Nearly 40 years ago, an 8-inch gasoline pipeline failed
north of Minneapolis, resulting in a serious incident with significant
environmental and property damages.
In its investigation,1 the National Transportation Safety Board (NTSB) said the pipeline, which had
been hydrostatically pressure tested above its normal operating pressure two years earlier, had
failed at a pressure below that.
Although the NTSB couldn’t say with
complete certainty what caused the failure, investigators concluded susceptibility
of the low frequency, electric resistance welded (LF-ERW) pipe to weld
corrosion was a contributing factor. LF-ERW pipelines installed during a
certain era often did not undergo a proper post-weld heat treatment, and NTSB
believed the corrosion had weakened the weld.
To
try to prevent something similar to this failure from occurring, the NTSB called for better data collection on pipeline leaks and failures,
including those potentially related to selective seam weld corrosion (SSWC). SSWC
occurs in the bond
line of the longitudinal seam weld,
where the corrosion attacks the long seam weld material preferentially compared
to the surrounding base pipe material. The NTSB recommendations evolved into
today’s pipeline integrity management requirements.
Though
there have been a small number of SSWC-related events since 1986, fortunately, none
have risen to the same impactful level. Much of the credit goes to coatings and
cathodic protection that help keep corrosion at bay. In addition, improved pipe
manufacturing practices have prevented LF-ERW pipeline material from being installed
since the late 1960s and early 1970s.2 Even
today, industry research organizations such as Pipeline Research Council
International (PRCI) continue to research SSWC susceptibility and integrity
management recommended practices.
Further,
given that 80% of U.S. pipelines were already in place before 1970, the ability
to detect, identify and size SSWC is just as important as ever — perhaps more
so, since SSWC is a time-dependent threat with the potential failure risk
increasing over time if the corrosion is still active.
In
fact, the problem is so urgent that the Pipeline Hazardous Materials and Safety
Administration (PHMSA) gives operators of hazardous liquid pipelines just 180
days to evaluate and remediate corrosion of or along the longitudinal seam weld
after it’s been discovered, regardless of the threat posed.
Interacting
Threats
SSWC can occur when a susceptible long seam
weld is subjected to an active corrosion environment. Because of its susceptibility to preferential
corrosion, the long seam weld material corrodes at a faster rate than the pipe
body material — Kiefner and Associates said between two to four times faster3 — resulting in a deep V-shaped groove aligned with the long seam weld axis.
Due to its aggressive corrosion depth growth
rate and orientation perpendicular to the primary stress, an SSWC anomaly poses
a greater threat to pipeline integrity than a similar volume of general
corrosion crossing the long seam weld does.
Kiefner also notes that SSWC is an
interacting threat. Typically, interacting threats are defined as two or more
coincident features or anomalies whose coincidence results in a greater
integrity threat to the pipeline system than what each feature or anomaly poses
individually.
In the case of SSWC, classic corrosion
mechanisms occur coincident with a susceptible seam material. This interacting
nature poses a greater threat risk for a number of reasons according to Kiefner,
including:
One Not Enough
There’s no doubt that over the years, advances in ILI technologies
have made it easier to detect
and characterize anomalies in the long seam weld. Complex sensor arrangements
on magnetizer assemblies that magnetize the pipe in complementary directions
have been particularly useful.
There is a key reason why. SSWC features have
a pair of nested feature geometries: In many cases, the seam weld, with its
narrow, axially oriented slotting feature, is embedded in a field of general
volumetric corrosion.
Further complicating matters, said T.D.
Williamson (TDW) Principal Data Science Engineer Adrian Belanger, is the fact
that the corrosion containing the selective seam weld corrosion can manifest
itself in a number of ways.
For example, he said, it’s possible for a
patch of corrosion to cross the seam without manifesting any preferential
growth along the seam but contain deep, narrow, axially aligned grooves that
follow the seam. Additionally, the corrosion could be preferential to the seam
with only shallow SSWC notching within it. In both cases, using only a single
direction of magnetization can mischaracterize the corrosion or miss it
altogether.
“Because the conditions vary so widely, it’s
extremely difficult, if not impossible, to consistently distinguish between
SSWC and general corrosion crossing the long seam by using a single
magnetization direction,” Belanger said. In essence, regardless of the number
of sensors thrown at the problem, there is only so much information to be
gleaned from a single magnetization direction.
“It’s not what you see, it’s what you don’t
that matters,” Belanger said.
Improving accuracy requires multiple magnetic
field orientations, and one of those fields must be sensitive to long, axially
oriented, narrow pipeline defects. Using multiple magnetic field directions
provides additional context about the actual anomaly geometry and position.
Spiral
MFL (SMFL) is the only technology that generates a helically oriented (spiral
or corkscrew-shaped) field in the pipe wall. Beyond the general detection of
the long seam, the SMFL data provides a complementary view of the SSWC defect —
“kind of like taking a photo from two different perspectives to create a 3D image,”
Belanger said.
Figure 1 displays the axial MFL and SMFL data streams for a section of pipe, the long seam weld is typically very clear in the SMFL data stream but much harder to detect in the MFL data.
As another example, Figure 2 shows the circumferential profile of
an SSWC feature for MFL and SMFL datasets. The ILI tool sensor data is on the
x-axis, and the median centered gauss values recorded by the tool’s Hall effect
sensors are on the y-axis. (Hall sensors are sensitive to the strength of the
magnetic field. As the magnetic field increases in intensity, the voltage
returned from the Hall sensor changes.)
The MFL profile is indicative of a general volumetric corrosion feature with
a smooth, parabolic-like shape. Conversely, the SMFL has a distinct peak within
the area of general corrosion indicating that a narrow axial feature is also present.
Without the SMFL dataset, there would be no indication that a narrow axial
feature is present.
TDW Data Science Engineering Manager Robert
Coleman said that SpirALL® MFL, one of six primary sensor technologies on the
company’s MDS® Pro platform, “provides a more thorough look at axial features, particularly
in or near the seam weld. It enables a more robust assessment of the signal than
traditionally axial fields.”
The
MDS Pro platform simultaneously delivers multiple data streams to
provide a comprehensive analysis of the pipeline. In addition to SpirALL MFL, the
MDS Pro platform includes:
First SSWC Classifier
Considering that MFL field is less impacted by axially oriented anomalies,
comparing its data to that of SMFL improves the characterization of SSWC
anomalies.
In fact, the difference in signal response between MFL and SMFL was critical
to the Department of Transportation (DOT)5 project that eventually led to the development of the industry’s first
validated SSWC classifier by TDW.
The classifier leverages the data collected by the MDS Pro platform to
characterize long seam weld corrosion anomalies. It utilizes an algorithm based
on the amplitude and width of both MFL and SMFL signals.
The project noted a large SMFL amplitude response that was sharply narrow
in the width direction as an indicator that the anomaly was circumferentially narrow
and axially oriented. When the SMFL signal response is uncharacteristically
large compared to the corresponding MFL response at the same location, the
anomaly is more likely to be SSWC.
To validate the model, TDW analyzed about 700
SSWC calls made on 75 ILI runs between 2013 and 2021. Dig feedback data was
available for 201 of these calls, and NDE only found three SSWC anomalies that
were not reported as SSWC by the classifier.
Additional features often misidentified by
ILI were used as controls during validation, including general corrosion
crossing the long seam weld, lack of fusion (LOF), mill anomalies, long seam
weld variations, axial planars and linear indications. The validation process
demonstrated the classifiers’ ability to accurately classify features as SSWC, distinguishing
them from features that might create similar signatures in MFL data alone.
Since TDW introduced the classifier, it has
been used by operators across the globe to properly classify seam weld
corrosion anomalies as either SSWC or general corrosion coincident with the
long seam. The validated performance
specification allows operators to confidently characterize the integrity risk
of each anomaly and prioritize their mitigative actions.
Seeing the True Nature of SSWC
Key to the success of the TDW classifier is
the approach leveraging two separate high-field MFL technologies, axial MFL and
SMFL, to characterize the seam weld corrosion anomalies.
The axial nature of the SSWC feature limits
the field response that is achievable with a pure axial MFL tool configuration,
independent of the axial tool sensor resolution. Improved classification is
achieved with a magnetic field crossing the axial SSWC anomaly, characterizing
the true axial nature of SSWC anomalies when present.
With better data, pipeline operators can
figure out what they’ve been missing. The result is safer, more compliant
systems. P&GJ
REFERENCES