strains for common points of pipe and ground (Datta
1999; ORourke and El Hmadi 1988; Stamos and Beskos 1995). Up to this time, analyses
of pipes were done using equivalent beam models. Altering the trend, Muleski and Ariman
(Stamos and Beskos 1995), Datta et al. and Wang et al. (Datta 1999), Takada and Tanabe
(1987), Takada and Higashi (1992), Takada and Katagiri (1995), Kouretzis et al. (2006), and
Halabian et al. (2008), among others used a shell model for pipe, in which pipe was modeled
with an isotropic elastic cylindrical thin shell making possible determining the response also
around the section. In addition, the behaviors of flexible and rigid straight pipelines have
been studied by ORourke et al. (2004), Wang et al. (2006) and Shi et al. (2008).
As of studying behavior of pipe bends, Shah and Chu (1974) and Goodling (1983) are
pioneers. They studied rigid and flexible bends and derived closed-form solutions (American
Society of Civil Engineers 1984; ORourke and Liu 1999). Other researchers like Shinozuka
and Koike assessed the same problem using a beam-on-elastic-foundation (Winkler) model
with giving their own analytical solution (ORourke and Liu 1999). In parallel, numerical
methods were employed to investigate the bent zone. Saionji and Taguchi (Takada et al. 1992),
Ogawa and Koike (2001), Mclaughlin and ORourke (2003, 2009), Lee et al. (2009), derived
numerical solutions for bends under seismic waves using Winkler models and calculated
elastic and plastic strains at bends. The response of an internally and externally pressurized
right angle elbow subjected to in-plane and out-of-plane was also investigated by Karamanos
et al. (2003), Karamanos et al. (2006) and Pappa et al. (2008).
Majority of the works cited in the relevant literature consist of a quasi-static two dimensional approach for analysis of pipe against incidence of waves. As a result, a complex
123
1375
three-dimensional (3D) time history dynamic analysis varying the incidence wave angle and
elbow angle in order to investigate the buried pipe behavior in various soil types is rare in the
literature thus needing more attention. Effect of soil type and geometrical specifications are
among important aspects having key roles in dynamic response of buried pipe bends. These
are the main ingredients of this research using numerical analysis on beam and beam-shell
hybrid models. Regarding the considerable time and effort necessary for developing such
complicated models, a simple regression based equation is desirable to predict the response
of buried pipes under wave propagation easily. This is another achievement of the current
study.
123
1376
Diameter
D (mm)
Thickness Bent
t (mm)
radius
(R)
Burial
depth
(m)
Mass
density
(Kg/m3)
Elastic
modulus
E(Pa)
Poisson
ratio
()
Yield
stress
y (Pa)
Steel
APIX65
400
9.5
1.5
7850
210e+9
0.3
465.4e+6 517.7e+6
3D
Ultimate
stress
u (Pa)
Specific
weight
(KN/m3 )
Internal
friction
angle (deg)
Pipe-soil
friction
angle (deg)
K0
Average shear
wave velocity
VS (m/s)
Loose sand
14
28
17
0.53
75
Medium sand
18
35
21
0.43
220
Dense sand
22
45
27
0.3
450
N70
Soil type
Specific
(KN/m3 )
Soft clay
16
10
02
75
Medium clay
18
50
610
220
Stiff clay
21
200
2030
450
Undrained shear
strength Su (KPa)
123
1377
Axial
(KN/m/m)
stiffness
Transverse-horizontal
stiffness (KN/m/m)
Loose sand
1,241.56
262.5
Medium sand
2,443.57
1,151.47
1,732.62
Dense sand
5,405.5
4,037.647
3,300
Soft clay
1,721
Medium clay
Stiff clay
305.88
7,585.3
1,911.76
18,551.18
10,196.078
653.33
80
533.33
3,200
Downward
2,538.66
8,654.4
24,970
333.33
2,000
10,000
The interaction of pipe and soil can be expected to have an essential effect on pipe response
against incident waves. In this research the behavior of soil around pipe is modeled in this
research using a number of springs extending in three perpendicular directions with respect
to pipe, as shown in Fig. 1. Apparently, this modeling is of Winkler type, differing in the fact
that the springs are compression-only. This condition is maintained by using such springs
on opposite sides of each node of the model. Characteristics of the soil springs are adopted
from American Lifeline Alliance (American lifelines alliance 2001).
The loaddisplacement behavior of the springs in longitudinal, transverse-horizontal and
transverse-vertical directions are defined with force-displacement curves, as shown in Fig. 2,
in which Fu and Xu are peak interaction force per unit length and yield displacement,
respectively. Although the longitudinal and transverse-horizontal springs have a symmetrical response relation (same response regardless of direction of movement), the transversevertical spring has an unsymmetrical response relation. That is, evaluation of the soil-pipeline
interaction for downward and upward movement must be performed separately. As is seen in
Fig. 2, an elasto-plastic behavior is assumed for the springs, enabling them to simulate the slip
of pipe in soil. The stiffness characteristics of soil springs used in this paper are as of Table 4.
Using the force-displacement relations of the type shown in Fig. 2, made it possible to
calculate the hysteretic behavior, i.e., complete loops of deformations, of the soil springs,
which is a great advantage.
In terms of damping, some energy dissipation mechanisms are considered. Since a considerable part of energy will be damped by slippage between pipe and surrounding soil, it
is essential for the soil springs to be able to simulate soil-pipe sliding properly. Moreover,
123
1378
Station
Shear wave
velocity Vs (m/s)
Transverse
Vertical
ChiChi
CHY041
Vs<180
0.639
0.302
0.123
ChiChi
CHY028
0.821
0.653
0.337
ChiChi
CHY080
0.902
0.968
0.724
Northridge
Montebello-Bluff Rd
Vs<180
0.179
0.129
0.076
Northridge
LA-Centinela St
0.465
0.322
0.109
Northridge
0.883
0.37
0.23
hysteretic behavior of soil material under dynamic loading is also greatly responsible for
system damping which is considered by aforesaid springs in the present paper. Material
damping is regarded to be of Rayleigh type using and coefficients for mass and stiffness proportional dampings, respectively. These coefficients have been computed through
a modal analysis identifying the modes with important contributions. The damping ratio is
assumed to be 4 % of the critical value according to Chopra (1995), Kishida and Takano
(1970), Rofooei and Qorbani (2008). In order to consider the radiation damping, an appropriate boundary condition is defined that takes care of dissipation of energy to infinity. This
aspect of modeling is described in the following.
2.1.3 The dynamic loading
As explained, a major purpose of this study is determining the time history of pipe bend
response under earthquakes, requiring selection and application of suitable ground motions
to the model. The main characteristics of the strong motions to be selected are: Earthquake
magnitude, the shear wave velocity of the soil on which the motion was recorded, and the
frequency content.
The soil types of this study were described in Tables 2 and 3. For selection of ground
motion, the (PEER) strong motion database was consulted. Before that, a modal analysis
was implemented for each of the six soil categories and the range of frequencies with large
mass participation factors was calculated. This was compared to the strong frequency band
of the Fourier amplitude spectrum of each record examined. Finally, the motions recorded on
a similar soil as of Tables 2 and 3 having appropriate frequency bands as described above and
being strong enough (magnitude > 6) were picked up. Among them, different recordings of
Northridge (1994) and ChiChi (1999) earthquakes on the corresponding soil sufficed for the
purposes of this study and will be used and compared afterwards. The characteristics of these
records are summarized in Table 5. The spectra of these ground motions are also illustrated
in Figs. 3 and 4.
An important parameter to be studied in this research is the incidence angle of seismic
waves. In the majority of the previous works, the basic assumption has been vertical propagation of input waves making one unable to study the effect of phase difference of input
motion at different parts of pipe. This is a limitation lifted in this study. To make this end, the
propagation medium is presumed to be homogeneous and elastic and the energy source (the
causative fault) is taken as being very far such that the wave propagation can be regarded
to be along parallel lines. Inclined propagation of waves in three dimensions can be defined
with incidence angles and , as is shown in Fig. 5.
123
1379
Fig. 3 Response spectra of longitudinal and transverse components of ChiChi earthquake in selected stations
(in logarithmic scale)
Vertical propagation is associated with = 90 , where all nodes of the model experience
the same input motion. For other values of , where is also defined, the same motion is
input to different nodes at different times regarding the arrival time of shear waves at each
node. This is a non-synchronized input motion making the response much more complicated
but more realistic especially in deep basins.
2.1.4 The boundary conditions
To have a realistic estimation of the dynamic response, it is necessary to somehow model the
namely infinite length of the pipe away from the bend. This is possible by taking a limited
123
1380
Fig. 4 Response spectra of longitudinal and transverse components of Northridge earthquake in selected
stations (in logarithmic scale)
length of pipe extending from bend to a suitable boundary condition. In this research the
boundary condition proposed by Liu et al. (2004) is adopted. In their study, they assume
that the lateral deformations of pipe in far distances do not affect the response of the portion
under study, but the longitudinal friction is important.
According to Fig. 6, friction force along part OB of pipe due to an axial force F consists
of two parts: (a) The static friction OC, (b) the slip friction CB. Point O is still. Relation of
the axial force F and the extension L is used for introducing a nonlinear spring at the pipe
boundary, as of Eq. (1).
1
3E A fs U 6 L 23
0 L U0
0
2
F(L) =
2
2E A f (L 1 U ) U L y A +
s
0
4 0
2E f s
123
U0
4
(1)
1381
Fig. 6 The boundary conditions simulating infinite length of pipe (Liu et al. 2004)
In Eq. (1), E is the elastic modulus, A is the pipe cross section, f s is the slip friction force
on unit length of pipe, u 0 is yield displacement and y is the yield stress of pipe material.
Note that in this study while the radiation damping is disregarded for the lateral relative
motion (considering its small amplitude in wave propagation analysis), the following sources
of damping are believed to be far more important and are taken into account: Hysteretic
behavior of soil material under dynamic loading and the Rayleigh type material viscous
damping (Sect. 2.1.2), radiation damping in the longitudinal direction and slippage between
pipe and the surrounding soil (current Section). This is in compliance also with the current
literature (ORourke and El Hmadi 1988; ORourke and Liu 1999; Yoshizaki et al. 2001;
Mclaughlin and ORourke 2003; Rofooei and Qorbani 2008; Mclaughlin and ORourke 2009;
Lee et al. 2009).
123
1382
123
1383
associated with the node on the longitudinal axis, or the center of section at the same place,
to account for consistency of displacements in the model.
3 Numerical results
In the following parts, results of pipe strain analysis varying propagation angle, bend angle,
soil type, pipe diameter to thickness ratio, and burial depth ratio are presented. Axial strains
are normalized to the critical and yield strains (cr and y respectively). Use of critical strain,
cr, is suggested by American lifelines alliance (2001) as of Eq. 2 and yield strain, y, is
evaluated by Hooks law as well. Therefore, according to Table 1, yield strain is calculated
to be around 0.002.
pD 2
t
wavepr opagation
cr
= 0.75 0.5 0.0025 + 3000
D
2Et
D =
D
1
3
D (D
Dmin )
(2)
123
1384
Fig. 9 Bend response for different straight lengths and boundary conditions
extensional strains are generally much lower than critical buckling or yielding strains and
can be simply ignored for vertical propagation. At the same time, the absolute maxima tend
to occur for softer soils of clay or sand, and for elbow angles larger than 90 , equal to about
135 in many cases. Note that elbow angle is the smaller angle between the two straight
branches of pipe.
3.3 Results for = 0 , = 0 , 45
According to Fig. 5, when = 0, the propagation plane is parallel to a pipe branch; then,
e.g., when also = 0 the wave itself propagates parallel to a branch. The analysis results are
shown in Fig. 11.
The trend is such that the strains are larger in stiffer soils. This is mainly due to the less
extent of slip of pipe in stiffer soils as will be justified when computing slip in the sections
to follow. A very important point to note is that strain maxima are multifold with respect to
vertical propagation (Sect. 3.2) to the extent that for horizontal propagation in some cases
yielding at bend and local damage occurs. Also, strain values for = 0 are considerably
larger than those for = 45 . This is a direct result of phase difference of input motion which
is highest for horizontal propagation of waves, when the apparent velocity of propagation is
a minimum. To make this point more clear, variation of strain for embedded pipe in stiff clay
is shown in Fig. 12 for an incremental variation of incidence angle. Again, strains are larger
for elbow angles different from 90 , being doubled in some cases especially around 135 .
There are two reasons for this behavior. In one hand with changing the elbow angle from
180 (straight pipe) to 90 , it is expected that the value of slippage (relative displacement
between soil and pipe) decreases due to presence of the elbow, resulting in an increase of
the response of the bend. On the other hand, when examining a right angle elbow under
propagation of seismic waves parallel to a pipe branch, the perpendicular components of
the seismic ground motion are each applied only along their corresponding soil springs, i.e.,
each component affects only the soil springs lying in its direction. However, when the elbow
angle is different from 90 (and 180 ), there is an interference between the longitudinal
and lateral seismic components in the direction of each soil spring around the pipe. This
combination of seismic motion components is likely to lead to a larger effective motion which
in turn increases the response of the pipe and the elbow. Therefore, there must be a critical
elbow angle between 90 and 180 with a highest response under the combination of the
both aforementioned phenomena. Besides, the combination of perpendicular components of
123
1385
Fig. 10 Longitudinal strains against elbow angle for vertical propagation (H=1.5, D=40 cm, t=95 mm); a
Loose sand, b Medium sand, c Dense sand, d Soft clay, e Medium clay, f Stiff clay
seismic motion and the fact that there is a critical angle of attack of earthquake at the system
to be analyzed, is a well-known fact. It is expected that when the two horizontal components
of ground motion are more or less equally strong, the critical angle of the ground motion
is around 135 degrees. When a pipe branch lies in the same direction, a peak response is
likely to occur. Figures 3 and 4 show that the earthquakes are strong in both directions. In
comparison to the study by Ogawa and Koike (2001), which concludes that the 90 bend is
the most critical for pipeline strains, the modeling and analysis in the present paper widely
overcomes their limitations. The analysis by Ogawa and Koike (2001) was two dimensional
and quasi-static. Moreover, they followed the guidelines of Japan Gas Association (2000)
for the seismic design load. The Aforementioned criteria could dramatically affect the elbow
response in comparison with the current research. As seen, the analyses in the present paper
are performed using the 3D dynamic time history procedure so that all three components
of ground motion records, which are combined of different wave types, are applied to the
models. Furthermore, in order to simulate the pipe behavior more accurately, effects of the
surrounding soil in all three directions are considered whereas Ogawa and Koike (2001) only
modeled soil in the longitudinal direction.
123
1386
Fig. 11 Effect of different incidence angles in the vertical plane on maximum axial strains of bend against
elbow angle (H=1.5, D=40 cm, t=95 mm); a Loose sand, b Medium sand, c Dense sand, d Soft clay, e Medium
clay, f Stiff clay
123
1387
Fig. 12 Effect of incidence wave angle in vertical plane, , on maximum axial strain of a bend
123
1388
Fig. 13 Effect of variable and incidence angles (as of Fig. 5) on the maximum strain at a 90 bend
(H=1.5 m, D=40 cm, t=9.5 mm). a sand, ChiChi, b sand, Northridge, c clay, ChiChi, d clay, Northridge
flexible pipes in soft to stiff clays while it is highly changing for stiff pipes in different clays.
It should be due to combinatorial effect of soil stiffness and cohesion which prevents pipe
from slipping in stiffer and more cohesive clays.
3.7 Effect of H/D
The results of analysis with different values of burial depth ratio (burial depth (H ) to pipe
diameter (D)) for the maximum axial strain of bend for a 90 bend and horizontal propagation
of waves are demonstrated in Fig. 16. Comparing with other factors discussed in previous
sections, the H/D ratio appears to be less affecting the pipe response, meaning that increasing
depth for the same pipe or decreasing diameter for the same depth is not going to drastically
change the maximum axial strain of pipe. Looking more closely at Fig. 16, it can be seen that
for sands increase of H/D increases the strain level. This is a result of smaller pipe slip for
larger H/D because in this case increasing H/D enhances the soil slippage capacity in Fig.2.
123
1389
Fig. 14 Effect of elbow angle on pipe slip (H=1.5 m, D=40 cm, t=9.5,mm, = 0 , = 0 ), a loose sand,
b medium sand, c dense sand, d soft clay, e medium clay, f stiff clay
On the other hand, the longitudinal and transverse soil resistances increase for larger
H/Ds resulting in a stiffer supporting medium around pipe and a larger response.
For clays, again a changing trend similar to effect of D/t in the previous section is
observed. For stiffer clays existence of cohesion force results in smaller slip and larger strain,
an effect that is not prevalent for softer clays.
3.8 Comparison of beam and hybrid models
Results of analysis of a pipe with a 90 bend under horizontal propagation of waves ( = 0
and = 0), for slip values and max. axial strains at bend are presented in Figs. 17 and 18 for
comparison. It is observed that the beam and hybrid models have given practically the same
results for slip values and the difference is even smaller for softer soils.
123
1390
Fig. 15 Effect of D/t on the maximum axial strain of bend (H=1.5 m, Elbow angle = 90 , = 0 , = 0 ),
a loose sand, b medium sand, c dense sand, d soft clay, e medium clay, f stiff clay
For axial strains, the difference is larger but again it is small in most cases. In order
to realize the effect of hoop strains, which could only be determined by the shell model,
the principal strains are compared in beam and hybrid models. Although the hybrid model,
123
1391
Fig. 16 Effect of burial depth ratio (H/D) on maximum axial strain of bend (Elbow angle = 90 , = 0 , =
0 ), a loose sand, b medium sand, c dense sand, d soft clay, e medium clay, f stiff clay
which is deemed to be more exact, results in larger principal strains which can be due to the
existence of hoop strains. Differences between principal strains in both aforesaid models are
practically small. Therefore, use of the axial stress/strain for estimating the yield behavior
123
1392
Fig. 17 Comparison of slip values in beam and hybrid models under ChiChi and Northridge earthquake
(H=1.5 m, D=40 cm, t=9.5 mm, Elbow angle=90 deg, = 0 , = 0 ), a sand, b clay
at bend is justified. The comparison between principal strains predicted by beam and hybrid
models for a pipeline buried in sandy soil and under Northridge earthquake is presented in
Fig. 19 as a sample case.
On the other hand, modeling the pipe with shells consumes much more time and cost and
is prone to a greater possibility of user error, while the enhancement in accuracy is small as of
Figs. 17 and 18. Use of hybrid modeling especially for pipe analysis under wave propagation
phenomenon is not recommended as a major result of this study, except when it is meant to
calculate the effect of local buckling or to evaluate the hoop strain.
The effect of the internal pressure of pipe on the strains is studied next. For an internal
pressure p, the fully plastic pressure is Py = 2 y t /Dm , where y and Dm are the yield
stress and the mean diameter of the undeformed pipe, respectively. Based on this, analyses
for an internal pressure being equal to 10, 20, 30, 40 and 50 % of the fully plastic pressure
are implemented and the results of the maximum principal pipe strain for one of the critical
cases, dense sand under Chichi earthquake, is presented in Fig. 20. As seen, there is a direct
relationship between the internal pressure and maximum principal strain in the elbow area.
4 A semi-analytical model
The modeling of soil-pipe interaction using the FE method is usually a time consuming
process. Furthermore, required equipments and convergence difficulties in numerical analyses are typical problems. As discussed in Sect. 3.8, the maximum axial strain is the key
parameter for estimating elbow response. It also plays a critical role in industrial pipe design
and is the principal strain in the beam model of elbow. The bulk of detailed FE analysis
presented above prepares ground for deriving a semi-analytical method to reach at simple
equations for easily predicting the maximum axial strain at elbow region. For this purpose,
a mathematical model must be developed first.
The parameters studied in response calculations of this paper were the elbow angle, the
shear wave velocity in the soil, the pipe diameter and the pipe embedment depth. Considering
the trend of response variation under the influence of the above factors as seen in the above
figures, the following general equation is set to calculate the maximum axial stress at the
elbow:
f
c
e
Vs
PG A h D k
Axial = a b
+d
S
S
VSBedr ock
g
t
123
(3)
1393
Fig. 18 Comparison of max. axial strains at bend for beam and hybrid models (H=1.5 m, D=40 cm, t=9.5 mm,
Elbow angle=90 deg, = 0 , = 0 ), a sand, ChiChi earthquake, b clay, ChiChi earthquake, c sand,
Northridge earthquake, d clay, Northridge earthquake
Fig. 19 Comparison of principal strains at bend for beam and hybrid models in sand under Northridge
earthquake (H=1.5 m, D=40 cm, t=9.5 mm, Elbow angle=90 deg, = 0 , = 0 )
123
1394
Fig. 20 Effect of internal pressure on Max. principal strains at bend for hybrid models in dense sand under
Chichi earthquake (H=1.5 m, D=40 cm, t=9.5 mm, Elbow angle=90 deg, = 0 , = 0 )
of the selected ground motion in g, g is the acceleration of gravity, D is the pipe diameter
and t is the pipes wall thickness.
As in the depth range considered in this research the depth to thickness ratio was proved
to have negligible effect on the response compared with the other parameters, it was omitted
from the mathematical model. The numerical data from the 3D FE analysis in this paper are
divided into two groups regarding the surrounding soil, as sand and clay data. Furthermore,
each group also includes three different soil types ranging from loose to dense for sand and
soft to stiff for clay, comprising a long range of soil types. Overall, as shown in Table 6, 48
different models for each soil type (sand and clay) are considered. Half of these models are
subjected to Chichi and the rest of them to Northridge ground motions. In all, about a hundred
cases out of the analytical models of this work are considered for regression analysis (Table 6).
Accordingly, a nonlinear multiple dimension regression analysis is performed to derive the
constants of the above equation in terms of the geometrical properties of the buried pipe, the
surrounding soil type, and the seismic input characteristics. Thus, four sets of equations are
developed as the following.
4.1 Regression analysis on sandy soils
The derived equation for the pipe embedded in sandy soils and subjected to the ground
motions of ChiChi and Northridge earthquakes and for their average are as mentioned in
Eqs. (4), (5), and (6), respectively.
ChiChil
N or thridge
Average
4.14
0.18
2
Vs
P G A 0.35 D 0.38
3.4
= 1.4 10
+ 4.44
(4)
S
S
VSBedr ock
g
t
6.6
0.6
2
Vs
P G A 0.01 D 0.1
5.4
2
0.25
Vs
P G A 0.2 D 0.45
5
2.71
= 9.5 10
+ 3.73
(6)
S
S
VSBedr ock
g
t
4
The above equations are illustrated for the different cases calculated in this paper in Fig. 21,
showing a very good agreement both for the individual records and for their average. While for
a large majority of cases the relative difference between the numerical and the semi-analytical
predictions of the average curve of Fig. 21 is less than 5 %, a larger difference is observed
123
Model number
1395
(degree)
Vs (m/s)
D/t
PGA(g)
ChiChi
Northridge
90
450
21
0.968
0.883
90
450
42
0.968
0.883
90
450
64
0.968
0.883
90
450
85
0.968
0.883
90
220
21
0.821
0.465
90
220
42
0.821
0.465
90
220
64
0.821
0.465
90
220
85
0.821
0.465
90
75
21
0.639
0.179
10
90
75
42
0.639
0.179
11
90
75
64
0.639
0.179
12
90
75
85
0.639
0.179
13
112.5
450
42
0.968
0.883
14
112.5
220
42
0.821
0.465
15
112.5
75
42
0.639
0.179
16
135
450
42
0.968
0.883
17
135
220
42
0.821
0.465
18
135
75
42
0.639
0.179
19
157.5
450
42
0.968
0.883
20
157.5
220
42
0.821
0.465
21
157.5
75
42
0.639
0.179
22
180
450
42
0.968
0.883
23
180
220
42
0.821
0.465
24
180
75
42
0.639
0.179
only for the cases 23 and 24 of Table 6 corresponding to a pipe embedded in medium and
loose sands (Elbow angle=180deg, D/t=42). This should make no practical problems since
it corresponds to small strains much less than the yield strain. This fact suggests that if the
response calculations of this paper are repeated for a suitably selected set of earthquake
records, an even better regression equation will be derived. The values of the parameters are
tabulated for each case number in Table 6.
Alternatively, the spectra of the above ground motions are assessed for reaching at applicable conclusions as shown in Figs. 3 and 4. As seen, the effective period range covered by
each of the above earthquakes is between 0.26 and 0.84 sec for ChiChi earthquake, and
0.12 and 0.24 sec for Northridge earthquake. Therefore, tentatively, Eq. (4) can be used for
T p 0.25sec and Eq. (5) for T p 0.25sec, in which T p is the governing period of the
record.
4.2 Regression analysis on clayey soils
A similar procedure was employed to derive equations for the pipe surrounded by the clayey
soils in this paper which are expressed in Eqs. 7, 8 and 9. in which the definitions of the
notations are similar to Eq. (3).
123
1396
Fig. 21 Comparison between calculated maximum elbow axial strains in sandy soil under seismic excitations,
a ChiChi; b Northridge; c Average
7.34
0.22
0.3
Vs
PG A
D
(7)
S
VSBedr ock
g
t
14.25
0.3
2
Vs
P G A 0.3 D 0.11
7
3.33
= 1 10
+ 5.12
(8)
S
S
VSBedr ock
g
t
7.4
0.2
2
Vs
P G A 0.6 D 0.3
2
+ 6.61
Again, the above equations are illustrated for the different cases calculated in Fig. 22, showing
generally a very good agreement. While for the governing majority of cases the relative
difference between the average numerical and semi-analytical predictions of Fig. 22 is less
than 5 %, a larger difference is observed only for the case of pipe buried in soft clay with an
elbow angle of 90 deg., and D/t=85 (model No. 12). Again, this case corresponds to very small
123
1397
Fig. 22 Comparison between calculated maximum elbow axial strains in clayey soil under seismic excitations.
a ChiChi; b Northridge; c Average
strains and possesses no practical importance. This agreement justifies similar calculations
for a multitude of earthquake records to attain a general response equation. As discussed in
Sect. 4.1, Eq. (7) is appropriate for Tp 0.25 sec and Eq. (8) for Tp 0.25sec. The values
of the parameters for each case number are again as Table 6.
4.3 Comparison with the existing methods
The strain values estimated by Eqs. (6 and 9) are compared also with those predicted by
two more widely used methods, namely, by the Japan Gas Association (Ogawa and Koike
2001) and Mclaughlin and ORourke (2003), Mclaughlin and ORourke (2009). For this purpose, the cases of a right angle elbow pipe embedded in dense sand (D=0.61m, t=0.0127m,
= 0 and = 0 ) at various ground strains and also a similar pipe with different wall
thicknesses are examined. As observed in Fig. 23, there is a good agreement between
the results predicted by the existing methods and the suggested method in this paper. It
should be mentioned that some discrepancies are expected in the results due to certain
123
1398
Fig. 23 Comparison between results predicted by the semi-analytical method and the two widely used methods; for the case of a pipe buried in dense sand under various ground strains and wall thicknesses (H=0.91 m,
D=61 cm, t=12.7 mm, Elbow angle=90 deg, = 0 , = 0 ). Comparison with the: a approaches separately, b
average of the approaches, c approaches separately, ground strain=0.001, d average of the approaches, ground
strain=0.001, e approaches separately, ground strain=0.002, f average of the approaches, ground strain=0.002
differences in the assumptions. For instance, in the Ogawa and Koike (2001) method, a
surface layer resting on half-space is considered, and this affects the pipe response considerably. There is an increase in ground strains in shallow depths and pipe strains as well,
for a soft layer over a relatively rigid halfspace. Moreover, in the current papers approach,
the analyses were three dimensional in comparison with other mentioned methods that are
2D. So, effect of the third component of earthquake ground motion records could lead to
certain changes in the pipe response. It is also worth mentioning that in quasi-static analysis
used by the aforesaid approaches, one type of waves, usually Rayleigh waves, is considered as the seismic wave, whereas in the time history analysis the pipe is subjected to a
combination of different wave types with various wave lengths and propagation velocities. Nevertheless, in spite of all the differing factors, there is still a very good agreement
between the results predicted by the proposed regression equations in the present study and
the widely used methods in the literature. It is noteworthy that while the effects of almost
all important factors in buried pipes analysis such as geometrical parameters of the pipe
elbow, soil type, and the seismic loading are considered in the suggested equations, they
123
Earthquake
1399
Soil type
Loose sand
Medium sand
Dense sand
Chichi
1.52
1.48
1.22
Northridge
1.54
1.03
1.01
Average
1.52
1.39
1.18
Design factor
1.5
1.4
1.2
Soft clay
Medium clay
Stiff clay
Chichi
1.40
1.08
1.04
Northridge
1.29
1.20
1.15
Average
1.39
1.09
1.05
Design factor
1.4
1.1
1.1
are very simple and fast to use and do not need complex and time consuming mathematical calculations. Therefore, in comparison with other widely used methods, the suggested
formulation is able to predict the elbow pipe response with a reasonable accuracy at low
cost.
123
1400
References
American lifelines alliance (ALA) (2001) Guidelines for the design of buried steel pipe (with addenda through
2005). American Society of Civil Engineers
American Petroleum Institute (2000) Specificarion for line pipe. API specification 5L, Forty-Second Edition
American Society of Civil Engineers (1984) Guidelines for the seismic design of oil and gas pipeline systems.
Committee on Gas and Liquid Fuel Lifelines Technical Council on Lifeline Earthquake Engineering, New
York
Berrones RF, Liu XL (2003) Seismic vulnerability of buried pipelines. Geofisica Internacional 42(2):237246
Bowles JE (1996) Foundation analysis and design. The McGraw-Hill Companies, NY
Chopra AK (1995) Dynamics of structures: theory and application to earthquake engineering. Prentice-Hall,
Englewood Cliffs
Dash SR, Jain SK (2008) An overview of the seismic considerations of buried pipeline. J Struct Eng 34(5):
349359
Datta TK (1999) Seismic response of buried pipelines: a state-of-the-art review. Nucl Eng Des 192:271284
Goodling EC (1983) Buried pipingan analysis procedure update. In: Proceedings international symposium
on lifeline earthquake engineering, Portland, Oregon, ASME, PVP-77: 225237
Halabian AM, Hokmabadi T, Hashemolhosseini SH (2008) Numerical study on soil-HDPE pipeline interaction
subjected to permanent ground deformation. The 14th world conference on earthquake engineering, Beijing,
China
Hart JD, Zulfiqar N, Lee CH, Dauby F, Kelson KI, Hitchcock C (2004) A unique pipeline fault crossing design
for a highly focused fault. In: Proceedings of IPC, international pipeline conference, Calgary, Alberta,
Canada
Hatzigeorgiou GD, Beskos DE (2010) Soil-structure interaction effects on seismic inelastic analysis of 3-D
tunnels. Soil Dyn Earthq Eng 30:851861
Hindy A, Novak M (1979) Earthquake response of underground pipelines. Earthq Eng Struct Dyn 7:451476
Horikawa H, Suzuki N (2009) Bending deformation of x80 cold bend pipe. In: Proceedings of the nineteenth
international offshore and polar engineering conference, Osaka, Japan
123
1401
Japan Gas Association (JGA) (2000) Seismic design guideline of high-pressure gas pipelines. Japan Gas
Association
Karamanos SA, Giakoumatos E, Gresnigt AM (2003) Nonlinear response and failure of steel elbows under
in-plane bending and pressure. J Press Vessel Technol 125(4):393402
Karamanos SA, Tsouvalas D, Gresnigt AM (2006) Ultimate bending capacity and buckling of pressurized 90
deg steel elbows. J Press Vessel Technol 128(3):348356
Kishida H, Takano A (1970) The damping in the dry sand. The proceeding of 3rd Japan earthquake engineering
simposium, Tokyo, Japan
Kouretzis GP, Bouckovalas GD, Gantes CJ (2006) 3-D shell analysis of cylindrical underground structures
under seismic shear (S) wave action. Soil Dyn Earthq Eng 26:909921
Kouretzis GP, Bouckovalas GD, Gantes CJ (2007) Analytical calculation of blast-induced strains to buried
pipelines. Int J Impact Eng 34:16831704
Lee DH, Kim BH, Lee H, Kong JS (2009) Seismic behavior of a buried gas pipeline under earthquake
excitations. Eng Struct 3:10111023
Liu AI, Hu YX, Zhao FX, Li XJ, Takada S, Zhao L (2004) An equivalent-boundary method for the shell
analysis of buried pipelines under fault movement. Acta Seismol Sinica 17:150156
Lysmer J, Kuhlemeyer AM (1969) Finite dynamic model for infinite media. J Eng Mech Div 95(4):859877
Mclaughlin PM, ORourke M (2003) Seismic response and behavior of buried continuous piping systems
containing elbows. Doctoral Thesis, Rensselaer Polytechnic Institute
Mclaughlin PM, ORourke M (2009) Strain in pipe elbows due to wave propagation hazard. Lifeline Earthquake
Engineering in a Multihazard Environment, ASCE
Muthmann E, Grimpe F (2006) Fabrication of hot induction bends from LSAW large diameter pipes manufactured from TMCP plate. Microalloyed steels for the oil and gas industry international symposium, Araxa,
Brazil
Newmark NM (1968) Problems in wave propagation in soil and rock. In: Proceedings of international symposium on wave propagation and dynamic properties of earth materials, August 2325, Univ. of New Mexico
Press, Albuquerque: 726
Ogawa Y, Koike T (2001) Structural design of buried pipline for sever earthquake. Soil Dyn Earthq Eng
21:199209
ORourke MJ, Liu X (1999) Response of buried pipelines subject to earthquake effects. MCEER Monograph
Series, Multidisciplinary Center for Earthquake Engineering Research
ORourke TD, Wang Y, Shi P, Jones S (2004) Seismic wave effects on water trunk and transmission lines.
In: Proceedings of the 11th international conference on soil dynamics and earthquake engineering and 3rd
international conference on earthquake geotechnical engineering. Berkeley, CA 2:420-428
ORourke MJ, El Hmadi K (1988) Analysis of continuous buried pipelines for seismic wave effects. Earthq
Eng Struct Dyn 16:917929
Pacific Earthquake Engineering Research Center, http://peer.berkeley.edu/, PEER
Pappa P, Tsouvalas D, Karamanos SA, Houliara S (2008) Bending behavior of pressurized induction bends.
Offshore Mechanics and Arctic Engineering Conference, ASME, OMAE2008-57358, Lisbon, Portugal
Rofooei FR, Qorbani R (2008) A parametric study on seismic behavior of continuous buried pipeline due to
wave propagation. The 14th world conference on earthquake engineering, Beijing, China
Sakurai A, Takahashi T (1969) Dynamic stresses of undergroud pipelines during earthquake. In: Proceeding
of the fourth wolrd confrenece on earthquake engineering, Santiago, Chile, pp 8195
Shah H, Chu S (1974) Seismic analysis of underground structural element. J Power Div 100:5362
Shi P, ORourke TD, Wang Y, Fan K (2008) Seismic response of buried pipelines to surface wave propagation
effects. The 14th world conference on earthquake engineering. Beijing, China
Stamos AA, Beskos DE (1995) Dynamic analysis of large 3-D underground structures by the BEM. Earthq
Eng Struct Dyn 24:917934
Takada S, Tanabe K (1987) Three dimensional seismic response analysis of buried continuous or jointed
pipelines. J Press Vessel Technol 109:8087
Takada S, Hassani N, Fukuda K (2001) A new proposal for simplified design of buried steel pipes crossing
active faults. Earthq Eng Struct Dyn 30:12431257
Takada S, Carlo A, Bo S, Wuhu X (1992) Current state of the arts on pipeline earthquake engineering in japan.
The construction engineering research institute foundation: 305326
Takada S, Higashi S (1992) Seismic response analysis for jointed buried pipline by using shell FEM model.
In: Proceedings of the tenth world conference on earthquake engineering pp 54875492
Takada S, Katagiri S (1995) Shell model response analysis of buried piplines. In: Proceedings of the fourth
U.S. confrence of lifeline earthquake engineering , pp 256263
Trifonov OV, Cherniy VP (2010) A semi-analytical approach to a nonlinear stress-strain analysis of buried
steel pipelines crossing active faults. Soil Dyn Earthq Eng 30:12981308
123
1402
Vazouras P, Karamanos SA, Dakoulas P (2010) Finite element analysis of buried steel pipelines under strikeslip fault displacements. Soil Dyn Earthq Eng 30:13611376
Wang Y, ORourke TD, Shi P (2006) Seismic wave effects on the longitudinal forces and pullout of underground
lifelines. In: Proceedings of the 100th anniversary earthquake conference: commemorating the 1906 San
Francisco Earthquake, San Francisco, California, Aplil 1822
Yoshizaki K, ORourke TD, Hamada M (2001) Large diformation behavior of buried pipeline with low angle
elbows subjected to permanent ground deformation. Proc JSCE (Japan Society of Civil Engineers) 675:
4152
123