STRUCTURAL INTEGRITY ANALYSIS OF BRIDGES EVALUATED

FROM OPERATIONAL VIBRATIONS

Tiago A. P. Souza
Marcelo Greco
[email protected]
[email protected]
Graduate Program in Structural Engineering, Federal University of Minas Gerais.
Av. Presidente Antônio Carlos, 6627, Zip-Code: 31270-901 / Escola de Engenharia / Bloco 1/
Sala 4127 / Belo Horizonte, Minas Gerais, Brasil.
Cláudio J. Martins
[email protected]
Graduate Program in Civil Engineering, Federal Center of Tech. Education of Minas Gerais.
Av. Amazonas, 7675, Zip-Code: 30510-000 / Escola de Engenharia / Prédio 18 / Belo
Horizonte, Minas Gerais, Brasil.

Abstract. This paper presents the methodology involved in the structural integrity analysis of
two bridges located in Brazil, from the excitations of operation (traffic, wind). The main
objective of this paper is to evaluate the structural integrity in terms of load tests by
numerical models, calibrated from the experimental parameters. The first bridge was built in
1983, whereas the second was built fourteen years later. For the measurement of vibrations in
each structure it was considered the natural actions as an excitation source, without
interruption of the traffic and without the use of impact equipment. Thus, a tri-axial
accelerometer with high sensitivity was positioned at various points on both bridges to
capture the natural frequencies of each mode of vibration. The spectra obtained were
processed by algorithm that provided the first natural frequencies and their structural
damping. Then, the numerical models were made in the structural analysis software followed
by the calibration step which consisted of the stiffness changing and the binding of certain
elements in the numerical models, until occur equivalence of natural frequencies with
experimental results. Thus, a numerical simulation was performed in terms of moving loads
by the numerical models calibrated, in order to assess the extent of displacement of the center
of each bridges platform. Therefore, the limits showed that both numerical models showed
displacement values below the standard established by the Brazilian codes, and no damage
that can affect the safety of users.
Keywords: Modal Analysis, Stochastic Subspace Identification, Structural Integrity

CILAMCE 2014
Proceedings of the XXXV Iberian Latin-American Congress on Computational Methods in Engineering
Evandro Parente Jr (Editor), ABMEC, Fortaleza, CE, Brazil, November 23-26, 2014
Template for CILAMCE 2014 (double-click to edit short title field)

1 INTRODUCTION
The development of technology addressed in this work is driven by its ability to provide
quick and effective solutions to real engineering problems with respect to the vibration
analysis of structures. Along with the development of numerical techniques, operational
modal analysis has emerged as a tool to assist in the solution vibrational problems and in the
analysis of structural integrity. This concept can be reasoned as a process of review of the
structure in terms of their modal characteristics, namely the natural frequencies, damping
ratios and mode shapes (Soeiro, 2001). The conception assumes that the vibration response of
a linear dynamic system (time-invariant) can be expressed as a linear combination of a set of
simple harmonic motion, called modes of vibration. These being natural, are specific to each
structural system and determined by their physical properties (mass, stiffness and damping)
and its spatial geometry (bindings and restrictions).
The process of applying controlled loads and subsequent measurement of vibration at
certain points of a structure with the purpose of identifying the dynamic properties (natural
frequencies, mode shapes and damping ratios), is called Experimental Modal Analysis. This
technique is initiated within the mechanical engineering, where the structures are small in
comparison with the civil, considering the monitoring of loads applied, either by vibrating or
impact hammers (Maia et al., 1998 and Martins et al., 2011).
The application in civil structures, from artificial excitations , presents a number of
difficulties, namely: the interruption of the traffic, the high cost of excitation equipment, the
risk of damage on site and the possibility of stimulating vibration modes less significant.
Because of such difficulties, since 1990 there is a new alternative on this process of analysis,
based only on measuring the response of the structure due to its operational activities (vehicle
traffic, wind, temperature changes, rain, peoples). Due to the difficult to obtain accurate
measured data, these excitation forces are idealized by a stochastic process of Gaussian white
noise (constant spectral density) and having zero mean (Rodrigues, 2004). For this analysis
technique and modal identification, in which the structures were excited from operations
actions (mainly traffic and wind), was designated as the theme of this paper: Structural
Integrity Analysis of Bridges evaluated from Operational Vibrations.

2 MATHEMATICAL PRINCIPLES
The matrix equations are based on Finite Element Method (FEM) and they govern the
dynamic behavior of structures as well as their representation of state. These methods
originated in systems and control engineering, science in which it is common to use the state
representation for the analysis of dynamic systems (He & Fu, 2001 and Ribeiro, 2002). The
method of stochastic modal identification, called Stochastic Subspace Identification, is used
to analyze the information obtained from the trials of dynamic response measuring of the
front structure to operational actions.

CILAMCE 2014
Proceedings of the XXXV Iberian Latin-American Congress on Computational Methods in Engineering
Evandro Parente Jr (Editor), ABMEC, Fortaleza, CE, Brazil, November 23-26, 2014
 F. Author, S. Author, T. Author (double-click to edit author field)

2.1 Equations of dynamic equilibrium

In the Finite Element Method, the domain of the physical problem is discretized by a
number of finite elements interconnected by its nodes and faces. The field variables, in turn,
are approximated by their nodal values, resulting in a discrete system of ordinary differential
equations given by, described in References (Zienkiewicz & Taylor, 1989 and Hughes, 2000):

  DU
MU   KU  F (1)

where M, D and K represent, respectively, the overall arrays of mass, damping and
stiffness, associated with vectors and correspond to the nodal displacement and load. The
superscripted dot above the variables indicates temporal derivative.
The determination of the behavior of the structure is performed under free vibration
condition to determine the modal shapes and the natural frequencies, through the solution of
the problem of eigenvalues and eigenvectors. Neglecting the effect of damping in the
structure, such problem can be expressed as:

(K   2j M)U j  0 (2)

where  j is the natural frequency of order j and U j is the vector of displacements

corresponding to the jth vibration mode.

2.2 State representation for continuous deterministic system

The system governed by second order partial differential equations can be described by a
state representation, given a vector as follows:

U 
X 
U 
(3)


The equation of dynamic equilibrium given by Eq. (1) can be expressed by:

  AX  BF
X
Y  CX
(4)

Where C corresponds to the observation of the system matrix and Y the vector of
observed responses, A and B corresponds respectively to the matrices of state and input, and
is given by:

 0 
A
 M D
I
 M K
1 1

 0 
(5)
B   1 
M 

CILAMCE 2014
Proceedings of the XXXV Iberian Latin-American Congress on Computational Methods in Engineering
Evandro Parente Jr (Editor), ABMEC, Fortaleza, CE, Brazil, November 23-26, 2014
Template for CILAMCE 2014 (double-click to edit short title field)

2.3 State representation for discrete stochastic system

The representation of the dynamic system shown in Eq. (4) states that the input vector F
is known and does not correspond to reality when are considered only system responses in
analysis. Due to operational actions corresponding to stochastic signals, the state
representation for discrete time can be given by:

k 1  AX k  Wk

X
Yk  CX k  Vk
(6)

Where the index k indicates a certain discrete-time analysis and Wk and Vk correspond
to the vectors of processed noise and measurement, respectively, idealized by the Gaussian
stochastic process using white noise with zero mean, described in Maia et al.(1998). Figure 1
shows a block diagram of the system indicated by the Eq. (6).

Figure 1. Blocks diagram

2.4 Stochastic subspace identification

Since Wk and Vk vectors correspond to stationary Gaussian processes, and observing
that the operations present in Eq. (6) are linear, one can assert that the state vectors and
response, and also feature distribution Gaussian zero mean as described in Andersen (1997).
Then, it was defined the following relations:

E[X k i X Tk ]  Σi (7)

E[Yk i YkT ]  Λi (8)

E[X k 1YkT ]  G (9)

W 
E  k  W  Q
VkT    T
S
R 
  Vk   S
T
k (10)

Where Σi and Λ i correspond to the covariance matrices of the state and response and
G corresponds to the covariance matrix of the state at time k +1 and response time k. The
matrices Q , S and R correspond to the covariance matrices of the vectors Wk and Vk .
Considering the Equation (6) to (10), the following relationships can be obtained:

CILAMCE 2014
Proceedings of the XXXV Iberian Latin-American Congress on Computational Methods in Engineering
Evandro Parente Jr (Editor), ABMEC, Fortaleza, CE, Brazil, November 23-26, 2014
 F. Author, S. Author, T. Author (double-click to edit author field)

Σ 0  AΣ 0 AT  Q (11)

G  AΣ0CT  S (12)

CΣ CT  R ,i  0
Λ i   0i 1
CA G ,i  0
(13)

Considering the correlation functions of the observed responses organized into Hankel
matrix H0 containing p blocks of lines, and observing the relationship shown in Eq. (13),
obtain the following expression:

 Λ0 Λ i 1   CA CAi 1G 
 Λ
Λ1
  
 CAG 
 1 Λ2  Λ i   CAG

CA2G  CAi G 
     
   
     
 Λ 2 p 1 Λ2 p  Λ 2 p i 1  CA2 p 1G CA2 p G  CA2 p i 1G 

or

H0  OΓ (14)

Where O and Γ are named, respectively, matrices observalidade and controllability of

the system. Such matrices are expressed by:

 C 

 
 CA 
O  and Γ  G A G  Ai 1G
  
(15)
 2 p 1 
CA 

Considering the decomposition of singular value matrix H0, obtain:

H 0  USV (16)

Therefore, the matrices O and Γ can be expressed as the matrices resulting from
decomposition, as follows:

O  US 2 and Γ  S 2 V T
1 1
(17)

From arrays observability and controllability of the system, C and G can be obtained
directly from the first rows and columns of O and Γ , respectively (see Eq. (15)).
The determination of the state matrix A can be performed considering a Hankel matrix
whose columns are H1 early a time interval relative to H0. Considering Equation (14) for H1,
we obtain:

H1  OAΓ (18)

Thus:

CILAMCE 2014
Proceedings of the XXXV Iberian Latin-American Congress on Computational Methods in Engineering
Evandro Parente Jr (Editor), ABMEC, Fortaleza, CE, Brazil, November 23-26, 2014
Template for CILAMCE 2014 (double-click to edit short title field)

A  S 2 UT H1VS
1  12
(19)

singular values of A ( A  ΨμΨ ). The natural frequencies i , damping ratios  i and mode
Finally, the modal parameters are determined by performing the decomposition into

shapes φ i are then given by:

ln(  i )
i  , i  i
t
(20)

Re(i )
i 
i
(21)

φi  CΨ i (22)

3 DESCRIPTION AND NUMERICAL MODELING OF BRIDGES

The structures were evaluated bridges of Mutuca’s mountains, located in Minas Gerais,
Brazil. As shown in Fig. 2, the bridge A, located at left, was built in 1983 and B, located at
right, was completed fourteen years later (in 1997).

Figure 2. Bridges A and B, respectively

CILAMCE 2014
Proceedings of the XXXV Iberian Latin-American Congress on Computational Methods in Engineering
Evandro Parente Jr (Editor), ABMEC, Fortaleza, CE, Brazil, November 23-26, 2014
 F. Author, S. Author, T. Author (double-click to edit author field)

3.1 Structural Typology

The superstructures of the bridges are composed by prestressed beams with concrete
boxes varying through the parabolic sections (heights between 2.2 to 6.2 m, Fig. 3), and
columns of reinforced concrete. The infrastructure consists of concrete blocks and caissons.
The bridge A has a length of 492 meters, and B bridge, 464m, both with 15.10 m wide. For
more details, see Cardoso & Rêgo (1979).

Figure 3. Cross section on span at left, and on the buttress at right

3.2 Numerical Modeling

Numerical models of the bridges were developed in a structural analysis program
SAP2000 (CSI, 2009) based on the Finite Element Method and in accordance with the
structural design. It is three-dimensional model (formed by quadratic bar elements and thin
shell) assuming homogeneous material and isotropic linear elastic behavior. It were
considered the mass of asphalt (15 cm thick) and the load of wheel guard (estimated at 580
kgf/m). The prospects of numerical models of each bridge are shown in Figs. 4 and 5.

Figure 4. Numerical model of the bridge A

CILAMCE 2014
Proceedings of the XXXV Iberian Latin-American Congress on Computational Methods in Engineering
Evandro Parente Jr (Editor), ABMEC, Fortaleza, CE, Brazil, November 23-26, 2014
Template for CILAMCE 2014 (double-click to edit short title field)

Figure 5. Numerical model of the bridge B

4 EXPERIMENTAL RESULTS

4.1 Equipment description and measurement sites

The equipment used in vibration measurements is called PIAS (Portable Integrity
Assessment System), shown in Fig. 6. This equipment allows measurements at low
frequencies and with great precision. The acceleration is achieved in three axial directions and
due to its high sensitivity it is possible to obtain natural vibrations of the structure by means
of its external actions (traffic, wind, temperature change). Therefore it was not necessary to
induce artificially impact.

Figure 6. Equipment for measuring vibration

The locations of accelerometer installation along the platform were defined in order to
avoid zero displacement points (pivot nodes), according to Reference NBR15307 (2006). The
bridge A 24 measuring points were adopted while on the bridge B, 22. The measurement
period at each position was 2 hours.

CILAMCE 2014
Proceedings of the XXXV Iberian Latin-American Congress on Computational Methods in Engineering
Evandro Parente Jr (Editor), ABMEC, Fortaleza, CE, Brazil, November 23-26, 2014
 F. Author, S. Author, T. Author (double-click to edit author field)

4.2 Spectra measurement and modal parameters

The spectra obtained for the measurement points considered in the platforms of the
bridges A and B, Figs. 7 and 8, were processed by the technique of stochastic ID subspaces
which provided the major natural frequencies and damping their experimental, Table 1 and
Table 2 .

Figure 7. Vibration spectrum (vertical direction of the bridge A)

Figure 8. Vibration spectrum (vertical direction of the bridge B)

CILAMCE 2014
Proceedings of the XXXV Iberian Latin-American Congress on Computational Methods in Engineering
Evandro Parente Jr (Editor), ABMEC, Fortaleza, CE, Brazil, November 23-26, 2014
Template for CILAMCE 2014 (double-click to edit short title field)

Table 1. Experimental frequencies and damping - Bridge A

Shape Frequencies (Hz) Damping (%) Considered shapes

1º 1,55 6,14 1º
2º 2,48 0,19 2º
3º 2,49 3,51 3º
4º 2,50 0,54 4º
5º 2,68 2,59 5º
6º 2,70 0,79 6º
7º 3,19 3,16 7º
8º 3,26 1,90 8º
9º 3,63 2,94 9º
10º 3,67 2,01 -
11º 6,96 1,09 -
12º 7,67 0,94 -
Mean 3,56 2,15

Table 2. Experimental frequencies and damping - Bridge B

Shape Frequencies (Hz) Damping (%) Considered shapes

1º 1,28 2,19 1º
2º 1,40 1,45 2º
3º 1,47 2,12 3º
4º 1,48 0,49 4º
5º 2,34 6,24 5º
6º 2,35 0,67 6º
7º 2,54 0,56 7º
8º 2,55 3,31 8º
9º 3,16 2,10 9º
10º 3,19 5,04 10º
11º 3,43 1,61 11º
12º 4,17 1,49 12º
Mean 2,45 2,27

5 NUMERICAL RESULTS

5.1 Calibration procedure

The calibration procedure was obtained from the change of stiffness and linkages of
elements by equivalence between the frequencies of the model to the experimental result.
Procedure which is based on Eurocode 8 (2009), as follows:
• The vibration modes that contribute significantly to the overall response of the
structure were considered.

CILAMCE 2014
Proceedings of the XXXV Iberian Latin-American Congress on Computational Methods in Engineering
Evandro Parente Jr (Editor), ABMEC, Fortaleza, CE, Brazil, November 23-26, 2014
 F. Author, S. Author, T. Author (double-click to edit author field)

• The total number of modes according to the total mobilized mass, considered as the
sum of modal masses effective in each X, Y and Z direction being so, has become the 1st
satisfied criterion when the sum of the accumulated mass to modes considered resulted in at
least 90% of the total mass deployed, as shown in Figs. 9 and 10.

Figure 9. Total mass mobilized (Numerical model A)

Figure 10. Total mass mobilized (Numerical model B)

The results of the calibration procedure are shown in Fig. 11 and Fig. 12, which represent
the equivalence between the experimental frequencies and the numerical frequencies for each
mode of vibration. As the stiffness expected by prestressing system was not observed in the
numerical models, an equivalent elasticity for the platform was established, in order to adjust
the numeric data to experimental data.

CILAMCE 2014
Proceedings of the XXXV Iberian Latin-American Congress on Computational Methods in Engineering
Evandro Parente Jr (Editor), ABMEC, Fortaleza, CE, Brazil, November 23-26, 2014
Template for CILAMCE 2014 (double-click to edit short title field)

Figure 11. Results of the calibration process (numerical model A)

Figure 12. Results of the calibration process (numerical model B)

5.2 Vibration modes

The mass factor mobilized accumulated, Figs. 9 and 10, was used to define the maximum
number of vibration modes to be treated, since the numerical model The mobilized 90% of its
accumulated mass (in the X, Y and Z) on the 9th mode vibration, whereas the model B, in the
12th mode. The comparison of the extracted vibration modes of numerical models A and B
are presented below:

Figure 13. 1st mode of vibration the bridges A and B, respectively

Figure 14. 2nd mode of vibration the bridges A and B, respectively

CILAMCE 2014
Proceedings of the XXXV Iberian Latin-American Congress on Computational Methods in Engineering
Evandro Parente Jr (Editor), ABMEC, Fortaleza, CE, Brazil, November 23-26, 2014
 F. Author, S. Author, T. Author (double-click to edit author field)

Figure 15. 3rd mode of vibration the bridges A and B, respectively

Figure 16. 4th mode of vibration the bridges A and B, respectively

Figure 17. 5th mode of vibration the bridges A and B, respectively

Figure 18. 6th mode of vibration the bridges A and B, respectively

Figure 19. 7th mode of vibration the bridges A and B, respectively

Figure 20. 8th vibration mode of the bridges A and B, respectively

Figure 21. 9th mode of vibration the bridges A and B, respectively

CILAMCE 2014
Proceedings of the XXXV Iberian Latin-American Congress on Computational Methods in Engineering
Evandro Parente Jr (Editor), ABMEC, Fortaleza, CE, Brazil, November 23-26, 2014
Template for CILAMCE 2014 (double-click to edit short title field)

5.3 Dynamic analysis

After calibration of the numerical models, the dynamics analysis of the structures was
performed in order to verify the structural integrity of bridges, from the using of actions
established by Brazilian standards.
According to the purposes of this study, the analysis focused on the verification of the
serviceability limit states (deformation of the platform that can cause discomfort in people),
by obtaining the maximum displacements at pre-selected points. Values were obtained
through of combinations actions of NBR 8681 (2003), ie own weight of each structure,
mobile cargo (type train of 45 ton and 0.5 ton of load evenly distributed on the platform) and
actions due to wind, where it was considered the situation more unfavorable to the safety of
the structure (NBR 7188, 1982), see figure 22. The points considered for obtaining the
maximum displacements at each calibrated numerical model were those located in center of
the spans, among which stood the accelerometer during the experimental test. These values
were compared with the limits of spans of the structure for sensory acceptability as
established by the Brazilian Standard (NBR 6118, 2003), see Tables 3 and 4.
Although the Brazilian standard NBR7188 (1982) specify that the impact should be
considered in the design of structural elements of road bridges, this study did not consider the
impact of moving load. The structural analysis by means of impact mobile load is purpose of
future work which will assess the transient behavior when considering effects of damage,
imperfection (geometric nonlinearity) and coupling (impact) of the moving masses in the
structural system (see Oliveira & Greco, 2014), according to criteria of ultimate limit states.

Figure 22. Numerical simulation with moving loads (type Train T45 and uniformly distributed load)

CILAMCE 2014
Proceedings of the XXXV Iberian Latin-American Congress on Computational Methods in Engineering
Evandro Parente Jr (Editor), ABMEC, Fortaleza, CE, Brazil, November 23-26, 2014
 F. Author, S. Author, T. Author (double-click to edit author field)

Table 3. Maximum displacements obtained from the bridge A platform and regulatory limit

Points Displacements (mm) Limitation of span

(mid-span) Long. (x) Trans. (y) Vert. (z) NBR6118 (mm)
A1 4,03 -0,19 -2,06 62,85
A5 5,30 0,60 -53,29 185,71
A9 1,86 0,21 -32,65 328,57
A13 -0,33 0,37 -43,52 328,57
A17 -3,18 0,87 -60,18 185,71
A20 -1,35 0,75 -11,53 57,14
A22 -1,02 1,02 -19,94 77,14
A24 -0,77 0,39 -11,99 57,14

Table 4. Maximum displacements obtained from the bridge B platform and regulatory limit

Points Displacements (mm) Limitation of span

(mid-span) Long.(x) Trans. (y) Vert. (z) NBR6118 (mm)
B1 1,55 -0,16 -4,50 62,85
B5 2,83 -0,27 -31,20 185,71
B9 0,47 0,56 -100,02 328,57
B13 -2,92 0,38 -90,28 328,57
B17 -4,12 -0,29 -25,32 185,71
B21 -2,82 -0,14 -10,50 105,71

6 CONCLUSIONS
The methodology used in this work allowed through contemporary techniques, to
characterize the dynamic properties of two bridges through its natural vibrations of operation.
The procedures for obtaining vibration signals without inducing artificial impacts or
paralyze the operation of the structure were presented, which represents a great advantage
compared to classical modal analysis, by not using large equipment excitement and accept
their functioning under normal conditions.
The numerical modeling and calibration procedures were also demonstrated, which
consisted of matching the natural frequency of each vibration mode to the experimental
frequencies by changing the stiffness of structural elements and the criteria mobilized mass.
These steps are essential for numerical models that express equivalent to the actual dynamic
behavior of structures.
According to the data of displacements obtained, it can be seen that both numerical
models have lower values than the maximum established by Reference A.B.N.T NBR6118
(2003). Therefore, it is concluded that these structures showed normal behavior and no
damage capable of affecting the security of people at the time of analysis.

CILAMCE 2014
Proceedings of the XXXV Iberian Latin-American Congress on Computational Methods in Engineering
Evandro Parente Jr (Editor), ABMEC, Fortaleza, CE, Brazil, November 23-26, 2014
Template for CILAMCE 2014 (double-click to edit short title field)

ACKNOWLEDGEMENTS
The authors would like to acknowledge CNPq (National Council of Scientific and
Technological Development), CAPES (Coordination of Improvement of Higher Education
Personnel), FAPEMIG (Minas Gerais State Research Foundation) and PROPEEs-UFMG
(Federal University of Minas Gerais) for the financial supports.

REFERENCES
A.B.N.T. NBR6118, 2003. Projeto de estruturas de concreto. Rio de Janeiro.
A.B.N.T. NBR7188, 1982. Carga móvel em ponte rodoviária e passarela de pedestre. RJ.
A.B.N.T. NBR8681, 2003. Ações e segurança nas estruturas. Rio de Janeiro.
A.B.N.T. NBR15307, 2006. Provas de cargas dinâmicas em grandes estruturas. RJ.
Andersen, P., 1997. Identification of Civil Engineering Structures using Vector ARMA
Models, PhD Thesis, Department of Building Technology and Structural Engineering,
University of Aalborg, Denmark.
Eurocode 8, 2009. B.S.I. Design of Structure for Earthquake Resistance – Part 2: Bridges. BS
EN 1998-2:2005+A1:2009. United Kingdom.
Cardoso, J. L., Rêgo, M. J. B., 1979. Projeto de Restauração e Duplicação Rodoviária -
Projeto de Execução de 2 (dois) Viadutos da Mutuca, Ministério dos Transportes,
Departamento Nacional de Estradas de Rodagem. Vol. 3.3. Belo Horizonte, 120 p.
CSI (Computer and Structures Inc.), 2009. Software SAP2000/Bridge. Version 14, 53p.
Berkeley, California, USA.
He, J., Fu, Z., 2001. Modal Analysis, 1st ed., Butterworth- Heinemann, Oxford, England,
291p.
Hughes, T. J. R., 2000. The Finite Element Method, Linear Static And Dynamic Finite
Element Analysis. 2ª ed. New Jersey, Prentice-Hall.
Maia, N., Silva, J. M., He, J., Lieven, N., Lin, N., Lin, R., Skingle, G., To, W. M., Urgueira,
A., 1998. Theoretical and Experimental Modal Analysis, Research Studies Press, London,
England.
Martins, C., Soares, T., Ortigão, A., Bujang, R., Azman, F. N., 2011. Dynamic loading test
by operational modal analysis for the Malaysian 4b SG. Bridge. Experimental Vibration
Analysis for Civil Engineering Structures, Evaces. University of Porto, Portugal.
Oliveira, F. M., Greco, M., 2014. Nonlinear dynamic analysis of beams with layered cross
sections under moving masses. Journal of the Brazilian Society of Mechanical Sciences. DOI
10.1007/s40430-014-0184-5
Ribeiro, M. I., 2002. Análise de Sistemas Lineares, IST Press, Portugal.
Rodrigues, J., 2004. Stochastic Modal Identification - Methods and Applications in Civil Eng.
Structures, Ph.D. Thesis, University of Porto, Portugal.
Soeiro, N. S., 2001. Análise Modal Experimental. Grupo de Vibrações e Acústica, Dep. de
Engenharia Mecânica, Univ. Federal do Pará, 61p.
Zienkiewicz, O. C., Taylor, R. L., 1989. The Finite Element Method, Vol. 1-2, Mc-Graw Hill.

CILAMCE 2014
Proceedings of the XXXV Iberian Latin-American Congress on Computational Methods in Engineering
Evandro Parente Jr (Editor), ABMEC, Fortaleza, CE, Brazil, November 23-26, 2014