J Mater Sci: Mater Electron (2022) 33:3348–3360

Study on preparation and simple calibration system

of acoustic emission sensor based on epoxy resin
Rui Guo1, Shifeng Huang1,* , and Fen Liu1

1
Shandong Provincial Key Laboratory of Preparation and Measurement of Building Materials, University of Jinan, Jinan 250022, P.
R. China

Received: 12 August 2021 ABSTRACT

Accepted: 5 December 2021 An epoxy-based acoustic emission sensor with low secondary reflection, high
Published online: precision, and strong corrosion resistance was prepared by using the acoustic
21 January 2022 characteristics of different combinations of sensing element, matching layer, and
backing layer. A simple calibration system for acoustic emission sensors is set
Ó The Author(s), under up, and the static and dynamic characteristics of the sensor are tested. The
exclusive licence to Springer experimental results show that the static and dynamic characteristics of the
Science+Business Media, LLC, sensor are obtained by analyzing the characteristic parameters of the acoustic
part of Springer Nature 2022 emission signal. In this static characteristic, sensitivity refers to the change in
acoustic emission energy value caused by every 1 mm change in the falling
distance of the zirconia ball, which is represented by the symbol E/mm. The
linear error of the sensor is 16.5%, the sensitivity is 444.05 E/mm, the return
error is 11.2%, and the non-repeating error is 19%. In order to evaluate the
characteristics of signal superposition, the concept of signal superposition error
value (DHt) is proposed based on error value between the sum of each impact
component and the total impact quantity. The signal stacking property of the
epoxy-based acoustic emission sensor is tested by using this index, and calcu-
lated DHt value is 0.059. This method can easily calibrate the acoustic emission
sensor and provide a new basis for the accuracy of information received by the
sensor. It is also found that the secondary reflexion signal of the polymer
acoustic emission sensor is smaller, the resonant frequency is lower, and it has
better corrosion resistance to rain water.

1 Introduction while the secondary acoustic emission source refers

to a kind of elastic wave that has no direct contact
Acoustic emission (AE) is a stress wave with typical with strain and fracture, such as fluid leakage, fric-
AE and secondary AE as sound sources. The phe- tion, impact, and combustion [1]. AE sensor is a
nomenon of rapid release of local energy caused by conversion device which converts the received elastic
strain and cracks is called a typical acoustic emission, wave signal into electrical signal which is easy to

Address correspondence to E-mail: [email protected]

https://doi.org/10.1007/s10854-021-07534-6
J Mater Sci: Mater Electron (2022) 33:3348–3360 3349

analyze and process [2]. It is a core part of the accuracy of calibration. In 2018, Markus et al. [20]
acoustic emission detection technology and determi- simulated wave propagation in media and related
nes the accuracy and reliability of original acoustic sensor calibration problems by modeling. However,
emission information [3]. The discovery of the 1950s these calibration methods are complicated and the
Kaiser effect opened research of the piezoelectric calibration process requires a high cost of equipment.
effect. Subsequently, sensitive materials for AE sen- In this paper, an epoxy-based AE sensor with
sors were developed using this effect. Since then, strong corrosion resistance and low secondary
commercial acoustic emission devices have been reflexion wave was prepared by comparing the
applied to detect the acoustic emission signals gen- effects of different material combinations on the
erated by the strain process of metal materials [4, 5]. acoustic performance of the sensor. The AE source is
In recent years, AE sensors have been widely used in simulated by the free falling motion of zirconia
fatigue damage of mechanical components [6, 7], ceramics ball, and the static and dynamic character-
medical field, and concrete evaluation and analysis istics of the sensor are calibrated by the simple
[8, 9]. M.S.H et al. [10] used AE sensors to measure change of output with input, which provides a new
the signal frequency during machining. Huang et al. basis for verifying the accuracy and stability of the
[9] used an embedded acoustic emission sensor to received signal of the AE sensor.
monitor cement mortar. Xu et al. [11] studied the
influence of different backing layers on the acoustic
characteristics by preparing acoustic emission sen- 2 Experimental process
sors with different backing layers. Reza et al. [12]
used acoustic emission technology to detect the cor- 2.1 Selection of materials
rosion of concrete pipes and found that acoustic
emission parameters changed significantly. AE sensors are generally composed of a sensing ele-
The sensitivity, linearity, repeatability, return error, ment, a matching layer, a backing layer, and a
and other static characteristics of the sensor are the package shell. The sensor prepared in this paper
key factors of AE detection. The evaluation of these adopts commercial lead zirconate titanate ceramic
performances of the sensor is based on the calibration (PZT, U12 9 6 mm) as the sensitive element of
method of the sensor, in which the piezoelectric epoxy-based AE sensor and commercial alumina
acoustic emission sensor is generally calibrated on (Al2O3) ceramic as the matching layer. The detailed
the solid medium, and the elastic wave in the solid performance indexes are shown in Table 1. Epoxy
medium is mainly body wave and surface wave [13]. resin with purity of 99.9% was used as shell material.
Acoustic emission signals will reflect and refraction The molecular formula and working conditions of
when they pass through the boundary of different epoxy resin are shown in Table 2. AE acquisition
materials, which leads to the complexity of acoustic system (PAC Micro-II, USA) was used to test the
emission signals in actual propagation [14]. There- acoustic attenuation characteristics of insulation foam
fore, surface waves and body waves are generally board. The AE source (pencil lead U0.5 mm,
used to calibrate sensors. The calibration of acoustic L = 2.5 mm, h = 30) was simulated by breaking the
emission sensors first appeared in Japan in 1976 [15]. lead at the interface between the aluminum plate and
In 1998, Hatano et al. [16] developed a reciprocity the foam plate. The acoustic emission signals
method based on the principle of piezoelectric sensor received by commercial AE sensors at different
sensing and drive consistency on a steel block to positions on the foam board were used to evaluate
calibrate sensitivity and other characteristics of the acoustic attenuation characteristics of the insu-
acoustic emission sensors. In 2012, Mclaskey et al. lating foam board, as shown in Fig. 1a. As can be
[17] found that sensors with built-in amplifiers could seen from Fig. 2b, with the increase of the distance
not be calibrated using reciprocity methods. In 2015, between the sensor and the sound source, the
Burks et al. [18] tested the acoustic characteristics of a received AE signal received also decreases. When the
standard sensor by directly coupling it face to face sensor is 12 cm away from the sound source, the
with the sensor under test. In 2016, Kober et al. acoustic emission signal is about 37 dB, which is very
[19]used time reverse acoustics to determine the close to the noise signal (35 dB). This means that
in situ calibration method, which improved the when the foam plate is placed on the aluminum plate,
3350 J Mater Sci: Mater Electron (2022) 33:3348–3360

Table 1 Performance indexes of each ceramic material

Material Density (103 kg m-3) Piezo voltage constant (10-3 V m N-1) Curie Temperature (Tc/°C) Hardness (HRa)

PZT 7.418 19.2 340 –

Alumina ceramics 3.555 – – 89

Table 2 Primitive agents needed for the experiment

Drug name Molecular Applicable temperature Mixing temperature (°C) Manufacturer

weight (°C)

Epoxy Resin (C11H12O3)n C5 12–20 Beijing Jinhong Century Technology Co. Ltd
Curing agent C4H13N3 C5

Fig.1 Acoustic attenuation characteristics of insulation foam board, a the field test picture, b the curve of detecting signal strength as a
function of sound source distance

the distance of 12 cm can almost absorb the acoustic the PZT, the received sound signal is distorted. Sec-
emission signal generated by the sound source (bro- ond, the PZT resonates with the received acoustic
ken lead signal). In order to prepare an acoustic signal. Third, PZT self-resonant wave and sound
emission sensor with excellent performance, the same wave propagate forwards together, and after reach-
acoustic emission signal is received by four different ing the critical surface between PZT and air, the
combinations of composite components, and the reaction occurs at the interface, in which part of the
waveform changes of received signals are compared, elastic wave is reflected back and received by the
as shown in Fig. 2. After receiving acoustic emission sensitive element, forming secondary wave. As can
signal 192 s, the pure PZT element received an be seen from Fig. 2b, the intensity of the secondary
obvious secondary wave signal with a signal inten- reflexion signal decreases slightly after the addition
sity of 9.13 mV, as shown in Fig. 2a. There are three of the matching layer. The reason for this phe-
reasons for the sensor to receive secondary AE sig- nomenon is that the addition of the matching layer
nals. First, the acoustic emission signal is transmitted makes the acoustic impedance between the PZT
from the aluminum plate to the PZT ceramic sheet in ceramic wafer and the aluminum plate tends to
the form of elastic wave [21]. Due to the mismatch of match, so that the signal can be better transmitted to
acoustic impedance between the aluminum plate and the sensing element. For the sensor prepared with
J Mater Sci: Mater Electron (2022) 33:3348–3360 3351

20 10.5 9.13
20 10.5
(a) 9.0
7.5
6.0
Rebound wave
(b) 9.0
7.5
6.0 3.28
15 4.5 15 4.5
3.0
3.0 1.5
1.5 0.0

Voltage (mV)
0.0 -1.5
10 -1.5
10
Voltage (mV)

-3.0
-3.0 -4.5
-4.5 -6.0 Rebound wave
-6.0 -7.5
-7.5 -9.0
5 -9.0
0 50 100 150 200 250 300 350 400
5 0 50 100 150 200 250 300 350 400

0 0

-5 -5

-10 -10
-125 0 125 250 375 500 625 750 875 1000 -125 0 125 250 375 500 625 750 875 1000
Time (μs) Time (μs)
20 9.0
20 10.5
(c) 7.5
6.0 4.80
(d) 9.0
7.5
6.0
15 4.5
3.0 15 4.5 2.12
1.5 3.0
0.0 1.5
Voltage (mV)

0.0

Voltage (mV)
-1.5
10 -3.0
-4.5
10 -1.5
-3.0
-4.5
-6.0 Rebound wave -6.0 Rebound wave
-7.5 -7.5
5 -9.0
0 50 100 150 200 250 300 350 400
5 -9.0
0 50 100 150 200 250 300 350 400

0 0

-5 -5

-10 -10
-125 0 125 250 375 500 625 750 875 1000 -125 0 125 250 375 500 625 750 875 1000
Time (μs) Time (μs)
Fig. 2 Sensor waveform in different combinations, a the pure PZT ceramics, b the Al2O3 and PZT composition, c the Al2O3 plus PZT plus
insulation foam board composition, d the PZT plus Al2O3 plus insulation foam board and cotton

alumina as the matching layer and foam as the of PZT plus Al2O3 plus insulating foam board and
shielding layer, the secondary reflexion signal cotton was adopted in the final sensor preparation.
received is reduced to 3.28 mV, as shown in Fig. 2c.
The main reason is that the sound attenuation is 2.2 Preparation of acoustic emission
serious when the sound wave travels in the insula- sensors
tion foam, and the secondary signal generated by
PZT continues to propagate backwards in the form of As verified in the previous section, it has been found
elastic waves. After transmitting to the foam, a large that when the insulation foam board and cotton were
amount of energy is absorbed, which makes the sig- used as the backing layer, the secondary rebound
nal reflected to PZT smaller. The sound attenuation of wave was very small and the acoustic emission signal
the foam is shown in Fig. 1. As can be seen from received was more pure. Therefore, the insulation
Fig. 2d, the addition of a small amount of cotton foam board is processed into a round barrel shape
between the PZT and the foam further improves the with an inner diameter slightly larger than the outer
acoustic attenuation capacity, and almost no sec- diameter of PZT ceramic, and a layer of cotton is
ondary reflexion signals are received, thus improving attached to the inner wall of the foam board as the
the static and dynamic characteristics of the sensor backing layer. The matching layer sensing element is
[22, 23]. In order to reduce the influence of the assembled together with the backing layer, and a
vibration wave of PZT ceramic on the sensor signal, coaxial cable is welded on the sensing element for
insulating foam board and cotton were selected as the transmitting the acquisition signal. The assembled
backing materials of the sensor, and the combination components are fixed in the square mold, and the
epoxy slurry is poured into the mold. After curing,
3352 J Mater Sci: Mater Electron (2022) 33:3348–3360

the epoxy-based acoustic emission sensor is obtained, Formula 2, where Sn is the sensitivity, DY is the
and the technological process is shown in Fig. 3. change value of the output on the Y axis, and DL is
the change value of the input on the X axis.
2.3 Performance of acoustic emission DY
sensors Sn ¼ ð2Þ
DL
Using zirconia sphere as acoustic emission source, If the output produced by the change of the sensor
the use of the output energy of signals is (AE sensor) input from small to large is inconsistent with the
to the input (the distance the ball off the ground) into output produced by the change of the sensor input
the degree of linearity of sensor linearity, sensitivity, from large to small, it means that the sensor has a
hysteresis, and repeatability of static characteristic certain return error (cH), and its calculation formula is
and the superposition of frequency stability and shown in Formula 3, where cH is the return error,
dynamic characteristics such as calibration, as shown DHmax is the maximum output difference, and YFS is
in Fig. 4. the full output range.
The ideal input–output characteristics of the sensor DH max
should be linear, but there is some non-linearity in cH ¼   100% ð3Þ
YFS
practice. For the linearity of the sensor, zero-crossing
rotation fitting and endpoint translation fitting are The sensor input varies from small to large (or
used for the input–output curve. The evaluation from large to small), and the consistency of the input–
index was linear degree (cL), and its calculation for- output curve obtained after many tests represents the
mula was shown in Formula 1, where cL is the non- repeatability of the sensor. The index of repeatability
linear error, DLmax is the maximum nonlinear is generally characterized by cR, and the smaller the
absolute error, and YFS is the output full scale range. cR, the better the repeatability of the sensor, and the
DLmax more accurate the acoustic emission signal received.
cL ¼   100% ð1Þ The calculation formula is shown in Formula 4,
YFS
where DRmax is the maximum output difference, and
Generally, the steeper the input–output character- YFS is the full output range.
istic curve is, the greater the sensitivity will be, while DRmax
the flatter the sensitivity will be, the smaller the cR ¼   100% ð4Þ
YFS
sensitivity will be. Sensitivity (Sn) refers to the ratio of
sensor output change to input change under The static characteristics of the sensor directly
stable state, and its calculation formula is shown in determine the detection accuracy of the sensor, and

Fig. 3 Flow chart of sensor manufacturing process

J Mater Sci: Mater Electron (2022) 33:3348–3360 3353

Fig. 4 Sensor calibration

system, a the experimental
device diagram, b the
monitoring schematic diagram

because the actual engineering conditions are com- when releasing zirconia balls to ensure that the balls
plex, the received acoustic emission signal is also a are protected from external forces and do not touch
complex excitation signal, which requires the sensor anything. The experimental data were fitted and
not only to have a good static characteristics but also calculated, and it was found that the input (ball
a good dynamic characteristics [24]. For example, the release distance) and output (acoustic emission
prepared sensor has certain superposition, which can energy) had a good linear relationship within a cer-
decompose the received complex acoustic emission tain range. When the distance between the ball and
signal into several simple excitation signals, thus the measured object was greater than or equal to
completing the processing of acoustic emission signal 20 cm, the acoustic emission energy received reached
data more easily and conveniently. the state of full scale, and the energy value was
6.55 9 10–5 V s, as shown in Fig. 5c. The static char-
acteristics of the AE sensor are obtained by using the
3 Experimental results and discussion detection and calculation of the device, as shown in
Fig. 6. Figure 6a shows the processing diagram of the
3.1 Static characteristic analysis tangents and secants of the sensor input–output
curve. It is found that the value of DLmax is 10837.31,
The linearity of the sensor was tested by the free and the nonlinear error (cL) of the sensor is calculated
falling motion of the zirconia ball, as shown in by using Formula 1 as 16.5%. Similarly, the input–
Fig. 5a. It is important to wear insulating gloves
3354 J Mater Sci: Mater Electron (2022) 33:3348–3360

Fig. 5 Test of input–output relationship of AE sensor, a the schematic diagram of test method, b the schematic diagram of zirconia pellets,
c the input–output characteristic fitting diagram, d the corresponding relation between impact force and acoustic emission energy

70k (a) 6.55×10-5 V·s 70k (b)

60k 60k
Energy (10-9 V·s)
Energy (10-9 V·s)

50k 50k

40k 40k
Δ Lmax

30k 30k ΔY=9185.164

20k 20k
ΔL=20.685
10k 10k

0 50 100 150 200 250 300 -25 0 25 50 75 100 125 150 175 200
Distance (mm) Distance (mm)

20 (c) Positive direction of travel 70k (d) YFS

18 Direction of reverse travel
16 60k ΔRmax
Energy (10-9 V·s)

14
Distance (cm)

50k
12 1
40k 2
10
3
8 30k 4
YFS
6 5
20k 6
4 ΔHmax
2 10k
0
10k 20k 30k 40k 50k 60k 70k 0 2 4 6 8 10 12 14 16

Energy (10 V·s) -9 Distance (cm)

Fig. 6 Static characteristics of polymer-based AE sensor, a the linearity, b the sensitivity, c the return trip error, d the repeatability
J Mater Sci: Mater Electron (2022) 33:3348–3360 3355

output curve is also treated with infinitesimals, and the frequency retention and stacking property were
then the sensitivity of the sensor is calculated as tested and verified. The test results show that the
444.05 E/mm through formula 2, as shown in Fig. 6b. received frequency information has a specific fre-
Acoustic emission energy changes with the increase quency retention under certain conditions, as shown
of input and the decrease of input are shown in in Fig. 7. The signal that the zirconia sphere hits the
Fig. 6c, and the return error value of the sensor is aluminum plate after being released at 25 cm from
11.2% after further calculation by using Eq. 3. The the aluminum plate is used as the signal source. The
sensor repeatability experiment is tested by control- signal was emitted three times intermittently at this
ling the change of the input signal of the sensor from position. At this time, the AE sensor receives three
zero to full scale. This experiment is repeated for 6 excitation signals of about 93 dB, and the frequency
times to change from zero to full scale, and the sensor characteristics at this time are obtained according to
repetition curve is obtained, as shown in Fig. 6d. The Fourier transform, as shown in Fig. 7a. Figure 7b
maximum error point is taken from the figure and the shows that the same zirconia sphere is released 10
non-repetition error (cr) value of the sensor is calcu- times at a position 25 cm away from the aluminum
lated by using Formula 4 as 19%. The impact force of plate, which means that 10 excitation signals of about
the ball on the aluminum plate is calculated accord- 93 dB are applied to the AE sensor, and the frequency
ing to the free fall motion of the zirconia sphere at characteristics at this time are obtained according to
different heights from the aluminum plate according the Fourier transform. Finally, the zirconia sphere is
to Formula 5, where M is the mass of the ball, and released for 7 times at any position, that is, 7 random
V is the velocity of the ball when it falls to the signals are given to the sensor. At this time, the
ground, F is the impact force of the aluminum plate, sound pressure level and spectrum characteristics of
t is the contact time between the ball and the alu- acoustic emission signals are shown in Fig. 7c. We
minum plate (set as 1), V0 is the initial velocity of the find that the peak positions and their variations in the
ball (set as 0 in this experiment), g is the acceleration spectra obtained in (a) and (b) are indeed almost the
of gravity, set as 9.8 m/s, and s is the height between same by comparing the entire Fig. 7. The reason for
the ball and the aluminum plate. It is found that there this phenomenon is that the input signal is a specific
is a certain correspondence between force and dis- signal with a fixed frequency (all of them are about
tance as well as between force and acoustic emission 93 dB excitation signals), which indirectly proves that
energy. With the increase of impact force, the energy the sensor has frequency retention under this condi-
of acoustic emission signal also shows an upward tion. However, in Fig. 7c, compared with Fig. 7a and
trend, as shown in Fig. 5d. It can be seen from the b, it is found that the peak value of frequency in
figure that there is a linear correspondence between Fig. 7c is significantly changed, especially the peak
the force and the distance as a whole, and there is a value in the high frequency position disappears
lowest point near 13 cm, which is caused by the seriously. The reason for this phenomenon may be
contact of the steel ruler in the release process of the that the input signals are random excitation signals
ball. This error value will not affect the overall trend with different frequency domain characteristics, and
of the experiment. their frequency domain information overlaps and
MV ¼ F  t cancels each other, resulting in the sensor receiving
spectral signals that are different from those in Fig. 7a
V 2  V 0 2 ¼ 2gs ð5Þ and b. The conclusion is that as long as the input
signal has a specific frequency, no matter how much
3.2 Dynamic characteristic analysis the input signal is, the acoustic emission signal
received by the AE sensor has this frequency char-
The ideal dynamic characteristic of the sensor mainly acteristic. There is a certain relationship between the
refers to the change of the output with the change of acoustic emission impact generated by the release of
the input without distortion, which generally has two a single zirconia sphere at 5 cm and 10 cm from the
important characteristics of superposition and fre- measured object and the number of collisions gener-
quency retention. In order to know more about the ated by the simultaneous release of two identical
dynamic characteristics of epoxy-based AE sensor, balls at 5 cm and 10 cm. The superposition of the AE
sensor was tested by comparing the corresponding
3356 J Mater Sci: Mater Electron (2022) 33:3348–3360

Fig. 7 Frequency retention test of AE sensor, a frequency signals of about 93 dB, c frequency characteristics generated by 7
characteristics generated by three excitation signals of about excitation signals of arbitrary intensity
93 dB, b frequency characteristics generated by 10 excitation

relationship between the acoustic emission impact emission signal impact number when each excitation
number generated by the three. The test results are signal is input separately and the acoustic emission
shown in Fig. 8. Figure 8a shows the figure of the signal impact number when each excitation signal is
field experimental device, where 1–3 is the acoustic input simultaneously is used to characterize. The
emission impact number generated by the release of a deviation is represented by signal superposition
single ball at the height of 10 cm and 5 cm, and the deviation (DHt) parameter, and the calculation
acoustic emission impact number generated by the method is shown in Eq. 6 (taking two excitation sig-
simultaneous release of two identical balls at the nals as an example). The DHt value calculated in this
position of 10 cm and 5 cm, respectively, corre- experiment is 0.059. The self-made AE sensor has a
sponding to the 1–3 in Fig. 8b in turn. It can be seen smaller DHt value and better signal superposition,
from the figure that the number of impacts at position where DHt is the signal stacking deviation value,
1 is 5, the number of impacts at position 2 is 11, and Ht1,2 is the acoustic emission signal collision number
the number of impacts at position 3 is 17, indicating when two excitation signals are input at the same
that there is an approximate superposition relation- time, Ht1 is the acoustic emission signal collision
ship when the two balls are released simultaneously number when the first excitation signal is input alone,
and the two balls are released separately. In order to and Ht2 is the acoustic emission signal collision
evaluate the signal superposition of the AE sensor, number when the second excitation signal is input
the deviation between the sum of the acoustic alone.
J Mater Sci: Mater Electron (2022) 33:3348–3360 3357

Fig. 8 AE sensor stacking test

diagram, a the experimental
device diagram, b the impact
time diagram; c the sensor
frequency characteristics

Ht1;2  Ht1  Ht2 received by the epoxy-based sensor is about 92 dB,

DHt ¼ ð6Þ which is not much different. The noise thresholds for
Ht1;2
epoxy-based AE float between 34 and 36 dB, while
the noise thresholds for commercial sensors float
3.3 Epoxy-based AE sensor application between 37 and 39 dB. Thus, an epoxy-based AE
performance testing sensor can receive weak acoustic emission signals
with smaller amplitudes, which would be difficult for
In order to evaluate the application performance of commercial sensors to distinguish. The two sensors
the epoxy-based AE sensor, the acoustic performance receive the signal energy difference is bigger, but in
of the sensor was compared with that of commercial the monitoring of stress concentration of cracks pro-
sensors, and the performance change experiment of duced and extended in practical engineering, can
the epoxy-based AE sensor under rainwater corro- only play a role in advance warning, received signal
sion was simulated. Figure 9 shows the performance energy is independent of the size, from the point of
difference between epoxy-based AE sensor and view, this kind of epoxy-based AE sensors can be
commercial AE sensor. The experimental device used in practical engineering structure health moni-
consists of AE acquisition instrument, signal ampli- toring. Contrast two sensor signals waveform fig-
fier, aluminum plate epoxy-based AE sensor, com- ure found that epoxy-based AE sensor voice signal
mercial AE sensor, and screen display, as shown in attenuation, and commercial sensors in 0.312 ms
Fig. 9a. In this experiment, the signal when the pencil location has obvious secondary reflexion wave as
broke was used as the analog signal, and the pencil shown in Fig. 9e, which may be commercial sensor is
lead at the center of the two sensors was used for 5 sensitive to metal shell reflected signal and vibration
times. The amplitude and energy of the signal caused by the components themselves, this will affect
received by the two sensors are shown in Fig. 9c and the accuracy of the received signal. This is one of the
d, respectively. The amplitude signal test results reasons the signals received by commercial sensors
show that the broken lead signal received by the are of high energy. By comparing the frequency
commercial sensor is about 93 dB, while the signal information of the two, it can be seen that the
3358 J Mater Sci: Mater Electron (2022) 33:3348–3360

Fig. 9 Comparison of the performance of self-made AE sensor and commercial sensor, a the diagram of experimental equipment, b the
simulation of acoustic emission sources, c amplitude comparison, d energy comparison, e waveform comparison; f frequency comparison

resonant frequency of the commercial sensor is 161 K, terms of the corrosion resistance of the epoxy-based
while the lower resonant frequency of the commer- AE sensor, the sensor can be used as the coarse
cial sensor is 130 K, as shown in Fig. 9f. This AE aggregate of concrete embedded in the concrete for
sensor with lower frequency is more suitable for long-term structural health monitoring of civil engi-
structural health monitoring of civil engineering such neering buildings.
as concrete. In this paper, the epoxy-based AE sensor
was immersed in water, and the sensor was taken out
every other week to detect the change of its ampli- 4 Conclusion
tude signal. The experimental results are shown in
Fig. 10. The results show that the amplitude of the This paper includes three aspects: material selection,
signal received by the sensor decreases gradually sensor preparation, and simple calibration test of the
with the increase of the time of the sensor immersed sensor. Through the comparison of different material
in water, but the signal reduction is very small, only combinations, the acoustic characteristics under dif-
0.35 dB decreased after 6 weeks of immersion, which ferent combinations were studied. Finally, the com-
has little impact on the performance of the sensor. In bination of PZT plus Al2O3 plus insulating foam
board plus cotton composition was selected to
J Mater Sci: Mater Electron (2022) 33:3348–3360 3359

92.1 This work was awarded by the National Natural

Science Foundation of China [Grant Numbers
92.0 U1806221, 51208228], the innovative team projects of
institutions of higher learning, piezoelectric com-
Amplitude (dB)

91.9 posite materials and engineering health monitoring

[Grant Number 2019 GXRC017].
91.8

91.7
References
1. J. Yu, P. Ziehl, F. Matta, A. Pollock, Constr. Steel Res. 86,
91.6 85–91 (2013)
2. N. Collier, P. Debreyne, G. Delaplace, B. Chen et al., Ultra-
1 2 3 4 5 6
sonics 89, 187–194 (2018)
Time (week)
3. P. Marina, T. Martinb, D.L. Tunnicliffeb, Sens. Actuators A.
Fig. 10 Simulation test of the self-made AE sensor against rain 99(3), 236–243 (2001)
water corrosion 4. H.L. Dunegan, D.O. Harris, C.A. Harris, Eng. Fract. Mech. 1,
105–110 (1968)
prepare the AE sensor. The secondary reflexion wave 5. H.L. Dunegan, D.O. Harris, J. Ultrason. 7(3), 160–166 (1969)
of the sensor prepared is small, which reduces the 6. E. Caso, F. Alfonso, P. Garcia et al., Appl. Acoust. 159,
influence of the vibration of the sensing element on 107092 (2020). https://doi.org/10.1016/j.apacoust.2019.
the acoustic emission signal received, and further 107092
improves the accuracy of the sensor. The dynamic 7. M. Hamel, A. Addali, D. Mba et al., Appl. Acoust. 79, 42–46
and static standard qualitative testing system of AE (2016). https://doi.org/10.1016/j.apacoust.2013.12.005
sensor is set up by adopting zirconia sphere, steel 8. B.J. Wright, S. Grigg, A.S. Bergsaker et al., Clin. Biomech.
rule and PAC acoustic emission acquisition device. 76, 105026 (2020). https://doi.org/10.1016/j.clinbiomech.20
The acoustic emission test results of the sensor show 20.105026
that the static index nonlinear error is 16.5%, the 9. H. Pan, M. Huang, Constr. Build. Mater. 254, 119307 (2020)
sensitivity is 444.05 E/mm, the return error is 11.2%, 10. M.S.H. Bhuiyan, I.A. Bhuiyan, M. Dahari, Y. Nukman,
and the non-repeating error is 19%. When the same Measurement 92, 208–217 (2016)
type of excitation signal is applied to the AE sensor, 11. W. Zhang, H. Jia, G. Gao, X. Cheng, P. Du, D. Xv, Appl.
the received acoustic emission signal has the same Acoust. 156, 387–393 (2019)
frequency, which means that it has a certain fre- 12. R. Goldaran, A. Turer, K. Mehdi, Constr. Build. Mater. 242,
quency retention under this condition. In order to 118–123 (2020)
evaluate signal superposition, the concept of signal 13. J.B. Liu, A.X. Guo, H. Li et al., J. Fluids Struct. 78, 197–214
superposition error value (DHt) is proposed. Com- (2018)
pared with commercial sensors, the acoustic signal 14. D. Xiao, Z. Zhang, H. Ren, Advances in Acoustic Emission
attenuation of epoxy-based AE sensors is faster, the Technology (Springer, New York, 2015), pp. 33–45
secondary reflexion signal is not obvious, and the 15. H. Hatano, E. Mori, Acoust. Soc. Am. 59, 344–349 (1976)
noise threshold is lower and can receive weak 16. H. Hatano, T. Chaya, S. Watanabe, K. Jinbo, Ultra Ferroe.
acoustic emission signals with smaller amplitudes. Freq. Control 45, 1221–1228 (1998)
Through simulated rain water corrosion test, it is 17. G.C. McLaskey, S.D. Glaser, Nondestruct. Eval. 31, 157–168
found that the self-made AE sensor has strong rain (2012)
18. B. Burks, M.A. Hamstad, Compos. Struct. 117, 411–422
water corrosion resistance, which can be applied to
(2014)
structural health monitoring in civil engineering
19. J. Kober, Z. Prevorovsky, M. Chlada, Sens. Actuators A Phys.
projects.
240, 50–56 (2016)
20. G.R. Markus, A. Markus, Sens. Actuators 269, 294–307
(2018)
Funding
3360 J Mater Sci: Mater Electron (2022) 33:3348–3360

21. J. Yang, W. Yang, W. Chen et al., Prog. Polym. Sci. 109, 24. J. Weiss, E. Jondeau, A. Giani et al., Sens. Actuators 265,
101289 (2020). https://doi.org/10.1016/j.progpolymsci.2020. 211–216 (2017)
101289
22. C. Wu, Y. Pan, T. Ueda, Constr. Build Mater. 174, 537–546 Publisher’s Note Springer Nature remains neutral with
(2018) regard to jurisdictional claims in published maps and
23. Y. Xi, B. Li, L. Gao, T. Tang, H. Liao, Appl. Acoust. 134, institutional affiliations.
131–137 (2018)