Materials and Structures, 1995, 28, 74-82

Determining the minimum crack width that can be detected

using the impact-echo method
Part 1: Experimental study
CHIA-CHI CHENG, MARY SANSALONE
School of Civil and Environmental Engineering, Cornell University, Ithaca, N Y 14853, USA

To determine the minimum crack width that can be detected usin9 the impact-echo method,
two types of laboratory experiment on concrete plate specimens were conducted. In each
experiment, a crack was propagated through the plate parallel to the plate surfaces, while
surface displacement caused by crack openin9 was measured and impact-echo tests were
performed along the crack trajectory. The relationship between the surface displacement and
the real crack openin 9 displacement was established usin9 the results of nonlinear fracture
analyses (reported elsewhere). The minimum crack width was determined by estimatin9 the
crack openin9 displacements for each impact-echo test location.

1. INTRODUCTION through the formation of a crack which could no longer

transmit stress across its faces.
Since the impact-echo method was developed for testing
concrete structures in the mid 1980s, one of the most
frequently asked questions is how open does a crack have
to be (i.e., how much crack opening displacement is 2. BACKGROUND
necessary) for it to be detected using the impact-echo
2.1 Impact-echo method
method? Finding the answer to this question became the
focus of a study in which carefully designed numerical In the impact-echo method, a transient stress pulse is
and experimental fracture and wave propagation studies introduced into a structure by mechanical impact at a
were carried out. The results of this study and the answer point on the surface. This pulse travels into the plate as
to the question are presented in this paper and in a dilatational (P) and distortional (S) waves and along the
subsequent paper [1]. surface as a Rayleigh (R) wave. The P and S waves
Two types of laboratory experiment on concrete plate propagate into the structure along spherical wavefronts
specimens were conducted. In the first, a crack was and are reflected by internal cracks, or voids or interfaces
initiated from an edge notch and propagated through a (such as reinforcing bars) and by the external boundaries
plate using mechanical jacks. In the second, cracks were of the structure. A displacement transducer located close
produced by filling small diameter plastic pipes placed to the impact point is used to monitor the surface
in concrete specimens with expansive cement mortar. displacements caused by the arrival of these reflected
This arrangement simulated cracking caused by cor- waves. These waves are, in turn, reflected at the free
rosion of reinforcing bars. During each experiment, a surface, and they propagate back into the test object to
crack was propagated through the plate parallel to the be reflected again by internal interfaces or boundaries.
plate surfaces, while both crack opening displacement Therefore, a transient resonance condition is set up by
was measured and impact-echo tests were performed multiple reflections of the waves between the free surface
along the crack trajectory. Displacements at the top and and internal defects or external boundaries. The P waves
bottom plate surfaces were measured using direct current are of primary importance in the impact-echo testing of
differential transformers (DCDTs). plate structures, because at points located close to the
These experimental studies are discussed in this paper. impact point the displacements caused by P waves are
In a subsequent paper [1], the relationship between the much larger than those caused by S waves [2]. The
surface displacement and the real crack opening dis- frequency of P wave arrivals at the transducer is
placement (COD) is established using the results of determined by transforming the recorded time-domain
nonlinear fracture analyses. The minimum crack width signal into the frequency domain using the fast Fourier
which could be detected using the impact-echo method transform technique. For plate structures, the fre-
is then determined by estimating the crack opening quency associated with the highest peak in the resulting
displacements for each impact-echo test location from amplitude spectrum represents the dominant frequency
the beginning of the experiment (uncracked condition) in the waveform [2]. Knowing the P wave speed in the
through the initial detection of a crack and eventually test object Cp, the depth d to an internal void or
0025-5432/95 9 RILEM
Materials a n d Structures 75

crack or an external boundary can be calculated width-to-thickness ratios large enough to avoid inter-
as [2] ference caused by wave reflections from the side
boundaries of the plate. To simplify both COD and
d - Cp (1)
2f impact-echo measurements and interpretation of results,
the objective was to propagate a crack through the plate
where f is the frequency of P wave reflections from the along a trajectory that was parallel to the plate surfaces,
internal interface or bottom plate surface. at least over a distance that permitted crack-opening
measurements to be made. Considering these criteria, two
2.2 Instrumentation types of specimen were designed.
The impact-echo test system used in current work is
composed of three components: impactors; a receiving 2.3.1 Mechanical splitting specimen
transducer; and a portable computer with a data
acquisition card. The specimens used in this set of studies were 1.5 m (5 It)
The force-time history of an elastic impact can be wide, 1.6 m (5 ft 3 in) long and 0.31 m (1 It) thick. The
approximated as a half-cycle sine curve [2]. The duration
(or contact time to) of the impact determines the frequency
content of the stress pulse that is generated [2]. Most of
the energy in the pulse can be considered to be contained
in frequencies less than about t.5/t c. A shorter duration
impact produces a broader range of frequencies in the
stress pulse; however, the amplitude of each component
frequency is lower. The impact duration determines the
lateral dimensions of the defect which can be detected
using the impact-echo method. Shorter duration impacts
produce pulses containing higher frequency (shorter
wavelength) components, and smaller defects or inter-
faces can be detected. The impact duration should be (a) _
chosen so that the pulse that is generated contains
components with wavelengths approximately equal to or
less than the lateral dimensions of the defect or interface
to be detected and twice the depth (2d) of the defect. In ~ " ' 0 31 m ~
current impact-echo research and field work, hardened
steel spheres on spring-steel rods are used as impact
sources. The durations of the impacts produced by these Section A-A
sources range from about 10 gs to 80 ps. F coil of DCDT
The receiver is a broadband displacement transducer core of DCDT
r Fixed
consisting of a small, conically shaped, piezoelectric
element cemented to a brass cylinder [3]. The output of 16 gauge wire
this transducer is proportional to normal surface dis- crack plane
placement. A thin sheet of lead is used between the conical (b)
element and the concrete surface to complete the
transducer circuit and to couple the transducer to the /--- Fixed
concrete surface. In the test system used by the authors,
the impactors and transducer are housed in a hand-held
unit which can be used to test horizontal, vertical, or
overhead surfaces [4]. The distance between the impactor
and receiver is 30 mm.
A portable computer-based data acquisition system artificial slot
is used to capture the output of the transducer, store the
digitized waveforms, and perform signal processing and notch , , ~ J
analyses. For all of the experimental results presented in
this paper, displacement waveforms contain 1024 points bo,t and nuts - / - " t ~-
recorded at a sampling interval of 2 gs. The resulting 3 rnm steel strip
(c)
digital amplitude spectra have a resolution of 0.488 kHz.
Fig. 1 The specimen used in the mechanical splitting
2.3 Laboratory specimens experiment: (a) the geometry of the specimens and the points
for displacement and impact-echo measurements;
The specimens for both experimental and numerical (b) positioning of a typical DCDT; and (c) detail of the
studies were chosen to be plain concrete plates, with mechanical jack configuration.
76 Cheng and Sansalone

/
//•I 4 space @ 0.05 5 space @ 0.05 3 space @ 0.05 ~ "

1.2. f
// i
? ," t / ~/.!'11 /z" ,,,TX'I I I , ,,~
/ / .k'.~,_'~,-,:z~JJ..,_'L~2",.,'~g, kZ ' ' -:.4~,,.'- I"] I ," t . " s ; , Z 1
/ / r/.~, ~'s s " s ~- /.i s % ~i r ~'J /j , //I .,(-y/ /
/ /
I
.,;:;.~%', ,
,,/'.~,,.//-/
., ,~'0."
~, / I /i ~"
v
./ ~
, ~,,.,
,~J /
,,
/
,7 -:,"

I o,o,e ,

I~'~ .. 1 . 4 \
t
L-----
0.022 dlarrmeter ~ll~stic pipes
I indicates the position of DCDT
All dimensions are in meters
Fig. 2 Typical example of the layout of the pipes in the expansive cement experiment.

maximum coarse aggregate size was 11 mm (9/20 in). The (1/8 in) diameter hole, which had been cast into the
geometry of the specimen and the points for displacement specimen, and the rod was fixed to the opposite surface.
and impact-echo measurements are shown in Fig. la. A Thus, when a crack propagated across the location of
0.03 m x 0.04 m notch was cast into one edge of the the DCDT, the relative displacement of the plate surfaces
specimen, and an artificial slot located at the centre of produced by the crack opening could be measured. All
the notch was cast into the concrete using a thin, rigid the DCDTs were connected to a data acquisition system
plastic sheet, to serve as a point for crack initiation. The which read and recorded measurements on demand and
crack opening force was produced by using mechanical converted the voltage change produced by the DCDTs
jacks made of 12.7 mm diameter bolts and nuts which to displacement.
were placed in the notch at 0.15 m intervals. Turning
the nuts a specified amount served to open the notch
2.3.2 Expansive cement specimen
and propagate a crack from the tip of the artificial
slot (see Fig. lb). To avoid a local failure caused A second way to initiate and propagate cracks in a
by the concentrated compression forces produced by the concrete specimen was used to verify independently the
mechanical jacks, a 25.4 mm wide (t in), 3.18 mm thick minimum crack width measured from the mechanical
(1/8 in) steel strip was inserted between each end of the splitting experiment. Several concrete plate specimens
jack and the wall of the notch. with thicknesses 0.2-0.31 m and lateral dimensions five
The cross-sectional geometry of the specimen was times the thickness were cast. In each specimen, several
similar to the geometry of the double-cantilevered beam 22 mm diameter plastic pipes were split in half in advance
specimens used in typical fracture experiments. It was and bound by masking tape before being placed in the
expected that after propagating parallel to the plate form for each specimen prior to casting the concrete. The
surfaces for a distance, the crack path would deviate pipe layout was varied until it produced cracks parallel
from the line of symmetry in this kind of specimen [6]. to the top and bottom plate surfaces. A typical example
In an effort to propagate a crack parallel to the surfaces of the layout of these pipes is shown in Fig. 2, where four
of the slab, the slab was positioned vertically as shown groups of three pipes were placed horizontally at
in Fig. la. This position caused the direction of the mid-depth in the concrete plate. After about one month
self-weight of the plate to be perpendicular to the crack curing time, the pipes were filled with an expansive
propagation direction, as well as to the direction of the cement called Bristar 150. After water is added, this
crack opening force. The effect of the self-weight was cement expands to a volume several times its original
symmetric with respect to the crack. The supports on volume over a period of a day. Thus bursting forces
the bottom of the plate were placed away from the edge similar to those produced by corroding steel over a period
where crack initiation and propagation were to occur, of many years are produced by the expansive cement in
to eliminate any restraint at the bottom surface of a matter of hours. Cracks initiated after a few hours,
the slab. when the bursting force caused by the expanding mortar
As shown in Fig. la, a series of DCDTs was placed caused stresses exceeding the tensile strength of the
along the anticipated crack trajectory to measure the concrete. Cracks propagated horizontally in the plane of
surface displacements caused by the propagating crack. the pipes and eventually a crack plane formed across
Fig. lc shows the positioning of a typical DCDT. The the entire slab. Three or four D C D T s were placed in the
coil of the D C D T was fixed on one plate surface, while space between each group of pipes before the pipes were
the core was attached tO a rod passing through a 3.18-mm filled with the expansive mortar, so that the crack opening
Materials a n d Structures 77

displacement could be measured. Impact-echo tests notch artificial slot

were performed on the surface at the same locations
where the displacement measurements were recorded.
Readings were taken every 20-30 min before cracking ~. P Q R S T U V W X -,,i---Impact
occurred and every 10 min thereafter. I I I I I I I I pos,ions
~.. O O O O O -4--- DCDTs'
1 2 3 4 5 positions,

3. EXPERIMENTAL STUDIES: M E C H A N I C A L i=l= i J l L J l l l l l l l = l l t JL

~.. 5 10 15 20 25
SPLITTING E X P E R I M E N T distance from tip of slot (cm)
3.1 Procedure 'V
The experimental procedure is shown in the flow chart Fig. 4 Elevation view of the positions of the DCDTs and the
presented in Fig. 3. To begin a test and for each impact-echo test points.
subsequent cycle of testing, the mechanical jacks in the
notch were turned a specified angle to produce a uniform
opening of the notch along its length. After the jacks were
extended, the relative surface displacements measured by 3.2 Results
the DCDTs were read and recorded and a series of
impact-echo tests were performed. As shown in Fig. 1,
3.2.1 Surface displacement
the locations of the impact-echo test points were very The surface displacement profiles measured by the
dose to the locations of DCDTs in order to make DCDTs for each notch-opening step beginning with step
the impact-echo response representative of the crack 10 and proceeding through to step 22 are shown in
opening displacements recorded by the DCDTs. The Fig. 5. The position of each DCDT shown in Fig. 5
positions of the DCDTs and the impact-echo test points corresponds to the perpendicular (horizontal) distance
are shown in Fig. 4, which is an elevation view of the from the tip of the artificial slot (see Fig. 4). Although
slab. The impact positions are denoted by letters, and five DCDTs were used, displacements were measured at
the positions of the DCDTs are denoted by circles. After only the first three, because the crack did not reach the
the impact-echo tests were completed, the mechanical fourth and the fifth sensors. The crack pattern along the
jacks were extended another specified amount, and upper side of the specimen is shown in Fig. 6; it confirms
the testing cycle was repeated. The experiment was that the crack propagated parallel to the plate surfaces
terminated when an open crack response (stress waves for about 0.12 m and then split. Each crack branch
were no longer transmitted across the crack) was propagated at an angle of about 40 ~ towards each slab
obtained from the impact-echo results at all test surface. Crack branching was expected in a double-
points. cantilevered beam-like specimen, and an explanation for
the observed crack pattern is given later. As shown in
Fig. 5, the displacement data obtained from the first three
sensers for each reading were connected by straight lines.
These lines represent the approximate surface profiles
caused by the opening of the crack.

a specified displacementI
v~l NotChmechanicalOpened
I by jacks.
0.7
I step 22 "N \

0.6 step 21
read DCDTs obtain impact-echo /
and calculate response along plate/ "" 0.5 step 2019- N " x - ~ ~
relative surface surface / ,~ step ensor 2
displacement \\\'<
data
I I ~o 0.4 step 18
step 17
,~ 0.3 step 16
~" step 15
yes
0.2 step 14
step 13 ~sensor 3
o.1 step 12
step 11
0.o , step 10
, i , , ,

0.05 0.1 0.15

Position of DCDT (m)
Fig. 3 Flow chart of experimental procedure of the Fig, 5 The surface displacement profiles measured by the
mechanical splitting experiment. DCDTs for notch opening steps 10-22.
78 Cheng and Sansalone

~ 9 591~ P9

.2.=_
<E

5 10 15 20 25 30
Frequency (kHz)
(a)

4.4 [54 112 P10

Fig. 6 Photograph of the crack pattern along the upper side

of the mechanical splitting specimen.

P0 5.9 Q0 0 5 I0 15 20 25 30
Frequency (kHz)
.g (b)

<E <

5 10 15 20 25 30 5 ~0 15 20 25 30
}<
Frequency (kHz) Frequency (kHz)
(a) (b)

59 SO 0 5 10 15 20 25 30
Frequency (kHz)
(c)
.~_ Fig. 8 Impact-echo spectra obtained at test point P for
< < notch opening steps: (a) P9, (b) PI0, and (c) P11.

5 10 15 20 25 30 0 5 I0 15 20 25 30 the notch opening increment (denoted by a number). For

Frequency (kHz) Frequency (kHz) example, P10 means the test is at position P and is for
(c) (d) the 10th increment of notch opening. Thus, the number
5.9 U0 in each of the impact-echo test result names is consistent
with the number shown next to the experimental crack
opening profiles in Fig. 5.
The impact-echo spectra for test points P - U prior to
the start of the experiment are shown in Fig. 7a-f,
<
respectively. The responses at positions V, W, and X are
not shown but are very similar to the response shown
for position U. For the uncracked condition, a large
; 1; 1; 20 2'5 30 o 5 ~o 15 20 25 30
Frequency (kHz) Frequency (kHz) amplitude peak appears at a frequency of 5.9 kHz.
(e) (f) This frequency is produced by multiple reflections
Fig. 7 Impact-echo spectra obtained at test points P-U of the P wave between the plate surfaces9 The P wave
prior to beginning the mechanical splitting experiment: speed in the specimen was calculated using this 'thick-
(a) P0, (b) Q0, (c) R0, (d) SO, (e) TO, and (f) U0. ness frequency' as Cp = 2Tf = 2 (0.31 m) (5900 H z ) =
3660 m s- 1. Some of the secondary peaks in the spectra
are produced by reflections from within the specimen.
39 Impact=echo responses
For example, in Fig. 7a, there is a secondary peak at a
After each notch opening step, a series of impact-echo frequency of 11.2 kHz which is produced by the P wave
tests was performed at each of the positions shown in reflected from the edge of the artificial slot.
Fig. 4. The name given to each of the impact-echo Notch-opening steps 1-9 produced no change in the
responses indicates the position (denoted by a letter) and impact-echo responses at the test points. After the 10th
Materials and Structures 79

5.4 Q 11 Rll

< < <

5 10 15 20 25 30 5 I0 15 20 25 30 5 I0 15 20 25 30
Frequency(kHz) Frequency(kHz) Frequency(kHz)
(a) (b) (a)

L9 ~ 11.2 Q12
JI 113.7
.-=
< < <

5 10 15 20 25 30 ; ,0"~; 20 25 30 0 5 10 15 20 25 30
Frequency(kHz) Frequency(kHz) Frequency(kHz)
(e) (d) fb)

;/l.
13.7 Q16 ] 137 Q19

e~
< <E <

5 10 15 20 25 30 5 10 15 20 25 30 0 5 10 15 20 25 30
Frequency(kHz) Frequency(kHz) Frequency(kHz)
(e) (0 (c)
Fig. 9 Impact-echo spectra obtained at test point Q for Fig. 10 Impact-echo spectra obtained at test point R for
notch opening steps: (a) Q10, (b) Qll, (c) Q12, (d) Q13, notch opening steps: (a) R11, (b) R12, and (c) R13.
(e) Q16, and (f) Q19.

step, the impact-echo responses began to change. As the For position Q, the responses for tests Q10, Q11, Q12,
notch was opened, a peak at a frequency corresponding Q13, Q16, and Q19 are shown in Fig. 9a-f, respectively.
to the depth of the crack appeared and then began to Again as in P9, the pattern in the spectrum obtained
increase in amplitude. Figs 8-14 show the impact-echo from test QI0 is very similar to the solid response at
responses for the loading steps just before and after the point Q (Fig. 7b). With increasing notch opening (steps
appearance of the peak corresponding to the depth of 11-13), the multiple peaks which exist around 14 kHz in
the crack for points P-T, respectively. the spectra became more and more dominant and the
Fig. 8a-c shows impact-echo spectra for tests P9, P/0, thickness frequency shifts down to 4.9 kHz. In Q13, Q16,
and P l l , respectively. The response shown for P9 is and Q19, the peak at 13.7 kHz becomes distinct and
similar to the solid response shown in Fig. 7a. For P10, dominant. Thus, at point Q, the calculated depth of the
the thickness frequency has shifted one digital point cracking plane is about 0.134 m. This is based on using
lower, to 5.4 kHz, and lower frequency peaks at 2.4 kHz 13.7 kHz in Equation (1). Similar results are shown in
and 4.4 kHz correspond to the flexurat vibrations of the Fig. 10a-c for Rll, R12, and R13, respectively. The
concrete section above the notch and slot 1-5]. The peak responses for test point S (not shown) are also similar
at 11.2 kHz, which originally corresponded to the to test points P, Q, and R with a peak at l l . 7 k H z
distance between point P and the top of the slot, becomes corresponding to the depth of the crack beginning at
dominant, because the crack has extended under point notch opening step 13. Thus, summarizing the experi-
P and the crack has opened enough to produce significant mental results, the first indication of crack response in
reflection of waves. The depth of the crack at point P the impact-echo spectra happened at notch opening steps
for the measured frequency of 11.2 kHz is about 0.16 m 10, 11, 12 and 13 for test positions P, Q, R, S, respectively.
(6.37 in), which is very close to the middle of the slab. In The minimum crack width when stress can no
P11, the peak corresponding to the depth of the crack longer be transmitted through the crack faces was also
becomes even more distinct. determined from the impact-echo responses. As the
80 Cheng and Sansalone

.= 2
e~
< <

5 10 |5 20 25 30 5 I0 15 20 25 30 5 I0 15 20 25 30
Frequency(kHz) Frequency (kHz) Frequency (kHz)
(a) (a) (b)

4'4A 7.3 . T14

"O
B
e~ e~
< < <E

5 I0 15 20 25 30 0 5 10 15 20 25 30 0 5 10 15 20 25 30
Frequency (kHz) Frequency(kHz) Frequency(kHz)
(b) (c) (d)

44A7.3 T15 Fig. 12 Impact-echo spectra obtained at positions T, U, V,

and W after the final (22nd) notch opening step: (a) T22,
(b) U22, (c) 722, and (d) W22.

< complicated and the thickness frequency at 5.9 kHz is

absent. In contrast, the spectra obtained at positions U,
V, and W shown in Fig. 12b-d, exhibit responses that
0 5 10 15 20 25 30 are representative of a solid response. There is a dominant
Frequency(kHz) peak at a frequency very close to the thickness frequency.
(c)
Thus, the crack did not propagate under these positions.
Fig. 11 Impact-echo spectra obtained at test point T for
notch opening steps: (a) T13, (b) T14, and (c) T15.
4. EXPERIMENTAL STUDIES: EXPANSIVE
C E M E N T EXPERIMENT
crack opening displacements become larger than a critical For the expansive cement experiments, the results
value, the measured impact-echo responses should obtained from the impact-echo responses and the
become constant and similar, because all the energy in corresponding crack opening readings for different
the stress waves is being reflected by the upper crack DCDT positions within a specimen are very similar to
face. In the experiment, it was found that the relative those obtained from the mechanical splitting experiment.
amplitude and the frequency of the peak corresponding Thus, only a series of representative spectra and the crack
to reflection from the crack face remained unchanged opening displacement readings for one position within
after notch opening steps 13, 14, and 17 for test positions two specimens are shown. The two concrete specimens
Q, R, and S, respectively. The spectra shown in Fig. 9e, f are 0.2 m and 0.31 m thick, respectively.
are the impact-echo responses at Q16 and Q19. Note For the 0.2 m thick slab, Fig. 13a shows the impact-
that these spectra are very similar to that shown in Q 13 echo spectrum obtained before cracking initiated in the
(Fig. 9d). All of these spectra have a dominant peak at specimen shown in Fig. 2, at the position near the DCDT
13.7 kHz which corresponds to the depth of the crack. marked A. The dominant response is the thickness peak
Finally, the response obtained at test points T13, T14, at a frequency of 8.8kHz. The P wave speed is
and T15 are shown in Fig. l l a - c , respectively. It is at 3630 m s- 1. The spectra obtained for the measured crack
this location that the crack branches and begins to bend opening displacements of 0.025 mm and 0.031 ram, at the
towards each surface. The spectra become more com- same position, are shown in Fig. 13b, c, respectively. A
plicated and multiple peaks appear. The branching of the comparison of these two spectra with the spectrum in
crack at this point was also evident on the surface (see Fig. 13a shows the following. First, the thickness
Fig. 6). frequency is no longer dominant in Fig. 13b. Second,
Fig. 12 shows the impact-echo responses at positions a lower frequency peak at 5.9 kHz corresponding
T, U, V, and W after the final (22nd) notch opening step. to the flexural vibration of the section of concrete
At position T, shown in Fig. 12a, the response is very above the crack is observed. Third, a peak at 17.1 kHz
Materials and Structures 81

8.8 Solid r e s p o n s e 5.4 COD=O.02 [ mm

-g

<
<

10 20 30
Frequency (kHz)
(a)
10 20 30
Frequency (kHz)
5 9 COD = 0.025 mm
(a)
. 17.1
.g
49 C O D = 0 037 mm

<

10 20 30
H
Frequency (kHz) <
(b)

I0 20 30

.5 Frequency (kHz)
(b)
<
Fig. 14 Impact-echo spectra obtained from the 0.3 m thick
specimen for the expansive cement experiment: (a) for
0 10 20 30 measured crack opening of 0.021 mm, and (b) crack opening
Frequency (kHz) of 0.037 ram.
(c)

Fig. 13 Impact-echo spectra obtained from the 0.2 m thick

specimen for the expansive cement experiment: (a) solid
response, (b) for measured crack opening of 0.025 ram, and
(c) crack opening of 0.031 mm.

corresponding to the depth of the crack becomes

increasingly dominant as crack opening displacement
increases.
For the 0.3 m thick specimen, the spectra obtained
when the D C D T readings were 0.021 m m and 0.037 m m
are shown in Fig. 14a, b, respectively. In the spectrum in
Fig. 14a, only a single, large amplitude dominant peak
at a frequency of 5 . 4 k H z exists. This frequency cor-
responds to the solid plate thickness. In Fig. t4b the
thickness frequency has shifted to a slightly lower Fig. 15 Test location and the crack location on the side
frequency and a peak at a frequency of about 17.1 kHz surface of a 0.3 m thick specimen for the expansive cement
is apparent. This frequency corresponds to a depth of experiment.
about 0.115 m. This is the observed depth of the crack
at this test location as seen from the side of the specimen,
of how open does a crack have to be in concrete for it
as shown in Fig. 15. The test location is indicated on the
to be detected using the i m p a c t - e c h o method.
photograph by an arrow.
In the mechanical splitting experiment, a crack was
produced by jacks located in a notch on the side surface
of a concrete plate. The crack propagated par~tllel to the
5. SUMMARY plate surfaces for about 0.12 m before splitting into
This paper has presented the results of two types of two branches. I m p a c t - e c h o responses at the four test
experiment which were designed to answer the question positions located directly above the parallel crack were
82 Cheng and Sansalone

used to identify the existence of a crack. Surface profiles continued support of the National Science Foundation
corresponding to each notch opening step were measured for impact-echo studies is gratefully acknowledged.
using DCDTs.
In the expansive cement experiments, cracks parallel
REFERENCES
to a plate surface were produced in a matter of hours by
filling small diameter tubes located inside the concrete 1. Cheng, C. and Sansalone, M., 'Determining the minimum
specimens with expansive cement. Surface displacements crack width that can be detected using the impact-echo
were measured using DCDTs, and impact-echo tests method. Part II: Numerical fracture analyses', Mater.
were conducted as in the previous experiment. Struct. (1995) in press.
To estimate the real crack opening displacement at 2. Sansalone, M. and Carino, N. J., 'Impact-Echo: A Method
for Flaw Detection in Concrete Using Transient
the impact-echo test locations for both the experiments,
Stress Waves', NBSIR 86-3452, National Bureau
nonlinear fracture analyses were conducted. The results
of Standards, Washington, DC/PB 87-10444/AS
of these analyses and the determination of the minimum (National Technical Information Service, Springfield,
crack width that can be detected using the impact-echo MA, 1986).
method are presented in a subsequent paper [1]. 3. Proctor, T. M., 'Some details on the NBS conical transducer',
J. Acoustic Emission 1, No. 3 (1982) 173-178.
4. Pratt. D. and Sansalone, M., 'Impact-echo signal interpre-
tation using artificial intelligence', Mater. J. ACI 89
ACKNOWLEDGEMENTS (1992) 178-187.
5. Cheng, C. and Sansalone, M., 'The impact-echo response
The assistance of Mr. Timothy Bond, manager of the
of concrete plates containing delaminations - numerical,
George Winter Structural Engineering Laboratory, is experimental, and field studies', Mater. Struct. 26 (1993)
gratefully acknowledged. The research work presented in 274-285.
this paper was funded by Grant No. MSM-8957131 from 6. Broek, D., 'Elementary Engineering Fracture Mechanics',
the United States National Science Foundation. The Chap. 5 (Martinus Nijhoff, 1986) pp. 137-140.

RESUME face, tandis que le d@lacement superficiel entrainO par

l'ouverture de la fissure Otait mesurO et des essais
Evaluation de la largeur minimale de fissure d6tectable par dYmission acoustique rOalis~s le long du trajet de la
la m6thode d'/~mission acoustique, ldre pattie:Etude fissure. En utilisant les rksultats des analyses de rupture
exp6rimentale non linbaire, on a Otabli la relation entre le dbplacement
superficiel et le d~placement de la v~ritable ouverture
On a exOcutO deux types d'essais en laboratoire sur causbe par la fissure. On a dkterrninb la largeur minimale
des ~chantillons de plaques de b~ton afin de d~terminer de la fissure par l' estimation des d@lacements d' ouverture
quelle est la fissure minimale dbtectable par la mbthode 3 chaque point par la mbthode dYmission acoustique. On
dYmission acoustique. Dans chaque essai, une fissure ktait prbsente iei les rksultats des essais. Dans un article ultbrieur
propagke h travers la plaque parallOlement d la sur- on donnera les rksultats des analyses de rupture.