Beam vibrations:

Discrete mass and stiffness models

Ana Cláudia Sousa Neves

[email protected]

Instituto Superior Técnico, Universidade de Lisboa, Portugal

May, 2015

Abstract

In the present work the dynamic behavior of several beams with different support conditions, forced or in free
vibration, is studied. Using the Discrete Element Method (DEM), the expressions governing the motion of the
blocks in which the beam is discretized are derived. A MATLAB program that calculates natural frequencies
and mode shapes is developed; the results are then compared with the exact solutions, in order to validate
the models. The program also allows to simulate the evolution of the dynamic systems in time, yielding
displacements, velocities and accelerations. The effect on the beam behavior due to the introduction of one or
more cracks is analyzed; cracks with different sizes and positions are considered. Three distinctive cases for
the studied models are considered: non-existence of cracks, permanently open cracks or “breathing cracks”
(cracks that open and close depending on the curvature sign). The obtained results are shown with the help of
tables and graphics and, when possible, compared with the exact solutions or with numerical or experimental
results found in literature.

Key-words: vibration of beams, rigid blocks, discrete stiffness, breathing crack.

1. Introduction
The method applied in the development of the DEM implemented in MATLAB.
models presented in this paper is the Discrete
Element Method (DEM) (Neild et al., 2001). Using One of the most important criteria to obtain good
this method one can represent the beam as a results using the DEM is the process adopted for the
discrete system of blocks (i.e. with a finite number of beam discretization. It is expectable that, as one
degrees of freedom) where the mass and the refines the mesh, the results tend towards the exact
moment of inertia of each block are lumped in its solutions. However, the number of blocks should not
respective middle point and where rotational and be indefinitely increased as that would lead to an
transverse springs, connecting adjacent blocks, increase of the time expended in the numerical
simulate respectively the bending and shear calculations. There should therefore exist a balance
distortion. Thus, the beam can be seen as a between the time and the effort used in the
sequence of rigid blocks linked by pairs of springs, as calculations and the aimed precision for the results.
observed in Figure 1. The easiness with which the DEM considers cases
where the beams are cracked should be noted.
Assuming the existence of cracks between rigid
blocks, the localized loss of stiffness (coincident with
the crack position) is taken into consideration by
changing the stiffness constants of the springs
Figure 1. Beam model using the Discrete Element Method. connecting the blocks; the amount of reduction of the
stiffness will be dependent on the crack depth
Once the model is defined, the differential equations (Okamura et al., 1969).
governing its evolution in time are established. For
the particular case of a beam, these equations In Section 2 a cantilever beam model is studied and
involve relative rotations and displacements between the system of ordinary differential equations that
blocks, as well as their derivates with respect to time. governs its motion is obtained. The action of an
The dynamic behavior in time of several beam external force on the beam and the existence of a
models is studied and simulated with the help of the crack (which introduces a local stiffness discontinuity

1
in the model) are also considered. The crack is either
assumed to be always open or to behave like a
breathing crack (a crack that opens and closes
accordingly to the sign of the curvature of the cross
section). The time integration of the equations that
govern the motion of a cracked cantilever beam is Figure 4. Discrete elements model of the cantilever beam
performed by the Runge-Kutta method and the time where the stiffness is also discretized in the interfaces
evolution of the beam’s dynamic response is between blocks.
presented and compared with experimental results. In the current study the shear distortion is neglected.
In Section 3, an analogous study is made for a So one only has the rotational springs between
suspended cracked beam submitted to an oscillatory blocks. The system of governing ordinary differential
external force acting perpendicularly to the beam’s equations is (Neves, 2015)
suspension plane and free of support conditions in its
plane of motion. 1
(𝐽 + 𝐿𝐵̲ 𝐴−1 𝑚 𝐴−𝑇 𝐵𝑇 𝐿) 𝜃̰ ̈ + 𝐴 𝐾 𝐴
̲𝑇 𝜃̰ = 0̰ , (1)
4
The last section is dedicated to the conclusions that
result from the performed numerical investigations
and to enumerate some aspects that are worth of a system of ordinary differential equations in terms of
future attention. the 𝑁 − 1 block rotations. The matrices (𝐽 +
1
𝐿𝐵̲ 𝐴−1 𝑚 𝐴−𝑇 𝐵𝑇 𝐿) and 𝐴 𝐾 𝐴
̲ 𝑇 are symmetric. System
4
(1) can also be written as
2. Cantilever beam
−1
1
2.1. Dynamics of a homogeneous 𝜃̰ ̈ = − (𝐽 + 𝐿𝐵̲ 𝐴−1 𝑚 𝐴−𝑇 𝐵𝑇 𝐿) 𝐴 𝐾 𝐴
̲𝑇 𝜃̰ (2)
4
cantilever beam
and shall be complemented with a set of suitable
In this section a cantilever beam of length 𝐿 with initial conditions
uniform rectangular transverse section 𝑏 × ℎ and
mass density 𝜌 (Figure 2) is considered. The beam is 𝜃̰ (0) = 𝜃̰ 0 , 𝜃̰ ̇ (0) = 𝜃̰ ̇ 0 . (3)
decomposed into 𝑁 blocks, as illustrated in Figure 3
One can use system (2) to estimate the exact natural
angular frequencies and the exact mode shapes.
Assuming a solution of the exponential type

𝜃̰ (𝑡) = 𝛩̰𝑒 𝜆𝑡 , (4)

Figure 2. Homogeneous cantilever beam with rectangular the following eigenproblem is obtained
transverse section.
−1
1
which corresponds to the discrete stiffness version [𝜆2 𝐼 + (𝐽 + 𝐿𝐵̲ 𝐴−1 𝑚 𝐴−𝑇 𝐵𝑇 𝐿) 𝐴 𝐾 𝐴
̲𝑇 ] 𝛩̰ = 0̰ , (5)
4
represented in Figure 4, where each pair of
consecutive blocks is connected by a pair of springs
from which it is possible to calculate approximations
(rotational and transverse). The first and last blocks
for frequencies and vibration mode shapes.
have a length that is half the length of the
𝐿 Considering an external concentrated varying force
intermediate blocks, the latter with length 𝑙𝑛 = , acting on the cantilever beam tip and Rayleigh
𝑁−1
mass 𝑚𝑛 = 𝜌𝑏ℎ𝑙 and moment of inertia 𝐽𝑛 = damping, the system of ordinary differential
𝜌𝑏ℎ𝑙𝑛
(ℎ2 + 𝑙𝑛 2 ) around an axis perpendicular to the equations (1) becomes (Neves, 2015)
2
plan of motion. The first block is considered to be
clamped. 1
(𝐽 + 𝐿𝐵̲ 𝐴−1 𝑚 𝐴−𝑇 𝐵𝑇 𝐿) 𝜃̰ ̈ + 𝐶𝜃̰ ̇ + 𝐴 𝐾 𝐴
̲𝑇 𝜃̰ = 𝐹(𝑡)𝑙̰ ′ . (6)
4

It is also important to consider the possibility of

existence of cracks in the span which will lead to a
nonlinear behavior of the beam. The DEM enables to
take into account the existence of one or more
cracks with variable depth. The modifications are
Figure 3. Scheme of the cantilever beam decomposition in
considered in the stiffness matrix as an open crack
blocks.

2
will produce a stiffness reduction in the section where The stiffness of the spring affected by the crack is
it is situated. determined based in (Okamura et al., 1969). If the
cross section has a positive curvature sign, the crack
A cantilever beam with a breathing crack in its upper is considered to be closed and the stiffness of the
part, with height 𝑎 and at a distance 𝑥𝑐 from the correspondent spring, like the other springs, follows
clamped end is modeled (Figure 5). As the crack is the known expression
located in the upper part of the beam, when the
curvature in that section is positive the crack is 𝐸𝐼
assumed to be closed (Figure 6) and there is no 𝑘𝑐 = (7)
𝑙
discontinuity in flexural stiffness; otherwise, if the
where 𝑙 stands for the length of each block or,
curvature in that section is negative the crack is open
equivalently, the length of influence of the 𝑛-th spring.
(Figure 7) and the local stiffness changes. It is
But if the cross section has a negative curvature
assumed that the crack thickness is negligible when
compared to its height and that it does not sign, the crack is considered to be open and the
propagate. stiffness of the correspondent spring is given by
(Okamura et al., 1969)

1
𝑘𝑐 = (8)
𝑙 72 𝑎
+ 𝐹( )
𝐸𝐼 𝐸𝑏ℎ2 ℎ
where

𝑎 𝑎 2 𝑎 3 𝑎 4
𝐹 ( ) = 1.98 ( ) − 3.277 ( ) + 14.43 ( ) −
ℎ ℎ ℎ ℎ
𝑎 5 𝑎 6 𝑎 7 (9)
31.26 ( ) + 63.56 ( ) − 103.36 ( ) +
ℎ ℎ ℎ
𝑎 8 𝑎 9 𝑎 10
147.52 ( ) − 127.69 ( ) + 61.50 ( ) .
ℎ ℎ ℎ

Three models of the cantilever beam were

computationally implemented:
Figure 5. Detail of the cantilever beam where the plan
crack of height 𝑎 and width 𝑏 is located.
 Nonexistence of cracks (the beam has a linear
behavior);
 An always open crack (the beam also has a
linear behavior);
 A breathing crack (the beam has a nonlinear
behaviour);

2.2. Results

Table 1 shows the geometric and material properties

of the cantilever beam.
𝑙
Figure 6. Closed crack, where 𝜃𝑙 + 𝜃𝑟 = 𝑀 . Table 1. Geometric and material properties of the
𝐸𝐼
cantilever beam.

Geometric properties
𝑏 [m] 0.06
ℎ [m] 0.22
𝐿 [m] 8.0
Material properties
𝐸 [Pa] 210 × 109
𝑙 72 𝑎 𝜌 [kg/m3 ] 7800
Figure 7. Open crack, where 𝜃𝑙 + 𝜃𝑟 = 𝑀 ( + 𝐹 ( )).
𝐸𝐼 𝐸𝑏ℎ2 ℎ

3
2.2.1. Natural Frequencies blocks were needed, which means the frequency
convergence is much slower for the cantilever beam.
Table 2 shows the exact natural frequencies of the This obvious difference in the convergence rates is
cantilever beam together with the ones obtained probably due to the fact that the mode shapes of a
through the DEM (approximated frequencies), for an cantilever beam are not the simple circular
increasing number of blocks 𝑁 = 24, 48, 96, 192, 384. trigonometric functions as for the simply supported
Figure 8 represents the ratio between the first five beam.
frequencies calculated with the program developed 2.2.2. Mode shapes of the cantilever beam
with MATLAB and their homologous exact natural without cracks
frequencies, as a function of the number of blocks 𝑁
mentioned in Table 1. Figures 9 and 10 represent the first five exact
normalized mode shapes and the first five mode
shapes obtained using the DEM for a beam
Table 2. Exact and approximated natural frequencies of the discretization of 96 blocks.
cantilever beam.
Exact
Approximated frequencies (𝑝𝑎 ) (rads−1 )
frequencies
(𝑝𝑒 )
𝑁 = 24 𝑁 = 48 𝑁 = 96 𝑁 = 192 𝑁 = 384
(rads −1)

18,10 17,35 17,72 17,91 18,01 18,05

113,45 108,73 111,01 112,17 112,75 113,04

317,67 304,30 310,46 313,65 315,27 316,08

622,51 595,65 607,28 613,40 616,54 618,13

1029,06 982,87 1001,47 1011,39 1016,51 1019,11

1537,23 1464,48 1491,51 1506,05 1513,60 1517,45

Figure 9. First, second and third mode shapes of the

cantilever beam (96 blocks).
___: 1st exact mode; o: 1st calculated mode;
___: 2nd exact mode; o: 2nd calculated mode;
___: 3rd exact mode; o: 3rd calculated mode.

Figure 8. The first five frequencies of the cantilever beam

as a function of the number of blocks (𝑁).
-----: 𝑝1𝑎 /𝑝1𝑒 -----: 𝑝2𝑎 /𝑝2𝑒 -----: 𝑝3𝑎 /𝑝3𝑒 -----: 𝑝4𝑎 /𝑝4𝑒
-----: 𝑝5𝑎 /𝑝5𝑒 .

From the observation of Table 1 and Figure 8 one

can conclude that 384 blocks are required to obtain Figure 10. Fourth and fifth mode shapes of the cantilever
an estimation of the first three exact frequencies with beam (96 blocks).
an error lower than 0.5% for the cantilever beam, ___: 4th exact mode; o: 4th calculated mode;
while for the simply supported beam in (Neves, ___: 5th exact mode; o: 5th calculated mode.
2015), to obtain the same limitation of error, only 12

4
Even without an excellent precision, with a
discretization of 96 blocks it is already possible to
replicate well enough the first three mode shapes.
Obviously, by increasing the number of blocks more
precise mode shapes would be obtained; but the big
improvement achieved for the first five frequencies
and modes due to the transition from a low number
of blocks to 96 blocks is hardly repeatable if the
number of blocks is increased from 96 to more: the
approximation improvement is not maintained with an
indefinite refinement.

2.2.3. Dynamic evolution of the oscillation of

the cracked cantilever beam

In this section the numerical results obtained using

DEM are compared with the results found in the
reference (Loutridis et al., 2005) where the only Figure 11. Experimental model of the cantilever: breathing
situation considered is that of one breathing crack. crack 𝑎 = 0.3ℎ (40 blocks). Acceleration at the free end;
damping factor 𝜉 = 15%.
With the purpose of studying the nonlinear response ____ DEM curve; ▲ Experimental values (Loutridis et
of the cracked cantilever beam, a Plexiglas model al., 2005).
with the characteristics found in Table 3 is considered
(Loutridis et al., 2005).

Table 3. Geometric and material properties of the

cantilever beam (Loutridis et al., 2005).

Geometric properties
𝑏 [m] 0.02
ℎ [m] 0.02
𝐿 [m] 0.230
Material properties
𝐸 [Pa] 2.5 × 109
𝜌 [kg/m3 ] 1200

A transverse crack is assumed at a distance

𝑙𝑓 = 70 mm from the clamped end of the cantilever, in
Figure 12. Experimental model of the cantilever: breathing
its upper part, with a height 𝑎 = 0.006 m (30% of the crack 𝑎 = 0.3ℎ (80 blocks). Acceleration at the free end;
beam’s height). damping factor 𝜉 = 15%.
____ DEM curve; ▲ Experimental values (Loutridis et
An exciting force is applied experimentally by a al., 2005).
15 mm diameter voice coil with a 3 g mass. The coil
was place in the field of a permanent magnet and Figures 11 and 12 represent the acceleration of the
was excited by an oscillator B&K type 2010 using a free end of the cantilever beam calculated with the
sinus signal. An accelerometer was mounted on the DEM, considering the existence of a breathing crack
free end of the beam to pick up the vibrational and a mesh of 40 and 80 blocks, respectively; the
response. It is considered that, after the set-up of the results found in the article (Loutridis et al., 2005) are
experimenting material, the beam is subjected to a also represented in the same figures.
harmonic force with amplitude 𝐹máx = 0.0294 N
(obtained from (3 g) × 10−3 × 9.8 m/s 2). The exciting From the observation of Figures 11 and 12 one can
frequency is 𝑓𝐹 = 44 Hz, approximately half of the see that the refinement from 40 to 80 blocks does not
first natural frequency (𝑓1 = 91 Hz). Rayleigh improve significantly the results. The approximation
damping (Clough, Penzien, 1993) is considered and to the experimental results is quite good: the
the damping coefficients for the two first frequencies irregularity of the acceleration near zero, due to the
are 𝜉1 = 𝜉2 = 0.15. change of the crack state, is well reproduced.

5
3. Suspended beam the following eigenproblem is obtained

3.1. Dynamics of a homogeneous 1

𝜆2 (𝐽 + 𝐿𝐵̲ 𝐴−1𝑚 𝐴−𝑇 𝐵 𝑇 𝐿) + 𝐷 𝑇 𝐾 𝐷
1
𝜆2 ( 𝐿𝐵̲ 𝐴−1𝑚 𝐴−𝑇 𝑎̰ 1′) 𝛩̰
4 2
2𝐽1 𝑇 2𝐾1 𝑇 { }=
suspended beam 𝜆2 (− 𝑎̰ ) − 𝑎̰ 𝜆2 𝑚1 𝑌1
[ 𝑙1 1 𝑙1 2 ] (12)
0
The dynamics of a suspended beam as shown in ={ }
Figure 13 is now studied. The beam is suspended by 0
its ends and is actuated by a force in the direction
from which is possible to calculate approximations for
perpendicular to the gravity acceleration,
frequencies and vibration mode shapes.
transversely to the beam’s axis. Under the action of
the force, the beam motion is free of support
conditions. A breathing crack is oriented transversely 3.2. Results
and it opens and closes according to the curvature
sign in the crack section. As shown in Figure 14, the Table 4 shows the geometric and material properties
beam is discretized in contiguous blocks connected of the suspended beam, which coincide with those in
by rotational springs (shear distortion is again reference (Saavedra, Cuitiño, 2001).
neglected) and the force 𝐹(𝑡) acts in the mass center
of the 𝑘-th block. Table 4. Geometric and material properties of the
suspended beam.

Geometric properties

𝑏 [m] 0.026
ℎ [m] 0.019
𝐿 [m] 0.9
Material properties

𝐸 [Pa] 210 × 109

Figure 13. Model of the cracked suspended beam acted 𝜌 [kg/m3 ] 7800
by a force transverse to the suspension plan.

3.2.1. Natural frequencies

Table 5 shows the exact natural frequencies of the

suspended beam together with the ones obtained
through the DEM (approximated frequencies), for an
increasing number of blocks 𝑁 = 8, 12, 24, 48, 96.
Table 5. Exact and approximated natural frequencies of
the suspended beam.
Exact
Figure 14. Discrete elements model of the suspended Approximated frequencies (𝑝𝑎 ) (rads−1 )
frequencies
beam acted by a force transverse to the suspension plan. (𝑝𝑒 )
𝑁=8 𝑁 = 12 𝑁 = 24 𝑁 = 48 𝑁 = 96
(rads −1)
The system of ordinary differential equations that
governs the motion of the model is (Neves, 2015) 786,07 785,27 785,34 785,36 785,36 785,37

2166,88 2160,69 2162,07 2162,46 2162,45 2162,51

1
𝐽 + 𝐿𝐵̲ 𝐴−1 𝑚 𝐴−𝑇 𝐵𝑇 𝐿
1
𝐿𝐵̲ 𝐴−1 𝑚 𝐴−𝑇 𝑎̰ 1′ 𝜃̰ ̈
4 2 4247,94 4217,72 4229,94 4232,95 4233,23 4233,27
[ 2𝐽1 ]{ } =
− 𝑎̰ 1𝑇 𝑚1 𝑦̈1
𝑙1 7022,07 6887,54 6969,13 6983,75 6985,00 6985,18
1 (10)
−𝐷𝑇 𝐾 𝐷 0̰ 𝜃̰ 𝐹(𝑡)𝐿𝐵̲ 𝐴−1 𝑒̰ 𝑘−1
2 10489,76 9888,98 10351,87 10407,30 10411,49 10412,10
=[ 2𝐾1 ]{ } + { }.
𝑎̰ 2𝑇 0
𝑙1 𝑦1 0
Figure 15 represents the ratio between the first five
Assuming 𝐹(𝑡) = 0 and a solution of the exponential
frequencies calculated with the program developed
type
with MATLAB and the first five homologous exact
natural frequencies, as a function of the number of
𝜃̰ (𝑡) = 𝛩̰ 𝑒 𝜆𝑡 and 𝑦1 (𝑡) = 𝑌1 𝑒 𝜆𝑡 (11) blocks 𝑁.

6
From the observation of Table 5 and Figure 15 one
can conclude that to obtain an error lower than 0.5%
in the estimation of the first three exact frequencies
of the suspended beam 12 blocks can be used. To
obtain an error lower than 1% in the estimation of the
first five frequencies at least 24 blocks are needed.

Figure 17. Fourth and fifth mode shapes of the suspended

beam (31blocks).
___: 4th exact mode; o: 4th calculated mode;
___: 5th exact mode; o: 5th calculated mode.

From the observation of Figure 16 one can see that

31 blocks reproduce well the first three mode
Figure 15. The first five frequencies of the suspended shapes; the same conclusion is taken from Figure 17
beam as a function of the number of blocks (𝑁). for the fourth and fifth mode shapes. This accuracy is
-----: 𝑝1𝑎 /𝑝1𝑒 -----: 𝑝2𝑎 /𝑝2𝑒 -----: 𝑝3𝑎 /𝑝3𝑒 -----: 𝑝4𝑎 /𝑝4𝑒 in agreement with the fast convergence of the
-----: 𝑝5𝑎 /𝑝5𝑒 . frequencies for a low number of blocks, as shown in
Table 5.
3.2.2. Modes shapes of the suspended beam
3.2.3. Dynamic evolution of the cracked
Figures 16 and 17 represent the first five exact suspended beam
normalized mode shapes and the first five mode
shapes obtained using the DEM, also normalized, for
In this section the numerical results obtained using
a beam discretization of 31 blocks.
the DEM are compared with the results of the
reference (Saavedra, Cuitiño, 2001). Table 4 shows
the geometric and material properties of the
simulated suspended beam.

A transverse crack is considered at a distance

𝑙𝑓 = 585 mm from the left end of the beam and with a
height 𝑎 = 0.0076 m (40% of the beam’s height). The
exciting force, applied at a distance of 270 mm from
the left end, has an amplitude 𝐹máx = 10N and a
frequency 𝑓𝐹 = 62 Hz, half of the first natural
frequency (𝑓1 = 124 Hz). Rayleigh damping is
considered and the damping coefficients for the two
first frequencies may have two different values:
𝜉1 = 𝜉2 = 0.1 and 𝜉1 = 𝜉2 = 0.2.

The decision to discretize the beam in 31 blocks is

justified by the fact that it is then possible to locate
Figure 16. First, second and third mode shapes of the the crack, the force and the accelerometer exactly at
suspended beam (31 blocks). the positions (see Figure 18) mentioned in
___: 1st exact mode; o: 1st calculated mode; (Saavedra, Cuitiño, 2001).
___: 2nd exact mode; o: 2nd calculated mode;
___: 3rd exact mode; o: 3rd calculated mode.

7
 One verifies that the approximations obtained with
the DEM, especially considering 2% of damping, are
quiet acceptable and are, in general, better than the
approximations obtained in the models described in
the articles (Saavedra, Cuitiño, 2001) and (Sinha &
Friswell, 2002).A shift between the response of the
model and the experimental response is observed.
However, the steady state response from the model
has a frequency equal to the frequency of the
exciting force. The shift between the responses can
Figure 18. Crack, force and response evaluation positions.
be due to the fact that in the experimental situation
Figures 19 and 20 represent the acceleration of gravity is present which may produce variations in
section A (at a distance of 810 mm from the left end the crack state.
of the beam) calculated with the DEM, considering
the existence of a breathing crack, a mesh of 31
blocks and a damping factor 𝜉 = 0.1 or 𝜉 = 0.2 4. Conclusions
respectively; the results found in the article
(Saavedra, Cuitiño, 2001) are also represented in the
4.1. Contributions
same Figures.
The crack detection in structures is a very important
issue and more practical and less onerous detection
methods are continuously investigated; this topic is
common to many engineering branches such as
Civil, Mechanical and Aeronautical. This work intends
to contribute to the characterization of the vibrations
that the existence of cracks of different
characteristics causes in the free or forced dynamic
response of some structures. Thus, from the data
collected from accelerometers strategically placed in
the beam the existence of cracks can be detected
and their depth and location may be assessed.
In the course of this work a program in MATLAB
environment that allows the application of the
Discrete Elements Method (DEM) to the analysis of
the dynamic behavior of some structures is
Figure 19. Acceleration at section A of the suspended developed. Taking as a starting point the model of a
beam (𝜉 = 1%): breathing crack 𝑎 = 0.4ℎ (31 blocks). simply supported beam, for which expressions were
____ DEM curve; ▲ Experimental values (Saavedra, obtained in (Neild et al., 2001), the expressions for
Cuitiño, 2001). the remaining models were derived.
The DEM is a valid and simple method to simulate
the behaviour of cracked beams. Based on the
presented tables and figures and also on the results
presented in (Neves, 2015), the following
conclusions could be inferred:

 The DEM leads to good approximations of the

natural frequencies and mode shapes;
 The approximations are better for the lower
frequencies and mode shapes, for a given
number of blocks used in the discretization,
though in some cases the convergence is not
monotonous;
 The DEM adapts well in the situations where
cracks exist and the programation in MATLAB is
Figure 20. Acceleration at section A of the suspended easily modified;
beam (𝜉 = 2%): breathing crack 𝑎 = 0.4ℎ (31 blocks).  The results obtained with DEM give good
____ DEM curve; ▲ Experimental values (Saavedra, approximations of the experimental results;
Cuitiño, 2001).

8
 The results obtained with DEM in situations cracked beams using instantaneous frequency.
where experimental results are not available NDT&E international, , 411-419.
come out as expected, namely:
Neild, S., McFadden, P., Williams, M. (2001). A discrete
o The stiffness of a cracked beam is lower than model of a vibrating beam using a time-stepping
the stiffness of an uncracked beam and that approach. Journal of Sound and Vibration,
condition is reflected in the reduction of the 239(1), 99-121.
natural frequencies of the cracked beam and
in its free dynamic response; Neves, C. (2015). Vibrações de vigas: Modelos de massa e
o The frequencies are more sensitive to the rigidez discretas. MSc Thesis in Civil Engineering,
existence of cracks when these are located Instituto Superior Técnico, Lisboa.
in regions of higher curvature of the
corresponding vibration modes; Okamura, H., Liu, H., Chu, C.-S., Liebowitz, H. (1969). A
o The natural frequencies of a beam where cracked column under compression. Engineering
there is a “breathing crack” have Fracture Mechnics, 1, 547-564.
intermediate values between those of an
Orhan, S. (2007). Analysis of free and forced vibration of a
uncracked beam and a beam with an always
cracked cantilever beam. NDT&E International,
open crack;
o The existence of “breathing cracks” located 443-450.
in regions of larger curvature yields more Rao, S. (2004). Mechanical Vibrations. Pearson Education
irregular responses when the crack state Inc., Prentice Hall.
changes (when the crack goes from being
open to closed or vice-versa); Saavedra, P., Cuitiño, L. (2001). Crack detection and
o For a cracked beam, the vibration amplitudes vibration behaviour of cracked beams.
in a forced vibration increase compared to Computers and Structures, 79, 1451-1459.
the case of an uncracked beam, according to
the verified reduction of stiffness. Saeedi, K., Bhat, R. (2011). Clustered natural frequencies
in multi-span. Shock and Vibration, 18, 697-707.
4.2. Future developments Sinha, J., Friswell, M. (2002). Simulation of the dynamic
response of a cracked beam. Computers and
One important aspect that would be interesting to Structures, 80, 1473-1476.
develop in future works is the consideration of the
shear distortion (which was ignored in the numerical
simulations of the presented models). Taking it into
account, the derivation of the expressions that rule
the blocks’ motions becomes more complex but the
achieved results will be more accurate, especially
when beams have small slenderness ratios 𝐿/ℎ.
It would also be interesting to study a larger number
of possibilities for the crack localization and quantity
of cracks, as a way to deepen the knowledge about
how the crack localization modifies the dynamic
behavior of a beam. The analysis could also be
extended to more complex structures, such as
continuous beams with two or more spans or even
frames.

References

Clough, R., Penzien, J. (1993). Dynamics of Structures.

McGraw-Hill International Edition (Civil
Engineering Series).

Loutridis, S., Douka, E., Hadjileontiadis, L. (2005). Forced

vibration behaviour and crack detection of