You are on page 1of 9

Proceedings of the ASME 2014 33rd International Conference on Ocean, Offshore and Arctic Engineering

OMAE2014
June 8-13, 2014, San Francisco, California, USA

OMAE2014-23690

PREDICTING HYDROSTATIC COLLAPSE OF PIPES USING FINITE ELEMENT


ANALYSIS
Ajit Bastola Junkan Wang
Det Norske Veritas Det Norske Veritas
London, UK London, UK

Ali Mirzaee-Sisan James Njuguna


Det Norske Veritas Robert Gordon University
London, UK Aberdeen, UK

ABSTRACT make determination of the exact collapse pressure difficult


A pipeline’s resistance to collapse is governed by through analytical method. Hence the finite element analysis
geometric imperfections, material properties and residual technique is often the preferred tool to estimate the critical
stresses. The offshore pipeline design code DNV-OS-F101 collapse pressure to optimize design at early stage if the actual
provides a method for predicting collapse of pipelines with geometry and material properties of the pipe are known.
diameter to wall thickness (D/t) ratios between 15 and 45. This
paper examined the various factors that could influence the LITERATURE REVIEW
collapse resistance of several pipe geometries, such as ovality, There has been substantial research since the 1980s on the
eccentricity, material stress-strain behavior and residual stresses pipe’s resistance to hydrostatic collapse [1-14]. Various
in the hoop and longitudinal directions. A total of 132 cases formulae exist that provide a good estimate on the collapse
were carried out, using 2D and 3D Nonlinear Finite Element pressure of pipes at intermediate D/t ratios. However, it is
Analysis, to predict the collapse pressure of several realistic understood that these formulae have originated from semi-
pipe geometries. Results of this study suggest that the DNV- analytical solutions covering both elastic collapse and plastic
OS-F101 predictions are conservative and applicable for a wide collapse for thin- and thick-walled sections [8]. During the
range of D/t ratios. While there is close correlation between installation phase, the pipeline must be designed to resist
Finite Element prediction and DNV-OS-F101 prediction, there external pressure combined with tension and bending loads,
is a degree of conservatism at low D/t ratios using DNV-OS- whereas in as-laid and operational conditions, pipeline collapse
F101 equations. Hence there would be scope for further is governed by external pressure only [4]. The operational
optimization of pipe wall thickness design against the collapse phase is the focus of this study.
limit state at low D/t ratios. DNV-OS-F101 (2013) [15] is one of the most advanced
and commonly used standards for designing offshore pipelines
worldwide. The equation prescribed by DNV-OS-F101 for
INTRODUCTION determining the critical collapse pressure of a pipe is based on
Deepwater gas transportation in excess of 2000m has been the method originally proposed by Haagsma and Schaap’s [9]
the frontier of hydrocarbon production and exploration for this which is a function of the pipe’s material properties, diameter to
decade. An example is the Cabiunas gas export pipeline, thickness ratio and ovality of its cross-section. By ignoring the
offshore Brazil from Santos Basin to Lagomar Beach with a influence on pipe’s collapse capacity by fabrication history,
maximum water depth of 2250m. temperature de-rating and material strength factor, as these are
Conventional pipelines are designed to withstand internal outside the scope of this study, the DNV-OS-F101 expression
pressure. However, the critical factor for deepwater pipelines is takes the simplified form of Equations (1):
its resistance to external pressure, or pipeline collapse,
particularly during installation when there is little or no internal D
pressure. External collapse is governed by material properties ( Pc  Pel )( Pc2  Pp2 )  Pc Pel Pp f o (1)
and geometry of the pipeline. Combinations of these properties t

1 Copyright © 2014 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/31/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


assembly. These stresses can be tensile or compressive and
where: these can be distributed over one, two or three dimensions [18].
𝑃𝑐 is the critical net external pressure that will cause collapse; In addition, no methods to predict quantitatively the
3 manufacturing-induced residual stress magnitude in the hoop or
2E  t 
Pel    is the pipe’s elastic collapse resistance other directions are detailed in the literature [19]. On the one
1  2  D  hand, Yeh & Kyriakides, (1986), [13] and Liessem et al.,
capacity; (2007), [20] stated that residual stresses tend to reduce the
 yield 2t collapse resistance, and their opinion was supported by
Pp  is the pipe’s plastic collapse resistance experimental and numerical work by Gresnigt et al., (2000),
D
capacity; [7], Fallqvist, (2010), [6] and Wang et al., (2013), [19]. On the
𝐸 is Young’s modulus of the pipe material; other hand, Baek, (2011), [3] and Huang et al., (2000), [21]
𝜈 is Poisson’s ratio; suggest that residual stress less than the yield stress of the
𝐷 is the pipe’s nominal outside diameter; material does not affect the collapse pressure.
𝑡 is the pipes’ nominal wall thickness;
METHODOLOGY
D  Dmin
f 0  max is the ovality of the pipe; In this study, Finite Element Analysis (FEA) based on the
D ABAQUS 6.12 platform was used for assessing the limiting
Dmax is the maximum outside diameter; and collapse pressure of pipes. ABAQUS is a widely used FEA
Dmin is the minimum outside diameter. tool for non-linear analysis. Both 2D plane-strain and 3D
models were considered. 8-noded biquadratic plane strain
Ovality is the most common geometric imperfection in the quadrilateral element with hourglass control (CPE8R) was used
pipe’s cross-section for UOE pipes and is detrimental to the for 2D models, whereas 20-noded quadratic brick element
pipe’s resistance to collapse. Ovality can be global, due to (C3D20R) was used for 3D models. Both methods used
general out-of-roundness of the pipe’s cross-section, or local, reduced integration. Mesh sensitivity study was carried out and
due to localized defects such as dents. showed that 10 elements through-thickness (for both 2D and
For seamless pipes, on the other hand, the most commonly 3D models) and 50 elements along the longitudinal direction is
observed geometric imperfection is eccentricity (ϵ). sufficient to ensure accurate results are achieved. Figure 1
Eccentricity describes the extent to which wall thickness varies shows the 2D and 3D meshes for 2 particular cases considered
around the pipe’s hoop direction. Like ovality, eccentricity can in this study.
be either global (such as uniform variation in thickness) or local
(such as gouges) depending on the source, and can be defined
generally as:

t t
Eccentrici ty (ϵ) = max min (2)
t avg

where 𝑡𝑚𝑎𝑥 , 𝑡𝑚𝑖𝑛 and 𝑡𝑎𝑣𝑔 are the maximum -, minimum - and
average thickness respectively.
The influence of local ovality and eccentricity on a pipe’s
fitness-for-service should be analyzed on a case-by-case basis Figure 1 Meshes used on 2D and 3D pipes
[16]. Hence, this paper focuses on the influence of global
ovality and eccentricity effects on a pipe’s resistance to A brief overview of the analysis program is presented in
collapse. Table 1. A total of 132 cases have been studied covering D/t
Apart from geometric imperfections, offshore pipes can ratios approximately ranging from 10 to 25, and with wall
undergo various loading, unloading and reloading cycles during thicknesses approximately between 8 mm and 22 mm. These
manufacturing, installation and operational processes. For cases are subdivided into 7 Case Groups. The collapse
example, the UOE manufacturing process induces tensile pressures for each of the individual cases considered in Table 1
plastic strain in the pipe’s hoop direction that reduces the were also compared with DNV-OS-F101 predictions.
compressive yield stress because of the Bauschinger effect, [17] 10 cases of Case Group I compared the collapse pressures
and is detrimental to the collapse resistance of the pipe. calculated using 2D models and 3D models for the ovality and
Material models used in pipe collapse modelling should capture eccentricity of 0.5% for a specific material model. 15 cases of
this effect to produce realistic results. Case Group II.1 and another 15 cases of Case Group II.2
Residual stress is another factor that can affect the collapse compared the effect of ovality for the range of D/t using
resistance of pipelines [6], [13]. The source of residual stress is different grades of steel and material models. Collapse
variable and can be because of manufacturing, processing and pressures calculated for 15 cases of Case Group III considered

2 Copyright © 2014 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/31/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


the effect of eccentricity only. The effect of eccentricity only on
the collapse pressure was not conclusive, hence, further 10
cases were compared to find the combined effect of ovality and
eccentricity on the collapse pressure. 57 cases considered on
Case Group V and 10 cases considered on Case Group VI
looked at the effect of residual stresses on collapse pressures.
Case Group V looked at distribution of residual stresses on the
hoop direction of the pipe and Case Group VI looked at the
residual stress distribution on the longitudinal direction of the
pipe. Figure 2 Ovality (left) and Eccentricity (right) as implemented in
this study
Table 1 Details of tests considered
Case No. of Material σr
D/t 𝒇 [%] 𝝐 [%]
Group Cases model [%σy ]
I 10 RO (X65) [10, 25] 0.5 0.5 -
II.1 15 RO (X65) [10, 25] 0.5, 1, 2 - -
II.2 15 LP (X70) [10, 25] 0.5, 1, 2 - -
1, 5, 10,
III 15 RO (X65) [10, 25] - - Figure 3 Combination of ovality and eccentricity considered in this
25
study
5, 10,
IV 10 RO (X65) 10, 20 2 15, 20, -
25 Throughout all 132 cases, only external hydrostatic load
was applied on the outer surface of the pipe to simulate air-
25, filled as-laid condition which is most critical for the collapse
V 57 RO (X65) [10, 25] 0.5 - 37.5,
50, 75 limit state. For 3D models, it is important to consider end-cap
effects where the difference between the force applied on the
25, internal cross-sectional area of the pipe and that applied on the
VI 10 RO (X65) 20, 25 0.5 - 37.5,
50, 75 external cross-sectional area leads to end-cap forces. This end
cap forces give rise to either tension or compression in the axial
direction of the pipe. Published literatures, [12], [13] and [23],
As mentioned previously, only global ovality and
suggest that end effects have little effect on the critical collapse
eccentricity are considered in this study. Ovality is
pressure when the model pipe’s length to diameter (L/D) ratio is
implemented by modeling the pipe’s cross section as two
greater than 10. In this study, all 3D studies were considered
concentric ellipses with the differences between their semi-
with L/D = 10 with both ends fully fixed.
major axes and semi-minor axes the same and equal to the
Two material models were considered in this study: the
pipe’s wall thickness. Eccentricity is implemented by modeling
Ramberg-Osgood model (RO), [24], and the Lüder's plateau
the pipe’s cross section as two non-concentric circles with the
(LP) model. The RO material model is a dynamic mathematical
distance between them equal to 0.5×ε×tavg. Figure 2 illustrates
representation of an experimentally obtained stress-strain curve
schematically how ovality and eccentricity are defined and
[24]. In this particular study, the RO model implemented has its
implemented in this FEA-based study.
yield stress (at 0.5% engineering strain) and tensile strength (at
Two types of ovality and eccentricity combination were
10.0% engineering strain) corresponding to that of a typical
considered. The largest thickness variation (tmax-tmin) and major
API 5L X65 steel [25]. The LP material model, on the other
axis of the elliptical pipe were in phase for one type.
hand, represents that of a typical API 5L X70 [25] steel
Meanwhile, the largest thickness variation (tmax-tmin) and the
featuring a yield strength of 537 MPa. This model incorporates
major axis of the elliptical pipe were out of phase by 90° for
a hardening behavior with a small plateau between 0.2% and
another type. These two types of ovality and eccentricity
1% true strain after which hardening occurs. Figure 4
combination are illustrated schematically in Figure 3.
illustrates, through true stress-strain plots, the two material
Calculated collapse pressures for a few cases of Case Group IV
models used in this study.
demonstrated identical results for these two combinations.
Therefore, results suggested that collapse pressures are
independent of ovality and eccentricity combination. Study by
Yeh et al., (1986) [13] had also reached to the same conclusion.

3 Copyright © 2014 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/31/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


800 800 direction as that was initially applied. It should be noted that
700 700 this study followed the sign convention that compressive
600 600 stresses were negative and tensile stresses were positive.
σ (True) [MPa]

σ (True) [MPa]
500 500
400 400 450

Residual Stress after equilibrium


350
300 300
250
200 200
150
100 100
50

[MPa]
0 0 -50
0 0.1 0.2 0.3 0 0.05 0.1 0.15
-150
ε (True) ε (True)
-250
a. Ramberg Osgood material model b. Lüder's Plateau material model for
-350
for X65 Grade steel X70 Grade steel
-450
Figure 4 Material models considered
0 10 20
thickness, t
Collapse pressures of the pipe models were determined by
Residual Stress-After equilibrium
the Riks method, commonly used for unstable collapse and Initial Stress-Before equilibrium
post-buckling analysis. This method is suitable if there are
Figure 6 Hoop stress before and after self-equilibrium for a case from
concerns about material and/or geometric nonlinearity prior to Case Group V.
buckling, or unstable post buckling response [26]. In this study,
the collapse pressure is determined by plotting the accumulated
external pressure increase against the displacement of the RESULTS
model pipe’s zenith relative to its nadir, as illustrated in Figure The effect of various factors that could influence the
5. When the curve reaches maximum external pressure, strain collapse resistance of several pipe geometries, such as ovality,
energy is released to maintain self-equilibrium, i.e. the collapse eccentricity, material stress-strain behavior and residual stresses
pressure is reached. in the hoop and longitudinal directions were analyzed.

25 Comparison between 2D and 3D Models


20 120
2D models
100
Pc [MPa]

15 3D models
80
10
Pc[MPa]

60
5
40
0
0 10 20 30 40 50 60 70 20
Displacement [mm] 0
Figure 5 Example showing how collapse pressure is determined 9 14 D/t 19 24
using the Riks method in this study (a case from Case Group II.1)
Figure 7 Comparison of collapse pressures obtained from 2D
models and 3D models (Case Group I)
In order to generate residual stress fields in the model pipes
for Case Groups V and VI, initial stresses were applied as an Case Group I compares the collapse pressures obtained
initial condition directly to elements before application of from 2D models and the same from 3D models. Figure 7 plots
external pressure, as shown in Figure 6. For Case Group V, the collapse pressure against the D/t ratios for all analysis cases
initial stress was applied in the hoop direction whereas for Case within this Case Group. The figure shows that collapse
Group VI, initial longitudinal stresses were applied. Before pressures obtained through 2D models are almost identical to
subjecting the model pipe to external pressure, an additional those obtained using 3D models. The maximum difference
load step was introduced so that the applied initial stress could between the two is less than 2% and at D/t = 25. At lower D/t
re-distribute and achieve a steady state residual stress field that ratios, the difference is around 0.5%. Results from this study
is in self-equilibrium. A few initial stress profiles have been have suggested that end effects are negligible for 3D models
assumed and applied to the pipe either in the hoop or with L/D ≥ 10, and that computationally efficient 2D models
longitudinal direction. Figure 6 illustrates the difference can be used for the purpose of this study. Figure 8 illustrates the
between the initial stress profiles applied and the eventual deformation mechanism right after collapse for both the 2D and
residual stress field in the pipe after self-equilibrium. Results 3D models. As is evident from the figure, hoop stress is tensile
from Case Groups V and VI show that, after equilibrium, the for the inner layers but compressive for the outer layers in the
resulting residual stress field was predominantly in the same collapsed pipe for both the 2D and 3D models.

4 Copyright © 2014 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/31/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


method is only applicable for D/t ranges between 15 and 45, it
may be concluded that, for thick-walled pipes with D/t ratios
below 15, the DNV-OS-F101 method is likely to under-predict
their resistance to collapse, and therefore there could be scope
for further optimization of this method for these thick-walled
pipes.

140
FE,fo=0.5%
(a) 120 FE,fo=1%
FE,fo=2%
100
DNV eqns,fo=0.5%
80

Pc[MPa]
DNV eqns,fo=1%
DNV eqns,fo=2%
60

40

20

(b) 0
8 12 16 D/t 20 24 28
Figure 8 Post-collapse hoop stress distribution for (a) a 2D model,
and (b) a 3D model with the same cross-sectional geometry (the Figure 10 Influence of ovality on collapse pressure for X70 grade
cases from Case Group I) steel

Effect of Ovality Effect of Material Model and D/t Ratio


Case Groups II.1 and II.2 examined the influence of initial 0.25
RO - fo=0.5%
ovality on the collapse pressure of pipes. Figure 9 plots the
RO - fo=1%
collapse pressures against D/t ratios for the analyses cases using 0.20
RO - fo=2%
the RO material model within these two Case Groups whereas LP - fo=0.5%
Figure 10 plots the same for all analyses cases using the LP 0.15 LP - fo=1%
Pc/σy

material model. In addition, calculated results using the LP - fo=2%


simplified DNV-OS-F101 method, i.e. Equations (1), were also 0.10
included in both figures for comparison.
0.05
120
FE, fo=0.5%
100 FE,fo=1% 0.00
FE,fo=2% 8 12 16 20 24 28
D/t
80 DNV eqns,fo=0.5%
Figure 11 Influence of ovality on normalised collapse pressure with
Pc[MPa]

DNV eqns,fo=1%
60 yield strength
DNV eqns,fo=2%

40 Figure 11 combines all data points in given in Figure 9 and


Figure 10 by normalizing the collapse pressure with the
20
material’s yield strength. It is evident that the trend between
0 normalized collapsed pressure and D/t ratio for the two material
8 12 16
D/t
20 24 28 models is very similar, suggesting that the pipe’s material
stress-strain behavior only has minor impact on its normalized
Figure 9 Influence of ovality on collapse pressure for X65 grade
resistance to collapse. The DNV-OS-F101 method supports this
steel
conclusion as it does not require detailed information on the
From both figures, it is evident that an increase in initial material’s strain hardening behavior.
ovality would lead to a decrease in the pipe’s resistance to
collapse, and that the DNV-OS-F101 method suggests a similar Effect of Eccentricity
trend. However, the DNV-OS-F101 method seems more DNV-OS-F101, (2013) provides detailed restrictions on the
conservative at lower D/t ratios than at higher ones. At the allowable thickness tolerances, i.e. variations, based on a line
lowest D/t ratio considered, D/t = 9.86, the DNV-OS-F101 pipe type and wall thickness. As eccentricity are most common
predictions can be up to 20% lower than the corresponding for seamless pipes, and that the range of pipe wall thicknesses
FEA analysis result for the same pipe. As the DNV-OS-F101 considered in this study is between 8.74 mm and 22.225 mm,

5 Copyright © 2014 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/31/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


the maximum allowable eccentricity is 27.5%, as demonstrated covered by analysis Case Group VI. As mentioned previously,
by Table 2 below. For the purpose of this study, four different Case Group V considers two different through-thickness
eccentricity values were hence considered: 1%, 5%, 10% and profiles of hoop residual stresses where the initially applied
25%. hoop stress at the pipe’s outer surface was either tensile (Type
A) or compressive (Type B). For analysis Case Group VI, only
Table 2 Tolerances for the thickest and thinnest pipe allowed by one through-thickness profile of longitudinal residual stress was
DNV OS-F101 examined where the initially applied residual stress was tensile
Tolerance at the pipe’s outer surface.
t D/t ϵmax
based on DNV-OS-F101 Figure 13 plots the collapse pressures from all analysis
[mm] [-] [mm] [%] cases within Case Group V normalized with the corresponding
8.74 25.07 +0.15t – 0.125t 27.5 residual-stress-free cases (Case Group II.1). It is evident that
22.225 9.86 ± 0.125t 25.0 residual hoop stresses start to have noticeable detrimental
effects on a pipe’s resistance to collapse when D/t is greater
Case Group III examined the effect of eccentricity alone on than 15. This effect is negligible for pipes with D/t ratios less
the collapse resistance of pipes containing no other geometric than 12, but becomes more significant as D/t ratio and/or the
imperfections. Analysis results showed that even at 25% magnitude of the residual hoop stress increases. For a pipe with
eccentricity, the resulting collapse pressure was still higher than D/t ratio of 25, presence of residual hoop stress could reduce its
that of a pipe containing 0.5% ovality but no eccentricity. resistance to collapse by nearly 20%. This conclusion is
Hence, ovality is more detrimental to a pipe’s resistance to supported by previous studies by Fallqvist, (2010) [6], Yeh and
collapse than eccentricity, and that eccentricity alone was not Kyriakides, (1986) [13] and Wang et al., (2013) [19]. Another
sufficient to induce collapse by external overpressure. interesting observation is that Type A (tensile at the pipe’s outer
Case Group IV considered the effect on collapse pressure surface) residual hoop stresses appear more detrimental than
from increased eccentricity combined with some finite ovality Type B (compressive at the pipe’s outer surface) residual
value (2%) for a low D/t case (D/t = 10) and a higher D/t case stresses. The cause of such difference is still under
(D/t = 20). The FEA results obtained and their comparison investigation.
with the DNV-OS-F101 predictions are illustrated in Figure 12.
1.05
100 1.50
90 1.40
80 1.30 1.00
Pc /Pc DNV

70
Pc [MPa]

1.20
60
1.10
Pc / Po

50 0.95
1.00 D/t=25-TypeA
40
30 0.90 D/t=20-TypeA
20 0.80 0.90 D/t=15-TypeA
5 15 25 5 15 25 D/t=12-TypeA
Eccentricity (ϵ) Eccentricity (ϵ)
D/t=10-TypeA
D/t = 19.69 D/t = 19.69 0.85 D/t=25-TypeB
D/t = 9.86 D/t = 9.86
D/t=20-TypeB
Figure 12 Effect of eccentricity with 2% initial ovality D/t=15-TypeB
0.80
D/t=12-TypeB
Figure 12 shows that, at D/t = 19.69, increasing the
D/t=10-TypeB
eccentricity from 5% to 25% only results in a 2.5% decrease in
the collapse pressure, and that at D/t = 9.86, the decrease is 0.75
0 20 40 60 80
even less at 1.35%. Hence, when combined with ovality, Applied residual stress after self-equilibrium (% of yield)
eccentricity has minimal effect on the collapse resistance of
pipes. The DNV-OS-F101 formulation supports this conclusion Figure 13 Effects of residual stress in hoop direction
as eccentricity does not affect the predicted collapse pressure.
Analysis results for Case Group VI, i.e. model pipes with
Effect of Residual Stress initial longitudinal residual stresses, are summarized in Figure
A few initial stress profiles were numerically introduced 14. The normalization method is the same as that adopted in
into the FEA models to investigate the effect of residual stress Figure 13. A comparison between Figure 13 and Figure 14
on pipe’s collapse resistance. Both hoop (S22) and longitudinal shows that residual hoop stress can have much greater influence
(S33) residual stress effects were studied. The former was on the pipe’s collapse pressure than residual longitudinal
covered by analysis Case Group V, whereas the latter was stresses. For example, considering D/t = 25, the collapse

6 Copyright © 2014 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/31/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


pressure was reduced by 17% for an applied residual hoop corresponding FE models in this study. In addition, detailed
stress of peak magnitude at 70% of yield stress, whereas the tensile properties were not provided for the experimental
reduction was only 5% for an applied residual longitudinal studies in [14] and [20], and hence generic Ramberg-Osgood
stress of the same peak magnitude. However, despite its minor stress-strain curves as well as minimum yield strength and
influence, the effect of residual longitudinal stress on the pipe’s tensile strength values were assumed based on the steel grades
collapse capacity cannot be ignored, particularly for pipes with given. Despite such uncertainties, the agreement between the
D/t ratios greater than 20. published experimental data and results from this FE study was
reasonably good.
1.01

1.00
CONCLUSION
0.99
This paper examined the various factors that could
Pc / Po

0.98 influence the collapse resistance of several pipe geometries,


0.97 such as ovality, eccentricity, material stress-strain behavior and
D/t=25 residual stresses in the hoop and longitudinal directions. A total
0.96
D/t=20 of 132 cases were carried out, using 2D and 3D Nonlinear
0.95
Finite Element Analysis, to predict the collapse pressure of
0.94
several realistic pipe geometries. Riks method was used to
0.00 20.00 40.00 60.00 80.00
determine the critical collapse pressure for each analysis case.
Applied residual stress after self-equilibrium (% of yield)
Results from this study have suggested that end effects are
small for 3D models with length to diameter ratios above 10,
Figure 14 Effects of residual stress in longitudinal (axial) direction
and that an increase in initial ovality would lead to a decrease
in the pipe’s resistance to collapse. This trend is supported by
Analysis results for Case Group VI, i.e. model pipes with
DNV-OS-F101.
initial longitudinal residual stresses, are summarized in Figure
Overall, good correlation between DNV-OS-F101
14. The normalization method is the same as that adopted in
predictions and FEA results is observed for all cases.
Figure 13. A comparison between Figure 13 and Figure 14
Furthermore, the DNV-OS-F101 predictions are conservative
shows that residual hoop stress can have much greater influence
and applicable for a wide range of D/t ratios. While there is
on the pipe’s collapse pressure than residual longitudinal
close correlation between Finite Element prediction and DNV-
stresses. For example, considering D/t = 25, the collapse
OS-F101 prediction, there is a degree of conservatism at low
pressure was reduced by 17% for an applied residual hoop
D/t ratios using DNV-OS-F101 equations. Hence there would
stress of peak magnitude at 70% of yield stress, whereas the
be scope for further optimization of pipe wall thickness design
reduction was only 5% for an applied residual longitudinal
against the collapse limit state at low D/t ratios.
stress of the same peak magnitude. However, despite its minor
In addition, it is found that, regarding geometric
influence, the effect of residual longitudinal stress on the pipe’s
imperfections, ovality is much more detrimental to a pipe’s
collapse capacity cannot be ignored, particularly for pipes with
resistance to collapse than eccentricity.
D/t ratios greater than 20.
It is found that the pipe’s material stress-strain behavior
only has minor impact on its resistance to collapse if the
Comparison with Published Experimental Results
Table 3 Comparison of FE results with published
collapse pressure is normalized by the material’s yield strength.
experimental data Analysis results, from cases with the few residual stress
profiles applied, suggest that residual hoop stress starts to have
Reference Description D/t 𝒇 [%] Difference*
a noticeable effect on a pipe’s resistance to collapse when the
5CC 17.8 0.17 7.6% pipe’s diameter to wall thickness ratio is greater than 15. Such
[14] 3RB 17.8 0.18 3.5% effect is negligible for pipes with the same ratios less than 12,
but becomes more significant as D/t ratio and/or the magnitude
8NB 18.1 0.15 12.6%
of the residual hoop stress increases. In addition, residual hoop
28" × 38
[20]
mm X65
18.9 0.17 0.6% stress has much greater detrimental influence on the pipe’s
collapse pressure than residual longitudinal stresses. Residual
Pc, FE  Pc, Experimental
*
Difference is defined as: Difference   100% stress assumed in this study was purely numerical.
Pc, Experimental
Experimental work is recommended to improve the
understanding of the distribution of residual stress in the pipe in
Table 3 compares results from this study with previously order to improve and validate the FEA model.
published experimental data in open literature [14], [20]. It is This paper highlights DNV GL’s on-going research on
worth noting that, although the maximum ovality values were thick-walled pipes. DNV GL has recently launched a JIP on the
the same, the real pipe-cross sections did not feature the same same topic which includes analytical studies, FE analyses and
idealized imperfection shapes as those assumed in the experimental validation.

7 Copyright © 2014 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/31/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


ACKNOWLEDGEMENT REFERENCES
The authors would like to thank Olav Aamlid from [1] DNV, Snamprogetti, and Sintef, 1996, "Buckling and
DNVGL for his kind support during this research and for Collapse Limit State," SUPERB Project.
reviewing this paper.
[2] Al-Sharif, A. M., and Preston, R., 1996 "Simulation of
Thick walled Submarine Pipeline Collapse under Bending
NOMENCLATURE and Hydrostatic Pressure," Offshore Technology
API American Petroleum Institute Conference (OTC 1996), Houston,Texas, p. no OTC 8212.
D Nominal diameter [3] Baek, J.H., 2011 "Effect of Ovality and Eccentricity on
DNV Det Norske Veritas Collapse Pressure of Subsea Pipelline," International Gas
Union Research Conference, Ansan, Korea, p.no
Dmax Maximum outside diameter
IGU2011-P2-34.
Dmin Minimum outside diameter [4] Benjamin, A. C., and Cunha, D. J. S., 2012, "Assessment
E Young's Modulus of Hydrostatic Collapse of Submarine Pipelines: Historical
fo Ovality Review of the Classic Methods," 9th International
Pipeline Conference (IPC2012), Calgary, Canada, p. no
L Length of pipe
IPC2012-90617.
LP Lüder's Plateau Material Model
[5] DeGeer, D., and Cheng, J. J., 2000, "Predicting Pipeline
OD Outer Diameter Collapse Resistance," International Pipeline Conference,
Pc Collapse Pressure Alberta, Canada, p. no IPC 00-0168.
Pc DNV Collapse Pressure by DNV method [6] Fallqvist, B,. 2010, "Collapse of thick deepwater pipelines
use to hydrostatic pressure," Master thesis report,
Pel Elastic Capacity
Department of Solid Mechanics, Royal Institute of
Reference collapse pressure without consideration
Technology, Stockholm, Sweden.
Po of relevant parameter
[7] Gresnigt, A. M., Foeken, R. J. V. and Chen, S., 2000
Pp Plastic Capacity
"Collapse of UOE Manufactured Steel Pipes," 10th
RO Ramberg-Osgood Material Model International Offshore and Polar Engineering Conference,
t Nominal thickness (ISOPE 2000), Seattle, USA p. no. ISOPE-I-00-138.
tavg Average thickness [8] Guarracino, F., Fraldi M., Freeman, R., and Slater, S.,
2011, "Hydrostatic Collapse of Deepwater Pipelines: A
tmax Maximum thickness
Rigorous Analytical Approach," Offshore Technology
tmin Minimum thickness Conference (OTC 1996), Houston, Texas, p. no OTC
ϵ Eccentricity 21378.
ε Strain [9] Haagsma, S., and Schaap, D., 1981 "Collapse resistance of
submarine lines studied," Oil and Gas Journal, 79 (5), pp.
εnom Nominal strain (engineering)
86-90, 1981.
εtrue True strain
[10] IntecSea, 2009, "South Stream Feasiblity Study - Scope of
σnom Nominal stress Work for Collapse Testing," Report, Delft, Netherlands.
σr Residual stress [11] Murphey, C. E., and Langner, C. G., 1985, "Ultimate Pipe
σtrue True stress Strength under Bending, Collapse and Fatigue,"
Proceedings of the 7th International Conference on
σy Yield Strength
Offshore Mechanics and Arctic Engineering (OMAE
1985).
[12] Tsuru, E., and Asahi, H., 2004, "Collapse Pressure
Prediction and Measurement Methodology of UOE Pipe,"
International Journal of Offshore and Polar Engineering,
14 ( 5), pp. 51-59.
[13] Yeh, M. K,. and Kyriakides, S., 1986, "Collapse of
Deepwater Pipelines," Offshore Technology Conference
(OTC 1986), Houston,Texas, p. no OTC 5215.

8 Copyright © 2014 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/31/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


[14] Stark, P., and McKeehan, D., 1995, "Hydrostatic Collapse 26] Dassault Systèmes, 2013, Abaqus 6.12 Documentation,
Research in Support of the Oman India Gas Pipeline," Dassault Systèmes Simulia Corp.
Offshore Technology Conference (OTC 1995), Houston, [27] Yasir, A., 2012, "The Cooling Rate Effect on the Buckling
Texas, p. no OTC 7705. Strength and Elastic Properties," International Journal of
[15] Det Norske Veritas, 2013, "Offshore standard DNV-OS- Mechanical & Mechatronics Engineering,12(03) pp.1.
F101: Submarine Pipeline Systems," Hovik, Norway. [28] British Standards, 2003, Code of practice for pipelines -
[16] Cosham, A., and Hopkins, P., 2005, "The Pipeline Defect Part 2: Subsea pipelines, (BSI PD 8010-2:2004), British
Assessment Manual (PDAM)," Penspen Limited, Standards Insitute, UK .
Newcastle upon Tyne, UK. [29] Chakrabarti, S., 2005, Handbook of Offshore Engineering,
[17] Chawla, N., Jester, B., and Vonk, D., 2003, "Bauschinger Elsevier B V, Amsterdam, Netherlands.
effect in porous sintered steels," Materials Science and [30] Jones, R., 2009, Deformation Theory of Plasticity, Bull
Engineering, Elsevier B V, pp. 266-272. Ridge Publishing, Blacksburg,VA, USA.
[18] Withers, P. J.,and Bhadeshia, H. K. D. H., 2001, "Residual [31] Navarro, J., 2008, "Improved Prediction of External
Stress Part 2 - Nature and origins," 17 ( 4), pp. 366-375. Pressure Collapse of Seamless Pipe," MSc Thesis,
[19] Wang, J., Saraswat, R., and Mirzaee-Sisan, A., 2013, Cranfield University, Cranfield, UK.
"Influence of Residual Stresses on Pipeline Integrity: A [32] Torselletti, E., Bruschi, R., Vitali, L., and Collberg, L.,
State-of-the-art Review," Rio Pipeline Conference and 2003, "Minimum Wall Thickness Requirements for Ultra
Exposition, Rio de Janeiro, Brazil. Deep-Water Pipelines," 22nd International Conference on
[20] Liessem, A., Grob-Weege, J., Knauf, G., and Offshore Mechanics and Arctic Engineering (OMAE
Zimmermann, S., 2007, "UOE Pipes for Ultra Deep Water 2003), Cancun, Mexico, p. no OMAE2003-37219.
Application — Analytical a nd FE Collapse Strength [33] Hines, J., Timms, C., and DeGeer, D., 2007, "Thermal
Prediction vs. Full-Scale Tests of Thermally Treated Line Ageing Effects of Thickwalled Line Pipe," Proceedings of
Pipe," Proceedings of the 7th International Society of the 26th International Conference on Offshore Mechanics
Offshore and Polar Engineers, Lisbon, Portugal. and Arctic Engineering (OMAE 2007), San Diego, CA,
[21] Huang, X., and Mihsein, M., 2000, "Finite element USA, p. no OMAE2007-29661.
prediction of the ultimate collapse strength of casings," [34] Zhou, Z., Nishida, A., and Kuwamura, A., 2011,
Proceedings of the Institution of Mechanical Engineers, "Applicability of Finite Element Method to Collapse
Part C: Journal of Mechanical Engineering Science 2000 Analysis of Steel Connection under Compression,"
214:1515, London, UK. Journal of Nuclear Science and Technology, 2, pp.
[22] Benjamin, A. C., and Cunha, D. J. S., 2012, "Assessment 481-485.
of Hydrostatic Collapse of Submarine Pipelines: The [35] Liu, G., and Quek, S., 2003, "The Finite Element
Classical Approach Revisted," Proceedings of the ASME Method: A Practical Course," Butterworth-Heinemann,
2012 31st International Conference on Ocean, Offshore Oxford UK, Chap 3, pp. 35-65.
and Artic Engineering , (OMAE2012),Rio de Janeiro,
[36] American Petroleum Institute, 2012, API Specification
Brazil, p. no OMAE2012-83833.
5L:Specification of Line Pipe, 45 ed., Washington: API
[23] Toscano, R. G., 2009, "Collapse and post-collapse Publishing Services.
behaviour of steel pipes under external pressure and
bending. Application to deep-water pipelines," PhD
Thesis, Universidad de Buenos Aires, Buenos Aires,
Argentina.
[24] Ramberg, W., and Osgood, W.,1943, "Description of
Stress-Strain Curves by Three Parameters," Technical note
no. 902, NASA, Washington, USA.
[25] American Petroleum Insitute, 2009, Design, Construction,
Operation and Maintenance of Offshore Hydrocabron
Pipelines (Limit State Design), 4 ed., Washington: API
Publishing Services.

9 Copyright © 2014 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/31/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

You might also like