Hindawi Publishing Corporation

Advances in Mechanical Engineering

Volume 2014, Article ID 792478, 16 pages
http://dx.doi.org/10.1155/2014/792478

Research Article
Free Vibration Characteristic of Multilevel Beam Based on
Transfer Matrix Method of Linear Multibody Systems

Laith K. Abbas and Xiaoting Rui

Institute of Launch Dynamics, Nanjing University of Science and Technology, Nanjing 210094, China

Correspondence should be addressed to Laith K. Abbas; laithabbass@yahoo.com

Received 18 August 2013; Accepted 28 November 2013; Published 20 January 2014

Academic Editor: Caishan Liu

Copyright © 2014 L. K. Abbas and X. Rui. This is an open access article distributed under the Creative Commons Attribution
License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly
cited.

In this paper, an approach based on transfer matrix method of linear multibody systems (MS-TMM) is developed to analyze the free
vibration of a multilevel beam, coupled by spring/dashpot systems attached to them in-span. The Euler-Bernoulli model is used for
the transverse vibration of the beams, and the spring/dashpot system represents a simplified model of a viscoelastic material. MS-
TMM reduces the dynamic problem to an overall transfer equation which only involves boundary state vectors. The state vectors at
the boundaries are composed of displacements, rotation angles, bending moments, and shear forces, which are partly known and
partly unknown, and end up with reduced overall transfer matrix. Nontrivial solution requires the coefficient matrix to be singular
to yield the required natural frequencies. This paper implements two novel algorithms based on the methodology by reducing the
zero search of the reduced overall transfer matrix’s determinate to a minimization problem and demonstrates a simple and robust
algorithm being much more efficient than direct enumeration. The proposal method is easy to formulate, systematic to apply, and
simple to code and can be extended to complex structures with any boundary conditions. Numerical results are presented to show
the validity of the proposal method against the published literature.

1. Introduction The vibration of systems composed of uniform double-beam

The vibration problem of beam-type structures is of partic-
ular urgent issue in many branches of modern aerospace,
mechanical, and civil engineering. Natural vibration frequen-
cies and modes are one of the most important dynamic
characteristics of these kinds of systems. For example, the
precision in manufacturing can be highly influenced by
vibrations. If the vibration characteristics cannot be solved or
preestimated exactly when designing a mechanism system,
it is often hard to obtain a good dynamic performance of
the mechanism system and consequently hard to control its
vibration.
There are different types of beam models. One of the well-
known models is the Euler-Bernoulli beam theory that works
well for slender beams. According to the Euler-Bernoulli
beam theory, the length of each beam section is much greater
than the height of each section and the shear and rotary
inertia effects are ignored.
The vibration theory of single-beam systems is well
developed and studied in detail in hundreds of contributions.
coupled by translational springs or elastic layers have been
studied extensively in the literature. Inceoğlu and Gürgöze
[1] studied the bending vibrations of a combined system
consisting of two clamped-free beams carrying tip masses to
which several double spring-mass systems are attached across
the span. Using Green’s function method, the frequency
equation of the system is established. Kukla [2] solved the
problem of free vibration of two axially loaded beams which
are connected by many translational springs. The solution
contains possible combinations of the classical boundary
conditions. The technique of the solution consists of devel-
oping a Green function. In [3], the vibrations of uniform
beams connected by homogeneous elastic layer are devoted.
Oniszczuk [4] discussed the free transverse vibrations of
two parallel simply supported beams continuously joined
by a Winkler elastic layer. The motion of the system was
described by a homogeneous set of two partial differential
equations, which was solved by using the classical Bernoulli-
Fourier method. Oniszczuk [5] is devoted to analyze the
undamped forced transverse vibrations of an elastically
2 Advances in Mechanical Engineering
connected complex double-beam system in the case of simply
supported beams. The classical modal expansion method was
applied to determine the dynamic responses of the beams
due to arbitrarily distributed continuous loads. Vu et al. [6]
presented an exact method for solving the vibration problem
of a damped double-beam system subjected to harmonic
excitation. The double-beam system consists of two identical
beams with the same boundary conditions on both sides.
The beams are connected by a viscoelastic layer. Natural
frequencies and mode shapes of vibration of the system
are determined and the forced vibrations are investigated.
Gürgöze and Erol [7] determined the natural frequencies
of a clamped-free double-beam system carrying tip masses
to which several spring mass systems are attached across
the span. However, there are only few contributions dealing
with the vibration of multibeam systems. That is, probably,
the general vibration analyses of an elastically connected
multibeam system are complicated and laborious in view
of a large variety of possible combinations of boundary
conditions, and thus, the solution of the governing coupled
partial differential equations is difficult [8].
Multibody system dynamics (MSD) has become an
important theoretical tool for wide engineering problems
analysis in the world. Lots of methods of MSD have been
studied by many authors on theory and computational
method [9–15]. Professor Rui Xiaoting and his students have
been enlightened by the method of letting state vectors (SVs)
be transferred into classical transfer matrix method and
built up a new multibody dynamics method called “Transfer
Matrix Method of Linear Multibody Systems-MS-TMM”
[16]. Using MS-TMM, the eigenvalue of linear multi-rigid-
flexible-body system is computed easily, the computational
ill-condition is overcome, and the computational efficiency is
increased. Over 20 years, MS-TMM has been developed and
used widely in engineering applications.
Motivated by the interesting study by Kukla [2] which
was published on the problem of the natural longitudinal
vibrations of two rods coupled by many translational springs
and by the two novel algorithms based on the new recursive
scanning approach presented by Bestle et al. [17], this paper
presents a unique yet simple scenario of obtaining the
exact free vibration characteristics of undamped/damped
multilevel beam coupled elastically. The scenario developed
in this paper is based on MS-TMM and Euler-Bernoulli beam
theory.
The text is organized as follows. The problem statement
is presented in Section 2. In Sections 3 and 4, the general
theorem brief of MS-TMM and problem solution scenario are
shown. In Section 5, some results calculated by MS-TMM and
the other method are given which can validate the proposed
method. The conclusions are presented in Section 6.
2. Problem Statement
The transversely vibrating system in a plane under considera-
tion consists of multi-level parallel, elastic, and homogeneous
Euler-Bernoulli beam with general boundary conditions.
Beam bending stiffness is 𝐸𝐼, where 𝐸 is the elastic modulus
and 𝐼 is the area moment of inertia. Beam mass per unit
length 𝑚 = 𝜌𝐴, where 𝜌 is the material density and 𝐴 is the
beam cross-section area. Of course, any different materials,
different dimensions, or different boundary conditions could
be considered in MS-TMM. The beams have the same length
𝐿 and are joined by the 𝑗-system of spring/dashpot located
at different positions (e.g., 𝑥1 and 𝑥𝑗 , where 𝑗 is the number
of spring/dashpot systems between two levels). 𝑘𝑦 and 𝑐 are
linear translational spring constant and damping coefficient,
respectively, as shown in Figure 1.
3. MS-TMM Strategy in the Context of
Free Vibration Characteristics
3.1. MS Topologies. According to the natural attribute of
bodies, a complicated MS can be represented by various
bodies (e.g., rigid bodies, elastic bodies, lumped masses,
etc.) interconnected by hinges (e.g., spherical joints, sliding
joints, cylindrical joints, dampers, springs, etc.). In MS-
TMM, there are different topologies based on a certain set
of modeling variables to formulate the dynamic equations of
MS. Such topologies are chain, tree, closed loop (as illustrated
in Figure 2 for reference), network, and so forth.
3.2. State Vector, State Variables, and Transfer Direction.
The state vector (SV) at a connection point of MS is a
column vector denoting the mechanics state of this point.
It includes the displacements of the point (including angu-
lar displacements) and the corresponding internal forces
(including internal moments). Therefore, the SV is given
by kinematics (displacements) and kinetics (internal forces)
quantities, called state variables. For convention in this paper,
z with bold lowercase represents the SV in the physical
coordinates and Z with bold capital represents the SV in
the modal coordinates. Vibrations in space are described
by displacement coordinates 𝑥, 𝑦, 𝑧 along the Cartesian axes
and angular rotations 𝜃𝑥 , 𝜃𝑦 , 𝜃𝑧 about Cartesian axes. Cutting
forces and moments are given by 𝑞𝑥 , 𝑞𝑦 , 𝑞𝑧 and 𝑚𝑥 , 𝑚𝑦 , 𝑚𝑧 ,
respectively. Positive directions at input points are shown
in Figure 3(a). Positive directions of forces and moments at
output points (Figure 3(b)) are opposite due to the principle
of action equals reaction. In 3D case with 𝑛𝑠 = 12 (𝑛𝑠 is the
number of state variables in the SV), the SVs in physical and
modal coordinates at the connection point 𝑝𝑖,𝑘 (where the
first subscript 𝑖 is the serial number of element for boundary
end and the second subscript 𝑘 is the serial number of the
hinge element and 𝑘 = 0 for boundary end) are summarized
in a vector, receptively:
󵄨
z𝑖,𝑘 󵄨󵄨󵄨physical coordinates
𝑇
= [𝑥, 𝑦, 𝑧, 𝜃𝑥 , 𝜃𝑦 , 𝜃𝑧 , 𝑚𝑥 , 𝑚𝑦 , 𝑚𝑧 , 𝑞𝑥 , 𝑞𝑦 , 𝑞𝑧 ]𝑖,𝑘 ,
(1)
󵄨
Z𝑖,𝑘 󵄨󵄨󵄨modal coordinates
𝑇
= [𝑋, 𝑌, 𝑍, Θ𝑥 , Θ𝑦 , Θ𝑧 , 𝑀𝑥 , 𝑀𝑦 , 𝑀𝑧 , 𝑄𝑥 , 𝑄𝑦 , 𝑄𝑧 ]𝑖,𝑘 .
Advances in Mechanical Engineering 3

y
General
boundary x Beam 1 m1 , (EI)1
conditions
ky c ··· ky c
Beam 2 m2 , (EI)2

ky c ··· ky c
..
x1 .
xj General
boundary
conditions
ky c ··· ky c
Beam m mm , (EI)m

Figure 1: A system of a multi-level beam with a distributed spring/dashpot system(s).

Hinges

Loop

Bodies
Loop
Base
(a) (b) (c)

Figure 2: Multibody systems (a) chain, (b) tree, and (c) closed loop topology.

In case of 1D or 2D applications, the SV will be reduced direction shown in Figure 3, the transfer equation between the
as shown later. Defining a boundary point of the MS as component input and output is
the transfer end, the direction from all other boundary points
to the transfer end is called transfer direction. Along the z𝑛,0 = U𝑛 z𝑛−1,𝑛
transfer direction, the nodes entering into elements are called
inputs denoted by 𝐼 and the nodes leaving from elements are z𝑛−1,𝑛 = U𝑛−1 z𝑛−2,𝑛−1
called outputs 𝑂. z𝑛−2,𝑛−1 = U𝑛−2 z𝑛−3,𝑛−2 (2)
..
3.3. Transfer Equation, Transfer Matrix, Overall System Trans- .
fer Matrix, and Overall System State Vector. A vibrating MS
comprised of 𝑛-components, see Figure 4, is used as an z1,2 = U1 z1,0 .
example to show how to deduce the overall transfer equation
The constant matrix U𝑘 is the transfer matrix of the
and overall transfer matrix of the system. In order to describe
𝑘th component. Transfer matrices of basic components are
conveniently the idea, the chain topology is considered
considered as building blocks, which can be assembled
in the following. This vibrating system is comprised of 𝑛
together to provide the transfer matrix of the whole system
components and 𝑛 + 1 connection points. The SVs of the
according to the chain MS-TMM topology as follows:
boundary right extremity and other boundary left extremity
of the system are expressed as z𝑛,0 and z1,0 , respectively. 𝑛−1
Transfer direction of the system is always from another z𝑛,0 = Tz1,0 , where T = ∏U𝑛−𝑘 . (3)
boundary end to may call it the root. Following the transfer 𝑘=0
4 Advances in Mechanical Engineering

yI , qyI yO , −qyO

𝜃yI , −myI −𝜃yO , myO

−𝜃xO , mxO
𝜃xI , −mxI

I xI , qxI O xO , −qxO

𝜃zI , −mzI −𝜃zO , mzO

zI , qzI zO , −qzO

(a) (b)

Figure 3: Sign convention at (a) input; (b) output.

Transfer direction

2 4 ··· n − 1 n 0 Root
0 1 3
Left end Right end
z1,0 zn,0

I 3 O
{

z2,3 z3,4

Figure 4: TMM-MS in the sense of chain topology. A vibrating system comprised of 𝑛-components with 𝑛 + 1 connections.

Rewrite (2) as from elimination of all columns of Uall (4c) associated zeros
󵄨 󵄨 in zall . For harmonic vibrations, solutions maybe written as
Uall 󵄨󵄨󵄨𝑛𝑠 ×(2×𝑛𝑠 ) zall 󵄨󵄨󵄨(2×𝑛𝑠 )×1 = 0, (4a) z𝑖,𝑗 = Z𝑖,𝑗 𝑒𝜆𝑡 where 𝜆 = −𝜆𝑟 ± 𝑖𝜆𝑖 , 𝜆𝑟 , 𝜆𝑖 ∈ R are the
eigenvalues. The real part (−𝜆𝑟 ) is related to the magnitude
where
of damping, where the imaginary part (𝜆𝑖 ) is related to the
𝑇 vibration frequency of the damped system. For undamped
z𝑇all = [z𝑇1,0 z𝑇𝑛,0 ] , (4b)
systems, 𝜆𝑟 = 0 and 𝜆𝑖 = 𝜔. Finally, Uall is only a function of
Uall = [T −I𝑛𝑠 ] . (4c) the unknown 𝜆 𝑖 of the system. For nontrivial solutions, the
Eigenfrequency equation
Herein, Uall is the overall system transfer matrix and zall is the
!
overall system state vector. Δ (𝜆) = det Uall = 0 (5)

3.4. Eigenfrequency Equation of the Whole MS. The overall
transfer equation (4a) only involves the boundary SVs, and
the SVs at all other connection points do not appear. The
SVs at the boundary are composed of displacements, rotation
angles, moments, and shears, which are partly known and
partly unknown. For common boundary conditions, half of
state variables of zall (4b) are zeros due to known constraints.
Thus, (4a) reduces to Uall zall = 0, where zall is composed of the
unknown state variables and Uall is a square matrix resulting
has to be fulfilled. The natural frequencies of the system can
now be computed.
3.5. Beam Transfer Matrix. The full derivation of transfer
matrices for the Timoshenko and Euler-Bernoulli beams
vibrating in a plane (with kinematics and kinetics’ SV defined
as z = [𝑦, 𝜃𝑧 , 𝑚𝑧 , 𝑞𝑦 ]𝑇 ) may be found in [18], which is an
open access article and the reader may download it from the
Internet. However, for completeness, only the transfer matrix
Advances in Mechanical Engineering 5

for the Euler-Bernoulli beam will be presented, Figure 5. The where 𝑌 (𝑥) = 𝐴 1 cosh 𝛽𝑥 + 𝐴 2 sinh 𝛽𝑥
differential equation of a Euler-Bernoulli beam is
+ 𝐴 3 cos 𝛽𝑥 + 𝐴 4 sin 𝛽𝑥,
(6)
𝜕4 𝑦 𝜕2 𝑦
𝐸𝐼 4 + 𝑚 2
𝜕𝑥 𝜕𝑡 𝐴 1 , 𝐴 2 , 𝐴 3 , and 𝐴 4 are arbitrary constants, and 𝛽 =
√4 −𝑚𝜆2 /(𝐸𝐼). For the Euler-Bernoulli beam, the linearized
𝜕4 𝑌 (𝑥) 𝑚𝜆2 relations in modal coordinates Θ𝑧 = 𝑌󸀠 , 𝑀𝑧 = 𝐸𝐼𝑌󸀠󸀠 , and
= 0 𝑦 (𝑥, 𝑡) = 𝑌 (𝑥) 𝑒𝜆𝑡 4
+ 𝑌 (𝑥) = 0,
󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀→ 𝜕𝑥 𝐸𝐼 𝑄𝑦 = 𝑀𝑧󸀠 maybe added to end up with the transfer relation:

𝑌 cosh 𝛽𝑥 sinh 𝛽𝑥 cos 𝛽𝑥 sin 𝛽𝑥 𝐴1

[ ] [ ][ ]
[ Θ𝑧 ] [ 𝛽 sinh 𝛽𝑥 𝜆 cos 𝛽𝑥 ]
[ ] [ 𝛽 cosh 𝛽𝑥 −𝛽 sin 𝛽𝑥 ][ 𝐴 2]
[ ] [ ][[ ]
]
[ ]
Z (𝑥) = B (𝑥) a or [
[𝑀 ]
] =[ ] [ ]. (7)
[ 𝑧] [𝐸𝐼𝛽2 cosh 𝛽𝑥 𝐸𝐼𝛽2 sinh 𝛽𝑥 −𝐸𝐼𝛽2 cos 𝛽𝑥 −𝐸𝐼𝛽2 sin 𝛽𝑥 ] [ ]
[𝐴 3 ]
[ ] [ ]
[ ] [ ][ ]
[
]
[ 3 3 3 3
]
𝑄𝑦 𝐸𝐼𝛽 sinh 𝛽𝑥 𝐸𝐼𝛽 cosh 𝛽𝑥 𝐸𝐼𝛽 sin 𝛽𝑥 −𝐸𝐼𝛽 cos 𝛽𝑥 𝐴4
[ ]𝑥 [ ][ ]

The coefficient vector a = [𝐴 1 , 𝐴 2 , 𝐴 3 , 𝐴 4 ]𝑇 summarizes (1) Break up the complicated MS into components with
the unknown constants to be adopted to boundary condi- simple dynamic properties, which can be expressed in matrix
tions. At input end Z𝐼 (𝑥 = 0), we get Z𝐼 = [B(0)]a. Thus, form and SVs (for each component, it is possible to obtain
the coefficient vector can be expressed as a = [B(0)]−1 Z𝑂 and the close form expression of the transfer matrix giving the
substituting it into (7) for the beam output end at 𝑥 = 𝑙, one displacements and the forces applied to one extremity to
gets the displacements and forces applied to the other extremity).
In other words, on the component level, the governing
Z𝑂 = [B (𝑙)] a = [B (𝑙)] [B (0)]−1 Z𝐼 = UZ𝐼 , (8a) partial and ordinary differential or algebraic equations are
transformed to algebraic transfer equations, where the output
where state results from a product of the input state and an
element specific transfer matrix. These component matrices
U = B (𝑙) B−1 (0) are considered as building blocks. In fact, the transfer matrix
of such components needs not to be rededuced but may be
𝑇 𝑈 𝑉
[ 𝑆
taken directly from a transfer matrix library.
𝛽 𝐸𝐼𝛽 2 𝐸𝐼𝛽3 ]
[ ] (2) Following the transfer direction that has been
[ 𝑇 𝑈 ]
[ ] designed already by the analyst and according to the topology
[ 𝛽𝑉 𝑆 ]
=[
[
𝐸𝐼𝛽 𝐸𝐼𝛽2 ] ,
]
of the MS, these component transfer matrices are then
[ 𝑇 ] assembled and end up with a system of linear algebraic
[𝐸𝐼𝛽2 𝑈 𝐸𝐼𝛽𝑉 𝑆 ]
[ 𝛽 ]
equations called the overall transfer equation.
[ ] (3) Substitute the boundary conditions into the overall
3 2
[ 𝐸𝐼𝛽 𝑇 𝐸𝐼𝛽 𝑈 𝛽𝑉 𝑆 ] transfer equation to construct the eigenfrequency equation.
Consequently, the vibration characteristics such as frequen-
𝑐ℎ + 𝑐 𝑠ℎ + 𝑠
𝑆= , 𝑇= , cies can be deduced as the roots of a transcendental equation.
2 2 Due to narrow couples of natural frequencies, the classical
𝑐ℎ − 𝑐 𝑠ℎ − 𝑠 zero search method is likely to fail. However, a new recursive
𝑈= , 𝑉= , scanning approach for minima of the absolute values of the
2 2
determinant shows much more efficiency and reliability than
𝑐ℎ = cosh (𝛽𝑙) , 𝑠ℎ = sinh (𝛽𝑙) , direct enumeration.
𝑐 = cos (𝛽𝑙) , 𝑠 = sin (𝛽𝑙)
4. Problem Solution Scenario
(8b)
Figure 6 illustrates the suggested scenario for the solution
is the transfer matrix of the Euler-Bernoulli beam compo- of the problem statement and as follows. There are (1 : 𝑚)
nent. multi-level beam. Each beam level is divided into (1 : n)
components, which have (0 : 𝑛 + 1) connection points
3.6. Summary. In the context of free vibration characteristic, and coupled with another beam level by a system (say 1 :
the general strategy of linear MS-TMM, in summary, is as 𝑗) of a viscoelastic material (modeled as a spring/dashpot
follows. system). The massless dummy body as shown in Figure 6
6 Advances in Mechanical Engineering

qyI
𝜃zO mzO
mzI 𝜃zI m, L, EI

qyO
yI yO

Figure 5: The direction of the state vector of Euler-Bernoulli beam component.

Massless dummy body

i−1 i n−1 n
0 1 2
A G
U1,1 U1,2 ··· U1,i ··· U1,n−1 U1,n 1

1, 1 j, 1

U2,1 U2,2 ··· U2,i ··· U2,n−1 U2,n 2

B C
.. ..
. .
.. 1, m − 1
.
..
.. j, m − 1 .
.
V m
Um,1 Um,2 ··· Um,i ··· Um,n−1 Um,n

Beam segment component

Spring/dashpot component

Figure 6: Structure of a multi-level beam with the distribution of spring/dashpot.

(e.g., 𝐴, 𝐵, 𝐶, and so on) is a connection point between Figure 7(a) sketches the first step. According to the continuity,
the two beam segment components and spring/dashpot 𝑗 has identical displacements, angles, and moments at input
system(s). The main key of MS-TMM is transferring the and output:
SV from one component to another following the general
transfer equation (2). For the beam segment component, 𝑌𝑂 = 𝑌𝐼 , Θ𝑧𝑂 = Θ𝑧𝐼 , 𝑀𝑧𝑂 = 𝑀𝑧𝐼 . (9a)
the transfer matrix is available (8b), while it is not for
spring/dashpot system at the connection point between two And from the force analysis shown in Figure 7(b), the
or multi-level beam in this paper. However, it is based on the spring/dashpot force 𝐹 changes the shear forces:
kinematics and kinetics of the spring/dashpot to formulate 𝑄𝑦𝑂 = 𝑄𝑦𝐼 + 𝐹. (9b)
the transfer matrix and it needs two steps. First, as an
example, let us consider a connection point 𝑗 between two Equations (9a) and (9b) in the view of the SV as a matrix form
beam segment components and system of spring/dashpot
may be grounded or connected to another connection point. Z𝑂 = Z𝐼 + e4 𝐹. (9c)
Advances in Mechanical Engineering 7

Massless dummy body

Beam segment I j Beam segment

O
Fsys yO
Qy,I F Qy,O

ky c
YI YO
ky c
I j O yI

Mz𝐼 Θz𝐼 Θz𝑂 Mz𝑂

Maybe grounded or connected
to another connection point ZI ZO Fsys

(a) (b) (c)

Z1,I 1 Z1,O
G Level
1

F
Fsys

ky𝑗,1

cj,1

Fsys

Z1,I 1 F
Z1,O Level
Level
A 1 C 2
Z2,I Z2,O
F F
Fsys
Fsys

Massless dummy body

ky1,1 ky1,𝑚−1

c1,1 c1,m−1

Fsys Fsys
F F
Level Level
B 2 V 3
Z2,I Z2,O Z3,I Z3,O

(d) (e)

Figure 7: (a) Two beam segments and spring/dashpot system connected at massless dummy body, (b) state vectors (SV) at the massless
dummy body 𝑗-connection point, (c) force analysis of a spring/dashpot system, (d) a system 1,1 connected two levels 1-2 through two
connection points 𝐴 and 𝐵 shown in Figure 6, and (e) two systems 𝑗, 1 and 1, 𝑚−1 connected with multi-level (1-2-3) through three connection
points 𝐺, 𝐶, and 𝑉 shown in Figure 6.
8 Advances in Mechanical Engineering

𝑇
e4 = [0 0 0 1] is a unit vector assigning 𝐹 to the points 𝐺, 𝐶, and 𝑉 shown in Figures 6 and 7(e) can be
transfer equation of the massless connection body. The obtained similarly:
spring compression and damping forces are given as (see
Figure 7(c))
U𝑑𝑗,1 U𝑐𝑗,1 0
Z [ ] Z
𝐹spring = 𝑘𝑦 (𝑌𝐼 − 𝑌𝑂) = 𝑘𝑦 Δ𝑌 = 𝑘𝑦 e𝑇1ΔZ { }
{ 1
} [ 𝑐 𝑑 𝑐 ] { 1}
{ }
{ Z2 } = [ 𝑗,1 𝑗,1/1,𝑚−1 1,𝑚−1 ]
[ U U U
] { Z2} ,
{ } [ ] { }
𝐹dashpot = 𝑐 (𝑦𝐼̇ − 𝑦𝑂̇ ) 𝑦 = 𝑌𝑒𝜆𝑡 = 𝑐𝜆 (𝑌𝐼 − 𝑌𝑂) = 𝑐𝜆e𝑇1 ΔZ [ ]
{Z3 }𝑂 0 U𝑐1,𝑚−1 U𝑑1,𝑚−1 {Z3 }𝐼
󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀→ [ ]
󳨀→ 𝐹sys = (𝑘𝑦 + 𝑐𝜆) e𝑇1ΔZ.
where U𝑑𝑗,1 ≡ I − D𝑗,1 , U𝑐𝑗,1 ≡ D𝑗,1
(10) (12d)

U𝑑1,𝑚−1 ≡ I − D1,𝑚−1 , U𝑐1,𝑚−1 ≡ D1,𝑚−1

Herein, e𝑇1 = [1 0 0 0] is a unit vector, which extracts 𝑌
from the transfer SV. Substitute (10) into (9c) and knowing
that 𝐹 = ± 𝐹sys , we ended the first step with U𝑑𝑗,1/1,𝑚−1 ≡ I − D𝑗,1/1,𝑚−1 ,

Z𝑂 = Z𝐼 ± DΔZ, D𝑗,1/1,𝑚−1 = ((𝑘𝑦 + 𝑐𝜆)𝑗,1 + (𝑘𝑦 + 𝑐𝜆)1,𝑚−1 ) e4 e𝑇1 .

0 0 0 0
[ 0 0 0 0] (11)
where D = (𝑘𝑦 + 𝑐𝜆) e4 e𝑇1 =[
[
]. Now we are in the position to end the solution scenario.
0 0 0 0] From the MS-TMM topology point of view, the system shown
[ 𝑘𝑦 + 𝑐𝜆 0 0 0] in Figure 1 or equivalently Figure 6 can be modeled as a mul-
tichain, following the similar procedures mentioned above
For system 1,1 connected two levels 1-2 through two in Section 3.3 to determine the overall transfer equation and
connection points 𝐴 and 𝐵 shown in Figure 6, the second step ending up with determinant solution of the reduced overall
is how to deduce the transfer equation between the input and transfer matrix.
output related to these two levels. For the connection point 𝐴
(see Figure 7(d)) and using (11) with 𝐹 = 𝐹sys ,
5. Numerical Examples
Z1,𝑂 = Z1,𝐼 + D1,1 (Z2,𝐼 − Z1,𝐼 ) Before the free vibration analysis of a multi-level elastic
beam coupled by a spring/dashpot system(s) is performed,
(I − D1,1 )Z1,𝐼 + D
= ⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟ ⏟⏟Z2,𝐼
⏟⏟⏟⏟⏟
1,1 the reliability of the suggested scenario and the Matlab
U𝑑1,1 U𝑐1,1
(12a)
computer program developed for this paper are confirmed
by comparing the present results with those obtained from
Z1,𝑂 = U𝑑1,1 Z1,𝐼 + U𝑐1,1 Z2,𝐼 . the existing literature.

Similarly, for connection point 𝐵 and using (11) with 𝐹 =

5.1. Reliability of the Suggested Scenario. Figure 8(a) consists
−𝐹sys ,
of two free-free and clamped-clamped undamped beams
connected to each other with two symmetrically distributed
Z2,𝑂 = Z2,𝐼 − D1,1 (Z2,𝐼 − Z1,𝐼 ) , linear translational springs. The hybrid system is vibrating
(12b) transversely in the 𝑥 − 𝑦 plane. The beams are supposed to
Z2,𝑂 = U𝑐1,1 Z1,𝐼 + U𝑑1,1 Z2,𝐼 . have the same length, material, and geometrical data, that is,
(𝐿 = 1 m, 𝑚1 = 𝑚2 = 𝑚 and (𝐸𝐼)1 = (𝐸𝐼)2 = 𝐸𝐼). It is worth
Combining (12a) and (12b) in one matrix form, we ended up mentioning here that the proposal scenario can be extended
the second step by writing the transfer equation and transfer to a system consisting of any number of uniform/nonuniform
matrix of a system located between two levels as beams coupled with any number of spring/dashpot systems.
However, the two springs 𝑘𝑦1,1 and 𝑘𝑦2,1 are located at distances
𝑥1 and 𝑥2 = 𝐿 − 𝑥1 , respectively, and are moving as a
U𝑑1,1 U𝑐1,1 function of 𝑥1 from 0 to 0.5 m as shown in Figure 8(a). This
Z1 ] {Z1 } .
{ } =[ 𝑐
Z2 𝑂 U1,1 U𝑑1,1 Z2 𝐼 (12c) example is presented by Kukla [2], which studied the dynam-
[ ]
⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟ ics of the longitudinal vibrations of two rods coupled by
transfer matrix several translational springs using Green’s functional method.
Figure 8(b) shows a vibrating two-level beam comprised of
The transfer equation for two systems 𝑗, 1 and 1, 𝑚 − 1 10 components with 4 boundaries, namely, Z1,0 , Z6,0 on the
connected with multi-level (1-2-3) through three connection system left side and Z5,0 , Z10,0 on the right side. Herein,Z𝑖,𝑘 =
Advances in Mechanical Engineering 9

y Transfer direction
x m1 , (EI)1
Beam 1 1 3 5
ky1,1 ky2,1 0 0 Level 1
Z1,0 Z1,2 Z2,3 Z3,4 Z4,5 Z5,0
m2 , (EI)2 Chain 1
Beam 2 ky1,1 ky2,1
2 4
6 7 8
x1 0 Level 2
0 Z Z6,2 Z2,7 Z7,4 Z4,8 Z8,0
x2 = 1 − x1 6,0 Chain 2
x1 L − 2x1 x1
L

Massless dummy body

Beam segement
(a) (b)

Figure 8: (a) Free-free and clamped-clamped beams connected with two linear translational springs, (b) chain TMM-MS: State vectors and
transfer direction of the suggested solution scenario.

[𝑌, Θ𝑧 , 𝑀𝑧 , 𝑄𝑦 ]𝑇𝑖,𝑘 . Chain MS-TMM topology is systematic to ↓

apply. Following the transfer direction from the left to right Z5,0
{ }
and applying (2) for two-level, the transfer equation can be Z8,0 𝑂
written as
U5 U4,𝑑 2,𝑑
2,1 U3 U1,1 U1 U5 U4,𝑑 2,𝑐
2,1 U3 U1,1 U6
[( ]
[ +U U4,𝑐 U U2,𝑐 U ) (+U U4,𝑐 U U2,𝑑 U )]
U5 0 [ 5 2,1 7 1,1 1 5 2,1 7 1,1 6 ]
Z5,0 Z4,5 [ ]
{ } =[ ]{ } =[ ]
Z8,0 𝑂 0 U8 Z4,8 𝐼 [ U8 U4,𝑐 U U 2,𝑑
U U U 4,𝑐
U U 2,𝑐
U ]
[ 2,1 3 1,1 1 8 2,1 3 1,1 6 ]
[( 4,𝑑 2,𝑐 ) ( 4,𝑑 2,𝑑 )]
↓ +U8 U2,1 U7 U1,1 U1 +U U U U
8 2,1 7 1,1 6 U
[ ]
U4,𝑑 U4,𝑐 Z1,0
Z4,5 [ 2,1 2,1 ] Z3,4
{ } = [ 4,𝑐 4,𝑑 ] { } ×{ }
Z4,8 𝑂 U2,1 U2,1 Z7,4 𝐼 Z6,0 𝐼
[ ] ↓
↓
T1,1 T1,2
Z5,0 ] {Z1,0 } .
Z3,4 U3 0 Z2,3 { } =[
Z8,0 𝑂 T2,1 T2,1 Z6,0 𝐼
{ } =[ ]{ } [ ]
Z7,4 𝑂 0 U7 Z2,7 𝐼
(13)
↓
U1 , U3 , U5 , U6 , U7 , and U8 represent the beam segment
U2,𝑑 U2,𝑐 transfer matrices (8b). U2,𝑑 2,𝑐 4,𝑑 4,𝑐
1,1 , U1,1 , and U2,1 , U2,1 are the
Z2,3 [ 1,1 1,1 ] Z1,2
{ } =[ 𝑐 ]{ } transfer matrices of a system between two connection points
Z2,7 𝑂 U1,1 U2,𝑑
1,1
Z6,2 𝐼
located between two beams (12c). The overall transfer equa-
[ ]
tion, overall SV, and overall transfer matrix are, respectively,
↓
󵄨 󵄨
Uall 󵄨󵄨󵄨8×16 Zall 󵄨󵄨󵄨16×1 = 0, (14a)
Z1,2 U1 0 Z1,0
{ } =[ ]{ } 𝑇 𝑇 𝑇 𝑇 𝑇
Z6,2 𝑂 0 U6 Z6,0 𝐼 where Z𝑇all = [Z ⏟⏟⏟ ⏟ Z
⏟⏟⏟1,0 ⏟⏟⏟ ⏟ Z
⏟⏟⏟6,0 ⏟⏟⏟ ⏟ Z
⏟⏟⏟5,0 ⏟⏟⏟8,0
⏟⏟⏟ ⏟] , (14b)
4×1 4×1 4×1 4×1
⇓ 󵄨 󵄨
T1,1 󵄨󵄨󵄨4×4 T1,2 󵄨󵄨󵄨4×4 −I4×4 O4×4
U4,𝑑 U4,𝑐 Uall = [ ]. (14c)
Z5,0 U5 0 U 02,1 2,1 󵄨󵄨 󵄨󵄨
{ } = [ ][ ][ 3 ] T 󵄨 T
[ 2,1 󵄨4×4 2,2 󵄨4×4 󵄨 O 4×4 −I4×4 ]
Z8,0 𝑂 0 U8 4,𝑐 4,𝑑 0 U7
[U2,1 U2,1 ]
Applying the boundary conditions listed in Table 1, half
U2,𝑑 2,𝑐
1,1 U1,1
of state variables of Zall (14b) are zeros due to known
×[ 𝑐 ] [U1 0 ] {Z1,0 } constraints. Thus, the overall transfer equation reduced to
U1,1 U2,𝑑
1,1
0 U6 Z6,0 𝐼
Uall |8 × 8 Zall |8 × 1 = 0 that is ready for eigenproblem (5).
[ ]
10 Advances in Mechanical Engineering

Table 1: Common boundary conditions for a beam vibrating in a grids 𝑁𝑥0 = 500, absolute precision tolerance 𝜀 = 10−6 ).
plane. Figure 9 shows log10 |Δ| obtained from the fMin1D algorithm
versus the first 10 dimensionless frequency values (𝜔 =
Support type Zero terms Nonzero terms
Fixed 𝑌, Θ𝑧 𝑄𝑦 , 𝑀𝑧 √4 (𝜆𝑖 ≡ 𝜔)2 𝑚𝐿4 /𝐸𝐼) for 𝑘𝑦 = 𝑘𝑦 = 𝑘𝑦 = 100(𝐸𝐼/𝐿3 ) and
1,1 2,1

Pinned 𝑌, 𝑀𝑧 𝑄𝑦 , Θ𝑧 𝑥1 = 0.3 m. 𝜔 values are plotted versus the location of the

Free 𝑄𝑦 , 𝑀𝑧 𝑌, Θ𝑧
Guided 𝑄𝑦 , Θ𝑧 𝑌, 𝑀𝑧
The natural frequencies of the system can now be com-
puted by zero search of the determinant, which is based on
sign change of Δ(𝜔) (in case of real numbers) or Δ(𝜆) (in
case of complex numbers) during a scanning of an interesting
frequency range. This procedure can be cumbersome for
several reasons. Therefore, reliable and efficient algorithms
called recursive scanning approach are applied (see [17] for
more details of the proposed algorithm) by switching from
zero search for Δ to minimization of the absolute value |Δ|
of the determinant, which is equally well applicable to both
the real and the complex cases. The algorithm general idea
is as follows. In a first iteration step, it divides an interesting
band of frequencies into a number of sample points and
searches for lower peaks. Each region having a lower peak is
then divided again into small intervals to find more narrow
regions of lower peaks as second iteration step. The algorithm
proceeds until the required precision of roots is achieved. MS-
TMM natural frequency results are obtained using fMin1D
[1, 3000] rad/sec, scanning
algorithm [17] (frequency ranges ⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟
𝜆𝑖 =𝜔
two springs in Figures 10(a) and 10(b) for 𝑘𝑦 = 100(𝐸𝐼/𝐿3 )
and 𝑘𝑦 = 1000(𝐸𝐼/𝐿3 ), respectively. The first and second
springs are moving opposite to each other from 𝑥1 = 0 at
the beams support ends and coincide together at the middle
of the beams (𝑥1 = 0.5 m). Reference [2] does not provide the
results as tables or specific data but only showed figures. The
reader may download the reference from the library or from
the Internet because the authors of the present paper cannot
include the figures to maintain the journal publication rights.
However, the MS-TMM results are in very good agreement
with Figures 2(a) and 2(b) [2, page 131].
5.2. Free Vibration of Damped Elastically Coupled Triple
Beams. The physical model of the transversely vibrating
system under consideration is composed of three paral-
lel uniform rectangular Euler-Bernoulli beams of homoge-
nous properties. Each two-level beam is joined (connected)
together by two spring/dashpot systems located at 𝑥1 and 𝑥2 ,
respectively (see Figure 1). The beams have the same length
and are pinned at their ends. The small damped vibrations of
the system are considered. In the sense of the chain MS-TMM
(see Figure 11), the overall transfer equation, overall transfer
matrix, and overall SV are, respectively,

Uall Zall = 0, (15a)

󵄨 󵄨 󵄨
T1,1 󵄨󵄨󵄨4×4 T1,2 󵄨󵄨󵄨4×4 T1,3 󵄨󵄨󵄨4×4 −I4×4 O4×4 O4×4
[ ]
[ 󵄨 󵄨 󵄨 ]
where Uall = [ T2,1 󵄨󵄨󵄨4×4 T2,2 󵄨󵄨󵄨4×4 T2,3 󵄨󵄨󵄨4×4 O4×4 −I4×4 O4×4 ] , (15b)
[ ]
󵄨 󵄨 󵄨
[ T3,1 󵄨󵄨󵄨4×4 T3,2 󵄨󵄨󵄨4×4 T3,3 󵄨󵄨󵄨4×4 O4×4 O4×4 −I4×4 ]12 × 24
𝑇
Z𝑇all = [Z1,0 𝑇 Z6,0 𝑇 Z11,0 𝑇 Z5,0 𝑇 Z10,0 𝑇 Z13,0 𝑇 ]24 × 1 , (15c)

T1,1 = U5 U4,𝑑 2,𝑑 4,𝑐 2,𝑐

2,1 U3 U1,1 U1 + U5 U2,1 U8 U1,1 U1 ,

T1,2 = U5 U4,𝑑 2,𝑐 4,𝑐 2/7,𝑑

2,1 U3 U1,1 U6 + U5 U2,1 U8 U1,1/1,2 U6 ,

T1,3 = U5 U4,𝑐 7,𝑐

2,1 U8 U1,2 U11 ,

T2,1 = U10 U4,𝑐 2,𝑑 4/9,𝑑 2,𝑐

2,1 U3 U1,1 U1 + U10 U2,1/2,2 U8 U1,1 U1 ,

T2,2 = U10 U4,𝑐 2,𝑐 4/9,𝑑 2/7,𝑑 9,𝑐 7,𝑐

2,1 U3 U1,1 U6 + U10 U2,1/2,2 U8 U1,1/1,2 U6 + U10 U2,2 U12 U1,2 U6 ,
(15d)

T2,3 = U10 U4/9,𝑑 7,𝑐 9,𝑐 7,𝑑

2,1/2,2 U8 U1,2 U11 + U10 U2,2 U12 U1,2 U11 ,

T3,1 = U13 U9,𝑐 2,𝑐

2,2 U8 U1,1 U1 ,

T3,2 = U13 U9,𝑐 2/7,𝑑 9,𝑑 7,𝑐

2,2 U8 U1,1/1,2 U6 + U13 U2,2 U12 U1,2 U6 ,

T3,3 = U13 U9,𝑐 7,𝑐 9,𝑑 7,𝑑

2,2 U8 U1,2 U11 + U13 U2,2 U12 U1,2 U11 .
Advances in Mechanical Engineering 11

ky = 100(EI/L3 ) and x1 = 0.3 m

15
10

ky1,1 ky2,1

1010

x1
x2 = 1 − x1
5
10
log10 |Δ|

100

10−5

10−10
0 2 4 6 8 10 12 14 16
𝜔

ky = 100(EI/L3 ) and x1 = 0.3 m ky = 100(EI/L3 ) and x1 = 0.3 m

8 10
10 10

106 108
log10 |Δ|

log10 |Δ|

104 106

102
104

10.6 10.7 10.8 10.9 11 11.1 11.2 11.3 11.4 14 14.05 14.1 14.15 14.2
𝜔
𝜔

Figure 9: fMin1D function determinant of a system consisting of free-free and clamped-clamped beams coupled with two springs. 𝑘𝑦 =
100(𝐸𝐼/𝐿3 ) and 𝑥1 = 0.3 m.

Herein, U1 , U3 , U5 , U6 , U8 , U10 , U11 , U12 , and to ⏟⏟U

⏟⏟⏟⏟⏟ ⏟⏟Z⏟⏟⏟
all all⏟⏟ = 0. The geometric and material properties
U13 represent the beam segment transfer matrices; 12 × 1212 × 1
U2,𝑑 2,𝑐 4,𝑑 4,𝑐 7,𝑑 7,𝑐 9,𝑑
1,1 , U1,1 , U2,1 , U2,1 , U1,2 , U1,2 , U2,2 , and U9,𝑐2,2 are the (steel) of the multi-level beam system are given as follows:
transfer matrices of a system between two connection 𝐿 = 1 m, 𝑏1 = 𝑏2 = 𝑏3 = 0.02 m, ℎ1 = ℎ2 = ℎ3 = 0.01 m,
points located between two beams (12c). U2/7,𝑑 4/9,𝑑 𝐸1 = 𝐸2 = 𝐸3 = 𝐸 = 2.069 × 1011 N/m2 , and
1,1/1,2 and U2,1/2,2 𝜌1 = 𝜌2 =𝜌3 = 𝜌 = 7850 kg/m3 . Here, 𝑏𝑖 and ℎ𝑖 represent
are the transfer matrices for two systems connected with
three-level through three connection points (12d). Applying the width and height of the beam, respectively. The cross-
sectional area and the moment of inertial of the beam
the boundary conditions, that is, Z1,0 = [0, Θ𝑧 , 0, 𝑄𝑦 ]𝑇1,0 ;
cross-section are 𝐴 𝑖 = 𝑏𝑖 × ℎ𝑖 and 𝐼𝑖 = 𝑏𝑖 × ℎ𝑖3 /12, respectively.
Z6,0 = [0, Θ𝑧 , 0, 𝑄𝑦 ]𝑇6,0 ; Z11,0 = [0, Θ𝑧 , 0, 𝑄𝑦 ]𝑇11,0 , Z5,0 =
For comparison, the lowest three “exact” dimensionless
[0, Θ𝑧 , 0, 𝑄𝑦 ]𝑇5,0 , Z10,0 = [0, Θ𝑧 , 0, 𝑄𝑦 ]𝑇10,0 , and Z13,0 = frequency values 𝜔 for the single pinned-pinned beam
[0, Θ𝑧 , 0, 𝑄𝑦 ]𝑇13,0 , the overall transfer equation (15a) reduced that might be found in structural or vibration text books
12 Advances in Mechanical Engineering

16 16

14 14

12 12

10 10

8 8
𝜔

𝜔
6 6

4 4

2 2

0 0
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5
x1 (m) x1 (m)

(a) (b)

Figure 10: Dimensionless frequency parameter 𝜔 values as a function of the spring locations (𝑥1 = 0 → 0.5 m) for a system consisting of
free-fee and clamped-clamped beams coupled by two springs. (a) 𝑘𝑦 = 100(𝐸𝐼/𝐿3 ); (b) 𝑘𝑦 = 1000(𝐸𝐼/𝐿3 ).

Transfer direction

1 3 5
0 0 Level 1
Z1,0 Z1,2 Z2,3 Z3,4 Z4,5 Z5,0 Chain 1

ky1,1 2 ky2,1 4
c1,1 c2,1

6 8 10
0 0 Level 2
Z6,0 Z6,2 Z2,8 Z8,4 Z4,10 Z10,0 Chain 2
or or or or
Z6,7 Z7,8 Z8,9 Z9,10
ky1,2 7 ky2,2 9
c1,2 c2,2
11 12 13
0 0 Level 3
Z11,0 Z11,7 Z7,12 Z12,9 Z9,13 Z13,0 Chain 3
x1 L − 2x1 x1

Figure 11: Chain TMM-MS: state vectors and transfer direction of pinned-pinned multi-level beam connected by spring/dashpot systems.

1010

105

100

10−5
log10 |Δ|

10−10

L 10−15

10−20

10−25

10−30
0 1 2 3 4 5 6 7 8 9 10 11 12
𝜔

Figure 12: fMin1D function determinant to evaluate the lowest three 𝜔 for undamped, uncoupled pinned-pinned three-beam.
Advances in Mechanical Engineering 13

ky1,1 ky2,1
10
ky1,2 ky2,2 9
8

x1 7

x2 = 1 − x1 6
5

𝜔
L
4
3
9.54
2
9.52
1
9.5 0
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5
9.48
𝜔

x1 (m)
9.46 (a) ky = 25(EI/L3 )
9.44

9.42 1010
0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5
x1 (m)
105
(a) ky = 25(EI/L3 )
log10 |Δ|

6.6
100
6.55
6.5
6.45 10−5

6.4
𝜔

6.35 10−10
0 2 4 6 8 10 12
6.3
𝜔
6.25
(b) ky = 25(EI/L3 ) and x1 = 0.25 m
6.2
0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5
x1 (m)

(a) ky = 25(EI/L3 )
102

100 100

10−2 10−2
log10 |Δ|
log10 |Δ|

10−4 10−4

10−6 10−6

10−8 9.41 9.42 9.43 9.44 9.45 9.46 9.47 9.48

6.25 6.3 6.35 6.4 6.45 6.5 6.55 6.6 𝜔
𝜔
(b) ky = 25(EI/L3 ) and x1 = 0.25 m
(b) ky = 25(EI/L3 ) and x1 = 0.25 m

Figure 13: (a) The lowest three 𝜔 values as a function of the spring systems locations (𝑥1 = 0 → 0.5 m) for undamped, coupled pinned-pinned
three-beam; (b) fMin1D function determinant to evaluate the dimensionless frequency parameter 𝜔 for 𝑘𝑦 = 25(𝐸𝐼/𝐿3 ) and 𝑥1 = 0.25 m.
14 Advances in Mechanical Engineering

ky1,1 ky2,1
c1,1 c2,1

ky1,2 ky2,2
c1,2 c2,2

x1
x2 = 1 − x1

2000 1000

1500 800
600
1000
400
500
200
𝜆r = 𝛿

𝜆r = 𝛿

0 0

−500 −200
−400
−1000
−600
−1500
−800
−2000 −1000
−200 0 200 400 600 800 1000 1200 1400 1600 −200 0 200 400 600 800 1000 1200 1400 1600
𝜆i = 𝜔 (rad/s) 𝜆i = 𝜔 (rad/s)
(a) (b)
500 600
400
500
300
200 400

100
300
𝜆r = 𝛿
𝜆r = 𝛿

0
200
−100
−200 100
−300
0
−400
−500 −100
−200 0 200 400 600 800 1000 1200 1400 1600 −200 0 200 400 600 800 1000 1200 1400 1600
i i
𝜆 = 𝜔 (rad/s) 𝜆 = 𝜔 (rad/s)

(c) (d)

Figure 14: fMin2D first step: damping (𝜆𝑟 = 𝛿) range scanning of a specific frequency band (𝜆𝑖 = 𝜔 ≤ 1500 rad/sec) for damped, coupled
pinned-pinned three-beam, (a) [−2000, 2000], (b) [−1000, 1000], (c) [−500, 500], and (d) [−100, 600].
Advances in Mechanical Engineering 15

the order of the frequency increases, the shape of the

8 frequency vibration versus the spring systems location
resembles a half or full sine wave with insignificant variation
6 in the frequency ranges. The strong ability of fMin1D
4 algorithm is shown in Figure 13(b) by capturing the very
narrow regions of lower peaks.
log10 |Δ|

2 The effect of the dashpot is included in the system

0 making the model more complicated. The two spring/dashpot
systems coupled with three-beam are located at position 𝑥1 =
−2 0.25 m. The spring/dashpot system in parallel is similar to
−4 viscoelastic material model. The given values of damping are
600 𝑐1,1 = 𝑐2,1 = 𝑐1,2 = 𝑐2,2 = 𝑐 = 5 √𝐸𝐼 𝑚/𝐿2 . Next, it is
400 1500
200 1000 the turn to implement fMin2D algorithm [17] for complex
0 500
𝜆r= −200 −500 0 eigenproblem solution. For undamped given system as it is
𝛿 (rad/ s)
𝜆 =𝜔
i
computed above, 𝜔 values do not exceed more than 10, which
is equivalent to (𝜆𝑖 = 𝜔 ≤ 1500 rad/sec). However, the
damping (𝜆𝑟 = 𝛿) is unknown in range. Therefore, within
the required system frequency band, a couple scanning
6 grids are demanded to understand and explore the damping
4 ranges of the system. Figure 14 shows different scanning
2 grids for damping ranges as follows: (a) [−2000, 2000], (b)
0 [−1000, 1000], (c) [−500, 500], and (d) [−100, 600]. It is
log10 |Δ|

−2 obviously seen that there are 9 promising regions distributed

−4
−6
−8
−10
600
400
2000
𝜆r =𝛿 0 0
500
(rad/
𝜆 =𝜔
i
Figure 15: fMin2D second step: function determinant to evaluate
the eigenvalues (𝜆 = −𝜆𝑟 ± 𝑖𝜆𝑖 ) for a damped, coupled pinned-
pinned three-beam.
are 3.1416, 6.2832 and 9.4248. MS-TMM results for
undamped, uncoupled pinned-pinned three-beam are
3.141592, 6.28318, and 9.424777 as shown in Figure 12
using fMin1D algorithm. For undamped elastically coupled
three-beam, Figure 13 shows that 𝜔 values were plotted
versus the location movement of spring systems. Systems
2 and 4 (𝑘𝑦1,1 = 𝑘𝑦1,2 = 𝑘𝑦 = 25 (𝐸𝐼/𝐿3 )) and systems 7 and
9 (𝑘𝑦1,2 = 𝑘𝑦2,2 = 𝑘𝑦 = 25 (𝐸𝐼/𝐿3 )) are moving opposite to
each other from (𝑥1 = 0 → 0.5 m). The system shows
symmetric and antisymmetric vibrations. For symmetric
vibration, the frequency parameters 𝜔1,1 , 𝜔1,2 , and 𝜔1,3
(where 𝑙 and 𝑝 are the beam number and vibration mode,
resp.) are constant throughout the spring systems movement
as seen in Figure 13(a). The antisymmetric of the first mode
vibration 𝜔2,1 and 𝜔3,1 started with equally values as 𝜔1,1 at
𝑥1 = 0 (spring systems are rigidly mounted at the support
ends) and increased significantly at 𝑥1 = 0.5 m (both spring
systems are in the same position). However, it is evident
that 𝜔2,1 and 𝜔3,1 are sensitive to the variation of spring
positions. The results presented of the other antisymmetric
vibration modes, 𝜔2,2 , 𝜔3,2 , 𝜔2,3 , and 𝜔3,3 , indicate that as
randomly on the surface. These regions represent the complex
eigenvalues of the system. In the following searching complex
roots, the damping range (Figure 14(d)) is considered. MS-
TMM for damped, coupled three-beam results are shown in
Figure 15 and Table 2 after using fMin2D algorithm coded
1500 under Matlab environment with the following input data:
1000
s) ([−1,
⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟ [−100, 600], 𝑁𝑥0 = 35, 𝑁𝑦0 = 20, 𝜀𝑥 = 10−8 ,
1500], ⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟
𝜆𝑖 𝜆𝑟
−8
𝜀𝑦 = 10 ).
Chain MS-TMM for undamped/damped coupled multi-
level beam with the suggested scenario and the two novel
recursive scanning algorithms provide a closed-form solu-
tion, not only presents the principles of the vibration problem
but also shed light on practical applications. Since the
solution is almost exact, it allows a complete understanding
of a problem.
6. Conclusions
Starting from the principle of mechanics and the elementary
formulations for the flexible beam, the free vibration analysis
of laterally vibrating system made up of a multi-level Euler-
Bernoulli beam to which spring/dashpot systems are attached
across the span is performed using one of the Transfer Matrix
Method of Linear Multibody Systems (MS-TMM) scenarios.
Although the number of coupling springs or spring/dashpot
systems considered in the examples given was limited to
three, there is no inherent difficulty in extending the current
method to solve the problems of vibration of systems con-
sisting of any number of uniform/nonuniform beams with
different boundary conditions and coupled with any number
of spring/dashpot systems. The numerical results obtained to
reveal that the eigenfrequencies calculated by this method are
in very good agreement with those obtained by the published
literature. Moreover, MS-TMM is encouraging for further
16 Advances in Mechanical Engineering

Table 2: Chain MS-TMM eigenvalues results of damped, coupled Journal of Sound and Vibration, vol. 269, no. 1-2, pp. 431–438,
pinned-pinned three-beam. 2004.
[8] M. Abu-Hilal and N. Beithou, “Free transverse vibrations of a
𝜆 = −𝜆𝑟 ± 𝑖𝜆𝑖 Chain MS-TMM results using fMin2D
triple-beam system,” Journal of Mechanical Engineering, vol. 58,
𝜆𝑖 (rad/sec)
no. 1, pp. 30–50, 2007.
𝜆1 −2.463512 × 102 + 4.010226 × 101 𝑖 [9] W. Schiehlen, Multibody Systems Handbook, Springer, Berlin,
𝜆2 −2.074369 × 10−9 + 1.462698 × 102 𝑖 Germany, 1990.
𝜆3 −7.409510 × 101 + 1.652322 × 102 𝑖 [10] W. Schiehlen, “Multibody system dynamics: roots and perspec-
𝜆4 −4.577982 × 102 + 4.615502 × 102 𝑖 tives,” Multibody System Dynamics, vol. 1, no. 2, pp. 149–188,
1997.
𝜆5 −2.074369 × 10−9 + 5.850790 × 102 𝑖
[11] J. Wittenburg, Dynamics of Systems of Rigid Bodies, Edited by:
𝜆6 −1.484833 × 102 + 5.861790 × 102 𝑖 B. G. Teubner, Stuttgart, Germany, 1977.
𝜆7 −2.159398 × 102 + 1.243120 × 103 𝑖 [12] J. Wittenburg, Dynamics of Multibody Systems, Springer, Berlin,
𝜆8 −7.448404 × 101 + 1.311750 × 103 𝑖 Germany, 2nd edition, 2008.
𝜆9 −2.074369 × 10−9 + 1.316428 × 103 𝑖 [13] A. A. Shabana, Dynamics of Multibody Systems, Cambridge
University Press, New York, NY, USA, 3rd edition, 2010.
[14] A. A. Shabana, Computational Dynamics, John Wiley & Sons,
investigations of more complex multibody systems of this New York, NY, USA, 3rd edition, 2010.
type with rigid bodies due to simplicity in the formulation [15] A. A. Shabana, “Flexible multibody dynamics: review of past
of the transfer equation, being systematic to apply, and being and recent developments,” Multibody System Dynamics, vol. 1,
easy to program. no. 2, pp. 189–222, 1997.
[16] X. Rui, L. Yun, Y. Lu, B. He, and G. Wang, Transfer Matrix
Method of Multibody System and Its Application, Science Press,
Conflict of Interests Beijing, China, 2008, (Chinese).
The authors declare that there is no conflict of interests [17] D. Bestle, L. K. Abbas, and X. Rui, “Recursive eigenvalue
search algorithm for transfer matrix method of linear flexible
regarding the publication of this paper.
multibody systems,” Multibody System Dynamics, 2013.
[18] L. K. Abbas, M. J. Li, and X. Rui, “Transfer matrix method for the
Acknowledgments determination of the natural vibration characteristics of realistic
thrusting launch vehicle—part I,” Mathematical Problems in
The research was supported by the Research Fund for Engineering, vol. 2013, Article ID 764673, 16 pages, 2013.
the Doctoral Program of Higher Education of China
(20113219110025), the Natural Science Foundation of China
Government (11102089), and the Program for New Century
Excellent Talents in University (NCET-10-0075).

References
[1] S. Inceoğlu and M. Gürgöze, “Bending vibrations of beams
coupled by several double spring-mass systems,” Journal of
Sound and Vibration, vol. 243, no. 2, pp. 370–379, 2001.
[2] S. Kukla, “Free vibration of the system of two beams connected
by many translational springs,” Journal of Sound and Vibration,
vol. 172, no. 1, pp. 130–135, 1994.
[3] S. Kukla and B. Skalmierski, “Free vibration of a system
composed of two beams separated by an elastic layer,” Journal
of Theoretical and Applied Mechanics, vol. 32, no. 3, pp. 581–590,
1994.
[4] Z. Oniszczuk, “Free transverse vibrations of elastically con-
nected simply supported double-beam complex system,” Jour-
nal of Sound and Vibration, vol. 232, no. 2, pp. 387–403, 2000.
[5] Z. Oniszczuk, “Forced transverse vibrations of an elastically
connected complex simply supported double-beam system,”
Journal of Sound and Vibration, vol. 264, no. 2, pp. 273–286,
2003.
[6] H. V. Vu, A. M. Ordóñez, and B. H. Karnopp, “Vibration of a
double-beam system,” Journal of Sound and Vibration, vol. 229,
no. 4, pp. 807–822, 2000.
[7] M. Gürgöze and H. Erol, “On laterally vibrating beams carrying
tip masses, coupled by several double spring-mass systems,”