2.

Principles of the Acoustic Emission (AE) Method and Signal Processing

2.1 Principles of the AE method

When an external force is applied to a solid material, the material deforms. In the case of low stress
due to a small external force, the deformed material elastically recovers to its original shape upon
unloading (Fig. 2.1 (a)). Such deformation is called elastic deformation. In the elastic range, the
external force leads to accumulation of energy inside the material as strain energy.

Fig. 2.1 Generation of AE due to strain energy release

There is a limit for energy accumulation in the solid material. If the strain energy stored in the
material because of the external force reaches the limit, it is released and results in plastic deformation.
That is, even when the external force is unloaded, the material cannot recover its original shape (Fig. 2.1
(b)).

If even larger external force is applied, cracking occurs in the material. In this case, the strain energy
accumulated inside the material is consumed by the generation and growth of cracks (Fig. 2.1 (c)).

When the solid material is deformed or cracked, it consumes strain energy. Thus, most of the strain
energy is released. Simultaneously, remaining energy is consumed to generate sound and heat. A
phenomenon in which sound is generated by the energy release is referred to as acoustic emission (AE),
and it is sometimes described as the “scream” of the material under stress.

Sound is a phenomenon of energy release in air. In principle, because of cracking inside the material,
elastic waves are generated and propagate through the material. The phenomenon is analogous to an
earthquake. In other words, AE can be referred to as a “micro-earthquake” in a solid material.

In summary, AE is defined as a phenomenon in which strain energy accumulated in a solid is released

because of deformation or cracking, and thus generates elastic waves. AE waves are detected usually
at the surface of the material by a sensor as illustrated in Fig. 2.2.

in solids

Fig. 2.2 Generation and detection of AE waves

Fig. 2.3 Generation mechanism of a plate earthquake

Elastic waves generated by cracking propagate through the material and are detected by an AE sensor
(vibration-to-electrical signal conversion element) placed on the surface. In this case, some portions of
surface vibrations are released to the air as sonic waves and can possibly be heard as a breaking sound.
 To explain the source mechanisms of AE phenomena as an analogy of an earthquake, the focal
mechanisms of an earthquake are illustrated in Fig. 2.3. A fault slip forms on a plate boundary between
an oceanic plate and continental plate of the Earth’s crust. As a result, a sliding failure or shear failure
is nucleated. In the case of AE phenomena, two types of dislocations in Fig. 2.4 can be referred to as
crack motions at a crack surface. A sliding or slip crack-motion (shear crack) corresponds to “in-plane
shear dislocation”, and tensile dislocation (tensile crack) is referred to as “opening dislocation”.

Fig. 2.4 Types of crack motions at a crack surface

Fig. 2.5 Frequency ranges of various elastic waves

The basic difference between the earthquake and AE is the scales of cracking (faults) and the related
frequency ranges. As shown in Fig. 2.5, the earthquake involves elastic waves of low-frequency
components up to several Hertz, while the AE phenomenon involves the emission of waves of high-
frequency components of up to several megahertz. In general, humans can physically sense seismic
events, while inaudible waves with low amplitudes are generated as AE phenomena.

In the case of the earthquake, seismic waves measured by a seismograph are analyzed to locate the
hypocenter, and the scale of the earthquake is estimated as the seismic magnitude. Similarly, in the case
of AE, the location of AE (AE source) and the scale of damage resulting in AE can be estimated by
analyzing measured AE waves.

2.2 AE testing as non-destructive testing

 AE is used for NDT, regardless of the amount of strain energy released during the deformation and
failure of materials. In this section, the reasons for choosing AE testing (AT) and the cases in which AT
is employed for NDT are described.
Failure, which does not occur instantly, generally begins at the micro-level, leading to final failure
after the gradual accumulation of micro-level cracks. AT is a method of monitoring or measuring this
process up to the final fracture. In other words, by detecting AE during the initial phases of the
generation and growth of any crack (repairable phases), we can stop ongoing operation and test a piece
or structure to avoid ultimate failure of the piece or structure (Fig. 2.6). This is why AT is used for NDT.

With AE monitoring Without AE monitoring

！ ！ AE sensor

AE AE
micro Micro
crack crack

AE sensor AE sensor
Continue
Suspend
pressurization
pressurization test
test

Leakage of
contents

Fig. 2.6 Application to pressure vessels

(a)Ultrasonic testing
Transducer Scanning

Reflection
wave
Crack (Flaw)
(b)AE testing
AE sensor
External force
AE Crack （flaw）
propagation
Fig. 2.7 Comparison between ultrasonic testing and acoustic emission testing
 An ultrasonic testing (UT) is widely used as a means to detect cracks (flaws) in a material. UT and
AT both involve the use of elastic waves and are often compared. Consequently, we describe the
characteristics of UT and AT and the difference between the two.

In UTs (Fig. 2.7 (a)), a reflected wave (echo) from a crack is detected by the excitation of an
elastic wave from a probe, and thus, the crack is detected. In ATs (Fig. 2. 7 (b)), a crack is discovered by
detecting an AE wave released from it. Both these methods involve the use of elastic waves, but there is
an essential difference between the two in this respect. Recently, research on an acoustic-ultrasonic (AU)
method in which ATs and UTs are combined has been promoted.

As mentioned above, the application of AT to structural diagnosis led to the prediction of

earthquakes and falling rocks and the establishment of the current monitoring technique. Consequently,
the following capabilities are being incorporated in commercial products.

(1) Detection of AE phenomena associated with micro-failure or indications of failure

(2) Detection of AE indicating damage by continuously observing noise and
vibrations
(3) Application of AE testing to various types of leak monitoring

A testing method, which is often used in structural diagnosis, measurement of ground pressure in ground
materials, and damage evaluation for pressure vessels, involves the application of the AE Kaiser effect.
The Kaiser effect is a phenomenon in which most of the AE is not detected until the preload on a
solid material is removed and a load is once again applied to it. Since this phenomenon
does not occur when the material is structurally unstable (for instance, in the developmental stages of
failure), the effect can be used as an indicator of stability.

2.3 Types and characteristics of AE sources

As a source of AE waves, a crack is shown in Fig. 2.2. In addition, a martensitic transformation and
metal transformations such as tin cry are well known as AE sources. On rocky and flat terrain
that is old enough to contain metals, collapse phenomena such as falling and
talking rock are considered to be sources of AE waves. These phenomena, essentially
caused by a failure phenomenon, can be said to result from crack formations.

Furthermore, AE waves are generated by rust formation and friction caused by rust. The generation of
AE waves by friction is not essentially different from the occurrence of an earthquake caused by a fault
slip. However, in the case of a composite material, delamination and fiber fracture occur in the material.
Therefore, AE is caused by a complicated combination of these events.

Next, the detection of abnormal noise for acoustic diagnosis has been carried out to monitor the safety
of equipment and determine when the equipment must be replaced. Abnormal noise caused by damage
to tool edges and turbine blades can also be considered as AE sources. Consequently, the AE method is
applicable to the detection of such noise. Furthermore, AE testing has already been employed in nuclear
facilities to monitor loose parts of systems. This method detects the existence of broken pieces and
fragments of parts (loose parts) that have slipped into the insides of pipes in a reactor.

Further, some of the materials reported so far do not generate AE upon their failure. This is considered
to be due to the minimum energy of failure or ultra-high-speed failure. This does not mean that AE
waves are not generated; rather, they cannot be as easily detected as in AE events.
 AE waves are essentially generated by failure phenomena and can be mostly attributed to the formation
of microcracks. This type of AE is sometimes called primary AE. On the other hand, AE generated by
rust formation and friction caused by inclusions and particles is called secondary AE. Typical
mechanisms that generate AE waves are conceptually illustrated in Fig. 2.8.

Fig. 2.8 Examples of sources of AE waves

Description of term ( Transformation )

The micro-features of metal materials indicate that their atoms are arranged systematically (crystal
structure). The crystal structure of a metal changes under certain conditions—this is called
transformation.

Conventionally, types of AE waves are classified as burst AE and continuous AE waves. In

principle, an AE wave is generated by the formation and growth of a crack at its source. Thus, a burst
AE wave is reasonable. For instance, an elastic wave emitted by an opening crack propagates, causing
amplitude damping. As a result, there is one waveform whose amplitude attenuates naturally over time.
However, in the case of continuous and consecutive generations of dislocations, the AE waveform
shown in Fig. 2.9 is observed, resulting from the overlapping effects.
 Fig. 2.9 Example of a detected AE waveform

As the time axis in the figure is extended, individual AE events can be discriminated in most cases.
Consequently, they can be considered to be overlapping burst AE waves. In particular, AE waves
associated with plastic deformation in metal materials are difficult to separate and are well known as
continuous AE waves.

2.4 Properties of AE waves

2.4.1 Propagation in solids

An AE wave is, in principle, defined as an elastic wave generated at an AE source. Wave motion is a
phenomenon in which particle motions are dynamically transferred to adjacent particles, as each particle
only vibrates at its own position and does not move. However, wave motion itself propagates with its
characteristic velocity.

Wave motions are defined on the basis of orientations of particle motions and directions of
propagations. In an earthquake, for instance, the motion that people first feel is longitudinal, and the
transverse motion arrives later. Major motion results from a Rayleigh wave, which is generated at the
surface of the Earth after the longitudinal and transverse waves arrive. Similarly, AE waves consist of
longitudinal, transverse and other waves.
 Fig. 2-10 Types of wave motion

(1) Longitudinal wave (P-wave, Primary wave)

In the case of longitudinal waves (Fig. 2.10 (a)), particles vibrate along the direction in which the wave
propagates. The wave is also called a dilatational wave since it is associated with volume change. The
longitudinal wave can propagate through all media types—solids, liquids, and gases—and is the fastest
of elastic waves.

(2) Transverse wave (S-wave, Secondary wave)

In the case of transverse waves (Fig. 2.10 (b)), particles vibrate perpendicular (in the “lateral direction”)
to the direction in which the wave propagates. The transverse wave is also called a shear wave, and can
propagate through solids but not through liquids and gases. The ratio of the velocity of the transverse

wave to that of the longitudinal wave is theoretically equal to , where  is Poisson’s ratio.

(3) Other waves

Longitudinal and transverse waves are always generated and propagate in an elastic solid. They are
sometimes called body waves. Following their arrival at the surface, other waves are generated in a
solid. For instance, Fig. 2.10 (c) shows the propagation of a Rayleigh wave. In the case of the
Rayleigh wave, particles near the surface move elliptically, and motion decreases with depth. It is thus
called a surface wave. In a thin plate, Lamb waves (plate waves) are generated, as the plate vibrates.
The Lamb wave has a symmetrical mode (S mode) in which the plate vibrates symmetrically, as shown
in Fig. 2.10 (d), and an anti-symmetrical mode (A mode) in which the plate vibrates asymmetrically, as
shown in Fig. 2.10 (e).
2.4.2 Wave velocity and wavelength

The wave velocity at which an elastic wave propagates through a solid depends on properties of the
materials and the wave type. For instance, the longitudinal wave propagates through aluminum at a
velocity of approximately 6350 m/s, while the transverse wave propagates at approximately 3130
m/s. Even for the same type of wave, the velocity varies, depending on the material. For instance, the
velocity of the longitudinal wave propagating through concrete is around 4000 m/s.

The velocity of the longitudinal wave, CL [m/s], is expressed in terms of Young’s modulus E,
Poisson’s ratio v and the density of the material through which the wave propagates:

. (2.1)

On the other hand, the velocity of the transverse wave, CT [m/s], is expressed as

. (2.2)

The velocities of the longitudinal and transverse waves propagating through typical materials are listed
in Table 2.1. The velocity of the Rayleigh wave is approximately 90% of that of the transverse wave.
The velocity of the Lamb wave varies with the plate thickness, frequency, and mode.

Table 2.1 Velocities of typical materials

(representative values)
Medium Density Velocity of Velocity of
(kg/m3) longitudinal- transverse-
wave (m/s) wave (m/s)
Aluminum 2700 6350 3130
Steel 7800 5900 3200
Concrete 2500 4000 2600
Water 1000 1430 -
Air 1.2 330 -

Wavelength is defined as the distance of one cycle in the sinusoidal motion of a wave. An example of
wave motion with constant frequency is shown in Fig. 2.11. The time interval between successive
motions of one cycle is defined as the period and expressed as T [s]. The number of cycles per second is
defined as the frequency and expressed as f in units of Hertz (Hz), which is equivalent to cycles per
second. Consequently, the period T and frequency f are related by

. (2.3)
In the case that the wave propagates with velocity C [m/s], the wavelength is derived from the wave
frequency f. Thus, the wavelength is expressed as λ [m], and the relation among the velocity C,
frequency f, and wavelength λ is

. (2.4)

When motion is repeated at f cycles per second, a wave with wavelength λ travels a distance of fλ per
second, which is equal to the velocity C as given in Eq. 2.4. For a velocity C, a higher frequency f
results in a reduced wavelength λ.

Period T
周 期 T
子の振動方向変位
粒 Displacement

Period
周 期 T T

0 10 20 30
時 間 μsec
Time, [ μ s]

Fig. 2.11 Wave motion and the period.

For instance, the velocity of the longitudinal wave is 4000 m/s in concrete. Therefore, the
wavelength of the 100-kHz component is obtained as 4000 m/s/100 kHz = 40 mm. A frequency band
lower than 100 kHz is often employed for AE testing in rock and concrete. Hence, it is noted that
wavelengths of several centimeters are normally measured. As a result, scattering due to inclusions and
aggregates becomes minor. In contrast, the velocity of the longitudinal wave is 5900 m/s in steel, and
the wavelength of the 1 MHz component is obtained as 5900 m/s/1 MHz = 5.9 mm. Consequently, the
wavelength often becomes greater than the plate thickness. In this case, dominant motions of AE waves
result in Lamb waves, instead of longitudinal and traverse waves. Therefore, it is necessary to pay
particular attention to selecting a velocity for locating AE sources (Section 6, Chapter 2).

2.4.3 Attenuation

Although waves can propagate through a completely elastic body without attenuation, waves
attenuate in most general materials. The attenuation occurs in time and in space. The former is related
to time-series motion of a material and is generally called viscous damping. The latter results in damping
of the amplitude with distance and is called distance attenuation.

As an AE wave propagates from its source, its amplitude decreases owing to attenuation (Fig.2.12).
Hence, it becomes necessary to determine the proper positions and number of AE sensors, depending on
sensor-to-sensor distances. Figure 2.13 shows a relation between detectable AE waves and frequencies
in concrete. In general, AE waves with higher frequencies can attenuate significantly. Therefore, the
sensors must be as well positioned so that the distance between the source and sensors is determined,
taking into account detectable AE waves at object frequencies.
 Attenuation

AE amplitude is
decreased with
AE source propagation length
Fig. 2. 12 Schematic illustration of attenuation

Fig. 2.13 Attenuation rate for concrete

2.4.4 Reflection and transmission

AE waves are generally detected by an AE sensor placed on the surface of a material. Therefore,
wave components affected by reflection on the sensor-installed surface are measured. It should be noted
that both longitudinal and transverse waves are generated by reflections of longitudinal waves. Even the
incidence of a transverse wave generates a longitudinal wave through reflection. Snell’s law is known as
a theory governing incident and reflection angles. This law is expressed as a relation between the
propagation velocity C and an incident angle :

, (2.5)

where θ1 is the incident angle, C1 is the propagation velocity, θ2 is the reflection angle, and C2 is the
propagation velocity of the reflection. In the case of an incident transverse wave (S wave) as shown in
Fig. 2.14 (a), we have
 . (2.6)

As a result, the incident angle θ becomes equal to the reflection angle θ'' since the velocities of the
transverse waves, CT, are equal even after reflection. Since the velocity of the longitudinal wave, CL, is
higher than that of the transverse wave, CT, the reflection angle θ' becomes larger than . When the
transverse wave (S wave) arrives at the AE sensor installed on the surface at a critical angle θc, the
reflected longitudinal wave (P wave) propagates along the surface. Since this wave (SP wave)
propagates faster than the direct transverse wave, AE waves are detected at the AE sensor as the incident
longitudinal wave, the SP wave, and the transverse wave successively. These waves are followed by the
Rayleigh wave.

incident
transverse
wave
reflected transverse wave

Reflected
longitudinal wave AE sensor
(2nd arrival)

incident transverse
wave (2) incident
longitudinal wave
(1st arrival)
incident transverse wave (1)

Fig. 2.14 Reflections of AE waves at the surface due to the incident of the transverse wave

Reflection can be considered to be a special case in which a boundary surface exists between two
materials (propagation media) and no upper layers exist in transmission. In the two layers shown in Fig.
2.15, an incident wave becomes a reflected wave at the boundary surface and then becomes a wave
transmitted to other layers. This transmitted wave causes refraction. An incident wave at the given
incident angle enters other layers at a given angle of refraction. The relation between the incident and
refractive angles is given by Equation (2.6).
 Fig. 2.15 Reflection and transmission of an AE wave

According to the theory of elastic wave motion, the amplitude of the incident wave and the
ratio of the amplitudes of the reflected and transmitted waves in Fig. 2.15 are
known to be related. Assuming that the amplitude of an incident wave is A, that of a transmitted
wave is AT, and that of the reflected wave is AR, these relations are

Reflection coefficient , (2.8)

Transmission coefficient , (2.9)

where ρ1 is the density of Medium 1 in Fig. 2.15, C1 is the propagation velocity of Medium 1, and ρ2 and
C2 are the density and propagation velocity in Medium 2, respectively. In the absence of Medium 2,
sinceρ2C2 = 0, the reflection coefficient AR/A becomes –1, while the transmission coefficient AT/A
becomes zero. On the other hand, if the relationρ2C2 =ρ1C1 is established in the same medium, no wave
will be reflected with a reflection coefficient zero and all waves will be transmitted with a transmission
coefficient of 1.

2.5 Signal processing

Detecting AE waves for inspection can be compared with estimating and comprehending the condition
of machines used daily on the basis of their operation sounds. A skilled machine operator can detect
anomalies in the machine on the basis of “unusual noises from the machine.” This indicates that
“something is wrong with the machine.” An engineer familiar with the components of the machine can
often discover the defect on the basis of unusual noises. For instance, intermittent low noise or rattling
sounds indicate that parts are rattling and bolts are loose, while a continuous high noise or rubbing sound
indicates the wear of rotating parts. The machine is diagnosed on the basis of noise. When people hear a
sound, they can recognize the sound through its attributes such as its interval (tone) and tune (rhythm)
and then assess the situation on the basis of the sound using their knowledge and experience.
 However, to conduct a series of tasks using the machine (for instance, to detect a sound, identify its
characteristics, and accurately evaluate the sound), it is required to operate the machine in many special
processing tasks. AE involves the propagation of weak waves through a solid and no sound can be
immediately heard by a person. Various devices are required for the detection of such waves and the
identification of their characteristics on the basis of acoustic phenomena.

2.5.1 What is an AE signal?

In scientific as well as AE measurements, physical quantities including sound, vibration,

temperature, and light intensity are called signals. These quantities must be observable and are generally
converted to electric signals using an appropriate sensor.

An electric circuit needed for AE measurement, called an AE channel, consists of a) an AE sensor,

b) a preamplifier or an impedance matching transformer, c) a filter, d) a main amplifier or other
necessary devices, e) a cable, f) a detector or processor, or a combination of devices that have the same
functions as these devices. In other words, an AE signal obtained from AE measurements is a physical
quantity determined after an AE wave propagating through a solid is detected by these devices and
converted to an electric signal. The quantity is generally a voltage value [mV].

d) Main amplifier

b) Preamplifier

c) Filter e) Signal processor

a) AE sensor
Fig. 2.16 Basic AE channel

Transverse wave

Micro-tremor
Amplitude

(Noise)

Longitudinal
wave

Time
Fig. 2.17 Example of an AE signal
 Next, the information included in the AE signal is described. AE is a phenomenon in which some of
the energy released by local changes in a material (an AE event) propagates through the material as a
wave. Consequently, the original AE signal must contain information on the nature of an AE event or the
source of the wave. A waveform of the type of AE signal shown in Fig. 2.17 is generally called a burst
waveform. As apparent from the name, any burst change in equilibrium generates AE. In addition, on
the basis of the scale of the wave (height in the figure), the degree of the change can be estimated. For
instance, assuming that the change is the burst breakage of given parts, the degree of the breakage or the
size of the broken parts can be inferred from analogy. If this breakage results from the collision of a
missile with an object, the information required to determine the degree of damage to the object is
included in the AE signal.

However, until the wave propagates from the AE source to a point where it can be detected or a point
where an AE sensor is present, depending on the property of the solid in which the propagation occurs,
some or all of the information on the source contained in the wave may be lost or information on the
material properties may be gained. Consequently, the information must be carefully collected.

2.5.2 Basics of waveform parameters

In the case of material evaluation and structure diagnosis based on AE, the relation between an AE
source event and the waveform characteristics in the form of the graphical representation of the AE
signal is generally understood.
(a) a burst waveform with clearly observable longitudinal and lateral waves;
(b) a harmonic continuous waveform with longitudinal and lateral waves that are difficult to observe;
(c) a burst waveform with longitudinal and lateral waves that are slightly difficult to observe.

(a) Burst waveform with clear arrival of

(a) 縦 波 と 横 波 と が 明 瞭 な 突 発 型 波 形
longitudinal and lateral waves

(b) Harmonic continuous waveform mixed

(b) 縦 波 も 横 波 も 不 明 瞭 な ハ ー モ ニ ッ ク な 連 続
with longitudinal and lateral waves
型波形

(c)縦 波Burst
(c) と 横 波 と が やwaveform
や 不 明 瞭 な 突 発 型 波 形with unclear longitudinal
and lateral waves
Fig. 2.18 Classified waveforms of AE signals
 Waveforms obtained for the AE signal vary widely but can often be visually classified into certain
patterns. To illustrate this, three waveforms are shown in Fig. 2.18. From the figure, a difference in the
wave intensity, duration of motion or type of propagation can be observed. In this manner, the
characteristics of a waveform are clearly expressed as values such as motion intensity, wave-
continuation time, and the time from the start of wave generation to the time when the largest wave is
observed or the time from the beginning of the reduction of wave intensity to the extinction of the wave,
which are called waveform parameters. Consequently the characteristics of the AE signal are described
by these waveform parameters.

2.5.3 Discrimination of an AE signal

A wave generated in one AE event comprises a block of various wave components such as
longitudinal, transverse, or surface waves, as shown in Fig. 2.17. Because of the difference in the
propagation velocities of different wave components, these components successively reach the AE
sensor placed far from the AE source with a time lag. Furthermore, there is a time lag between the
gradual reduction in wave intensity to the extinction of these waves. A series of pulses is called a wave
packet. Furthermore, the wave packets of the AE signal corresponding to the AE waves released from a
certain source are separated and extracted from signals that are continuously output by the AE sensor.
This signal processing is called discrimination.

Actual AE signals, in contrast to deterministic signals with a regular time period and amplitude, are
random signals in which the changes in the signal values after a certain time instant cannot be predicted.
In addition, as shown in Fig. 2.18, because several types of noise generated by many factors are included
in AE signals to be measured, it is unclear what the start and end of AE signals are. Therefore,
instruments for AE measurements rely on unique methods to differentiate between AE signals.

As shown in Fig. 2.19, when a voltage threshold is set slightly higher than the voltage level of the
background noise during AE measurements and the amplitude of the AE signal exceeds this threshold,
the electric signal is recognized as an AE signal. In addition, this voltage threshold can be set at the
dashed line for which the voltage is negative in Fig. 2.19. Furthermore, these two methods can be
concurrently used.
 Fig. 2.19 Signal discrimination with a voltage threshold

However, in this simple processing, the discriminated AE signals form a pulse train, resulting in
potential loss of critical information on AE. Consequently, in this case, a method illustrated in Fig. 2.20
is sometimes used for signal processing. In particular, the amplitude of an AE signal is measured while
an envelope is detected from the AE signal. The start of the AE signal is determined to be the point at
which the voltage level of the detection signal first exceeds a set voltage threshold. On the other hand,
the end of the AE signal is determined to be the point at which the voltage level of the envelope
detection signal falls below a set voltage threshold. One wave packet of continuous AE signals between
the start and end of the AE signal is called an AE hit.

Furthermore, there is another method for the concurrent use of this voltage threshold and time
discrimination by a timer. In particular, assuming that the start of the AE signal is the point at which the
voltage level of the AE signal first exceeds the voltage threshold, the AE signal levels and the voltage
thresholds that are continuously input are compared. If the conditions for the voltage of an AE
signal to be higher than a certain threshold are satisfied within a given time, the
above comparison is repeated. If the conditions cannot be satisfied even when the time has elapsed, this
point is determined to be the end of the AE signal.
 Fig. 2.20 Signal discrimination by envelope detection

It is believed that the discrimination of AE signals using the above mechanism will be easiest when the
steps below are followed.
1) Set a voltage threshold Vt and detection time limit Tt that specify the start of the AE signal.
2) Assume the point at which the level of the AE signal, Vae, first exceeds Vt as the start of the AE
signal, and start the timer.
3) If Vae exceeds Vt again before the timer has reached the detection time limit Tt, reset the timer at this
point.
4) Repeat the comparison between Vae and Vt until the timer reaches Tt.
5) If the timer reaches Tt without Vae exceeding Vt, consider this point as the end of the AE signal end.

Regardless of the discrimination processing of any AE signal, a given “dead time” during which the AE
signal is not detected after the end of the AE signal is generally set. This time is established so that a
wave packet of an AE signal already identified is not detected as a wave packet of another AE signal by
the same AE sensor because of bypassing and reflection.
 Fig. 2.21 Signal discrimination by timing parameters

2.5.4 AE parameters

It was mentioned earlier that an discriminated AE signal is believed to include information on an AE

event in the source generating the wave. Therefore, information on AE events and physical phenomena
is generally inferred indirectly; i.e., by calculating waveform parameters representing the
characteristics of a waveform of the discriminated AE signal, examining temporal variations in specific
waveform parameters during measuring AE, obtaining distributions for one or more predetermined
evaluation thresholds, and investigating the correlation between different waveform parameters.

The focus of this section is the waveform parameters defined in terms of the AE standard, ISO
12716: 2001. Figure 2.22 illustrates the significance of the main parameters through schematic
waveforms of an AE signal.

In general, the voltage threshold is only focusing on the instantaneous amplitude of AE. Though,
evaluation threshold which is different from voltage threshold is used in some cases when calculating
AE parameters. In Fig. 2.22, the evaluation threshold has the same value as the voltage threshold. In
some cases, the evaluation thresholds are set to higher than voltage thresholds to calculate the AE
parameters.
 Fig. 2.22 Waveform parameters

The time that elapses between the start and end of an AE signal is called the AE signal duration, and
the peak amplitude of the AE signal is called the AE signal peak amplitude or simply the AE signal
amplitude. The time that elapses between the start of an AE signal and the time corresponding to the AE
signal peak amplitude is called the AE signal rise time.

The frequency at which the AE signal amplitude exceeds a predetermined threshold within a
specified time range is called the AE count, count, ring-down count, or emission count. In general,
the AE count is related to the frequency at which the AE signal amplitude exceeds a
predetermined evaluation threshold within the AE signal duration. The AE count is used in AE testing to
determine the number and frequency of AE signals.

The AE count rate or emission rate is calculated from the AE count per unit time, as a parameter that
indicates the increase/decrease in the frequency associated with the AE count time.

It is important to understand how AE signal processed in the AE measurement instrument that is

currently in use, since waveforms and AE parameters may be changed when inappropriate frequency
filter is used. It is also noted that evaluation threshold also affects AE parameters.

Comparisons of individual measurement results carried out by repeating the same AE measurement
and continuous or intermittent AE measurements of the same object over a relatively long time would
not be valid if the same measurement instrument is not used.
 The purposes of AE tests are to quantify the characteristics of waveforms of AE signals using various
methods and to parameterize them, to guess the (unknown) events that have occurred, determine the
number and frequency of AE signals, and estimate when AE occurred. AE parameters are summarized
as follows.

(a) Parameters based on the signal waveform level

1) AE signal (peak) amplitude
The maximum voltage in an AE signal (hereafter, referred to as the discriminated signal) detected
between the start and end of the AE signal discriminated on the basis of a given measured threshold; the
signal belongs to a series of AE signals generated by one AE event.

The voltage value of the AE signal can be expressed as a common logarithm on the basis of a reference
value of 1 μV.

AE signal peak amplitude [dBAE] = 20 log10 (A1/A0)

Here, A0 = 1 μV, which is the output from the AE sensor without any amplification, and A1 is the
measured maximum voltage of the AE signal.

2) AE count (ring-down count)

The frequency when the measured voltage of the AE discriminated signal exceeds an evaluation
threshold. The AE count at a given time is divided by the time that has elapsed since the start of the AE
signal, and the count result per unit time is called the AE count rate.

3) AE energy
The AE signal energy is determined as the square of the AE signal amplitude, the integral of the square
of the instantaneous amplitude of the AE discriminated signal over a certain duration, or the integral of
the instantaneous AE signal amplitude determined by envelope detection over a certain duration.
However, these values differ from the total energy (AE event energy) released due to an AE event.

(b) Parameters based on the shape of the signal waveform

4) AE signal duration
Time that has elapsed between the start and end of the AE discriminated signal.
5) AE signal rise time
Time that has elapsed between the start of the AE discriminated signal and the time at which the AE
signal peak amplitude is attained.

6) Ratio of rise time to amplitude

The ratio of the AE signal peak amplitude to the AE signal rise time.

(c) Parameters based on the level of the continuous AE signal

7) Effective value of the AE signal
The effective value (square root) is obtained by calculating the root-mean-square of an AE signal. The
effective value is also called the root-mean-square value. The effective value of the AE signal represents
the energy level of the AE signal and also enables us to evaluate the rate of occurrence of AE in the
same manner as the AE count rate does, particularly in the case of continuous AE (refer to the
description on page 32).
8) Average signal level of the AE signal
The average signal level represents the average energy calculated by integrating the absolute value of the
amplitude of a rectified AE signal over a given time and dividing the result by the time that has elapsed
since the start of the signal.

The average level of the AE signal is considered to be also effective for evaluating the root-mean-square
value of a continuous AE signal (refer to the description on page 34)

(d) Other AE parameters

AE hit time
AE hit count
AE hit count rate
AE event count
AE event count rate
Damping factor based on propagation distance
Average frequency
Frequency spectrum
Energy moment

2.6 AE source location

2.6.1 One-dimensional and two-dimensional location methods

A great advantage of the AE method is that the location of an AE source can be determined
relatively easily. In contrast to the case for a UT, it is not necessary to scan the whole of a n object using
a probe when determining the position of damage to the object. The location of the damage can be
measured using a fixed sensor. This method is the same as the location method for determining an
earthquake center. In the location method, the wave velocity of the object and the arrival time of the AE
wave must be known. Consequently, since a clear rising point of the waveform is observed in the
detected burst AE waveforms generated by micro-cracking, it is easy to detect the arrival time of the AE
wave, thereby enabling relatively accurate location. On the other hand, since a clear rising point in the
continuous AE waveforms associated with plastic deformations is not observed, it is difficult to
determine the arrival time of the AE wave. Therefore, an accurate location cannot be expected.

It is necessary to determine beforehand the velocity of the AE wave to identify the location of
damage to the object using the AE wave. Since the first-arriving longitudinal wave can often be clearly
detected when the object is sufficiently thick, only the velocity of the longitudinal wave is used. Because
the longitudinal waves often cannot be clearly observed in the case of a thin-plate object, it would be
better to use the velocity of a Rayleigh wave or Lamb wave (refer to Fig. 2.10); these velocities can be
clearly observed. Before AE measurement, it is necessary to check whether the correct location can be
found using a simulated AE source (breakage of pencil lead and pulse generator) after installing an AE
sensor.

In general, it is desirable to find the three-dimensional location in determining the damage position.
However, when the number of measurable channels and the number of sensors are limited or when the
shapes of the object are specific, it is difficult to determine the three-dimensional location of the damage.
With a known velocity, at least an arrival time interval is needed only for the order of a location to be
determined; i.e., the minimum numbers of sensors required for one-dimensional location, two-
dimensional location, and three-dimensional location are two, three, and four, respectively.
 A method for determining the arrival time of the AE wave is important. When a clear ris ing point
in the AE waveform is first observed, this point can only be the arrival time, but the rising point
generally cannot be clearly identified in most cases because of noise. In this case, it is practical to
calculate the arrival time assuming that it is given by the point exceeding a threshold or the point at
which the waveform attains the initial peak. However, it would be also necessary to evaluate the
precision of location achieved using the simulated AE source.

One-dimensional location in which a burst waveform is detected is now discussed. Assuming that
there is AE between two AE sensors, the location of an AE source is x, and the locations of the two
sensors are x1 and x2, we derive an equation for t1 and t2 when the AE wave reaches the sensors:

, , (2.10)

where C is the velocity of the AE wave. Furthermore, assuming that there is a difference between the
arrival times at the two sensors, we have

. (2.11)

The velocity and the positions of the sensors are known. From this equation, the one-dimensional
location of the AE source, x, can be determined.

AE sensor
(Ch.1) t1 t2
(Ch.2)
AE source
AE wave

x1 x x2 X axis

Fig. 2.23 Linear source location

Similarly, three sensors enable us to determine the planar location of the AE source or the two-
dimensional location (x, y). Assuming that the velocity is C, the positions of the three sensors are (x 1,
y1), (x2, y2), and (x3, y3), the arrival times at the sensors are t 1 and t2, and the differences between the
arrival times at the sensors are t12 and t13, as mentioned above. We thus have (Fig. 2.24)

. (2.15)
 These two equations with two unknowns x and y can be solved. The numerical solution of non-linear
equations obtained using a computer enables us to easily determine the two-dimensional location of the
AE source. The precision of location achieved by these methods depends mainly on the resolution of the
arrival times, sensor sizes, and the sensor-to-senor interval.

Ch.2 y
t2 (x 2,y 2) x
AE source
Ch.1 t3
(x,y) Ch.3
t1
(x 1,y 1) (x 3,y 3)

Fig. 2.24 Planar source location

2.6.2 Guard sensor

As described in the previous section, if a sufficient number of measurement channels are available for
measuring the arrival time of an AE wave, the location of an AE source can be determined. However,
when many extraneous noise signals are measured from sections other than the object to be measured, it
is inefficient to conduct location in the presence of assumed noise signals with the above method.

The following method is useful for eliminating noise from the chucking of a test specimen, as required
in a materials tensile test, or for removing noise generated by the test. It also contributes to the
elimination of clear noise propagating from outside the object, even in a large structure.

x X
Monitoring range
Noise Noise
AE

Guard Sensor Sensor Guard

sensor (Ch.1) (Ch.2) Sensor
(Ch.0) (Ch.3)
Fig. 2.25 Arrangement of a guard sensor for noise elimination

The measurement method involving the use of a guard sensor for one-dimensional location is described
below. As shown in Fig. 2.25, an AE source between the sensors at x 1 and x2 is to be monitored. Guard
sensors for noise elimination are installed at x 0 and x3 outside the monitored area. If any noise is
generated outside the area bounded by x0 and x3, the first signal will always reach a guard sensor at x 0
or x3. Consequently, when comparing the arrival times of all AE signals in the channel, if a guard
sensor at x0 or x3 indicates the arrival time of the first signal, then the noise in this event can be
concluded as originating from outside the monitored area. Thus, the one-dimensional location of events
except eliminated events can be efficiently carried out using the sensors at x1 and x2.

2.6.3 Zone location

The concept of the first-hit channel described in the previous subsection is also useful in zone location,
particularly in the case of large structures. The frequency band used in AE measurement is lower than
that used in a UT; further, the signal damping in AE measurement is less than that in the UT. Therefore,
AE measurement can be used to monitor large structures. However, in the case of a larger object or a
weaker generated signal, all AE signals cannot be recorded in all measured channels because of signal
attenuation. In this case, it is impossible to locate the AE source when there is a difference between the
arrival times of AE signals. The AE signal that arrives first corresponds to the AE source nearest a
sensor in the channel; this enables approximate location of the AE source.

A method for two-dimensional zone location of the object is described below. As shown in Fig. 2.26,
sensors are equally spaced in measuring AE.

CH1 CH2 CH3

AE
sources

・ ・・・ ・
CH4 CH5
Cumulative number
of first-hit

1 2 3 4 ・・ ・・
Channel
Fig. 2.26 Zone location

If any AE signal is detected, it can be concluded that an AE source exists near the sensor. For
instance, if a histogram showing the number of events observed in each channel is plotted, the area with
many AE sources can be identified. Consequently, the precision of locating AE sources will depend on
the sensor-to-sensor intervals. Although this precision cannot be compared with that obtained from the
difference in arrival times, this method significantly helps determine the damage to a large structure in a
realistic manner.

2.6.4 Leak location

AE location can be applied to determine the leakage points in various pipes. If liquids leak from a
pipe and a burst AE signal is detected, it is possible to determine the leakage point in the pipe employing
the above location method. When the arrival time is clearly obtained in each channel, as described
earlier, the difference in arrival times enables accurate location. In addition, when clear signals cannot be
obtained in all channels because of signal attenuation, zone location enables the determination of any
leakage point.

On the other hand, gas leakage will generate a continuous rather than burst AE signal in most cases.
Therefore, location employing the time axis is impossible in such cases. However, any leakage point can
be roughly determined by comparing the amplitudes or RMS voltages of channels, since a strong signal
must be obtained at the sensor near the leakage point in the same manner as in the above case of zone
location.

Reference
(1) Masayatsu Ohtsu, Characteristics and Theory of Acoustic Emission (AE) (Ver. 02), Morikita
Publishing, 2005 (written in Japanese)