Short Course 101-1111-00L:Fundamentals and Applications of

Acoustic Emission

Download :

1. October 4 2. October 11

3. October 18 4. October 25

5. November 1 6. November 15

[] Assignment for a credit:

opening quiz (30%) homework (40%)

final report (30%)

Short Course 101-1111-00L:Fundamentals and Applications

of Acoustic EmissionOctober 4, 2007

1-1 Introduction – History and Fundamental

1-2 Measurement – Sensor

1-3 Measurement – Instrument

1-1 Introduction October 4, 2007 (1/3)

Invitation Discovery of AE Founders and Terminology AE in Concrete Development of AE Fundamentals of AE

Measurement Remarks on

Mathematical Backgrounds

What is acoustic emission (AE) ? In a similar manner to earthquakes, elastic waves are generated due to cracking inside a material.

Elastic waves reach to the surface (boundary) of the medium, and then are converted into sonic waves in air.

The sonic waves are audible as acoustic waves. Thus, due to cracking, “acoustic waves are emitted”. In the measurement, elastic waves are referred to as

AE waves.

Discovery of AE Audible phenomena associated with generation of

elastic waves are known as rock-bursts in mines. In metallurgy, the first AE phenomenon was

considered to be an audible “tin-cry”, which is produced by twinning of pure tin during plastic deformation.

Martensite transformation in steel is also accompanied by large audible noises.

Other AE phenomenon is creaking of timber prior to break.

Typical sources of AE phenomena

The oldest (first) report on a scientifically planned AE experiment, 1934

T. F. Drouillard,”Anecdotal History of Acoustic Emission from T. F. Drouillard,”Anecdotal History of Acoustic Emission from Wood,” Journal of AE, Vol. 9, No. 3, 155-176, 1990.Wood,” Journal of AE, Vol. 9, No. 3, 155-176, 1990.

Prof. Fuyuhiko Kisinoue

Early research

F. Foerster and E. Scheil,”Acoustic Study of the Formation of Martensile Needels,” Zeitschrift fur Metallkunde, Vol. 28, No. 9, 245-247, 1936.

L. Obert,”Use of Subaudible Noises for Prediction of Rock Burst,” Report of Investigations 3555, U. S. Bureau of Mines, 1941

Founders and Terminology J. Kaiser,”An Investigation into the

Occurrence of Noises in Tensile Tests,” UCRL-Translation of Kaiser’s Dissertation(1950), Lawrence Radiation Labo., Livermore, 1964

B. H. Schofield,”Acoustic Emission under Applied Stress,” Report ARL-150, Lessels and Associates, 1961

AE in Concrete Engineering

It was reported that H. Ruesch had known J. Kaiser in the institute. So, he studied the noise emitted during application of compressive load in concrete [1959]. This has been known as one of the first studies on “the Kaiser effect” in engineering materials. He found that the Kaiser effect was observed up to around 75% load level of failure load,

AE behavior AE behavior under under compression compression in concretein concrete

L’Hermite R G (1960) Volume change of concrete. Proc. 4th Int. Symp. Chemistry of Cement, V-3, NBS Monograph 43:659-694

Development of AE 　 Technology Following Schofield’s study, A. T. Green and

H. L. Dunegan were known to develope standard procedures and devices.

Proof tests of

Tanks

Rocket motor-cases

Pressure vessels

AE Committees & Conferences

H. R. Hardy, Jr. organized and held the series of five Conferences on Acoustic Emission/Microseismic Activity in Geologic Structures and Materials [1974, 1978, 1981, 1985 and 1991].

In the U. S. A., the Acoustic Emission Working Group (AEWG) was conceived in 1967 by J. C.Spanner.

The 6th International Conference on Acoustic Emission: 50th AEWG Meeting - 40th Year AEWG Anniversary

Lake Tahoe, Nevada, USA, October 29 to November 2, 2007

AE Committees & Symposia . The Japanese Committee on AE was founded in 1969.

Since 1980 the committee has been organized as an ad hoc technical committee in the Japanese Society for Nondestructive Inspection (JSNDI). The International Acoustic Emission Symposia (IAES) were inaugurated in 1972 and has been biennially held in Japan.

In Europe, research activity on AE led to the European Working Group on AE (EWGAE). The first meeting was held in 1972, and in 2006, the 27th meeting was held

Journal of Acoustic Emission

Professor Kanji Ono at University of California, Los Angeles has been editing the Journal of AE, based on the exponentially increasing number of papers.

The first issue was published in 1982. Since then, the journal has kept providing the state of the art on AE researches.

Fundamentals of AE Measurement

Basic system

Amplification: dB = 20 log10(Aoutput/Ainput)

40 dB – 100 dB Band-pass filter: high-pass around 10 kHz

low-pass 300 kHz, 1 MHz

Fundamentals - 1

F. Kishinoue already made comments on the problems with background or environmental noises. Care is needed during the experiment.

This is because detection of AE signals is affected by vibrations from a string wind, passers-by and a truck passing on a nearby street. Many of these problems have been eliminated with development of instrumentation systems.

Fundamentals - 2 In updated equipments, the frequency range of

the measurement is normally set above that of audio or environmental noises, which are substantially minimized by grounding the equipments.

Owing to advances of measuring systems, the use of a band-pass filter effectively eliminates background noises and allow meaningful tests under usual laboratory environments.

Remarks on Mathematical Backgrounds

Tensor notationEquation of equilibriumFourier transformsLinear systemEigen-value analysis

1-2 Measurement – Sensors 1-2 Measurement – Sensors October 4, 2007 (2/3)October 4, 2007 (2/3)

Introduction Sensor and System Response Response of PZT Sensors Calibration and Detection

The elastic (AE) waves propagate inside a material and are detected by an AE sensor. Except for contactless sensors, AE sensors are directly attached on the surface.

Typical AE sensor

In the most cases, a piezoelectric element in a protective housing is applied. The sensors are exclusively based on the piezoelectric effect out of lead zirconate titanate (PZT). PZT sensors provide the best combination

of low cost, high sensitivity, ease of handling and selective frequency responses.

Although PZT sensors are not normally suited for broad-band detection in basic studies of AE waveform analysis, they are practically useful for most AE experiments and applications.

System Response: Linear system

The convolution integral is defined,  s(t) = ∫f(t-)w()d = f(t)*w(t).   Then Dirac's delta function (t) plays an important role. f(t) = f(t)*(t). In the case of the linear system,   g(t)= L[f(t)*(t)] = f(t)*L[(t)]. Setting L[(t)] as w(t), we have, g(t) = f(t)*w(t).  Introducing the Fourier transform,  G(f) = F(f)W(f)

g(t) = L[ f(t) ]

Linear SystemLinear System

a(t) = wf(t)*wa(t)*w(t)*f(t)

A(f) = Wf(f)Wa(f)W(f)F(f)

Electro-mechanical vibrations of PZT bodies can be solved in a manner similar to the corresponding mechanical vibration problem, but with additional variables. The constitutive laws of PZT materials are represented,

{} = [C]{} + [d]T{E}, {D} = [d]{} + [p]{E}. Here { are the elastic strains, { are the stresses, {E} are the electric

potentials and {D} are the electric displacements. [C] represents the adiabatic elastic compliance tensor at constant electric filed, [d] is the adiabatic piezoelectric tensor and [p] is the adiabatic electric permittivity.

One-dimensional motion assumed leads to nominal frequencies as,

a thickness resonance,

a radial resonance etc.

Comparison of Resonant FrequenciesShape

Free vibration (kHz) Fixed boundary (kHz)

Nominal FEM analysis Experiment Nominal FEM analysis Experiment

Cylindrical element

71(shear)

160(comp.)

201(radial)

109

196

207

217

298

133

210

251

312

71(shear)

160(comp.)

201(radial)

61

167

191

245

316

42

152

204

Disk-shaped element

201(radial)

597(shear)

116

206

322

541

581

42

91

146

166

231

536

1343(comp.) 382

766

1219

1476

2245

291

516

720

Conical

element

80

262

414

567

774

66

278

494

713

404

803

878

1099

1874

404

500

570

766

Absolute calibration by a Capacitive Transducer (NIST)

Relative calibration by the reciprocity method (Hatano)

Step-function force:

Glass-capillary break

(Breckenridge)

Pencil-lead break

(Hsu)

Capacitive transducer:

=== Lamb’ problem ===

Laser-Doppler vibrometer

Lamb’s solutions of velocity motions detected and calculated.

Time (sec)

Sensor calibration by NIST Fs(f):Spectrum of a

detected wave by AE sensor

Fa(f):Spectrum of a detected wave by a capacitive transducer

[Both due to pencil-lead break]

R(f) = Fs(f)/Fa(f)[Displacement calibration]

Coupling with AE sensorsCoupling with AE sensors For the PZT sensors, coupling between sensors and a

member is important due to the low amplitudes of AE signals.

Various methods exist for fixing the sensors to the structure. Adhesives or gluey coupling materials and couplant like wax or grease are often used due to their low impedance. If the structure has a metallic surface, magnetic or immersion techniques are widely used.

In general, the coupling should reduce the loss of signal energy and should have a low acoustic impedance compared to the material to be tested.

Sensor mounting An essential requirement in mounting a sensor is

sufficient acoustic-coupling between the sensor’s surface and the structure surface.

Application of a couplant layer should be thin, so it can fill gaps caused by surface roughness and eliminate air gaps to ensure good acoustic transmission.

Commonly used couplants are vacuum greases, water-soluble glycols, solvent-soluble resins, and proprietary ultrasonic couplants.

Sensor stationary In addition to coupling, the sensor must always be

stationary. One way to achieve this is to use glue, which can also serve as a couplant.

To prevent attenuation, air bubbles and thick glue layers should be avoided.

Another way to help a sensor stationary is to use a holding devices such as tapes, elastic bands, springs, magnetic hold-downs, and other special fixtures.

It is important that any mechanical mount does not make electrical contact between the sensor case and the structure. Accordingly, grounding the case is often necessary.

1-3 Measurement – Instrument1-3 Measurement – InstrumentOctober 4, 2007 (3/3)October 4, 2007 (3/3)

Detection and Other Type Sensors

Instrument Data Acquisition and

Parameters Flaw (Source) Location

WaveguidesWhen the sensor can not be attached to directly the structure, waveguides are employed. It is noted that the use of waveguides introduces further complexity to frequency contents of AE waves.

Other-type sensor 1 Laser system has been

applied for AE detection. It is a contactless measurement but less sensitive than the PZT sensors.

PZT sensors have a limitation in application at elevated temperature, because PZT has Curie point.

Other-type sensor 2

An optical fiber sensor is a new and an attractive AE sensor as alternative to the PZT sensor.

It can offer a number of advantages such as the long-term monitoring, the condition free from electro-magnetic noises, and the use of corrosive and elevated environments.

Instrument - Amplifier

Because cables from the sensor to the amplifier are subjected to electro-magnetic noise, specially coated cables of short length shall be used.

Amplifiers with a flat response in the frequency range are best use.

AE signals are normally amplified both by a pre-amplifier and by a main-amplifier.

The gain of the amplifier is given in dB (decibels),

dB = 20 log10(Vo/Vi).

In concrete and rocks,

60 dB to 100 dB in total

A filter of variable band-width between a few kHz and 2 MHz is generally employed. The choice of the frequency range depends on noise level and attenuation property of concrete.

When the wave with energy level E is attenuated by E over one-wavelength propagation,

parameter Q is defined as,

Q = 2E/E.  

In the case of a pure elastic material, E = 0 and Q = infinite.

The larger Q is, the lower the attenuation is.

Q is larger than 1000 for typical metals, while Q is reported as lower than 100 in concrete.

Attenuation – dependent on distance and frequency

When AE waves propagate for distance D, the amplitude U(f) of frequency components f attenuates from U0 to,

 

U(f) = U0exp (-fD/vQ)

Substituting frequency f = 1 MHz, distance D = 1 m, velocity of P wave v = 4000 m/s, and Q = 100,

the attenuation U(f)/U0 becomes –68 dB/m.

The higher frequency components are, the more quickly they attenuate.

Data Acquisition

Main concern for data acquisition results from the A/D(analog to digital) conversion and the triggering. Fast A/D units have to be used to ensure that a large number of events are recorded

A/D converter is equipped for each channel of the recording unit. Anti-aliasing filters are required so that the signals can be properly transformed to the frequency domain.

A monitoring system can analyze such parameters as count, hit, event, rise time, duration, peak amplitude, energy, RMS (root mean square) voltage, frequency spectrum, and arrival-time difference as discussed later.

Normally AE signals are processed after the amplitude becomes larger than the threshold level.

AE waveform parameters

Threshold Rise time Duration Maximum amplitude Energy

(RMS) voltage Hit Counts (ringdown) Arrival times

Flaw (Source) Location Procedure

Equation of hyperbolas

1

2221

21 )()( tVyxbyax p

2222

22

2 )()( tVyxbyax p

3222

32

3 )()( tVyxbyax p ,

Ideal Solution

Three hyperbolas meet at one point.

This is the case where no errors are contained to measure the arrival-time differences.

Solution procedure

)()()()(2)(2 21

212

22

2211

222

21

221122112 batbatttttVytbtbxtata P

Taking a square of hyperbolic equations and substracting them,

)()()()(2)(2 22

223

23

2322

233

22

232233223 batbatttttVytbtbxtata P

Solution of the first approximation

The solution is obtained as the intersection of two lines L1 and L2, because three hyperbolas can not meet at the one point.

Technical problem Although the derivation of these equations is simple

and the software is commercially available, errors are generally inevitable because of mis-reading the arrival times. Usually, the arrival times are estimated from the time when AE signal amplitude become higher that the threshold voltage.

It definitely depends on the frequency contents of the signal. In the case that the noise level is high, the arrival time is often contaminated and difficult to be read.

Treatment for improved solution by Taylor’s expansion

ymxlRR ai

ai

aii

22 )()( ia

ia

i byyaxxR

22 )()( ia

iaa

i byaxR

ai

iaai

R

axl

a

i

iaai

R

bym

Iteration procedure

By applying the least-square method, a series of linear algebraic equations on Dx and Dy are solved. By employing an iteration procedure, a converged solution can be obtained after estimating the error, i

Ti.

ymmxllRRtV aaaaaaP )()()( 01010111

Treatment for Velocity Anisotropy

2222 mVlVV yx

0

22

1

21

21

1

)()()()(

V

yyxx

V

byyaxxt

aaaa

21

221

21 mVlVV yx 2

022

02

0 mVlVV yx

Iteration for solution

Applying the least-square method, the above equations can be solved.

0

0

0

0

0

0 )()(V

R

V

Rty

V

m

V

mx

V

l

V

l a

i

ai

i

a

i

ai

a

i

ai

Sensors and AE equipments are already commercially available. In this concern, important aspects are to select proper devices and systems.

Depending upon materials and structures, selection of sensors, decision of frequency range, techniques to eliminate noises and conditions for system setting may change.

Basic treatment of the flow (source) location is presented. Although the procedure is commercially is available, it is essential to know a mathematical viewpoint. This is because sources are not always determined at proper locations.

[Homework : Nr. 1] reply to e-mail: mohtsu@ethz.ch  

Short Course 101-1111-00L: Fundamentals and Applications of Acoustic Emission by Prof. Masayasu Ohtsu

October 4, 2007 1-1 Considering the case where AE sensors, T0, T1, T2, T3 are arranged at (0,0), (0,1), (1,0) and (-1,0), respectively. Assuming that the arrival-time differences times the velocity; T1 from T0,

T2 from T0 and T3 from T0 are 1/2, 4

117 , and

4

117 , respectively, obtain three

hyperbolas and solve them to determine the source coordinates. 1-2 Determine the source coordinates, in the case the arrival-time differences times the velocity

are mistakenly estimated as d, 4

117 , and

4

117 .