Proceedings of the 2022 14th International Pipeline Conference
IPC2022
September 25 – September 29, 2022, Calgary, Alberta, Canada
UNDERSTANDING PRINCIPAL DRIVERS TO BURST PRESSURE AND LOCAL DEFORMATION OF
PIPES WITH SCC COLONIES
Yong-Yi Wang and Alex Wang
Center for Reliable Energy Systems
Dublin, OH, USA
ywang@cres-americas.com
Stephen C. Rapp and Gary Vervake
Enbridge
Houston, TX, USA
ABSTRACT
Two pups with SCC colonies were removed from service.
The most dominant cracks in the colonies are short and deep.
One of the pups had a bulge in the area of the SCC colony. The
two pups were subjected to hydrostatic burst tests and detailed
post-test metallurgical examination. A companion paper covers
the formation, dormancy, and growth of these SCC colonies
using forensic analysis. This paper covers two main elements
aimed at understanding the behavior of pipes with SCC colonies
and supporting the work covered in the companion paper:
(1) Burst pressure prediction, and
(2) Analysis of the bulge.
Burst pressure predictions were made with available
material properties and flaw dimensions measured by MPI, UT,
and PAUT prior to the actual burst tests. After the burst tests,
the predictions were updated with more relevant material
property data and the flaw dimensions obtained from fracture
surfaces exposed. The modified Ln-Sec, CorLAS™, MAT-8,
and Level 2 of API 579 burst pressure models and the associated
Charpy to fracture toughness correlations were used. Two flaw
interaction rules, CEPA and PRCI-CRES, were used to
determine the single equivalent crack dimensions for input into
the burst pressure models. The burst pressure predictions were
compared with the experimentally measured burst pressures.
Of the multiple factors affecting the burst pressure
prediction, the selection of flaw interaction rules has the most
prominent impact on the accuracy of the burst pressure
prediction. The selection of burst pressure models has a
secondary impact with the exception of API 579, which tends
to give lower burst pressure predictions than other models.
1 © 2022 by ASME
IPC2022-87338
Brian Leis
B. N. Leis Consultant, Inc.
Worthington, OH, USA
The formation of the bulge was simulated under different
longitudinal/axial loading conditions and two levels of internal
pressure. It is shown that the level of the residual bulge has a
strong dependence on the severity of SCC (length, depth, and
spatial distribution), the level of maximum internal pressure
before depressurization, and the longitudinal stress state.
Compressive longitudinal stress reduces the level of internal
pressure needed to produce a bulge of the same magnitude when
the severity of SCC remains constant. Multiple possible
conditions could have existed to produce the observed bulge.
KEYWORDS
SCC, burst pressure prediction, burst pressure models, flaw
interaction rules, bulge at SCC, effects of longitudinal/axial
loads
1 INTRODUCTION
Two pups with SCC colonies, named Dig 1777 and Dig
2692, were removed from a pipeline transporting natural gas in
2021. The most dominant cracks in the colonies are short and
deep. A bulge in the area of the SCC colony was visible on Dig
1777.
The pipeline was constructed in 1957 with API Grade X52
pipes with 30-inch OD and 0.375-inch WT. A hydrostatic test
was done in 1986. The two pups were on the discharge side of
a compressor station before the flow direction was reversed in
2015. The MAOP before and after the 2015 flow reversal was
1040 psi (80% SMYS) and 936 psi (72% SMYS), respectively.
The two pups were subjected to hydrostatic burst tests and
detailed post-test metallurgical examination in 2021. A
companion paper covers the formation, dormancy, and growth
of these SCC colonies using forensic analysis [1]. This paper
covers two main elements aimed at understanding the behavior
of pipes with SCC colonies and supporting the work covered in
the companion paper:
(1) Burst pressure prediction, and
(2) Analysis of the bulge in Dig 1777.
The work covered in this paper was produced in stages over
a course of several months. The material properties used in
different stages were slightly different, reflecting the best
available information at the time the work was performed. The
differences are generally small and should not affect the major
outcomes and conclusions from this work.
2 BURST PRESSURE PREDICTION
2.1 Overview of the Burst Pressure Prediction
Process
The overall burst pressure prediction process follows the
established process at CRES [2,3,4]. Each burst pressure
prediction goes through the following steps:
1. Extract crack sizes and locations from photos of a
colony,
2. Assign depths to extracted cracks using colony depth
data,
3. Apply interaction rules to cracks to determine
dimensions of equivalent cracks,
4. Use equivalent crack dimensions and pipe properties
as inputs to various burst pressure models to obtain
predicted burst pressures, and
5. Define the predicted burst pressure as the lowest burst
pressure produced in Step 4.
The Modified Ln-Sec, CorLAS™, MAT-8, and API 579
burst pressure models were used to produce burst pressure
predictions.
2.2 Burst Pressure Prediction of Dig 1777
Two predictions were produced for Dig 1777: one
prediction before the full-scale burst test using estimated pipe
properties and NDE crack depths, and one prediction after the
burst test using actual pipe properties and pre-test NDE crack
depths that have been augmented with measurements from the
post-test fracture surfaces.
2.2.1 Pipe Properties
The pipe tensile properties and Charpy impact energies
used to produce burst pressure predictions are summarized in
Table 2-1. Two sets of mechanical properties, listed as Pre-Test
Estimate and Subject Pipe, were obtained from information
available at the corresponding times.
Pipe properties from either the pipe containing the colony
or its adjacent joints were not available at the time the pre-burst
predictions were produced. Instead, pipe properties from pipes
of the same pipeline segment were used to produce estimated
pipe properties that were assumed to be close to the actual
properties of the Dig 1777 pipe. These estimated properties are
listed under the “Pre-test Estimate” column. The actual pipe
properties were procured from small-scale mechanical tests
using samples taken from the pipe containing the SCC colony
after the burst test. This second set of mechanical properties is
listed under the “Subject Pipe” column.
Table 2-1: MECHANICAL PROPERTIES USED FOR DIG 1777
BURST PRESSURE PREDICTION
Pre-test
Subject Pipe
Estimate
YS (psi) 55800 59000
UTS (psi) 79500 78900
Full-Size Energy
20 28
(ft-lb)
Charpy
Test Temp (°F) 41 70
Shear Area (%) 22 76.7
2.2.2 NDT and Flaw Size
Crack length and location data were extracted from a
combination of two photos of the colony. The first photo, shown
in Figure 2-1, was provided to CRES for an initial assessment
of the colony. The second photo, shown in Figure 2-2, was
provided later along with field UT data. Neither photo
contained all the data needed to make a burst pressure
prediction, as the markings indicating locations at which UT
depth measurements were taken were only present on the
second photo, but that photo did not have a ruler or markings
that could be used as such. As a result, the two photos were
combined using an image editing program, resulting in the
photo shown in Figure 2-3. An in-house program was then
used to extract the in-plane dimensions of the cracks and their
spatial locations from the photo.
A location-to-depth mapping was then used to assign
depths to the extracted cracks. This mapping was constructed
by using the field UT markings on the photo in Figure 2-2 and
the depth measurements in the accompanying Excel file
provided to CRES to produce an initial mapping, then
superimposing PAUT data for locations where such data was
available.
Both the CEPA and PRCI-CRES interaction rules were
then separately applied to the extracted dimensions and spatial
locations of cracks and their corresponding depths to produce
two sets of equivalent cracks. These, along with the pipe
properties described in Table 2-1, were fed into burst pressure
models to produce two sets of predicted burst pressures. The
equivalent cracks from each set that produced the lowest burst
pressures were then used for the final burst pressure prediction.
The dimensions of those equivalent cracks are shown in Table
2-2.
2 © 2022 by ASME
 Table 2-2: EQUIVALENT CRACK DIMENSIONS FOR THE
DIG 1777 PRE-TEST BURST PRESSURE PREDICTION

Interaction Rule Crack Length (in) Crack Depth (in)

CEPA 3.870 0.288

PRCI-CRES 1.697 0.288

The derivation of the equivalent crack sizes for the post-

test burst pressure prediction was done using a procedure
similar to that for the pre-test burst pressure prediction. The
only difference is that the depths of the cracks exposed by the
post-depth fracture surface were updated with measurements
taken from the photo of the post-burst fracture surface shown in
Figure 2-4. The dimensions of the resulting equivalent cracks
are shown in Table 2-3. There is a very small increase in the
depth of the equivalent cracks.
Figure 2-1: THE FIRST OF TWO PHOTOS USED TO LOCATE
AND SIZE CRACKS FOR DIG 1777

Figure 2-2: THE SECOND OF TWO PHOTOS USED TO

LOCATE AND SIZE CRACKS FOR DIG 1777

Figure 2-4: EXPOSED FRACTURE SURFACE AFTER BURST

TEST OF DIG 1777
Table 2-3: EQUIVALENT CRACK DIMENSIONS FOR THE
DIG 1777 POST-TEST BURST PRESSURE PREDICTION

Interaction Rule Crack Length (in) Crack Depth (in)

CEPA 3.870 0.296

PRCI-CRES 1.697 0.296

2.2.3 Comparison of Predicted Burst Pressure and

Full-Scale Test Results
Four burst pressure prediction models were used to predict
burst pressures using the pipe properties and the equivalent
crack sizes from the CEPA and PRCI-CRES rules. For the
Figure 2-3: THE COMBINED PHOTO OF THE ENTIRE DIG
predictions with the MAT-8 model, two Charpy to Kmat
1777 COLONY
correlations were used, one from API 579 and the other from
Wallin [5]. The Modified Ln-Sec and CorLAS™ burst pressure
models require flow stress as an input parameter. The arithmetic

3 © 2022 by ASME
mean of the YS and UTS values in Table 2-1 was used for the
flow stress for these two burst pressure models.
The predicted burst pressure and the burst pressure of 1455
psi from full-scale test are compared in Figure 2-5 and Figure
2-6. The only differences between the two rounds of the
predictions were the material properties and flaw depths used
for burst pressure prediction. Compared to the predicted burst
pressure prior to the burst test, the post-burst predictions are
noticeably higher with the CorLAS, MAT-8, and API 579
Level 2 burst pressure models, and are closer to the
experimentally measured burst pressure. The equivalent crack
dimensions (i.e., the choice of interaction rule) have the largest
effect on the accuracy of the predicted burst pressure. The
choice of burst pressure models has a smaller effect on the
accuracy of the predicted burst pressure.
Figure 2-5: COMPARISON OF BURST PRESSURE
PREDICTIONS MADE PRIOR TO THE FULL-SCALE BURST
TEST WITH THE MEASURED BURST PRESSURE OF DIG 1777
Figure 2-6: COMPARISON OF BURST PRESSURE
PREDICTIONS AFTER THE FULL-SCALE BURST TEST WITH
THE MEASURED BURST PRESSURE OF DIG 1777
2.3 Burst Pressure Prediction of Dig 2692
2.3.1 Pipe Properties
Pipe properties from either the Dig 1777 or Dig 2692 joints
were not available at the time the Dig 2692 pre-test burst
pressure prediction was made. Material properties from a Dig
1777-adjacent joint as described in Table 2-4 were used under
the assumption that those properties would be close to those for
the Dig 2692 joint.
Table 2-4: MATERIAL PROPERTIES OF A JOINT ADJACENT
TO THE DIG 1777 JOINT USED FOR DIG 2692 PREDICTION
YS (psi) 59700
UTS (psi) 80200
Flow Stress (psi) 69950
Full-size Charpy @
27.1
70°F (ft-lb)
2.3.2 NDT and Flaw Size
An in-house program was used to extract crack length and
location data from the photo of the Dig 2692 SCC colony shown
in Figure 2-7.
Figure 2-7: DIG 2692 SCC COLONY
Since no other depth information was available, the PAUT
markings on the photo were used to construct a location-to-
depth mapping which was then used to assign depths to the
extracted cracks. The CEPA interaction rules were then applied
to these cracks to obtain a set of equivalent cracks. Those, along
with the pipe properties described in Table 2-4, were fed into
the Modified Ln-Sec, CorLAS, and MAT-8 burst pressure
models to obtain corresponding burst pressures for each
equivalent crack. The equivalent crack that produced the lowest
burst pressure prediction was then used for the final burst
pressure prediction.
The depth of the largest single crack was out of range for
the PRCI-CRES interaction rules, so a fallback method of
treating single cracks as equivalent cracks was used. The
dimensions of those cracks used for the final burst pressure
prediction are described in Table 2-5.
4 © 2022 by ASME
 Table 2-5: CRACK DIMENSIONS FOR THE DIG 2692 PRE-
TEST BURST PRESSURE PREDICTION
Interaction Rule Crack Length (in) Crack Depth (in)
None 1.610 0.330
CEPA 3.425 0.330
2.3.3 Comparison of Predicted Burst Pressures with
Full-Scale Test Results
The Dig 2692 pipe did not burst. A leak developed at 1570
psi, after which the test was stopped. The maximum pressure
attained during the test was 1574 psi.
The Dig 2692 pre-test burst pressure predictions are shown
in Figure 2-8, along with lines indicating the maximum pressure
attained during the test and the MAOP after the 2015 flow
reversal.
Figure 2-8: PRE-TEST BURST PRESSURE PREDICTIONS
FOR DIG 2692
2.3.4 Observations from the Burst Pressure
Predictions
The predicted burst pressure was lower than the observed
highest pressure from the burst test by a fair amount. From the
perspective of equivalent crack dimensions, the pressure values
from the single largest crack are the highest value possible as
the application of any flaw interactions would have produced
an equivalent crack no smaller than the largest single crack.
The impact of the variations of mechanical properties is
examined. Various estimates and tests put the yield strength in
the range of 59.7-60.2 ksi and the range of UTS in the range of
80.2-83.6 ksi. Mechanical tests of pipes adjacent to the Dig
1777 and Dig 2692 indicate the tensile properties are within a
very narrow band. Therefore, possible variations in mechanical
properties are unlikely to explain the burst pressure
underpredictions.
The other possibility is the variations in flaw dimensions
from various measurements. The dimensions used for the
prediction of Figure 2-8 were from images of MPI and PAUT
depth measurement. The examination of the fracture surfaces
produced slightly different crack dimensions, with a maximum
depth of 0.333 inches and a length of 1.503 inches [1]. These
dimensions would have produced predicted burst pressures of
1354, 1335, and 1340 psi from Modified Ln-Sec, CorLAS,
and MAT-8, respectively, slightly higher than that given in
Figure 2-8, but the overall difference is rather small
At a depth of 0.330 inches, the crack depth to wall
thickness ratio (a/t) is 0.880. The trends of predicted burst
pressure for deep cracks from the three models in Figure 2-8
are examined in Figure 2-9. There is an accelerated drop in
predicted burst pressure from all three models from an a/t of
0.80 to 0.90. These trends are expected as the models are
expected to predict burst pressure of zero when the crack depth
approaches the wall thickness. As a short and deep crack would
be more likely to produce a leak rather than a burst, it is quite
likely the current setup of the burst pressure models is
underpredicting the burst pressure for relatively short and very
deep cracks. Indeed, as indicated by Leis, et al., the burst
pressure can be more precisely predicted by using a predictive
model referenced to UTS when the flaw becomes short and
deep [1].
Figure 2-9: PREDICTED BURST PRESSURE AS A FUNCTION
OF RELATIVE CRACK DEPTH (a/t) FROM THREE BURST
PRESSURE MODELS
3 ANALYSIS OF BULGE FORMATION
3.1 Background and Objective
A heat map of the bulge in the SCC area of Dig 1777 was
obtained from a laser scan and is shown in Figure 3-1. The
laser scan done at zero internal pressure before the 2021 burst
test indicated a maximum bulge of 0.0596 inches. Figure 3-2
is an overlay of the SCC colony on the bulge contour.
The sequence of events leading to the formation of the
bulge was one of the focuses of this investigation. The initial
thought was that the bulge was formed while the pipeline was
in service. If this were the case, the pipe was expected to fail at
a pressure level not much higher the MAOP as the bulge would
be an indication of imminent failure. Since the pup failed at a
5 © 2022 by ASME
pressure much higher than the MAOPs either before or after The work reported here was meant to examine possible
flow reversal in the 2021 burst test, it was thought that the bulge scenarios of bulge formation.
could have been formed during the 1986 hydrostatic test.

Figure 3-1: LASER SCAN CONTOUR PLOT OF THE BULGE REGION

Figure 3-2: OVERLAY OF THE SCC CRACKS ON THE BULGE

6 © 2022 by ASME
3.2 Analysis Approach and Outcomes value on the x-axis after the internal pressure is reduced to zero
3.2.1 FE Model Setup (depressurized). It is evident that the magnitude of the bulge
increases at a higher rate at higher pressure. Figure 3-6 shows
Figure 3-3 shows the engineering stress-strain curve from
the variation in the magnitude of the residual bulge as a function
the tensile test as well as the true stress-strain curve used for the
of crack length. The magnitude of the residual bulge increases
FE modeling.
at a fast rate with the increase of crack length.
100
1600
90
1400
80

Internal Pressure (psi)

70 1200
Stress (ksi)

60 1000
50
800 2c=1.250"
40
600 2c=2.250"
30
2c=3.000"
20 True stress-strain curve (input to FE) 400 2c=4.000"
10 Engineering stress-strain curve (test) 2c=5.000"
200
0 2c=6.000"
0 0.05 0.1 0.15 0.2 0.25 0
0 0.025 0.05 0.075 0.1 0.125 0.15 0.175 0.2
Strain (in/in)
Bulge Magnitude (inch)
Figure 3-3: STRESS-STRAIN CURVES USED IN THE BULGE
Figure 3-5: RELATIONSHIP BETWEEN THE INTERNAL
ANALYSIS
PRESSURE AND THE MAGNITUDE OF BULGE FOR CRACKS
OF DIFFERENT LENGTHS AT A FIXED a/t RATIO OF 0.789
The FE modeling approach can be summarized as follows: 0.12

The overall geometry of the model has the nominal

Depressurized Bulge Magnitude

1.
0.1
pipe diameter and wall thickness. The length of the
model was set to 20D. 0.08
2. A quarter of the pipe was modelled by applying
(inch)

appropriate symmetry boundary conditions. 0.06

3. The cracks were modelled as semi-elliptical surface-

0.04
breaking flaws on the OD side as shown in Figure 3-4.
0.02

0
0 1 2 3 4 5 6 7
Crack Length (inch)

Figure 3-6: MAGNITUDE OF RESIDUAL BULGE AS A

FUNCTION OF CRACK LENGTHS AT A FIXED CRACK DEPTH
RATIO a/t OF 0.789
3.2.3 Bulge Formation When Pressurized to 1986
Hydrostatic Test Pressure

Figure 3-4: SCHEMATIC OF A FE MODEL AND THE As shown in Section 3.2.2, crack size could have a major
BOUNDARY CONDITIONS impact on the residual bulge. To represent the SCC colony more
accurately, models with four dominant cracks, shown in Figure
3.2.2 Effects of Crack Size and Internal Pressure on
3-7, were generated for further analysis. The maximum crack
Residual Bulge depth was changed slightly to a/t=0.75 based on the updated
To understand the effect of crack size and pressure on the information at the time of the analysis.
magnitude of the bulge, analysis was first done with models
containing a single semi-elliptical crack with a fixed depth of
a/t=0.789 and multiple lengths.
The relationship between the internal pressure and the
magnitude of bulge for cracks of different lengths at a fixed a/t
ratio of 0.789 is shown in Figure 3-5. The residual bulge is the

7 © 2022 by ASME
Figure 3-7: CRACK DIMENSIONS AND LOCATIONS IN THE
BULGE AREA USED IN THE FE MODEL
Three longitudinal/axial boundary conditions were applied
and maintained during the application of internal pressure:
(1) Restrained end, equivalent to a buried pipeline which is Figure 3-9: RESIDUAL BULGE SHAPE WITH A FREE-END
restrained from longitudinal/axial movement, BOUNDARY CONDITION AND A MAXIMUM PRESSURE OF
(2) Free end condition, i.e., pipe is free to contract in the axial 1336 PSI
direction when internal pressure is applied, and
(3) Axial compressive force equivalent to one-half of yield
force.
The maximum internal pressure applied was the location-
specific 1986 hydrostatic test pressure of 1,336 psi. The bulge
magnitude was obtained after the pipe was depressurized.
Figure 3-8, Figure 3-9 and Figure 3-10 show the bulge
region under three longitudinal/axial boundary conditions.
Table 3-1 lists the magnitude of the residual bulge. The results
indicate that the magnitude of the residual bulge is sensitive to
the longitudinal loading condition. The fixed-end condition
produces tensile longitudinal/axial stress due to Poisson’s
effect. The free-end condition would have zero
longitudinal/axial stress, whereas the axial compression Figure 3-10: RESIDUAL BULGE SHAPE WITH A
scenario would have produced compressive longitudinal/axial COMPRESSIVE LONGITUDINAL/AXIAL LOAD BEING 50%
stress. It is evident that when the mode of the longitudinal/axial YIELD LOAD AND A MAXIMUM PRESSURE OF 1336 PSI
load changes from tension to compression, the magnitude of Table 3-1: MAGNITUDE OF RESIDUAL BULGE WITH
residual bulge increases remarkably. DIFFERENT LONGITUDINAL (AXIAL) LOADING
CONDITIONS WHEN PRESSURIZED TO THE 1986
HYDROSTATIC TEST PRESSURE OF 1336 PSI

Axial
End Condition Fixed End Free End
Compression

Magnitude of
0.047 0.084 0.195
Bulge (in)

3.2.4 Bulge Formation with Pre-Flow-Reversal MAOP

The change in flow direction was thought to have halted
the high-pH SCC as the SCC sites went from the discharge to
suction side of the compressor station. Therefore, the relevant
in-service bulge formation would be prior to flow reversal, in
part due to higher MAOP before the flow reversal.
Figure 3-8: RESIDUAL BULGE SHAPE WITH A FIXED-END
BOUNDARY CONDITION AND A MAXIMUM PRESSURE OF
1336 PSI

8 © 2022 by ASME

Figure 3-11: RESIDUAL BULGE SHAPE WITH A
LONGITUDINAL LOAD BEING 25% OF YIELD LOAD AND A
MAXIMUM PRESSURE OF 1040 PSI
Figure 3-12: RESIDUAL BULGE SHAPE WITH A
LONGITUDINAL LOAD BEING 50% OF YIELD LOAD AND A
MAXIMUM PRESSURE OF 1040 PSI
Three conditions of compressive axial loads were
investigated: 0.25, 0.50, 0.75 × yield loads. These loads were
maintained while the pipe sections were pressurized to the pre-
flow-reversal MAOP of 1040 psi. The bulge was examined after
depressurization.
Figure 3-11 and Figure 3-12 shows the residual bulge
shapes for two different longitudinal/axial loading conditions.
The maximum magnitude of the bulge is given in Table 3-2.
Table 3-2: MAXIMUM MAGNITUDE OF BULGE WITH
A MAXIMUM PRESSURE OF 1040 PSI AND THREE
DIFFERENT AXIAL LOAD CONDITIONS
Axial Load Axial Compressive Load / Yield Load
Condition 0.25 0.5 0.75
Magnitude of
0.033 0.046 0.075
Bulge (in)
3.3 Observations
Possible bulge formation under two different internal
pressures and multiple possibilities of longitudinal/axial load
were examined. It is evident that the level of the residual bulge
has a strong dependence on the severity of SCC (length, depth,
and spatial distribution), the level of maximum internal pressure
before depressurization, and the longitudinal stress state. The
state of one or two conditions can affect the condition(s) of
other factors necessary to achieve the same level of residual
bulge. At least three possibilities exist for the formation of the
bulge with the magnitude measured by laser scan:
(1) If the longitudinal stress condition at the location of
the bulge is similar to an end-restraint condition,
equivalent to a long buried pipeline restrained from
longitudinal movement, the bulge could exist if (1)
there was an internal pressure equal to the 1986
hydrostatic test and (2) crack morphology (length,
depth, and spatial positions) was similar to that
discovered in 2021, implying minimal crack growth
from 1986 to 2015.
(2) If the longitudinal stress condition at the location of
the bulge had less tension than the conditions of
Scenario 1 or some level of compression, the bulge
could exist if (1) there was an internal pressure equal
to the 1986 hydrostatic test and (2) a less severe crack
morphology (length, depth, and spatial positions) than
Scenario 1, implying some level of crack growth from
1986 to 2015 would be permitted.
(3) The formation of the bulge during service is possible
prior the 2015 pressure reversal if the location had a
moderately high level of compressive load.
The role of longitudinal/axial stress is not commonly
considered in burst pressure predictions. There is a perception
that high longitudinal stress could only exist in areas with
geohazards. This perception is not always true [6]. At least two
sources could have contributed to compressive
longitudinal/axial load/stress:
(1) Changes in pipe support conditions from normal
settlement and construction practice can introduce
bending deformation, leading to compressive
longitudinal stress in certain o’clock positions of a
pipe. This becomes more likely due to the hilly terrain
at the location of Dig 1777.
(2) Compressive longitudinal/axial loads can be produced
by the high temperature at the location. Dig 1777 was
on the discharge side of a compressor station with a
relatively high temperature that led to high pH SCC.
An increase in temperature of 60 ºF would produce a
compressive stress of 23% SMYS.
The compressive load/stress produced by the above-
mentioned possible sources would have been compensated for
by the tensile stress produced by the Poisson’s effect due to
internal pressure. On balance, it is possible to have a moderately
high compressive load/stress, although this possibility is
difficult to verify after the fact.
9 © 2022 by ASME
4 CONCLUDING REMARKS
As observed in prior work, the choice of flaw interaction
rules, and thus the resulting equivalent crack dimensions, has a
much greater impact on the accuracy of the burst pressure
prediction than the choice of burst pressure models. The burst
pressure of Dig 1777 was accurately predicted using the PRCI-
CRES flaw interaction rules. The predicted burst pressure of
Dig 2692 is lower than the observed maximum pressure during
the burst test which produced a leak. The causes of the under-
predictions are not definitive after examining multiple
possibilities. The most likely cause is the burst pressure models
which are under-predicting the burst pressure for pipes with
short and very deep flaws.
The formation of the bulge was simulated under different
longitudinal/axial loading conditions and two levels of internal
pressure. It is shown that the level of the residual bulge has a
strong dependence on the severity of SCC (length, depth, and
spatial distribution), the level of maximum internal pressure
before depressurization, and the longitudinal stress state.
Compressive longitudinal stress reduces the level of internal
pressure needed to produce a bulge of the same magnitude.
Multiple possible conditions could have existed to produce the
observed bulge. With a compressive longitudinal stress,
potentially caused by elevated temperature and settlement and
construction stresses, the bulge could have formed during the
1986 hydrostatic test or later in service with additional growth
of SCC before the flow reversal in 2015. The formation of the
1 Leis, B., Wang, Y.-Y., Eshraghi, A., Rapp, S. and Vervake,
G., “The Practical Benefits of Detailed Forensic Analysis,”
Proceedings of the 14th International Pipeline Conference,
Paper No. IPC2022-87028, September 26-30, 2022,
Calgary, AB, Canada.
2 Wang, B., Liu, B., Wang, Y.-Y., Wang, A., and Rapp, S.,
“Burst Pressure Prediction of Pipes with SCC Colonies -
Development of Intelligent Flaw Interaction Rules,”
Proceedings of the 13th International Pipeline Conference,
Paper No. IPC2020-9693, September 28 – October 1, 2020,
Calgary, AB, Canada.
3 Wang, B., Liu, B., Wang, Y.-Y., Wang, A., and Rapp, S.,
“Burst Pressure Prediction of Pipes with SCC Colonies –
Evaluation of Intelligent Flaw Interaction Rules Using
Full-Scale Burst Tests,” Proceedings of the 13th
bulge of observed magnitude is also possible during the 1986
hydrostatic test with a tensile longitudinal stress if the severity
of the SCC was close to that observed in 2021, implying very
minimal or no growth after the 1986 hydrostatic test.
Understanding the possible scenarios of bulge formation
has implications on the margin of safety while the pipe was in
operation. If the bulge was formed during the 1986 hydrostatic
testing, the pipeline would have had a greater margin of safety
than if the bulge was formed during service after the 1986
hydrostatic testing.
The possible role of compressive longitudinal/axial
load/stress on the bulge formation also has implications on the
interpretation of laboratory burst tests. These tests are almost
always done with end-caped vessels, producing burst pressures
higher than what it would be if the longitudinal/axial stress is
less tensile or even compressive.
5 ACKNOWLEDGMENT
The permission by Enbridge to publish this work is greatly
appreciated. Mr. Jiawei Wang assisted the analysis of the
residual bulge. Mr. David Warman provided valuable
comments of the work. The authors would also like to
acknowledge the contributions of Dr. Amin Eshraghi. The
assistance of Dr. Dan Jia and Mr. Paul Pianca on the paper
submission process is much appreciated.
6 REFERENCES
International Pipeline Conference, Paper No. IPC2020-
9696, September 28 – October 1, 2020, Calgary, AB,
Canada.
4 Wang, Y.-Y., Wang, B., Liu, B., and Wang, A., “Burst
Pressure Prediction of Pipes with SCC Using Intelligent
Flaw Interaction Rules,” Paper No. 102, Pipeline Pigging
and Integrity Management Conference, February 2-4,
2022, Houston, TX, USA.
5 Wallin, K., Fracture Toughness of Engineering Materials:
Estimation and Application. EMAS Publishing, Birchwood
Park, Warrington, UK, 2011
6 Wang, Y.-Y, Liu, B., and Fleck P., “Strain-Based Design
and Assessment of New and Ageing Pipelines,”
Technology for Future and Ageing Pipelines, October 19-
21, 2021, Gent, Belgium
10 © 2022 by ASME