See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/228571110

Pipeline Inspection with the High Resolution

EMAT ILI-Tool: Report on Field Experience

Article October 2002

CITATIONS READS

4 672

2 authors, including:

Thomas Beuker
ROSEN Germany GmbH
32 PUBLICATIONS 97 CITATIONS

SEE PROFILE

All content following this page was uploaded by Thomas Beuker on 27 July 2015.

The user has requested enhancement of the downloaded file.

 Proceedings of IPC 2006
6th International Pipeline Conference
September 25-29, 2006, Calgary, Alberta, Canada

Draft IPC2006-10156

PIPELINE INSPECTION WITH THE HIGH RESOLUTION EMAT ILITOOL:

REPORT ON FIELD EXPERIENCE

Dr.-Ing. Martin Klann Thomas Beuker

ROSEN Technology GmbH & Co. KG ROSEN Technology GmbH & Co. KG
Physics Department Product Development
Am Seitenkanal 8 Am Seitenkanal 8
49811 Lingen (Ems), Germany 49811 Lingen (Ems), Germany
Email: [email protected] Email: [email protected]

ABSTRACT INTRODUCTION
In succession of a research project to develop an EMAT in- Pipelines are commonly inspected by inline inspection (ILI)
line inspection technology for the detection of stress corrosion tools equipped with sensors for nondestructive evaluation and
cracking (SCC) and disbonded coating, a 16 EMAT inspection testing. These sensors utilize different nondestructive techniques
tool was manufactured and tested. like Magnetic Flux Leakage (MFL), Ultrasonics Testing (UT),
This 16 EMAT Crack Detection (ECD) Tool establishes a or Eddy Current (EC). ILI tools available in the market today
new high-resolution approach to effective and accurate crack in- are primarily based on the MFL principle. These tools have ma-
spection technology especially for gas lines. The performance of tured over the years to provide superior mapping of corrosion.
this tool was tested on several pipe samples containing natural On the other hand, adequate and cost effective techniques to de-
SCC as well as artifical anomalies and laminations. Further- tect cracks especially in gas pipelines have not been realized yet.
more, a field test is currently being conducted, which will detail Piezoelectric ultrasonics have been utilized for wall thick-
the tools performance under operational conditions. The new ness measurements and crack detection. These tools work well
technology will be compared to the results from the established in liquid filled pipelines since the fluid being transported is cou-
Magnetic Flux Leakage inspection (MFL) technology. pling the pressure wave generated by the piezoelectric transduc-
EMAT inspection technology, originally designed for detec- ers to the pipe wall. However, in natural gas pipelines this cou-
tion of SCC, is also sensitive to areas of disbonded coating, other pling mechanism does not work since nearly all of the acoustical
crack-like features, and anomalies like gouging and channeling. energy generated by the transducer in the gaseous product is re-
To distinguish the different types of pipeline threats, a qualita- flected by the pipe wall and does not travel into the pipe wall.
tive defect sizing process is under development incorporating the In a research project conducted by ROSEN and Sonic Sen-
EMAT data and the information from the other in-line inspection sors, Inc. [1], a new contact free sensor based on Electro-
technologies. Magnetic Acoustic Transduction (EMAT) was developed. The
Following an introduction into EMAT guided wave theory, EMAT sensor generates ultrasonic waves inside the pipe wall it-
data from a series of full-scale tests obtained with 34 prototype self through a combination of Lorentz forces and magnetostric-
sensors and the complete 16 ILI-EMAT tool, will be presented. tion, no matter what medium is present in the pipe. The material
The inspection tool itself, as well as the process for evaluating the being inspected is its own transducer, eliminating the need for a
recorded data, are undergoing continuous improvements. This liquid couplant as required in traditional UT methods a good
paper will discuss the development status of these technologies. solution for natural gas pipelines.

1 Copyright
c 2006 by ASME
a)

Figure 2. SIMPLIFIED PRESENTATION OF PARTICLE MOVEMENT

b) FOR HORIZONTAL SHEAR AND LAMB WAVES.

c)
Cracks exist often in colonies with various lengths and
depths. Typically there are numerous shallow cracks within one
colony. Their depth growth happens by coalescing of initially
d) shallow and short cracks. A distinct irregular surface, especially
for the inter-granular crack types can be expected. SCC occurs
Figure 1. EXAMPLES OF STRESS CORROSION CRACKING FOUND
in gas and in liquid pipelines, typically linked to sections with
IN TEST SAMPLES FROM 16 PIPE, VISUALIZED USING MAGNETIC
disbonded coating.
PARTICLE INSPECTION (MPI) METHOD.

EMAT BASICS
NOMENCLATURE During the last two years, several investigations on EMAT
EMAT Electro-Magnetic Acoustical Transducer transducer technology and design were conducted. Remarkable
ECD EMAT Crack Detection experimental results were obtained from laboratory as well as
UT Ultrasonic Testing from pulltests that proved the feasibility of using EMAT trans-
MFL Magnetic Flux Leakage ducers and receivers to detect cracks and other structural defects
SCC Stress Corrosion Cracking in pipelines.
SHn Shear Horizontal guided wave of order n
EDM Electro-Discharge Machining Guided-wave Mode Selection
MPI Magnetic Particle Inspection Fundamentally, guided waves can be categorized into Lamb-
waves and horizontal shear waves of different order. A guided-
wave solution is well suited to detect SCC, because the SCC is
STRESS CORROSION CRACKING (SCC) a general degradation of the material, and the guided-wave is a
Stress corrosion cracking (SCC) is a term used to describe mechanical elastic wave whose propagation depends on the di-
service failures in engineering materials that occur by slow envi- mensions and properties of the material itself.
ronmentally induced crack propagation. This phenomenon is as- Guided-waves, as generated by an EMAT transducer, prop-
sociated with a combination of stress (applied or residual) above agate between the external and the internal pipe surface as a
some threshold value, specific environmental and in some sys- boundary condition. The presence of SCC disturbs the guided-
tems metallurgical conditions, which lead to surface cracks with wave and a significant change in the dimensions of the wave
a high aspect ratio (long and shallow). SCC has been recognized guide, such as that from a crack, will reflect the wave and cause
as a cause of failures in high pressure gas and oil transmission an echo, which may then be detected. The design of an EMAT
lines since mid 60s. SCC on the external surface of pipelines allows to control very well the type of guided wave and its prop-
has occurred in several countries throughout the world and con- agation properties.
tributes to major failures in pipelines [2]. A permanent magnet circuit based on a common MFL tool
Although the mechanisms and causes for SCC in pipelines generates a magnetic field that is orientated axially in plane with
are still an ongoing research topic, the shape and appearance of the pipeline wall. A meander coil electromagnetic acoustic trans-
SCC is well known. To investigate natural SCC features from an ducer (MC-EMAT) is placed in the magnetic field and driven
inline-inspection point-of-view, several North American pipeline with a short tone burst of adjustable frequency. By Lorentz-
operators provided pipe samples to ROSEN. Fig. 1 shows some force and magnetostriction several types of waves are induced
examples of SCC regions in these samples, visualized using inside the pipeline wall. The majority of these waves are largely
Magnetic Particle Inspection (MPI). damped, except for guided waves that propagate in so-called

2 Copyright
c 2006 by ASME
modes of the structure. The different modes exhibit dispersion, 1000
SH0
so there exist certain frequency-wavelength or frequency-coilgap SH1

frequency, kHz
relations, where the preferred mode is generated with sufficient 800 SH2
effiency. Since the acoustical wavelength of the induced guided SH3
SH4
wave is provided by the constant gap between the meander coil
600
loops, one of the guided wave modes is discriminated against all
others by adjusting the frequency of the applied tone burst.
The preferred mode is in general a horizontally polarized 400
shear wave, a so-called SHn wave, where n stands for the number
of nodes in the particle velocity amplitude distribution as seen in 6 8 10 12 14 16
wall thickness, mm
cross-section of the pipeline wall, see Fig. 2. The modes with
order n>0 have an especially high energy concentration at the
surfaces of the wall and hence are sensitive for possible cracks Figure 3. GUIDED WAVE FREQUENCY SPECTRUM AS A FUNCTION
or slots inside the sound beam that will lead to reflection and OF THE WALL THICKNESS.
refraction of the shear wave.

The EMAT excitation of distinct guided waves is based on

Dispersion Relation. the preselection of certain modes by adjusting the excitation fre-
Based on the common textbook by Achenbach [3] an quency f according to a given wave length in order to satisfy
overview on the theoretical formulation of guided wave propa- Eq. 3 with suitable values. The ideal case of monochromatic ex-
gation in plates will be developed. citation of a single mode at a single frequency would require a
From geometrical considerations the frequency equation for meander coil with an infinite number of loops and an excitation
harmonic SH waves in an elastic layer can be written as waveform with infinite length. If the temporal frequency of the
excitation does not correspond to the given wave length with re-
2 = n2 + 2 (1) spect to Eq. 3, either no guided wave mode is excited at all or an
unwanted mode, e.g. a Lamb mode or a Rayleigh wave is excited.
Due to the finite length of both the meander coil and the ex-
While n = 0, 1, 2, . . . is the order of the guided wave mode, the citation tone burst, the excitation is not monochromatical but ex-
dimensionless frequency and the dimensionless wavenumber hibits several frequencywave length combinations that might lie
are defined as inside the excitation bandwidth and can therefore be observed. It
is desirable to select an optimum frequencywave length combi-
2W T f 2W T f 2W T nation so that only one mode of order n is actually excited while
= , = = (2) all the adjacent modes do not lie inside the excitation bandwidth
cT cSHn SHn
and hence are not excited.
Figure 3 also shows that the SH0 mode frequency does not
The wavelength SH of the SHn-waves is determined by the depend on the wall thickness, whereas the frequencies of higher
EMAT meander coil geometry actually implemented. Their modes do significantly depend on the wall thickness. This means
modal frequencies depend on the order n, the wall thickness W T , that a short sound burst containing higher order modes will elon-
and on the transverse sound velocity cT , that is cT 3255 m/s for gate with increasing travelling distance. This effect hinders the
constructional steel. Substitution of those entities relates all rel- use of higher order modes if large areas have to be covered with
evant parameters to each other. a single sound burst as it was implemented in previous EMAT
Some relations are of particular interest for the operation of ILI tools.
an EMAT. A few simple calculations lead to the following equa- Compared to the classical Ultrasonic Testing (UT) approach
tion describing the SH frequency spectrum of the steel layer: that uses short pulses of longitudinal waves, the transverse shear
wave burst generated by an EMAT transmitter is much longer

n 2
 in time. Classical UT obtains a high spatial resolution by us-
2 2
fSHn = c2T + SHn (3) ing short pulses, but cannot use frequency information since the
2W T
spectrum of the pulses is very broadband. EMAT generated
waveforms, on the other hand, are very narrowband in terms of
At a fixed acoustical wavelength, this equation relates the fre- their frequency content. Taking into account the dispersion rela-
quency of the guided wave modes to the wall thickness of the tion displayed in Fig. 3, the frequency spectrum obtained from
pipeline under investigation, see Fig. 3. the transmission signal can be used to estimate the actual wall

3 Copyright
c 2006 by ASME
 Pixel Lab Test piece, WT = 8 mm
Crack Run Direction /
Sender
Pipe axis Depth:
0.1 0.5 1.0 2.0 mm

4 EDM Notches (l=20 mm)

Echo Transmission
Receiver Receiver

time
Figure 4. CONFIGURATION OF INDIVIDUAL EMAT SENSOR REPRE-
SENTING ONE OF 72 PIXELS ON THE 16 EMAT ILI TOOL.
Echo Envelopes
thickness. Furthermore, it is known that the interaction of shear position
waves with features of varying depths depend on the mode of the
shear wave actually generated. This will be used to estimate the Integral of Echo

dt
depth of a defect from the spectrum of the echo signal obtained
in a run.

position
Laboratory Tests
Figure 5 shows a sample plate with EDM-notches of con-
Figure 5. LABORATORY TEST MEASUREMENT WITH PROTOTYPE
stant length and width, but varying depths, that was used in lab-
EMAT SENSORS.
oratory measurements with plain prototype EMATs. The basic
EMAT sensor configuration incorporating sender, transmission
receiver and echo receiver is displayed in Fig. 4. In the mid- The ECD sensors carriers are spring-loaded integrated de-
dle of Fig. 5, the resulting A-scan echo amplitudes are displayed vices with a total of 8 electronic assembly boards for pulse gener-
as a function of position (x-axis) and EMAT time (y-axis). The ation, receiver pre-amplifying, data processing, and communica-
bottom part of Fig. 5 shows the echo data integrated along the tion, respectively. Extensive monitoring of sensor performance
EMAT time axis and gives a scalar representation of the acousti- parameters like supply voltages, charge currents and tempera-
cal energy reflected by the notches. tures ensure optimum reliability.
The reflected energy is increasing with increasing depth of The actual EMAT sensors themselves are spring-loaded
the notches. A more detailed investigation of the frequency con- modules with an integrated ceramic wear surface that ensures the
tent of the reflected waveform reveals that different modes be- necessary low lift-off distance to the pipe wall. Since the EMAT
have differently with regard to the depth of the notches. This modules run completely independent from the magnets, wear is
spectral information is used in a sizing algorithm to classify the greatly reduced and long inspection distances become feasible.
depth of features from the echo data obtained from the ILI tool.

High-Resolution Pixel Definiton

EMAT ILITOOL DESIGN CONCEPT Low-resolution EMAT tools are equipped with few EMAT
EMAT inline inspection systems have been around for some sensors on the circumference only. Such a design utilizes the
years; however, ROSENs EMAT Crack Detection (ECD) tool ability of a guided wave to travel around the perimeter but de-
takes a different approach to achieve optimum performance. The mands a cumbersome interpretation of a single waveform that
first aspect in the new approach is to design the magnetic config- contains information from a large area. Moreover, signal quality
uration similar to conventional MFL ILI tools. As a result, the and reliability is reduced since the guided wave is attenuated and
magnetic energization of the pipe-wall is obtained from a typ- dispersed on its way between transmitter and receiver.
ical magnetic circuit design. There are no magnets located in At this stage it was decided to utilize a high-resolution ap-
the ECD sensor housing itself. This approach makes the sen- proach with a large number of individual EMAT sensors distrib-
sors much smaller and lighter then alternative EMAT designs, uted around the circumference of the pipeline. Each individual
enabling excellent dynamic performance under a wide range of sensor is sampling a distinct area that is a fraction of the whole
operating conditions. circumference.

4 Copyright
c 2006 by ASME
 1. filtering the received echo and transmission waveforms with
matched filters to enhance the signal-to-noise ratio; a good
estimate of a matched filter is given by the actual excitation
waveform that is obtained and digitized for each shot indi-
vidually.
2. Fourier transforming the filtered data to obtain the frequency
spectrum of both the transmission and the echo signal.
3. deriving the envelope of the filtered data; the Hilbert trans-
form turns out to be the standard procedure to obtain the
envelope of a quasi-periodical waveform.

Data Evaluation Process

The evaluation of an actual inline inspection tool run con-
Figure 6. ROSEN 16 EMAT HIGH RESOLUTION CRACK DETECTION sists of the following main steps:
TOOL EQUIPPED WITH 72 EMAT SENSORS.
1. check lift-off channel to ensure proper mechanical contact
of the EMAT sensors to the inner pipeline surface
The newly developed 16 EMAT tool as displayed in Fig. 6 2. check transmission signal to ensure acoustical energy is
is equipped with an improved resolution of 72 overlapping present in the pipe, obtain wall thickness from transmission
EMAT channels, evenly distributed around the circumference. frequency analysis
By aligning adjacent EMAT pixels slightly oblique to each other, 3. find girth welds in transmission and echo signal
a circumferential array is composed were each pixel covers an 4. find regions with disbonded coating by adaptively analysing
area of only few square inches. the transmission signal amplitude and integral
Since the sound path is limited to a short distance between 5. find regions with significant values in echo signal, use corre-
transmitter and receiver, both propagation loss and dispersion ef- lation and database to sort out installations and other known
fects between transmitter and receiver are neglectable. This pro- features
vides a superior signal-to-noise ratio of the EMAT sensor which 6. investigate remaining echo signals in more detail with regard
simplifies the subsequent data evaluation and avoids misinterpre- to time waveform, arrival times, and frequency content
tations. The generated waves are travelling in both clockwise and 7. correlate results to previously obtained inspection data
counterclockwise direction and are reflected preferentially by ax- 8. classify echo based on a correlation algorithm taking into
ially oriented features. This interrogates irregular cracks from account all available integrity management information
two different sides.
It is known that different feature types have different reflection
properties, e.g. seam welds are good reflectors for SH0 waves
Data Processing while higher-order modes are preferably reflected at shallow
An individual ECD carrier captures about 10,000 digitized cracks.
samples per logdistance, e.g. every 5 mm. Since memory capac- Since the guided wave modes involved are well defined, it is
ity is limited, it is not possible to store all raw A-scan data that possible to derive the properties of the feature itself aided by a
accumulates during a run. The overall amount of raw data poses conversion method. Hence it is possible to distinguish between
a severe challenge to the data transfer capabilities. different feature types, e.g. long seams, cracks, general corro-
The features of interest in the pipe wall under investigation sion, lamination, installations, etc.
may be obtained during runtime by appropriate data processing
inside the inspection pig. Actually, not all of the data is necessary
to display the results of the inspection run; in normal conditions
while ne feature is detected the amount of data to be stored can be FIELD TESTS
reduced significantly. So memory capacity is preserved to store Starting early in 2005, full-scale pulltests were conducted
critical signals in greater detail to be investigated in the after-run with prototype EMAT sensors in a 34 pulltest line containing
evaluation. artificial features as well as disbonded coatings and natural SCC.
The data pre-processing is implemented utilizing high- Upon completion of the 16 ECD tool, pulltests in several
performance field-programmable gate arrays (FPGAs) for each 16 lines containing a variety of features were performed. This
pixel individually and essentially consists of three steps: section reports on the results of these test runs.

5 Copyright
c 2006 by ASME
 ECD data analysis
Feature detection

Coating Disbondment Detection

8
7

(-dB versus bare pipe)

6
60 degrees

Signal Effect
5
4
3
2
1
0
5m
FBE Coating PE Coating Bitumen/Tar Tape coating
27 Pulltest data (integrated along EMAT axis)
Coating disbonded on 20 mm length
Figure 7. RESULTS FROM 34 PULLTEST: EDM-NOTCHES; THE UP-

EMAT transmission, a.u.

WT = 5 mm
1.2
MOST PIXEL SHOWS REFLECTIONS FROM A LONG SEAM. WT = 6.2 mm
1.1 WT = 10 mm

1
Artifical EDM-Notches 0.9
The 34 line contains natural SCC as well as a variety of 0.8
artificial electro-discharge machined (EDM) notches. 0.7
Figure 7 shows the integrated echo data captured during the 0 100 200 300 400 500
runs. Significant echo signals were obtained from the EDM- logdistance, mm

notches with a length ranging from 15 to 50 mm, a depth between

10% and 80% WT, and a width of 0.2 mm. A good correlation Figure 8. TOP: SIGNAL ATTENUATION FOR DIFFERENT TYPES OF
between depth and frequency content was found that aids the data COATING; BOTTOM: INCREASE OF TRANSMISSION SIGNAL IN RE-
evaluation process currently under development. GIONS WITH DISBONDED COATING (FROM 16 PULLTEST).

Coating Disbondment
A coating present on the outer surface of the pipeline atten-
uates the guided shear wave significantly. This is a serious issue
in a low-resolution approach since this attenuation decreases the
signal amplitude obtainable if the EMAT receiver is positioned
at a large distance to the transmitter.
The high-resolution approach uses short distances between
transmitter and receiver so the signal amplitude is only slightly
attenuated with a neglectable effect on crack detection capabil-
ities. In this configuration the absolute transmission amplitude
obtained depends on the type of coating present, see Fig. 8 top.
Furthermore, coating disbondment is detectable as an increase
of the transmission amplitude, see Fig. 8 bottom. The minimum
detectable disbondment area is approx. 20 mm long and 50 mm
wide.
Figure 9. COMPARISON BETWEEN ECD AND MFL SIGNALS OF THE
SAME REGION OF A PULLTEST LINE CONTAINING ARTIFICIAL FEA-
Comparison between ECD and MFL data TURES.
Figure 9 shows a comparison of ECD and MFL data. The
area of the pulltest line displayed in Fig. 9 contains a total of
three artificial features. The prominent signal centered at about
37.1 m, 170 degrees is induced by a grinding machined into the Two other features are visible solely in the ECD data at 220
outer surface posing a significant loss of wall thickness. This is degrees, 36.8 m and 37.2 m, respectively. These are echos from
clearly visible in both MFL and ECD data; taking a closer look 0.2 mm wide EDM-notches with a length of 20 mm and 10 mm,
at the ECD data reveals a corresponding degradation of the trans- respectively. There are no signals in the MFL data of this regions.
mission signal that indicates a large-area feature with significant This comparison is a valueable step in the evaluation process to
depth. distinguish between general corrosion and crack-like features.

6 Copyright
c 2006 by ASME
Natural SCC colonies
Figure 10 shows the EMAT echo signals obtained from four
pixels each at different locations of a pulltest line . The sus-
picious regions were investigated with MPI showing significant
SCC crack colonies. The depth of these cracks was found to
be up to 0.120 at 5-6 mm wall thickness (prominent peaks in
a)
Fig. 10a) but also much shallower subcritical cracks induced sig-
nificant echo amplitudes. An excellent correlation between the
SCC regions made visible with MPI and the EMAT data was Pixel1
found. Signals were obtained from both sides of the cracks.
Pixel2

FURTHER DEVELOPMENTS Pixel3

The time axis of individual EMAT channels can be bro-
Pixel4
ken down to achieve an intra-pixel resolution. Therefor the
acoustical information will be correlated with acoustical pat-
terns of available defect types. A sophisticated data evaluation
is currently under development, correlating all available inspec- b)
tion data including MFL data, geometry data, and pipe log data
contributed by the operator. The first inspection run under op-
erating conditions is scheduled for April 2006. Detailed results Pixel1
of this run will be presented upon availability in the conference
Pixel2
presentation.
Pixel3
SUMMARY Pixel4
A novel high resolution EMAT ILI tool was presented. The
tool has an improved detection capability for SCC colonies due
to an appropriate selection of ultrasonic guided-waves and a two- c)
directional circumferential pulse sequence. By additional mea-
surement results using a variety of guided wave features, in-
Pixel1
dependent of the SCC measurement, information about the in-
tegrity of the coating bond can be obtained. Pixel2
Tests on artificial notches showed an excellent sensitivity on
shallow and short defects (depth 10% of wall thickness, length of Pixel3
15mm). SCC colonies were detected with good signal-to-noise
ratio in a variety of real pipe samples. Thus, this technology com- Pixel4
bines proactive and reactive measures to improve the reliability
of a pipeline that is susceptible to SCC.

d)
REFERENCES
[1] Beuker, T., Brown, B., Alers, R., and Alers, G., 2004. SCC
Pixel1
detection and coating disbondment detection improvements
using the high resolution EMAT ILI-technology. In Pro- Pixel2
ceedings of IPC 2004, pp. IPC040697.
[2] Manfredi, C., and Otegui, J. L., 2002. Failures by SCC in Pixel3
buried pipelines. Engineering Failure Analysis, 9(5), Oct.,
pp. 495509. Pixel4
[3] Achenbach, J. D., 1975, 8th imp. 1999. Wave Propagation
in Elastic Solids, Vol. 16 of Applied Mathematics and Me- Figure 10. MEASUREMENT RESULTS IN SCC SAMPLE REGIONS.
chanics. Elsevier Science.

7 Copyright
c 2006 by ASME

View publication stats