Gong et al.

/ J Zhejiang Univ-Sci A (Appl Phys & Eng) 2011 12(8):627-636 627

Journal of Zhejiang University-SCIENCE A (Applied Physics & Engineering)

ISSN 1673-565X (Print); ISSN 1862-1775 (Online)
www.zju.edu.cn/jzus; www.springerlink.com
E-mail: jzus@zju.edu.cn

Buckling response of offshore pipelines under combined tension,

bending, and external pressure*

Shun-feng GONG†, Lin YUAN, Wei-liang JIN

(Institute of Structural Engineering, Zhejiang University, Hangzhou 310058, China)
†
E-mail: sfgong@zju.edu.cn
Received Dec. 4, 2010; Revision accepted Mar. 21, 2011; Crosschecked July 4, 2011

Abstract: The buckle and collapse of offshore pipeline subjected to combined actions of tension, bending, and external pressure
during deepwater installation has drawn a great deal of attention. Extended from the model initially proposed by Kyriakides and
his co-workers, a 2D theoretical model which can successfully account for the case of simultaneous tension, bending, and external
pressure is further developed. To confirm the accuracy of this theoretical method, numerical simulations are conducted using a 3D
finite element model within the framework of ABAQUS. Excellent agreement between the results validates the effectiveness of
this theoretical method. The model is then used to study the effects of several important factors such as load path, material prop-
erties, and diameter-to-thickness ratio, etc., on buckling behaviors of the pipes. Based upon parametric studies, a few significant
conclusions are drawn, which aims to provide the design guidelines for deepwater pipeline with solid theoretical basis.

Key words: Deepwater, Offshore pipeline, Buckle, External pressure

doi:10.1631/jzus.A1000489 Document code: A CLC number: X703

1 Introduction economic loss.

The buckling response of the pipes subjected to
Submarine pipeline is an important part of off- installation loads is of paramount importance, and as
shore oil and gas exploitation projects. The installa- a result this subject has received much attention in the
tion of pipelines, especially in deep water, can induce past few decades. Gellin (1980) investigated the ef-
rather severe loads to the structure, which, as a result, fect of nonlinear material behavior on the buckling of
will lead to the unforeseeable risks and challenges. an infinitely-long cylindrical shell under pure bending.
Pipes in the sag-bend region have to withstand com- Subsequently, the response and stability of elasto-
bined tension, bending, and ambient pressure re- plastic circular pipes under combined bending and
gardless of the installation method used (S-lay, J-lay external pressure were analyzed by Kyriakides and
or reeling-lay) (Kashani and Young, 2005; Li et al., Shaw (1982). They determined the maximum mo-
2008). Under these combined loads, deepwater pipe- ment and curvature as a function of the material and
lines are vulnerable to local buckling, which will, in geometric parameters for different pressures. Corona
turn, have the potential of initiating a propagating and Kyriakides (1988) further studied the stability of
buckle. Once the propagating buckle occurs, the circular tubes under combined bending and external
pipeline will suffer from catastrophic flattening, re- pressure. The curvature-pressure interaction collapse
sulting in failure of whole structure as well as huge envelopes were generated for two different loading
paths involving bending followed by pressure and
*
Project supported by the National Natural Science Foundation of
pressure followed by bending. It was found that the
China (No. 51009122), the Fundamental Research Funds for the loading path strongly affected buckling response of
Central Universities (No. 2010QNA4030), and the National High- the tube, critical collapse loads, and the nature of
Tech R&D (863) Program of China (No. 2006AA09A105)
© Zhejiang University and Springer-Verlag Berlin Heidelberg 2011 instabilities. Dyau and Kyriakides (1992) developed a
628 Gong et al. / J Zhejiang Univ-Sci A (Appl Phys & Eng) 2011 12(8):627-636

2D model to examine the buckling response of tubes tail corresponding to a certain load path denoted T→
under combined bending and tension, in which two Radial(κ, P), in which the pipe is first tensioned to a
types of combined loadings were investigated in de- chosen value T in the longitudinal direction, and then
tail. Al-Sharif and Preston (1996) proposed a deter- the curvature κ and external pressure P are applied in
ministic model to calculate the collapse of the tube proportion. This load path considered in the present
under combined bending and pressure, and then de- study can more approximate the case of practical
veloped a numerical model to simulate the plastic pipe-laying. Besides, a parametric study concerning
collapse of thick-walled pipe. Moreover, it was found several important influence factors was conducted,
that the simulation results agreed well with those of and some significant conclusions were drawn.
the deterministic model. Studies on the collapse of
thick-walled tubes under three different load combi-
nations, i.e., bending and pressure, tension and pres- 2 Theoretical formulations
sure, and tension and bending, were performed re- 2.1 Kinematics
spectively by Kyriakides and Corona (2007). Nu-
merical models with simultaneous tension, bending, The tube considered is a long, circular, thick-
and external pressure applied were carried out to walled tube, with its mean radius R and thickness t.
simulate the behavior of thick-walled tubes using Uniform tension T, curvature κ and external pressure
ABAQUS by Bai et al. (1997). They also proposed a P are assumed to be applied along the length. As
set of interaction equations accounting for some ma- shown in Fig. 1, it is noted that the coordinate z is the
jor factors affecting collapse envelopes based on the radial distance from the mid-surface of the tube wall
analyses. Lately, Yuan et al. (2009) respectively in- rather than from the centre, and the displacements of a
vestigated the buckling performance of deepwater point on the mid-surface are denoted as u, v and w
pipes subjected to pure bending, and combined with respect to axial, circumferential and radial co-
bending and external pressure. It is indicated that the ordinates x, θ and z, respectively. The plane sections
buckling response of the tube is closely related to the are assumed to be normal to the mid-surface of the
diameter-to-thickness ratio, and the existence of ini- tube cross-section before and during deformation.
tial curvature will weaken the load-carrying capacity Besides, small strain and finite rotations about the
of the pipe to resist external hydrostatic pressure. axes are assumed here (Gellin, 1980).
The present study aims to extend the general
theory proposed by Kyriakides and his co-workers to
further investigate the buckling response of thick- 1/κ
walled tubes under simultaneous tension, bending,
and external pressure. It is assumed that the buckling T M P A M T
is symmetric about the vertical plane and deforms
uniformly along the length of pipeline. Based upon P
A

the nonlinear ring theory, which could accommodate t v

large deformation and circumferential extension w
ς
scenario, the relationship of strain and displacement is v
R R θ
obtained. Then, a virtual work approach is adopted to ς
θ
formulate a set of equilibrium equations, which are w, z
solved using the Newton-Raphson method. Mean-
A-A
while, a 3D numerical model is developed under the
same loading conditions using ABAQUS, and com- Fig. 1 Geometric parameters and coordinate system
ς is the distance from a point in the tube wall to the neutral
parison of the results between the two methods is axis; M is the bending moment
carried out. It is found that the growth of ellipticity
corresponding to theoretical model is somewhat The axial strain of the tube can be denoted as
slower when the applied loads are relatively high.
Furthermore, the buckling response is studied in de- ε x = ε x0 + ςκ , (1)
 Gong et al. / J Zhejiang Univ-Sci A (Appl Phys & Eng) 2011 12(8):627-636 629

where ε x0 is the axial strain of neutral axis, ς is the where

distance from a point in the tube wall to the neutral ⎧ 0, σ e ≤ σ e max ,
⎪
axis, and can be obtained from Fig. 1: Q=⎨ 1 ⎛ E ⎞ (8)
⎪ 4σ 2 ⎜ E − 1⎟ , σ e > σ e max ,
⎩ e ⎝ t ⎠
ς = ( R + w) cos θ − v sin θ + z cos θ . (2)
where () denotes an increment in (), σe is the equiva-
The circumferential strain can be expressed as
lent stress, and Et = Et (σ e ) is the tangent modulus of
εθ = εθ0 + z ⋅ κθ , (3) material. They are given as follows:
where
3
σ e2 = Sij Sij , (9)
2 2 2
⎛ v′ + w ⎞ 1 ⎛ v′ + w ⎞ 1 ⎛ v − w′ ⎞
εθ0 = ⎜ ⎟+ ⎜ ⎟ + ⎜ ⎟ , (4) 1 1 ⎡⎢ 3 ⎛ σ e ⎤
n −1

⎝ R ⎠ 2⎝ R ⎠ 2⎝ R ⎠ ⎞
= 1+ n⎜ ⎟⎟ ⎥, (10)
and Et E ⎢ 7 ⎜⎝ σ y ⎠ ⎥
⎣ ⎦
2
⎛ v′ − w′′ ⎞ ⎛ v − w′ ⎞ where
κθ = ⎜ ⎟ 1− ⎜ ⎟ , (5)
⎝ R
2
⎠ ⎝ R ⎠ 1
Sij = σ ij − σ kk δ ij , (11)
3
where ()′ denotes the differential with respect to θ.
where Sij is the deviatoric stress tensor, σij is the stress
2.2 Constitutive model tensor, σkk is the first invariant of stress tensor, and δij
The material of deepwater pipelines exhibits is the Kronecker Delta function.
good plastic deformation capacity, and the pipe can be
modeled as an elastoplastic solid. The Ramberg- 500
Osgood model is used to characterize nonlinear
400
stress-strain relationships of the material shown in
Fig. 2, which is given as
σ (MPa)

300

200
σ ⎡⎢ ⎤
n −1
3 σ ⎥,
ε = 1+ (6) 100
E ⎢ 7 σy ⎥
⎣ ⎦
0
0.000 0.002 0.004 0.006 0.008 0.010

where E is Young’s modulus, σy is the effective yield ε

Fig. 2 Stress-strain curve for the Ramberg-Osgood model
stress, and n is the material hardening parameter. σy=400 MPa, E=206 GPa, n=10.7
The incremental J2 flow theory of plasticity with
isotropic hardening is used to model the plastic be- 2.3 Principle of virtual work
havior of material. The radial stress σ r and shear
According to the principle of virtual work, the
stress { σ xθ , σ θr , σ rx } are neglected due to the fact equation below must be satisfied when the tube is in
that these components are quite small when compar- an equilibrium state:
ing with those in the axial and circumferential direc-
tions. Therefore, the incremental constitutive model π t/2
2R ∫ ∫ (σˆ xδεx + σˆθ δεθ )dθ dz = δ W , (12)
can be simplified into: 0 −t / 2

⎧εx ⎫ 1 ⎡1 + Q(2σ x − σ θ ) 2 −υ + Q(2σ x − σ θ )(2σ θ − σ x ) ⎤ ⎧σ x ⎫

⎨ ⎬= ⎢ ⎥ ⎨ ⎬, (7)
⎩εθ ⎭ E ⎣ −υ + Q (2σ x − σ θ )(2σ θ − σ x ) 1 + Q (2σ θ − σ x )2 ⎦ ⎩σθ ⎭
630 Gong et al. / J Zhejiang Univ-Sci A (Appl Phys & Eng) 2011 12(8):627-636

where (•ˆ ) ≡ (• + • ), and δ W is the virtual work of the The main steps of solution procedure for com-
external loads, which can be given by bined loading case are shown in Fig. 4. The tube can
be loaded by axial tension T, curvature κ as well as
external pressure P. The loading process is controlled
δ W = Pˆδ V + Tˆδεx0 , (13)
by prescribing the increments of three loading pa-
rameters {ΔT, ΔP, Δκ}. When P is specified to zero,
where δV is the change of volume per unit length of the case would be that of combined bending and ten-
the tube considered. In expanded form, Eq. (12) sion. If the prescribed P and T both equal zero, it
becomes: would reduce to the pure bending case.
The converged solution of the previous step is
2R ∫
π t/2 regarded as the initial estimate of nodal displacements
0 ∫
−t / 2
(σˆ xδεx + σˆθ δεθ )dθ dz
for the next step. Subsequently, the strain increment
ˆ 2π [δ w +(2 wˆ δ w + 2vˆδ v + wˆ δ v′
= PR ∫ 0
(14) can be obtained through the increments of nodal dis-
placements and curvature, and then the stress incre-
+ vˆ′δ w − vˆδ w ′ − wˆ ′δ v) / (2 R)]dθ + Tˆδεx0 . ment can be achieved. Note that this procedure also
involves nested iteration of the constitutive relation-
It is assumed that the deformations of the pipe ship. After obtaining the stress components of each
cross-section, i.e., the in-plane displacements w and v, integral point, the problem can be solved using the
are symmetric about the axis θ=0, and they are the Newton-Raphson method. Strains, stresses as well as
functions of θ. Therefore, w and v can be approxi- displacements corresponding to every integral point
mated by the following series expansions (Gellin, are updated when the converged solution is achieved.
1980): After each converged solution, the moment can
be obtained:
N N
w ≅ R ∑ an cos(nθ ), v ≅ R ∑ bn sin(nθ ). (15)
n =0 n=2
σy

κ
Substituting Eq. (15) into Eq. (14), a series of
2N+1 nonlinear algebraic equations in term of
{a0 , a1 ," , a N , b2 , b3 " bN , εx0 } may be obtained. T P

2.4 Solution method κ

The resultant 2N+1 nonlinear algebraic equa-

tions are solved iteratively in the present solution. U i

Some parameters should be prescribed in advance,

namely geometric dimensions, material parameters, εi

and load sizes. The number of Gaussian integral

σ i
points, for the half cross-section of the tube, along the σe
i

circumferential direction and through the thickness is ΔD / D ,

ε x0
k and l, respectively (Fig. 3). σ e ≥ σ emax
i

F (U i +1 ) = 0 U = U + U ,
ε = ε + ε,
σ = σ + σ ,
σ emax
k U i +1

U i = U i +1

Fig. 3 Distribution of Gaussian integral points Fig. 4 Flow chart of numerical solution procedure
 Gong et al. / J Zhejiang Univ-Sci A (Appl Phys & Eng) 2011 12(8):627-636 631

M = 2R∫
π t/2 κ = ϕ / L. (17)
0 ∫ −t /2
σ xς dθ d z. (16)

To facilitate the development of buckling de-

It is found that the solution can meet the preci- formation, the length of the pipe, L=3D is considered
sion requirements when N equals 4 to 6. In the case of to be suitable. The pipe model is meshed into 6 parts
pure bending, k=12 and l=5 is sufficient. While for the through the thickness, 100 parts around the half cir-
combined case, the mesh should be finer, therefore, cumference and 100 parts along the length, which is
k=12 and l=7 is found to be adequate. found to be adequate. Fig. 5 illustrates a typical finite
element mesh used in the analyses. Furthermore, the
Nlgeom option is selected for the nonlinear calcula-
3 Numerical simulations tion, and the Riks algorithm (arch length method) is
adopted here.
A finite element model is developed within the
framework of the software ABAQUS to simulate the
buckling behavior of pipes under simultaneous ten-
sion, bending, and external pressure. 3D, eight-node
incompatible solid element, C3D8I, is chosen to M T

model the pipe. Since this type of element is enhanced

by incompatible modes to bending behavior, it is best
Y
suited for the present problem (Simo and Armero, X
1992; Hibbitt et al., 2006). The J2 flow theory of O
Z
plasticity with isotropic hardening proposed by Co- L
T M
rona et al. (2006) is adopted to describe the plastic
behavior of material, and the Ramberg-Osgood con-
stitutive model is used by multi-linear approximations Fig. 5 Finite element mesh and loadings
of the stress-strain curve shown in Fig. 2.
The symmetry of the loads and deformations
reduces the problem to a quarter of a pipe. As a result, 4 Results and discussion
symmetrical boundary conditions are applied at the
4.1 Illustrative example using theoretical formu-
mid-span (X=0) and Z=0 planes (Fig. 5). Besides,
lations
additional spring constraints along vertical direction
(Y) are applied at the mid-span plane. This kind of The maximum curvature in the sag-bend region
elastic constraints is desirable for this problem since it often occurs close to the seabed where the maximum
can avoid the stress concentration phenomenon which water depth is reached. Considering that the curvature
is inevitable if rigid constraints are applied. and hydrostatic pressure exerted on the pipes in-
*KINEMATIC COUPLING relationship is im- creases with the depth of the water, while axial ten-
posed between the nodes on the right end of the tube sion is nearly maintained constant, the case of
and a reference point (the central node or the bottom T→Radial(κ, P) loading path is examined.
one are both suitable). The right end plane is con- The pipe is first tensioned incrementally to a
strained to remain plane in the loading process, and at chosen value T=1000 kN, and then curvature and
the same time the cross-section should be free to external pressure are increased proportionately until
deform. The curvature is applied by prescribing the the values of κ=0.15 and P=10 MPa are reached. The
angle of rotation at the reference point, φ. Likewise, main features of the pipe response subjected to the
uniform tension is applied to the model through this combined loads are illustrated in Fig. 6 for a pipe with
reference point, and hydrostatic pressure is imple- its diameter D=254 mm (10 inch) and D/t=20. The
mented on the external surface of the pipe. Thus, the predicted ellipticity-water depth, ellipticity-curvature,
average curvature of the section can be given by axial strain-curvature, and moment-curvature curves
632 Gong et al. / J Zhejiang Univ-Sci A (Appl Phys & Eng) 2011 12(8):627-636

2.5
depth at the beginning. However, the nonlinearity
(a) becomes more and more notable as the loads augment.
2.0 As to the axial strain of the pipe, it nearly experiences
a linear growth with curvature. In addition, it can be
ΔD/D (%)

1.5
seen from moment-curvature response that there ex-
1.0 hibits a limit moment before collapse. Once attaining
the limit moment, localized deformation would
0.5 quickly develop in a region of about 5 to 6 times of
the tube diameters, which can be taken as the critical
0.0
0 200 400 600 800 1000 state of buckling.
Water depth (m)
2.5
(b) 4.2 Comparisons of finite element analysis results
2.0 with theoretical solutions
Numerical simulations and theoretical calcula-
ΔD/D (%)

1.5
tions are carried out respectively for the scenario of
1.0 Radial(T, P, κ) loading path. In other words, three
loading parameters {ΔT, ΔP, Δκ} are simultaneously
0.5
applied to the model. The analyses are performed for
0.0 the pipe model based on the parameters of D=254 mm,
0.00 0.04 0.08 0.12 0.16
κ (1/m) D/t=20, n=10.7, σy=400 MPa, T=600 kN, P=35 MPa
1.0 and κ=0.013. The sequences of deformed configura-
(c) tion and stress distribution during the loading process
0.8
are depicted in Fig. 7.
0.6
εx (%)

0.4

0.2

0.0
0.00 0.04 0.08 0.12 0.16
κ (1/m)
400
(d)

300
M (kN·m)

200

100

0 Fig. 7 Deformed configuration and stress distribution

0.00 0.04 0.08 0.12 0.16
during loading process
κ (1/m)
Fig. 6 Predicted responses for T→Radial(κ, P) loading
path. σy=400 MPa, D/t=20, n=10.7
The comparison of responses calculated by the
(a) Ellipticity-water depth; (b) Ellipticity-curvature; (c) Axial
strain-curvature; (d) Moment-curvature
two methods is shown in Fig. 8. The predicted ellip-
ticity of two models is quite close in the elastic range.
However, the increase of theoretical result slightly
are shown in this figure. The increase of ellipticity is lags behind that of finite element simulation at high
approximately proportional to the curvature and water values of loadings.
 Gong et al. / J Zhejiang Univ-Sci A (Appl Phys & Eng) 2011 12(8):627-636 633

1.0
stress and shear stress, the equivalent stress will be
(a) smaller compared with the practical situation, hence,
ABAQUS result
0.8 Theoretical result later occurrence of plastic plateau. Likewise, the
growth of ellipticity is somewhat delayed. The suit-
ΔD/D (%)

0.6
ability of the theoretical method used in predicting the
0.4 buckling response of deepwater pipes has been vali-
dated herein.
0.2
4.3 Parametric study
0.0
0 1000 2000 3000 4000 The theoretical model is adopted to examine the
Water depth (m)
effects of several important factors including tension
1.0
(b) T, strain-hardening parameter n, yield stress σy as well
ABAQUS result
0.8 Theoretical result as diameter-to-thickness ratio D/t. T→Radial(κ, P) is
the loading path considered in the present section.
0.6
Besides, some discussions and comparisons are made
ΔD/D (%)

0.4
concerning the design of pipes in engineering
practice.
0.2 The buckling of tube is related to several factors,
such as the diameter D, wall-thickness t, material
0.0
0.000 0.003 0.006 0.009 0.012 0.015 properties, initial ellipticity ΔD/D, and load history.
κ (1/m) In addition, residual stress induced in the manufac-
1.0 turing process as well as yield anisotropy play an
(c)
ABAQUS result important role in the occurrence of tube buckling. For
0.8 Theoretical result
offshore applications, a D/t value ranging from 10 to
0.6 70 is recommended. While for deepwater application,
ΔD/D (%)

a D/t value ranging from 10 to 35 is more suitable. In

0.4
addition, the yield strength of steel for typical off-
0.2 shore pipelines is commonly between 276 and
448 MPa. Besides, the tubes, with initial ellipticity
0.0
0 100 200 300 400 500 600
exceeding 0.5%, should be avoided in the deepwater
T (kN) applications (Ju and Kyriakides, 1991).
Figs. 9 and 10 show that axial tension has a sig-
Fig. 8 Comparisons of finite element analysis results with
nificant effect on bending moment carrying capacity
theoretical solutions
(a) Ellipticity-water depth; (b) Ellipticity-curvature; (c) of a pipe. The tension is prescribed to 500, 1000 and
Ellipticity-tension 1500 kN respectively, and the ratio of (P/P0):(κ/κ0)
ranges from 5:1, 1:1, 1:3 to 1:8, respectively denoted
The main reason for the difference is that as Radial 1–Radial 4, which consists of 12 different
ABAQUS uses a finite deformation J2 flow theory of load combinations. The result indicates that the
plasticity whereas the theoretical formulation in presence of tension impairs bending moment carrying
Eqs. (7)−(11) is small deformation. In addition, the capacity greatly. With the increase of tension applied,
theoretical model simplifies this 3D problem to a 2D the limit moment Mc drops. Furthermore, it can also
one, which only takes into account the stresses along be observed that the increase of external pressure will
the axial and circumferential directions. The disre- cause the value of limit moment to decrease. Addi-
gard of the secondary radial stress and shear stress tionally, as can be seen in Fig. 10, the predicted
will not generate much error in the elastic range. critical pressure Pc and critical curvature κc become
However, with the increase of stress in the radial smaller when the value of tension increases. It is
direction, the discrepancies become more and more important to note that the results are normalized to
notable. Moreover, due to the disregard of radial dimensionless factors by the following variables:
634 Gong et al. / J Zhejiang Univ-Sci A (Appl Phys & Eng) 2011 12(8):627-636

M 0 = σ 0 D02 t , P0 = 2 σ 0 t / D0 , κ 0 = t / D02 , (18) the critical pressure and curvature is examined in

Fig. 13. Three D/t values 15, 20 and 25 are adopted,
where mean diameter D0=D−t, and σ0 is API yield while keeping other parameters constant. Just as ex-
stress (API, 2004), i.e., the stress at a strain of 0.005. pected, the limit values corresponding to lower D/t
tubes are higher than those of larger D/t ones. In ad-
dition, note that the degree of its influence varies with
1.0
different combinations of loads applied.
0.8
0.5
0.6 Radial 1 σy/E=0.0023
Mc/M0

0.4 σy/E=0.0019
0.4 σy/E=0.0015
Radial 1 Radial 2

Pc/P0
Radial 2 0.3
0.2
Radial 3
Radial 4 Radial 3
0.2
0.0
0 500 1000 1500 2000
Radial 4
T (kN)
0.1
Fig. 9 Limit moment versus applied tension
0.0 0.2 0.4 0.6 0.8 1.0
κc /κ0
0.6
Fig. 11 Effects of yield stress on critical pressure and
0.5 Radial 1 curvature
T=500 kN
T=1000 kN
0.4 T=1500 kN
Radial 2 0.5
Pc/P0

0.3
Radial 3 0.4 Radial 1 n=10.7
0.2 n=17
Radial 4 Radial 2 n=30
0.3
0.1
Pc/P0

Radial 3
0.0 0.2
0.0 0.2 0.4 0.6 0.8 1.0 1.2
κc /κ0 Radial 4
0.1
Fig. 10 Effects of tension on critical pressure and
curvature 0.0
0.0 0.2 0.4 0.6 0.8 1.0
κ c / κ0
Fig. 11 shows how the critical pressure and
Fig. 12 Effects of strain-hardening parameter on critical
critical curvature vary with the material yield stress σy pressure and curvature
with other parameters kept constant. Clearly, the
tubes with larger yield stress possess higher critical
pressure and curvature. In addition, it is also worth 0.5
noting that at higher curvatures the effect of yield Radial 1 D/t=15
stress is less pronounced compared with the cases of 0.4 D/t=20
D/t=25
lower curvatures. 0.3
Radial 2
Pc/P0

Larger strain-hardening parameter n means lar-

Radial 3
ger strain-hardening effect. Fig. 12 presents the 0.2
Radial 4
variation of critical pressure and critical curvature
0.1
with the strain-hardening parameter n. It can be ob-
served that tubes with larger n can sustain larger 0.0
0.0 0.2 0.4 0.6 0.8 1.0
critical pressure and curvature, i.e., higher load- κc /κ0
carrying capacity.
Fig. 13 Effects of D/t on critical pressure and curvature
The effect of diameter-to-thickness ratio D/t on
 Gong et al. / J Zhejiang Univ-Sci A (Appl Phys & Eng) 2011 12(8):627-636 635

5 Conclusions periments under simultaneous tension, bending, and

external pressure should be carried out, and thus, the
The buckling behavior of the pipes under effectiveness of this theoretical method can be care-
T→Radial(κ, P) load path is presented in the fact that fully examined.
this load path could approximate the loading history
experienced by the pipeline in the sagbend region for References
deepwater pipe-laying operation. Comparisons be- Al-Sharif, A.M., Preston, R., 1996. Simulation of Thick-
tween the results of theoretical analysis and those of Walled Submarine Pipeline Collapse under Bending and
numerical simulations demonstrate that the theoreti- Hydrostatic Pressure. Proceedings of Offshore Technol-
cal formulation can accurately predict the buckling ogy Conferences, Houston, Texas, USA, OTC8212,
p.589-598.
behavior of deepwater pipes under simultaneous ten-
API (American Petroleum Insitute), 2004. API Specifications
sion, bending, and external pressure.
5L: Specifications for Line Pipe (43rd Ed.). API Pub-
A series of parametric study on the buckling lishing Services, Washington DC, USA.
response of deepwater pipes subjected to T→Radial Bai, Y., Igland, R.T., Moan, T., 1997. Tube collapse under
(κ, P) load path is conducted adopting the theoretical combined external pressure, tension and bending. Marine
formulation developed. The following conclusions Structures, 10(5):389-410. [doi:10.1016/S0951-8339(97)
can be drawn. 00003-8]
Corona, E., Kyriakides, S., 1988. On the collapse of inelastic
1. The load-carrying capacity of the tube is sig-
tubes under combined bending and pressure. Interna-
nificantly affected by the tension applied. With the tional Journal of Solids and Structures, 24(5):505-535.
increase of tension, the limit moment obviously drops, [doi:10.1016/0020-7683(88)90005-4]
and the predicted critical pressure and curvature Corona, E., Lee, L.H., Kyriakides, S., 2006. Yield anisotropy
would become smaller. In the case of equal propor- effects on buckling of circular tubes under bending.
tional loading between external pressure and curva- International Journal of Solids and Structures, 43(22-23):
7099-7118. [doi:10.1016/j.ijsolstr.2006.03.005]
ture, the effect of tension on load-carrying capacity of
Dyau, J.Y., Kyriakides, S., 1992. On the response of elastic-
the tube is less conspicuous compared with other plastic tubes under combined bending and tension.
loading ratio. Journal of Offshore Mechanics and Arctic Engineering,
2. The buckling behavior and load-carrying ca- 114(1):50-62. [doi:10.1115/1.2919952]
pacity of pipes is quite sensitive to material properties. Gellin, S., 1980. Plastic buckling of long cylindrical shells
Larger yield stress σy and strain-hardening parameter under pure bending. International Journal of Solids and
n always lead to higher limit pressure and curvature, Structures, 16(5):397-407. [doi:10.1016/0020-7683(80)
90038-4]
i.e., stronger resistance to pipe buckling. The critical
Hibbitt, H.D., Karlsson, B.I., Sorensen, P., 2006. ABAQUS
pressure of the pipe is more susceptible to yield stress Theory Manual, Version 6.3. Pawtucket, Rhode Island,
rather than strain-hardening parameter, whereas the USA.
critical curvature is just the contrary. Therefore, the Ju, G.T., Kyriakides, S., 1991. Bifurcation buckling versus
high strength steel is preferred to improve the resis- limit load instabilities of elastic-plastic tubes under
tance to external pressure for deepwater pipes in the bending and external pressure. Journal of Offshore Me-
practical engineering. chanics and Arctic Engineering, 113(1):43-52. [doi:10.
1115/1.2919895]
3. Diameter-to-thickness ratio D/t plays a very Kashani, M., Young, R., 2005. Installation load consideration
important role in buckling response of pipes. In gen- in ultra-deepwater pipeline sizing. ASCE Journal of
eral, pipes with lower D/t values possess stronger Transportation Engineering, 131(8):632-639. [doi:10.
capability to resist the buckling deformation. But, the 1061/(ASCE)0733-947X(2005)131:8(632)]
degree of its influence varies with different combina- Kyriakides, S., Shaw, P.K., 1982. Response and stability of
tion of loads applied. elastoplastic circular pipes under combined bending and
external pressure. International Journal of Solids and
In summary, it can be concluded that the theo-
Structures, 18(11):957-973. [doi:10.1016/0020-7683(82)
retical formulation and solution method described in 90086-5]
this context could provide a reasonably-accurate es- Kyriakides, S., Corona, E., 2007. Mechanics of Offshore
timate of the buckling and collapse of deepwater Pipelines, Volume 1: Buckling and Collapse. Elsevier
pipes. In addition, it should be mentioned that ex- Science, Oxford, UK and Burlington, Massachusetts.
636 Gong et al. / J Zhejiang Univ-Sci A (Appl Phys & Eng) 2011 12(8):627-636

Li, Z.G., Wang, C., He, N., Zhao, D.Y., 2008. An overview of Methods in Engineering, 33(7):1413-1449. [doi:10.1002/
deepwater pipeline laying technology. China Ocean En- nme.1620330705]
gineering, 22(3):521-532. Yuan, L., Gong, S.F., Jin, W.L., Li, Z.G., Zhao, D.Y., 2009.
Simo, J.C., Armero, F., 1992. Geometrically non-linear en- Analysis on buckling performance of submarine pipelines
hanced strain mixed methods and the method of incom- during deepwater pipe-laying operation. China Ocean
patible modes. International Journal for Numerical Engineering, 23(2):303-316.