Pipeline Technology Conference 2021, Berlin

A New Phased Array Sensor for Pipeline Inspection – Optimization and Quantitative
Performance Evaluation

Martin Spies, Olaf Müller, Irina Lachtchouk, Martin Tschuch

Baker Hughes, Process & Pipeline Services, Lorenzstraße 10, 76297 Stutensee, Germany

Abstract

The phased array technique allows for a flexible adaption of the ultrasonic inspection
techniques for a large range of applications. By using it in the area of pipeline inspection
various inspection modes can be performed simultaneously. That was the reason for Baker
Hughes (previously PII Pipeline Solutions) to develop an inspection tool based on the phased
array technique which is successfully operated since more than a decade. In order to make
optimal use of the various testing modalities supplied by the phased array technique,
especially in view of the identification of different defect morphologies, a new generation of
phased array transducers has been designed in the course of simulation-based investigations.

In this contribution, we report on the different optimization objectives, which served as input
for the simulations and which have been accounted for accordingly. The performed
simulations aimed at the optimization of the beam fields generated by different virtual
apertures. We describe the iterative procedure used for the optimization and we present
selected results of the simulation calculations and of the experiments performed for
validation. Special focus of the experimental investigations was on the validation of the beam
field simulation results. For various operating modes (beam field steering and/or focusing)
beam profiles have been recorded in water using a hydrophone and compared against the
simulation results. For quantitative evaluation an acceptance criterion has been defined via a
‘performance number’ which has been determined for the various examined cases.

We present an overview of the various steps of this study and show representative results
which prove the performance capabilities of the new phased array sensor.

1. Introduction

New developments in the area of ultrasonic inspection are mainly based on the phased array
technology which allows to steer and focus the ultrasonic beam fields. Thus, it bears a
tremendous potential for a flexible adaptation of ultrasonic inspection to a wide spectrum of
applications and constitutes an efficient alternative to inspection techniques using various
conventional single-element transducers. By using it in the area of pipeline inspection various
inspection modes aiming at crack detection and the detection of metal loss can be performed
simultaneously. In account of these advantages, Baker Hughes (formerly PII Pipeline
Solutions) have developed a pipeline inspection tool based on the phased array technique.
Meanwhile, the UltraScanTM DUO tool ([1], Figure 1) is successfully operated since more than
a decade.

Figure 2 schematically illustrates the various inspection modes which can be executed by
correspondingly applying adapted delay laws when exciting the respective active array
elements. The inspection for metal loss due to corrosion is performed using longitudinal waves

1
 Pipeline Technology Conference 2021, Berlin

at perpendicular incidence, while the detection of external and internal cracks in the pipe wall
is performed using transverse waves at an angle of incidence of, e.g., 45°.

Figure 1. View of the phased array sensors and carriers of the UltraScanTM DUO (left) and tool
receive after an inspection run (right).

Figure 2. Schematic representation of the beam field configurations applied for the detection of the
different defects. For wall thickness measurement (left), operation in focused mode can be beneficial,
while for crack detection (right) the beam field can be steered to one or several angles of incidence.

Moving from this schematic to a realistic view, it gets apparent that threats to pipelines are
becoming more complex and interactive. Cracks can exist in pipelines and can often remain
stable until acted upon by an external force, hence the crack will grow in service. Mechanisms
that stimulate crack growth are mechanical forces (e.g. fatigue, strain) or environmental effects
or a combination of both (e.g. stress corrosion cracking, Figure 3). Certain cracks are difficult
to detect and to distinguish from non-crack anomalies (e.g. geometric features), crack sizing
being an even more difficult challenge.

It is obvious that inspection techniques utilizing one or two parameters to describe simpler
features can match the Probability of Detection (POD), Identification (POI) and Sizing (POS)
requirements. However, cracks can have significantly different morphologies, requiring much
more sophisticated inspection techniques to push the boundaries to include these defect types
in ILI tool performance specifications (Figure 4).

2
 Pipeline Technology Conference 2021, Berlin

Figure 3. Micrograph of a typical stress corrosion crack [2].

Figure 4. Schematic illustration of tool performance enhancement by extracting the maximum

information from ILI data made available by single sophisticated techniques or their combination.

To optimally tailor such sophisticated ultrasonic inspection techniques with enhanced

performance capabilities the phased array principle can be applied. Accordingly, a new
generation of phased array sensors has been developed in the course of simulation-based
investigations. In this contribution, we report on the optimization targets which have been
considered as input for the simulations performed using the Generalized Point Source
Superposition technique (GPSS [3]). We show representative results including a selection of
experiments performed for validation.

2. Sensor Optimization - Targets and Applied Approach

The new generation of phased array sensors has been optimized for inline inspection, covering
a wide range of pipe diameters and wall thicknesses. The optimization has been performed with
respect to crack detection using transverse waves at various angles of incidence aiming at the
further enhancement of the defect signal amplitudes. For wall thickness measurement using
longitudinal waves at perpendicular incidence beam field focusing to different depths has been
considered. The optimization approach has addressed different aspects which have been
worked off step-wise, as briefly described in the following.

3
 Pipeline Technology Conference 2021, Berlin

2.1 Suppression of Grating Lobes

A well-known problem in configuring phased array sensors is the generation of side lobes of
higher order, the so-called grating lobes. These are due to interference effects when steering
and focusing is applied by time-delayed excitation of the single array elements using the
corresponding delay laws. A disadvantageous configuration can result in grating lobe
formation with an amplitude which is even higher than the one of the main lobe. This can lead
to misinterpretation of the inspection data.

2.2 Beam Field Homogenization

A second problem concerns the formation of the beam field in the component in dependence
of the water path (stand-off). If the pipe surface is located, e.g., in the extreme near-field of the
virtual aperture, the beam field in the steel pipe exhibits the well-known interference structure
which has to be avoided for efficient detection and sizing of defects. For adaptation of the near-
field length the variation of the frequency/wavelength as well as the reduction of number and
length of the active elements are suited, the latter accordingly leading to a smaller effective
aperture and thus to a decrease of the near-field length.

2.3 Beam Field Optimization Accounting for the Inspection Specification

If the first two optimization steps have been successfully performed to avoid grating lobe
formation and to enhance the beam field characteristics, a further aspect has to be considered.
In order to ensure optimal coverage of the inspection area, minimum requirements are specified
for the width and height of the amplitude dynamic curves for external and internal notches of
different depths. Further important aspects refer to the sensitivity of the beam field to tilting,
e.g. when the sensor carrier passes over excess weld metal.

The optimization process is thus performed step by step, potentially requiring several iterations.
Based on this approach, the new sensor has been optimized, its efficiency is illustrated in the
following via representative results for the beam fields and the amplitude dynamic curves.

3. Beam Field Simulation and Validation

3.1 Simulated and Measured Beam Fields in Water

Since simulations were the basis for the new sensor design, beam field measurements were
carried out for validation. Two-dimensional scans were performed in water at various distances
from the sensor. For the 2D scans a commercially available measurement system was used. It
consists of a water basin, an x-y-z drive and gear, a hydrophone, an oscilloscope and an
ultrasonic measurement device. Figure 5 shows the water basin and the sensor manipulation
device. The measurements were carried out for a variety of phased array steering and focusing
modes as presented in the following.

We compared and quantitively assessed the beam profiles determined by simulation and by
measurement. For this comparison, line profiles were extracted from the 2D beam field data.
Due to the experimental set-up, the recorded datasets are slightly shifted as compared to the
simulated ones. Therefore, the simulated and the measured profiles were fitted by overlapping
and were aligned by 0.2 mm steps (measurement resolution) until the mean error of the sum of

4
 Pipeline Technology Conference 2021, Berlin

the square errors of each individual data point was minimal. This minimum error was then the
overall performance of the simulation, scaled in [dB]. Considering N data points of the ith
overlap fit, the error for each fit was calculated according to

𝑒 = (𝐴 − 𝐴̅ ) /𝑁

where e : average error for each measured and simulated line profile,
𝐴 : amplitude of the measured data point,
𝐴̅ : amplitude of the simulated data point.
We then calculated the performance number ‘s’ in [dB] according to 𝑠 = min 𝑒 .

Figure 5. Setup of the beam field measurement system.

Results

From the experimental validation results acquired we show two representative cases addressing
beam field steering and focusing. For each case, the results are illustrated via a two-dimensional
plot of the simulated beam field in water (x-z-plane), the two-dimensional image (x-y-plane)
of the simulation and the measurement at the specific distances from the sensor and the
comparison of the specified lateral line profiles.
Case 1 – Insonification 19°, unfocused, profile at z = 28 mm
The 2D images are plotted in logarithmic scaling and qualitatively exhibit common features.
Since the simulations were performed for monochromatic excitation, the features are higher
resolved. Contrary to that the measurements were performed using pulsed ultrasonic excitation
which leads to a smoothening of the beam field. For the quantitative comparison, line profiles
have been selected from the 2D plots. The y-position of the selected line profiles are indicated
by the black lines in the measured images. This applies to Case 2 as well.
The simulated and the measured beam pattern shown in Figure 8 exhibit common features.

For the quantitative comparison, only data points with amplitudes above -9 dB were considered
for the comparison as for lower values the measurement was impacted by noise. The simulation
of the beam field in the x-z-plane (Figure 6) shows the generation of a grating lobe, however,
in our evaluation we concentrate on the main lobe. The lateral profiles at the y-positions 9.2
mm and 11.2 mm were used (Figure 7) for comparison, the calculated performance numbers
‘s’ are 1.87 dB and 1.50 dB, respectively.
5
 Pipeline Technology Conference 2021, Berlin

Figure 6. Case 1: Simulated beam field in water in the x-z-plane (linear scale, left). The simulated
(middle) and measured (right) beam fields in the x-y-plane are shown in logarithmic scaling. The line
profiles selected for the quantitative comparison are indicated by the black lines in the right image.

Figure 7. Case 1: Simulated and measured line profiles at y = 9.2 mm and y = 11.2 mm selected for
the quantitative comparison.

Case 2 – Insonification 0°, focus at z = 60 mm, profile at z = 60 mm

As in the previous case, the simulated and the measured 2D images quantitatively agree well
(Figure 8). For the quantitative comparison, the profile at the position y = 12 mm of the
measured beam field was evaluated (Figure 9). Using data points with amplitudes above -12
dB, the calculated performance number ‘s’ is 0.98 dB.

Conclusion
Usually, the repeatability of ultrasonic measurements is within 1 to 2 dB, also from many
comparisons performed in the past, the same agreement can be expected between simulated
and experimental results. All compared profiles are showing a performance number ‘s’ of less
than 2 dB. Therefore, this is considered to be an acceptable result to prove the accordance of
measurement and simulation and to prove the efficiency of the new transducer design in terms
of its steering and focusing capabilities.

6
 Pipeline Technology Conference 2021, Berlin

Figure 8. Case 2: Simulated beam field in water in the x-z-plane (linear scale, left). The simulated
(middle) and measured (right) beam fields in the x-y-plane are shown in logarithmic scaling. The line
profile selected for the quantitative comparison is indicated by the black line in the right image.

Figure 9. Case 5: Simulated and measured line profiles at y = 12.0 mm selected for the quantitative
comparison.

3.2 Simulated Beam Fields in Steel

An important part of the optimization process after each iteration was the revision of the sensor
characteristics in terms of the inspection specification. The operation range of the new sensor
covers a wide range of pipe diameters and wall thicknesses. Out of the large number of
calculations, we have selected simulation results for a representative pipe diameter.

Figure 10 shows the beam fields for transverse waves, which are generated in the pipe wall by
steering to angles of incidence from 30° to 70°. Here, the array elements in the center of the
array probe are used as a virtual sensor. In the presented depth range, the beam fields display a
homogeneous appearance with only slight differences in the maximum amplitudes when
compared to each other. Furthermore, no grating lobes are generated, only in the 30° case a
weak lobe can be recognized which, however, far below the maximum amplitude of the main
lobe.

A further criterion for the effectiveness of the sensor is the beam field homogeneity when the
virtual sensor is shifted along the complete array aperture. In Figure 11, the beam fields are

7
 Pipeline Technology Conference 2021, Berlin

plotted for an angle of incidence of 45° for the cases where the virtual aperture is located at the
center, at the outmost left and the outmost right position. It can be seen that by shifting the
virtual sensor an electronic scanning in circumferential direction can be performed.

Figure 10. Beam field simulations for transverse waves with angles of incidence from a) 30°, b) 45°,
c) 60° to d) 70°. The amplitudes are plotted in logarithmic scale

4. Calculation of Amplitude Dynamic Curves for Notches as Model Defects

For the calculation of amplitude dynamic curves (ADCs) for notches of different depths,
positioned internally and externally at the pipe wall, the GPSS simulation code has been
accordingly modified. External notches are detected using the half skip reflection, while for
internal notches the full skip reflection is considered. Thus, for the latter the reflection of the
insonified transverse wave at the outer pipe wall has to be accounted for. Moreover, the
implementation also considers the scan along the circumferential pipe direction.

In Figure 12, the simulated amplitude dynamics for two representative pipe diameters are
shown for the inspection with 45° transverse waves, the pipe wall thickness is 8 mm. Due to
the optimized sensor characteristics, the maximum amplitudes and the shape of the ADCs are
very similar for both the two pipe diameters and the external as well as internal notches.

8
 Pipeline Technology Conference 2021, Berlin

Figure 11. Simulated transverse wave beam field for different positions of the virtual sensor: position
a) outmost left, b) centered and c) outmost right, the angle of incidence is 45°.

Figure 12. Simulated ADCs for two different pipe diameters (left and right), the detected defect amplitudes are
plotted against the scan angle. The diagrams display the respective curves for external notches (a) and c)) as
well as for the internal notches (b) and d)).

5. Experimental Validation of the ADC Simulations

The validation measurements were performed in the laboratory using a test rig which allows to
scan test specimens of various diameters in circumferential direction using immersion
technique. The test rig is shown in Figure 13 including the measurement set-up where the

9
 Pipeline Technology Conference 2021, Berlin

sensor insonifies onto the inner surface of a pipe segment. For the validation of the simulated
ADCs and defect amplitudes, a series of test specimens of different diameters and wall
thicknesses were fabricated with notches of 1 mm, 2mm and 4 mm depth.

Figure 13. Lab set-up for the validation measurements consisting of the water basin, sensor holder, manipulator
and mounting element for the positioning of test specimens of various diameters. In the image on the right, the
prototype of the new sensor can be seen, which has been fabricated by Waygate Technologies, a Baker Hughes
business (Hürth, Germany).

For two different diameters, the diagrams in Figure 14 show the maximum amplitudes of the
simulated ADCs as a function of notch depth in comparison with the experimentally
determined values. For the internal notches, the values agree well, also for the external notches,
except for the 1 mm deep notch where the experimental values exceed the simulated ones by
approximately 4 units. This larger difference could due to the influence of the notch width in
relation to the low notch depth or to the quality of the notch preparation.

Figure 14. Maximum amplitudes as a function of notch depth for two different pipe diameters. The values are
normalized to the maximum amplitude of a 1 mm external notch in a 42” pipe of 8 mm wall thickness.

6. Summary

In this contribution, we report on the optimization of a phased array sensor for pipeline
inspection. The specification of the target parameters in terms of a further enhancement of the

10
 Pipeline Technology Conference 2021, Berlin

sensor currently used on the UltraScan TM DUO ILI tool was the basis for the optimization
process. The simulation results for the beam fields and the amplitude dynamic curves have
been validated in comparison with experimental results.

The new generation of phased array sensor is the basis for the implementation and application
of sophisticated inspection techniques with enhanced performance as compared to standard
techniques. Using several of these simultaneously during an inspection run allows to extract
even more information from the acquired data. This is essential to achieve enhanced tool
performance in terms of POD, POI and POS of complex defect morphologies.

The steady progress in computer performance allows to even consider the application of
sophisticated phased array imaging techniques, such as Plane Wave Imaging. Respective
studies and implementations have already been performed [4] and will further be addressed in
due course.

The presented work is part of our ongoing efforts to support the technically and economically
efficient integrity programs of our customers. Enhanced accuracy and reliability of our ILI data
help to reduce the potential threat due to certain critical defect types and to reduce the amount
of unnecessary dig and repair costs.

References

[1] A. Hugger, et al.; Ultrasonic Phased Array Crack Detection Update; 4th Pipeline Technology Conference 2009
[2] P.G. Fazzini, J.L. Otegui; Experimental Determination of Stress Corrosion Crack Rates and Service
Lives in a Buried E.R.W. Pipeline; International Journal of Pressure Vessels and Piping (2007)
[3] M. Spies; Efficient Optimization of Single and Multiple Element Transducers for the Inspection of Complex-
shaped Components; NDT&E International, 37, 455-459 (2004)
[4] R.K. Rachev, et al.; Plane Wave Imaging Techniques for Immersion Testing of Components With Nonplanar
Surfaces; IEEE Transactions on Ultrasonics, Ferroelectrics and Frequency Control, 67, 1303-1316 (2020), doi:
10.1109/TUFFC.2020.2969083.

11