Issam 2
Issam 2
net/publication/354518966
CITATIONS READS
3 179
5 authors, including:
All content following this page was uploaded by Hatim Fakhreddine on 13 September 2021.
Abstract
The main objective of this work is to study the geometrically non-linear free vibration of
stepped beams carrying multiple masses. These beams are studied on the basis of the Euler-
Bernoulli beam theory and the Von Karman geometrical nonlinearity assumptions. The dis-
crete expressions for the beam total strain and kinetic energies are derived. By applying
Hamilton's principle, the problem is reduced to a non-linear algebraic system solved by an
approximate method (the so-called second formulation). A parametric study is performed to
explore the effect of non-linearity on the dynamic behaviour of stepped beams with several
added masses. The free vibration case is discussed by considering three types of stepped
beams which differ in the cross-section type.
5331
I. El Hantati, A. Adri, H. Fakhreddine, S. Rifai, R. Benamar
1 INTRODUCTION
Stepped beams are commonly encountered in practical structures and extensively used in
many domains, including mechanical and civil engineering, such as high-rise buildings, robot
arms, etc. These structures, often subjected during their lifetime to transverse vibrations, may
contain concentric elements that can be assimilated to point masses or to linear or rotational
springs. Considerable work has been done to study the dynamic behaviour of these types of
beams by means of linear theory, making the analysis relatively easy, but leading to inaccu-
rate results when the beams are subjected to large vibration amplitudes. In such situation, non-
linear analysis is needed to obtain with accuracy the deformation of the beam for each ampli-
tude level considered. A review of the literature goes back to Naguleswaran who presents an
overview of the existing studies performed to analyse stepped beams using linear
approaches[1]. El Hantati[2], [3] presents a literature review of works related to non-uniform
beams in the case of linear and non-linear vibrations. A significant number of studies have
been conducted on the dynamic response of single and multi-mass beams, using linear and
non-linear analyses for both free and forced vibrations. These studies have been mostly re-
ported in an extensive literature review presented by Fakhreddine[4]. Regarding non-uniform
beams with concentric masses, many studies have been carried out to analyse their response to
free and forces vibrations, considering both the Euler-Bernoulli and the Timoshenko beam
theories. Karami[5] investigated free vibrations of arbitrary non-uniform Timoshenko beams
with their concentrated mass and rotary inertia and resting on elastic supports based on the
differential quadrature element method. Based on the same method, Torabi et al. studied in [6]
free vibrations of a non-uniform cantilever Timoshenko beam with multiple concentrated
masses. Kohan et al. [7]studied a non-uniform beam carrying a concentrated masses and sup-
ported by linear and rotational springs at both sides. The analytical formulation exposed in
this work was based on the Ritz method and the orthogonal polynomials within the frame-
work of the first order shear deformation beam theory. Jong-Shyong[8]used the continuous-
mass transfer matrix method to investigate free vibrations of a multi-step Timoshenko beam
axially loaded and carrying arbitrary concentrated elements. K.Torabi et al [9] investigated
multiple-stepped Bernoulli Euler and Timoshenko beams, carrying concentrated masses with
rotary inertia at arbitrary points based on the same method. Yusuf Yesilce[10]studied free
vibrations of a Timoshenko multiple-step beam under axial load, carrying a number of inter-
mediate lumped masses and rotary inertias based on the Differential transform method and
numerical assembly technique. Farghaly and El-Sayedpresented in[11] parametric studies for
an axially loaded stepped Timoshenko beam carrying several attachments. All of the afore-
mentioned studies were carried out assuming linearity and none of them took into account the
effect of geometrical non-linearity. This work, which is a continuation of the work previously
carried out by Benamar et al.[12]–[19], is intended to contribute to a non-linear modal analy-
sis of structural vibration by studying the geometrically non-linear free vibration of stepped
beams carrying multiple masses using the Euler-Bernoulli beam theory and the Von Karman
geometrical nonlinearity assumptions. The beam total strain and kinetic energies are presented
as discrete expressions and then derived. By applying Hamilton's principle, the problem is
reduced to a non-linear algebraic system solved by an approximate method previously applied
to similar problems (the so-called second formulation) [16]. A parametric study is performed
to explore the effect of non-linearity on the dynamic behaviour of stepped beams with several
added masses. The free vibration case is discussed by considering three types of stepped
beams which differ in the cross-section type.
5332
I. El Hantati, A. Adri, H. Fakhreddine, S. Rifai, R. Benamar
1- General formulation
The transverse free vibration in each span is governed by the following equation:
d 4Wi +x , 4
0 ε i Wi +x , ≅ 0 ;x ⊥0 L(For i=1,2…, 12) (1)
dx 4
Where
υj Sj
ε ij4 ≅ζ i 2 (For j=1,2 and 4) (2)
Ej I j
After putting the parameters of equation (1) into a non-dimensional form, Equation (1) can
be written in the following form:
5333
I. El Hantati, A. Adri, H. Fakhreddine, S. Rifai, R. Benamar
d 4Wi +x* ,0 ε 4W
4 i i +x ,≅ 0 *
; x* ⊥0 1
dx (3)
Where
xj Lj ε ij S I
x j* ≅ ; R j* ≅ ; zj ≅ ;u j ≅ j ;vj ≅ j (4)
L L ε i1 S1 I1
Each step of the beam is characterized by a ratio of sections, moments of inertia and eigen-
value parameters which are denoted by ( uj, vj and zj ).
The solution of Eq. (1) presents the linear modes form which can be expressed at each span
as:
Wi j +x* , ° ∼x ∞ x* ∞ x j * ϒ ½
j 01
*
∫∫ ∫∫
Wi +x* , ® } ¾
∫ } ∫
¯∫ ∫¿
+ , + , +
Wi j +x* , ≅ Aj cosh ε i j L +x* 0 x j 01* , . B j sinh ε ij L +x* 0 x j01* , . C j cos ε i j L +x* 0 x j 01* , . D j sin ε i j L +x* 0 x j 01* , (5) , + ,
In equation (5), the constants Aj ,B j ,C j and D j can be solved from the compatibility condi-
tions and the boundary conditions.
The boundary conditions at the left side:
dWi1 +x* ,
Wi1 +x *
, ≅0 ; *
≅0 (6,7)
*
x ≅0 dx
x* ≅ 0
The compatibility conditions at each change in the cross-section are given by:
Wij +x* , * *
≅Wij . 1 +x* ,
x ≅xj x* ≅ x j*
(10)
dWij +x* , dWij . 1 +x* ,
*
≅ *
dx dx
x* ≅ x j* x* ≅ x j*
(11)
d 2Wij +x* , d 2Wij . 1 +x* ,
EI j *2
≅ EI j . 1 *2
dx dx
x* ≅ x j* x* ≅ x j*
(12)
5334
I. El Hantati, A. Adri, H. Fakhreddine, S. Rifai, R. Benamar
The compatibility conditions for an attached mass with a rotary inertia are given by:
W ji ( x* ) = W(j+1)i ( x* ) (14)
x* ≅ xm
*
x* ≅ xm
*
dW ji ( x* ) dW(j+1)i ( x* )
= (15)
dx* x* ≅ xm
* dx* x* ≅ xm
*
d 2W ji ( x* ) d 2W(j+1)i ( x* ) dW ji ( x* )
= + J j ω2 (16)
dx*2 x* ≅ xm
* dx*2 x* ≅ xm
* dx* x* ≅ xm
*
d 3W ji ( x* ) d 3W(j+1)i ( x* ) M j ω 2W ji ( x* )
= - (17)
dx*3 x* ≅ xm
* dx*3 x* ≅ xm
* EI x* ≅ xm
*
Non-trivial solutions corresponding to the natural frequencies are derived by stating that
the determinant of the homogeneous system obtained from the boundary and compatibility
conditions vanishes. The transcendental equation obtained is solved iteratively by the New-
ton-Raphson method and the constants Aj ,B j ,C j and D j are determined by classical algebra.
2.3. Non-linear formulation
T=
2 0
S +x ,¨
¨ δt ¸
dx+ M c ¨
2 © δt
¸
¹
(18)
© ¹
The total deformation energy V is the sum of the deformation energy due to bending and
the axial deformation energy due to the non-linear stretching forces induced by the large de-
flections, respectively denoted by Vf and Va and written as follows:
V = V f + Va (19)
· § d wj ·
2
1 L § d 2 wi
V f = ⌡E I +x ,¨ 2 ¸ ¨¨ 2 ¸¸ dx (20)
2 0 © dx ¹ © dx ¹
L
1 Nx2 dx
Va = ⌡ (21)
2 E 0 S +x ,
Where N x present the non-linear stretching forces in the beam. The dynamic behaviour of
the system is studied by applying Hamilton's principle, stated formally as follows:
2ω
δ ⌡ +V -T , dt = 0
t
(22)
0
5335
I. El Hantati, A. Adri, H. Fakhreddine, S. Rifai, R. Benamar
Assuming that the system has a harmonic motion, the transverse displacement can be writ-
ten as:
w +x,t ,= aiWi +x , sin +ωt , +For i = 1,2…,n , (23)
The linear modes, the basic function contribution coefficients and the associated frequency
are denoted by Wi +x , , ai and ζ . After Substitution of equation (23) into equations (18, 20
and 21), the expressions for the kinetic energy, the axial deformation energy and the bending
strain energy can be written as follows:
1 2 2
T≅ ζ ai ai + cos +ζ t ,, mij ; (24)
2
1 1 2
Va ≅ ai a j ak al (sin +ζ t ,) 4 bijkl ; V f ≅ ai ai +sin +ζ t ,, kij (25,26)
2 2
Where kij , bijkl and mij denote the stiffness tensor due to Vf , the non-linearity tensor due to
Va and the mass tensor attributable to T . Equation (23) is extended by adding equations (24 to
26) into it, which leads to the following new form:
2 ai kir . 3 ai a j ak bijkr0 2 ζ 2 ai mir ≅ 0 ; r ≅ 1,...., n (28)
In which +kir ,, +mir , , +a j ak bijkl , and +ai , are the general terms of the matrices
⊥K ,[M ], ª¬ B +∼Aϒ,º¼ and ∼Aϒ. Setting the dimensionless parameters as follows:
Wi +x* , ≅ h ; ζ 2 ≅ E I1 M ij
; ≅ υ S1h0 2 (30,31,32)
Wi *
+x , 0 ζ 2 υ S1L4
* *
M ij *
E I1h0 2 * E I1h0 2
kij ≅ 3
k ij ; bijkl ≅ 3
bijkl * (33,34)
L L
After substitution of the dimensionless parameters, Equation (29) becomes:
3
ª¬ K * º¼ ∼Aϒ. ª¬ B* +∼Aϒ ,º¼ ∼Aϒ0 ζ *2 [ M * ]∼Aϒ≅ 0 (35)
2
In order to determine the frequency ω and the column vector of the contribution coeffi-
cients {A}, which are the unknowns of equation (35), the so-called second formulation, pre-
viously developed by EL KADIRI et al. in [16], is used to solve equation (35).
NUMERICAL RESULTS AND DISCUSSION:
In order to validate the numerical method mentioned above using the available data, the re-
sults corresponding to a uniform beam carrying a point mass at the middle are compared to
those obtained in [20], where the section ratios of the the 1st and 2nd steps, designated by u1
and u2, are equal to u1= u2= 1 and the mass magnitude M c ≅ 1
5336
I. El Hantati, A. Adri, H. Fakhreddine, S. Rifai, R. Benamar
Wmax
r Present work Reference [20] Rel Diff[20]
0.4 1.0035 1.0034 0.01%
0.8 1.0142 1.0137 0.05%
1 1.0221 1.0213 0.08%
2 1.0851 1.0814 0.34%
Table 1: §¨ ζ NL ζ ·¸ for clamped beams, in which a concentred mass is located at its middle with r = I S .
© ¹
The results show an excellent agreement, with a relative difference not exceeding 0, 34%.
The cases considered in all figures below are for three stepped beams with two cross-section
changes and carrying a concentrated mass at the middle as shown in figure (1). The mass
magnitude M c ≅ 1 , the cross-section ratios at the 1st and 2nd step u1 and u2 equal to u1=0.8 and
u2= 0.6. The three beams keep the same ratios u1 and u2, but each of them is expressed dif-
ferently. The figures below show the effect of the cross-section variations on the frequency
ratio.
Figure (3): Backbone curves of the three stepped beams studied, corresponding to the first nonlinear mode.
Figure (4): The normalized first non-linear mode of beam (a) for different values of a1.
5337
I. El Hantati, A. Adri, H. Fakhreddine, S. Rifai, R. Benamar
Figure (5): The non-linear curvature distribution of beam (a) associated to the normalized first non-linear
mode for different values of a1.
Figure (6): The normalized first non-linear mode of beam (b) for different values of a1.
5338
I. El Hantati, A. Adri, H. Fakhreddine, S. Rifai, R. Benamar
Figure (7): The non-linear curvature distribution of beam (b) associated to the normalized first non-linear
mode for different values of a1.
Figure (8): The normalized first non-linear mode of beam (c) for different values of a1.
Figure (9): The non-linear curvature distribution of beam (c) associated to the normalized first non-linear
mode for different values of a1.
Curvature
beam Non-linear theory Linear theory percentage correction %
a 49,33 61,43 19,69
b 54,18 81,44 33,47
c 58,56 70,12 19,74
Table 2: The percentage correction between the curvatures estimated at the right end via the linear and nonlinear
theories for a1=2.
The backbone curves in Figure (3) show that by increasing the maximum non-dimensional
amplitude W*max, the frequency ratios also increase. Also, the hardening effect of non-
linearity is more accentuated in beam (b) compared to the other. It can be noticed that the ge-
ometrical non-linearity is not only related to the section ratio but also to the way it varies. For
instance, the results obtained from Table (2) show that the percentage correction of the linear
and non-linear bending moment of beam (a) at the right end with a contribution coefficient
5339
I. El Hantati, A. Adri, H. Fakhreddine, S. Rifai, R. Benamar
a1=2 is 19.69%, and remains higher than the two others, where the percentage correction of
beam (c) and beam (a) are respectively 19.74% and 19.69%. Once more, these results reveal
that the assumption of neglecting geometrical non-linearity for the same beam can be mislead-
ing.
2 CONCLUSION
The geometrically non-linear transverse vibrations of multi-stepped fully clamped beams
carrying masses at the middle was studied analytically based on the Euler-Bernoulli's beams
theory and the Von Karman's nonlinearity assumptions. The solution of the linear problem
was obtained for the stepped beams considered. After determining the linear mode shapes,
these were employed as basic functions for non-linear vibration analysis. To solve the non-
linear algebraic system, the approximate method called the second formulation was applied. A
comparison was performed between the result obtained from the numerical method used in
the present work and those of Reference [18]. This comparison showed an excellent agree-
ment. Then, the non-linear dynamic behaviour of multi-stepped beams in the free vibration
case was examined by considering three types of fully clamped multi-stepped beams carrying
masses at the middle. The results were illustrated by backbone curves, amplitude dependent
non-linear modes and curvatures, showing a clear hardening non-linear behaviour. The as-
sumption of neglecting the geometrical non-linearity is also illustrated and discussed.
3 REFERENCES
5340
I. El Hantati, A. Adri, H. Fakhreddine, S. Rifai, R. Benamar
5341