NDT&E International 35 (2002) 399406

www.elsevier.com/locate/ndteint

A new non-contacting non-destructive testing method for defect detection

in concrete
K. Mori a, A. Spagnoli b,*, Y. Murakami c, G. Kondo c, I. Torigoe a
a
Department of Mechanical Engineering, Kumamoto University, 2-39-1 Kurokami, Kumamoto, 860-8555 Japan
b
Department of Civil Engineering, University of Parma, Parco Area delle Scienze 181/A, 43100 Parma, Italy
c
Department of Mechanical Science and Engineering, Kyushu University, Fukuoka, Higashi-ku, 812-81 Japan
Received 4 October 2001; revised 24 January 2002; accepted 3 February 2002

Abstract
A new non-destructive testing (NDT) method for defect detection in concrete structures is presented. The method is based on the dynamic
response of awed concrete structures subjected to impact loading. Conversely to similar NDT techniques, such as the impact-echo method,
the present method uses non-contacting devices for both impact generation (a shock tube producing shock waves) and response monitoring
(laser vibrometers measuring concrete surface velocity). Experimental and numerical (nite element) studies have been carried out for
concrete specimens containing articial defects (penny-shaped cracks parallel to the free surface) with varying length and depth. According
to the experimental and numerical results, it appears that the present method enables an effective detection of defects, particularly in the range
of shallow defects. q 2002 Elsevier Science Ltd. All rights reserved.
Keywords: Concrete; Defect detection; Impact-echo testing; Laser vibrometer; Non-destructive testing method; Shock wave

1. Introduction a characteristic `hollow' sound is heard, whereas a `ringing'

Numerous concrete structures such as bridge decks, slabs,
tunnels, etc. require periodic inspections through non-
destructive testing (NDT) methods to assess their structural
integrity. Typical geometries encountered in concrete struc-
tures vary largely from bars and hollow cylinders to plates
and shells, while their common defects include: internal
cracks, voids, shallow delamination, honeycombing, and
surface-opening cracks. There is a large number of well
known NDT methods for concrete structures, such as:
infrared thermography, ground-penetrating radar, acoustic
impact method (sounding), ultrasonic pulse-velocity
method, ultrasonic pulse-echo method, impact-echo
method. Except for the rst two methods, the other methods
are based on the mechanical principle of stress wave
propagation.
Among the stress wave propagation-based NDT methods,
the acoustic impact method is the oldest [1]. Free concrete
surface is stroke with a metal object such as a hammer or a
heavy chain and an expert operator listens to the sound that
is produced. The generated stress waves are very low in
frequency (less than 1 kHz). When a awed area is struck,
* Corresponding author. Fax: 139-0521-905-924.
E-mail address: spagnoli@unipr.it (A. Spagnoli).
0963-8695/02/$ - see front matter q 2002 Elsevier Science Ltd. All rights reserved.
PII: S 0963-869 5(02)00009-9
sound is produced by striking over a solid area. The method
is time-consuming and requires the subjective judgement of
the operator to interpret the results. Sounding becomes less
reliable as the depth of the defect increases.
The ultrasonic pulse-velocity method [2] introduces repe-
titive pulses (with frequency of 80 kHz and above) on the
free surface of concrete by a transmitter. A receiver is used
to detect the arrival of stress waves, so that their travel time
can be determined. Hence, knowing the length of the travel
path, the velocity of stress waves can be calculated. From
such a calculated value of velocity, the condition of the
material and/or the presence of defects can be evaluated.
In plate-like structures, the method is generally used for
through-thickness measurements by placing the receiver
on the back-wall surface.
Conversely to the above method, with the ultrasonic
pulse-echo method [2] the receiver is used to detect the
arrival of reected waves, or echoes, so that the travel
time (round-trip travel period) of these waves can be calcu-
lated. Wave reections are produced by internal defects,
interfaces between materials with different densities and
elastic moduli and boundaries of the solid (typically back-
wall surfaces). This method is used to determine the location
of defects or interfaces by knowing the velocity of stress
waves and measuring their travel time. In concrete, because
400 K. Mori et al. / NDT&E International 35 (2002) 399406

is described. Then, the results of an experimental campaign

Nomenclature along with a numerical simulation are presented and
a half-length of defect discussed.
CP normal wave (P-wave) velocity
CR Rayleigh wave (R-wave) velocity
2. Principle of the method
CS shear wave (S-wave) velocity
d depth of defect from free surface
The proposed method is based on the dynamic response
D thickness of plate; side length of cubic speci-
of awed concrete structures subjected to impact loading.
men
The impact loading generating small stress levels is applied
fd frequency of thickness vibration due to defect
on the free surface, and normal-to-surface displacement or
reection
velocity against time is monitored by transducers located on
fD frequency of thickness vibration due to back-
the free surface itself at different locations. Typically, a
wall reection
single transducer is positioned near the central point of
ffx rst natural frequency of exural vibration
impact loading, but congurations with multiple output
E Young's modulus
reading points can also be considered.
tc duration of impact
The dynamic response of awed concrete structures
Td period of thickness vibration due to defect
subjected to impact loading is governed by stress wave
reection
propagation. Two kinds of vibrations can be distincted:
TD period of thickness vibration due to back-wall
thickness vibration and in the presence of shallow defects,
reection
plate-like exural vibration. Both these vibrations deter-
Tfx rst natural period of exural vibration
mine characteristic displacement/velocity time histories on
n Poisson ratio
the free surface, which permit the detection of defects.
r density
2.1. Thickness vibration
of the high-frequency stress waves exploited, strong
attenuation and wave reections due to the heterogeneous Let us consider a semi-innite solid subjected to a surface
nature of concrete are produced. point impact. The pulse travels along the spherical wave-
According to the impact-echo method, a transient stress fronts through concrete as P (dilatation or normal) and S
pulse is introduced by mechanical point impact on the free (distortion or shear) body waves and R (Rayleigh) surface
surface [3,4]. Stress waves propagate into concrete along wave (Fig. 1) [7,8]. The dynamic problem is governed by
spherical wavefronts and are reected by defects interfaces the Theory of Elasticity, since, for the applied stress
between materials and boundaries. A broadband transducer levels, concrete can be treated as a linear-elastic, iso-
located on the free surface close to the impact point is used tropic and homogeneous medium. The velocity of
to monitor the normal-to-surface displacement against time, P-waves is given by:
caused by the arrival of the reected waves. A transient s
1 2 nE
resonance condition is set up by multiple reections of the CP 1
1 1 n1 2 2nr
waves between the free surface and defects, interfaces or
boundaries. S-waves and R-waves have lower velocities, equal to
From a theoretical viewpoint, the analysis of eld or about 0.61 and 0.56CP, respectively, for n 0:20:
laboratory data obtained by means of a NDT method Now let us consider an innite plate subjected to a point
requires the solution of an inverse problem. Generally impact (Fig. 2). When body stress waves reach back-wall
speaking, inverse problems in mechanics are those, which
concern the estimation of some input data (e.g. shape of
domain and boundaries, presence of unknown inner bound-
aries, distribution of material properties, etc.) from output
(response) [5]. In particular, the relevant inverse problem
for defect detection (for instance, the identication of an
internal defect from the dynamic response of a structure
subjected to an impact loading) is a domain-boundary one.
The aforementioned stress wave propagation-based NDT
methods require contacting devices for both mechanical
impact and response measurement. In this paper, a new
non-contacting NDT method using shock waves and laser
vibrometers [6] is presented. The principle of the method is Fig. 1. Wavefronts produced by point impact on a semi-innite solid
explained and the adopted instrumentation (testing technique) (schematic).
 K. Mori et al. / NDT&E International 35 (2002) 399406 401

Fig. 2. Wavefronts produced by point impact on an innite plate (sche-

matic).

surface, reection and mode conversion occur. 2P-wave is

the reected P-wave and PS-wave is the S-wave produced
by P-wave (mode conversion). P-wave reections are of
primary importance because in the vicinity of the impact
point, where usually transducers are located, the values of
normal-to-surface displacement/velocity produced by
P-waves are larger than those produced by S-waves. In
particular, a thickness vibration due to a stress P-wave
resonance, and characterised by a transient period (round-
trip travel period of P-wave from point C to A, and back-
wards), occurs. Note that the forward P-wave (from point C
to A) is a compression wave, while the backward 2P-wave
(from point A to C) is (for a traction-free back-wall surface)
a tension wave. Hence, when 2P-wave reaches the free
surface, such a surface in the vicinity of point C undergoes
inward normal-to-surface displacements.
The transient period TD (and frequency fD) of thickness
vibration is given by:
1 kD
TD
fD CP
The parameter k is equal to 2, if the back-wall surface is a
solid/air interface (traction-free surface), and k 4 for a
solid/rigid solid interface (a solid can be regarded as rigid
when its acoustic impedance, r CP, is much higher than
that of the solid subjected to the impact loading). Note that
in the case of a nite square plate P-wave velocity
approaches, with decreasing side length of the plate, that
of plane waves travelling along a bar, characterised by a
p p
CP equal to 0.95CP (for n 0:20), i.e. CP
velocity
p
E=r [7,8].
In the case of an innite plate with a defect parallel to the
free surface, reections of stress body waves occur at the
upper surface of defect (Fig. 3), so that a thickness vibration
is established. Such a vibration is characterised by the
following period Td (and frequency fd):
Fig. 3. Wavefronts produced by point impact on an innite plate containing
a defect parallel to surfaces (schematic).
2.2. Flexural vibration
Let us consider a penny-shaped crack parallel to the free
surface in an innite plate. In the case of a shallow defect
(low values of the depth d), the solid portion above the crack
might act as a plate-like structure undergoing exural vibra-
tions. In the specic case under consideration, such a
portion can be treated as a circular plate having diameter
2a and thickness d (Fig. 4). The analytical expression for
the rst natural period Tfx (and frequency ffx) of exural
vibration of a circular plate is as follows [9]:
s
1 2p 121 2 n2 ra4
Tfx 4
ffx K1 Ed2
2 For simple supported boundary conditions, the rst natural
frequency is obtained for K1 4:99: Such a value repre-
sents a lower bound for the actual natural frequency, since
the considered circular plate has some degree of rotational
constraints along its boundaries, due to the presence of the
remaining portion of the awed solid.
Typically, the velocity CP in concrete ranges from 3000 to
4000 m/s. For instance, assuming d 50100 mm; the
frequency fd of thickness vibration varies from 15 to
40 kHz. On the other hand, the value of the frequency ffx
for a defect of 200 mm in diameter and 50100 mm in depth

1 2d
Td 3
fd CP

Obviously, Eq. (3) is valid only if the acoustic impedance of

the defect is smaller than that of the solid, such as in the case Fig. 4. Flexural vibration of a plate-like portion of an innite plate contain-
of a traction-free surface of a crack-like defect. ing a shallow penny-shaped crack parallel to surfaces (schematic).
402 K. Mori et al. / NDT&E International 35 (2002) 399406
Fig. 5. Schematic view of the proposed NDT method.
ranges from 3.4 to 9.2 kHz. Generally speaking, the value of
the frequency ffx is lower than that of the frequency fd, and
the amplitude of surface displacement/velocity due to ex-
ural vibration is higher than that due to thickness vibration.
3. Testing technique
A schematic view of the proposed non-contacting NDT
method for concrete is shown in Fig. 5. Impact on the free
concrete surface, inducing thickness vibration and/or ex-
ural vibration, is produced by shock waves. Such waves are
generated by a shock tube containing high-pressure air [6].
Air is compressed into the shock tube vessel by an air
compressor. A metal lm holds the high-pressure air, so
that a needle driven by a solenoid fractures the lm to
produce shock waves. In addition, so-called expansion
waves are generated by the expanding compressed air in
the shock tube. Shock waves are characterised by a travel
velocity approximately equal to that of sound, while
expansion waves travel with a relatively lower velocity.
Thickness and exural vibrations are detected by laser
Doppler vibrometers, measuring time history of surface
velocity components. It should be noted that both devices
used for generating impact and measuring vibrations are
non-contacting ones. Length and location of defects can
be detected by analysing the output velocity data. In order
to allow a better identication of the involved periodic
phenomena, measured time histories of velocity can be
transformed to frequency domain via fast Fourier transform
(FFT) [10,11].
The depth of the defect can be obtained from Eq. (3), in
which the measured peak frequency of thickness vibration is
introduced and an estimated (from the Young's modulus E
and the density r ) or measured value of the velocity CP is
known. Once the depth d has been determined, the length 2a
of the defect can be calculated using Eq. (4) from the value
of exural vibration frequency. Length of defects can also
be estimated when a grid of measuring points of surface
velocity is considered. As a matter of fact, peak frequencies
related to thickness vibration or exural vibration can be
detected only when the measuring surface point is located
above the inside of the defect.
It should be emphasised that there are a number of
controlling parameters which determine the defect detection
limits of the present method. If the pressure acting over the
free concrete surface be treated for a pulse of duration tc
(typically tc 1580 ms), then the maximum frequency
content of impact energy is equal to about fc 1:5=tc
(typically fc 18100 kHz). Moreover, the duration tc is
correlated with the minimum detectable defect length and
depth, that is, usually, min{2a; d} CP tc =1:5: However,
short impact duration (e.g. having maximum frequency fc of
80 kHz or higher) has the disadvantage of inducing short-
length waves that are scattered by concrete inhomogene-
ities, and exhibit limited penetrating ability [4].
4. Experimental study
Experimental tests on cubic concrete specimens, with
side D of 430 mm, containing articial defects were carried
out. The adopted laboratory equipment is depicted in Fig. 6,
showing the concrete specimen and the shock tube. Three
laser Doppler vibrometers were placed to measure velocity
components at three points on the specimen surface, i.e.
specimen centre x y 0 mm; and at a distance from
the centre of 50 and 100 mm, respectively (x 50 and
100 mm, y 0 mm) (Fig. 7). A sampling frequency of
500 kHz and a sample number of 50,000 (the total sampling
time is equal to 100 ms) was adopted for velocity measure-
ment.
As a reference, a solid concrete specimen was tested
along with four specimens containing a central penny-
shaped crack parallel to the free surface (with nite width
equal to 10 mm), having the following geometric
parameters (Fig. 7): d 25 mm and 2a 100 and
200 mm (shallow defects), d 50 mm and 2a 200 mm;
d 100 mm and 2a 200 mm (deep defect).
The value of P-wave velocity CP in the concrete mix
under study was determined from the measured P-wave
travel time through the thickness D in the solid specimen.
Note that an alternative measurement method of P-wave
velocity can be performed by measuring R-wave velocity
[12,13], as for instance, when the back-wall surface of the
actual concrete structure is not accessible. A value of CP
 K. Mori et al. / NDT&E International 35 (2002) 399406 403

Fig. 6. Laboratory testing equipment: view of the specimen and shock tube.

equal to 3770 m/s was measured for the present concrete

mix. From the measured value of the concrete density r
2380 kg=m3 ; an estimated value of the Young's modulus E
is calculated to be equal to about 33,800 MPa E < CP2 r:
This value has been adopted in the following numerical
study.
Before performing NDT tests, the pressure on concrete
surface produced by shock tube was measured. By placing a
steel plate with embedded pressure sensors at a distance of
500 mm from the head of the shock tube, the time history
of the surface pressure was monitored (Fig. 8). Two tests of
pressure measurement were performed: one recorded the
total pressure phenomenon (time interval 010 ms,
Fig. 8(a)), while the other analysed the initial time instants
of impact with relatively higher sampling resolution (time
interval 01 ms, Fig. 8(b)). Pressure was measured at four
different locations: Ch1 (x y 0 mm; specimen centre),
Ch2 (x 50 mm; y 0 mm), Ch3 (x 100 mm;
y 0 mm) and Ch4 (x 150 mm; y 0 mm). Note that
Ch1 is also located at the central axis of the shock tube.
Surface pressure generated by shock tube acts for more
than 10 ms and it is characterised by a rather complex time
Fig. 7. Geometric characteristics of the tested specimens.
history. Expansion waves inuence signicantly the overall

Fig. 8. Time history of the pressure produced by shock tube on the specimen surface (x distance from specimen centre): (a) overall trend inuenced by
expansion waves; (b) initial trend generated by shock waves, with a close-up of initial time instants of impact for Ch1.
404 K. Mori et al. / NDT&E International 35 (2002) 399406

trend of the pressure time history (see, for instance, the

pressure peak of about 130 kPa at 2 ms for Ch1, Fig. 8(a)).
In the initial time instants of impact, shock waves, which are
characterised by a higher travel velocity in comparison to
that of the expansion waves, determine the initial pressure
peaks (see, for instance, the pressure peak of about 30 kPa
for Ch1, Fig. 8(b)). There is also a non-uniform distribution
of pressure over the specimen surface: pressure intensity
decreases as we move away from the centre of specimen
(Fig. 8). In addition, there is a time shift of the pressure
peaks, which appears to be a function of the distance from
the specimen centre (Fig. 8). Fig. 9. Normalised velocity spectrum at central surface point for the solid
Conversely to conventional contacting NDT methods specimen.
(e.g. impact-echo method), characterised by a point impact,
the proposed impact method using shock tube produces a four defect congurations were modelled. In FE models, the
pressure loading which acts over a certain portion of the actual nite width of defects (equal to 10 mm) was input.
concrete surface. Moreover, the time history of the induced
surface pressure is very different from, for instance, a simple
6. Results and comparisons
half sinusoidal wave characterising the time history of the
impact of a steel ball [4]. In order to strengthen the results
Results from experimental and numerical studies are
obtained with the present impact method, a point impact
herein presented in terms of normalised spectrum of
produced by a hammer was also used to determine exural
normal-to-surface velocity at the central surface point C
vibration frequencies in the awed specimens.
shown in Fig. 7 (also see Figs. 913). The presented
spectrums are plotted in the overall (080 kHz) and low
5. Numerical study (010 kHz) frequency ranges to highlight peak frequencies
due to thickness vibration and exural vibration, respec-
Three-dimensional nite element (FE) analyses were
tively. In Table 1, detected peak frequencies corresponding
performed to simulate the experimental tests. The commer-
to both thickness vibration and exural vibration are
cial package ANSYS 56 was used [14]. A transient dynamic reported according to experimental and FE results as well
analysis, based on the explicit central difference method,
as theoretical predictions (Eqs. (2)(4)).
was carried out. Eight-node isoparametric linear solid
For the solid specimen (Fig. 9), both experimental and FE
elements were employed. Concrete was treated as a linear-
models show a peak value of frequency at 3.9 kHz, which is
elastic, isotropic and homogeneous material. Young's
in good agreement with the theoretical value fD (Eq. (2) and
modulus was assumed to be equal to 33,800 MPa, while a
Table 1).
typical value of Poisson ratio for concrete n 0:15 and
As far as the awed specimens are concerned, it is seen
the measured value of density r 2380 kg=m3 were input
that there is a good agreement between theoretical and FE
in FE models. results in terms of the rst natural frequency of exural
The element size was kept smaller than about 0.33CP/fmax,
vibration (Table 1). The theoretical values of ffx should be
where fmax is the maximum frequency of interest
a lower bound for FE results, although this is not always the
(fmax CP =2dmin ; with dmin 25 mm for the tested
case because of the relatively course resolution of FE spec-
specimens). The total analysis time T is equal to 1024 or
trum, equal to about 0.9 kHz. Both proposed (shock tube)
2048 ms, in order to have a fairly ne spectrum resolution Df
Df 1=T: To avoid numerical instabilities, the time step
was assumed to be equal to 0.9hmin/CP, with hmin being the
minimum element size.
For symmetry conditions about xz and yz planes (Fig. 7), a
quarter model was considered. Minimum constraints were
applied to the quarter model to prevent rigid-body motion. A
uniform pressure, with a time history corresponding to a
triangular pulse of 35 kPa peak and 28 ms duration, was
applied to the entire face z 0 of the model. This is a strong
simplication of the actual pressure time history generated
by shock waves (Fig. 8), but it is deemed not to affect
signicantly the following comparison between experimen- Fig. 10. Normalised velocity spectrum at central surface point for specimen
tal and FE results in terms of dominant frequencies. with 2a 100 mm and d 25 mm : (a) thickness vibration; (b) exural
The solid specimen and the awed specimens with the vibration.
 K. Mori et al. / NDT&E International 35 (2002) 399406 405

Fig. 11. Normalised velocity spectrum at central surface point for specimen Fig. 13. Normalised velocity spectrum at central surface point for specimen
with 2a 200 mm and d 25 mm : (a) thickness vibration; (b) exural with 2a 200 mm and d 100 mm : (a) thickness vibration; (b) exural
vibration. vibration.

and conventional (hammer) impact devices show a perfect Table 1

agreement in the exural vibration frequency (Table 1). As First natural frequency (kHz) of thickness vibration (fD, fd) and exural
vibration (ffx) ((A) shock tube; (B) hammer)
is shown in Figs. 10(b)13(b) and Table 1, there are discre-
pancies between experimental and numerical values of 2a d (mm) Thickness vibration Flexural vibration
exural vibration frequency. Such discrepancies might
partly be related to differences in the actual length of the Theory FE a Exp. Theory FE a Exp.
defect in laboratory specimens with respect to the nominal (A) (A) (B)
one input in FE models (e.g. a defect having an actual length
10% higher than the nominal one, produces a reduction in Solid 4.4 3.9 3.9 N/A N/A N/A N/A
100 25 75.4 N/A N/A 8.7 8.8 2.2 2.8
the value of exural vibration frequency of about 20%).
200 25 75.4 N/A 73.2 2.2 2.9 1.3 1.4
Note that the large frequency peak at about 4.0 kHz 200 50 37.7 36.3 37.1 4.4 3.9 2.2 2.3
shown in Fig. 13 is related to the thickness vibration of 200 100 18.9 19.1 26.4 8.7 9.0 3.0 3.4
the specimen as a whole. a
Spectrum resolution <0.9 kHz.
Now, considering the peak frequencies due to thickness
vibration (Figs. 10(a)13(a)), the comparison between
experimental and FE results appears to be more difcult waves. Then the velocity of the beam changes and the
than for exural vibration. This might be due to the low frequency measured by the laser uctuates. The best corre-
frequency content of impact energy (as is illustrated in lation among the cases analysed is for 2a 200 mm and
Section 3, the relatively long duration of shock wave impact d 50 mm (Fig. 12(a)).
generates a low frequency content of impact energy) which When frequencies of thickness vibration are measured,
in turn encounters difculties in detecting the high frequen- two aspects contribute to produce interference: both stress
cies of thickness vibration. When frequency peaks are waves in concrete and expansion waves in air are charac-
detectable, a general discrepancy possibly caused by inter- terised by high frequencies; thickness vibration attenuates
ference phenomena between laser beam and expansion rapidly in concrete, so that signicant values of surface
waves occurs. As a matter of fact, the laser beam measuring velocity are measured in the initial time instants of impact,
surface velocity travels through the air subjected to pressure while expansion waves are acting [6]. On the other hand, in
the case of exural vibration, peak frequencies are lower and
vibration continues further after expansion waves are
extinguished, so that the above interference phenomenon
is negligible [6].

7. Conclusions

A new NDT method for defect detection in concrete

Fig. 12. Normalised velocity spectrum at central surface point for specimen
with 2a 200 mm and d 50 mm : (a) thickness vibration; (b) exural
vibration.
structures is presented. The method is based on the dynamic
response of awed concrete structures subjected to impact
loading. A periodic response due to thickness vibration and,
in the presence of shallow defects, plate-like exural vibra-
tion can be detected in the time history of the measured
surface velocity. The method uses non-contacting devices
406 K. Mori et al. / NDT&E International 35 (2002) 399406
for both impact generation (a shock tube producing shock
waves) and response monitoring (laser vibrometers measur-
ing surface velocity).
Experimental and numerical (nite element) studies have
been carried out to assess the effectiveness of the proposed
NDT method. A solid specimen and four awed specimens
containing articial defects (penny-shaped cracks parallel to
the free surface) with varying length and depth have been
analysed. When shallow defects are considered, a fairly
good agreement between experimental and numerical
results, as well as theoretical predictions in terms of natural
frequencies of exural vibration is obtained. In addition, it is
seen that traditional impact devices (e.g. hammer) yield
similar values of natural frequencies. On the other hand,
some difculties, related to the low frequency content of
impact energy, for both experimental and numerical studies
in detecting the peak frequencies of thickness vibration
occur.
Acknowledgements
The present study has been carried out within the frame-
work of a project on dynamic NDT analysis of defects in the
concrete structures of Shinkansen (Japanese high-speed
train) tunnels. The nancial support from the Japanese
Ministry of Transport under Grant No. 99-11 is gratefully
acknowledged.
References
[1] ASTM D 4580-86. Standard practice for measuring delaminations in
concrete bridge decks by sounding, Annual book of ASTM standards,
vol. 04.03. Philadelphia: ASTM, 1987. p. 74951.
[2] Bray DE, Stanley RK. Nondestructive evaluation: a tool in design,
manufacturing, and service. Boca Raton: CRC Press, 1996.
[3] Sansalone M, Carino NJ. Detecting delaminations in concrete slabs
with and without overlays using the impact-echo method. ACI Mater J
1989;86(2):17584.
[4] Sansalone M. Impact-echo: the complete story. ACI Struct J
1997;94(6):77786.
[5] Kubo S. Inverse problems in fracture mechanics. In: Nisitani H,
editor. Computational and experimental fracture mechanic. South-
ampton: Computational Mechanics Publications, 1994.
[6] Mori K, Spagnoli A, Torigoe I, Murakami Y, Itoh S. A new method
for non-destructive evaluation of concrete structures using a shock
wave and laser vibrometers. Proceedings of the Fourth International
Symposium on Impact Engineering (ISIE/4). Kumamoto, Japan,
2001.
[7] Kolsky H. Stress waves in solids. New York: Dover Publications,
1963.
[8] Graff K. Wave motion in elastic solids. New York: Dover Publica-
tions, 1991.
[9] Meirovitch L. Elements of vibration analysis. New York: McGraw-
Hill, 1986.
[10] Abraham O, Leonard C, Cote P, Piwakowski B. Time frequency
analysis of impact-echo signals: numerical modeling and experimen-
tal validation. ACI Mater J 2000;97(6):64557.
[11] Maji AK, Paul S, Wang ML. Improved impact-echo technique by
signal processing. Res Nondestruct Eval 1990;2(1):4556.
[12] Lin JM, Sansalone M. A procedure for determining P-wave speed in
concrete for use in impact-echo testing using a Rayleigh wave speed
measurement technique. Innovations in nondestructive testing, SP-
168, Farmington Hills, MI: American Concrete Institute, 1997.
[13] Popovics JS, Song W, Achenbach JD, Lee JH, Andre RF. One-sided
stress wave velocity measurement in concrete. J Engng Mech, ASCE
1998;124(12):134653.
[14] ANSYS/LS-DYNA 56. Theory and user's manual. Canonsburg, PA:
Ansys Inc, 1999.