You are on page 1of 10

Proceedings of the ASME 2022 14th International Pipeline Conference

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

IPC2022-87676

NEW APPROACHES IN UTILIZING EDDY CURRENT TESTING TO ADDRESS PIPELINE


INLINE INSPECTION REQUIREMENT

Sylvain Cornu, Raymond Karé Ahmed Sweedy, Michael Sirois


NDT Global Eddyfi Technologies
Stutensee, Germany Quebec, Canada

ABSTRACT development will make measurements in 2-D, longitudinal and


Eddy current testing is one of the most widely used NDT hoop directions, and will include a full array of sensors to
inspection techniques for both ferrous and non-ferrous provide a full coverage 2-D stress map of the pipe wall.
materials. This may explain why it was one of the very early The latest development will address accurate detection and
techniques implemented on an inline inspection tool (ILI) back sizing of defects on the internal wall of the pipeline. Most crack
in the 1970s, following the first implementation of Magnetic Flux detection tools are focused on the outer wall, e.g. Stress
Leakage (MFL) tools. Eddy current testing is primarily Corrosion Cracking. However, the growing interest in hydrogen
developed to accurately detect and size surface breaking defects. transportation and the potential for Hydrogen Induced Cracking
The technology is now mature and has been deployed in various (HIC) on the internal surface of pipelines, has focused the
applications, most notably for pipeline in-ditch inspection. As of industry on developing tools for internal crack detection.
today, the technology as a crack detection technique is not Detection and sizing of surface breaking defects are at the core
deployed commercially on ILI tools as it can only detect cracks of eddy current technologies.
on the inner pipe wall.
The most common use of the eddy current technology in ILI Keywords: Eddy Current, Stress, Geohazard, Materials,
tools is the simple configuration for lift-off measurement. This Internal Crack
offers a way to discriminate between internal and external
corrosion on a volumetric tool (MFL) or enhance caliper-based NOMENCLATURE
geometry measurements. The most advanced applications of ART Acoustic Resonance Technology
eddy current testing are targeting measurement of EDM Electrical Discharge Machining
electromagnetic properties of the pipe wall to either measure ERW Electric Resistance Welding
material properties (pipe grade) or pipe stress due to external ECA Eddy Current Array
loading. Both pipe grade or pipe stress eddy current technologies EMAT Electromagnetic Acoustic Transducer
can only operate successfully in combination with a FEA Finite Element Analysis
magnetically saturated pipe wall, either prior to or during the HIC Hydrogen Induced Cracking
measurement. The first objective of this development is to remove ILI Inline Inspection
the requirement for magnetic saturation, thus removing the need IMU Inertial Measurement Unit
for MFL magnets or large electromagnets. LO Lift-Off
The second objective is to progress the established Eddy MFL Magnetic Flux Leakage
current development to allow for the measurement of an SCC Stress Corrosion Cracking
additional parameter. Current commercial stress measurement SNR Signal to Noise Ratio
method is uniaxial and aligned in the axial/longitudinal direction TECA Tangential Eddy Current Array
with limited coverage around the pipe circumference. The new UT Ultrasonic Technology

1 © 2022 by ASME
1. INTRODUCTION weak magnetic field H and, therefore, the magnetostriction effect
Pipeline systems have been delivering energy across the is negligible in our magneto-elastic application.
globe for well over 50 years. To maintain safe operation, a
technology market was created to support ILI tools and has now
reached a mature status in terms of capabilities and reliability to
detect the most common threats such as corrosion or mechanical
damage. With a drive towards zero harm to people and the
environment, what may have been considered a secondary threat
has now become the new objective for pipeline operators and
regulators. For example, with ageing infrastructure, geohazard
type events like slow moving landslides have become a dominant
threat to a pipeline [1] and must be detected and their effects on
the pipeline accurately characterized.
Moreover, the energy industry is moving towards newer and
greener systems by introducing projects to convert existing
FIGURE 1: MAGNETO-ELASTIC COUPLING
pipelines used to transport petroleum and natural gas to transport
hydrogen or for CO2 capture and storage. To accomplish this
2.1 Stress Hysteresis
shift, the industry requires the repurposing of ageing pipeline
The general principle of stress measurement is relatively
networks. If we can imagine an operator having a vintage
well understood. However, the application in a commercial setup
pipeline, with incomplete or unclear pipe grade that they would
has been marginal due to inherent constraints in its deployment.
like to eventually utilize for hydrogen transport and with the
One of the main difficulties is how to interpret or compensate for
added internal cracking risk associated with such a pipeline.
material hysteresis response due to load. Load or stress
Even worse, the line runs through hilly terrain known for subtle
hysteresis is different to magnetic hysteresis under a variable
but consistent pipeline movement. Current MFL or UT based
magnetic field, for example, it is referred to as a BH curve [6].
technology will be of limited benefit in this scenario and new
Stress hysteresis refers to the difference in magnetic response
routes will need to be explored.
between a loading and unloading cycle on the same material at
Eddy current is a proven method to detect and characterize
the same location. The effect, hysteresis, is related to the energy
stress from geohazard [2], material properties [3] and provide an
loss in the magnetic domain movement responding to stress
answer for cracking [4].
excitation. Magnetic domains are regions within a magnetic
By leveraging eddy current technology, an operator could
material which have uniform magnetization. Domain structure is
gain additional information to help determining the suitability of
responsible for the magnetic behavior of ferromagnetic
injecting hydrogen into the line. Before they ever injected
materials.
hydrogen, they would know the material properties throughout
the line, areas of increased stress and therefore crack
susceptibility, and if there are any potential injurious features
(crack or crack-like) when converted to hydrogen.

2. STRESS & GEOHAZARD


Load measurement, also referred to as stress or strain
measurement, is a niche application of electromagnetic
measurement known for decades under the term of
Magnetoelastic effect (or Villari Effect [5] or Inverse FIGURE 2: STRESS HYSTERESIS PRINCIPLE, PRIOR
Magnetostrictive Effect). Summarized in Figure 1 below, the (RIGHT) AND AFTER (LEFT) DEMAGNETISATION
magneto-elastic coupling principle relies on the application of a
known magnetic field H over a ferromagnetic material (pipeline Figure 2 (Right) presents the idealized stress hysteresis
steel) where the local magnetization level M will reflect the curve for a load cycle in tension. One can see that for one value
stress level σ the material is experiencing. Based on elastic of stress there are two potential magnetic levels corresponding to
properties of the material and therefore, for measurement below loading in section 1 and unloading in section 2. Without prior
the yield strength, stress and strain are interchangeable knowledge of the stress history, one cannot determine which
parameters using the Elastic modulus E. Similar but opposite, the solution is correct. To overcome this issue, solutions were
magnetostriction is the generation of dimensional change or developed around magnetic shake up of the magnetic domain in
strain following the application of a large magnetic field. This is order to remove the load history bias. In practice, this often
the principle behind acoustic wave generation in an EMAT involves the use of a demagnetization process or the opposite by
sensor. In principle, the magnetic field H applied to measure operating near or at magnetic saturation. After demagnetization,
stress will also generate strain, but the effects are several orders the hysteresis collapses as is shown in Figure 2 (Left), producing
of magnitude smaller as the load measurement is conducted in a a unique curve for both loading and unloading conditions.

2 © 2022 by ASME
Stress measurement is currently available in ILI application differentiate between a lift-off effect and a stress effect. An
under the axial strain name and was developed by some of the example is presented in Figure 3 for one of the injected
authors over a decade ago. It will be referred to as the legacy frequencies. Using a Lissajous (real vs imaginary) plot of the raw
technology. To overcome the issue of stress hysteresis, the measurement, from an unstressed and no lift-off operating point
measurements are conducted behind the MFL tool. The large on steel (orange), an increase in stress (green) will shift our
magnet produces a saturation level sufficient to shake the operating point in a different direction than an increase in lift-off
magnetic domain and remove stress hysteresis. This can be (black). The angle, also referred to as phase, used when
grossly referred to as a demagnetization even if the material is in converting real and imaginary number into a magnitude
reality left with a fixed but low residual magnetic level. Having measurement allows one to decouple lift-off from stress when
the MFL tool in order to conduct stress measurement is a perfect orthogonality is achieved. However, perfect
considerable structural constraint for the ILI tool deployment. In orthogonality can only be achieved in a system using a single
a geohazard context where corrosion is not the target, a simpler frequency with a single coil arrangement. In a multi-frequency
tool, referred to as MFLess, similar to a combined geometry- approach injected in the same sensors, compromise must be
IMU tool but with improved performance for stress made to have near orthogonality across the frequencies range.
measurement, provides a valuable step forward in the pipeline
operator toolbox.
Extensive research was conducted to develop a solution that
removes the need for demagnetization (removal of MFL
magnetizer) for stress measurement during an ILI run. The
achieved solution is embodied in a transmit and receive eddy
current configuration, but also uses a new method for signal
injection combined with advanced data processing.
Using multi frequency injection, 2 or more, an array of
magnetic parameters, which reflect different magnetic
properties, can be measured at different skin depths. One can also
reconstruct the local BH curve to fully characterize the pipe steel
magnetic properties. The multi frequencies injection can be done
sequentially or simultaneously for higher data acquisition speed.
Multi-frequencies eddy current is sometimes used in standard
FIGURE 3: EDDY CURRENT LISSAJOUS FIGURE TO
eddy current testing for crack detection to compensate for
DEMONSTRATE PHASE SEPARATION BETWEEN STRESS AND
secondary effects such as temperature or material properties. The LIFT-OFF MEASUREMENT
idea was extended and developed further for the hysteresis
removal effect. The multi-frequency array of measurements is 2.3 Longitudinal and Hoop Stress Measurement Over
then fed through a proprietary analysis algorithm to combine the Full Circumference
frequencies and provide a unique sensor response, collapse As mentioned previously, the legacy technology to measure
hysteresis, as illustrated in Figure 2. pipeline strain using electromagnetics was unidirectional and
limited in the number of sensors around the circumference. The
2.2 Lift-off Compensation unidirectionality was removed by developing new directional
As well as the removal of the MFL requirement, further sensor geometry allowing a measurement in 2D, longitudinal and
development was conducted to enhance the technology. Eddy circumferential/hoop directions. In effect, two sets of sensors are
currents are prone to the lift-off effect, and this is therefore one present per probe. The new circumferential direction will allow
of their primary uses. However, for accurate stress measurement, for a better interpretation of the local effect (below 10 m or
the lift-off must be null or decoupled from stress measurement. around the circumference only) but also a new understanding of
The primary factor to reduce lift-off effect is to develop a sensor the internal pressure effect on the pipe hoop stress.
geometry producing a clear difference in signal due to lift-off The number of sensors in the legacy technology deployed
versus stress. This is referred to as phase shift, ideally by 90° or around the pipe circumference was limited due to the coil design.
orthogonality. The approach is used commercially in most high- The coils used are wound around a steel core to enhance the field
end eddy current crack detection systems, as presented later in strength due to increased permeability. This is often done in eddy
this paper. It was not available at the time of the design of the current sensors using steel or ferrite core. A disadvantage is
legacy stress measurement technology. Reducing the effects of adjacent sensors affect one another if they are moving in close
lift-off relied purely on the design and capabilities of the proximity to each other due to the metal inside the probe. The
deployment arms to always maintain the sensors in a well- new directional sensors are air core, the steel core was removed,
defined position to the pipe wall. The lift-off compensation and the sensor design was enhanced to produce stronger field
technique from the crack detection systems was transferred without the need for the steel core. This was the result of
across and adapted to the stress measurement system. By theoretical investigations combined with lab experimentation to
adjusting data processing parameters, one can clearly

3 © 2022 by ASME
produce the right combination of sensor size, number of turns, example, a geohazard type threat is expected to require lower
wire diameter and power dissipation based on the level of electric resolution than a girth weld cracking threat.
current injected. Air core sensors allow full circumferential 2.5 Geohazard Tool
coverage similar to a full array of MFL sensors. The technology will be known as ETEC (Eddy Current
Air core sensors have another benefit of being more Technology), developed in response to the geohazard threat to
consistent in their manufacturing process. In a steel core sensor, pipelines. The geohazard tool is based on 16 probes equally
the strength of the field will be linked to the magnetic spaced around the pipe circumference for a 24” pipe size. This
permeability of the core. Maintaining near identical steel configuration allows the detection of large scale geohazard
permeability over large production batches is a challenging task events (1m to several 100 m) producing mainly axially oriented
to accomplish, requiring complex steel qualification processes. load in the pipeline. To fully assess geohazard events, the tool is
designed on an ART platform. The ART tool offers caliper
measurement in gas or liquid media for high resolution pipeline
geometry (ovalities, dents, buckles, or wrinkles). The geohazard
tool is equipped as standard with an IMU to measure bending
strain (or movement in run-to-run analysis) and provide pipeline
mapping. The eddy current sensors, not requiring MFL
magnetizer, can be deployed in any tool combination (UT, ART,
MFL) or even standalone. However, the combined ART caliper,
IMU and eddy current setup is believed to offer beyond current
data to answer pipeline operator geohazard management.

FIGURE 4: EDDY CURRENT SENSORS TRANSMIT-


RECIEVE CONFIGURATION IN BOTH AXIAL AND HOOP
DIRECTIONS FIGURE 5: GEOHAZARD COMBO TOOLS USING EDDY
CURRENT(FRONT) / IMU(MIDDLE) / ART CALIPER (REAR)
2.4 High-Resolution Data
The last enhancement is related to new multi-frequencies 2.6 Technology Validation – Full Scale Testing
made possible with new electronics capabilities. The legacy The tool has undergone full scale testing from static load
stress measurement system was based on 5kHz alternating measurement to pump test in a water-filled test pipeline. The
current field. This limited the speed of data acquisition. For this challenge in stress measurement testing on a pipeline is to
application, the primary target was to remove stress hysteresis. generate a large stress level in a controlled and repeatable
To do that, we successfully used higher field frequencies, using manner, ideally in the longitudinal direction to match target
alternating current signals, a higher frequency allows higher data geohazards. The initial investigation pointed to similar testing
sampling, more cycles happen in the same amount of time. conducted during the development of the IMU tool to measure
However, in eddy current, the field penetration into the materials pipe bending. This includes large scale bending of a pipe section
is linked to the injected field frequency. Higher frequency means using custom hydraulic fixtures [7] or crane to lift a full pipe
lower skin depth. Tests showed that, below a certain limit of section [8]. The complexity of setting-up such testing led to a
several hundred hertz, stress measurements could be conducted refocus on the goal. The sensors, being a stress measurement
at higher frequency without compromising the data quality due device, and the direction of the stress itself is not primordial like
to skin depth effects. Measurements using over five times the in a bending strain scenario for IMU. Moreover, the high-
field frequency could produce measurable stress measurement resolution data and extended circumferential coverage do not
results as demonstrated later on in this paper. Even if the number require large scale stress events over 10 or 100 meters to obtain
of frequencies is now higher, simultaneous injections can be sufficient data points. Therefore, the test plan was to rescale
used to remove the impact. The stress measurement deliverable towards generating more local stresses similar to stress produced
is now moving from a legacy target of 1 point per meter at 5 m/s by an ovality on the pipe. The stress would mostly be oriented
tool speed down to under 25 mm for the same tool speed. For towards the circumferential direction. Using a hydraulic
low-speed lines (<1m/s), axial sampling can be collected with a workshop press of large capacity (45ton) as the basis of the test
couple of millimeter resolution. plan, additional fixtures were designed using FEA analysis. The
The resolution needed for full circumferential coverage or result is a loading frame capable of deforming the pipe section
axial spacing could be adjustable to provide the optimum up to or above yield stress. FEA results are presented in Figure
response to specific stress related threat requirements. For 6.
The selected pipe section was then strain gauged using
rosette gauges (0/45/90°) for validation of the calculated stress,

4 © 2022 by ASME
initially on the internal surface with two additional control gauge measurement. In real life applications, the onboard IMU
gauges on the external surface at 3 (90°) & 9 (270°) clock is used for pipeline mapping, features geo-localization and
positions. The internal gauges were used only for initial stress bending strain (or movement).
validation and load calibration and then removed to allow proper The validation campaign has therefore produced the
tool passage. The stress was then controlled using the outside following: a theoretical stress distribution from the FEA
gauges only and the pump pressure gauge. analysis, a local stress distribution using strain gauges and
measurement from the newly developed tool in static position
and dynamic results using pull testing and pump testing. Direct
comparison is possible and provides valuable information to
prove the technology and derive an initial performance
specification.

FIGURE 6: PIPE SECTION UNDER OVALIZATION LOAD

The setup was used to validate the stress measurement first


in a static position, inside the pipe at the loading frame location
and in line with the previously mounted strain gauge. Then the FIGURE 8: LOADING FRAME AND STRAIN GAUGE PIPE
tool was tested dynamically during pull testing. The tool was SECTION AS PART OF THE PUMP TEST PIPELINE SETUP
pulled along the inside length of the pipe using mechanical
winching power. Pull tests were repeated with data collected for However, it is important to note important caveats when
the full pipe length at incremental loading stages up to yield making these comparisons, for example, the FEA analysis
strength. An example of results captured by the tool is presented assumes a perfect pipe with homogenous material properties and
in Figure 7. The data was captured by the sensor at the 3 o'clock no seam weld. The strain gauge measurement represents only a
position and shows the incremental increase in loading stages. discrete location around the pipe and in-between the strain must
As the load increased, the stress profile shifted accordingly, be interpolated. Also, the loading frame and pipe section were
towards a higher compressive level. mounted and dismounted as part of the pull versus pump test
setup. The exact position was maintained as best as possible, but
error is prone to be introduced due to misalignment. Temperature
was monitored throughout the tests, especially during the pump
testing, as it was conducted outside where variations above 15°C
were monitored and assessed in terms of equivalent thermal
stress. Pressure was also controlled but the pump test was done
at internal pressure below 1 bar (14.5 psi), in effect producing
negligible hoop stress. However, the weight of water in the
partially supported pipe produced an added strain of about 100
microstrains as measured by the control strain gages compared
FIGURE 7: RAW SINGLE SENSOR MEASUREMENTS to the static and pull tests, which were conducted with an empty
ALONG PIPE SECTION AT INCREMENTAL STRESS VALUE pipe. The total experimental errors are therefore estimated to be
in the range of 50 MPa in a worst-case scenario.
Following the pull test, the instrumented pipe section and In Figure 9, the correlation between the different stress
loading frame were mounted in line on a 24” pipeline pump test measurements and tests is presented in terms of von Mises stress.
setup shown in Figure 8. The pipeline setup comprises of 60m of As a prerequisite, the grey section represents the pipe section
diverse pipe spools and two 3D bends back-to-back. The ILI tool under the loading plate. It is expected that local plastic
travels autonomously from the launch to receive trap in the deformation occurs there and the profile deviates from the FEA
water-filled pipeline using an industrial pump. analysis and over time. This is illustrated by the change in profile
The pump test consists of running the tool through the pipe in these sections between the initial static test and the pull/pump
section at various loading stages. The IMU onboard is used to test. Outside the loading plate, in the unrestraint and free moving
measure the tool rotation and to correlate with the initial strain pipe section where strain gauge data was collected, the

5 © 2022 by ASME
correlation is within changes in operating parameters described Furthermore, the ART caliper results illustrate the
above. One can derive an initial performance specification with difficulties in setting up and understanding full scale testing or
an anomaly detection threshold at 90% POD of 45 MPa and later-on real-life application. The FEA analysis provided
based on repeat measurements from multiple pull tests at the guideline of the load distribution around the pipe section, but
same stress level, an accuracy at 80% certainty of 20 MPa. The manufacturing effects, corrosion or previous loading history
profile and amplitude match the strain gauge data whereas the could not be predicted by the FEA. The strain gauges gave a
FEA grossly underestimates the effect. This is partially better estimation of the stress amplitude but only locally within
explained by the lower wall thickness from general corrosion its limits, for example in the 250° region where local variation
seen on the pipe surface compared to the nominal wall thickness may be important. The true stress effect on the pipe can only be
value used in the FEA analysis. assessed by a continuous measurement as provided by a
geometry tool when suitable deformation occurs, or a stress tool
as demonstrated here. Conceptually the combination of
geometry and stress measurement data would provide a more
comprehensive dataset to reflect pipeline condition.
The next step of the tool validation is planned on
commercially operated pipelines which have been subjected to
geohazard type events. Suitable pipelines have been selected
following discussions with operators to identify geohazard site
with extended knowledge on the location to test the technology.
A combination of stress amplitude, events with specific length
and also available information (strain gauges data for example)
make for a suitable candidate pipe section to inspect. Results will
be presented in a follow-up paper.
FIGURE 9: MEASUREMENT FROM FEA VS STRAIN
GAUGES VS ETEC SENSORS DURING THE DIFFERENT 3. MATERIAL PROPERTIES
VALIDATION PHASE (STATIC/PULL/PUMP TEST) Eddy current has been used commercially in various
industries to identify differences in steel properties, like hard
The ART caliper tool as seen in Figure 5 provides an extra spot during the manufacturing process [9]. The technology has
set of data validation and helps to explain some of the stress also been developed in combination with MFL type magnetizer
patterns. However, ART caliper inspections were only to identify different pipe grades during an ILI inspection [10].
conducted as part of the pump test and if plastic deformation Pipe grade identification is in effect a measurement of the
occurred locally in the static or pull test, the effect is not homogenous pipe materials properties along the length of a pipe
quantifiable using the ART caliper. The measured profile on the joint. The effect should be clearly segregated by the girth weld
unloaded pipe shows clear ovalization, offset by 10-20° location.
compared to the 3 and 9 o’clock positions. The profile also The same effect is measured when eddy current technology
presents a local flat spot at the seam weld location. While under is used for crack detection. The effect is compensated to produce
load, the 3 o'clock position presents a mostly uniform accurate sizing, an example in [4] where magnetic permeability
deformation along the length of the loading frame, the 9 o'clock compensation is applied to accurately detect and size SCC.
position has sharper local variation matching internal corrosion. Lastly, eddy current sensors are tuned for hard spot detection on
the pipe outside surface [11]. Hard spots are a local change in
material properties, as opposed to the global change across the
full pipe joint as is the case for pipe grade.

3.1 Material Properties Development


The eddy current technology developed here is sensitive to
steel properties and, as part of the normal analysis process, a pipe
type boxing is done to extract the stress component. Pipes are
grouped into different pipe types based on their outside diameter,
wall thickness, seam weld type, pipe tally information and, more
importantly, magnetic response. The required development is to
make it deployable in an ILI environment and to produce
FIGURE 10: ART CALIPER MEASURMENT BASELINE OF accurate material properties deliverables.
UNLOADED PIPE AND PROFILES AT MAX LOAD ALONG THE The stress sensors presented in the previous section were
LOADING PLATES (1 LINES PER SENSOR) designed with this application in mind but as a secondary
objective. Parallel testing was conducted to maintain a multi-
purpose sensor to deliver both stress measurement and materials

6 © 2022 by ASME
properties data utilizing the same tool. Another more interesting
application is to use the system to identify local variation in
properties in the pipeline. An example of this application is to
identify internal hard spots.
The main challenge is around the definition of what the
deliverable would be. A pipe grade is required by the newly
introduced Mega Rule in North America. However, pipe grade
is a way of grouping pipe materials with properties falling into
predefined ranges. But there is no one set of unique properties
matching a pipe grade. A pipe grade identification would be
derived from the measurements obtained.
In contrast, yield strength, chemical composition or
FIGURE 12: PUMP TEST MATERIALS IDENTIFICATION -
hardness value are measurable parameters which are directly BOXING PER PIPE POPULATION
related to magnetic properties. It is likely that a pipe grade
identification using eddy current will predict the grade based on 4. CRACKS
specifically measured parameters like yield strength. One such Cracks are a major threat to pipelines. On the external
example is presented in Figure 11. Hardness test using surface, cracks can take the form of SCC in pipeline body,
conventional mechanical indenter were conducted on a range of fatigue cracks or defects resulting from the manufacturing
available pipe and compared to predicted hardness based on the process in the welds. The industry has devoted a great amount of
magnetic response using the eddy current sensors. Using effort to address the detection and sizing of cracks in the external
adequate measured magnetic parameters, one can predict pipe surface of pipelines.
hardness with a satisfactory level of confidence. Hardness is a Migration toward green energy is accompanied by
well-known parameter that is related to ultimate tensile strength. conversion of gas pipelines to transport hydrogen or CO2. This
Both are measurable parameters which can help define a pipe has resulted in new risk or higher consequence of rupture, for
grade. example associated with hydrogen induced cracks on the internal
surface of the pipelines. Pre-existing internal cracks, if any,
would be required to be detected prior to injection of hydrogen
leading them to potentially grow into a serious risk for pipeline
integrity. Then, during operation of a hydrogen pipeline, early
detection and sizing of potential HIC would be crucial for the
operator to mitigate the risk as soon as identified.

4.1 Selected Approach and Technology


Following an internal technology review, eddy current was
identified as the most suitable approach to accurately detect and
size surface breaking crack on the internal surface of a pipeline
during ILI runs. A feasibility study was carried out to evaluate
the possibility of using eddy current array as a primary technique
FIGURE 11: HARDNESS MEASUREMENT - UNITY PLOT for the detection and characterization of any surface breaking
linear indications that may appear on the internal surfaces of
3.2 Material Properties Validation pipelines. The final solution had to meet the following
The tool validation presented in chapter 2.5 uses a pump test requirements:
pipe section built with different pipes sourced from pipes • Capability of detecting 50 mm long x 0.75 mm deep
recovered from operating pipelines, but also new sections or indications, based on longitudinal anomalies left in seam
even purpose-built sections. The results give an insight into the weld after line pipe fabrication
new probe identification capabilities as seen in Figure 12. The • Target flaw must still be detected at high lift off (up to 8
different pipe sections composing the pump test can be split into mm), to cope with ILI deployment potential condition
populations which each population showing different mean • Eddy current solution should provide a proper signal
levels. For example, the fabricated section for launch and receive signature for depth sizing
trap (#6 and #7) or the 3D bend (#4 and #5) shows clear • Target scan speed to match ILI tool speed in most lines
differences compared to the main line (#1 and #2).
Material identification is not only valuable in the context of Several solutions were evaluated during the feasibility
pipeline traceable records. Validating pipe grade in a pipeline is study, using different coil configurations, topologies, and
an important prerequisite for fitness for service assessment prior operating frequencies. However, one solution outclassed others
to converting the lines for products such as hydrogen or CO2. with regard to the requirement set above. This solution was based

7 © 2022 by ASME
on TECA configuration [12], an established technology fact that lift-off is limited for the SCC application by using hand
especially in the oil and gas industry to detect and size surface deployed conformable sensors, generally less than 1mm, small
cracks in carbon steel alloys. Although TECA was primarily coils worked for SCC. However, for internal cracks, possibly
developed for the inspection of carbon steel welds in components HIC, isolated cracks are expected rather than high-density
like pressure vessels, diverse types of tanks, pipes and piping, colonies, and this, with potentially high lift-off from ILI
wind towers, etc., along the road it was adapted to also tackle deployment means that going with larger sensors would be the
challenging problems for the pipeline industry, like SCC [13]. preferred embodiment.
One of the main reasons why TECA was identified as the The rescaling and development of the new TECA
preferred candidate resides in its very particular signal signature. configuration was applied first on a single element using iterative
Lift-off compensation is a key characteristic in eddy current phase with tests on reference sample. The testing was conducted
measurement as explained for the stress measurement on EDM notches ranging from 1 to 5 mm min depth and up to
application, this is also a crucial parameter for crack detection 25 mm in length. The notches were machined on carbon steel
and sizing. When correctly calibrated, crack signals present plates and pipe coupons. All the notches were well below the 50
unique characteristics: mm length target resolution. Parameters were isolated and their
• Horizontal and flat lift-off signal effects evaluated theoretically, but also tested on real defects in
• Crack-like features nearly orthogonal with lift-off order to produce the best response for the application. Once all
• For a given lift-off, crack-like indications have the same the parameters were perfectly optimized, i.e., coil size,
phase, only the amplitude changes with depth arrangement, operating frequency, etc., results obtained are
shown in Figure 14. The main observation is the perfect
orthogonality between flaw signals and lift off and this, for
multiple lift offs. As discussed before, this characteristic is key
to the possibility of providing accurate depth sizing with this new
TECA configuration in a future ILI tool.

FIGURE 13: TECA DEPTH AND LIFT-OFF SIGNALS


ORTHOGONALITY. 1, 2 AND 3 MM-DEEP CRACKS ARE SHOWN
IN GREEN, VIOLET, AND BLUE, RESPECTIVELY, AT MULTIPLE
LIFT-OFF. A DEPTH SIZING EXAMPLE IS SHOWN (LIFT-OFF =
0.4 MM, DEPTH = 1.2 MM). REPRODUCED FROM [4].

Any operating point in the impedance plane (Lissajous)


corresponds to a specific combination of a given crack depth
linked to a precise amount of lift-off measured during an
acquisition – the result is similar to applying a dynamic
compensation for lift-off on depth sizing as demonstrated in FIGURE 14: RESCALED TECA DEPTH AND LO SIGNALS
Figure 13. In addition to this, the TECA configuration is also ORTHOGONALITY. EDM NOTCH SCANNED AT MULTIPLE LO.
affected by material properties, as seen in previous materials
identification chapter, but allows compensation for property TECA technology is made up of an arrangement of three
changes, mainly for magnetic permeability, which significantly coils: one driver and two receivers. As previously discussed, the
impacts depth sizing accuracy. first receiver provides a signal for depth sizing (depth channel),
and in addition to that, a second receiver is used for length sizing
4.2 TECA Solution Development for ILI (length channel). This second receiver, having its axis oriented
Based on the feasibility results, it was decided that the best perpendicular to the surface inspected, is only sensitive to crack
approach to detect internal cracks in pipelines would be to ends. Thus, once encoded, length measurements can be
rescale a TECA element to better manage detection and depth accurately achieved from this channel. For the purposes of an ILI
sizing at high lift-off when sensors are running over seam or girth deployable system, a built-in lift-off was added to the design in
welds. With eddy current sensors design, there are always the form of a 2 mm offset from the bottom of the receiver coils.
tradeoffs to make between several parameters to achieve This represents the sensor wear plate. It is expected that the
optimum results. As an example, increasing coils size helps to version for the ILI tool will use similar deployment mechanism
tolerate more lift-off, but the main drawbacks would result in and probe envelope to the stress and materials sensors. It will
losing resolution and sensitivity over small flaws. For SCC on also use a ceramic shoe as a wear plate to protect the sensor as it
the external surface of a pipeline, large coils would not be will have to be in contact with the pipe wall for the majority of
advisable because of colonies of high-density cracks. Given the the inspection.

8 © 2022 by ASME
The single sensor testing showed particularly good internal surface, the pipe sections were inspected using a portable
detection capabilities at high lift-off, resulting in high SNR. The UT system. All three coupons had defects but were mostly not
length of the channel is a bit more sensitive than the depth surface breaking defects or only over a few millimeters in length.
channel on the prototype meaning an initial better capability in It was therefore decided to try to open the defects to make them
detection. Results of tests carried out on 25 mm-long x 1 mm– fully surface breaking on the inner. A hydraulic press under
deep EDM are shown in Figure 15. At lift-off of 2 mm, which controlled conditions was used to open the pipe curvature
corresponds to the ceramic scraper, the new TECA exhibited slowly, resulting in creating large stress and breaking open the
high SNR of 43 and 35 dB for length and depth channels, defects.
respectively. At lift-off of 8 mm, SNR of length and depth Each plate was loaded using this technique, the resulting
decreased to 32 and 29 dB, allowing detection of the target defect crack length and size were purely a product of the initial internal
at the max targeted lift-off. defect subjected to accelerated growth rate resulting from
external loading. The sections were first scanned using
commercial eddy current systems for in-ditch inspection to size
the cracks. The cracks were 0.8 mm-D x 29.4 mm-L, 0.4 mm-D
x 20.5 mm-L, 1.7 mm-D x 31.5 mm-L.

Depth C-scan Length C-scan

FIGURE 15: TECA LIFT-OFF TOLERANCE OF LENGTH


AND DEPTH CHANNELS Depth signal Length signal

To conclude the study, an array prototype was designed and


manufactured using the optimum design parameters of the newly
developed single-element TECA. This array prototype
comprised five TECA elements, each has its own driver, depth,
and length receivers. This array prototype had a scraper thickness (a)
of 2 mm as a provisional wear plate to simulate the ILI probe. To
provide length measurement, it was also fitted with encoder
Depth C-scan Length C-scan
whereas the ILI configuration will rely on tool odometers. Using
five elements enables the possibility of displaying C-Scans, a
powerful view where eddy current signals from each individual
channel in the array are all displayed side-by-side on a single
map. Color changes on C-Scans can be interpreted as impedance
variations.
The single element and the following array prototype were
hand deployed or used in combination with an XY scanner in Depth signal Length signal

order to cover the full surface under inspection. It is expected


that in an ILI application a series of those five elements array
will be deployed around the pipe circumference in order to obtain
full coverage. For the testing purpose, a smaller sized array is
sufficient when combined with an appropriate scanning device (b)
to reconstitute the full C-scan. FIGURE 16: A-SCAN USING THE NEW TECA ARRAY
PROBE AT LIFT-OFF EQUALS (a) 2 MM AND (b) 7 MM.
4.3 Internal crack detection validation All these cracks were successfully detected using the newly
To validate the performance of this array prototype, an developed TECA probe at lift-off as high as 7 mm. Two of these
extracted pipe section from a pipeline with internal real cracks real cracks were smaller than the target defect, hence, detecting
was needed. The pipe was a 16” diameter, 8.7 mm wall thickness them proves the ability of the new sensor arrangement to detect
ERW (trimmed) section. The selected defects were all reported the target's internal cracks. The detection scans of the smallest
to be on the internal ERW seam weld, not in the base metal. two defects at lift-off of 2 and 7 mm are shown in Figures 16(a)
Defects were likely to be due to lack of fusion. After cutting the and (b), respectively. As shown in these figures, despite the
3 coupon sections from the main spool for ease of testing on the rusted internal surface of the pipe, the depth and length signals

9 © 2022 by ASME
had high SNR. The peaks and troughs of the length signals [2] Westwood, S., Jungwirth, D., Nickle, R., Dewar, D.,
indicate the beginning and end of the cracks. Therefore, they are Martens, M., “In line inspection of geotechnical hazards”,
used to determine the length of the cracks (i.e. 29.3, 18.3 and Proceedings of International Pipeline Conference, IPC 2014-
29.5 mm). 33245.
The sensors have been designed with the orthogonality [3] Molenda, D., Thale, W.., Reining, B., “Method for the
between the depth and lift-off signal required for accurate depth contactless determination of a mechanical-technological
measurement. However, depth measuring algorithm is still characteristic variable of ferromagnetic metals, and also
required to be developed to allow depth sizing and lift-off apparatus for said method”, United States Patent Application
compensation for this new coil's arrangement. This is to be Publication, US 2016/0231277A1.
conducted in the subsequent phase of development, but this time [4] Sirois, M., Bouchard, M.., Sweedy, A., “Advanced eddy
a larger number of defects will be required than that used for current array tools for stress corrosion cracking direct
detection validation. In parallel, development is ongoing in terms assessment on pipelines”, Proceedings of International Pipeline
of integration to an ILI tool. The similarity between both the Conference, IPC 2020-9335
stress/materials sensors and crack sensors are used to transfer [5] Villari, E., "Change of magnetization by tension and by
deployment knowledge from one technology to another. The electric current", Ann. Phys. Chem., 126 (1865), 87-122.
next phase of development is to build a mock-up tool to be [6] Nestleroth, J. B., Crouch, A. E., “Variation of Magnetic
deployed at speed in pull test scenarios. Results will be presented Properties in Pipeline Steels”, Report No. DTRS56-96-C-0010,
in a follow-up paper. Subtask 1.1 Report, to the U. S. Department of Transportation,
March 1998.
5. CONCLUSION [7] Czyz, J., Fraccaroli, C., Sergeant, A., “Measuring
The research presented has been the result of several years pipeline movement in geotechnically unstable areas using an
of development projects to evolve and enhance a technology inertial geometry pipeline inspection pig”, Proceedings of
portfolio in order to help operator to better respond to new and International Pipeline Conference, IPC 1996-1839.
growing pipeline integrity threats. By using existing knowledge [8] Russell, D., Latimer, R., “Development of a pig based
developed in other industries as well as developing innovative inspection tool utilizing maps stress measurement technology”,
approaches, a suite of inspection techniques was developed on PPSA Seminar, 2007.
what may have been considered dormant technology, eddy [9] Wolter, B., Gabi, Y., Conrad, C., “Nondestructive
current, for ILI inspection. We have shown that, using the same Testing with 3MA - An Overview of Principles and
core technology but developing innovative solutions and Applications”, Applied Science, March 2019.
opportunities, it is possible to capture new datasets during ILI [10] Eiken, T., Thale, W., “A Novel Non-Destructive
inspection of pipelines: Technology for Pipe Grade Determination and MAOP/Design
1. Stress measurement in multi-directions without the Pressure Validation of Operating Pipelines”, 19th World
additional overhead of requiring MFL to be part of the Conference on Non-Destructive Testing 2016
tool. [11] Sirois, M., “Pipeline Hard Spots: Eddyfi Technologies
2. Materials identification is possible again without the Presents Spyne™, the Proven Solution for In-Ditch Detection”,
need for the MFL tool. e-Journal of Nondestructive Testing (NDT) ISSN 1435-4934,
3. Internal Crack detection and sizing capable of being February 2021.
deployed on ILI tool [12] Sirois, M., Parmentier, S.., Grenier, M., Decourcelle,
Moreover, all those approaches have the potential to be used N., “Eddy current array probe with independent transmitters”,
as a standalone or combined with existing technology, UT, or United States Patent, US 10,794,864 B2.
ART for example. In an environment where ageing pipelines [13] Sirois M., Bouchard, M., “Recent Advances in Depth
repurposed to transport new products (hydrogen, or CO2) is set Assessment of Stress Corrosion Cracking Using Tangential
to become a trend in the next decades, having all the necessary Eddy Current Array on Carbon Steel Pipelines” AMPP Annual
datasets to make informed decisions is key. The proposed Conference & Expo, C2022-17893.
technology will be able to provide this added set of information
to the operator.

ACKNOWLEDGEMENTS
The authors would like to thank the efforts of the
development and testing teams who contributed to this project
across the wider organization.

REFERENCES
[1] “Guidelines for management of landslide hazards for
pipelines”, The INGAA Foundation and a Group of Sponsors,
August 17, 2020.

10 © 2022 by ASME

You might also like