ARTICLE IN PRESS

Optics and Lasers in Engineering 42 (2004) 131–140

A new set-up for pulsed digital shearography

applied to defect detection in composite
structures
Fernando Santosa, Ma! rio Vazb, Jaime Monteiroc,*
a
NDT Expert, Toulouse, France
b
Faculdade de Engenharia da Universidade do Porto, DEMEGI/SMAP, Rua Dr Roberto Frias,
4200-465 Porto, Portugal
c
#
Instituto de Engenharia Mecanica * Industrial, DEMEGI/SMAP, Rua Dr Roberto Frias,
e Gestao
4200-465 Porto, Portugal

Received 30 April 2003; accepted 7 July 2003

Abstract

An optical set-up based on the combination of new CCD technology, a ruby laser and a
Mach-Zehnder shear interferometer is presented for defect detection in composites. Transient
loads are used to excite defects in a sandwich structure and image-processing routines allow
improving the detection resolution. Phase calculation is obtained by FFT algorithms applied to
interference patterns with a spatial carrier in the primary fringes. Independent control of the
shear and the frequency of the spatial carrier can be obtained with this set-up. The experimental
implementation of the set-up and the results obtained with it are presented and discussed.
r 2003 Elsevier Ltd. All rights reserved.

Keywords: Ruby laser; Mach-Zehnder; Interferometry; Pulsed digital shearography; FFT

1. Introduction

The high stiffness/weight ratio presented by the composite materials, when

compared with metallic ones, makes them well adapted for the construction of
aeronautical and aerospatial structures. Recent aeronautic projects, like Tiger, EFA
or Rafale, have most of its components made in different kinds of composite
materials.

*Corresponding author. Faculdade de Engenharia da Universidade do Porto, DEMEGI/SMAP, Rua

Dr Roberto Frias, 4200-465 Porto, Portugal. Tel.: +351-22-5081597; fax: +351-22-5081584.
E-mail address: jmont@fe.up.pt (J. Monteiro).

0143-8166/$ - see front matter r 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/j.optlaseng.2003.07.002
 ARTICLE IN PRESS
132 F. Santos et al. / Optics and Lasers in Engineering 42 (2004) 131–140

Being non-homogeneous, the composites have frequently, in its interior, defects

that can slow down its mechanical performances. As they present low energy
absorption coefficient they are vulnerable to damages due to impact loads, like tool
drops during maintenance, bird strike or impacts with other projectiles during
service. This way, the need for Non-Destructive Techniques (NDT) that allows a
quick monitoring of the structure integrity is of crucial importance in industries
working with this kind of materials.
The great development of experimental mechanics during the last decades was
mainly due to the optical techniques. The video recording of interferometric patterns
and the introduction of image processing techniques for data interpretation were
crucial in this development. Nowadays it is possible to build systems which are based
on methods that allow real time high-resolution field measurements with no contact
with the inspected components.
The shearography is a special case of the holographic interferometry techniques in
which the interfering wave fronts, both speckled, are at the same time reference and
object [1,2], leading to a fringe pattern corresponding to the displacement gradient of
the object surface in two different states. Being a common path interferometric
technique it has the advantage of allowing short coherence length illumination and is
less sensitive to environmental disturbances than other holographic techniques [3].
Internal defects, if properly excited, can be revealed by its effect in the surface
displacement field. Debondings and delaminations will cause localised displacement
gradients that can be easily detected. With shearography these gradients are directly
measured with no need for numerical differentiation.
To excite the defect the structure must be stressed in some non-destructive way.
When CW illumination is used a gentle heating of the surface (1
or 2
) or a small
underpressure load can be successfully used in most of the situations.
In each interferometric pattern a huge quantity of information become available.
The interferometric pattern corresponds to the distribution of displacement gradient
that occurs between the two recordings. As this information is codified in the phase
of the fringe pattern the calculation of the phase maps is crucial. Using CW
illumination temporal phase shift is normally used to perform phase calculation.
With this technique at least 3 recordings of the fringe pattern have to be obtained
after adding known phase steps. For that a computer controlled phase modulator is
included in the set-up. All the set-up and environment should be kept very stable to
minimise errors.
By using pulsed lasers the influence of environmental perturbations is drastically
reduced [4]. Two consecutive states of the structure are recorded by double exposure
and correlated afterwards. During the recording of the images two parameters will
bias the characteristics of the measurement:
* Displacement of the different points of the structure during each exposition.
* Relative displacement between both instants of recording.

To minimise the first one, the relative displacement of the structure during the
recording should be less than a tenth of the light wavelength. A pulse duration of
 ARTICLE IN PRESS
F. Santos et al. / Optics and Lasers in Engineering 42 (2004) 131–140 133

about 20 ns to Ruby lasers or 10 ns to Nd:YAG can be easily achieved. So, for

displacement speeds of some m/s no special precautions are needed. In what
concerns the second parameter, the amplitude of the relative displacement between
two states is limited by controlling the pulse separation. Typically, this time interval
Dt varies between 1 and 800 ms for a ruby laser. Displacement amplitude of 7 mm is
obtained for a speed of about 7 m/s if a pulse separation of 1 ms is used. This way,
combining short exposure times and small pulse separations it is possible to record,
in situ, the dynamic behaviour of a structure neglecting the problems of hazardous
vibrations and other environmental noise.
When transient phenomena have to be studied temporal phase calculation
techniques (phase shift) are not the best approach because they demand set-ups that
are difficult to adjust due to the presence of a double reference in the reconstruction
process [5]. In this case good results can be obtained using pulsed techniques with
spatial phase algorithms. These techniques rely upon the use of a spatial carrier to
codify the modulation induced by the object deformation. Depending on the set-up
used the carrier can be introduced in the primary or in the secondary fringes. Several
ways have been proposed to generate the carrier and to perform its demodulation
[6–8].
When the carrier is in the secondary fringes, the demodulation involves only one
interferogram and leads directly to the deformation [9–11]. If the carrier is in the
primary fringes, its demodulation leads to the phase distribution of each wave front
that, after subtraction, leads to the deformation pattern [12,13]. Spatial algorithms
used to access phase maps in pulsed recordings are sinusoidal fitting [14] and FFT
phase calculations [15].

2. Phase extraction of an interferometric image by Fourier transform

The light intensity of an interferogram is given by

gðx; yÞ ¼ aðx; yÞ þ bðx; yÞ cos ðfðx; yÞÞ; ð1Þ
where, aðx; yÞ is the background light intensity, bðx; yÞ the interference fringes
amplitude and fðx; yÞ the wanted phase, ðx; yÞ being the image plane co-ordinates.
The previous equation can be written as
gðx; yÞ ¼ aðx; yÞ þ cðx; yÞ þ c ðx; yÞ ð2Þ
with
bðx; yÞ ifðx;yÞ
cðx; yÞ ¼ e ð3Þ
2
where c ðx; yÞ is the conjugate complex of cðx; yÞ:
It is possible to isolate the term cðx; yÞ; containing the phase information, in the
frequency domain. Eq. (2), after one-dimensional Fourier transform, has the
following form:
Gðf ; yÞ ¼ Aðf ; yÞ þ Cðf ; yÞ þ C  ðf ; yÞ: ð4Þ
 ARTICLE IN PRESS
134 F. Santos et al. / Optics and Lasers in Engineering 42 (2004) 131–140

As the terms C and C  are symmetric a filter containing only the positive
frequencies has to be used to keep only the term Cðf ; yÞ: Using the inverse FFT is
possible to obtain cðx; yÞ from where the 2p modulus phase can extracted by the
equation
Imðcðx; yÞÞ
fðx; yÞ ¼ arctg : ð5Þ
Reðcðx; yÞÞ
The necessary condition to obtain the phase is that the background frequency varies
little when compared with the carrier one. Otherwise it will be very difficult to eliminate
the term Aðf ; yÞ and to isolate one of the signal lobes. If closed fringes are obtained,
both lobes will mix and signal inversions will occur. This implies that the carrier to be
used should be properly adjusted to allow a special separation between C and C  :
The phase obtained with only one interferogram has a signal indetermination.
This problem can be overtaken if the direction of the deformation is known. This can
be achieved interactively by creating a signal map or by introducing a known carrier
frequency [16].
The light intensity of an interferogram with a carrier frequency is described by
gðx; yÞ ¼ aðx; yÞ þ bðx; yÞ cos ðfðx; yÞ þ 2pf0 xÞ; ð6Þ
where aðx; yÞ is the light intensity of the background and bðx; yÞ the modulation term.
fðx; yÞ is the wanted interferogram phase from the signal gðx; yÞ: f0 is the information
carrier frequency introduced by the generation of a parallel fringe pattern.
Using
bðx; yÞ ifðx;yÞ
cðx; yÞ ¼ e ð7Þ
2
bðx; yÞ ifðx;yÞ
c ðx; yÞ ¼ e ð8Þ
2
one obtains
gðx; yÞ ¼ aðx; yÞ þ cðx; yÞ ei2pf0 x þ c ðx; yÞ ei2pf0 x ð9Þ
which has the following form in the frequencies domain:
Gðf ; yÞ ¼ Aðf ; yÞ þ Cðf  f ; yÞ þ C  ðf þ f ; yÞ:
0 0 ð10Þ
In the case of aðx; yÞ; bðx; yÞ and fðx; yÞ having a frequency much smaller then the
carrier frequency f0 ; the Fourier spectrum will represent two lobes C and C  really
separated and symmetric in relation to the 0 frequency A and separated by 7f0 ; as
shown in Fig. 1.

3. Mach-Zehnder shearography

Shearography is normally performed using a set-up based on a Michelson

interferometer. This is a very compact and simple solution that allows an easy
alignment. When pulsed lasers are used this solution becomes very complex regarding
the introduction of a carrier frequency necessary for spatial phase calculation [17].
 ARTICLE IN PRESS
F. Santos et al. / Optics and Lasers in Engineering 42 (2004) 131–140 135

G (f, y)

fo fo

A (f, y)

C * ( f + fo, y) C ( f _ fo, y)

Fig. 1. Schematic representation of 1D spectrogram after introduction of carrier fringes.

The proposed shearographic set-up is based in the Mach-Zehnder interferometer.

This configuration allows for an independent control of the two arms of the
interferometer. The introduction of carrier fringes becomes independent of the light
source and adjustable in the interferometer [18]. This interferometric set-up allows
the generation of shear images through a tilt of one mirror and the introduction of a
spatial carrier frequency by translating the other mirror. This way it is possible to
introduce a carrier frequency in the primary fringes allowing phase calculation.
Considering cylindrical wave fronts, the introduction of shear between images and
the carrier frequency can be described in a simple way by the schematic
representations presented in Fig. 2. Thus, considering a small tilt, a; in one of the
mirrors one observes two wave fronts overlapping as shown in Fig. 2(a). Considering
a translation, e; in one of the mirrors, see Fig. 2(b), the two wave fronts overlap
creating interference fringes. This means that one of the mirrors controls the amount
of shear through its rotation and the other mirror controls the carrier frequency
through its translation.
The aperture of the system is a vertical slit that is placed at the interferometric
system entrance. The aperture of the system is longitudinal instead of circular to
increase the speckle size in one direction. The spatial carrier is introduced
perpendicular to the horizontal direction of the speckle.
In this set-up a CCD Kodak ES1.0, camera allowing double acquisition, was used.
The time delay control was done through a commercial trigger system using a firing
pin to load the object and start the measurements. A Lumonics HSL2 model ruby
laser emitting 800 mJ per pulse was used as laser source. A schematic presentation of
the set-up is depicted in Fig. 3.
The interferometer used was assembled in the top of a small optical table as shown
in Fig. 4. In Fig. 4(a) the position control systems for both mirrors can be seen. In
Fig. 4(b) the system aperture is perfectly visible.

4. Experimental results

The analysed object was a composite plate with a sandwich structure; a core of
Nomexs with skins made of fiberglass. Two circular defects were artificially made
 ARTICLE IN PRESS
136 F. Santos et al. / Optics and Lasers in Engineering 42 (2004) 131–140

Beam-splitter
ε
Mirror Beam-splitter
Mirror with
translation ε

Mirror with
rotation α
Mirror
Beam-splitter
(a) α (b) Beam-splitter

Fig. 2. (a) Shearing of images and (b) introduction of carrier fringes in a Mach-Zehnder interferometer.

Mirror
Ruby Laser
Object
Lens
Mach-Zehnder
Shear Interferometer

Slit aperture
KODAK

Image lens
Fig. 3. Set-up used for validation of the experiment.

Fig. 4. (a,b) Images of the prototype.

one with 2.5 cm diameter and the other with 1 cm in diameter. A picture of the tested
plate can be seen in Fig. 5.
A transient load applied with an electro-dynamic hammer (firing pin) was used to
excite the defects. A schematic diagram of the complete set-up is shown in Fig. 6.
This set-up includes a ruby laser, an optical head and all the necessary electronics
for data acquisition and synchronisation.
 ARTICLE IN PRESS
F. Santos et al. / Optics and Lasers in Engineering 42 (2004) 131–140 137

Fig. 5. Image of the object under study.

Firing pin

Accelerometer

Object

Digital
oscilloscope PC

Optical
head
Frequency
analyser
Illumination
device

Band pass filter

Double pulse ruby laser

Trigger system

Laser control
Video signal

Fig. 6. Complete set-up for pulsed shear.

Using a digital oscilloscope to store two signals, one coming from the
accelerometer placed in the plate and the other from the photodiode of the ruby
laser, it is possible to observe the instant at which the laser shots occurred compared
to the beginning of the impact.
The optical head is a Classical Mach-Zehnder interferometer in which the object
observation is made through a narrow slit placed at the entrance of the optical
system and a conventional objective placed before the CCD. In Fig. 7, the first image
represents the interference speckle pattern observed through the proposed optical
 ARTICLE IN PRESS
138 F. Santos et al. / Optics and Lasers in Engineering 42 (2004) 131–140

system and in which, we can observe with clarity the primary carrier fringes over the
elongated speckle grains introduced by the vertical slit. The following images
represent the phase calculation by FFT of the previous interference pattern.
Images presented in Fig. 8 represent some of the most significant results obtained
in these tests. For all the images the time delay between the two shots was 30 msU The
first image presents the surface phase map in an instant, 0.025 ms after impact, where
the shock wave is still in the beginning of the propagation. So, as the defect is
supposed to be a little bit more to the left it is still possible to stimulate it in the way
to make it visible. The second figure represents an instant, 0.060 ms after impact,
where the shock wave just passed through the defected zone clearly revealing the
defect. As the size of shear was superior to the defect dimension, the two lobules of
the gradient function are separated, which in any case, makes the interpretation of
the image more difficult.
After observing the last four images it is possible to conclude that the smaller
defect is the one that is easily revealed. This is due to the fact that these techniques
search for local displacement gradients. So, the smaller defect since it is more
localised, is revealed in a more evident way.
We must refer to the importance of the accelerometer position in the plate to be
tested. So, the accelerometer must be in the zone to be inspected and near the shock

Fig. 7. Interference pattern and phase calculation by FFT.

Fig. 8. Shock wave propagation captured in 6 different instants, between 0.025 and 0.775 ms after impact.
 ARTICLE IN PRESS
F. Santos et al. / Optics and Lasers in Engineering 42 (2004) 131–140 139

zone so that the registers of the oscilloscope are the most exact possible. We should
also test the plate with a less intensity shock and with a time delay between pulses
also less. Through the observation of the signal in the oscilloscope coming from the
accelerometer, it is possible to correct the time shot of the laser. So, if we pretend to
obtain few fringes the two shots must occur at the time when the velocity is
minimum, meaning when the difference of displacements between the two different
instants in time is minimum.

5. Conclusions

The obtained results allow the conclusion that Pulsed Shearography can be used in
the detection of defects in composite materials. The need to create displacement
gradients that propagates at high speed compels to use the stimulation by shock.
However, these are low-energy impacts that do not compromise the integrity of the
components, keeping the non-destructive character of these techniques. As main
advantage of the pulsed techniques we can point out the reduced influence that
environmental perturbations have in the measurements.

Acknowledgements

Authors would like to thank the financial support from FCT (Funda@*ao da
Ci#cncia e Tecnologia) under the projects: PRAXIS/C/EME/13180/98, POCTI/EME/
35901/99 and the co-operation with Holo3 (France). Part of this work was
performed where the author was on leaving from Holo3.

References

[1] Hung Y, Taylor C. A tool for measurement of derivatives of surface displacement. Proc Photo-opt
Instrum Eng (SPIE) 1973;41:169–75.
[2] Leendertz JA, Butters JN. An image-shearing Speckle Pattern Interferometric Technique for
measuring bending moments. J Phys E 1973;6:1107–10.
[3] Owner-Peterson M. Decorrelation and fringe visibility: on the limiting behaviour of various electronic
speckle pattern correlation interferometers. J Opt Soc Am 1991;8:1082–9.
[4] Spooren R, Dyrseth A, Vaz M. Electronic shear interferometry: application of a (double-) pulsed
laser. Appl Opt 1993;32(25):4719.
[5] Crawforth L, et al. Application of pulsed laser holographic interferometry to the study of magnetic
disk drive component motions. In: Proceedings of the 1990 Fall Conference of the SEM, Bethel, 1990.
p. 404–12.
[6] Hariharan P, Ramprasad B. Wavefront tilter for double-exposure holographic interferometry. J Phys
E 1973;6:173–5.
[7] Matthys D, et al. A windowing technique for the automated analysis of holo-interferograms. Opt
Laser Eng 1988;8:123–36.
[8] Pedrini G, Froning P, Fessler H, Tiziani H. Transient vibration measurement by using multi-pulse
digital holography. Opt Laser Technol 1997;29:505–11.
 ARTICLE IN PRESS
140 F. Santos et al. / Optics and Lasers in Engineering 42 (2004) 131–140

[9] Farrant D, Kaufmann G, Petzing J, Tyrer J, Oreb B, Kerr D. Measurement of transient deformations
with dual-pulse addition electronic speckle pattern interferometry. Appl Opt 1998;37(31):7259–67.
[10] Davila A, Kerr D, Kaufmann G. Fast electro-optical system for pulsed ESPI carrier fringe
generation. Opt Commun 1996;123(4–6):457–64.
[11] Fern!andez A, Kaufmann G, Doval A, Blanco-Garcia J, Fern!andez J. Comparison of carrier removal
methods in the analysis on TV holography fringes by the Fourier transform method. Opt Eng
1998;37:2899–905.
[12] Santos F, Chousal J, Vaz M, A dual beam pulsed ESPI system for measurements of out-of-plane
displacements. Proceedings of the 11th International Conference on Experimental Mechanics,
Oxford, 1998.
[13] Pedrini G, Pfister B, Tiziani HJ. Double-pulsed-electronic speckle interferometry. J Modern Opt
1993;40:89–96.
[14] Pedrini G, Tiziani HJ. Double-pulse electronic speckle interferometry for vibration analysis. Appl
Opt 1994;33:7857–63.
[15] Takeda M, Ina H, Kobayashi S. Fourier-transform method of fringe-pattern analysis for computer-
based topography and interferometry. J Opt Soc Am 1982;72:156–60.
[16] Kreis TM, Juptner WPO. Fourier-Transform Evaluation of interference patterns: the role of filtering
in the spatial frequency domain. Laser Interferometry: Quantitative Analysis of Interferograms, SPIE
1989;1162:116–25.
[17] Pedrini G, Zou Y, Tiziani H. Quantitative evaluation of digital shearing interferogram using the
spatial carrier method. Pure Appl Opt 1996;5:313–21.
[18] Fern!andez A, Doval AF, D!avila A, Blanco-Garc!ıa J, P!erez-Lopez! C, Fern!andez JL. Double-pulsed
carrier speckle-shearing pattern interferometry for transient deformation analysis. In: Kujawinska M,
Pryputniewicz RJ, Takeda M, editors. Laser Interferometry IX: Techniques and Analysis; Proc.
SPIE. vol. 3478, San Diego, USA; 1998. p. 352–58.