ULTRASONIC CHARACTERIZATION OF DEFECTS

AND DAMAGE IN THICK COMPOSITES

Isaac M. Daniel and Shi-Chang Wooh

The Robert R. McCormick School of

Engineering and Applied Science
Northwestern University
Evanston, Illinois 60208

INTRODUCTION

The fabrication and use of composites in thick sections create special needs for
flaw detection and characterization which cannot be met by conventional nondestruc-
tive evaluation (NDE). Thick composite sections are susceptible to a variety of fab-
rication defects and in-service damage. Typical fabrication defects include matrix
cracking, porosity, delaminations, fiber misalignment and waviness, fiber fractures,
and nonuniform matrix distribution (fiber volume ratio). A consequence of fabri-
cation defects is the variation of mechanical properties through the thickness and
the buildup of residual stresses. In-service defects result from environmental factors,
such as thermal gradients, and mechanical loading, such as impact.

A variety of NDE techniques are used for evaluating the integrity of composite
materials [1]. Although ultrasonic methods have some limitations related to attenu-
ation and penetration, they still offer the most promising approach to NDE of thick
composites. Techniques for characterization of some manufacturing defects in thick
composites were discussed before [2]. Previously developed techniques were extended
to three-dimensional imaging and to detection of transverse matrix cracks.

The material investie;ated here was AS4/3501-6 graphite/epoxy (Hercules, Inc.).

Three 10.16 X 15.24 em (4 X 6 in.) plates with different types of defects were fabri-
cated and studied.

EXPERIMENTAL PROCEDURE AND RESULTS

Specimen No.1- Embedded Teflon Inclusions

This specimen was a 150-ply thick unidirectional graphite/epoxy plate with

7.9 mm (0.31 in.) diameter and 0.025 mm (0.001 in.) thick Teflon inclusions. These
inclusions were embedded between the plies at various locations through the thick-
ness.

The conventional pulse-echo method with a single focused transducer is not

suitable for thick composites because the focal depth-of-field is much shorter than
the thickness of the specimen; thus, flaws in the out-of-focus region cannot be de-
tected well. Furthermore, the focused ultrasonic beam cannot penetrate the thick
section. For this reason, the modified through-transmission technique described be-

Review of Progress in Quantitative Nondestructive Evaluation, Vol. 9 1489

Edited by D.O. Thompson and D.E. Chimenti
Plenum Press, New York, 1990
 transmitter

- ---- h

receiver
Fig. 1 Illustration of specimen with embedded inclusions and modified through-
transmission setup.

fore, was developed [1,2]. An illustration of the specimen cross section and the modi-
fied through-transmission scanning arrangement is shown in Fig. 1.
The RF signal received by the receiving transducer was fed through a gated
peak detector. The output peak voltage was recorded by a low speed A/D converter
(HP7090A) and transmitted to a micro-computer (IBM PC/ AT). The movement of
the scanning bridge with the transducers was programmed by the micro-computer.
The peak voltage recorded was considered as an image function or gray scale of the
spatial coordinates, f(x, y). The continuous function f(x, y) was approximated by
an equally spaced array of 256 discrete gray levels. A gray-level image depicting a
planar view of the specimen with its defects was obtained.
The original C-scan as obtained by the through transmission technique dis-
cussed before seemed blurry and the edges of the embedded inclusions were not
well defined. In order to enhance the contrast and definition a technique was used
combining thresholding, logarithmic transformation and contrast stretching [2, 3].
The result of enhancement operations performed is shown in the enhanced image of
Fig. 2.

Fig. 2 Enhanced through-transmission C-scan of 150-ply laminate with embedded

inclusions.

1490
 - (b)
-
-
Q) (a)
- -
as
C) - -
fll
-,,aA, -r---; Jil. ,,.
.h.
:>.. I'
- -
'Cd"'
....c...,'"' -
-
-
-
'as"' I I I I I I I I I

- (c) (d)
-
-
.a At IIIlo
.,, .,,
_aA, 11.

-
-
" -
-
- -
I I I I I I I I I I I I
0 5 10 15 20 0 5 10 15 20
Time, t, J..Lsec Time, t, J..LSec
Fig. 3 A-scans of 150-ply laminate with embedded inclusions at various locations.
(a) Unflawed location, (b) Through nearest inclusion, (c) Through middle
inclusion, and (d) Through farthest inclusion.

In addition to the two-dimensional C-scan obtained as described before, tech-

niques were further extended to three-dimensional imaging of the flaws. Three-
dimensional imaging for relatively thin composite laminates was developed and de-
scribed elsewhere in this volume [4]. However, this method had to be modified and
adapted to the case of thick composites.

The specimen was scanned twice with the focused/unfocused transducer pair.
The first scanning was performed in through-transmission mode as described before
and the peak-to-peak values of the received signal were saved and used later in the
second scanning. In the second scanning, which is done in the same pattern as the
first, only the unfocused transducer is used in the pulse-echo mode to provide time-
of-flight information. All waveforms, such as those shown in Fig. 3, were stored for
subsequent analysis.
The analysis of the stored waveform consists of two parts, detection of the flaws
and determination of their through-the-thickness location. The C-scan image ob-
tained by the through-transmission scanning is reviewed in a bilevel (black and
white) form and an appropriate flaw discrimination threshold value yielding the best
image is selected. The through-transmission database is interrogated on a point-by-
point basis by comparing the peak-to-peak value with the selected threshold value.
If a flaw is indicated, the corresponding pulse-echo waveform at that location is re-
called and the through-the-thickness location of the flaw is determined by measuring
the time of flight from the front surface reflection. Two different time gates were
used, one between the front and backface reflections to monitor the flaw and the
other on the backface to monitor the backface reflection. Since the transmitter is
unfocused, multiple reflections from the same planar flaw are possible. Thus, at any
given point only the first internal reflection was taken as a flaw indicator.

The construction of the necessary database for three-dimensional imaging re-

quires a huge amount of storage. This problem was resolved by using data compres-
sion and pattern matching techniques described in detail elsewhere in this volume

1491
 Fig. 4 B-scan of 150-ply laminate along a line through inclusions.

[4]. One disadvantage of the procedure used here was the long scanning time for the
consecutive scans. This can be resolved by using two receivers connected to both
transducers and thus acquiring the through-transmission and pulse-echo data simul-
taneously.
Once the gate parameters (location and threshold value) are determined, B-
scans are constructed from the database, representing the through-the-thickness
location of the flaws along a scan line. A typical B-scan along a line through three
inclusions is shown in Fig. 4.
A three-dimensional isometric view of the flaw map is obtained by projecting
stacks of B-scans onto the viewport plane for a set of given viewing angles. This
technique combines the best attributes of both B- and C-scans to produce a quasi-
three-dimensional or isometric projection of the object. The position coordinates of
the transducer ( x, y) and the time-of-flight information (or depth, z) at each point
are used to create the illusion of a 3-D image on a 2-D display plane by transforming
the x, y, z coordinates onto the eye coordinates x., Ye as follows:
x. = Px + m ( x cos B + y sin B) (1)
Ye = pll + m [(y cos B- x sin B) sin</>+ z cos</>]
where 8 is the rotation of the x- and y-axes, </> is the tilt of the z-axis, m is the zoom
factor, and Px, Py are the pan coordinates. Three-dimensional views of the specimen
for four sets of rotation and tilt angles are shown in Fig. 5.

Specimen No.2- Matrix Cracks

This specimen was a 40-ply thick crossply graphite/epoxy laminate of
[Ow/90w]s layup. The large difference in longitudinal and transverse coefficients
of thermal expansion results in high transverse tensile stresses in each layer after
cooldown from the elevated stress-free temperature. If these thermal stresses are
sufficiently high, they induce cracks in the matrix. Transverse matrix cracking in
all layers was induced and maximized by immersing the specimen in liquid nitrogen.
The depth of the transverse matrix cracks was approximately 2.54 mm (0.1 in.).
An X-radiograph of the cracked specimen was obtained by using an X-ray
opaque penetrant (Zinc Iodide solution) for subsequent comparison with ultrasonic
results. The ultrasonic backscattering technique with oblique incidence was not
found suitable in this case. Because of the depth of the crack tip below the surface
and the resulting attenuation of the ultrasonic beam, less energy is scattered back
from the crack tip than from the surfaces of the specimen. The as-obtained image
showed the pattern of the bleeder cloth on the surface of the specimen produced in
the fabrication process.
The normal incidence pulse-echo method was more successful. The beam was
focused at the crack tip causing scattering at this tip and resulting in attenuation

1492
 Fig. 5 Three-dimensional views of embedded inclusions in 150-ply laminate.
Clockwise from top left, rotation and tilt angles: (oo, oo), (oo, goo),
(30°, 60°), (goo, goo).

of the backface reflection. The intensity of this back-wall echo determines the de-
gree of scatter at the crack tip. Thus, normal incidence pulse-echo gives a C-scan of
the cracks although the physical interaction of the wave and the cracks is different
from that for planar defects, such as delaminations. Figures 6(a) and 6(b) show the
X-radiograph and as-obtained C-scan, respectively. The latter was subsequently en-
hanced by sharpening and one-dimensional segmentation [3]. Sharpening is obtained
by subtracting from the original image function (gray scale) a multiple of its Lapla-
cian. One-dimensional segmentation was used to transform the image into segments
of lines representing loci of high gradient in the x- and y-directions, Fig. 6(d).

Specimen No.3- Porosity and Cracks

This specimen was a 200-ply thick crossply graphite/epoxy plate of [0 25 /90 25 hs
layup. The mismatch in coefficients of thermal expansion between the various layers
and the increased thickness resulted in matrix cracking and a high degree of porosity
throughout. The X-radiograph shown in Fig. 7 shows the extent of matrix cracking
in a cumulative way, since the radiograph shows the projection of all cracks through
the thickness. Attempts to obtain an ultrasonic image of the specimen were not suc-
cessful as the ultrasonic beams of 2.25 MHz and 1 MHz transducers were attenuated
entirely and could not penetrate the specimen in the thickness direction.
Since ultrasound propagates along the fiber direction with much lower attenua-
tion, the specimen was examined by means of in-plane wave propagation. A contact
1 MHz transducer applied to one edge was used in the pulse echo mode. A typical
pulse containing five consecutive echos is shown in Fig. 8. The measured density and
wave propagation velocity, elL, were used to obtain the longitudinal modulus E 1 of
the 0° layer and the fiber and void volume ratios from the following relations:
E1 = PcC~L
E 1 = E,v, + Em(l- Vf- Vv) (2)
Pc = P!Vf + Pm(1- Vf- Vv)
where

1493
 (a)

(b)

(c)

(d)

Fig. 6 Thermally induced matrix cracks in crossply graphite/epoxy laminate.

(a) X-radiograph, (b) As-obtained ultrasonic C-scan, (c) Sharpened C-scan,
and (d) Enhanced image by one-dimensional segmentation.

1494
 Fig. 7 X-radiograph of [02s/902shs graphite/epoxy specimen.
f!b Pm, Pc =fiber, matrix and composite densities, respectively.
l'j, V., = fiber and void volume ratios, respectively.
Ef> Em =fiber and matrix moduli, respectively.

From the measured values of Pc = 1, 490 kg/m3 and elL = 9,144 ms-I, the
following were obtained:

E 1 = 124.6 Gpa {18.1 x 106 psi)

YI = o.53
v., = 0.046
Multiple echo records such as those of Fig. 8 were obtained in the 0° direction
for an unflawed specimen, a specimen with less approximately 1% porosity and the
crossply specimen. The attenuation in decibels per unit length was obtained from
the following relationship:

_ 5 An~+I
a - - 1og 2 2 (3)
h AnAn+2 + An+! An+2
where An, An+I, and An+ 2 are the amplitudes of three consecutive echos. Results for
the three specimens above are:

Unidirectional unflawed specimen: a =0.048 dB/em (0.122 db/in.)

Unidirectional specimen with 1% porosity: a = 0.236 dB/em (0.600 dB/in.)
0° layer of crossply specimen: a = 0.62 ± 0.18 dB/em
(1.58 ± 0.46 dB/in.)
Although no quantitative relation has been obtained yet between attenuation
and porosity, it seems that this should be pursued as attenuation is a sensitive mea-
sure of porosity.

SUMMARY AND CONCLUSIONS

Ultrasonic techniques with appropriate enhancement procedures were used
to detect and characterize defects in thick composite laminates. Embedded film
inclusions, simulating delaminations, were characterized by means of a through-
transmission technique employing a flat transmitting and a focused receiving trans-
ducer. A three-dimensional image of the defects was obtained by combining and pro-
cessing pulse-echo and through-transmission signals obtained from the same setup.

1495
 -
'+->
.._
fiJ

(1.)
"'d

-
;::!
.....
.....
~

~
0 10 20 30 40 50 60
Time, t, /-LSec
Fig. 8 Five consecutive echos of in-plane wave along 0° layer of 200-ply crossply
specimen.

Deep cracks in a crossply laminate were imaged by means of nonnal incidence pulse-
echo technique with enhancements, such as sharpening and one-dimensional segmen-
tation. Thicker crossply laminates containing extensive porosity in addition to ma-
trix cracks resulted in too high attenuation for ultrasonic imaging. Fiber and void
volume ratios were obtained from density and wave propagation velocity measure-
ments. Attenuation was found to be a sensitive measure of porosity.

ACKNOWLEDGEMENT
The work described here was sponsored by the Office of Naval Research. We are
grateful to Dr. Y. Rajapakse of ONR for his encouragement and cooperation.

REFERENCES
1. I. M. Daniel, S. C. Wooh, and J. W. Lee, "Defect and Damage Characteriza-
tion in Composite Materials," Review of Progress in Quantitative Nondestruc-
tive Evaluation, Vol. 6B, ed. by Donald 0. Thompson and Dale E. Chimenti,
Plenum Publishing Corp., 1987, pp.ll95-1202.
2. I. M. Daniel and S. C. Wooh, "Ultrasonic Techniques for Characterization of
Manufacturing Defects in Thick Composites," Review of Progress in Quantita-
tive Nondestructive Evaluation, Vol. 8B, ed. by Donald 0 . Thompson and Dale
E. Chimenti, Plenum Publishing Corp., 1989, pp.1605-1612.
3. S. C. Wooh and I. M. Daniel, "Enhancement Techniques for Nondestructive
Evaluation of Composites," ASME Journal of Engineering Materials and Tech-
nology, October, 1989.
4. S. C. Wooh and I. M. Daniel, "Three-Dimensional Imaging of Impact Damage
in Composite Laminates," Review of Progress in Quantitative Nondestructive
Evaluation, Vol. 9, ed. by Donald 0 . Thompson and Dale E. Chimenti, Plenum
Publishing Corp., 1990.

1496