Engineering Structures 30 (2008) 34973504

Contents lists available at ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

Damage detection by means of structural damping identification

R.O. Curadelli a, , J.D. Riera b , D. Ambrosini a , M.G. Amani a
a

Laboratory of Experimental Dynamic, National University of Cuyo, CONICET, Mendoza, Argentina

Laboratory of Structural Dynamic and Reliability, Federal University of Rio Grande do Sul Porto Alegre, Brazil

article

info

Article history:
Received 12 November 2007
Received in revised form
18 May 2008
Accepted 19 May 2008
Available online 2 July 2008
Keywords:
Structural damage
Dynamic properties identification
Instantaneous frequency
Instantaneous damping coefficient
Wavelet transform

a b s t r a c t
Vibration measurements in structures under ambient or controlled excitation open a way to damage
assessment by correlating changes in the dynamic system parameters natural frequencies, modal
shapes or damping with damage. Most methods proposed in the literature are based on the
measurement of natural frequencies or modal shapes, associating damage to local reduction of the
structural stiffness while completely ignoring damping, both in the identification procedure as well
as in the correlation of changes in the spectral properties of the structure with damage. In numerous
applications these methods, however, present some practical limitations on account of the low sensitivity
of natural frequencies or modal shapes to damage. The main objective of this paper is to present a novel
scheme to detect structural damage by means of the instantaneous damping coefficient identification
using a wavelet transform. Laboratory tests as well as numerical simulations showed that in many
commonly used structural systems damage causes important changes in damping. Thus, parameters that
characterize structural damping can be used as damage-sensitive system properties. It is further shown
that the wavelet transform can be conveniently employed in a procedure for instantaneous frequencies
and damping identification from free vibration structural response.
2008 Elsevier Ltd. All rights reserved.

1. Introduction
Structural monitoring and damage detection are areas of current interest in civil, mechanical and aerospace engineering. Most
damage detection methods are visual or localized experimental
procedures such as acoustic or ultrasonic methods, magnetic field
methods, radiographs, eddy-current and thermal field methods. All
these experimental techniques require that the location of damage be known a priori and that the portion of the structure under
inspection be easily accessible. These limitations led to the development of global monitoring techniques based on changes in the
vibration characteristics of the structure.
Damage or fault detection by monitoring changes in the
dynamic properties or response of the structure has received
considerable attention in recent literature. The basic idea springs
from the notion that spectral properties, described in terms of
the so-called modal parameters (frequencies, mode shapes, and
modal damping), are functions of the physical properties of the
structure (mass, energy dissipation mechanisms and stiffness).

Corresponding author.
E-mail addresses: ocuradelli@fing.uncu.edu.ar (R.O. Curadelli),
riera@cpgec.ufrgs.br (J.D. Riera).
URLs: http://www.fing.uncu.edu.ar (R.O. Curadelli), http://www.ufrgs.br
(J.D. Riera).
0141-0296/$ see front matter 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.engstruct.2008.05.024

Therefore, changes in the physical properties will cause changes in

the modal properties. Literature reviews on damage identification
and health monitoring of structures based on changes in their
measured dynamic properties were published, for example, by
Doebling et al. [1,2], Zou et al. [3], Sohn et al. [4], Salawu [5] and
Chang et al. [6]. Moreover, in large civil engineering structures
it is usually impossible to measure all components of the input
excitation. Experimental modal identification through ambient
vibration response records, in which only the structural output is
monitored, has consequently become a very attractive alternative.
Structural damage usually results in a decrease in mass,
stiffness and strength of structural elements. In consequence, its
detection and localization have great practical importance and led
to the rapid growth of a research field rather incorrectly known as
Structural Health Monitoring (SHM) [7], although what is actually
monitored is the structural integrity and it would sound awkward
to qualify any structure as healthy. At any rate, four SHM categorical
levels were proposed to quantify damage in engineering structures
[8]: Level 1: damage detection; Level 2: damage location; Level 3:
quantification of the degree of damage; Level 4: estimation
of the remaining service life. For SHM purposes, perhaps the
most frequently used techniques are vibration-based methods.
Vibration methods are based on the fact that damage in a structure
usually implies a local stiffness reduction or increase in flexibility,
which induce changes in its dynamic properties. These changes can
be used for damage identification.

3498

R.O. Curadelli et al. / Engineering Structures 30 (2008) 34973504

The amount of literature describing damage detection schemes

based on shifts in the natural frequencies is quite large. The
forward problem, which usually falls into the category of Level 1
damage identification, consists of calculating frequency shifts for a
known type of damage. Typically, damage is modelled numerically
by a reduction in the stiffness and then the measured frequencies
are compared to the frequencies predicted by the model. Moreover,
this technique is used extensively in damage diagnosis and health
monitoring of existing highway bridges, seismic behaviour and
residual capacity of structures.
It should be noted that frequency shifts have practical
limitations in several applications, especially in case of large
structures. Farrar et al. [9,10] conducted tests on the I-40 Bridge
and verified that for a reduction in the bending stiffness of the
overall bridge cross section equal to 21%, no significant reduction in
the modal frequencies was observed. Chen et al. [11] investigated
changes in natural frequencies on steel space structures subjected
to atmospheric corrosion showing that atmospheric corrosion does
not perceptibly affect the natural frequencies of the structure.
Several references cited in literature reviews by Doebling et al.
[1], Salawu [5] and Sohn et al. [4] stress that the usually small
frequency shifts caused by damage require either very precise
measurements or high levels of damage. Ongoing research may,
however, provide ways to circumvent these difficulties.
Damping properties, on the other hand, have seldom been
used for damage diagnosis. Crack detection in a structure based
on damping, however, has advantages over detection schemes
based on frequencies and modal shapes. In fact, when cracks
result in small or no frequency variations, damping changes may
render the detection of the nonlinear, dissipative effects that
cracks produce. Modena et al. [12] show that visually undetectable
cracks cause very little change in resonant frequencies and
require higher mode shapes to be detected, while these same
cracks cause larger changes in damping. In some cases, damping
changes of around 50% were observed. Their study focuses on
identifying manufacturing defects causing structural damage in
precast reinforced concrete elements. The particular dynamic
response of reinforced concrete justifies the use of damping as
a damage-sensitive property, proposing two new methods based
on changes in damping to detect cracking. These techniques are
employed to locate cracks in a 1.20 5.80 m precast hollow floor
panel excited with stepped sins and shocks, and the diagnosis
results are compared with those of frequency and curvature-based
approaches.
Kawiecki [13] measures damping in a 90 20 1 mm metal
beam and metal blanks used to fabricate 3.5-inch computer hard
disks, noting that damping can be a useful damage-sensitive
indicator. Modal damping is determined using the half-power
bandwidth method applied to the FRF within the frequency range
of 59 kHz obtained from measured data. The author claims that
the approach should be particularly suitable for structural health
monitoring of lightweight structures and microstructures.
After conducting vibration tests on prestressed reinforced
concrete hollow panels, Zonta et al. [14] observe that cracking
produces a frequency splitting in the frequency domain and the
beat phenomenon of the free decay signals in the time domain. It
is argued that crack formation in prestressed reinforced concrete
causes a no viscous dissipative mechanism, making damping more
sensitive to damage, proposing thereby to resort to this dispersive
phenomenon for damage diagnosis. Zonta et al. [14] point out
that the dispersive phenomenon cannot be represented by the
standard linear model of a single degree of freedom system and
emphasize that the oscillator has a variable stiffness, and this
variability is proportional to the frequency. They recognize the
need for further research using the dispersion phenomenon for
damage detection because additional effects such as hysteresis and

other non-classical dissipative mechanisms must be considered. It

is important to note that these difficulties to accurately identify
modal damping led to a scattering higher than the frequency, as it
is showed below.
Recent advances in signal processing based on the Hilbert
Huang transform [15] and wavelet analysis opened a new perspective on the development of the system identification methodologies
[1619] and on SHM [2022].
In this paper a new scheme to detect structural damage employing the instantaneous damping coefficient identification using
the wavelet transform is presented. Experimental and simulated examples confirm that, in the structures analyzed, damage causes
important variation of the coefficient that characterizes damping, which proves therefore to be a generally damage-sensitive
system property. Because the wavelet estimation technique requires free decay response observations, which is rarely feasible
on large structures, the modal identification herein proposed is indeed an appealing technique. Ambient excitations such as traffic
or wind have the advantage that no equipment is needed to excite
the structure. To convert ambient random response to free decay
responses the well established Random Decrement Technique was
used [38,39,18].
2. Identification of instantaneous undamped natural frequency and damping coefficient
Although various complex phenomena occur when different
materials/structural elements are damaged, principally under
seismic excitation, only two of those phenomena are commonly
associated with damage: stiffness and strength degradation. Hence
most models describing the hysteretic behaviour of structural
members have been derived from the shape of experimental
forcedisplacement relationship (hysteretic curves). Examples of
macro models proposed to represent the complex behaviour of
reinforced concrete beams or beam-columns are Takeda-type [23]
and their variations [24,25], which simulate phenomena such as
stiffness and strength degradation and pinching. An oscillator
with such behaviour can be modelled by second-order systems
with nonlinear restoring force k(x)x and nonlinear damping force
ho (x) x and represented by the equation x + 2ho (x) x + k (x) x = 0,
which leads to a free response of the form:
x(t ) = A(t ) cos((t )).

(1)

Note that functions ho and k represent apparent (viscous)

damping and apparent (elastic) stiffness coefficients, respectively.
In the following they will be referred to simply as damping and
stiffness coefficients.
Feldman [26,27] showed, by applying the multiplication
property of the Hilbert Transform for overlapping functions to
the equation of motion for viscously damped systems, that the
instantaneous undamped natural frequency and the instantaneous
damping coefficient may be calculated according to the formulae:

o2 (t ) = 2

A
A

2A 2
A2

,
(2)

ho (t ) =
A
2
in which o (t ) is the instantaneous undamped natural frequency
and ho (t ) the instantaneous damping coefficient of the system.
A(t ) and (t ) are the envelope (instantaneous amplitude) and the
instantaneous damped angular frequency of the vibratory system


solution with their first and second time derivatives ,
A , A .
The envelope A(t ) and the instantaneous phase (t ) can be
A

extracted from the vibration signals employing the continuous

wavelet transform [28,29], as follows:

R.O. Curadelli et al. / Engineering Structures 30 (2008) 34973504

According to analytic signal theory, the analytic signal xa (t ) =

x(t ) + jH (x(t )) of a large number of processes including the
vibration of a under-critically-damped system described by Eq. (1),
where it is assumed that the phase varies much faster than the
amplitude (asymptotic signal), can be approximated by a complex
signal (Torresani [30]):
xa (t ) = A(t )e

j(t )

(3)

in which A(t ) and (t ) were defined before and H (x(t )) is the

Hilbert transform of x(t ):
H [x (t )] =

x ( )

1
t

d .

(4)

The continuous wavelet transform of an asymptotic signal x(t ) is

defined as (Chui [31]):
+
1
Wg [x] (a, b) =
x (t ) g
a

t b


dt

(5)

where g (t ) is the analyzing or mother wavelet and g (t ) the

complex conjugate of g (t ), any function g (t ) can be used as
an analyzing wavelet if it satisfies the admissibility condition
(Chui [31]); b is the parameter localizing the wavelet function in
the time domain: Wg [x](a, b) shows the local information about
x(t ) at the time t = b (translation) and a is the dilatation
or scale parameter defining the analyzing window stretching.
The idea of the continuous wavelet transform is to decompose a
function x(t ) into wavelet coefficients Wg [x](a, b) using the basis
of wavelet functions. The wavelet coefficients represent a measure
of the similitude between the dilated and shifted wavelet and the
function x(t ) at time b and scale a.
In terms of the analytic signal the continuous wavelet transform
of an asymptotic signal x(t ) is:
Wg [x] (a, b) =

2 a

A ( t ) ej t g

t b
a

dt .

(6)

Developing, the Eq. (6) by Taylors formula for t = b, using the

Fourier transform G (a) of g [(t b)/a] and neglecting terms of
superior order:

Wg [x] (a, b) =

A(b)ejb G (a) .

(7)

From Eq. (3) the concept of instantaneous angular frequency as

the time varying derivative of the phase can be derived: (t ) =
(t ).
Using the complex Morlet wavelet in the frequency domain,
1
2
G (a) = e 2 (ac ) as analyzing function; Eq. (7) can be written
as:

Wg [x] (a, b) =

b)c )
A(b)ej(b) e 2 (a(

(8)

in which c is the wavelet central frequency.

The square of the modulus of the continuous wavelet transform
can be interpreted as an energy density distribution over
the timescale plane. The energy of the signal is essentially
concentrated on the timescale plane around a region called the
ridge of the continuous wavelet transform. In other words, the ridge
of the continuous wavelet transform is the region containing the
points defined by a = a(b), where the amplitude of the continuous
wavelet transform is maximum. The ridges are identified by
seeking out the points where the continuous wavelet transform
coefficients take on local maximum values: for each value of b, we
obtain a value as |Wg [x] (a(b), b) | = maxa Wg [x] (a, b). To obtain
the ridge, the dilatation parameter a = a(b) has to be calculated in
order to maximize the analyzing wavelet G (a(
b)), that is, using

3499

the complex Morlet wavelet, for a = a(b) = c /(

b). It can be
obtained:

Wg [x] (a(b), b) =

a (b)
2

A(b)ej(b) .

(9)

The values of the continuous wavelet transform that are

restricted to the ridge are the skeleton of the continuous wavelet
transform.
It is important to note that the real components of the continuous wavelet transform along the ridge are directly proportional to
the signal given by Eq. (1) and from Eq. (9) we obtain
A(b) = 2

|Wg [x](a(b), b)|

a(b)

(b) = Arg Wg [x] (a(b), b) .

(10)
(11)

Finally, using the Eqs. (10) and (11) it is possible to estimate

the instantaneous amplitude A(t ) and the phase (t ). Being the
derivative of A(t ), A (t ) and the derivative of (t ), the instantaneous
damped angular frequency (t ) = (
t ), in conjunction with
the Eq. (2), the instantaneous undamped natural frequency and
damping of the system can be determined.
This procedure establishes an identification technique to determine the undamped natural frequency and damping coefficient of
the system as instantaneous functions of time at every point of the
vibration process.
Employing this method, the evolution of the system parameters
with increasing damage will be analyzed in numerical and
experimental examples.
3. Changes in frequency and damping with damage
In nonlinear systems (nonlinear dissipative and elastic forces)
the damping coefficient and the natural frequency become functions of the vibration amplitude. With the aim of determining the
evolution of the dynamic system parameters (undamped natural
frequency and damping coefficient) with vibration amplitude and
increasing damage, three examples were analyzed.
3.1. Numerical simulation of the response of 2D reinforced concrete
frame with strength and stiffness degradation
In order to assess changes in the system parameters caused
by damage, in this section, a 2D reinforced concrete frame and
modelled with inelastic elements described by Takeda-based [23
25] constitutive rules capable of representing hysteretic stiffness
degradation and pinching was analyzed using the Drain 2D [32]
software. The example is a three bays, six-stories high reinforced
concrete frame, designed for a peak ground acceleration (PGA)
of 0.17 g (value that presents a 10% probability of exceedance
in 50 years, in accordance with the provisions of the Argentine
Code INPRES-CIRSOC 103 [33] or ACI, which in this example result
in similar designs (Curadelli and Riera [34]). The total mass per
floor is 100 t, Youngs modulus of concrete E = 24.8 GPa which
lead to a fundamental period T1 = 1 s. The internal damping
ratio was assumed equal to 5% of critical. In the nonlinear
constitutive relations, the yield strength of reinforcing steel and
the compressive strength of concrete were assumed equal to fy =
413 MPa and fc = 27.6 MPa, respectively. Fig. 1 shows the basic
dimensions of the frame.
The structure was subjected to the Caucete, San Juan, Argentina
1997 seismic acceleration time history. From the last 10 s of the
roof response, which is actually a free vibrations record, changes
in the system parameters were determined by the procedure
described above. In order to assess different degrees of structural

3500

R.O. Curadelli et al. / Engineering Structures 30 (2008) 34973504

Fig. 3. Instantaneous undamped fundamental frequency.

Lowest excitation level (PGA = 1.0 m/s2 , undamaged),
highest excitation level (PGA =
4 m/s2 , severe damage).

Fig. 1. Reinforced concrete 2D frame structure.

damage, the record was normalized to four intensity levels [1, 2, 3

and 4 m/s2 (collapse)], defined in terms of their PGA.
Fig. 2(a) and (b) show the free response and the corresponding
envelopes, determined by the identification technique, for the first
and ultimate degrees of damage.
Figs. 3 and 4 illustrate the dependence of the instantaneous
fundamental frequency and instantaneous damping coefficient on
the amplitude of vibration, respectively, for each damage level.
Figs. 3 and 4 provide valuable evidence for the influence of damage on the dynamic properties of the reinforced concrete frame.
The maximum undamped fundamental frequency reduction, observed for a PGA = 0.4 g, reaches nearly 25%, increasing linearly
with the PGA. Damping, on the other hand, presents more pronounced and typical variations.
3.2. Experimental study of a reinforced concrete beam
The performance of the identification procedure was also
assessed using experimental data, for which purpose a reinforced
concrete beam tested by Palazzo [35] was analyzed. The specimen
was a 0.20 m 0.10 m 5.60 m reinforced concrete beam in
flexure under two points loading (Fig. 5(a) and (b)). The yield

Fig. 4. Instantaneous damping coefficient.

Lowest excitation level (PGA =
1.0 m/s2 , undamaged),
highest excitation level (PGA = 4 m/s2 , severe
damage).

strength of reinforcing steel bars and the compressive strength of

concrete were fy = 420 MPa and fc = 17 MPa, respectively.
Four KYOWA AS-GB accelerometers (accuracy 100 mV/g) were
used to measure the dynamic response (vertical acceleration at
different points) of the structure, sampled at 500 Hz each channel.
The signals generated by the accelerometers were amplified by
a dynamic strain amplifier KYOWA DPM-612B and recorded by

Fig. 2. Free vibration response and envelope. (a) PGA = 1 m/s2 and (b) PGA = 4 m/s2 .

R.O. Curadelli et al. / Engineering Structures 30 (2008) 34973504

3501

Fig. 5. Sketch of reinforced concrete simply supported beam tested (b) load location.
Table 1
Load step values (from Palazzo [35])
Load
step

Load (kN)

Fundamental frequency measured after each load step

0
1
2
3
4
5

(undamaged)
2 1.46
2 2.25
2 3.23
2 5.19
2 7.25a

4.52
4.34
4.29
4.25
4.01
3.63

a
Maximum load considered (correspond to 3.50 /00 , maximum compress strain,
INPRES-CIRSOC 201 [41]).

Fig. 6. Reinforced concrete beam test setup.

a Computerboars PCM-DAS16D/16 data acquisition board. For

details, see Palazzo [35]. Loads were increased monotonically until
the intensities indicated in Table 1 were reached, as schematically
shown in Fig. 6. At that point loads were removed and free vibration
tests conducted to determine dynamic properties. The procedure
was then repeated for the next loading step. Fig. 7(ab) shows the
observed free vibration response of the undamaged state and after
the last loading step, respectively.
The correlations between the instantaneous fundamental frequency and instantaneous damping coefficient and the amplitude
of vibration, for the healthy and damaged condition (after Load
Step 5), are presented in Figs. 8 and 9, respectively. The scattering of parameters for ten samples in each state represented by one
standard deviation to both sides of mean values are indicated in
dashed line.

Figs. 8 and 9 show the instantaneous modal parameters along

time for two different levels of damage. It is clear that damping
has an important variation.
3.3. Experimental 3D frame model
As a second experimental example, the one bay, six-storeys
high aluminium 3D frame model tested by Amani et al. [36] was
analyzed (Fig. 10). Geometric properties are indicated in Table 2
(more details are reported in Amani [37]. Structural damage was
caused by subjecting the model to a horizontal unidirectional base
motion in a shaking table. The excitation consisted of a series of
nine simulated acceleration time histories (Gaussian white noise)
with increasing intensity. The standard deviations of the base
acceleration in the tests were a = 1.8; 2.1; 6.1; 12; 20; 32;
38; 44; and 48 m/s2 . As it was mentioned before, the wavelet
transform procedure for modal parameters identification operates
on the free vibration. Thus, in order to obtain free decay response,

Fig. 7. Free vibration response and his envelope: (a) for healthy condition and (b) after load step 5 test. (Palazzo [26]).

3502

R.O. Curadelli et al. / Engineering Structures 30 (2008) 34973504

Fig. 8. Instantaneous undamped fundamental frequency.

Fig. 10. One bay, six-storeys high aluminum 3D frame model tested by Amani et al.
[37].

every example is similar to those obtained by harmonic linearization of a bilinear oscillator [40]; after reaching a maximum value
for a certain value of amplitude, a smooth decrease takes place.
4. Conclusions

Fig. 9. Instantaneous damping coefficient.

Table 2
Geometrical and mechanical properties of the model
Total height
Story height
Span length
Columns cross section

Total mass
Mass density (columns)
Youngs modulus

Stories 1&2
Stories 3&4
Stories 5&6

0.50 m
0.083 m
0.10 m
0.6 18 (mm)
0.6 12 (mm)
0.6 6 (mm)
0.665 kg
2698 kg/m3
67 GPa

the measured horizontal acceleration response at the top floor was

processed by random decrement technique [38,39,18].
Fig. 11(ab) show free vibration normalized acceleration
response at the top floor obtained from random decrement
technique for the lowest and highest excitation level, respectively.
Figs. 11 and 12 show for both, lowest excitation level ( a =
1.8 m/s2 , undamaged) and highest excitation level ( a =
48 m/s2 , severe damage), the dependence of instantaneous
fundamental frequency and instantaneous damping coefficient
on the instantaneous amplitude of vibrations, respectively. The
scattering is indicated by one standard deviation to both sides of
the mean values.
Similarly, Figs. 12 and 13 show that increasing excitation yield
to a damage level reducing the undamped fundamental frequency
by about 9%, while damping reveals more marked variations.
It is important to point out that the relationship between
instantaneous damping coefficient and vibration amplitude for

For the kind of structures analyzed, a damage detection technique using the instantaneous damping coefficient as sensitivedamage feature, has shown that it is potentially useful. It seems
quite clear that, in systems as analyzed, damping is a promising
damage indicator in structural health monitoring because it has
more sensibility to damage than the natural frequency. As shown
on Figs. 8 and 9 for reinforced concrete beam and 12 and 13 for
the experimental model, the statistical variations of damping are
higher than the natural frequencies, however, the results are not
masked. The paper describes an approach, based on a wavelet transform, to detect structural damage by means of the instantaneous
damping coefficient identification.
With examples using experimental results and numerical
simulation on structures subjected to seismic base excitation,
it was demonstrated that the identification technique based on
wavelets is useful to assess changes in the vibration characteristics
due to incremental damage of nonlinear systems.
Since the wavelet transform is a signal decomposition procedure
working as a filter in the timefrequency domain defined by the
localizing parameter and dilatation parameter, it is possible to
analyse a signal only locally. Thus, a multidegrees of freedom
system can be decoupled into single degrees of freedom. For a fixed
value of the dilatation parameter (a = ai ), which maximises the
amplitude of the continuous wavelet transform, only the mode
associated with ai gives a relevant contribution, while all the other
terms are negligible. Hence, it is possible to choose the value of the
dilatation parameter ai corresponding to the analysed mode and
by means of Eqs. (10) and (11) to determine the amplitude and the
phase variations in the time domain of this mode.
It is also stressed that this work is a start point for damage
identification based on damping evaluation, and further research
is needed in the subject.
Acknowledgements
The authors wish to thank the financial aid of CONICET and
SECyT, Argentina and CNPq and CAPES, Brazil.

R.O. Curadelli et al. / Engineering Structures 30 (2008) 34973504

3503

Fig. 11. Free vibration normalized acceleration response and the envelope for: (a) lowest and (b) highest excitation level. One bay, six-storey high aluminium 3D frame
model tested by Amani et al. [37].

Fig. 12. Instantaneous undamped fundamental frequency.

Lowest
excitation level ( a = 1.8 m/s2 , undamaged),
highest excitation level
( a = 48 m/s2 , severe damage).

Fig. 13. Instantaneous damping coefficient.

Lowest excitation level ( a =
1.8 m/s2 , undamaged),
highest excitation level ( a = 48 m/s2 , severe
damage).

References
[1] Doebling SW, Farrar Ch, Prime MB, Shevitz DW. Damage identification and
health monitoring of structural and mechanical systems from changes in their
vibration characteristics: A Literature Review. Los Alamos National Laboratory
Report. LA-13070-MS. UC900; 1996.

[2] Doebling SW, Farrar CR, Prime MB. A summary review of vibration-based
damage identification methods. Shock and Vibration Digest 1998;30:91105.
[3] Zou Y, Tong L, Steven GP. Vibration-based model-dependent damage
(delamination) identification and health monitoring for composite structures
a review. J Sound Vibration 2000;230(1):35778.
[4] Sohn H, Farrar CR, Hemez FM, Shunk DD, Stinemates DW, Nadler BR. A review
of structural health monitoring literature: 19962001, Los Alamos National
Laboratory, USA; 2003.
[5] Salawu OS. Detection of structural damage through changes in frequency: A
review. Eng Struct 1997;19(7):71823.
[6] Chang PC, Flatau A, Liu SC. Review Paper: Health monitoring of civil
infrastructure. Struct Health Monit 2003;2(3):25767.
[7] Chang FK. Structural health monitoring 2000. In: Proceedings of the Second
International Workshop on SHM. 1999.
[8] Rytter A. Vibration based inspection of civil engineering structures. Ph.D.
Thesis. Aalborg University. Denmark; 1993.
[9] Farrar CR, Baker WE, Bell TM, Cone KM, Darling TW, Duffey TA. et al. Dynamic
characterization and damage detection in the I-40 Bridge over the Rio Grande.
Los Alamos National Laboratory report LA-12767-MS; 1994.
[10] Farrar CR, Jauregui DA. Comparative study of damage identification algorithms
applied to a bridge: I. experiment. Smart Mater Struct 1998;7(5):70419.
[11] Chen B, Xu YL, Qu WL. Evaluation of atmospheric corrosion damage to steel
space structures in coastal areas. Int J Solids Struct 2005;42:467394.
[12] Modena C, Sonda D, Zonta D. Damage localization in reinforced concrete structures by using damping measurements, Damage assessment of structures. In:
Proceedings of the international conference on damage assessment of structures, DAMAS 99. 1999, p. 132141.
[13] Kawiecki G. Modal damping measurements for damage detection. European
COST F3 conference on system identification and structural health monitoring.
Madrid, Spain; 2000, p. 651658.
[14] Zonta D, Modena C, Bursi OS. Analysis of dispersive phenomena in damaged
structures. European COST F3 conference on system identification and
structural health monitoring. Madrid, Spain; 2000, p. 801810.
[15] Zhou L, Yang G. HHT method for system identification and damage detection:
An experimental study. Smart Struct Syst 2006;2(2):14154.
[16] Staszewski WJ. Identification of damping in MDOF systems using timescale
decomposition. J Sound Vibration 1997;203(2):283305.
[17] Staszewski WJ. Identification of non-linear systems using multi-scale ridges
and skeletons of the wavelet transform. J Sound Vibration 1998;214(4):
63958.
[18] Meo M, Zumpano G, Xiaolin Meng, Cosser E, Roberts G, Dodson A.
Measurements of dynamic properties of a medium span suspension bridge by
using the wavelet transforms. Mech Syst Signal Process 2006;20:111233.
[19] Minh-Nghi Ta, Lardis J. Identification of weak nonlinearities on damping and
stiffness by the continuous wavelet transform. J Sound Vibration 2006;293:
1637.
[20] Ovanesova A, Surez L. Applications of wavelet transforms to damage
detection in frame structures. Eng Struct 2004;26:3949.
[21] Haase M, Widjajakusuma J. Damage identification based on ridges and maxima
lines of the wavelet transform. Int J Eng Sci 2003;41:142343.
[22] Rucka M, Wilde K. Application of continuous wavelet transform in vibration
based damage detection method for beams and plates. J Sound Vibration 2006;
297(35):53650.
[23] Takeda T, Sozen M, Nielsen N. Reinforced concrete response to simulated
earthquakes. J Struct Div ASCE 1970;96: ST12.
[24] Tang X, Goel SC. DRAIN-2DM Technical notes and users guide, Research
Report UMCE 88-1. Department of Civil Engineering, University of Michigan
at Ann Arbor, MI; 1988.
[25] Hueste M, Wight J. Nonlinear punching shear failure model for interior slabcolumn connections. J Struct Eng, ASCE 1999;125(9):9971008.
[26] Feldman M. Non-linear free vibration identification via the Hilbert transforms.
J Sound Vibration 1997;208(3):47589.

3504

R.O. Curadelli et al. / Engineering Structures 30 (2008) 34973504

[27] Feldman M. Non-linear free vibration analysis using Hilbert transform I. Free
vibration analysis method freevib. Mech Syst Process 1994;8(2):11927.
[28] Minh-Nghi Ta, Lardis Joseph. Identification of weak nonlinearities on
damping and stiffness by the continuous wavelet transform. J Sound Vibration
2006;293:1637.
[29] Kim H, Melhem H. Damage detection of structures by wavelet analysis. Eng
Struct 2004;26:34762.
[30] Torresani B. Analyse Continue par Ondelettes, CNRS Editions, Paris, 1995.
[31] Chui CK. Wavelets analysis and applications. In: An introduction to wavelets.
New York: Academic Press; 1992.
[32] Kanaan A, Powel G. DRAIN-2D, a general-purpose computer program for
dynamic analysis of inelastic plane structures. Technical Report UCB/EERC 7306. University of California. Berkeley, CA, USA; 1973.
[33] INPRES-CIRSOC 103 Part II. Argentinean code for reinforced concrete
earthquake resistant constructions. 2000 [in Spanish].
[34] Curadelli R, Riera J. Reliability based assessment of the effectiveness of metallic
dampers in buildings under seismic excitations. Eng Struct 2004;26:19318.

[35] Palazzo G. Systems identification and damage structural assessment, Master

Dissertation National University of Tucumn, Argentina. 2000 [in Spanish].
[36] Amani MG, Riera J, Curadelli O. Identification of changes in the stiffness and
damping matrices of linear structures through ambient vibrations. Structural
Control and Health Monitoring 2007;14(8):115569.
[37] Amani MG. System identification and structural damage assessment. Ph.D.
thesis National University of Tucumn, Argentina. 2004 [in Spanish].
[38] Ibrahim SR. Incipient failure detection from random decrement time
functions. Int J Anal Exp Modal Anal 1986;1:18.
[39] Ibrahim SR, Brincker R, Asmussen J. Modal parameter identification from
responses of general unknown random inputs. In: Proceedings of the 14th
IMAC; 1996.
[40] Curadelli RO. Structural vibration control by metallic dampers. Ph.D. thesis,
Federal University of Rio Grande do Sul, Brazil. 2003 [in Portuguese].
[41] INPRES-CIRSOC 201. Argentinean code for reinforced concrete structures.
2000 [in Spanish].