AE source location and

o r i e n t a t i o n d e t e r m i n a t i o n of
tensile cracks f r o m surface
observation
M. Ohtsu and K. Ono

A technique to determine the location and orientation of an acoustic emission (AE)

source due to a tensile crack was studied. Theoretical AE waveforms from tensile
cracks were synthesized on the basis of the generalized theory of AE. Numerical
experiments were conducted using the waveforms. We considered a source location
technique using only surface observations and took into account the effects of
velocity anisotropy. The technique is based on the radiation pattern of P-waves and
incorporates a procedure to determine the orientation of the AE source. The results
accurately reproduced the location and orientation of simulated tensile cracks as
well as the values of their moment tensor representation.

Keywords: acoustic emission, tensile crack, source location, velocity anisotropy,

P-waves, numerical experiments

L a m b ' s problem of determining transient elastic disturb- cannot be readily employed because of a huge number
ances at a surface due to a dynamic point force in a of combinations of feasible source characteristics and
half-space is well known in seismology. Breckenridge possible propagation paths. In order to completely
et alm applied this concept to the study of acoustic determine the source characteristics and locations of AE,
emission (AE) sources. They developed a transient point one may have to use a large array of sensors.
force generator and a wide-band capacitive sensor of
In the present paper we discuss the development of a new
surface displacement, and showed experimental AE wave-
technique for determining the location and orientation
forms similar to theoretically predicted ones. Their success
of a tensile crack from only surface observations of
bred theoretical treatments of AE waveform analyses
AE waveforms. The last restriction is realistic as this
based on dislocation models and Green's functions 12-m. In
condition is typically imposed in practical applications
these studies, AE sources are represented by dislocation
of AE.
models or equivalent force models and correspond to
dislocation movements or cracking in a solid. The models
are prescribed by the discontinuities of displacements or Mathematical background
forces. In addition, since these motions are dynamic, time Governing equations and characteristics of
dependencies at AE sources (also known as source elastic waves
functions) must be given or found.
In elastodynamics we can ignore the body force in the
When source characteristics are specified, AE waveforms equilibrium state. In terms of displacement u(x, t) or its
can be calculated by convolving the source function with component u~(x, t) the equilibrium equation becomes
an appropriate Green's function. This approach is known
as a forward problem or simulation analysis. When one (.~ + I~)Uj,ji(x, t) + ttui,jj(x, t) = pffi(x, t) (1)
attempts to obtain the source characteristics from a
measured AE waveform and the Green's function, this is where 2 and # are Lam6 constants and p is the material
referred to as an inverse problem or deconvolution density. Here ui4(x, t) means a differentiation of ui with
analysis. respect to x j, and a dot means a differentiation with time.
To find the fundamental characteristics of a solution for
Some attempts at obtaining source characteristics from
Equation (1), we consider plane waves propagating
observed AE waveforms have been made t7'81. However,
through a medium with velocity v[9]:
since such deconvolution analyses are burdened with a
number of discrepancies between theoretical assumptions
and experimental conditions, only limited success has ui(x, t) = Ji(I-x - vt) (2)
been reported.
where I is the direction cosine of the propagation
In practical applications of AE, so far, inverse problems direction. Substituting (2) into ( 1 ), we have a relationship

0308 9126/88/030143 08 $3.00 (c) 1988 Butterworth & Co (Publishers) Ltd

NDT International Volume 21 Number 3 June 1988 143
 [(;'.+/~)lilk+(/~ t)r:)bik].!~ = 0
// 1 ~
where a prime indicates a differentiation. This leads to . , \\
the characteristic equation / i \ "\
~rlciden~ , / ' ~ . .... ' ~ "N ReflecteG
(2+21l---pr2)(IL pV2)2=0 (4)

When the function ./i(x,t) in (2) has velocities given

by solutions in (4), the equilibrium equation (1) is
automatically satisfied. F r o m the solutions of Equation (4) i Reflected
we have characteristic wave velocities S-wove
x3
%=[(~,+2/~)/p] '`z v ~ = ( # / p ) ','2 Fig. 1 Reflections of P-wave at the boundary in a half-space

where Vp and v~ are the velocities of the primary wave

(P-wave) and the secondary wave (S-wave) respectively.
Half-spece
D y n a m i c motions in an elastic body always generate
P-waves, which arrive first at any observation point. In
an AE experiment it is often difficult to detect P-waves
because of their small amplitudes. They are, however,
non-dispersive, whereas almost all the surface waves and
other waves generated by diffraction are dispersive. Since
the velocities of dispersive waves are dependent on
frequency components, AE arrival time differences of
P-waves are most suited for a source location technique.

Potential formulation and reflection

coefficient
Since AE wave motions are detected at a stress-free
surface, we must consider the effects of reflection. Solutions
of E q u a t i o n (1) can be presented in the form of
Equation (2). By using a similar formulation, we can
a
obtain the reflection coefficient at a surface.
In the case of the incident P-wave shown in Figure 1, AE transducer
solutions u~(x,t) and u3(x,t) are represented by using
potentials G and H in the steady state tl°J
D
G = A 1 exp [(io)/Vp)(X 1 sin c~- x 3 cos ~ - vpt)]
(5)
+ A z exp[(ioo/vp)(xl sin ~ + x 3 cos ~ - Vpt)]

H = B 2 exp[(ko/v,)(x 1 sin/3 + X 3 COS/J - - vst)] (6)

where A~, A 2 and B 2 are the amplitudes of the incident AE source

P-wave, the reflected P-wave and the reflected S-wave
respectively, and 09 is the angular frequency of the incident b
P-wave. Displacements u a (x, t) and u 3 ( x, t) are obtained Fig. 2 (a) Tensile crack embedded in a half-space. (b) Sketch of
from the relations AE wave propagation

Ul(X,t ) = G1 + H.3 (7) incident P-wave:

u3(x,t)=G 3-H1 (8) u3/u o = 2k2(k 2 - 2 sin2~)/[(k 2 - 2 sinZa) 2

+ 4 sin2~ cos ~(k 2 - sin2~)1/2 ] (11)
F r o m Snell's law with respect to the reflection at a
stress-free surface (x 3 = 0 in Figure 1), we have the where u 0 is the incident displacement in the x3 direction
condition due to the incident P-wave, and k = vp/v~.
vp/sin c( = Vs/sin [:¢ (9) Integral representation and radiation pattern
We can write [11] the displacement field Ui(x,t) due to a
and at the stress-free surface
dislocation in terms of the Burgers vector bk(y,t)(see
Figure 2a):
G13 = G33 = 0 at x3= 0 (10)

Using Equations ( 5 ) - ( 1 0 ) , we obtain the reflection ui(x,t)= Tik(x,y,t)*bk(y,t)dS(y ) 112)

coefficient for the resultant displacement u 3 due to the

144 NDT International June 1988

where T~k is a Green's function of the second kind (indicating T0(O,0,0)
the traction field associated with the Green function Gip),
* indicates a convolution and F indicates the area of the
dislocation. For an isotropic solid, the integrand in
Equation (12) is also expressed by using a moment tensor
M, a component of which is mpq, as

Tik* b k = G ip.q*mpq (13)

Y Ti+l(Ot÷l,bi+l,O )
i , )
where

mpq = ~.bknk6pq + Itbpnq + #bqnp (14)

z
Here n k is a component of the unit normal to the Fig. 3 AE source and transducers array for surface observation
dislocation surface F. From Figure 2a we can set
located at the origin of the coordinate system, and
nk = (cos 0 cos 4>, cos 0 sin ~b, sin 0) transducers T i at (ai, bi, O) and Ti+l at (ai+l,bi+x,O)
are located on a surface. The distances Ro(x,y,z) and
We obtain a relationship between the amplitude of Ri(x, y,z) are expressed as
the incident P-wave and the crack orientation from the
radiation pattern t12'131. Consider a tensile crack, which Ro(x,y,z ) = (x 2 + y2 + Z2)1/2
can be represented by the Burgers vector of a crack (17)
dislocation, bk; that is, Ri(x,y, z ) = [ ( x - ai) 2 + ( y - bi) 2 + z z ] 1/2

bk(y, t) = bnkf( y ) H ( t ) (15) Considering the time difference tl between P-wave

propagation to transducer Ti and to transducer T o, we
where 6(y) is the Dirac delta function and H(t) is the obtain N - 1 non-linear equations with three unknowns
Heaviside step function. x, y and z as follows:
AE wave motions propagate as shown in Figure 2b. Ri(x,y,z)/vp-Ro(x,y,z)/vp=t i (i=I,N-1) (18)
However, reflected waves from other surfaces have
essentially lost the information on the crack orientation. where N is the number of AE transducers. In order to
Therefore we must consider only direct P-waves to linearize the non-linear equations, we square the ith
determine the AE source characteristics. Under this equation and subtract it from the square of the (i + l)th
restriction, we can account for the effects of reflections equation. We now have N - 2 linear algebraic equations
at an observation point by using the reflection coefficient with unknowns x and y, which are expressed as
given in Equation ( 11 ). xPn + YPI2 = qi or, in matrix form,
Because we consider only the dominant terms due to
P-waves, a Green's function of the second kind in an infinite
space is employed instead of that in a half-space. By the
substitution of Equation (15), Equation (12) gives the
incident wave displacement field u0(x,t ) in the x 3 where
direction due to a tensile crack as follows:
Pil = --2(aiti+l --ai+lti) Pi2 = -2(biti+l --bi+ltl)
Uo(X, t) = p b [ 2 v / ( 1 - 2v) + (l,,n,.) 2 ] cos ot6(t)/(47tpv3r)
(16) q i : v p ( t2i + l t i
2 - - t i t 2 + 1 ) -(a~+b~)ti+ 1 +(a~+ 1 +bE+1)t,
(i----1, S - - 2 )
where r is the distance from a crack to the observation
point. A component of its direction cosine is 1,,, and v is Here [ P ] is an N - 2 by 2 matrix and [q] is a column
Poisson's ratio. matrix with N - 2 elements. When more than five
transducers are employed, the least-square technique can
Source location technique be used in solving Equation (18). By multiplying [ p ] T
on both sides of (19), we get the algebraic equation
Least-square method
The basic concepts of a source location technique are [p]J[p]{~} = [p]T[q] (20)
well known. Devices for two- and three-dimensional AE
source location are commercially available. The three-
dimensional source location technique has been applied where [ p ] T represents a transposed matrix. By solving
successfully in a laboratory [a4] and in an experiment for Equation (20), we can determine unknowns x and y.
source characterization [as]. To provide needed back- Using Equation (18), unknown z is then determined.
ground for the technique developed, we briefly discuss a
three-dimensional source location technique using surface Effect of velocity anisotropy
observations. The elastic wave velocity of materials is usually anisotropic.
Figure 3 shows an AE source S at (x, y, z) and a transducer When the P-wave velocity is only weakly anisotropic - for
array To ..... T~, T i + 1, where T o is the reference transducer example, within approximately 20 % variation - the effect

NDT International J u n e 1988 145

of the velocity anisotropy can be accounted for by an solving for dx, dy and dz, we repeat the same procedure
iteration procedure [161. as represented by Equation (22). After several iterations
we can obtain the converged solution for x, v and z.
We denote the P-wave velocities in the x, y and z
Ohtsu 116j confirmed the convergence of this procedure.
directions by vx, vy and v= respectively. The direction
cosine from the AE source S to a transducer T i is
(1~,1i2,1i3). The P-wave velocity v~ for this direction is
given by the relation [~v] Determination of crack orientation
S i m u l a t i o n analysis
12i = [(l)xlil )2 + (/)y112)2 q_ (V21i3)2] 1/2 (21)
We conducted numerical experiments in order to locate a
We apply the iteration procedure after we have obtained tensile crack and determine its orientations from the
a solution of Equation (20) as the first approximation surface AE observations. We calculated AE waveforms
(Xa, Ya, Za) using the mean value of the P-wave velocity, due to a tensile crack in a half-space as shown in Figure
(vx+vy+v=)/3. Next we adopt a solution (x,y,z)= 2a, using the procedure discussed previously II~l. Some
(x. + dx, y. + dy, z. + dz). After Taylor's expansion we of the calculated dynamic surface displacements due
obtain the following perturbation relation with respect to inclined tensile cracks which open in the direction
to the distance R~: n = (cos 0 cos ~b, cos 0 sin qS, sin ~b) and are located at a
depth of 0.05 m are shown in Figure 4. The geometry
R i ( x , y , z ) = Ri(xa, Ya, Za) + 1il d x + llzdy + li3dz (22) of such a crack is shown in Figure 2a. In these
examples q5 = 0 and 0 = 10 °. AE waveforms are observed
where at ( x l , x : ) = ( 0 . 0 , 0 . 0 ) , ( - 0 . 0 5 , - 0 . 0 5 ) , (0.1,0.1) and
( - 0 . 1 , - 0 . 1 ) . The distance is given in metres and the
amplitudes of these figures are relative. The scale depends
Ill = (X. -- ai)/ R i ( x a, ya,za)
on the magnitude of a crack opening and the area of a
li2 = (Ya -- bi)/ Ri(xa, Y~, z.) crack. The following source function flu is used in the
calculation:
li3 = za/ Ri(xa, Ya, Za)

Substituting (21 ) and (22) into (18), we have a set of linear

algebraic equations with unknowns dx, dy and dz. After

~ P

0 I I
b I I
0 20 40 0 20 40

e I a
d I I
0 20 40 0 20 40
Time (10"e s) Time (10"es)

Fig. 4 SimulatedAEwaveformsobservedat(a)(0.0,0.0),(b)( 0.05,-0.05),(c)(0.1,0.1)and(d) (-0.1,-0.1).Sourcedepth=0.05m

146 NDT International June 1988

The rise time of the source function, ~, was 1 /~s. In all observation of AE in a thick plate. A procedure for the
the waveforms in Figure 4 the P-wave is observed as the source location is based on Equation (20). For the
first positive peak. In Figure 4a the S-wave is seen as a determination of the orientation of a crack, Ohtsu [131
negative peak of small amplitude. Only P- and S-waves previously attempted to do this in the two-dimensional
are detected at the epicentre. In the three other waveforms problem by directly solving Equation (16). In general
the P-wave is followed by an SP-wave of small positive three-dimensional cases, however, Equation (16) has
amplitude. The S-wave is always observed as a negative more than one solution for the unit normal n,, and
peak. Because of the inclination of the crack, the cannot be solved directly to obtain a unique solution.
waveforms at (0.1,0.1) and ( - 0 . 1 , - 0 . 1 ) are different, Consequently, we developed a new exhaustive procedure
although the distances from the crack to the observation to determine the crack orientation. Here we compute the
points are identical (Figures 4c and d). modified radiation pattern based on Equations (11 ) and
(16) for all cases of inclined angles with a 2.5 ° increment.
When a boundary exists as shown in Figure 2b, reflected
(The increment can be reduced to refine the computation.)
waves are superimposed on the waveforms shown in
These are compared with the spatial distribution of the
Figure 4. This implies that the quantitative treatment of
amplitude of a P-wave. The orientation corresponding
AE waveforms may be feasible only for P-waves, because
to the best-fit pattern is selected as the crack orientation.
P-waves are readily discriminated from other waves.
Radiation pattern Results
The radiation pattern of P-waves in a half-space was We evaluated the procedures outlined in the previous
obtained by calculating u 3 components on the free surface. subsection using calculated waveforms of known location
Figure 5 shows two representative cases. The solid points and orientation. Table 1 lists the positions of nine sensor
in the figure give the amplitude of P-waves normalized locations. The inclination angles of the tensile cracks
by that at the origin, u~. The crack is oriented at 0 = 0 tested are shown in Table 2. The cracks are placed
and 10 °. The displacement was obtained at x ~ ~< + 0.1 m at the origin. Simulated AE waveforms at the nine
and x 2 ~<0.1 m. In order to deduce a simpler expression observation points were computed in each case. Taking
for the amplitude, we employ the far-field approximation the waveforms as the AE signals from an unknown source,
in an infinite space [18] in conjunction with the reflection we attempted to locate and orient the source. We assume,
coefficient. For a tensile crack we combine Equations ( 1 1) however, that it is a tensile crack. By employing the
and (16). The results are again normalized by the program developed, AE sources were located from time
displacement u~ observed at the origin and are plotted differences of P-waves observed. Next, crack orientations
by solid lines in Figure 5. These correspond to cross- were searched from the spatial distribution of P-wave
sections of the radiation pattern of P-waves. It is obvious
that the radiation pattern in an infinite space modified Table 1. Locations of transducers
by the reflection coefficient agrees well with the radiation
pattern of P-waves obtained in a half-space. This implies
Number (Xl, x2) (m)
that the crack orientation can be recovered on the basis
of the modified radiation pattern of P-waves.
1 (-0.05, 0.05)
Location and o r i e n t a t i o n 2 (0.0, 0.05)
We consider the geometry of a tensile crack as shown in 3 (0.05, 0.05)
Figure 2a. It simulates the condition of the surface 4 (-0.05, 0.0)
5 (o.o, o.o)
6 (0.05, 0.0)
IC = 7 ( -o.05, -o.05)
8 (0.0, -0.05)
9 (0.05, -0.05)
On
o~
Table 2. Comparisons of crack orientations
b e t w e e n cases tested and results
oE 0.6

+ Cases tested Results obtained

(3. (e,~) (e,~)
0.4
(o.o, o.o) (o.o, o.o)
Z (10.0, 0.0) (10.0, 0.0)
(20.0, 0.0) (20.0, 0.0)
0.2
X2 =Oil m x2 = 0.1 rn (90.0, 9o.o) (9o.o, 90.0)
(30.0, 0.0) (32.5, 0.0)
(45.o, o.o) (5o.o, o.o)*
, , ; , , , , , , , (0.0, 15.0) (0.0, 15.0)
0 -0.1 -0.05 0.05 OI -0.1 -005 0 0.05 0.1
(0.0, 45.o) (0.0, 45.0)
Distance x I (m)
(10.0, 15.0) (10.0, 15.0)
Fig. 5 Comparisons between c o m p u t e d P - w a v e amplitudes and (45.0, 45.0) (50.0, 45.0)
those of the radiation pattern

NDT International June 1988 147

 . . . . . . . . . . . . .
amplitudes. A typical output of the computer program
is shown in the Appendix. The upper half of the
computer output shows the results of determining crack
orientations. First, the x and y coordinates are determined 0 = 45 °
by the source location. Next, the z coordinate is
determined from Equation (18). Because solutions are O_
5o-~-~ -
obtained by the least-square method, Equation (18) is
not necessarily satisfied with respect to all the observation <~

points. Therefore we determined the z coordinate as the 0.5

minimum error solution. The output in the Appendix
shows the minimum error which was encountered at the
eighth observation point in Table 1. The error distance
is equal to zero, although the source depth z 1 is obtained
0 I 2 3 4
as 0.051 instead of the exact solution z =0.050. The
accuracy in depth is generally limited using a transducer Time (lO-6s)
array located only on a surface 119]. Fig. 6 Result of the deconvolution analysis

After determining the location of an AE source, the

coordinates are transformed so as to place the epicentre
of the AE source at the origin of the coordinate
system. Next, the locations of the observation points are crack orientations with a limited precision ( + 2.5 °) can
rearranged as transducer locations (x 1, Yl )- Comparing be effective in AE source characterization.
these results with observations in Table 1, it is seen that
the accuracy with respect to the x and y coordinates is Elements of m o m e n t tensor
perfect. Source characteristics can also be investigated using the
moment tensor. The representation of the moment tensor
Next, inclination angles are determined from the best-fit
M or its components mp~ is shown in Equation (14). Kim
pattern. Results are listed for both horizontal angle (oh and Sachse/7] determined a moment tensor rigorously
in Figure 2a) and vertical angle (0). In the bottom portion
based on Equation (13) and the spatial derivatives of
of the output (see Appendix), normalized amplitudes of
Green's functions in an infinite plate. It is also possible to
P-waves are indicated for both amplitudes obtained from
utilize the radiation pattern of the P-wave for determining
the radiation pattern (UP) and those from the simulation
the elements of a moment tensor.
analysis (UR).
We attempted to determine the elements of a moment
By employing this program, crack orientations were
tensor for three cases of different 0 and ~b as indicated in
determined in ten cases. Results obtained are summarized
Table 3. We considered only the dominant term of the
in Table 2. In almost all cases, good agreement with the
P-wave in Gip,q*mpqof Equation (13) and obtained the
original crack angles ('cases tested') is observed. The
representation
horizontal angle (4~) was always obtained accurately.
In three cases (0,~b)=(30.0°,0.0°), (45.0~,0.0 °) and Uo(X) = blplqmpq/4~pv3rcos (23)
(45.0 ~, 45.0'~) - slight errors in the vertical angle (0) were
observed. The errors were observed only in the cases
It follows that displacement U3(X ) at an observation
where the vertical angle was intermediate between the
point is represented by the reflection coefficient in
horizontal plane and the vertical plane. This probably
Equation (11). Since we have obtained synthesized
results from the same reason as that which produced a
waveforms at nine observation points, theoretical ampli-
poor accuracy of the depth in the source location.
tudes u3(x ) of P-waves are known. Substituting these
data calibrated by the reflection coefficient into the
Discussion left-hand side of Equation (23), we obtained a set of linear
algebraic equations and determined elements of the
Effect on d e c o n v o l u t i o n analysis moment tensor, mpq. In Table 3, results obtained were
Ohtsu and Ono |2°1 showed that an error in the inclination normalized by the maximum value and compared with
angles of a crack leads to a false source time function in elements obtained by substituting values of 0 and q~ into
deconvolution analysis. We performed the deconvolution Equation (14). These are referred to as theoretical values
analysis in the case marked * in Table 2. For 0 = 45 °, in the table. Although we employed only the radiation
the AE waveform observed at the epicentre was decon- pattern of P-waves, the elements of a moment tensor
volved by using Green functions of the second kind in a are recovered with quite a small error. In all cases,
half-space computed for 0 = 4 5 ° and 50 ° . The latter discrepancies between theoretical values and results are
corresponds to the case where the crack orientation was less than 10%. This implies that a simple procedure based
determined 5 ° in error and the deconvolution analysis on the radiation pattern of P-waves can be employed for
was carried out with an incorrect 0 of 50 °. Both results determining the elements of a moment tensor of not only
are shown in Figure 6. The source time function obtained a tensile crack but also of a shear crack.
from the incorrect inclination angle ( 0 = 5 0 ° ) differs
slightly in the plateau region from the exact solution
(0 = 45"), but the initial portion was identical. This result
Conclusions
suggests that the 5 ° error in the vertical angle can be A program to determine source locations and crack
tolerated in the determination of the source time function orientations from the surface observation of AE signals
and that the present technique for the determination of is developed. The feasibility of this technique is investi-

148 NDT International June 1988

Table 3. Elements of t h e m o m e n t tensors

(0, 4') Theoretical values Results obtained

1.000 0.000 0.099 1.000 0.000 0.101

(10.0, 0.0) 0.439 0.000 0.394 0.000
sym. 0.454 sym. 0.437

1.000 0.149 0.000 1.000 0.156 0.000

(0.0, 15.0) 0.485 0.000 0.446 0.000
sym. 0.446 sym. 0.427

1.000 0.149 0.100 1.000 0.154 0.101

(10.0, 15.0) 0.492 0.027 0.452 0.027
sym. 0.471 sym. 0.455

gated by numerical experiments. Results obtained are References

summarized as follows.
1 Breckenridge, F.R., Tschiegg, C.E. and Greenspan, M. "Acoustic
• The mathematical background for determining the emission: some applications of Lamb's problem" J Acoust Soc Am
57 3 (1975) pp 626 631
locations and orientations of tensile cracks is presented.
2 Pan, Y.H. 'Theory of acoustic emission' Elastic Waves and
It includes characteristics of P-waves, the reflection Nondestructive Testing of Materials, A M D Vol 29 ASME, New
coefficient at a stress-free surface, a generalized theory York (1978) pp 107 128
and the radiation pattern. All concepts are employed 3 Wadley, H.N.G., Scruby, C.B. and Shrimpton, G. 'Quantitative
to develop a technique for the determination of acoustic emission source characterization during low temperature
cleavage and intergranular fracture' A ERE-R9644 AERE, Harwell
locations and orientations of tensile cracks from
(1980)
surface observations. 4 Hsu, N.N. and Eitzen, D.G. 'AE signal analysis laboratory
• A source technique based on the least-square method experiments examining the physical process of acoustic emission'
is briefly discussed. A technique for a material with Proc 5th lnt AE Syrup JSNDI, Tokyo (1980) pp 67 78
elastic anisotropy is also presented. 5 Ohtsa, M. "Source mechanism and waveform analysis of acoustic
emission in concrete' J Acoust Emission 1 2 (1982) pp 103-112
• Simulation analyses for the surface observation of AE 6 Ohira, T. and Kishi, T. 'Quantitative characterization of dynamic
signals are carried out. AE waveforms due to tensile fracture process by means of acoustic emission source wave
cracks in a half-space are computed and assumed to analysis" Progress in Acoustic Emission eds M. Onoe et al, JSNDI,
be detected at observation points on a surface. Results Tokyo (1982) pp 208 217
7 Kim, K.Y. and Saehse, W. 'Characteristics of acoustic emission
show that the calibrated radiation pattern of P-waves signals from indentation cracks in glass" Progress in Acoustic
is reasonably coincident with the amplitude distribution Emission I1 eds M. Onoe et al, JSNDI, Tokyo (1984) pp 163 172
of P-waves obtained in a half-space. 8 Ohira, T. and Pao, Y.H. "Acoustic emission source characterization
• A F O R T R A N program to determine locations and of microcracking in A533B steel' Review of Quantitative Non-
destructive Evaluations Williamsburg, VA (1985)
orientations of tensile cracks from surface observations 9 Love, A.E. A Treatise on the Mathematical Theory ~ Elasticity
is developed on the basis of the source location Dover Publications, New York (1944)
technique and the radiation pattern. The applicability 10 Graft, K.F. Wave Motion in Elastic Solids Oxford University Press,
of the new technique is confirmed by numerical Oxford (1975)
experiments. 11 Ohtsu, M. and Ono, K. 'A generalized theory of acoustic emission
and Green's functions in a half space' J Acoust Emission 3 1 (1984)
• The technique developed is based only on the amplitude pp 27-40
distribution of the P-waves. Therefore we attempt to 12 Mal, A.K. 'Interaction of elastic waves with a penny shaped crack'
determine elements of a moment tensor from the lnt J Eng Sci 8 (1970) pp 381 388
amplitude distribution of the P-waves. This shows that 13 Ohtsu, M. 'Radiation pattern of acoustic emission" Zairyo 32 356
(1983) pp 577 583
information on source characteristics is recovered 14 Yanagidani, T. et M ' L o c a t i o n of dilatancy in Ohshima granite
primarily from the P-wave portion. This concept under constant uniaxial stress" J Geophys Res 90 B8 (1985)
makes the theoretical approach so simple that the pp 6840 6858
technique developed is readily implemented on a 15 Scruhy, C.B. 'Acoustic emission measurements using point-
contact transducers' J Acoust Emission 4 1 (1985) pp 9 18
microcomputer. 16 Ohtsu, M. 'Acoustic emission characteristics of concrete and
fundamental mechanisms' PhD Thesis (in Japanese), Kyoto
University (1982)
17 Rothman, R.L., Greenfield, R.J. and Hardy, Jr, H.R. 'Errors in
Acknowledgement hypocenter location due to velocity anisotropy' Bull Seism Soc
Am 64 6 (1974) pp 1993-1996
The authors are grateful for the financial support of this 18 Aki, K. and Riehards, P.G. Quantitative Seismology Theory and
study by the Office of Naval Research, Physics Program. Methods II W.H. Freeman and Company, San Francisco (1980)

NDT International June 1988 1 49

19 Ge+ M.C. and Hardy, Jr, tt.R. +Effect of velocity data quality A p p e n d i x . O u t p u t of a c o m p u t e r p r o g r a m
on source location and analysis of optimum transducer array
geometry' Prnc 4th Con./" on A E / M S in Geologic Structures and
~'* SOURCE LOCATION OBTAINED I,=
Materials, The Penn. State University (1985). Acoustic Emission.
IV, Trans Tech PubLishers OOOm÷O0
MINIMUM ERROR AT | 8 ERROR DISTANCE-
20 Ohlsu, M. and Ono, K. +The generalized theory and source
representations of acoustic emission' J Acou.st Emission 4 2/3 • 11 SOURCE DEPTH I1~ ZI- .05i

(1985) pp s50 s53 NEAREST(REFERENCE) TRANSDUCER X- 000 Y- 000

mAs TRANSDUCER LOCATION OBTAINED *tt
! [ XI- -.O~O YI- .0S0
I 2 Xl- .00O YI- .050
! 3 XI- .050 ¥1- .050
! 4 XI- -.O~O YI- ,OOO
| 6 XI- .0S0 YIm .000
! 7 Xl- -,0~0 YI- -.0~0
t S El= .00O Yt= -,OS0
I f Xl- .0~O YI- -.050
• ** FINAL SOLUTION in,

HORIZONTAL ANG- 45.000 DEG. VERTICAL ANC- ~0.000 DEG I

Authors
I! THEORY ii ~l OBSERVED ts (NORMALIZED RATIO)
M. Ohtsu is in the Department of Civil Engineering, UP- .2St UR- .23~
UPm .306 UR- .2~1
Kumamoto University, Kumamoto 860, Japan. K. Ono UP- ~Z0O UR- .204
UP- .~72 UR- .~81
is in the Department of Materials Science and Engineering, UP- .306 UR- .291
School of Engineering and Applied Science, University UP-
UP=
.447
.$72
UR-
UR-
.4~S
.SS|
of California, Los Angeles, CA 90024, USA. UP- .2~1 UR= .239

Paper received 25 August 1987

150 NDT International June 1988