You are on page 1of 10

Proceedings of the 2018 12th International Pipeline Conference

IPC2018
September 24-28, 2018, Calgary, Alberta, Canada

IPC2018-78293

ASSESSMENT OF PIPELINE SPIRAL WELD CRACKS SUBJECTED TO


INTERNAL PRESSURE

Mark C. Neuert Thomas J. Dessein Millan Sen


C-FER Technologies (1999) Inc. C-FER Technologies (1999) Inc. Enbridge Pipelines Inc.
Edmonton, Alberta, Canada Edmonton, Alberta, Canada Edmonton, Alberta, Canada

ABSTRACT these manufacturing anomalies may include porosity, undercut,


Spirally welded pipelines can make up significant portions hot cracking, or lack of fusion.
of operator transmission systems, and may contain
manufacturing anomalies that are susceptible to fatigue growth.
Modifications to inputs of crack assessment models, such as
CorLAS®, are required to account for the angle these cracks
make with respect to the longitudinal pipe axis, given that these
crack assessment models were developed for longitudinally
orientated cracks. Two such modifications were investigated and
are discussed in this paper.
One approach considered the normal stress component
perpendicular to the angled crack, for which a stress
transformation calculator was developed. Another approach,
adapted from API 579 and BS7910 standards, used an effective
crack length calculated as the longitudinal projection of the full
length of an angled crack.
Failure pressures calculated using these approaches were
compared to validated finite element (FE) results. For both FIGURE 1. EXAMPLE OF SPIRAL PIPE MANUFACTURING
modifications, the pressure capacity increased for angled cracks
versus longitudinal cracks. The transformed normal stress The CorLAS® crack model is often used to assess the threat
approach resulted in non-conservative failure pressure of cracks and manufacturing anomalies on pipeline systems. It
predictions with respect to the FE models, whereas the modified has been previously validated for cracks oriented along the
crack length approach was conservative. longitudinal pipe axis, but not for spirally oriented cracks.
Additionally, the extended finite element method (XFEM) Moreover, current crack detection tools cannot reliably detect
was used to investigate the propagation behavior of angled these angled cracks. However, these limitations may be
cracks. It was found that the general tendency was for mitigated by the reduction in crack opening stresses from
propagation parallel to the longitudinal pipe axis; however, when pressure loads in comparison to those for longitudinal cracks due
considering weld residual stresses, the crack propagation would to the relative angle between the cracks with respect to the hoop
be directed toward the direction of the spiral seam. direction.
Past investigations [1] into flaws in the vicinity of spiral
INTRODUCTION welds have sought to calculate their burst pressure capacity by
Spiral welded pipe is used extensively in liquids calculating an equivalent longitudinal crack length of an angled
transmission pipeline systems, especially on more modern lines. flaw using the procedure outlined in BS7910, in addition to other
It is manufactured from hot rolled steel strips, as shown in Fig. crack transformation procedures produced for corrosion defects
1, that are welded together using the double-sided submerge arc [2, 3]. These investigations found that using the equivalent
welding process. Similar to longitudinal seam welds, spiral longitudinal crack lengths in a failure assessment diagram (FAD)
welded pipe may exhibit manufacturing anomalies in either the procedure, as outlined in BS7910, yielded conservative
base metal heat affected zone or within the weld itself. Some of

1 Copyright © 2018 ASME

Downloaded From: https://proceedings.asmedigitalcollection.asme.org on 12/11/2018 Terms of Use: http://www.asme.org/about-asme/terms-of-use


predictions for burst pressure in comparison with experimental in this study to calculate the modified toughness-based critical
tests; however, no such comparison has been made for CorLAS. stress values, σtough, for cracks angled with respect to the
This study was undertaken to explore modifications to longitudinal pipe axis, but may also be used by pipeline operators
CorLAS model inputs to allow for the model to be used for the in determining the stress state at cracks on spiral welded
evaluation of angled cracks in spiral pipe. One approach pipelines in their system.
modified the nominal stress, using the normal stress component The tool uses the transformation equations that define
perpendicular to the angled crack instead of the full hoop stress. Mohr’s Circle to transform normal and shear stresses between
The second approach calculated an equivalent longitudinal crack rotated coordinate systems (Fig. 2). Hoop stresses are calculated
length by projecting an angled crack onto the longitudinal pipe from internal pressure using Barlow’s formula considering
axis. The change in failure pressure with increasing spiral angle diameter, wall thickness, and operating pressure. If the pipe is
was compared to validated finite element (FE) models. assumed longitudinally constrained (as was the case in this
Furthermore, the propagation behavior of angled cracks in study; see next section for discussion of this assumption),
the vicinity of the seam weld on spiral pipe was investigated longitudinal stresses are calculated from the resistance to
using the extended FE method (XFEM). The results may be used longitudinal strains induced by thermal expansion and the
to support mitigation of crack failure on spirally welded Poisson effect. Longitudinal and shear stress values from other
pipelines. sources can also be included. A sample of the stress values
produced by the calculator, given various pipeline attributes, is
THE CORLAS MODEL provided in Table 1.
CorLAS is a widely used pipeline assessment model
developed to determine the critical pressure and/or crack size TABLE 1. SAMPLE OF STRESS VALUES PRODUCED BY
that will result in the failure of pipeline cracks [4]. It is typically THE STRESS CALCULATOR FOR A GIVEN SET OF
used to evaluate longitudinal cracks on pipeline systems. Given PIPELINE PARAMETERS
a particular set of crack, pipe, and operating parameters, CorLAS
will determine the corresponding burst pressure, P , as: Diameter (mm) 914.4
Wall Thickness (mm) 8.9
2
(1) Grade (MPa) 483
Weld Angle (deg) 40
where t is pipe wall thickness, D is outer diameter, and FRS is the Internal Pressure (MPa) 9.4
remaining strength factor calculated from the remaining material
area at the crack location. The critical stress, σcrit, is the minimum Installation Temp (°C) 20
of the material’s flow strength (σf) and the toughness-based Operating Temp (°C) 30
critical stress (σtough): Stress normal to weld
(all sources - MPa) 328
min , (2) Stress normal to weld
(pressurization - MPa) 279
The flow stress is taken to be the average of the material’s Stress normal to weld
yield and ultimate strength, while the toughness-based critical (axial loads - MPa) 49
stress criterion depends on the value of the J-integral.
The J-integral is an established parameter in fracture The calculator also plots stresses normal to the angled crack
mechanics used for assessing the potential for a crack to face, along with the relative contributions from longitudinal and
propagate. It is similar to the concept of stress in that failure hoop stresses (Fig. 2).
occurs once the J-integral reaches a critical value, Jc. Unlike
stress analysis, however, Jc depends on crack loading and NON-PROPAGATING CRACK ANALYSIS
geometry in addition to the material itself. For a longitudinally- Longitudinal Cracks
oriented crack in the CorLAS model, σtough is the stress at which A validation procedure was performed by comparing FE and
the J-integral reaches Jc. Note that, through Eq. (1), the J-integral CorLAS predicted failure pressures under typical pipeline
can be expressed in terms of pressure, which is more relevant to operating loads for longitudinally orientated cracks.
a pipeline operator than σtough. FE models of 36 inch by 0.35 inch NPS X70 pipe were
created using the FE software Abaqus® 2016. Eight-node brick
(3) elements (C3D8R) were utilized, along with a layer of six-node
wedge elements (C3D6) to model the high stress and strain
gradients surrounding the crack tip at a longitudinal weld. Figure
STRESS TRANSFORMATION CALCULATOR 3 shows the mesh refinement around the crack tip (element size
A calculator was developed to determine the state of stress of ~0.075 mm) to ensure accuracy and mesh-independence of the
for a coordinate system rotated at an arbitrary angle with respect FE-computed J-integral value where high stress variation occurs.
to the longitudinal/hoop pipe directions. The calculator was used

2 Copyright © 2018 ASME

Downloaded From: https://proceedings.asmedigitalcollection.asme.org on 12/11/2018 Terms of Use: http://www.asme.org/about-asme/terms-of-use


corresponding to the predicted failure pressure. This is shown in
Fig. 4, where PCorLAS and PFEA correspond to the failure pressures
predicted by CorLAS and the FEA, respectively.
Elastic-plastic material properties were modeled using a
power law relation and a Charpy V-notch number (CVN) of
100 J was prescribed, corresponding to a Jc of 1250 N/mm. Both
the material plasticity definition and CVN-toughness
relationship were taken from the CorLAS model [4].
Both FEA model and CorLAS longitudinal crack failure
pressures were calculated for rectangular crack depths between
50% and 75% through-wall, with checks performed to ensure
(a) Conceptualization of stress transformation, where τ is the shear mesh independence of the FE results. The crack lengths were
stress parallel to an angled crack and σw is the normal stress
chosen such that the failure pressure, as calculated by CorLAS,
perpendicular to an angled crack, with contributions from hoop
(σwP) and longitudinal (σwA) components.
corresponded to a hoop stress of 70% and 80% of specified
minimum yield strength (SMYS). Table 2 displays the cases
assessed, along with the corresponding FE-predicted failure
pressures; these results are further illustrated in Fig. 5.

(a) Detail View of FE Model Mesh For Crack Tip

(b) Graphical output from the calculation tool showing the normal
stress (σw) acting perpendicular to an axis rotated 40° with respect
to the longitudinal pipe axis, broken down by contributions from
hoop (σwP) and longitudinal (σwA)stresses. 
FIGURE 2. CONCEPTUALIZATION AND GRAPHICAL
OUTPUT OF THE STRESS TRANSFORMATION
CALCULATOR

Constraints were placed at the model longitudinal and


circumferential faces (illustrated in Fig. 3 for a spiral model,
though the same boundary conditions were employed for
longitudinal crack models). Longitudinal nodal movement was
set to zero, as it was assumed that fixed longitudinal boundary
conditions predominate over the total pipeline length due to the (b) Spiral Crack Model Geometry Showing Faces Fixed for FE
fact that virtual anchors caused by pipe/soil friction can develop Boundary Conditions
over a few hundred meters (a relatively short distance when FIGURE 3. ISOMETRIC VIEW OF MESH (A) AND SPIRAL
compared to pipelines many kilometers long). Circumferential PIPE GEOMETRY (B) FOR THE CRACK FE ANALYSIS
nodal movement was also set to zero as it was assumed the pipe
would not experience any torsion. These were relaxed on the
interior of the model to allow the crack to open freely. Radial
nodal strain was allowed everywhere throughout the model.
The J-integral demand for a given pressure was calculated
using both the CorLAS crack model and FE analysis (FEA), with
the pressure at which the J-integral demand reached Jc

3 Copyright © 2018 ASME

Downloaded From: https://proceedings.asmedigitalcollection.asme.org on 12/11/2018 Terms of Use: http://www.asme.org/about-asme/terms-of-use


models of pipe bodies containing cracks running parallel to spiral
welds were created. The angle of the spiral weld, with respect to
the longitudinal axis of operating pipelines, was determined to
range between approximately 40° and 50° based on magnetic
flux leakage (MFL) image data and manufacturing procedure
specifications. Thus, spiral angles of 40°, 45°, and 50° were
selected for analysis. The same 36 inch by 0.35 inch pipe
dimensions were used as in the longitudinal analysis.
For a longitudinally-oriented crack, hoop stresses acting
normal to the crack are the only stresses considered in the
CorLAS model (i.e. Mode I cracking: opening). This is not the
case with spiral cracks, and so several input modifications to the
CorLAS model were explored.
It has been reported that the application of internal pressure
FIGURE 4. J-INTEGRAL AS A FUNCTION OF PRESSURE in seam cracks on spiral welded pipe greatly increases the
Mode I crack driving force (i.e. stresses perpendicular to the
The percent difference between CorLAS and the FE crack opening) in comparison to Modes II and III, in-plane and
longitudinal crack models ranged between 6% and 15%. This out-of-plane shear, respectively [5] (Fig. 6). Thus, the first
error tended to increase as the crack depth approached 75% method for modifying the inputs to the CorLAS model (referred
through-wall thickness. The FEA, therefore, provided a to herein as the “resolved stress method”) calculated burst
reasonable estimate of crack burst pressure. pressure by calculating the stresses acting perpendicular to an
angled crack, σr (see Fig. 7) and using this value when
TABLE 2. COMPARISON OF FAILURE PRESSURES FOR calculating σtough for Eq. (2).
LONGITUDINAL CRACKS PREDICTED BY CORLAS AND
THE FEA
Crack Crack PCorLAS PFEA % Diff.
Depth Length (%SMYS) (%SMYS) (PCorLAS and
(%WT) (mm) PFEA)
50% 379 80% 85% 6%
60% 186 80% 89% 11%
70% 124 80% 92% 15%
75% 104 80% 91% 14%
60% 381 70% 77% 11%
70% 190 70% 79% 12%
75% 151 70% 79% 13% FIGURE 6. ILLUSTRATION OF THE THREE FRACUTRE
MODES1

FIGURE 7. ILLUSTRATION OF THE “RESOLVED STRESS


FIGURE 5. COMPARISON OF FAILURE PRESSURES METHOD”, WHERE THE NORMAL STRESS, σR, RESOLVED
PREDICTED BY CORLAS AND THE FEA CRACK MODEL PERPENDICULAR TO THE CRACK IS USED IN
CALCULATING σTOUGH INSTEAD OF THE HOOP STRESS σH
Angled Cracks
In order to determine the effect of crack angle on the Alternatively, both API 579 [6] and BS7910 [7] specify
difference of the FE and CorLAS predicted failure pressures, FE procedures in which an effective longitudinal crack length and

1
Figure obtained from the Wikipedia article on fracture mechanics, used under the
public domain

4 Copyright © 2018 ASME

Downloaded From: https://proceedings.asmedigitalcollection.asme.org on 12/11/2018 Terms of Use: http://www.asme.org/about-asme/terms-of-use


the full hoop stress are applied (referred to herein as the Of particular note is the near identical results of the BS7910
“projected crack length method”) for crack integrity and API 579 length modification procedures, indicating that
calculations. For BS7910, this modification is a simple either method could be adopted when implementing the
projection of the angled crack onto the plane perpendicular to a projected crack length method.
principal stress (i.e. the longitudinal direction, see Fig. 8), Discussion of Results
whereas API 579 also incorporates the ratio of the values of For the longitudinal flaw cases, it was shown that CorLAS
principal stresses. These concepts were adopted in this study, predicted a lower failure pressure than the corresponding FE
where CorLAS was used to calculate the failure pressure of models, which is an expected outcome for an empirically
cracks using a modified crack length value, as defined in BS7910 calibrated equation and consistent with the results of studies used
and API 579, and applying the full hoop stress loading in in the development of CorLAS itself [4]. This result was shown
calculating the toughness-based failure stress in Eq. (2). to be consistent across varying crack depths and length
dimensions. Of note is the fact that the FE and CorLAS predicted
pressures diverged as crack depth approached 80% of the wall
thickness, which is similar to the findings reported by Jaske and
Beavers [4] when they compared CorLAS model results to the
FE results generated by Yagawa et al. [8]. This also corresponds
to the depth at which the API 579 specification recommends
converting a surface crack to a full through-wall crack during the
Failure Analysis Diagram (FAD) procedure to maintain accuracy
due to the high strain and plasticity effects.
Angling the cracks had the effect of increasing the pressure
capacity of a pipe in comparison to a longitudinal crack of the
same dimensions. This is consistent with what would be
expected for the resolved stress method, as the stress normal to
FIGURE 8. ILLUSTRATION OF THE “EFFECTIVE LENGTH” an angled crack is typically less than the full hoop stress, which
APPROACH, WHERE THE FULL HOOP STRESS AND THE
would thus increase the pressure capacity of the pipe at the crack.
CRACK LENGTH PROJECTED ON THE LONGITUDINAL
AXIS lP IS USED IN PLACE OF THE FULL CRACK LENGTH l On the other hand, the FE models predicted a lower failure
WHEN CALCULATING σTOUGH. pressure than the resolved stress method. This indicates that,
despite the decreased Mode I crack driving force (opening)
Figure 9 compares the failure pressures, as calculated from incurred from pressurizing the pipe, shear stresses acting on the
the FEA and using the above described modifications to crack crack front in the case of an angled crack (Mode II), which are
length and σtough, for X70 and X60 pipe in CorLAS. Spiral angles not accounted for in the CorLAS modification, but are accounted
of 0° (identical to the validation case), 40°, 45°, and 50° for by the FEA, add significantly to the crack-driving force.
measured from the longitudinal pipe axis were used, and the The projected length method, as outlined in BS7910 and
crack depth was 75% through-wall. In all cases, the failure API 579, assumes a shorter crack length that is subject to the full
pressure increased with the crack angle in comparison to the hoop stress. The trend of increasing failure pressure with an
longitudinal case. increase in crack angle was maintained, as would be expected for
The resolved stress method (blue line) leads to dramatic a shorter crack. In this case, however, the results from the
increases in predicted burst pressure that are non-conservative CorLAS model with these modified inputs was more
with respect to the lower FE predicted burst pressures (green conservative than the FE results. This is consistent with Chen et
line). In the case of the projected crack length method (BS7910 al. [1], which showed conservative burst pressure predictions
and API 579 shown as red and purple lines, respectively), the when using the BS7910 modification, as opposed to alternative
pressure capacity still increases with increasing crack angle, but angled crack modifications developed for corrosion defects. This
remains conservative with respect to the FE predictions, which shows that the pressure capacity of a pipe is much more sensitive
are relatively higher. to a reduction in stresses acting normal to the pipe than it is to a
The above analyses were also performed for a crack at a change in crack length.
spiral angle of 50° with a depth of 60% through-wall and length Thus, in the CorLAS model, the more conservative option
sized such that failure for a longitudinal crack would occur at when assessing angled cracks in spiral welded pipe is to use the
80% SMYS. As shown in Fig. 10, the relative difference in longitudinally projected length as the crack length parameter,
predicted failure pressures when using FE versus the projected combined with the full applied hoop stress, to calculate the
length and resolved stress methods is consistent with the 75% toughness-based critical stress (projected length method), as
through-wall cases. opposed to using only the stresses acting perpendicular to the
angled crack with the full angled crack length (resolved stress
method).

5 Copyright © 2018 ASME

Downloaded From: https://proceedings.asmedigitalcollection.asme.org on 12/11/2018 Terms of Use: http://www.asme.org/about-asme/terms-of-use


(a) X70 (a) X70

(b) X60 (b) X60


FIGURE 9. PREDICTED FAILURE PRESSURE FOR A 75% FIGURE 10. PREDICTED FAILURE PRESSURE FOR A
THROUGH-WALL CRACK IN 36 INCH × 0.35 INCH (A) X70 CRACK ANGLED AT 50° IN 36 INCH × 0.35 INCH FOR (A)
AND (B) X60 PIPE X70 AND (B) X60 PIPE

6 Copyright © 2018 ASME

Downloaded From: https://proceedings.asmedigitalcollection.asme.org on 12/11/2018 Terms of Use: http://www.asme.org/about-asme/terms-of-use


PROPAGATING CRACK ANALYSIS maximum cohesive traction Tmax, is the maximum stress capacity
An FE model was also developed to predict spiral crack of an element before damage due to cracking is initiated. Once
propagation behavior that incorporated factors that would the principal stress in an element reaches Tmax, phantom nodes
potentially influence the direction of crack growth. These are introduced and crack opening across that element is enabled.
included weld geometry, material properties of the pipe body, The crack opening distance, δ, increases until it reaches δmax, at
heat-affected zone (HAZ), weld, and residual stresses in the which point the stress carrying capacity of that element is
vicinity of the pipe weld. The results from this evaluation would reduced to zero. The area under the stress-separation curve
provide guidance on whether a spirally orientated crack under represents the energy released during the crack propagation
typical operating stresses would propagate in the longitudinal or process, corresponding to the fracture toughness of a material,
off-axis direction. The direction of growth could potentially Jc. By specifying Tmax and Jc, the entire damage evolution process
affect the in-line inspection (ILI) detection probabilities and is defined. The relationship between Tmax, and δmax is illustrated
other related management techniques. in Fig. 12.

Pipe Construction and Geometry


Pipe dimensions were kept consistent with the
non-propagating crack models (i.e. 36 inch diameter pipe with
0.35 inch wall thickness). A spiral angle of 50° with respect to
the longitudinal pipe axis was used. The boundary conditions at
the model’s longitudinal and circumferential faces were also
identical to those of the non-propagating crack models.
A 60° bevel at the edge of the steel skelp was modeled as
representative of the manufacturing procedure specification and
a 3 mm wide HAZ adjacent to the weld was included.
The crack itself was placed at the mid-point between the
HAZ/pipe body and HAZ/weld boundaries. Crack depth and
width were taken from one of the validation cases, with a depth
of 75% through-wall and a length corresponding to a failure
pressure of 80% SMYS, as predicted by CorLAS for X70 steel.
Details of the weld geometry and crack placement are illustrated
in Fig. 11.
FIGURE 12. COHESIVE TRACTION (I.E. STRESS) AND
CRACK OPENING DISPLACEMENT BEHAVIOR

Three material zones were defined for each of the pipe body,
HAZ, and weld material, with specified values of Tmax and Jc as
listed in Table 3. The value of Tmax was set at 555 MPa, which
corresponds to the yield stress plus 70 MPa (10 ksi). The
strength reduction in the HAZ due to heating effects and
overmatch of the weld material were modeled by decreasing Tmax
by 10% for the HAZ and increasing Tmax by 10% for the weld.
A fracture toughness corresponding to 100 J CVN was assigned
FIGURE 11 WELD GEOMETRY AND CRACK PLACEMENT
to the pipe body, with a fracture toughness corresponding to 50
J CVN for the weld and HAZ. The relationship between CVN
MATERIAL PARAMETERS
and fracture toughness was taken from the CorLAS model [4].
Most crack-growth modeling using the FE method forces
crack propagation to occur at element edges, requiring the user TABLE 3. MATERIAL PARAMETERS USED IN THE FE
CRACK PROPAGATION MODEL
to know the crack propagation path beforehand. Abaqus’ XFEM
capability, however, allows for crack growth across mesh Material Tmax Jc
elements, with the direction of crack growth determined by Zone (MPa) CVN (J) (N/mm)
model outputs, such as stress or strain [9]. This allows the model Weld 611 50 625
to predict the crack propagation direction for a wide set of pipe HAZ 500 50 625
characteristics and operating parameters. Pipe Body 555 100 1250
Surface-based cohesive material behavior was used to
model the progressive weakening of material during the cracking
process, which is governed by several parameters. The

7 Copyright © 2018 ASME

Downloaded From: https://proceedings.asmedigitalcollection.asme.org on 12/11/2018 Terms of Use: http://www.asme.org/about-asme/terms-of-use


Residual Stresses longitudinally along the pipe axis and along the weld,
The FE model uses the principal stress state at the crack tip respectively. Unlike Cases 1 and 2, the model was terminated
to determine whether or not a crack Tmax has been reached and if before the crack was able to deepen due to convergence issues
the crack will propagate, as well as the direction of propagation. encountered in this highly non-linear system.
Therefore, the residual stresses in the pipe body and HAZ
resulting from the cooling of the weld bead will impact crack Discussion of Results
propagation behavior. These stress may vary depending on Each of the models encountered convergence issues once
factors such as weld cooling rate and geometry, but can reach as the cracks started to grow depth-wise. In order to determine
high as yield stress in the HAZ [10]. Thus, residual stresses were whether depth growth would continue without lengthening
incorporated into the model by adding a step in which the recurring, a fourth case was run which modelled a longitudinal
elements in the weld material zone were cooled, resulting in crack on a straight section of pipe using the same pipe and crack
tensile stresses of 100% SMYS acting perpendicular to the weld. geometry as Cases 1 to 3. In this case, the model was able to run
to completion, with the crack growing through-wall before
Analysis Cases growing longitudinally along the pipe axis. This is consistent
Three analysis cases were created to explore how the with the results of the non-propagation cases where the largest J-
differences in properties between material regions and the integral value (i.e. crack-driving force) was at the centre of the
magnitude of residual stresses work together to determine crack crack front. For these reasons, the authors are confident that the
propagation behavior. crack modelled in Cases 1 to 3 would deepen and continue to
Case 1 was performed on a homogeneous model using only progress through the pipe wall before lengthening further.
the pipe body material properties throughout all material zones, It was shown in both Cases 1 and 2 that any lengthening of
and did not include residual stresses. This case established the an angled crack in the pipe body occurs longitudinally. This
baseline propagation behavior of an angled crack in the pipe suggests that any lengthening of angled cracks within the pipe
body. The crack initially turned to propagate longitudinally body will occur longitudinally in a way that could be detectable
along the pipe axis for 3 to 5 mm at either end of the crack. The by conventional crack in-line inspection tools.
crack then began to deepen, and the model was terminated. It was also shown that a material with a lower toughness is
These results are illustrated in Fig. 13 not necessarily the path of least resistance for a propagating
Case 2 incorporated all features of Case 1, with the addition crack. This was illustrated in Case 2, where once the crack
of different material properties for each material zone, according encountered the HAZ/pipe body boundary (where the HAZ had
to Table 3 and as shown in Fig. 14. This case was designed to the lower strength and toughness) and the HAZ/weld boundary
establish the propagation behavior of a crack encountering a (where the HAZ had the lower strength but higher toughness),
boundary between adjacent material zones. Similar to Case 1, the higher stress capacity of the pipe body and weld resulted in
the initial propagation behavior of the crack was to align the stress capacity being exceeded by the largest margin at the
perpendicular to the principal hoop stresses in the pipe and grow crack base, causing the crack to grow depth-wise sooner than in
longitudinally along the pipe axis. However, for Case 2, Case 1. Further studies investigating the impact of the magnitude
longitudinal crack growth ended once the crack front of the material strength differential may provide insight into
encountered a change in material at the HAZ/pipe body and ways in which crack lengthening might be arrested.
HAZ/weld boundaries, at which point the crack began to deepen. Finally, Case 3 showed that residual stresses acting in
This was due to the higher Tmax value for the weld and pipe body, tension perpendicular to the direction of the seam weld tend to
which temporarily arrested the crack and allowed the stress at the divert the direction of crack propagation away from the pipe axis
deepest part of the crack to increase to the point where damage and parallel to the angled seam weld. This illustrates that, in the
initiation would occur. These results are illustrated in Fig. 15, region near the weld, there are competing forces influencing the
where red is the weld material zone, beige is the HAZ material crack propagation direction: the hoop stresses caused by
zone, and green is the pipe body material zone. As in Case 1, the pressurization, which tend to extend the crack longitudinally,
model was terminated once the crack started growing and the residual stresses in the vicinity of the weld, which tend
depth-wise. to cause propagation parallel to the weld.
Case 3 included all the features of Case 2, with the addition There have been some experimental investigations into the
of the cooling step in the pipe weld. This case was designed to use of cohesive-zone modeling applied to X70 pipe; however,
establish the influence that the residual stresses near the seam these experiments were performed on machined test specimens
weld would have on crack propagation. As illustrated in Fig. 15, in a drop-weight tear test (DWTT) [11]. Furthermore, FE models
the initial crack propagation direction was not along the pipe based on these experiments were two-dimensional and did not
axis, but at an angle between that of the spiral weld and the use the XFEM method to predict crack propagation (i.e. the crack
longitudinal direction (material zone colours are consistent with propagation direction was determined beforehand from
Fig. 14). The crack growth of Case 1 is overlaid in orange for experimental tests). While experimental methods for validating
reference. This indicates that the principal stresses induced by the type of crack propagation model presented in this paper exist
the pipe pressure and those resulting from the residual stresses [12], such validation was considered outside the scope of this
represent competing forces that tend to direct the crack

8 Copyright © 2018 ASME

Downloaded From: https://proceedings.asmedigitalcollection.asme.org on 12/11/2018 Terms of Use: http://www.asme.org/about-asme/terms-of-use


project. For these reasons, quantitative values from the made herein to facilitate construction of the FE model itself.
propagation analyses are not reported. These assumptions were evaluated as reasonable by the authors,
but without experimental data they could not be fully validated.
The crack placement in the HAZ was not made precisely on
the weld/HAZ boundary in order to accommodate meshing of
the FE model. While this does not accurately reflect the location
of some types of cracks (e.g. incomplete weld fusion), this
placement allowed the crack to be influenced by the residual
stresses in the vicinity of the weld, as illustrated in Case 3. Based
on these results, the authors are confident that similar results
would have been obtained had the crack been placed directly on
the weld/HAZ boundary.
In addition, the transition of material properties between the
HAZ and pipe body is gradual and not abrupt. This would make
it easier for a crack to grow from the HAZ into the pipe body
than was shown in the present models, although the sharp
FIGURE 13. DETAIL VIEW OF EXTENT OF CRACK discontinuity between the HAZ and weld is realistic.
LENGTHENING FOR CASE 1
SUMMARY
The use of spiral welded pipe in liquids transmission lines is
a common industry practice. Cracks extending parallel to the
spiral seam weld at an angle to the pipe axis have been observed;
however, crack assessment and detection techniques are
generally designed for longitudinally aligned cracks. The study
was undertaken to explore how one such crack assessment
technique, the CorLAS crack model, could have its inputs
modified to support evaluation of the threat associated with
angled cracks in spiral welded pipe, as well as to anticipate the
crack growth direction of an angled crack to support spiral crack
management decision making.
FE models of longitudinal cracks were assessed by
FIGURE 14. DETAIL VIEW OF EXTENT OF CRACK comparing FE predicted failure pressures with those predicted by
LENGTHENING FOR CASE 2 the CorLAS crack model. Failure pressures predicted by
CorLAS were consistently lower than those predicted in the FE
analysis. Agreement between failure pressure predictions ranged
between 6% and 15%, with agreement falling off as the crack
depth approached 80% through-wall, which is consistent with
the findings reported by Jaske and Beavers [4] and the
recommendations in API 579. Due to these factors, the FE
models were considered to be capable of predicting crack failure
pressure.
Several approaches to applying the CorLAS model to angled
cracks were investigated by comparing predicted failure
pressures in FE models of angled cracks and the application of
the CorLAS model with modifications to the input parameters
used in calculating the toughness-based critical stress.
FIGURE 15. DETAIL VIEW OF THE EXTENT OF CRACK Modifications included only considering stresses perpendicular
LENGTHENING FOR CASES 1 AND 3 to the angled crack while using its full angled length (resolved
stress method), as well as procedures outlined in standards API
Model Limitations 579 and BS7910, which specify modification of crack length
Crack propagation is a highly complex process. Factors that while using the full hoop stress (projected length method). In all
influence crack growth include material inclusions, the cases, burst pressure increased with crack angle; however,
anisotropy of the pipe material, and the impact of pipe forming, pressures predicted by CorLAS when considering only
welding, and expansion. In addition, the transition of material perpendicular stresses (the resolved stress method) were non-
properties between the HAZ and pipe body is gradual and not conservative with respect to the FE results, and this approach is
abrupt. Correspondingly, several simplifying assumptions were not recommended. Little difference was observed in failure

9 Copyright © 2018 ASME

Downloaded From: https://proceedings.asmedigitalcollection.asme.org on 12/11/2018 Terms of Use: http://www.asme.org/about-asme/terms-of-use


pressures predicted by API 579 and BS7910 when their International Journal of Pressure Vessels and Piping Vol. 46
respective crack length modifications were used in the CorLAS No. 2(1991): pp.195-216.
model (the projected length method), and both were found to be [3] Bai, Y., Xu, T. and Bea, R. “Reliability-Based Design and
more conservative than the FE analyses while maintaining the Requalification Criteria for Longitudinally Corroded
increased pressure capacity observed with increasing crack Pipelines.” Proceedings of the International Offshore and
angle. Thus, it is recommended that these procedures be used to Polar Engineering Conference. ISOPE-I-97-424. Honolulu,
modify crack length when assessing angled cracks using the Hawaii, 1997.
CorLAS model. [4] Jaske, C.E and Beavers, J.A. “Integrity and Remaining Life
The crack propagation behavior of angled cracks was also of Pipe with Stress Corrosion Cracking.” PR-186-9709.
assessed using the extended-FE method capabilities of Abaqus
Prepared for the Materials Technical Committee of Pipeline
in consideration that the direction of growth could potentially
Research Council International Inc. (PRCI), Chantilly, VA.
affect the in-line inspection (ILI) detection probabilities and
2001.
other related management techniques. Due to the lack of
experimental data on the propagation of angled cracks in spiral [5] Liu, M., Wang, Y., Collins, L. “Tensile Strain Capacity of
welded pipe, the numerical values of outcomes produced in the Spiral Pipes.” Proceedings of the International Pipeline
propagation analyses, such as pressure at which propagation Conference. IPC 2012-90661. Calgary, Alberta, 2012.
initiates or crack final lengths, were not considered as predictive [6] American Petroleum Institute (API). Fitness for Service. API
of actual physical quantities. However, the results of the crack Recommended Practice 579-1, 3rd ed., June, 2016.
propagation models were used to highlight several qualitative [7] British Standard Institution (BS). Guide to Methods for
conclusions that could be considered for future pipeline design Assessing the Acceptability of Flaws in Metallic Structures.
considerations, or as a guideline for future investigations and British Standard BS 7910:2013+A1:2015, 3rd ed., 2013.
crack management decision making. [8] Yagawa, G., Kitajima, Y., and Ueda, H. “Three-Dimensional
It was observed that in the absence of material boundaries Fully Plastic Solutions for Semi-elliptical Surface Cracks.”
or residual stresses, it was the tendency of the spirally orientated International Journal of Pressure Vessels and Piping Vol.
cracks to propagate in the longitudinal pipe direction (i.e. 53 (1993): pp. 457-510.
perpendicular to the largest tensile principal stress). Introducing [9] Abaqus. Analysis User’s Guide. Section 10.7.1., 2016.
higher ultimate strengths in the pipe body and weld materials, [10] Dewees, D. J. “Comparison of 2D and 3D Welding
relative to the HAZ, contained the crack within the HAZ, acting Simulations of a Simple Plate.” Proceedings of the Pressure
as a barrier for crack propagation into these regions. The Vessels & Piping Conference. PVP 2012-78573. Toronto,
introduction of residual stresses due to cooling of the weld seam Ontario, 2012.
after manufacturing had the effect of orienting the crack [11] Dunbar, A., Wang, X., Tyson, W.R., and Xu, S. “Simulation
propagation direction more toward the spiral weld direction, of Ductile Crack Propagation and Determination of CTOAs
reflecting the change in local principal stresses. in Pipeline Steels Using Cohesive Zone Modelling.”
Fatigue & Fracture of Engineering Materials & Structures
REFERENCES
Vol. 37 No. 6 (2014): pp.592-602.
[1] Chen, J., Feng, Q., Wang, F., Zhang, H, and Song, H.
[12] Cornec, A., Scheider, I., and Schwalbe, KH. “On the
“Research on Burst Tests of Pipeline with Spiral Weld
Practical Application of the Cohesive Model.” Engineering
Defects.” Proceedings of the International Pipeline
Fracture Mechanics Vol. 70 No. 14 (2003): pp. 1963-1987.
Conference. IPC 2012-90089. Calgary, Alberta, 2012.
[2] Mok, D.H.B., Pick, R.J., Glover, A.G., and Hoff, R.
“Bursting of Line Pipe with Long External Corrosion.”

10 Copyright © 2018 ASME

Downloaded From: https://proceedings.asmedigitalcollection.asme.org on 12/11/2018 Terms of Use: http://www.asme.org/about-asme/terms-of-use

You might also like