Original Article

Structural Health Monitoring

2019, Vol. 18(5-6) 1903–1914

Open Guided Waves: online platform Ó The Author(s) 2018

Article reuse guidelines:
sagepub.com/journals-permissions
for ultrasonic guided wave DOI: 10.1177/1475921718817169
journals.sagepub.com/home/shm
measurements

Jochen Moll1 , Jens Kathol2, Claus-Peter Fritzen2,

Maria Moix-Bonet3, Marcel Rennoch4 , Michael Koerdt4,
Axel S Herrmann4, Markus GR Sause5 and Martin Bach6

Abstract
Ultrasonic guided waves have been used successfully in structural health monitoring systems to detect damage in isotro-
pic and composite materials with simple and complex geometry. A limitation of current research is given by a lack of
freely available benchmark measurements to comparatively evaluate existing methods. This article introduces the
extendable online platform Open Guided Waves (http://www.open-guided-waves.de) where high-quality and well-
documented datasets for guided wave-based inspections are provided. In this article, we describe quasi-isotropic
carbon-fiber-reinforced polymer plates with embedded piezoelectric transducers as a first benchmark structure.
Intentionally, this is a structure of medium complexity to enable many researchers to apply their methods. In a first step,
ultrasound and X-ray measurements were acquired to verify pristine conditions. Next, mechanical testing was done to
determine the stiffness tensor and sample density based on standard test procedures. Guided wave measurements were
divided into two parts: first, acoustic wave fields were acquired for a broad range of frequencies by three-dimensional
scanning laser Doppler vibrometry. Second, structural health monitoring measurements in the carbon-fiber-reinforced
polymer plate were collected at constant temperature using a distributed transducer network and a surface-mounted
reversible defect model. Initial results serving as validation are presented and discussed.

Keywords
Guided waves, composite structures, signal processing, structural health monitoring, scanning laser Doppler vibrometry

Introduction
A literature review published by Mitra and
Gopalakrishnan1 shows the recent developments in the
1
area of guided wave techniques for structural health Department of Physics, Goethe University Frankfurt, Frankfurt am Main,
monitoring (SHM). Multiple methods were reported in Germany
2
Department of Mechanical Engineering, University of Siegen, Siegen,
that paper ranging from signal processing techniques to Germany
statistical and machine learning methods. The important 3
Institute of Composite Structures and Adaptative Systems,
question for the practical application of guided wave– Multifunctional Materials, German Aerospace Center (DLR), Germany
4
based diagnostics is the performance of the methods rela- Measurement Systems and Monitoring, Faserinstitut Bremen e.V.
(FIBRE), Bremen, Germany
tive to each other. In other words, what is the diagnostic 5
Mechanical Engineering, Institute for Materials Resource Management,
accuracy of those methods, for example, in terms of University of Augsburg, Augsburg, Germany
damage detection sensitivity and localization correctness. 6
ETLN—NDT Engineering & Process Simulation, Airbus Helicopters
The lack of permanent and publicly available bench- Deutschland GmbH, Donauwörth, Germany
mark models motivated the research presented in this
Corresponding author:
article. By means of well-documented and freely acces- Jochen Moll, Department of Physics, Goethe University Frankfurt, Max-
sible measurements, existing algorithms for SHM and von-Laue-Straße 1, 60438 Frankfurt am Main, Germany.
wave field processing can be fairly compared without Email: [email protected]
1904 Structural Health Monitoring 18(5-6)

uncertainties related to different measurement equip-

ment, different transducer technology, and so on. In
this sense, such datasets might be considered as a refer-
ence standard. The data base might also be helpful for
guided wave beginners and those researchers with smart
ideas but no access to expensive measurement equip-
ment. The overall goal of the Open Guided Waves
(OGW) online platform is to support research and
developments in the field of guided wave technology
for SHM.
Basically, SHM methods must be at least as reliable
as competing non-destructive testing (NDT) tech-
niques.2,3 To proof reliability, appropriate requirements
for damage detection need to be defined and validated
with respect to the structural items of interest. This pro- Figure 1. Workflow illustrating data generation, deposition,
cess is well established for the verification or ‘‘technical and dissemination.
qualification’’ of conventional ultrasonic testing.4,5 For
active guided wave approaches, the differences are in composite materials.9–16 This wave field data are also
the required elaborate consideration of structural fea- beneficial for the verification of image processing tools
tures at the structural item, variations of environmental for acoustic wave field analysis and spectroscopy.17–22
and operational conditions (EOCs) in conjunction with Section ‘‘Guided wave measurements part II: SHM
a potential dependency on previous measurement data measurements’’ presents the SHM measurements of the
(e.g. reference or baseline data), and the interaction of CFRP plate at constant temperature where a surface-
guided waves with structural damage in terms of dam- mounted reversible defect model was placed at several
age type, position, and inspection frequency.6–8 positions on the structure. Multiple frequencies were
Most current works only partly address those recorded in a round-robin fashion at each structural
aspects. There are major differences in numerical model- condition. This dataset is well suited to test baseline-
ing procedures or the way experiments were performed. dependent and baseline-free damage detection and
Due to this high degree of variation, a comparison of damage localization methodologies.23–27 Finally, con-
results is nearly impossible, for example, due to different clusions are drawn at the end in section ‘‘Summary.’’
modeling strategies, or different actuators and data
acquisition systems used. This shows the need to estab-
lish benchmark tests that are valuable to compare and Description and characterization of the
validate guided wave–based methods for SHM. test structures
The underlying idea of the proposed OGW platform
is shown in Figure 1 which illustrates the whole process
Test structure preparation
from data acquisition to its distribution. A first test Four CFRP plates were manufactured for this study with
structure, described in section ‘‘Description and charac- the dimensions of 500 mm 3 500 mm and a thickness of
terization of the test structures,’’ is given by carbon- 2 mm. The specimens are based on prepreg material
fiber-reinforced polymer (CFRP) plates with clearly HexplyÒ M21/34%/UD134/T700/300. Depending on its
defined geometry and embedded piezoelectric transdu- purpose, as shown in Table 1, the laminates are either
cers. To characterize the baseline state, ultrasound and quasi-isotropic with layup ½45=0=  45=90=  45=0=
X-ray testing were performed. Material properties in 45=90S or unidirectional.
the form of stiffness tensor and sample density A first specimen, called ‘‘wave field plate,’’ with a
were measured, validated, and compared with theoreti- single piezoelectric transducer in the center was used for
cal predictions. A discussion of the reversible damage acoustic wave field measurements. A second specimen,
model used in this work is presented in section called ‘‘SHM plate,’’ was equipped with 12 piezoelectric
‘‘Relevance of reference damage.’’ Next, section transducers for the acquisition of guided wave data in
‘‘Guided wave measurements part I: 3D acoustic wave pitch-catch configuration. Both specimens are shown in
field measurements’’ describes the acoustic wave field Figure 2. Two additional laminates are divided in cou-
measurements by a three-dimensional (3D) scanning pons and undergo mechanical testing to determine the
laser Doppler vibrometer characterizing guided wave material properties, and to validate the stiffness tensor.
propagation in the structure for a broad range of fre- DuraAct piezoelectric transducers are co-bonded to
quencies. This dataset will be helpful in the verification the plate during the curing process in the autoclave.
of numerical methods simulating wave propagation in These transducers are composed of a lead zirconate
Moll et al. 1905

Table 1. Labeling and description of the laminates.

No. Layup Transducers Labeling Purpose

1 Quasi-isotropic 12 ‘‘SHM plate’’ Acousto-ultrasonics

2 Quasi-isotropic 1 ‘‘Wave field plate’’ Full-wave field analysis
3 Unidirectional No – Mechanical testing (stiffness tensor measurement)
4 Quasi-isotropic No – Mechanical testing (validation)

Figure 2. (a) ‘‘Wave field plate’’ with single transducer in the middle of the structure; (b) ‘‘SHM plate’’ equipped with 12
transducers. One row with six equally spaced transducers is arranged on top and one row at the bottom of the plate. In both plates,
the horizontal direction denotes a fiber orientation of 90° and the vertical direction represents a fiber orientation of 0°.

titanate (PZT) circular disk embedded in a ductile poly- Table 2. Capacitance of piezoelectric disks of the ‘‘SHM plate’’
mer along with the required electrodes, electrical con- measured with PeakTech Capacitance Tester 3710.
tacts, and insulators.28 The piezoelectric disk measures
Transducer Capacitance (nF)
0.2 mm in thickness and 5 mm in diameter. The embed-
ding of the piezoelectric transducer provides electrical T1 4.54
insulation and mechanical pre-compression, making T2 4.52
the piezoceramic robust against deformations. The T3 4.58
capacitances of the 12 transducers of the ‘‘SHM plate’’ T4 4.54
T5 4.45
are listed in Table 2 and show only small variability. T6 4.60
T7 4.51
T8 4.41
Baseline state characterization using ultrasonic testing. All T9 4.56
specimens were inspected by ultrasonic NDT to verify T10 4.53
the undamaged state. The inspection has been per- T11 4.59
T12 4.60
formed with a USPC 3040 ultrasonic imaging system
(Ingenieurbüro Dr. Hillger, Braunschweig, Germany)
equipped with a 5-MHz transducer (Olympus V309) by
means of immersion testing in pulse-echo Maaß.30 In that technique, a composite laminate is consid-
configuration. ered damage-free if the backwall echo does not drop 6 dB
Figure 3 depicts the amplitude of the backwall echo or more in comparison with a defect-free area within the
in decibels for the ‘‘SHM plate.’’ The specimen has same material, thickness, and stacking sequence. The ampli-
been evaluated for damage following the 6-dB method tude of the backwall echo in Figure 3 remains between 24
described in Schnars and Henrich29 and Brandt and and 210 dB, indicating the pristine state of the specimen.
1906 Structural Health Monitoring 18(5-6)

representing attenuation of X-rays in each pixel. This is

why regions of higher density appear brighter and are
distinguishable from material parts with lower X-ray
density.
Digital radiography was performed in this work for
the ‘‘wave field plate’’ and the ‘‘SHM plate’’ using a
phoenix v|tome|x m (research edition) CT, developed
by GE Sensing & Inspection Technologies GmbH
(Wunstorf, Germany). This instrument is equipped with
both a microfocus X-ray tube with a reflection target
up to a maximum of 240 kV at 320 W and a nanofocus
X-ray tube with a transmission target up to a maximum
of 180 kV at 15 W. The test structures were exposed for
500 ms to a beam, emitted from the microfocus tube
with a source voltage of 50 kV and a source current of
130 mA. According to the US standard ASTM E2597-
07, the X-rays passing through are then detected by a
temperature-stabilized digital GE DXR detector array,
with 2000 3 2000 pixels of size 200 mm on a 400 mm 3
400 mm surface. The signal-to-noise ratio was further
Figure 3. Backwall echo of ultrasonic NDT of the ‘‘SHM plate’’ increased by averaging five frames per image after one
with 12 co-bonded piezoelectric transducers. skip.
Figure 4 shows the X-ray images for the ‘‘wave field
plate’’ and the ‘‘SHM plate.’’ Both images show a black
Baseline state characterization using X-ray testing. Industrial region at the bottom corresponding to the sample
radiography and computed tomography (CT) are well holder to fix the sample during radiographic measure-
established in material science for non-destructive eva- ments. A soft masking tape protects the bottom edge of
luation of technical components to detect defects such the plate, which can be seen as a dark shade around the
as cracks, delaminations, or voids.31 The object of inter- holding device. The piezoelectric transducers can be
est is placed between an X-ray tube and an array detec- clearly identified by their black shape. To provide
tor. As different materials variously absorb the X-rays orientation for the ‘‘SHM plate,’’ an upward arrow of
passing through it, a grayscale image can be generated, copper tape was bonded onto the surface and, hence,

Figure 4. X-ray images of the ‘‘wave field plate’’ (left) and the ‘‘SHM plate’’ (right). The black region at the bottom of both images
corresponds to the sample holder used to fix the samples during X-ray measurements. The ‘‘SHM plate’’ shows an upward arrow of
copper tape in the center of the plate that was bonded to the specimen to provide correct orientation.
Moll et al. 1907

Table 3. Measurement values and test standards used to obtain material properties.

Quantity Value Test standard Source

E11 125.5 6 2.4 GPa DIN EN ISO 527-5 –

E22 8.7 6 0.1 GPa DIN EN ISO 527-5 –
G12 4.135 GPa EN ISO 14129 Petersen et al.32
n12 0.37 6 0.08 DIN EN ISO 527-5 –
n23 0.45 6 0.02 DIN EN ISO 527-5 –
r 1571 6 2 kg/m3 – –

can be recognized in the center of the corresponding

image. In addition, in both images, lines in the direc- Table 4. Values of stiffness tensor for unidirectional Hexply
tion angles 0°, 90°, and 645° occur, indicating the M21/34%/UD134/T700/300 material.
directions of the laminate plies. Note that there are no
Unidirectional laminate (GPa)
further grayscales, which differ from the homogeneous
gray that is generated by the unvarying absorption of C11 = 130.0
the material. Commonly, this is considered as intact in C12 = C13 = 6.1
the range of the given resolution. Therefore, it is note- C23 = 5.2
C22 = C33 = 11.2
worthy that in fact there may be defects smaller than C44 = 3.0
the provided resolution of 120 mm. However, those will C55 = C66 = 4.2
have no influence on the wave propagation due to the
chosen ultrasound wavelength. Hence, both samples
are assumed as intact, that is, suitable to provide base-
lines for further characterization. E11 E22
= ð1Þ
n21 n12

Mechanical parameter measurement and the in-plane relationship of E22 , E23 , and G23 is

In order to implement modeling of guided waves for E22

fiber-reinforced laminates, the stiffness tensor and the G23 = ð2Þ
2(1 + n23 )
density of the material are required. In this study, we
establish the stiffness tensor for the unidirectional plies which leads to the stiffness tensor of the unidirectional
of the Hexply M21/34%/UD134/T700/300 material layer as listed in Table 4.
used for fabrication of the test structure. This is partly Validity of the stiffness tensor was evaluated by ten-
based on measurements and partly based on literature sile tests following DIN EN ISO 527-4 using off-axis
values published for this batch of this material.32 angle measurements at incremental angles of 15° for a
Table 3 lists the measured material properties as well quasi-isotropic plate. Within the margin of error, the
as the corresponding test standards. All material sam- calculated values coincide with the measurement results
ples were conditioned and tested at 23°C and 50% rela- (cf. Figure 5(a)). For the test frequency of 50 kHz, the
tive humidity (RH), to obtain material properties at the phase velocity was extracted from full-field laser vib-
same test conditions as used for measurements of the rometer measurements (see section ‘‘Guided wave mea-
‘‘wave field plate’’ (see section ‘‘Guided wave measure- surements part I: 3D acoustic wave field
ments part I: 3D acoustic wave field measurements’’) measurements’’). The result is compared to phase velo-
and ‘‘SHM plate’’ (see section ‘‘Guided wave measure- cities calculated by finite element modeling following
ments part II: SHM measurements’’). All tests were the approach in Sause35 and shown in Figure 5(b).
conducted in accordance with the corresponding stan- Within the margin of error, these calculations agree
dards, except for the use of digital image correlation reasonably well with the experimental results. The
techniques for strain measurements. To obtain the n23 obtained stiffness tensor may thus be assumed valid for
Poisson’s ratio, the strain field was evaluated at the the test conditions of the ‘‘wave field plate’’ in section
edge of the laminate (cf. approach presented in ‘‘Guided wave measurements part I: 3D acoustic wave
Sause33). field measurements’’ and the ‘‘SHM plate’’ in section
Applying the Maxwell–Betti relationship34 for E11 , ‘‘Guided wave measurements part II: SHM
E22 , n12 , and n21 measurements.’’
1908 Structural Health Monitoring 18(5-6)

plate theory agrees well with the exact 3D theory, espe-

cially at lower frequency-thickness products.37 A polar
representation of the group velocity for three represen-
tative frequencies is shown in Figure 6. Given by the
quasi-isotropic stacking sequence, the velocity is almost
independent of the direction of wave propagation.

Relevance of reference damage

The propagation of guided ultrasonic waves in CFRP
material is adequately described, for example, in
Rose,38 Su et al.,39 and Wang and Yuan.40 For simpli-
(a) fied cases, the interaction of guided waves with damage
can be modeled as well, as shown for a symmetric dela-
mination in Ramadas et al.41 The geometry of a realis-
tic impact damage is more complex, usually described
with a pine-tree-shaped delamination with additional
matrix and fiber cracks.42 The interaction of guided
waves with realistic damages is of high importance for
verification of damage detection and assessment relia-
bility, as the underlying algorithm needs to extract
appropriate features from this interaction.
Giving an example for conventional ultrasonic
inspection, appropriate features are the occurrence of
intermediate echoes between interface echo and back-
(b) wall echo, or the attenuation of the backwall echo.
Both features are very stable and easy to analyze.
Figure 5. (a) Comparison of stiffness values predicted by
This means that reference measurements can be done
classical laminate theory and measurement values at 15° angle
on any structure with similar thickness, attenuation,
increment for static tests and (b) comparison of phase velocity
obtained by finite element modeling at 50 kHz and obtained by and wave velocity. In addition, ultrasound features
full-field laser vibrometer measurements. can be easily reproduced by flat bottom holes or a
separating foil.43
By transferring this example to guided wave propa-
In addition, the theoretical group velocities were gation, the conceptual differences become clearer: As
computed by the well-known global matrix method most state-of-the-art algorithms for damage detection
exploiting third-order plate theory.36 The third-order or assessment are using a baseline subtraction method,

Figure 6. Polar representation of the group velocity (m/s) for (a) the A0 -mode and (b) the S0 -mode. The frequencies shown here
reflect the minimum, intermediate, and maximum frequencies during the experiments of the undamaged ‘‘wave field plate’’ (see Table 5).
Moll et al. 1909

Figure 7. Photograph of the reversible damage model in the

form of an aluminum disk with a diameter of 10 mm coupled to
the structure by tacky tape. White circles indicate four closely
spaced damage positions (cf. Figure 12).

the reference measurement needs to be done on a very

silmilar structure or even the structure itself with same
EOC. Even more important are the features for damage
identification that can be different as soon as the dam-
age type changes. Figure 8. Experimental setup with the scanning laser Doppler
In this work, we employ a reversible defect model vibrometer and the ‘‘wave field plate.’’
where an aluminum disk is mounted on the surface of
the CFRP plate by a tacky tape as introduced in Beard
et al.44 (cf. Figure 7). Although this reference damage is Table 5. Carrier frequencies used for acoustic wave field
simplified with respect to the geometry of an actual measurements.
delamination, its interaction with guided waves behaves Description Carrier frequencies (kHz)
similarly in terms of dedicated features such as change
in time of flight or decrease in amplitude as quantified Undamaged structure 30–100 (Df = 10)
in Bach et al.45 120–240 (Df = 20)
Damaged structure 30, 50, 100, 150, 200, and 250
The underlying idea of using this reversible damage
is to provide measurement data at various places on the The reference damage (see Figure 7) was placed at location x = –
structure so that model-assisted probability of detection 0.059 m and y = 0.045 m (in the coordinate system of the laser Doppler
(MAPOD) techniques can be developed.7,8 vibrometer setup shown in Figure 10).

Guided wave measurements part I: measurement noise, every dataset is averaged 100
3D acoustic wave field measurements times. The temperature was kept constant at room tem-
perature, in this case, 23°C.
As shown in Figure 8, full-wave field measurements Many experiments have been performed on the
were carried out by a 3D scanning laser Doppler vib- ‘‘wave field plate’’ for the undamaged and the damaged
rometer PSV-400-3D from Polytec GmbH specimen as listed in Table 5. The carrier frequencies in
(Waldbronn, Germany) using a measurement range of the experiments range from 30 to 240 kHz in steps of
6200 mm/s. To increase sampling rate to 2.56 MHz, Df . The reference damage shown in Figure 7 is used to
the PCI 6110 National Instruments measurement card model a structure with a defect. For reasons of symme-
was used. Measurement subject was the ‘‘wave field try, only the lower-left quarter of the plate was exam-
plate’’ introduced before with a central transducer ined. Hence, the transducer is located in the upper-
placed exactly in the middle of that structure. The exci- right corner of the measured wave field.
tation signal is a 5-cycle Hann-filtered sine wave ampli- Figure 9 depicts the measured wavenumbers of the
fied to 6150 V. To minimize the influence of random ‘‘wave field plate’’ along 0° direction (see Figure 2) as a
1910 Structural Health Monitoring 18(5-6)

Figure 9. Comparison between theoretically predicted and

experimentally measured dispersion properties of the ‘‘wave
field plate’’ in 0° direction (see Figure 2). Multiple narrowband
measurements were combined to model a broadband pulse. The
experimental dispersion graph was normalized to the maximum
amplitude of the antisymmetric wave mode. Theoretical
dispersion curves were computed with the global matrix
method.36

result of a 2D Fourier transform in relation to theoreti-

cally predicted dispersion curves in the multilayered
laminate. The theoretical dispersion curves are in good
agreement with the experimental dispersion curves. In
addition, mode tuning behavior can be observed which Figure 10. Snapshot of the acoustic wave field excited at 50
describes the capability of piezoelectric transducers to kHz after 155.8 ms. Snapshot (left) for the undamaged structure
and (right) the damaged structure showing antisymmetric wave
excite and detect Lamb waves.46 In the present case,
mode scattering at the defect. The damage position is marked
the maximum of A0 -mode excitation is in the region of by a red circle.
about 80 kHz and decreases to lower and higher fre-
quencies, respectively.
In addition, Figure 10 shows the snapshots measured the wave field measurements described in the previous
at 50 kHz for the intact structure and for the structure section, a Hann-filtered sine wave with 5 cycles was
with the reference damage placed at x = –0.059 m and used with an amplitude of 6100 V. A dedicated device
y = 0.045 m. Antisymmetric wave mode scattering can described in Neuschwander et al.47,48 acquires all
be clearly observed. actuator–sensor pairs in a round-robin fashion, that is,
time-division multiplexing.
Guided wave measurements part II: The process of data acquisition for the SHM mea-
surements is listed in Table 6 and consists of six phases.
SHM measurements In the first phase, 20 baseline measurements of the
Figure 11 shows the experimental setup with the ‘‘SHM intact structure were recorded. After that, the model
plate’’ placed in a climate chamber at a constant tem- defect was placed at 11 different positions on the plate.
perature of 23°C and 50% RH (DIN EN ISO 291). On These measurements correspond to damage positions
the top and at the bottom of the structure, Pt-100 tem- D1 to D11 as shown in Figure 12. In each case, only a
perature sensors were attached to the plates’ surface to single defect model is attached to the structure at the
assess potential temperature gradients. The tempera- same time. Another 20 baseline measurements were
ture sensors were coupled to a PT-104A device (Omega recorded in the third phase, followed by measurements
Engineering GmbH, Deckenpfronn, Germany) that of damage positions D12 to D20 in phase 4. Phase 5
allows temperature measurements with a resolution of consists of 20 additional baseline measurements of the
0.001°C and an accuracy of 0.01°C. One temperature pristine structure. The large number of baseline mea-
measurement was recorded for each dataset. Similar to surements enables the analysis of statistical variations
Moll et al. 1911

Figure 12. Geometry of the transducer positions T1 to T12 and

the defect locations D1 to D28 during the SHM measurements.
Figure 11. Photograph of the ‘‘SHM plate’’ in the climate
chamber, where the tests were performed at 23°C and 50% RH.
Two temperature sensors were installed on the top left and Figure 13 depicts the exemplary guided wave mea-
bottom left to measure the surface temperature of the samples. surements at 60 kHz for the transducer pairs T1  T7
and T6  T12 . Please note that the geometry of the
transducer pairs is symmetric for this case with respect
to the boundary of the plate (cf. Figure 12). This leads
Table 6. Process of data acquisition for the SHM to similar time-domain signals from both transducer
measurements. pairs. The signals marked in blue correspond to 60
unique baseline measurements from the undamaged
Phase Description structure including datasets from phases 1, 3, and 5.
1 20 baseline measurements The signal marked in red corresponds to a measure-
2 Damage scenarios D1 to D11 ment where the model defect was placed at damage
3 20 baseline measurements position D1 (in the path of transducer pairs T6  T12 ).
4 Damage scenarios D12 to D20 It is interesting to see from the differential representa-
5 20 baseline measurements tion shown at the bottom of Figure 13 that the mea-
6 Damage scenarios D21 to D28 , D25 =D28 , and
D14 =D25 =D28 sured signals for the intact structure is almost pure
measurement noise while a distinct waveform can be
observed in the differential signal of the damaged struc-
ture. This waveform corresponds to guided wave scat-
in the baseline measurements, compared with Attarian tering at the defect, mainly based on the interaction of
et al.49 who studied the baseline changes in the case of the fundamental antisymmetric wave mode with the
temperature variations. In the last phase, damage posi- surface damage. This leads to changes in the measured
tions D21 to D28 were measured, plus two additional signals with respect to the baseline signals that can be
datasets with two and three concurrent surface defects. analyzed for damage assessment.
The data acquisition took about 3 days. During that The reconstruction algorithm for probabilistic
time, the temperature in the climate chamber was quite inspection of damage (RAPID), described by Hay et
stable with a maximum temperature variation of 0.49°C al.,52 is well known as a damage localization technique
(sensor on top) and 0.37°C (sensor at the bottom). This in composite materials. This method was used here to
means that temperature compensation techniques, such demonstrate the localization capability of guided waves
as the ones proposed by Croxford et al.50 or Douglass in the proposed CFRP plate. Figure 14 shows the
and Harley,51 are not needed here. two exemplary image reconstruction results based on
1912 Structural Health Monitoring 18(5-6)

Summary
This article introduced the OGW platform with a first
benchmark dataset from quasi-isotropic CFRP plates
with embedded piezoelectric transducers. The article
described the fabrication of the test structures as well
as their characterization by means of ultrasound and
X-ray testing. In addition, the stiffness tensor and the
density of the samples were measured and documented,
and verified by numerical simulations and experimental
measurements.
Acoustic wave field measurements for multiple fre-
quencies were performed on the so-called ‘‘wave field
plate,’’ where a single piezoelectric transducer was
placed in the middle of the test structure. Initial analy-
sis showed good agreement between experimentally
and theoretically predicted dispersion properties.
Subsequent measurements on the ‘‘SHM plate’’ were
conducted at constant temperature conditions in a cli-
mate chamber. The analysis showed the high quality of
Figure 13. (a, b) Sixty Guided wave measurements at 60 kHz
differential signals as well as the possibility for damage
from the intact structure (marked in blue) and one guided wave
measurement from the damaged structure (marked in red) for
localization employing guided wave tomography
two different transducer pairs. (c) Differential signal after techniques.
baseline subtraction. While the differential signals from the All guided wave measurements can be downloaded
intact structure are measurement noise only, a distinct freely from the project’s website (http://www.open-
waveform can be observed in the differential signal due to wave guided-waves.de) in HDF5 format supported by exam-
scattering at the defect corresponding to damage case D1 . All ple scripts. Based on these datasets, existing guided
signals were high-pass filtered using a Butterworth filter with a wave techniques can be comparatively evaluated.
filter order of nF = 3 and a cut-off frequency of 20 kHz. Future work aims at additional datasets for the OGW
platform including measurements with a more complex
40 kHz measurements. It is shown that damage D4 and geometry or recorded under changing EOCs. The
D16 were localized in both cases. In the analysis, only OGW platform is open for contributions from research-
those datasets that have transducer pairs on opposite ers worldwide. Most important are the technical qual-
sides (top and bottom) were included, that is, where the ity, scientific rigor, same file format, and soundness of a
damage is in the direct path of an actuator and its cor- preliminary analysis.
responding sensor.

Figure 14. Guided wave tomographic image reconstruction at 40 kHz for damage D4 (left) and damage D16 (right). The scaling
parameter of the RAPID algorithm was defined here as b = 1:1.
Moll et al. 1913

Acknowledgements Papadopoulos M (eds) Smart intelligent aircraft structures

The authors are all involved in the expert committee (SARISTU). New York: Springer, 2016, pp. 631–642.
Structural Health Monitoring of the German Society for Non- 7. Eckstein B, Fritzen CP and Bach M. Considerations on
Destructive Testing (DGZfP): https://www.dgzfp.de/ the reliability of guided ultrasonic wave-based SHM sys-
Fachausschüsse/Zustandsüberwachung. The authors want to tems for CFRP aerospace structures. In: Proceedings of
thank DGZfP for providing the expert committee platform. the 6th European workshop on structural health monitor-
ing, Dresden, 3–6 July 2012, pp. 957–964.
8. Moix-Bonet M, Eckstein B, Bach M, et al. Damage clas-
Declaration of conflicting interests
sification in aeronautic structures using guided waves. In:
The author(s) declared no potential conflicts of interest with Proceedings of the 11th international workshop on struc-
respect to the research, authorship, and/or publication of this tural health monitoring, Stanford, CA, 12–14 September
article. 2017.
9. Shen Y and Cesnik CES. Modeling guided wave propaga-
Funding tion in composite structures using local interaction simu-
lation approach. In: Liu WK, Hutchings I and Huiskes
The author(s) disclosed receipt of the following financial sup- R. (eds) Computational and experimental methods in struc-
port for the research, authorship, and/or publication of this tures, vol. 8. London: World Scientific Publishing, 2018,
article: J.M. gratefully acknowledges the financial support of pp. 47–91.
this research by the Federal Ministry for Economic Affairs 10. Leckey CA, Wheeler KR, Hafiychuk VN, et al. Simula-
and Energy (grant no. 03SX422B). M.R. gratefully acknowl- tion of guided-wave ultrasound propagation in composite
edges the financial support of the Bremen Economic laminates: benchmark comparisons of numerical codes
Development within ‘‘WERTFASER’’ (QS 1005) as well as and experiment. Ultrasonics 2018; 84: 187–200.
the partial financial support of this work by AiF under IGF- 11. Bulling J, Prager J and Korme F. Application of the
Project Nr. 18651 N/2 through the Research Association Scaled Boundary Finite Element Method (SBFEM) for a
Deutsche Forschungsvereinigung für Mess-, Regelungs- und numerical simulation of ultrasonic guided waves, 2017,
Systemtechnik within the funding scheme Industrielle https://www.ama-science.org/proceedings/details/2603
Gemeinschaftsforschung implemented by the Federal 12. Glushkov EV, Glushkova NV, Eremin AA, et al. Ultra-
Ministry of Economic Affairs and Energy based on a decision sonic guided wave characterization and inspection of
from the German Bundestag. J.K. gratefully acknowledges laminate fiber-reinforced composite plates. In: Parinov
the financial support of this research by the Center for Sensor IA, Chang SH and Topolov VY (eds) Advanced materi-
Systems (ZESS). Additional thanks goes to D. Schmidt als. Cham: Springer, 2016, pp. 449–457.
(DLR) for supporting this work. 13. Samaratunga D, Jha R and Gopalakrishnan S. Wavelet
spectral finite element for modeling guided wave propa-
ORCID iDs gation and damage detection in stiffened composite
panels. Struct Health Monit 2016; 15(3): 317–334.
Jochen Moll https://orcid.org/0000-0003-2299-2250
14. Ostachowicz WM, Kudela P, Krawczuk M, et al. Guided
Marcel Rennoch https://orcid.org/0000-0001-7183-561X
waves in structures for SHM: the time-domain spectral ele-
ment method. Hoboken, NJ: John Wiley & Sons, 2012.
References 15. Schulte R, Fritzen CP and Moll J. Spectral element mod-
elling of wave propagation in isotropic and anisotropic
1. Mitra M and Gopalakrishnan S. Guided wave based
shell-structures including different types of damage. IOP
structural health monitoring: a review. Smart Mater
Conf Ser: Mater Sci Eng 2010; 10(1): 012065.
Struct 2016; 25(5): 053001.
16. Moser F, Jacobs LJ and Qu J. Modeling elastic wave pro-
2. SAE International. Guidelines for implementation of
structural health monitoring on fixed wing aircraft. SAE pagation in waveguides with the finite element method.
paper ARP6461, 2013. NDT&E Int 1999; 32(4): 225–234.
3. Aldrin J, Medina E, Lindgren E, et al. Protocol for relia- 17. Michaels JE. Ultrasonic wavefield imaging. In: Ida N and
bility assessment of structural health monitoring systems Meyendorf N (eds) Handbook of advanced non-destructive
incorporating model-assisted probability of detection evaluation. Cham: Springer, 2018, pp. 1–32.
(MAPOD) approach. In: Proceedings of the 8th interna- 18. Yu L, Tian Z, Li X, et al. Core–skin debonding detection in
tional workshop on structural health monitoring, Stanford, honeycomb sandwich structures through guided wave wave-
13–15 September 2011, pp. 1–8. Lancaster, PA: DEStech field analysis. J Intel Mater Syst Struct. Epub ahead of print
Publications. 27 February 2018. DOI: 10.1177/1045389X1875 8180.
4. US Department of Defense. Military handbook 1823: 19. Harley JB and Chia CC. Statistical partial wavefield ima-
nondestructive evaluation of system reliability assessment. ging using Lamb wave signals. Struct Health Monit 2018;
Washington, DC: US Department of Defense, 2009. 17(4): 919–935.
5. Berens AP. NDE reliability data analysis. ASM Hand- 20. Keshmiri Esfandabadi Y, De Marchi L, Testoni N, et al.
book 1989; 17: 689–701. Full wavefield analysis and damage imaging through
6. Buethe I, Dominguez N, Jung H, et al. Path-based MAPOD compressive sensing in lamb wave inspections. IEEE T
using numerical simulations. In: Wölcken PC and Ultrason Ferr 2018; 65(2): 269–280.
1914 Structural Health Monitoring 18(5-6)

21. De Marchi L, Marzani A, Moll J, et al. A pulse coding 38. Rose JL. A baseline and vision of ultrasonic guided wave
and decoding strategy to perform Lamb wave inspections inspection potential. J Press Vess T ASME 2002; 124(3):
using simultaneously multiple actuators. Mech Syst Sig- 273.
nal Pr 2017; 91: 111–121. 39. Su Z, Ye L and Lu Y. Guided Lamb waves for identifica-
22. Flynn EB, Chong SY, Jarmer GJ, et al. Structural ima- tion of damage in composite structures: a review. J Sound
ging through local wavenumber estimation of guided Vib 2006; 295(3–5): 753–780.
waves. NDT&E Int 2013; 59: 1–10. 40. Wang L and Yuan F. Group velocity and characteristic
23. Kudela P, Radzienski M, Ostachowicz W, et al. Struc- wave curves of Lamb waves in composites: modeling and
tural health monitoring system based on a concept of experiments. Compos Sci Technol 2007; 67(7–8):
Lamb wave focusing by the piezoelectric array. Mech 1370–1384.
Syst Signal Pr 2018; 108: 21–32. 41. Ramadas C, Balasubramaniam K, Joshi M, et al. Inter-
24. Alguri KS, Melville J and Harley JB. Baseline-free guided action of the primary anti-symmetric Lamb mode (A0)
wave damage detection with surrogate data and diction- with symmetric delaminations: numerical and experimen-
ary learning. J Acoust Soc Am 2018; 143(6): 3807–3818. tal studies. Smart Mater Struct 2009; 18(8): 085011.
25. Memmolo V, Maio L, Boffa ND, et al. Damage detec- 42. Abrate S. Impact on laminated composite materials. Appl
tion tomography based on guided waves in composite Mech Rev 1991; 44(4): 155.
structures using a distributed sensor network. Optical 43. Sedov A, Schmerr LW and Song SJ. Ultrasonic scattering
Eng 2015; 55(1): 011007. by a flat-bottom hole in immersion testing: an analytical
26. Lee SJ, Gandhi N, Hall JS, et al. Baseline-free guided model. J Acoust Soc Am 1992; 92(1): 478–486.
wave imaging via adaptive source removal. Struct Health 44. Beard SJ, Kumar A, Qing X, et al. Practical issues in real-
Monit 2012; 11(4): 472–481. world implementation of structural health monitoring sys-
27. Moll J, Schulte R, Hartmann B, et al. Multi-site damage tems. San Diego, CA: SPIE, 2005, p. 196.
localization in anisotropic plate-like structures using an 45. Bach M, Pouilly A, Eckstein B, et al. Reference damages
active guided wave structural health monitoring system. for verification of probability of detection with guided
Smart Mater Struct 2010; 19(4): 045022. waves. In: Chang FK and Kopsaftopoulos F (eds) Struc-
28. Wierach P, Monner HP, Schoenecker A, et al. Applica- tural health monitoring. Lancaster, PA: DEStech Publica-
tion-specific design of adaptive structures with piezoceramic tions, 2017.
patch actuators. San Diego, CA: SPIE, 2002, pp. 333–341. 46. Giurgiutiu V. Tuned lamb wave excitation and detection
29. Schnars U and Henrich R. Applications of NDT methods with piezoelectric wafer active sensors for structural
on composite structures in aerospace industry. In: Pro- health monitoring. J Intell Mater Syst Struct 2005; 16(4):
ceedings of the conference on damage in composite materi- 291–305.
als, Stuttgart, 2006, pp. 1–8. 47. Neuschwander K, Moll J, Memmolo V, et al. Simulta-
30. Brandt C and Maaß P. A state space approach for the neous load and structural monitoring of a carbon fiber
non-destructive evaluation of CFRP with ultrasonic test- rudder stock: results from a quasi-static tensile test. J
ing. In: Proceedings of the 7th international symposium on Intell Mater Syst Struct. Epub ahead of print 26 March
NDT in aerospace, 2016, pp. 1–8. 2018. DOI: 10.1177/0954406218764226.
31. Singhal A, Grande JC and Zhou Y. Micro/nano-CT for 48. Neuschwander K, Shrestha A, Moll J, et al. Multichannel
visualization of internal structures. Microscop Today device for integrated pitch catch and EMI measurements
2013; 21(2): 16–22. in guided wave structural health monitoring applications.
32. Petersen E, Cuntze R and Hühne C. Experimental deter- In: Proceedings of the 11th international workshop on
mination of material parameters in Cuntze’s failure- structural health monitoring, Stanford, CA, 12–14 Sep-
mode-concept-based UD strength failure conditions.
tember 2017, pp. 1723–1730. DEStech Publications, Inc.
Compos Sci Technol 2016; 134: 12–25.
49. Attarian VA, Cegla FB and Cawley P. Long-term stabi-
33. Sause MGR. Digital image correlation. In: Sause MGR
lity of guided wave structural health monitoring using
(ed.) In situ monitoring of fiber-reinforced composites, vol.
distributed adhesively bonded piezoelectric transducers.
242. Cham: Springer, 2016, pp. 57–129.
Struct Health Monit 2014; 13(3): 265–280.
34. Schürmann H. Konstruieren mit Faser-Kunststoff-Verbun-
50. Croxford A, Moll J, Wilcox P, et al. Efficient temperature
den. 2nd ed. Berlin: Springer, 2007.
compensation strategies for guided wave structural health
35. Sause MGR. Acoustic emission. In: Sause MGR (ed.) In
monitoring. Ultrasonics 2010; 50(4–5): 517–528.
situ monitoring of fiber-reinforced composites, vol. 242.
51. Douglass ACS and Harley JB. Dynamic time warping
Cham: Springer, 2016, pp. 131–359.
temperature compensation for guided wave structural
36. Torres Arredondo M, Ramirez Lozano M and Fritzen
health Monitoring. IEEE T Ultrason Ferr 2018; 65(5):
CP. DispWare toolbox—a scientific computer program for
851–861.
the calculation of dispersion relations for modal-based
52. Hay TR, Royer RL, Gao H, et al. A comparison of
acoustic emission and ultrasonic testing (technical report).
embedded sensor Lamb wave ultrasonic tomography
Siegen: University of Siegen, 2011.
approaches for material loss detection. Smart Mater
37. Torres Arredondo M. Acoustic emission testing and
Struct 2006; 15(4): 946–951.
acousto-ultrasonics for structural health monitoring. PhD
Thesis, University of Siegen, Siegen, 2013.