See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/281105572

41 Caw Ad JSA Vol 14 Issue 2

Data · August 2015

CITATIONS READS
0 1,140

2 authors:

Peter Cawley R. D. Adams

Imperial College London University of Bristol
383 PUBLICATIONS   19,105 CITATIONS    230 PUBLICATIONS   13,013 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

University of Bristol Innovative Approaches to Composite Structures, 1995-1998 View project

NDT of Adhesive Joints View project

All content following this page was uploaded by R. D. Adams on 20 August 2015.

The user has requested enhancement of the downloaded file.

The Journal of Strain Analysis for Engineering
Design
http://sdj.sagepub.com/

The location of defects in structures from measurements of natural frequencies

P Cawley and R D Adams
The Journal of Strain Analysis for Engineering Design 1979 14: 49
DOI: 10.1243/03093247V142049

The online version of this article can be found at:

http://sdj.sagepub.com/content/14/2/49

Published by:

http://www.sagepublications.com

On behalf of:

Institution of Mechanical Engineers

Additional services and information for The Journal of Strain Analysis for Engineering Design can be found at:

Email Alerts: http://sdj.sagepub.com/cgi/alerts

Subscriptions: http://sdj.sagepub.com/subscriptions

Reprints: http://www.sagepub.com/journalsReprints.nav

Permissions: http://www.sagepub.com/journalsPermissions.nav

Citations: http://sdj.sagepub.com/content/14/2/49.refs.html

>> Version of Record - Apr 1, 1979

What is This?

Downloaded from sdj.sagepub.com at SAGE Publications on December 6, 2011

 THE LOCATION OF DEFECTS IN
STRUCTURES FROM MEASUREMENTS OF
NATURAL FREQUENCIES

P. CA WLEY University of Bristol

R. D. ADAMS University of Bristol. Fellow of the Institution

A method of non-destructively assessing the integrity of structures using measurements of the structural natural
frequencies is described. It is shown how measurcments made at a single point in the structure can be used to detect,
locate and quantify damage. The scheme presented uses finite-element analysis, since this method may be used on any
structure. The principle may, however, be used in conjunction with other mathematical techniques. Only one full
analysis is required for each type of structure.
Results are presented from tests on an aluminium plate and a cross-ply carbon-fibre-reinforced plastic plate. Excel-
lent agreement is shown between the predicted and actual damage sites and a useful indication of the magnitude of the
defect is obtained.

1 INTRODUCTION Location and assessment of the severity of the damage,

A d a m et al. (1)t found that, with fibre-reinforced
plastics, a state of damage could be detected by a reduc-
tion in dynamic stiffness and an increase in damping,
whether this damage was localized, as in a crack, or distri-
buted through the bulk of the specimen as many micro-
cracks. Changes of stlffness, whether local or distributed,
lead to changes in the natural frequencies of the structure.
Also, since the stress distribution through a vibrating
structure is non-uniform and is different for each natural
frequency (mode), any localized damage would affect each
mode differently, depending on the particular location of
the damage. Thus, the measurement of the natural fre-
quencies of a structure at two or more stages of its life
offers the possibility of locating damage in the structure
and of determining the severity of the damage. If one set of
frequencies were measured before the structure was put
into service, subsequent frequency measurements could be
used to test whether the structure was still sound. Changes
in damping could be used in exactly the same manner for
locating defects.
Measurement of the dynamic characteristics, natural
frequencies and damping, of a structure is potentially a
very attractive method of non-destructive testing, since
these properties can be measured at one point of the struc-
ture and are independent of the position chosen.
The authors ( 2 ) successfully tested the frequency-
measurement principle on structures which could be
treated as one-dimensional and which were, therefore, cap-
able of solution by closed-form analytical techniques. It
was found to be possible to detect damage equivalent to
the removal of one per cent of the cross-sectional area of
the structure at a single location. Excellent agreement was
also obtained between the predicted and actual damage
sites. The presence of damage can be detected simply by
noting changes in the natural frequencies of the structure.
The MS. of this paper was received at the Institution on 20th July 1978
and acceptedfor publication on 15th November 1978.
t References are given in Appendix 2 .
however, requires that a full dynamic analysis of the struc-
ture be carried out. This analysis must be sufficiently
accurate to give a reasonable indication of the mode
shapes.
The present work describes the extension of the
frequency method to two-dimensional and, potentially, to
three-dimensional structures. The analysis of the one-
dimensional structures described in (2) was carried out
using receptance techniques (3). There is no two-dimen-
sional equivalent of the receptance technique, so another
method had to be used. Finite-element analysis was
chosen as it is well-documented (4), accurate, and can be
used to analyse any structure. The method described in
this paper is therefore applicable to all structures, and the
proposed technique can be used with any type of
structural vibration, that is torsional, axial, flexural or
combinations of these. When testing one-dimensional
structures, it was convenient to use axial or torsional
modes, but this is not practicable with shell-type struc-
tures, so the results presented here were obtained using
flexural vibration.
Measurements of structural damping were not under-
taken in the tests described here, though the results of (1)
indicate that the damping would increase with damage. It
was felt that changes in dynamic stiffness would be easier
to measure than changes in damping, though it is intended
to investigate the latter at a later stage.
1.1 Notation
A Area of damage
B Matrix of derivatives of element shape func-
tions
D Elasticity matrix
De Average elasticity matrix for layered material
Dk Elasticity matrix for kth layer of layered
material
D* Elasticity matrix for damaged volume
E Total normalized error
e Error

JOURNAL OF STRAIN ANALYSIS VOL 14 NO 2 1979 OIMechE 1979 49

Downloaded from sdj.sagepub.com at SAGE Publications on December 6, 2011

0309-3247/19/0400-0049$02.00
 P. CAWLEY AND R. D. ADAMS

Functions Thus, provided that the change in stiffness is independent

z co-ordinate of top of kth layer of layered offrequency,
material
Functions of x, y
Global stiffness matrix
Element stiffness matrix
The ratio of the frequency changes in two modes is
Global mass matrix
Number of layers therefore only a function of the damage location. Posi-
tions where the theoretically determined ratio 6oi/60,
Volume of damage
Position vector equals the experimentally measured value are therefore
Strain energy possible damage sites.
Sensitivity of mode i to damage at r It is now necessary to compute the change in the
Thickness natural frequency of the structure due to damage at a
given site. The basis of the method is to consider damage
Volume
Weighting factor as a local decrease in the stiffness of the structure. One
Mode shape vector way of computing the changes in the natural frequencies
Co-ordinate in thickness direction due to damage in a given element of the finite-element
mesh would be to reduce the stiffness of that element and
Small change in quantity X
Strain vector to repeat the dynamic analysis. This could be done for
o2
each element in turn, so that the theoretical changes in
Stress vector frequency for damage at a number of sites would be
Frequency found. The problem with this method is that the amount of
computer time involved would be prohibitive.
Subscnpts
i,j Mode i,j 2.3 Use of sensitivity analysis
r Position given by vector r A very attractive alternative to repeating the full dynamic
min Minimum analysis in order to compute the changes in the natural
k Layer number frequencies due to localized damage is to use a sensitivity
(perturbation) analysis. The basic principles of the method
2 THEORY are described by, for example, Courant and Hilbert (8). By
this method, the sensitivities of the natural frequencies of a
2.1 Dynamic analysis system to small changes in the stiffness are calculated
The computer program used for the basic dynamic from the mode shapes of the undamaged structure pro-
analysis has been described in (5). The element used was duced by the initial full dynamic analysis.
the eight-node, 40 degrees-of-freedom finite-element The basic eigenvalue equation which is solved in the
described by Zienkiewicz et al. (6)(7). This element was dynamic analysis was
chosen because it can accommodate irregularly-shaped
boundaries and is readily generalizable to shells. It can (K - MI)x = 0 (6)
therefore be used in the analysis of most two-dimensional Consider the effect of a small change, SK, in the stiffness
structures. All the results presented in this paper have been matrix with similarly denoted changes in the other para-
obtained using this element. None of this analysis, how- meters. Equation (6) then becomes
ever, is specifically dependent on this particular element
but can be applied to any, suitable, type of element. {(K+ 6 K ) - ( M + 6M)(A+ SI)} (X + 6 ~=) 0 (7)
Multiplying out and neglecting second-order terms yields
2.2 Method of damage location
The change in the natural frequency of mode i of a K x - W X - I ~ M X~+ K x
structure due to localized damage is a function of the posi-
tion vector of the damage, r, and the reduction in stiffness -~IMx+K~x-W~X=O (8)
caused by the damage, 6K. Thus But K x - IMx = 0 from equation (6) and the model of
damage used does not include a change in mass, so 6M =
= f ( 6 K r)
h, (1) 0. Equation (8) therefore reduces to
Expanding this function about the undamaged state (6K =
0), and ignoring second-order terms, yields ~ K -x6AMx + K 6 ~ A
- M ~=x0 (9)

8s Multiplying through by X-T gives

60, = f ( 0 ,r) + 6K -(0,r)
WK) X' 6K x - 6A X'MX+ ( J K - A x ' M ) 6~= 0 (10)
But f ( 0 , r) = 0, for all r, since there is no frequency Since K and M are symmetric matrices (4), the transpose
change without damage. Hence, of equation (6) is

60, = 6Kg,(r) (3) J ( K - IM)= 0

Similarly, Post-multiplying by 6x gives
60, = 6K g,(r) (4) ( J K- A i M ) 6x = 0

50 JOURNAL OF STRAIN ANALYSIS VOL 14 NO 2 1979 CIIMcchE 1979

Downloaded from sdj.sagepub.com at SAGE Publications on December 6, 2011
 THE LOCATION OF DEFECTS IN STRUCTURES FROM MEASUREMENTS O F NATURAL FREQUENCIES

Hence, equation (10) reduces to define the damage site uniquely or to reduce the number of
x~~Kx-~Ax~Mx=O
possible sites to the minimum dictated by symmetry. How-
ever, because the frequency changes tend to be small and
or
the model of damage is unsophisticated, it is desirable to
XT 6 K x use a larger number of mode pairs, so that some averaging
6A = ~

of the results is obtained. This scheme has been imple-

XTMX
The sensitivities of the natural frequencies to changes in
stiffness, 6K, may therefore be readily computed from the
mode shape vectors and the mass matrix of the un-
damaged system. These may be output from the dynamic
analysis.
It is now necessary to compute the change in stiffness,
6K, due to damage at a point in the structure. The element
stiffness matrix is given by (4)
Ke = Jv BTDB d V
Consider a change in the stiffness of the element over
some volume R. Then the change in the stiffness matrix is
given by
6K = IRBT(D*- D)B d V
(13)
In two-dimensional structures, the damage is modelled as
being of constant severity through the thickness and
affecting an area A . If the damaged area is small, then
6K z !iB'(D* - D)B],A dz
where [BT(D* - D)BI, denotes that the expression within
the brackets is evaluated at the position of the damage site,
r. For a three-dimensional structure, the corresponding
mented, but it was found difficult to automate the plotting
of the loci, which meant that they had to be plotted
manually. This procedure was very tedious, particularly
when a large number of mode pairs was used.
An alternative was to output the sensitivities at a series
of grid points and to compute an error function at each
point, which was a measure of the error in assuming the
damage to be at that point. The point at which the error
was a minimum gave the approximate position of the
damage.
(12) Define the error in assuming the damage to be at posi-
tion r, given frequency changes d o i and 6oj in modes i
andj, respectively, as
srilsrj
errJ..=-- 1,
6w;l6w, s,; 6w;
(14) The total error, e,, in assuming the damage to be at posi-
tion r is the sum of the errors in all the mode pairs. Thus,
e,= e,ij
allpairs i . j

expression would be To aid interpretation, this error is normalized with

respect to the minimum value of the total error. The final
6K z RIBT(D* - D)Bl, value output to the damage location chart is
If it is assumed that the damage affects the stiffness by
the same proportion in all directions, then (D* - D ) K D.
If the damage is equivalent to a hole, then D* = 0 and er
equation (14) reduces to The most probable damage site(s) is (are) therefore that
6K z - I,(Br DB),A dz (those) where E, = 100.
(15)
This expression may be readily evaluated using the Hole
routines which calculate the stiffness matrix in the I
dynamic analysis. Equation (1 1) may then be used to out-
put the sensitivities of the natural frequencies to damage at
any point of the structure.
The results of the sensitivity analysis when the damage
has been taken to be the equivalent of removing a whole
element have been compared with those obtained by
setting the stiffness of the element to zero and repeating
the dynamic analysis. Excellent agreement was obtained,
so the sensitivity analysis method was adopted for all the
tests reported here.

2.4 Presentation of the results of the location scheme

2 2; 4 5 ; 57 1001 100 57; 5 4; 2 2
Measurement of the frequency changes in one pair of I I I I I
modes will yield a locus of possible damage sites, that is, 2 21 4 41 28 1 8 1 IS 201 4 41 2 2
points where the ratio of the experimentally determined
I 2; 4 5 ; 6 6 ; 6 6 ; 5 4; 2 I
changes equal the theoretical ratio. The loci for several I I I I I
pairs of modes may be superimposed, the actual damage 2 21 2 31 3 31 3 31 3 21 2 2
site being given by the intersection of the curves. With
symmetrical structures, two or more sites will be pre- - - - - Finite-elementmesh
dicted, the number depending on the degree of geometric Plate boundaries
symmetry and the elastic symmetry of the structural
material. At least two mode pairs must be used in order to Fig. 1. Location chart for rectangular aluminium plate with hole

51
JOURNAL OF STRAIN ANALYSIS VOL 14 NO 2 1979 QIMechE 1979
Downloaded from sdj.sagepub.com at SAGE Publications on December 6, 2011
 P. CAWLEY AND R. D. ADAMS
It remains to decide how many grid points should be
used. The finite element used in the work described here
gives a quadratic displacement variation across it, which
allows for a linear stress variation. The effect of damage at
a point in the structure is dependent on the stress at that
point, so it would be wasteful of computer time to use
more than two grid points along one side of the element.
Values between the points could be computed by inter-
polation. The results presented here have been computed
using a 6 x 6 finite-element mesh with two grid points per
element side, giving 144 points in all. The accuracy
obtained has not warranted the use of interpolation
between the points.
An example of the chart output is given in Fig. 1, the
hole shown being the actual damage in this case. The
dotted lines show the finite-element mesh. In this case, four
possible sites are indicated because of symmetry.
2.5 Problem of directional damage
The model of damage used in the preceding sections has
assumed that the local stiffness of the structure is reduced
by the same proportion in all directions. This would be the
case with a hole in an isotropic material, but the assump-
tion is no longer valid with a crack or with a hole in an
anisotropic material.
Consider the situation shown in Fig. 2. The crack
shown will have negligible effect on the frequency of mode
B, since there is little stress across it, whereas a hole at that
site would have a significant effect. O n the other hand, the
crack would have a relatively large effect on modes A and
C. The damage location scheme described in the previous
section would, therefore, be expected to give erroneous
results with mode pairs A and B or B and C. O n the other
hand, modes A and C should give good results, since the
stress direction at the crack is similar in the two modes.
The damage location scheme could therefore be improved
by giving high weight to modes in which the directions of
the stress vectors at the point of interest are similar, and
low weight to modes where there is low similarity. This is
readily achieved by weighting the error according to the
scalar product of the normalized stress vectors for the two
modes. Thus, equation (17) becomes
e, = e," Wrij
all palrs 1.1
where
?IFIF1
Displacement nodal lines
_ _ _ _ -
----
I I
Crack
Fig. 2. Illustration of the problem of directional damage
52
Downloaded from sdj.sagepub.com at SAGE Publications on December 6, 2011
Thus, for the example shown in Fig. 2, W,,, ",0; W,,, z
1.
The values of stresses at the grid points may be com-
puted during the dynamic analysis using standard tech-
niques (4). The values of the stresses obtained depend on
the position through the thickness at which they are calcu-
lated. In flexural vibration, the direct stresses would be
zero at the midplane. The stresses are therefore calculated
at a point away from the midplane, the precise position
being immaterial, since only the direction of the stress
vector is required, not the magnitude. The transverse shear
stresses were sufficiently small to be neglected in all the
structures tested. Most of the results presented in this
paper were obtained on structures fabricated from layered
materials. If the stress-weighting procedure described
above is to be used with these materials, it is necessary to
compute a n average stress vector. This is because the
direction of the stress vector at a given position through
the thickness depends on the orientation of the layer. A
mean value may be computed by defining an average elas-
ticity matrix for the ply orientations used. The derivation
of this matrix is described in Appendix 1.
In practice, in order to reduce the amount of data
required to be output from the dynamic analysis and
hence to save on the amount of disc space relquired, the
stresses were computed at the mid-points of the elements.
These values were then assigned to all the grid points
within the element. This procedure is sufficiently accurate
provided the stress directions d o not change rapidly with
position. (Rapid change of magnitude does not matter as
the stresses are normalized.) This is the case with the lower
modes of vibration, such as those used in this investi-
gation.
The stress-weighting procedure has been adopted in all
the results presented here. The results obtained without
weighting were fairly similar to those presented, parti-
cularly in cases where more modes were used. This is
probably because the results tend to average-out to give
the correct solution. The results with stress-weighting
tended to be better defined and there was a marked
improvement in some cases where only a few rnodes were
used.
2.6 Problem of erroneous frequency measurements
Occasionally, it was found that the results using the
damage location scheme described in the preceding sec-
tions were incorrect but that the errors were rectified if the
results were computed without using the readings from
one mode. This could be due to an error in thr: frequency
measurements, the appearance of a double resonance
which was not noticed, a relatively large change in mode
shape due to the damage (the location scheme assumes
that the mode shape is essentially unchanged) or the model
of damage not being good enough to predict the frequency
change accurately. A method was therefore sought to
automate the procedure of checking whether one mode
was causing the final result to be incorrect.
The technique developed uses the fact that if one mode
is causing errors in the final results while the osther modes
give correct results, then removing this mode will tend to
make the damage site more clearly defined on the location
chart. The sum of the numbers on the location chart will
therefore be reduced. The modified location scheme com-
putes the results with all the modes included and with each
J O U R N A L OF STRAIN ANALYSIS VOL 14 NO 2 1979 @IMechE 1979
 THE LOCATION OF DEFECTS IN STRUCTURES FROM MEASUREMENTS OF NATURAL FREQUENCIES
mode neglected,in turn. The final solution is taken as the
one with the lowest sum of numbers on the chart. If less
than five modes are measured, then all the readings must
be used.
2.7 Estimation of the magnitude of damage
The sensitivity analysis uses equations (11) and (15) to
output the frequency change in each mode due to a hole of
area A centred on each of the grid points in turn. Compari-
son of the measured frequency change in a given mode
with that given by the sensitivity analysis for a hole of area
A at the predicted damage site therefore gives an indica-
tion of the severity of the damage in terms of the size of a
hole equivalent to the damage. The result varies consider-
ably between the modes, particularly if the damage is
directional, so an average value is taken.
The method gives a rough indication of the magnitude
of the damage, but is clearly not accurate. The calculation
is performed at the grid point where the error is a mini-
mum, but the damaged area may be between grid points.
This can introduce a substantial error into the calculation.
The model of damage used does not take into account the
mass of material removed by a hole. This is reasonable
with most types of damage, since damage does not usually
reduce the mass of the structure. An error is, however,
introduced in the case of a hole. The sensitivity analysis is
probably only accurate for relatively small changes in the
local stiffness, since it does not take account of the effect
of stress concentrations around the damage. This means
that the effect of a hole is underestimated, resulting in the
area predicted from the measured frequency changes
being larger than the actual size of the hole. Nevertheless,
the method gives a useful indication of the severity of the
damage. The results show that the predicted severity
always increases as the amount of damage is increased.
2.8 Final damage location scheme
The method implemented for these tests involved per-
forming the dynamic analysis, the sensitivity analysis and
the stress calculations on the University ICL 4-70
computer, the results being output to disc or tape. The
time taken by this program was approximately 50 per cent
higher than that required for a single dynamic analysis, so
the saving in computer time over repeating the dynamic
analysis with each element damaged in turn was very
large.
These results, together with the experimental measure-
ments, were then used as input to the program which com-
puted the relative errors and produced the location chart.
This program could be readily adapted to run on a micro-
computer. The .major computational effort is in per-
forming the original analysis. However, this need only be
performed once for a given type of structure.
3 TEST PROCEDURE
The basic requirement of the method is to determine repro-
ducibly several natural frequencies of the structure under
consideration. It does not matter whether the boundary
conditions are exactly the same as those assumed in the
dynamic analysis provided that they are reproducible and
the mode shapes are not greatly altered from those pre-
dicted by the theoretical model. Since this criterion is
JOURNAL OF STRAIN ANALYSIS VOL 14 NO 2 1919 QlMechE 1979
Downloaded from sdj.sagepub.com at SAGE Publications on December 6, 2011
easily met, a variety of methods of exciting and detecting
the vibrational response may be used. The best method to
use in industrial applications would probably be the tran-
sient test, in which the response of the structure to an
impulse is recorded and the natural frequencies computed
by Fourier transform techniques. Unfortunately, the
necessary equipment was not available when these tests
were carried out, so the results presented here were
obtained using steady-state measurements, which are
slower and more difficult to perform. It is hoped to report
some results obtained using the transient test later.
The system used in this investigation was identical to
that described in (5). The structures were suspended on
nylon thread passed through small holes near the edge of
the structure. Excitation was provided by a moving-coil
loudspeaker fed by a high-stability, variable-frequency
oscillator and a power amplifier. Resonance was located
by tuning to the maximum output of a piezoelectric strain-
gauge cemented to the structure at a point where the
stresses across it were predicted to be non-zero in the
modes of interest.
The tests were carried out in a constant-temperature
enclosure, which kept the structure under test at 25 &
1OC. This was found to be sufficiently stable to enable
accurate frequency changes to be recorded. With metallic
structures, it was found to be possible to dispense with the
constant-temperature enclosure and to use a correction
factor based on the temperature coefficient of Young’s
modulus, as described in (2). This would be much more
difficult with composite materials, since the different
moduli have different temperature coefficients and the
coefficients of expansion would be direction-dependent.
Probably the easiest way to determine the correction
factors with composite materials would be to measure
them experimentally from the variation of the natural
frequency with temperature. The signal from the oscillator
was stable to & 0.01 Hz and resonance could be found to
within 0.1 Hz with most of the modes used. With metallic
structures, the resolution was typically better than 0.05
Hz.
The damage location scheme requires that the mode
shapes of the modes used be similar to those predicted by
the dynamic analysis. Some symmetrical structures have
two or more modes at the same frequency, the mode
excited in practice being a combination of the two depend-
ing on the position and nature of the excitation. In these
cases, it is impossible to determine accurately the mode
shape. Modes of this type were therefore not used in the
location scheme.
The mode shapes of the structures tested were checked
roughly by determining the nodal patterns. The structure
was placed on sponge-rubber blocks and excited by an
electromechanical shaker. Dry sand was spread lightly
over the surface of the structure. When the system was
tuned to resonance, the sand particles migrated to the
nodal lines, giving a check on the theoretical predictions of
their position.
The test procedure used was to measure the initial
frequencies of the structure, remove it from the constant-
temperature enclosure, apply the damage and return it to
the enclosure. After allowing time for the temperature of
the structure to stabilize, the frequencies were again
measured. The frequencies quoted in Tables 1 and 2 are
the averages of two or more readings.
53
 P. CAWLEY AND R. D. ADAMS

Undamaged frequencies (Hz)

Area removed
Damage 125.05 158.94 278.29 301.52 361.44 466.98 (percentage

Frequency reductions (Hz) Actual Predicted

Rectangular 0.41 0.77 1.18 0.37 0.72 0.81 0.31 0.47

hole

-
Undamaged frequencies (Hz)
Predicted area.
Damage 104.32 260.98 312.74 415.29 584.48 693.24 794.70 936.50 1062.25 1150.93 removed with respect
to virgin condition
Frequency reductions from virgin condition (Hz) (percentage of total)
-
First saw cut 0.06 0.39 0.29 0.14 -0.07 0.14 2.70 0.62 0.20 0.28 0.19
Secondsawcut 0.85 5.46 4.50 2.08 1.78 4.32 5.29 3.70 1.20 1.68 2.17
Crushdamage 1.24 5.66 5.91 3.44 3.93 4.38 9.02 6.90 10.41 6.51 3.65
-

4 RESULTS Saw cut

The tests reported here were of a preliminary nature and
were on simple, two-dimensional structures; these tests
were carried out in order to test the accuracy of the pro-
posed technique. The results from tests on more compli-
cated structures with forms of damage which are more
representative of those likely to be encountered in service
will be presented later (9).
The first test was carried out on an aluminium plate of
dimensions 450 x 350 x 6 mm with damage produced by
cutting a rectangular hole at the position shown in Fig. 1,
which also shows the location chart produced from the
frequency measurements set out in Table 1. It can be seen
that the damage was successfully located within the con-
straints imposed by symmetry. The discrepancy between
the actual area removed and that predicted by the loca-
tion routine is probably due to the factors referred to in
Section 2.1.
The second set of tests was performed on a trapezoidal,
cross-ply, carbon-fibre-reinforced plastic (CFRP) plate,
which was cut from a square plate of side 250 mm and
thickness 2.8 mm, fabricated from HT-S carbon fibre in
DX 210 epoxy-resin. This plate was damaged with saw
cuts, the first being 2.5 cm long at the position shown in
y 4
- - - - Finite-element mesh
Plate boundaries

Fig. 3. The location chart shown in Fig. 3 relates to this

case. Since this structure was asymmetrical, the damage
site was uniquely defined on the location chart. The
second cut was of a similar size but was approximately at
right angles to the first and at the same site. The location
chart for damage comprising the two cuts together is
shown in Fig. 4.
If the technique described here were to be used as an in-
service test with measurements being made periodically, it
would be important to be able to detect and locate changes
in the condition of the structure, not only with respect to
the virgin condition but also with respect to a stage where
some damage was present. In order to test the method
under these conditions, a location chart was produced
using the frequency changes between the first and second
sets of damage. This is shown in Fig. 5 . It can be seen that
the damage is again successfully located.
Fig. 3. Location chart for CFRP plate with single saw cut
The plate was then damaged again at the site shown in
Fig. 6, by supporting the plate on a thick-walled steel tube
and pressing a 25-mm diameter steel ball into the surface
of the plate at the centre-line of the tube using a hydraulic
press. The location chart shown in Fig. 6 was produced
from the frequency changes caused by the cru:shing with
the ball-bearing alone, that is, using the frequencies
obtained with the two saw cuts as the initial condition.
The model of damage used assumes that the damage is
at one site only. It would be difficult to amend the
technique to enable the location of damage at two or more
sites, and this case would probably be unusual i n practice.
The location scheme was tested with the case of damage at
two sites by using the changes between the virgin fre-

54 JOURNAL OF STRAIN ANALYSIS VOL 14 NO 2 1979 OlMechE 1979

Downloaded from sdj.sagepub.com at SAGE Publications on December 6, 2011
 THE LOCATION OF DEFECTS IN STRUCTURES FROM MEASUREMENTS OF NATURAL FREQUENCIES
Saw cuts
/- I
- - - Finite-element mesh
Plate boundaries
Fig. 4. Location chart for CFRP plate with t w o saw cuts; frequency
changes computed with respect to virgin frequencies
Second saw cut
/ First saw cut
/ /
/ _ _ _ _ Finite-element mesh
-Plate boundaries
Fig. 5. Location chart for CFRP plate with two saw cuts;
frequencies after first cut taken as 'undamaged' values
quencies and those measured after the crush damage had
been applied. The location chart for this case is shown in
Fig. 7. As expected, erroneous results were produced. The
presence of damage was, however, still detected from the
frequency changes.
The frequency measurements and the equivalent areas
removed as predicted by the location routine for this series
J O U R N A L O F S T R A I N A N A L Y S I S VOL 14 N O 2 1979 O l M e c h E 1979
Downloaded from sdj.sagepub.com at SAGE Publications on December 6, 2011
Saw cuts
Fig. 6. Location chart for CFRP plate with t w o saw cuts and crush
damage; frequencies after both saw cuts taken as 'undamaged'
values
Saw cuts
Fig. 7. Location chart for CFRP plate with t w o saw cuts and
crush damage; virgin frequencies taken as 'undamaged' values
of tests are set out in Table 2. It can be seen that the
frequency changes produced by the first saw cut are much
smaller than those produced by the second, in spite of the
fact that the cuts were of similar length. The predicted
equivalent area removed was 0.19 per cent for the first saw
cut and 2.17 per cent after the second had been made. It
would appear to be difficult to reconcile the fact that two
55
 P. CAWLEY AND R. D. ADAMS

cuts of similar length produce such disparate areas of APPENDIX 1

damage. The explanation is that the routine which esti-
mates the severity of the damage assumes that the damage DERIVATION OF MEAN ELASTICITY MATRIX FOR A
is in the form of a hole. The stress at the position of the LAYERED MATERIAL
saw cuts in the modes of interest was mainly in a direc-
The strain energy of a structure at any instant is given by
tion parallel to the first cut, which was close to, and
parallel to, the plate edge. Thus, a small hole at that site S = J&'E dV (2 1)
would produce the same frequency changes as this cut. and stress and strain are related according to the equation
The second cut, however, was approximately per-
pendicular to the directions of the major stresses, so a 6 = DE (22)
much larger hole would be required to produce an equi- Thus, the strain energy is given by
valent effect, that is, a relatively large frequency reduc-
tion. The value of the equivalent area removed produced ,
S = $ 1 E ~ DdEV (23)
by the analysis of Section 2.7 can therefore only be It is required to derive a mean elasticity matrix for a struc-
expected to give an idea of the order of the severity of a ture fabricated from n layers so that an average stress
defect unless the damage has the same effect on the stiff- vector may be found. This may be done by specifying that
ness of the structure in all directions. The analysis also the strain energy of the structure computed using the mean
assumes that the damage is at a single site defined by the elasticity matrix according to equation (23) is the same as
grid point at which the location chart shows a value of that obtained by summing the strain energies of the indivi-
100. In the case where the saw cuts and crush damage are dual layers. That is,
combined and compared with the virgin condition, n
erroneous results must therefore be expected from the jJ,eTDeEdV=$ 1 J,ETDkedV (24)
routine, both in terms of defect location and size. It should k=l

be noted that the case given in Table 2 for crush damage
corresponds to Fig. 7 and not to Fig. 6. The predicted
equivalent area removed corresponding to the location
chart shown in Fig. 6, which relates to the case where the
frequencies after the crush damage were compared with
those after the application of the saw cuts, was 1.02 per
cent of the plate area. The actual area of the tube used in
producing this damage was approximately 0.5 per cent of
the area of the plate, and it is fair to assume that the actual
damaged area would spread somewhat beyond this. Thus,
the predicted damaged area of 1.02 per cent was of a
similar order, bearing in mind the assumptions and
simplifications used, to the actual damaged area.
One of the modes tested showed a frequency increase
after the first saw cut. This was probably due to an error in
the initial frequency measurement, so this mode was not
used in the damage location scheme.
5 CONCLUSIONS
The test results show that the proposed method can be
used to detect, locate and roughly quantify damage in
structures. The technique is applicable to all systems
which are amenable to finite-element analysis. The final
scheme has the advantage that only one dynamic finite-
element analysis need be performed on a given type of
structure. The results of the dynamic analysis may be
stored on disc or tape and used as input to the damage
location program, along with the experimentally-deter-
mined natural frequencies. This program could be easily
run on a microcomputer.
The method requires access to only one point of the
structure and, if the measurement of the natural fre-
quencies were carried out using transient techniques, the
test time would be very short. The presence of damage is
where De is the mean elasticity matrix and Dk is the elasti-
city matrix of the kth layer, the integral on the right-hand
side of the equation being taken over the volume of the
layer and that on the left being over the volume of the
structure. Cartesian co-ordinates have been used in the
work described here and, in this case, equati.on (24)
becomes
n
111&'Dee d x dy dz = 111E'D~Ed x dy dz (25)
k=l
The elasticity matrices are not dependent on x and y , so
equation (25) reduces to
n
1,eTDe&dz = 1 E ~ Ddz~ c
k=l
where h, is the z co-ordinate of the top of the kth layer.
Flexural vibration has been used in the tests reported
here, so it is reasonable to assume a linear direct-strain
variation through the thickness and a parabolic transverse
shear-strain variation. The strain vector for the element
used here is therefore given by
where k,, . . ., k , are functions of x and y.
Substituting equation (27) into equation (26) and equa-
ting similar terms gives

i
indicated immediately from changes in the natural fre- 15 8
quencies without the need for any computation.
It is possible to use the method to monitor the growth of
0; = -
8t
1Dfj((hk
k=l
- hk- --
3t 2
(hi - h i - (28)

damage without the need to have measured the fre-

16
quencies of the virgin structure by using a damaged state + -(hi - hi- ,)
as the baseline for future measurements. 5P

56 JOURNAL OF STRAIN ANALYSIS VOL 14 NO 2 1979 OlblechE 1979

Downloaded from sdj.sagepub.com at SAGE Publications on December 6, 2011
 THE LOCATION O F DEFECTS IN STRUCTURES FROM MEASUREMENTS OF NATURAL FREQUENCIES

The C F R P structure used here had a symmetrical ply (3) BISHOP, R. E. D. and JOHNSON, D. C. The mechanics of vibra-
configuration about the midplane. This means that fion 1960 (Cambridge University Press).
(4) ZIENKIEWICZ, 0. C. The finite element method in engineering
De - D e - D e - D ; , = D e - D e -0 science 197 I (McGraw-Hill).
13- 14- 23- 3s - 4s -
(5) CAWLEY. P. and ADAMS. R. D. ‘The theoretical and experi-
The average elasticity matrix, De, defined by equation mental natural modes of free-free C F R P plates’, Jnl of Composite
(28) has been used to compute the stress vectors for the M t l ~1978 12,336-347.
C F R P plate used in these tests. (6) AHMAD, S., IRONS, B. M. and ZIENKIEWICZ, 0. C. ‘Analysis
of thick and thin shell structures by curved finite elements’, In!. J.
Numerical Methods in Engng 1970 2,419.
APPENDIX 2 (7) ZIENKIEWICZ, 0. C., TAYLOR, R. L. and TOO, J. M.
‘Reduced integration technique in general analysis of plates and
shells’, Int. J. Numerical Methods in Engng 1911 3, 275.
REFERENCES
(8) COURANT, R. and HILBERT, D. Methods of mathematical
ADAMS, R. D., WALTON. D.. FLITCROFT, J. E. and SHORT. phvsics 1953 1 (Interscience).
D. ‘Vibration testing as a non-destructive test tool for composite (9) CAWLEY, P. and ADAMS, R. D. ‘A vibration technique for
materials’, Composite Reliability, ASTM STP 580 1975, 159-175 non-destructive testing of fibre composite structures’, Jnl of
(Amer. SOC. for Testing and Mtls). Composite Mtls 1978 (to be published).
ADAMS. R. D.. CAWLEY, P., PYE. C. J. and STONE, B. J. ‘A
vibration technique for non-destructively assessing the integrity of
structures’, J. mech. Engng Sci. 1978 20 (No. 2), 93-100.

57
J O U R N A L OF S T R A I N A N A L Y S I S VOL 14 N O 2 1 Y 7 Y OIMecht 1979
Downloaded from sdj.sagepub.com at SAGE Publications on December 6, 2011

View publication stats