RECENT IMPROVEMENTS REGARDING ULTRASONIC

CRACK INSPECTION OF PIPELINES

Herbert Willems, Thomas Hennig
NDT Global, Stutensee, Germany

ABSTRACT
Crack inspection of pipelines using conventional ultrasonic technology has become a standard
application for in-line inspection (ILI) of liquid pipelines. Crack inspection tools have proven very
successful for the detection of various types of cracks (e.g. SCC) or crack-like anomalies present in
many pipelines worldwide. The first inspection tools were developed for axial crack inspection (UC), as
most cracks or crack-like defects in pipelines are axially orientated. In some cases, however,
circumferential cracking can occur prompting the development of tools for circumferential crack
inspection (UCc). Standard crack inspection tools can be applied in most liquid pipelines transporting
typical crude oils or products (e.g. diesel).

Over the years, specific inspection requirements came up that were not covered by the first tool
generations. These requirements are related to different aspects of the inspection process ranging from
tool-related characteristics to inspection-related challenges such as crack inspection in liquid gas.
Consequently, those challenges are addressed by the latest tool developments allowing an inspection
performance not possible before with regard to inspection speed and measuring resolution. In the paper,
the achieved progress including enhanced depth sizing is described and illustrated by examples from
inspection runs.

1 INTRODUCTION
The first ultrasonic inspection tools for inline crack detection using the 45° shear wave technique were
developed in the early nineties and their commercial application started in 1994 [1]. The development
was driven by the increasing demand for an alternative to hydrotesting as a means of proving the
integrity of a pipeline. The minimum crack size to be detected was determined from fracture mechanics
calculations. As a result, a minimum length of 30 mm and a minimum depth of 1 mm were defined
ensuring a sufficient safety margin with regard to critical crack sizes. Inline crack inspection proved to
be quite successful over the years and has become one of the standard applications for ILI. Nowadays,
several vendors are providing ultrasonic crack inspection tools for a wide range of pipelines. As an
example, NDT Global offers ILI tools for axial crack inspection as well as circumferential crack inspection
covering all relevant sizes from 6" upwards with an inspection track record of more than 100,000 km as
per today (starting 2003). These tools use conventional, piezoelectric sensors which limits their
application to liquid pipelines. Gas pipelines can be inspected, too, by using a liquid batch, which
however requires considerable additional efforts including a shut-down of the line.

2 INSPECTION METHOD
The main target of inline crack inspection is the reliable detection of surface-breaking cracks or crack-
like anomalies with predominantly radial orientation. The detection limit was initially defined by a
minimum length of 30 mm and a minimum depth of 1 mm where the limit size typically refers to a
probability of detection (POD) of 90 %. Secondly, precise sizing of the crack dimensions (length and
depth) is equally important in order to provide suitable input data for crack assessment. As a viable
solution complying with the restrictive conditions of inline inspection, the well-known 45° shear wave
method [2] was chosen when the first crack inspection tools were developed [1]. Even though this
method has some inherent limitations regarding depth sizing [3], it is still the standard method applied
with current ILI tools for crack inspection in liquid pipelines.

2.1 Principle
The principle of the 45° shear wave method is explained in Fig. 1. A piezoelectric transducer generates
a longitudinal ultrasonic wave (center frequency  4 MHz) which propagates through the liquid coupling
medium into the pipe wall. The angle of incidence in the medium is selected such that a refracted shear
wave is obtained propagating through the wall at an angle of approx. 45°. Using water as a couplant,

Copyright © 2017 Pigging Products & Services Association

 PPSA Seminar 2017

the angle of incidence is then approx. 18°. If the pulse hits a radial crack a strong reflection is obtained
(corner reflection) that is received by the same sensor (pulse-echo method). Depending on the time-of-
flight of the crack signal relative to the surface signal one can readily determine whether the crack is
internal or external. The received signal is displayed as an A-scan showing the measured reflection
amplitudes as a function of time-of-flight or distance (Fig. 1).

Figure 1: Inspection geometry and A-scan signal (schematic)

The amplitude of the crack signal depends on the propagation angle of the refracted shear wave. The
path of the reflected signal includes the refraction from the liquid medium into the wall and vice versa as
well as the reflection at the backwall and the reflection at the crack. This sound path is depicted in Fig. 2a
together with the result of the calculated reflection amplitude for a (deep) crack as a function of the
refraction angle β (Fig. 2b). The calculation is based on the plane wave approximation using formulas
given in [2]. Fig. 2b also shows the result of modelling using the finite difference (FD) method. The
agreement between both calculations is quite good. Some deviations become noticeable for refraction
angles above 60° (dashed line) where the plane wave assumption is limited due to the finite wall
thickness.

a) Inspection geometry b) Angular dependency of reflection signal

α – angle of incidence in water

β – refraction angle in steel

Figure 2: a) Sound path of crack signal b) calculated signal amplitude as a function of refraction angle β for
shear wave in steel (blue line: plane wave calculation; red dots: modelling using the finite difference (FD) method)

As can be seen from Fig. 2b, the amplitude of the reflection signal is rather constant within an angle
range from approx. 40° to 50°. The amplitude dips around 30° and 60° are caused by the mode
conversion from shear wave to longitudinal wave taking place at β  30° at the backwall and around
β  60° (= 90° - 30°) at the crack. Inspection angles outside the angle range indicated in Fig. 2b by the
green shading have to be avoided.

3-2
 PPSA Seminar 2017

2.2 Pipeline Medium

Most liquids transported in pipelines can be used as coupling medium provided that the liquid is free of
amounts of gas that might impede ultrasonic propagation. These liquids include most crude oils,
products and even liquid gases (e.g. propane or butane). The ultrasonic properties (ultrasonic velocity
& attenuation) of the medium need to be known prior to the inspection in order to ensure optimum
inspection settings. The ultrasonic velocity in the pipe wall is considered to be constant which is in
particular true for ferritic pipeline steels (the velocity of shear waves is normally in the range from
3200 m/s to 3260 m/s). In a standard sensor carrier the angle of incidence is fixed by the mechanical
design of the sensor holders. The standard angle of incidence for crack inspection in crude oils is 17°,
which can be used for medium speeds from approx. 1200 m/s to 1500 m/s. In contrast, crack inspection
e.g. in propane would require an incidence angle of 11° (based on a velocity of 850 m/s) and thus a
modified sensor carrier has to be applied. Especially for crack inspection in liquid gases, the dependence
of the ultrasonic speed on temperature and pressure needs to be taken into account in order to ensure
proper inspection conditions (see section 3.3).

3 RECENT IMPROVEMENTS
The progress of in-line inspection tools is strongly correlated to the progress in electronics, data
processing and data storage capacity. Highly integrated electronic components allow for compact size
and reduced power consumption at the same time. As a result, an increased number of sensor channels
can be accommodated in less space which is in particular important for small diameter tools. In general,
the corresponding improvements regarding the quality of the inspection data are:

• Enhanced axial resolution

• Enhanced amplitude resolution and time resolution
• More sensor channels thus providing better circumferential solution
• More parallel processing of receiving channels enabling higher inspection speed

It should be mentioned that also improvements of ultrasonic transducers have taken place over the last
two decades. In particular, composite transducers are nowadays applied, which offer an increase of
sensitivity by typically more than 10 dB compared to standard piezo-electric transducers when used in
a liquid environment.

3.1 Resolution
In general, the resolution of an ultrasonic ILI tool can be described by the following four components:
1. Axial resolution: The axial resolution is defined by the axial distance between two consecutive
measurements of the ultrasonic sensors.
2. Circumferential resolution: The circumferential resolution is defined by the circumferential
distance between two adjacent ultrasonic sensors. Both, axial & circumferential resolution
determine the scanning grid.
3. Sampling frequency of ADC: The sampling frequency determines the resolution of the time-of-
flight measurement of ultrasonic indications as well as the maximum amplitude error of the peak
amplitude measurement.
4. Sampling depth of ADC: The sampling depth determines the resolution of the amplitude
measurement of ultrasonic indications. It also relates to the dynamic amplitude range that can
be covered.

Fig. 3 shows the scanning grid for the standard crack inspection UC (Fig. 3a) and for the latest high-
resolution version UCx (Fig.3b). Here, the scanning grid is defined by the axial resolution and the
circumferential resolution. The ultrasonic shot density is by a factor of four higher for the UCx inspection,
which in turn increases the data volume by the same factor. The improvement of the circumferential
resolution can also be recognized from the increased sensor density (Fig. 3c,d). Doubling the number
of sensors requires considerable efforts on the construction side as the amount of cabling and
transmitting channels needs to be doubled as well.

3-3
 PPSA Seminar 2017

a) Scanning grid UC (3 mm x 10 mm) b) Scanning grid UCx (1.5 mm x 5 mm)

c) Example UC sensor carrier d) Example UCx sensor carrier

Figure 3: Scanning grid for UC and UCx and corresponding examples of sensor carriers

The higher circumferential resolution also provides a more homogeneous sensitivity distribution over
the pipe circumference as illustrated in Fig. 4. Here, the maximum sensitivity drop is reduced from - 6 dB
(UC) to approx. -2 dB (UCx). As a result, the measuring error regarding the maximum reflection
amplitude from a crack is reduced.

a) UC

b) UCx

Figure 4: Improvement of circumferential sensitivity for axial crack inspection

The advantage of the increased sensor density is demonstrated in Fig. 5 showing B-scans from a crack-
like anomaly located at the longitudinal weld as recorded during two different inspections. Using
standard resolution UC the anomaly is picked up by three sensors while the same anomaly is detected
by six sensors when the UCx resolution is used. The improved resolution not only ensures a higher POD

3-4
 PPSA Seminar 2017

but also provides more detailed information allowing for better signal classification as well as for more
accurate sizing.

Figure 5: B-scans from two different inspections showing indications from a crack-like reflector located at
the longidudinal weld: run 1 (UC) – 3 sensors; run 2 (UCx) – 6 sensors

Fig. 6 shows a result from a special inspection that was carried out in a gas pipeline by running the tool
in a water batch. Here, the inspection task was to find in particular very short stress corrosion cracking
(SCC) located near the girth weld. The minimum crack length to be detected was 20 mm, which was
achieved by appropriate settings of the detection criteria together with an axial resolution of 1.5 mm.
The example shows a single crack starting next to the girth weld. Here, the verified length was 20 mm
and the depth was already 5.2 mm.

a) .Axial SCC at girth weld b) B-scan showing crack indication

Figure 6: Axial SCC near girth weld (length: 20 mm, depth: 5.2 mm) and corresponding B-scan.

Table 1 summarizes some characteristics indicating the improvement of the inspection resolution from
the first tool generation to the latest tool generation.

3-5
 PPSA Seminar 2017

Table 1: Improvement of resolution characteristics for crack inspection

Variable First Generation Latest Generation
Axial resolution (mm) 3 (1.5 opt.) 1.5 (0.75 opt.)
Circ. resolution (mm) 10 5
Sampling frequency (MHz) 40 80
Time resolution (ns) 25 12.5
Max. amplitude error (dB) 1 0.2
Sampling depth (bit) 8 14
Dynamic range (dB) 60* 78**
*using logarithmic amplifier **using one bit for sign

3.2 Inspection Speed

The relationship between axial resolution aR and the maximum inspection speed that is acceptable to
ensure a specified axial resolution is given by the relation
vmax = aR / (Nm * tus)
with:
vmax – max inspection speed at given resolution a R
Nm – no. of multiplexed ultrasonic receiving channels / unit,
tus – time window for recording a single A-scan.
As the time window to record the ultrasonic signal is more or less fixed (typically 60 µs to 100 µs), the
main option to increase the maximum inspection speed for a given axial resolution is to parallelize the
data processing on the receiving side; i.e., to reduce the number N m of multiplexed channels. The
corresponding design is implemented in NDT Global’s new EVO Series 1.0 electronics. Compared to
the old design the new design allows for an increase of inspection speed by a factor of up to four
depending on the total number of sensors used. Consequently, the high resolution tools can now be
operated at inspection speeds that in most cases do not require the pipeline operator to reduce the
pumping speed and thus the throughput of his pipeline.

Table 2: Maximum inspection speed obtained with EVO Series 1.0 as depending on axial resolution
for axial crack inspection UC and UCx and circumferential crack inspection (UCc); old values shown in
brackets.

Type of Inspection Axial Resolution (mm) Max. Inspection Speed (m/s)

Axial cracks (UC) 3.0 4.0 (1.6)

1.5 2.0 (0.8)
Axial cracks (UCx) 3.0 2.0
1.5 1.0
Circ. cracks (UCc) 1.5 2.0 (0.8)

3.3 Medium Properties

Ultrasonic ILI requires a liquid couplant, which is usually provided by the pipeline medium itself (crude
oil, products etc.). The ultrasonic properties of the medium (ultrasonic velocity, ultrasonic attenuation)
need to be known prior to inspection in order to ensure optimized tool settings. During an inspection,
these properties used to be treated as constant in the past although it is known that there is some
temperature & pressure dependency. However, knowing these dependencies allows for reducing
uncertainties in the inspection data otherwise ignored.

3-6
 PPSA Seminar 2017

Figure 7: Change of medium attenuation during inspection run (blue line); the resulting amplitude
change is compensated by proper gain adjustment (red line).
In order to monitor the medium properties during an inspection run the new generation of inspection
tools are equipped with reference sensors. These sensors together with attached reference reflectors
are immersed into the liquid medium allowing for a continuous recording of its ultrasonic velocity and
attenuation. Based on this data, the saturation amplitude used for crack depth sizing can be corrected,
if necessary. An example is shown in Fig. 7. In this case, the medium attenuation has changed during
the inspection by an (unusual) amount of approx. 12 dB caused by larger temperature and pressure
variations. The change in attenuation was compensated in real time by an automated gain adjustment
(see Fig. 7). One of the benefits of this improvement is the fact that the same saturation amplitude
required for the depth sizing of crack-like indications is readily available for data analysis without further
post-processing of the data. Furthermore, the automatic gain control prevents the risk of degraded data
quality due to a loss of inspection sensitivity caused by increasing medium attenuation during an
inspection.

3.4 Enhanced Depth Sizing

The determination of the crack depth from ILI data mainly relies on the recorded amplitudes from the
corner reflection (see Fig. 1). The typical amplitude behavior of the corner reflection as a function of
crack depth is shown in Fig. 8 where the inspection geometry is similar to the situation illustrated in
Fig.1, i.e., using EDM notches in a plate and water as coupling medium.

Figure 8: Amplitude of corner reflection as a function of crack depth for external cracks (EDM notches
in plate with 10 mm wall thickness; transducer diameter 15 mm)

3-7
 PPSA Seminar 2017

The measured dependency is very well reproduced by the 2D - FD modelling results also shown in
Fig. 8. In the depth range between 1 mm and approx. 4 mm an amplitude change of about 10 dB is
noticed. For depths larger than 4 mm the amplitude of the reflection signal levels off approaching a
saturation level. The actual shape of such an amplitude vs. depth relation depends on properties of the
ultrasonic probe (e.g. diameter) as well as on diameter and wall thickness of the pipe. In order to size
crack indications detected by ILI, the saturation level and the actual depth dependency need to be
known. Then, a depth can be recalculated from the measured amplitudes at least in the depth-sensitive
range below 4 mm. As the deeper cracks, however, are most relevant for safe pipeline operation, it is
of utmost importance that such cracks shall be detected with highest priority by in-line crack inspection.
One of the characteristics of ultrasonic pulse-echo inspection that is in particular sensitive to deeper
cracks is related to the reflection taking place via the wall side opposite to the crack. In contrast to the
direct corner reflection, we refer to this echo in the following by the acronym ICE (indirect crack echo).

The approach of using this type of signal is explained in Fig. 9 assuming an external crack. If the crack
is located at the half skip distance, the usual corner echo is obtained (see Fig. 1). The ICE signal
becomes visible at the full skip distance when the crack is extending into the sound beam (Fig. 9a).

a) Principle of indirect crack echo (ICE) b) Signal amplitude vs. crack depth (modelling
results, wt = 10 mm)

Figure 9: a) Explanation of the indirect crack echo (ICE) b) ICE amplitude as a function of crack depth
(echo amplitude from corner reflection for comparison)

The modelling results for the amplitude of this signal are shown in Fig. 9b using a wt of 10 mm and a
probe diameter of 15 mm. At a sensitivity level of approx. - 20 dB, the ICE becomes visible for depths
above approx. 4 mm. The amplitude of the corner reflection is shown for comparison. While the corner
reflection is sensitive in the depth range below 4 mm, the ICE is sensitive over the depth range from
4 mm to 10 mm thus covering in particular the range where the corner echo is saturated. It should be
mentioned that the practical applicability of the ICE is subject to certain constraints regarding transducer
diameter D, wall thickness wt and a sensitivity related minimum crack depth d min.

3-8
 PPSA Seminar 2017

a) former depth sizing specification b) enhanced depth sizing specification

Figure 10: Old (a) and new (b) specification range for crack depth sizing

By exploiting the amplitude of the ICE signal, the range of crack depth sizing can be extended over the
full wall thickness (wt) where the covered wt range is mainly depending on the probe diameter. This new
approach was verified by modelling studies as well as by comprehensive experimental work including a
variety of different pipe diameters and wall thicknesses [3]. As a result, a tolerance of ± 1.3 mm at a
certainty of 80 % was determined (Fig. 10b). Compared to the old sizing specification (Fig. 10a), where
the depth range is limited to the saturation depth of approx. 4 mm, the enhanced sizing approach
represents a major step forward regarding the reliability of inline crack inspection.
The enhanced sizing was applied using inspection data from a blind test performed earlier [4]. The test
pipeline contained a large set of fatigue cracks which were generated at the circumferential side of
welded-on anode pads. Such anode pads are used for corrosion protection e.g. in offshore flowlines.
The test pipeline was inspected by a tethered tool equipped with standard UCc technology for crack
detection and with a TOFD unit for precise crack depth measurement [4]. The true depth of the fatigue
cracks was verified afterwards by destructive examination. A typical example of this type of fatigue crack
is depicted in Fig. 11a showing the surface indication as well as the cross section of a 62 % deep crack
as determined by destructive testing. The ultrasonic B-scan recorded for this crack including the
amplitude dynamics of the reflection signals are illustrated in Fig. 11b.

a) Example of circumferential fatigue crack at b) Amplitude dynamics (top) and B-scan (bottom)
the fillet weld of an anode pad showing ICE signal in the center

Figure 11: Example of fatigue crack showing surface indication and cross section (a) and ultrasonic B-scan (b)
with corner echoes and ICE indication in the center

3-9
 PPSA Seminar 2017

The ICE signal could be identified for 13 cracks (out of 14) having a depth above approx. 30 % wt at a
wall thickness of 12.9 mm. One crack with a verified depth of 36 % wt did not exhibit an ICE signal. The
reason was probably that this crack had an intermittent shape where the length of the deeper section
was below the specified minimum length of 20 mm. Based on the modelling result presented in Fig. 9b,
the crack depths were determined from the maximum amplitude of the ICE signal. The results are shown
in Fig. 12 together with the TOFD results. All the depth values determined by this procedure are lying
within a tolerance band of ± 10% wt, which is in agreement with the specified tolerance of ± 1.3 mm (see
Fig. 10b).

Figure 12: Unity plot showing results of crack depth sizing using the ICE amplitudes. Results from
earlier TOFD measurements [4] are shown for comparison.

4 PERFORMANCE SPECIFICATION
The performance specification for ultrasonic crack inspection is indicated in Table 3 comparing some
actual characteristics with the preceding ones. In particular, the minimum crack length for UC inspection
has been reduced and the enhanced depth sizing has been introduced (see section 3.4).

3-10
 PPSA Seminar 2017

Table 3: Excerpt of performance specification for crack inspection (old values in brackets)

Axial Inspection (UC) Circ. Inspection (UCc)

Min. Crack Length* @ 90 % POD

• Axial Resolution 1.5 mm 20 mm (30 mm) 30 mm
• Axial Resolution 3.0 mm 25 mm (30 mm) /

Min. Crack Depth @ 90 % POD

• Base Material & at Weld 1.0 mm
• in Weld 2.0 mm

Length Sizing @ 90 % Certainty ± 10 mm ± 15 mm

Depth Sizing @ 80 % Certainty

• Depth range 1 mm – 4 mm ± 1.0 mm
• Depth range > 4 mm ± 1.3 mm (new)

*smaller length optional

5 SUMMARY
Ultrasonic crack detection has become one of the standard applications for the inspection of liquid
pipelines. Today, inspection tools are available that allow for the reliable inspection of most crack issues
present in pipelines including in particular axial cracking but also circumferential cracking.

Although the inspection technology itself, which is based on using 45° shear waves, is still the same as
used in the first tool generation, the technical progress achieved by a variety of new developments is
remarkable from many perspectives. The main improvements coming with the latest crack tool
generation and the related benefits are summarized in Table 4. Further improvements are already
available today by combining more than one inspection technology in one tool (e.g. crack inspection &
metal loss inspection). Future progress of inline crack inspection will address the implementation of
additional testing modes to reduce current uncertainties that may affect POD, POI and sizing.

3-11
 PPSA Seminar 2017

Table 4: Latest technical improvements of inline crack inspection and related benefits

Improvement Benefit

Enhanced axial resolution • More detailed crack profiles

• Smaller minimum crack length
Enhanced circumferential resolution • Increased POD & POI
• Reduced risk of incomplete coverage
• More accurate maximum reflection amplitude
Higher signal dynamics • Wider range of medium attenuation
• Better sensitivity & signal quality
Increased inspection speed • Reduced costs by avoiding loss of throughput during
inspection run
• Less operational interference
Enhanced depth sizing • Full wall coverage of crack depths
• More accurate and less conservative crack
assessment
• Reduction of excavation costs
Online monitoring of medium • Reduced risk of failed run due to change of medium
properties properties during inspection
• Better data quality by adaptive signal gain

6 ACKNOWLEDGMENT
We thank our colleagues Rogelio Guajardo for providing inspection examples and Dr. Gerhard Kopp for
performing the ultrasonic modelling calculations.

7 REFERENCES
1. H. Willems, O.A. Barbian, "Ultrasonic Crack Detection in Pipelines by Advanced Intelligent Pigging",
Proceedings of the 2. International Conference on Pipeline Technology, Ostend (Belgium),
September 11-14, 1995.
2. J. Krautkrämer, H. Krautkrämer, Werkstoffprüfung mit Ultraschall, 5th ed., Springer-Verlag Berlin
Heidelberg, 1986.
3. H. Willems, G. Kopp, V. Haro, "Sizing Crack Indications from Ultrasonic ILI: Challenges and
Options", 12th Pipeline Technology Conference, Berlin, May 2-4, 2017.
4. T.S. Kristiansen, H. Willems, H.P. Bjørgen, G. Wieme, "Ultrasonic Tethered Tool for Inspection of
Weld Cracks in Offshore Pipelines: Tool Qualification and First Inspection Results", 10th Pipeline
Technology Conference, Berlin, June 8-10, 2015.

3-12