You are on page 1of 10

Proceedings of the ASME 2022 14th International Pipeline Conference

IPC2022
September 26-30, 2022, Calgary, Alberta, Canada

IPC2022-83791

FINITE ELEMENT ANALYSIS OF COLD FIELD BENDS FOR SOUR SERVICE


APPLICATIONS

Ismael Ripoll Carlos Sicilia Kevin Williams


Xodus Group TotalEnergies Oil and Gas Verification
London, UK Paris, France London, UK

ABSTRACT the influence of the residual longitudinal and hoop stresses,


In onshore pipelines, cold field bends are regularly used for models are run with pipe, elbow and shell element models.
planned and in-situ pipeline route adjustments. The cold field Based on the results of these analyses, this paper
bending operation consists of curving permanently a straight recommends additional modelling and testing requirements for
pipe until the desired change of alignment is achieved. The bends cold field bends for more sensitive applications such as sour
are achieved by locally bending the pipe against a die using a service. These requirements intend to complement the approach
pipe bending machine. Bending against the die imposes a currently adopted by the industry and ensure the fitness for
transversal plastic deformation over a short distance, many such service of cold field bends.
local deformations are made to form the required cold field bend. Keywords: cold field bends, bending machine, finite element
Thus, cold field bends experience plastic strains from the analysis, ABAQUS, residual stress, plastic strain, ovalisation,
formation process and residual stresses are left in the finished sour service, fitness for service
product.
The onshore pipeline industry typically disregards residual NOMENCLATURE
stresses from cold bending as these are difficult to evaluate; ABAQUS Software for Finite Element Analysis
mainly, because the main pipeline stress software packages used ALA American Lifelines Alliance
by the industry do not capture them. This empirical approach has API American Petroleum Institute
provided a good record of pipeline service reliability for many AS/NZS Australian/ New Zeeland Standard
ASME American Society of Mechanical Engineers
years; although, international codes such as ASME B31.8
CONT Continuous
transfer the ultimate responsibility to determine whether such EAF Effective Axial Force
stresses should be evaluated to engineers (projects). FE Finite Element
For applications in sour service environments, where FEA Finite Element Analysis
additional plasticity during operation generally needs to be FEM Finite Element Model
avoided, the influence of these residual stresses may be FFS Fitness for Service
significant, and the robustness of the current practice should be HAZ Heat Affected Zone
considered. HIC Hydrogen Induced Cracking
In this context, this paper presents a comparative finite KDF Knockdown Factor
NACE National Association of Corrosion Engineers
element analysis (FEA) using ABAQUS to evaluate the impact of
NPS Nominal Pipe Size
the residual stresses from cold field bending on the overall stress OD Outer Diameter
state during operation. Analyses are performed using a PEEQ Equivalent Plastic Strain
representative single 14” x 19.05mm x 80D x 14.7˚ cold field RESID Residual
bend which is buried in a non-cohesive soil. To bound the range SEG Segmented
of local curvatures that the bend has to withstand during the SMYS Specified Minimum Yield Stress
formation process, models are run with a bend uniformly bent to SSC Sulphide Stress Cracking
52D which relaxes to 80D, and a bend with 24 short sections SWC Stepwise Cracking
bent to 17D (die radius) which relaxes to an average of 80D with TOP Top of Pipe
UK United Kingdom
a non-constant curvature along the bend length. To capture the
UR Utilization Ratio
impact of the type of element over the ovalisation and capture WT Wall Thickness

1 © 2022 by ASME
1. INTRODUCTION 10% lower than the ASME B31.8 allowable ratio. This was
In onshore pipelines, cold field bends are regularly used for selected to illustrate the impact of considering residual stresses
planned and in-situ pipeline route adjustments. The cold field in a scenario in which the allowable stress limits in the design
bending operation consists of curving permanently a straight code are maintained with some margin when the standard design
pipe by cold bending (i.e., ambient temperature) using a pipe approach (ignoring residual stresses) is adopted.
bending machine (FIGURE 1). Thus, cold field bends experience
residual stresses and plastic strains from the formation process. Parameter Value
The onshore pipeline industry typically disregards residual Linepipe Data
stresses from cold bending as these are difficult to evaluate; NPS 14 inches
mainly because the stress analysis software used by the industry OD 355.6 mm
does not capture them (AutoPIPE and Caesar II). This empirical WT 19.05 mm
approach has provided a good record of reliability in pipeline Internal Corrosion Allowance Not Considered
service for many years; although, international codes such as Fabrication Tolerances Not Considered
ASME B31.8 [1] transfer the ultimate responsibility to determine Material Data
whether such stresses should be evaluated to engineers Material Grade API 5L X65
(projects).
Elastic Modulus 207 GPa
In addition, the onshore pipeline industry does not have a
Poisson’s Ratio 0.3
consistent approach to address the impact of fluid sourness on
SMYS 450 MPa
the pipeline mechanical response (allowable stress limits);
Thermal Expansion Coefficient 1.17·10-5 °C
mainly because international codes such as ASME B31.8 [1] do
not provide guidance on how to address it. However, there are an Installation and Operating Data
increasing number of pipelines under design/operation in which Pipeline Installation Method Onshore, Trenched
additional requirements and limit states have been imposed as a Service Sour Service
result of specific testing programs. Indeed, testing on carbon Minimum Installation Temperature 10 °C
steel weld HAZs operating in severe sour service (FIGURE 4) Maximum Installation Temperature 80 °C
has shown propensity to SSC when tensile stress levels (residual Design Temperature 100 °C
or applied stresses) are above 80% of the SMYS, and fatigue Design Pressure 219 barg
KDF close to 60 [2]. In addition, the authors are aware of at least Shutdown Temperature -15 °C
one system of sour gas injection pipelines currently under Shutdown Pressure 0 barg
construction that has been designed to incorporate these findings. Contents Density 0 kg/m3
The objective of this paper is to present a comparative FEA Soil Data
performed to evaluate the impact of the residual stresses and Cover Depth to TOP 2m
strains due to bend formation on the overall stresses and strains Effective Unit Weight 15 kN/m3
during operation and shutdown for a representative, single cold Angle of Friction 35°
bend schedule. This paper then recommends the type of stress Cohesion 0 kPa
analysis (modelling) and material testing that should be adopted Coating Dependent Factor 0.8 (Rough Steel)
in onshore pipeline projects where cold field bends operate in TABLE 1: PIPELINE DATA
sour service. It is noted modelling recommendations could be
applied to cold field bends operating in sweet service if the 2.2. Cold Bend and Bending Machine Data
Fitness for Service (FFS) approach presented herein was TABLE 2 presents the cold field bend and bending machine
considered more appropriate than the empirical approach data used in the assessment of segmented bends. FIGURE 1
typically used by the onshore pipeline industry, in which residual presents an image of a typical bending machine used to cold bend
stresses from cold field bending are neglected. pipes between 6” and 20” diameter. FIGURE 2 and FIGURE 3
present 2D and 3D images of the typical bending machine
2. MATERIALS AND METHODS components. The inner diameter of the pin-up shoe and stiffback
match the total outer diameter of the pipe. It is noted, however,
2.1. Pipeline Data that the bending machine is not modelled in the FEM and the
TABLE 1 presents the pipeline data used in the comparative deformed shape of the cold bend is generated imposing plastic
FEA. This data is based on the authors’ experience and is not deformations (rotations) to match the geometry of the bending
taken from any specific reference. As base case, FEMs are run machine.
with elastic, perfectly plastic stress-strain curves, which are not An internal pipe mandrel is not used in this assessment;
temperature de-rated. Additional sensitivities to stress-strain although, mandrels are often used to optimize the cold bending
curves with hardening and temperature de-rating are then operation and avoid rippling of the pipe wall.
performed. Soil data is selected to obtain a peak equivalent stress
equal to about 80% of the pipe SMYS, when the contribution of
the residual stresses is not considered. This target ratio is about

2 © 2022 by ASME
Parameter Value
Cold Bend Data
Pipe Joint Length 12.2 m
Angle (Residual) 14.743°
80 x OD
Radius
28.448 m
Tangents 2.440 m
Arc Length 7.320 m
Bending Machine Data
Step Length 0.305 m
Step Angle (Residual) 0.614°
Number of Steps 24
17 x OD
Die Radius
6.05 m FIGURE 4: ISO 15156 SOUR DOMAINS [3]
Pin-up Shoe Length 0.337 m
Stiffback Length 3.061 m 2.3. Methodology
TABLE 2: COLD BEND AND BENDING MACHINE DATA This section presents the methodology used to evaluate the
impact of the residual stresses on the overall stresses during
operation and shutdown and to determine the level of local
strains expected in a typical design where ASME B31.8 [1] stress
limits are imposed.

ABAQUS Models
ABAQUS models are used to evaluate cold bend stresses as
the software commonly used by the industry to perform the stress
analysis (AutoPIPE and Caesar II) does not take them into
consideration. The implicit solver in ABAQUS 2019 is used to
perform the simulations, with geometric and material non-
linearity. The response of the steel is modelled using an elastic,
perfectly plastic stress-strain curve to enable the formation of the
FIGURE 1: 6”-20” BENDING MACHINE IMAGE (TYPICAL) cold bends.
The study is performed using global and local models of the
pipeline and cold bend. FIGURE 5 presents an overall view of
the ABAQUS global and local models. The geometry consists of
a single 12.2m long, 80D and 14.743° cold bend. In the global
models, the cold bend is in the centre with 500m tangents at
either side to ensure fully restrained conditions are achieved at
both ends.
The pipeline is modelled using PIPE31, ELBOW31B,
FIGURE 2: KEY BENDING MACHINE COMPONENTS – 2D ELBOW31 and S4 (shell) elements. These elements allow
VIEW (TYPICAL) consideration of the coupling between pressure and external
force and moment. ELBOW31 elements capture the variation of
ovalisation along the pipeline length, whereas ELBOW31B do
not capture it. The formulation of these elements is based on thin
wall theory as the ratio between pipe outer diameter and nominal
wall thickness is 18.7 (the design Effective Axial Force (EAF) is
only underestimated by 1.2%).
In PIPE31 and ELBOW31B models, element lengths are
constant and equal to the step length of the cold bend (0.305m).
In ELBOW31 and S4 models, element lengths are equal to one-
eighth of the step length. Around the cross-section, pipe and
elbow elements have 32 section points, whereas shell element
models have 32 shell elements. Through the wall thickness, pipe
FIGURE 3: KEY BENDING MACHINE COMPONENTS – 3D elements have 1 section point, whereas elbow and shell elements
VIEW (TYPICAL)
have 5 sections points.

3 © 2022 by ASME
In the global models, the interaction between buried pipeline o Models with segmented curvature along the cold
and surrounding soil is modelled using pipe-soil interaction bend length. In these models, 24 die curvatures are
elements, PSI34. The FEMs are contained in a horizontal plane imposed over one quarter of the step length
and thus only axial and lateral soil resistances contribute to the (0.305m) and no curvature is imposed over the
equilibrium. Axial and lateral soil resistances are calculated remainder of the step length. The average residual
using the equations recommended by ALA 2001 [4]. The build- curvature of the models is also 80D; although, it is
up of both resistances follows an elastic-perfectly plastic bilinear not constant. In each step, pipe is bent to 17D and
curve. then it relaxes to an average of 80D. In these models,
The EAF, shear force and bending moment at the ends of the the short sections of straight pipe at either side of the
cold bends (FIGURE 6) are extracted from the global models and bent pipe constrain the ovalisation of the bent
applied to the local models. The lateral soil resistance along the section.
cold bend is also extracted from the global model and applied to Curvatures are applied by imposing nodal rotations to the
the local model as follows: nodes of pipe and elbow element models. In the shell element
- In the pipe and elbow models, the lateral soil resistance models, curvatures are applied by imposing nodal rotations to the
is applied as a DLOAD perpendicularly to the element nodes located at the three o’clock position of the cross-sections,
longitudinal axis. which are tied to the remainder cross-sectional nodes using the
- In the shell models, the lateral soil resistance is applied FORTRAN subroutine that ensures that cross-sections remain
as an external pressure to the extrados (half cross- plane.
section).
- The contribution of the axial soil resistance along the Residual Stress from Cold Bending
cold bend is deemed to be negligible and thus is not Model Element Yes
considered. Type Type No Constant Variable
In the ELBOW31 models, a no warp boundary condition is Curvature Curvature
imposed at each end of the bent sections. In the S4 models, a PIPE 31 ✓ ✓ 
FORTRAN subroutine is used to ensure that the cross sections Global ELBOW 31B ✓ ✓ 
remain in a plane. ELBOW 31   ✓
The following models are run to evaluate the influence over PIPE 31 ✓  
Local
the pipeline mechanical response of element type and cold S4 ✓ ✓ ✓
bending residual stresses: TABLE 3: MATRIX OF SENSITIVITIES RUN
- Models with PIPE31, ELBOW31B, ELBOW31 and S4
elements to evaluate the influence of the cross-section Models are run considering the following two loadcases:
distortion due to ovalisation over the stresses during - LC1: From minimum installation temperature and no
operation. pressure to design temperature and pressure.
- Models with and without residual stresses to evaluate - LC2: From maximum installation temperature and no
the influence of the cold bending residual stresses over pressure to shutdown temperature and pressure.
the stresses during operation. In the models without Local models are only run with LC1. In the shell element
residual stresses, the initial geometry of the model models, longitudinal stresses around the elements that form
consists of a single bend in its centre and tangents at each cross-section are integrated to determine axial force
either side. In the models with residual stresses, the and bending moment and shear stresses to determine the
initial geometry of the model is straight, and the cold shear force.
bend shape is created by imposing and releasing a Longitudinal, hoop and equivalent (Mises) stresses are
curvature beyond the pipe elastic range. extracted, compared and checked for information against the
- Models with constant and segmented curvatures to allowable stresses presented in the following section.
evaluate the influence of the bend formation curvature
over the cross-section distortion and stresses during Allowable Stresses
operation. These two cases are expected to bound the This paper mostly compares the FEM results, but allowable
local curvatures which the cold bend will have to stress checks are included to compare against a standard design
withstand during the formation process as the initially and assess the onset of plasticity in operation. The following
straight sections will be bent against the radius of the ASME B31.8 [1] limit states are checked:
bending machine die by imposing a transversal - Hoop stress check: Hoop stresses are limited to 72% of
deformation from a relatively short distance: the pipe SMYS (324MPa) [location class 1, division 2
o Models with constant curvature along the cold bend which is intended to reflect sparsely populated areas].
length. The residual curvature of these models is - Longitudinal stress check: Longitudinal stresses are
constant and equal to 80D. Pipe is bent to 52D and limited to 90% of the pipe SMYS (405MPa).
then it relaxes to 80D. - Equivalent stress check: Von Mises stresses are limited
to 90% of the pipe SMYS (405MPa).

4 © 2022 by ASME
ELBOW
PIPE 31 ELBOW 31B
Peak Stress to 31
SMYS Ratio & NO
RESID,
NO
RESID, RESID,
PEEQ (1, 2, 3) CONT CONT SEG
RESID RESID
CURV CURV CURV
Long. LC1 0.52 0.73 0.54 0.79 1.00
Ratio LC2 0.63 0.98 0.64 1.00 1.12
Hoop LC1 0.41 0.41 0.42 0.52 0.86
Ratio LC2 0.00 0.00 0.01 0.12 0.50
Mises LC1 0.80 1.00 0.81 1.00 1.00
Ratio LC2 0.63 0.98 0.64 1.00 1.00
Bending N/A 0.65 N/A 0.70 3.11
0.65 0.70 3.11
PEEQ LC1 0.00 0.00
[0.05] [0.06] [0.05]
(%)
0.65 0.70 3.16
LC2 0.00 0.00
[0.00] [0.00] [0.06]
FIGURE 5: OVERVIEW OF GLOBAL AND LOCAL MODELS (1) Values in red denote stress URs equal to or larger than the ASME

B31.8 allowable stresses.


(2) Values in blue denote equivalent plastic strains that increase under

operating or shutdown conditions.


(3)
In brackets, additional plasticity during operation or shutdown.
TABLE 4: STRESS UR & PEEQ, MAX THROUGH THICKNESS

ELBOW
PIPE 31 ELBOW 31B
Peak Stress to 31
SMYS Ratio & NO
RESID,
NO
RESID, RESID,
PEEQ (1, 2, 3) CONT CONT SEG
RESID RESID
CURV CURV CURV
Long. LC1 0.51 0.79 0.95
Ratio LC2 0.62 0.92 1.12
Hoop LC1 Values not 0.42 0.40 0.86
Ratio LC2 available at inner 0.01 0.10 0.50
FIGURE 6: TANGENT AND END POINT LOCATIONS (LOCAL
MODEL) Mises LC1 surface (thin wall 0.81 1.00 1.00
Ratio LC2 pipe element 0.62 0.95 1.00
Bending considered, so all N/A 0.61 2.73
3. RESULTS AND DISCUSSION integration points 0.61 2.73
PEEQ LC1 at mid-thickness) 0.00
[0.05] [0.05]
3.1. Stresses under Design Conditions (%)
0.61 2.77
This section presents a summary of the results obtained LC2 0.00
[0.00] [0.06]
using the nine ABAQUS models presented in TABLE 3. Results (1) Values in red denote stress URs equal to or larger than the ASME

are organized as follows: B31.8 allowable stresses.


(2) Values in blue denote equivalent plastic strains that increase under
- TABLE 4 and TABLE 5 present a comparison between
the ratios of uniaxial and equivalent stress to SMYS and operating or shutdown conditions.
(3) In brackets, additional plasticity during operation or shutdown.
accumulated equivalent plastic strains. For information,
ratios are compared against the allowable ASME B31.8 TABLE 5: STRESS UR & PEEQ, INNER PIPE SURFACE
ratios.
- FIGURE 7 to FIGURE 9 present a comparison between From the results presented in this section, the following is
the typical stress fields along the length of the bend for obtained:
each of the three types of bends considered in the - The agreement between stresses and strains obtained
analysis: 80D with no residual stresses, bend section using global and local models is very good when the
uniformly bent to 52D which relaxes to 80D, bend boundary and loads applied to the local model are as
section with 24 short, segmented sections bent to 17D described in Section 2.3. Thus TABLE 4 and TABLE 5
which relaxes to an overall 80D. present results only from global models.
- FIGURE 10 and FIGURE 11 present comparative - Both pipe (PIPE31) and elbow (ELBOW31B and
contour plots among stresses and strains obtained when ELBOW31) element models capture the residual
the cold bending is imposed using the shell models with longitudinal stresses that arise from the cold bending
constant and segmented curvatures. process. In addition, elbow (ELBOW31) and shell (S4)
- FIGURE 12 and FIGURE 13 present the additional element models capture the residual hoop stresses that
equivalent plastic strain accumulated during operating arise from the restraint imposed by the unbent sections
(LC1) and shutdown (LC2) loadcases. at either side of the bends (TABLE 6).

5 © 2022 by ASME
Residual
Model Residual
Element Longitudinal
Description Hoop Stress
Stress
PIPE31 ✘ ✔
with Residual ELBOW31B ✘ ✔
Stresses ELBOW31 ✔ ✔
Shell S4 ✔ ✔
TABLE 6: CAPTURED RESIDUAL STRESSES BY ELEMENT
TYPE

- The influence of the residual stresses on the overall


stresses during operation is significant. When residual
stresses are included, local hoop stresses increase from
42% up to 86% SMYS, local longitudinal stresses from
64% up to 112% SMYS, and peak equivalent stresses
from 81% up to 100% SMYS (limited by the
FIGURE 7: GLOBAL MODEL – PIPE 31 – NO RESIDUAL
STRESSES – LC1 assumption of elastic-perfectly plastic material). It is
noted that this comparison is made using results from
the global models with elbow elements and an elastic
perfectly plastic stress strain curve without temperature
derating.

FIGURE 8: GLOBAL MODEL – PIPE 31 – RESIDUAL


STRESSES – CONSTANT CURVATURE – LC1 FIGURE 10: S4 MODEL – CONSTANT CURVATURE –
RESIDUAL STRESS AND STRAIN DURING COLD BENDING

FIGURE 9: GLOBAL MODEL – ELBOW 31 – RESIDUAL FIGURE 11: S4 MODEL – SEGMENTED CURVATURE –
STRESSES – SEGMENTED CURVATURE – LC1 RESIDUAL STRESS AND STRAIN DURING COLD BENDING

6 © 2022 by ASME
- Plasticity additional to that introduced during cold
bending is accumulated during operation and shutdown
when residual stresses are considered (0.06% in the
case presented), see FIGURE 12 to FIGURE 14.
In the case modelled with constant curvature, the
additional in-service plasticity is not accumulated at the
location of peak plasticity (intrados and extrados), but
near the neutral axis (see FIGURE 12 & FIGURE 14).
In the case modelled with segmented curvature, the
additional in-service plasticity is similarly accumulated
near the pipe neutral axis, but also at the bend intrados
during shutdown (a location of peak plasticity during
bend formation) (see FIGURE 13).
It should be noted that additional plasticity develops at FIGURE 14: ELBOW 31, CONTINUOUS CURVATURE,
the inner surface of the pipe, which is in contact with ADDITIONAL PEEQ DURING SHUTDWON (LC2) AT PIPE
the production fluid. This could introduce a risk of INNER SURFACE (%)
failure under some sour service conditions that is not
identified when the residual stresses from bend Additional sensitivities to the stress-strain curve
formation are not considered. (considering strain hardening) and temperature derating also
- Peak equivalent stresses under operation (LC1) occur show additional plasticity in operation (LC1) and shutdown
close to the neutral axis of the cross-section (lower (LC2) at the inner wall surface of the bends. The introduction of
half), whereas peak equivalent stresses under shutdown strain hardening (Ramberg Osgood stress-strain curve) has a
(LC2) close to the neutral axis of the cross-section negligible effect on the amount of additional plasticity with
(upper half) and at the bend intrados (see FIGURE 13). regards to the base case analysis with elastic perfectly plastic
conditions. Temperature derating increases very slightly the
amount of in-service plasticity (0.07% vs 0.06%).

3.2. Residual Stresses from Cold Bending


This section presents a more detailed summary of the
residual stresses induced from cold bending, as these have been
shown to be significant when results of the models are compared.
Three bend types were considered in the analyses:
- Type 1: 80D bend formed with no residual stresses.
- Type 2: Bend section uniformly bent to 52D, which
relaxes to 80D (FIGURE 10).
- Type 3: Segmented bend, formed with short sections
bent to 17D, which relax to an average 80D bend but
FIGURE 12: ELBOW 31, SEGMENTED & CONTINUOUS with local variation (FIGURE 11).
CURVATURE, ADDITIONAL PEEQ DURING OPERATION (LC1) The following plots compare the model behaviour during
AT PIPE INNER SURFACE (%) forming (Time Step 0-1), relaxation (Time Step 1-2) and
application of operating loads (LC1) (Time Step 2-3). The
notation for the plots is shown in FIGURE 15.

FIGURE 13: ELBOW 31, SEGMENTED CURVATURE, FIGURE 15: NOTATION FOR RESULT EXTRACTION
ADDITIONAL PEEQ DURING SHUTDWON (LC2) AT PIPE
INNER SURFACE (%)

7 © 2022 by ASME
FIGURE 16 presents the mid wall longitudinal strain during could affect susceptibility to SSC. Identical hoop behaviour is
bending, relaxation and operation. As expected, the segmented also seen in the elbow model (ELBOW 31 elements).
bend sees a significantly larger strain.

FIGURE 16: COMPARISON OF LONGITUDINAL STRAIN, S4


MODEL, TOP: TYPE 2 & BOTTOM: TYPE 3

FIGURE 17 presents the associated diameter change during


bending. The segmented bend ovalises less than the uniform
bend since the undeformed sections restrain ovalisation. This
effect also occurs at the tangents of the uniform bend.

FIGURE 18: COMPARISON OF STRESSES, S4, TYPE 2

FIGURE 17: COMPARISON OF DIAMETER AND OVALITY, S4


MODEL, TOP: TYPE 2 & BOTTOM: TYPE 3

FIGURE 18 and FIGURE 19 present the axial and hoop


stress history for the uniform and segmented bend S4 models.
The uniform bend has a hoop stress that is expected given the
pressure applied (about 180MPa). However, the segmented bend
has a very different response at the bend intrados and extrados.
The constrained ovalisation generates a high residual tensile and
compressive hoop stress (see FIGURE 20), which induces a
tensile hoop stress at both mid and inner wall during operation
(shell element model). This is a significant increase in tensile FIGURE 19: COMPARISON OF STRESSES, S4, TYPE 3
stress on the inner surface of the pipe (about 320MPa), which

8 © 2022 by ASME
As shown above, whether the cold bend is deformed as type 4. RESULTS SUMMARY
3 using steps in which the radius of curvature of the bending A series of finite element analyses of a typical cold bend
machine die are imposed locally or as type 2 with a constant have been presented to assess the residual stresses resulting from
curvature applied along the entire length influences the residual bend formation. These analyses have been performed for a cold
stresses and therefore the stresses in operation. It is therefore bend under loading conditions that, when residual stresses are
important that the more realistic segmented modelling is not considered, are within the requirements of ASME B31.8 with
considered. margin (80% vs. 90% SMYS allowable equivalent stresses).
The residual longitudinal, hoop and equivalent stresses from From the results of the analyses, the following is found:
cold bend formation in the model ELBOW31 elements with - Maximum equivalent stresses in operation increase from
segmented curvature are illustrated in FIGURE 20. In order to 80% to 100% of SMYS when residual stresses from cold
capture accurately the residual hoop stresses in the figure below, bend formation are considered.
the segmented bend formation needs to be considered. - As a result of this, plasticity additional to that introduced
during cold bending is accumulated during operation and
shutdown. In the case presented in this paper, the additional
plastic strain is a modest 0.06%. Indeed, if the cold bend
was designed to satisfy the allowable equivalent stress of
ASME B31.8 of 90% SMYS (ignoring residual stresses),
the additional plastic strain in operation (considering
residual stresses) would be expected to be higher.
As shown in FIGURE 12 and FIGURE 13, this additional
plasticity takes place at the inner pipe surface, which is in
contact with the production fluid. Under some sour service
conditions, this could introduce a risk of failure that would
not be identified if the residual stresses from bend
formation were not considered (conventional design).
- When the modelling considers cold bending with uniform
curvature, the additional plasticity is limited to a very
localised area near the pipe neutral axis (see FIGURES 12
& 14). However, when the modelling considers the more
realistic segmented curvature, a wider area of additional
plasticity is observed at the pipe bend intrados (see
FIGURE 13). This is the result of the residual hoop stresses
generated by the bend formation with segmented curvature.

5. CONCLUSIONS
The results presented indicate that modelling of residual
stresses from cold bend formation can lead to local stresses in
excess of the code allowable limits in scenarios in which the
conventional design approach (ignoring residual stresses) shows
code compliance. Successful practice however suggests that
despite these limitations, the conventional design approach
should be acceptable for most pipelines. However, for critical
designs, the validity of the conventional approach should be
confirmed.
In the case of pipelines in sour service, and in particular
severe sour service, the effect of residual stresses from bend
formation and associated increase of plasticity in operation
should be assessed to evaluate the risk of SSC in operation.
In order to address these challenges, the following is
recommended for the analysis of critical pipeline designs:
FIGURE 20: ELBOW31 SEGMENTED CURVATURE MODEL - - Complement or replace stress analysis software commonly
RESIDUAL STRESSES AFTER COLD BENDING, used by the onshore pipeline industry (AutoPIPE and
TOP: LONGITUDINAL, MID: HOOP, BOTTOM: EQUIVALENT Caesar II) by a general-purpose FE package capable of non-
linear modelling to capture cold bending residual stresses.
- Model the segmental formation of cold bends instead of the
simplified continuous bending.

9 © 2022 by ASME
- Model cold bends using elements such as ABAQUS - Full scale testing of cold bend after ageing (captures points
ELBOW31, to capture the variation of ovalisation along the 1 & 2 above)
pipe length and thus capture residual longitudinal and hoop - Samples taken from cold bend extrados and intrados and
stresses when the segmented cold bending process is aged (captures point 1 above, but not point 2)
simulated. In addition, impose a NO WARP boundary - Samples taken from parent pipe, strained axially to match
condition at each end of the bent sections to ensure they conditions in cold bend intrados and extrados and aged
remain plane. (expected to capture point 1 above, but not point 2)
- If local models with shell or brick elements are used in - Samples taken from parent pipe, strained axially to match
order to evaluate a particular detail of the cold bend conditions in cold bend intrados and extrados and aged. The
mechanical response, apply end forces and moments and load applied to these samples during SSC testing will be
distributed soil reactions along the local model length. increased by the maximum residual stress due to cold
It is proposed that these recommendations are followed at bending (expected to capture point 1 above and point 2 –
least until the conventional or another simplified approach is probably too conservatively)
shown to be safe. - Samples taken from parent pipe, strained in bending to
In the next section, a testing program intended to support the match conditions in cold bend intrados and extrados,
development of simplified approaches is discussed. straightened by reverse bending and aged (potentially
captures points 1 and 2 above)
6. SOUR SERVICE COLD BENDS MATERIAL TESTING - Samples taken from parent pipe, strained in bending to
ISO 15156 [3] specifies requirements for use of carbon steel match conditions in cold bend intrados and extrados
in sour service conditions but has no specific guidance or (without straightening by reverse bending) and aged
requirements for use of cold field bends in sour service. Pipeline (potentially captures points 1 and 2 above)
design codes [1, 6 or 7] contain requirements for cold field bends From the comparisons of the results of these tests a
but focus on avoidance of wrinkling and provide no guidance on simplified testing approach could be proposed. Until such
qualification of cold field bends for use in sour service. comparative testing is complete, it is expected that cold bends
The closest code identified to deal with cold field bends in would need to be produced and tested to demonstrate
sour service is ISO 15590-4 [5], which addresses factory cold acceptability of the residual strains and stresses under sour
bends and contains in Annex B additional provisions required for service.
cold bends ordered for sour service, including HIC/SWC, SSC
and hardness testing. Compliance with these provisions requires REFERENCES
cold bends to be available before the construction activities [1] ASME B31.8, Gas Transmission and Distribution Piping
commence (as samples need to be taken from the cold bend), Systems, January 2020
which is not normal for conventional onshore projects. [2] NACE International, Paper No. 11105, SSC Susceptibility of
Moreover, the scope of ISO 15590-4 specifically states that the Local Hard Areas in Carbon Steel Weld Heat Affected Zones
document is not applicable to field cold bends. [3] ISO 15156:2020, Petroleum and Natural Gas Industries –
The response of cold field bends to loading under sour Materials for Use in H2S Environments in Oil & Gas Production.
service conditions and resistance to SSC can be influenced by [4] ALA 2001, Guideline for the Design of Buried Steel Pipe
two aspects in addition to those relevant for standard linepipe: [5] ISO 15590-4:2019, Petroleum and Natural Gas Industries -
- (1) Strained and aged condition of material, due to plastic Factory Bends, Fittings and Flanges for Pipeline Transportation
deformation imposed during formation of the cold bend Systems Part 4: Factory Cold Bends
- (2) Residual stresses in the longitudinal and hoop directions [6] ISO 13623:2017, Petroleum and Natural Gas Industries -
(also due to the formation of the cold bend), that have an Pipeline Transportation Systems
influence on the stress state and potential plastic straining [7] AS/NZS 2885.1:2018, Pipelines – Gas and Liquid Petroleum
in operation (whilst exposed to sour product), illustrated in
Section 3.1.
Both effects can be addressed by full scale testing of cold
field bends, but this is both impractical (cold bends needed at
early stage) and costly (large testing facilities required).
In order to seek an alternative to full scale testing, a test
programme could be developed to assess the validity of
qualification of cold field bends for sour service by small scale
testing of specimens extracted from pipe joints (before any actual
cold bending).
In order to develop this simplified approach, SSC tests on
the following types of specimens are proposed. Loads to apply
to these samples during SSC testing to be calibrated by FE.

10 © 2022 by ASME

You might also like