Engineering Structures 196 (2019) 109288

Contents lists available at ScienceDirect

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

Analytical and experimental studies on dynamic instability of simply T

supported rectangular plates with arbitrary concentrated masses
Zilin Zhonga, Airong Liua, , Yong-Lin Pia,b, Jian Dengc, Hanwen Lua, Sen Lia
⁎

a
Guangzhou University-Tamkang University Joint Research Centre for Engineering Structure Disaster Prevention and Control, Guangzhou University, Guangzhou 510006,
China
b
Centre for Infrastructure Engineering and Safety, The University of New South Wales, NSW 2052, Australia
c
Department of Civil Engineering, Lakehead University, Ontario P7B 5E1, Canada

ARTICLE INFO ABSTRACT

Keywords: This paper presents innovative analytical and experimental investigations of transverse dynamic instability of a
Dynamic instability simply supported rectangular plate attached with arbitrary concentrated masses owing to parametric resonance
Parametric resonance excited by an in-plane uniformly distributed periodic load along two opposite edges, which has not been re-
Simply supported rectangular plate ported in the literature. Based on the von-Kármán large deflection theory of thin plate, differential equations of
Arbitrary concentrated mass
transverse motion are established by using the Galerkin method. The analytical results of dynamic instability
Modal displacement
Swept frequency tests
regions and nonlinear response curves of the plate having arbitrary concentrated masses for various instability
modes are obtained. Tests investigating the influence of concentrated mass on the out-of-plane dynamic in-
stability of various modes of a plate under the in-plane periodical loading presented in the paper are first time
reported in the literature. Independent swept frequency tests are carried out to investigate instability frequency-
amplitude regions for transverse dynamic instability of the plate and the test results agree well with the theo-
retical counterparts. It is shown that concentrated masses significantly influence out-of-plane dynamic instability
of the plate. The investigation leads to the following novel findings: (1) When the concentrated masses are
located at positions corresponding to non-zero modal displacements, the masses increase the rate of energy
dissipation and damping ratio of the plate, leading to an increase of the critical excitation amplitude for dynamic
instability of the plate, but to a decrease of the natural frequencies of the plate and the critical excitation
frequencies for dynamic instability of the plate. (2) Under the same excitation amplitude, the excitation fre-
quencies and the widths of excitation frequency region for dynamic instability of the plate as well as excitation
frequency intervals for nonlinear dynamic instability of the plate decrease with an increase of the concentrated
masses, which are located at positions corresponding to non-zero modal displacements. (3) When the con-
centrated masses are located at positions corresponding to zero modal displacements, the concentrated masses
almost do not affect the critical excitation frequency and the widths of excitation frequency region for dynamic
instability of the plate as well as excitation frequency intervals for nonlinear dynamic instability of the plate. (4)
The widths of excitation frequency regions for dynamic instability of the plate as well as excitation frequency
intervals for nonlinear dynamic instability of the plate increases when masses move away from the position of
the largest modal displacement of the plate.

1. Introduction frequency and amplitude. In particular, the dynamic instability may

occur below the static critical load of the plate under in-plane com-
Plates have been widely used in modern structural systems, because pressions over wide ranges of excitation frequencies. The dynamic in-
the desired performance can be achieved by properly adjusting the stability of rectangular thin plates without concentrated masses has
shape, materials, and external loadings of the structures [1]. Vibration been extensively studied [2–4]. On the other hand, plates with con-
analysis of plates has been an active research subject in the field of centrated masses can be found in many practical engineering structures,
engineering. It is known that plates subjected to in-plane periodic loads such as cladding panels in building structures, bridge deck and slabs
may suffer dynamic instability for certain combinations of excitation with additional components in watercraft structures. Investigations on

⁎
Corresponding author.
E-mail address: [email protected] (A. Liu).

https://doi.org/10.1016/j.engstruct.2019.109288
Received 25 January 2019; Received in revised form 4 June 2019; Accepted 7 June 2019
0141-0296/ © 2019 Elsevier Ltd. All rights reserved.
Z. Zhong, et al. Engineering Structures 196 (2019) 109288

free vibrations, dynamic responses, and dynamic stability of plates with the Mathieu-Hill equation becomes more complicated and the coupled
concentrated masses are briefly reviewed as follows. coefficients of sine and cosine terms of the transverse displacement
Free vibration analysis of rectangular plates with a concentrated function cannot be solved by the conventional coefficients separation
mass or line-concentrated masses was investigated by Cha [5] and Li method. There is no knowledge about how the size and position of the
[6]. Chai [7] used the Rayleigh-energy method to obtain the theoretical concentrated masses influence the out-of-plane nonlinear dynamic in-
formulation of free vibrations of plates with and without a concentrated stability of plate under the in-plane periodical loading available in the
mass and the different conditions of the plate were analyzed. An ap- literature. It is also not known that how the aspect ratio of a plate in-
proximate solution for the influence of concentrated mass on large teracts with the mass position and how the interaction influence the
amplitude and free flexural vibration of plates were obtained by Gu- dynamic instability of the plate. A novel method is needed for solving
tierrez and Laura [8]. In addition to rectangular plates, the free vibra- the coupled problem of the critical excitation frequency to determine
tion of a circular plate with arbitrary mounted masses was studied in the accurate boundaries of the stability regions for the plate with
[9] using an integral equation method. It was shown in [10] that the concentrated masses. The new analytical solutions for dynamic in-
characteristics of free vibrations of a cantilever anisotropic plate with a stability of a plate with masses based on the method can be used to
concentrated mass can be solved by approximate methods. Wu and Luo solve the following important issues in the practical control of the dy-
[11] used an analytical-and-numerical-combined method in the free namic instability of the plate: How the magnitude and position of the
vibration analysis of a rectangular plate with any number of point concentrated masses influence the out-of-plane dynamic instability and
masses and translational springs. More recently, they [12] also de- nonlinear traction instability of a plate under the in-plane periodical
termined the natural frequencies and mode shapes for a uniform rec- loading; and how the aspect ratio of a plate influences the out-of-plate
tangular plate with concentrated masses and translational springs. dynamic instability region, the excitation frequencies and the nonlinear
Dynamic responses of an elastic plate with a concentrated mass are traction instability of the plate with concentrated mass under the in-
another topic being extensively investigated [13]. For example, Doh- plane periodical loading.
erty and Dowell [14] experimentally studied the asymptotic modal In addition, to authors’ knowledge, although experiments on dy-
response of a rectangular plate with concentrated masses, which is namic stability of plates have been reported, the experimental in-
applicable for the measurement of frequencies for a plate with con- vestigations about the effect of concentrated masses on the out-of-plane
centrated masses, even with limited conditions. An effective method parametric resonance instability of various modes for plates under the
was developed to exactly predict the natural frequencies and mode in-plane periodical loading do not appear to be reported in the litera-
shapes of plates with mass, stiffness and damping modifications [15], ture.
because closed form solutions rarely exist. Ostachowicz et al. [16] This paper, therefore, aims to use analytical and experimental
presented a genetic algorithm search technique for recognition of the methods to investigate the transverse dynamic instability mechanisms
location of a concentrated mass on plates from the changes of natural of simply supported rectangular thin plates with arbitrary concentrated
frequencies. Maiz et al. [17] used Ritz method to analyze the effect of masses under in-plane uniformly distributed periodic loading along two
concentrated masses on the natural frequencies and mode shapes for a opposite edges owing to parametric resonance. The dynamic governing
clamped elliptical plate. equations of motion for various instability modes are derived by the
Earlier research on dynamic stability of plates usually focused on Galerkin method, and dynamic instability equations are established as
problems without concentrated masses [18–24]. For example, Bolotin Mathieu–Hill equations. The critical excitation frequencies-amplitudes
[18] studied the dynamic stability of the simply supported rectangular of the dynamic instability region and the stable solution as well as the
plate without concentrated mass under an in-plane periodic uniform unstable solution of nonlinear response curves are solved. More im-
load and obtained the critical boundary frequencies of the excitation portantly, the paper explores the effects of the size, positions and dis-
load. Pierre [20] presented the incremental harmonic balance method tribution of concentrated masses as well as the aspect ratio on the cri-
to solve the unstable region and nonlinear response curve for the simply tical excitation frequencies and amplitudes of the dynamic instability
supported rectangular plate under in-plane periodic uniform load. region for a plate in the first and second instability modes. In addition,
Mathiua equation of laminated composite thin plates under in-plane the effect of nonlinearity on dynamic instability and the influence of the
periodic uniform load was derived by Darabi and Ganesan [24], and the excitation amplitude on the nonlinear response are also investigated.
influences of different parameters on dynamic instability of laminated Experimental investigations are carried out to verify the accuracy of
composite thin plates were also analyzed by Darabi and Ganesan [24]. analytical solutions and to independently explore the effects of the size
Free vibration and dynamic stability of various plates were studied, and positions of concentrated masses on the critical excitation fre-
such as plates on elastic foundation [25]. For a nonlinear elastic base, quencies and amplitudes for the dynamic instability and nonlinear in-
the oscillations of a plate with concentrated masses were first addressed stability response of the plate.
by Bondarev [26]. Eshmatov and Khodzhaev [27] analyzed the dy-
namic stability of a viscoelastic plate with concentrated masses under
rapidly increasing loads. Azarboni et al. [28] employed the Galerkin 2. Derivation of dynamic instability equations
method to analyze the nonlinear dynamic buckling of imperfect rec-
tangular plates with different boundary conditions subjected to various 2.1. Assumption and geometry of the plate
pulse functions. Under thermal environment, the nonlinear dynamic
characteristics and stability of composite orthotropic plate on elastic The transverse dynamic instability of a simply supported rectan-
foundation has been studied by Gao [29]. More recently, the problem of gular thin plate with arbitrary concentrated masses (Fig. 1) induced by
flutter of viscoelastic rectangular plates and cylindrical panels with an in-plane uniformly-distributed periodic load P(t) along two opposite
concentrated masses was studied in a geometrically nonlinear for- edges is investigated in this paper. For convenience of investigation, a
mulation by Eshmatov et al. [30]. right-handed Cartesian system oxyz shown in Fig. 1 is used to facilitate
Although the analytical solutions for the dynamic instability of the analysis, the origin of the system is located at the upper left corner
plates are available in the literature, analytical solutions for the dy- of the plate, and the x-y plane coincides with the mid-plane of the
namic instability of various modes of plates with concentrated masses undeformed plate. The length, width and the thickness of the plate are
have not been reported in the literature hitherto. Differently from the denoted by a, b and h, the mid-plane displacements in the direction x, y,
plate without concentrated masses, when the concentrated masses are and z are denoted by U, V, and W respectively and the concentrated
included in the dynamic stability analysis of a plate, the derivation for mass on the plate is denoted by Mu (Fig. 1). The material of the plate is
various instability modes and critical excitation frequencies based on assumed to be elastic, isotropic and homogeneous.

2
Z. Zhong, et al. Engineering Structures 196 (2019) 109288

the in-plane boundary conditions and transverse boundary conditions

are related to both the transverse displacement W(x,y,t) and the stress
function Ф(x,y,t) [35], so that the in-plane boundary conditions can be
expressed as
2
x y
=0
2 1
= P (t )
y2 h (7)

at x = 0, a, and
2
x y
=0
2
=0
x2 (8)

at y = 0, b, where a and b are the length and width of a plate in x and y

directions (Fig. 1).
Concurrently, the transverse boundary conditions can be expressed
Fig. 1. The mechanical model.
as
2W 2W
W = 0, +µ =0
2.2. Differential equation of motion x2 y2 (9)

at x = 0, a, and
Based on the von-Kármán large deflection theory of thin plate [31],
the equilibrium equations in the z-axial direction of a simply supported 2W 2W
W = 0, +µ =0
rectangular plates with arbitrary concentrated masses under in-plane y2 x2 (10)
uniformly-distributed periodic load P(t) along two opposite edges can
be expressed as at y = 0, b.
The approximate solutions for the transverse displacement W(x,y,t)
4W 2W 2W 4W 2W
and the stress function Ф(x,y,t) of governing equations can be expressed
D +2 + +m
x4 x2 y2 y4 t2 in terms of shapes of flexural waves as
2W 2W 2W
= Nx + Ny + 2Nxy W (x , y , t ) = Wkl (t ) k (x ) l (y ) (11)
x2 y2 x y (1)
and
and
y2
2N 2N 2W 2 2W 2W (x , y , t ) = pq (t ) Xp (x ) Yq (y ) P (t )
y
+ 2Nx Ny + x
= Eh p q
2h (12)
x2 y2 x y y2 x2 (2)
where Wkl(t) are the time-dependent generalized coordinates, and
where because the in-plane displacements U and V are much smaller Φpq(t) are the load factor functions of t, the subscript k, p, l and q are the
than the out-of-plane displacement W, the in-plane inertia is neglected numbers of half-waves in x and y directions, Xp(x), Yq(y), ϑk(x) and ψl(y)
in the Eqs. (1) and (2) [18,24,32,33]; Nx, Ny are membrane stresses per are the eigenfunctions of free transverse vibration in the x and y di-
unit length in the direction of x and y, Nxy is membrane shear stress in rections of the plate and they are defined as
the direction of y within the xoy plane and Nx, Ny, and Nxy are given by
x x x x
2 2 2 Xp (x ) = cosh p cos p p sinh p sin p ,
Nx = h , Ny = h , Nxy = h a a a a
y2 x 2 x y (3)
(13)
3 2
in which Φ(x,y,t) is the stress function; D = Eh /12(1 − μ ) is the
y y y y
bending stiffness with Poisson’s ratio μ; in addition, m(x,y) is defined by Yq (y ) = cosh q cos q q sinh q sin q ,
h h h h (14)
N
m (x , y ) = h + Mu (x x u ) (y yu ) k x
u=1 (4) k (x ) = sin
a (15)
with ρ being the density of the plate material, Mu being the con-
centrated mass, δ(x-xu) and δ(y-yu) being the Dirac-delta function [34]. and
Substituting Eqs. (3) into Eqs. (1) and (2), the equilibrium equations l y
can be expressed as l (y ) = sin
b (16)
4W 2W 2W 4W 2W
D +2 + +m respectively, in which αp and αq are the coefficients acquired from the
x4 x2 y2 y4 t2 solution of transcendental equations
2W 2 2 2W 2 2W
=h + 2 1 cos( p, q ) cosh( p, q ) =0 (17)
x2 y2 x2 y2 x y x y (5)
Moreover, in the case of simply supported at all edges, the coeffi-
and cients γp,q can be defined by
4 2 2 4 2W 2 2W 2W cosh( p, q ) cos( p, q)
+2 + =E p, q =
x4 x2 y2 y4 x y x2 y2 (6) sinh( p, q) sin( p, q ) (18)

According to the von-Kármán large deflection theory of thin plate, Substituting Eqs. (11) and (12) into Eqs. (5) and (6) leads to

3
Z. Zhong, et al. Engineering Structures 196 (2019) 109288

(4) (4) Fkl = diag(F11, F12, ...,Fkl ), Epq

D [Wkl ( k l +2 k l + k l )]
N 1 1 1 2 k l
2W
kl
+ h+ Mu (x x u ) (y yu ) k l = diag Epq pq (t ), Epq pq (t ),..., Epq pq (t ) ,
t2
u=1 p q p q p q

= h Wkl
p q
pq (X p k Yq l + Xp k Yq l
1
h
P (t ) Wkl k l
1 1 1 2 k l
Vpq = diag Vpq pq (t ), Vpq pq (t ),..., Vpq pq (t ) ,

)
p q p q p q
2X p k Yq l and the elements in matrices Ukl, Vpq, Dkl, Epq and Fkl are given by
(19)

and Ukl =
D k l
i j ( k4 4
a4
+
l4 4
b4 )+2 k2l2 4 ik
a2b2
G1 H1jl
,
N
(4)
pq (X p Yq + 2Y p Yq + Xp Yq(4) ) = E [Wkl2 ( k k l l k k l l )]
h+
4
ab
u=1
Mu sin ( k
a ) ( x ) sin ( y ) sin ( y )
x u sin
i
a u
l
b u
j
b u
(26)
p q

(20) 2 p q kl 2 ipk 2
q k2 2 ipk
2
q l2 2 ipk
G4 H4jql G2 H3jql G H2jql
a2b2 b2 a2 a2 b2 3
respectively, where for X and ϑ, X′ = dX/dx, ϑ′ = dϑ/dx and so on, Vpq = N
,
while for Y and ψ, Y′ = dY/dy, ψ′ = dψ/dx and so on. h + abh
4 k
Mu sin a ( ) ( ) ( ) ( )
x u sin
i
a
x u sin
l
y
b u
sin
j
y
b u
(27)
Multiplying 4ϑi(x)ψj(y)/(ab) and Xm(x)Yn(y)/(ab) on both sides of u=1

Eqs. (19) and (20) and then integrating Eqs. (19) and (20) with respect k2 2 ik
G1
to x from 0 to a and with respect to y from 0 to b lead to Dkl = a2
,
N

D k l
i j Wkl ( k4 4
+
l4 4
) + 2W k 4l 4 4 ik
kl a4b4 G1 H1
jl
h+
4
ab
u=1
Mu sin ( x ) sin ( x ) sin ( ) sin ( )
k
a u
i
a u
l
y
b u
j
y
b u
(28)
a4 b4
N 4 4 2 2
( ) ( x ) sin ( y ) sin ( y )
4 k i l j 2W p q p q
+ h+ Mu sin x u sin u b u b u t2
kl
Epq = p q
m n + +2 G5mp H5nq
ab
u=1
a a a b a b (29)
2
q k2 2 ipk
2
p l 2 2 ipk and
Wkl pq G2 H3jql + G H2jql
b2 a2 b2 a2 3
p q
k 2l2 4 mkk nll k2l2 4 mkk nll
Fkl = E G6 H6 G7 H7
=h p q kl 2 ipk , a2b2 a2b2 (30)
2 G H4jql
b2 a2 4

p (t ) k2 2
Since Φpq(t) are the pq-independent load factors that can be sepa-
Wkl 2 G1ik rated from matrices Epq and Vpq, substituting Eq. (25) into Eq. (24) and
h a (21)
eliminating Φpq(t) leads to an infinite spatial modes of general ordinary
differential equations with cubic nonlinearity in term of time as
p q
m n
p q
pq ( ) +( )
a
p 4 q 4
b
+2
p q
pq ( )( )G
a
p 2 q 2
b
mp nq
5 H5
V pq
Ẅkl (t ) + Ukl Wkl (t ) + P (t ) Dkl Wkl (t ) + Fkl W3kl (t ) = 0
E (31a)
= E Wkl Wkl ( k2l2 4 mkk nll
G H6
a2b2 6
k2l2 4 mkk nll
a2b2
G7 H7 ) where
pq

(22)
1 1 1 1 k l
It can be shown that eigenfunctions of free transverse vibration in E pq = diag Epq, Epq,..., Epq
the x and y directions of the plate given in Eqs. (13)–(16) have the p q p q p q (31b)
following orthogonal properties:
1 1 1 2 k l

a if p = m b if q = n V pq = diag Vpq, Vpq,..., Vpq

a b
0
Xp Xm dx = , Yq Yn dy = , p q p q p q (31c)
0 if p m 0 0 if q n
a b and the periodic load P(t) is defined as
a 2
if k = i b if l = j
k i dx = , l j dy = 2
0
0 if k i 0
0 if l j
P (t ) = 0 Pcr + 0 Pcr cos( t) (31d)
(23)
with α0 and β0 being the static and dynamic coefficients, and Pcr being
It can be seen from Eqs. (21), (22) and (23) that the static buckling load of the plate for the corresponding mode.
δik= δjl = δmp = δnp = 1 in Eqs. (21) and (22) when i = k, j = l, m = p, However, Eq. (31a) does not include non-conservative forces of the
and n = q; and δik = δjl = δmp = δnp = 0 when i ≠ k, j ≠ l, m ≠ p, plate system, such as damping force. In practice, damping has a sig-
n ≠ q. In addition, the expression of coefficients Gn and Hn (n = 1–7) nificant influence on the dynamic instability of a plate and needs to be
can be written in an integral form as shown in “Appendix 1”. Conse- considered. For this, damping of the plate is assumed to satisfy the
quently, Eqs. (21) and (22) can be rewritten in the matrix form as linear Rayleigh dissipation function [36,37]. By substituting the ex-
pression of the periodic load P(t) and linear damping of a plate, Eq.
Ẅkl (t ) + Ukl + P (t ) Dkl Wkl (t ) + Vpq Wkl (t ) = 0, (24) (31a) can be written as the following Mathieu–Hill equations
V
Epq = Fkl Wkl (t ) Wkl (t ) (25) Ẅkl (t ) + 2 kl Wkl (t ) + 2
kl (Ikl 2 kl cos( t )) Wkl (t ) +
pq
Fkl W3kl (t ) = 0
E pq
where (32)

Wkl (t ) = diag[W11 (t ), W12 (t ), ...,Wkl (t )], Ukl = diag(U11, U12, ...,Ukl ), Dkl where

= diag(D11, D12 , ...,Dkl ), kl = kl 1 0, kl = Uijkl , kl = Ikl [ /2(1 0 )] (33)

4
Z. Zhong, et al. Engineering Structures 196 (2019) 109288

in which Ωkl represents the vibration frequency of the kl mode of a and

rectangular plate under the constant component of the in-plane force,
A = [A11, A12, ...,A21, ...,Akl ]T (39)
ωkl is the natural frequency of the kl mode of the plate, Λkl is the ex-
citation coefficient of the kl mode, Ikl is the identity matrix of kl mode, with Cs and Akl being the Fourier series and harmonic coefficient vec-
ξkl is the damping coefficient of the kl mode, which can be expressed as tors given by

kl kl 1 0 Cs
=
kl
2 (34)
(2i 1) 3 3
= ...,sin t , ...,sin t , sin t , cos t , cos t , ...,
where Δkl is the corresponding damping ratio of the simply supported 2 2 2 2 2
rectangular plate loaded by the constant component of the in-plane (2i 1)
load. It is worth pointing out that the last term of Eq. (32) represents the cos t , ...
2 (40)
nonlinear effects.
In analyses of the effects of the sizes and distributions of con- and
centrated masses on the instability region and nonlinear response, the
Akl = [...,akl,(2i 1) , ...,akl,3, akl,1, bkl,1, bkl,3, ...,bkl,(2i 1) , ...] (41)
logarithmic decrement method can be used in calculation of damping
ratios when the decay records are obtained in hammer tests [36,37,38]. Substituting Eq. (37) into Eq. (32) without the nonlinear terms and
applying the harmonic balance method to set the terms having common
3. Solution of dynamic instability factors sin(nθt/2) and cos(nθt/2) (n = 1, 3, 5,…) to be equal to zero,
leading to a group of homogeneous linear equations as
3.1. Equations of dynamic instability regions
Bkl Akl = 0 (42a)
It has been shown in [32,33] that for the dynamic system described in which
by Eq. (32), only one mode can be excited by an in-plane harmonic load
Bkl,11 Bkl,12
under the boundary condition considered in this investigation. There- Bkl =
fore, each spatial mode contributes independently to dynamic in- Bkl,21 Bkl,22 (42b)
stability of the plate. According to Liapunov Principle [18], Ma-
where matrices Bkl,11, Bkl,12, Bkl,21, and Bkl,22 are given in “Appendix 2”.
thieu–Hill equations (32) can be used to determine the dynamic
The Eq. (42a) can be rewritten in a quadratic form as
instability regions for the plate. The Bolotin’s method of [18] has also
been successful in determining the instability region of out-of-plane 2
Bkl Akl = B1 B2 B3 Akl = 0
dynamic instability of structures such as laminated composite plates, 2 2
kl kl (43)
circular arches, and columns under a periodic load [23,24,36,37,39].
T
The accuracy of the Bolotin’s method has also been verified by ex- By introducing X = [Aklθ/(2Ωkl) Akl] , Eq. (43) can be transformed
periments for example in [36,37] and so it is used in this investigation as
to seek the solution of dynamic instability regions of the plate.
B3 0 0 B3
The periodic solutions of the dynamic system can be found from X=0
0 B1 2 B3 B 2 (44)
periodic T and 2T associated with dynamic load of P(t). Owing to the kl

larger instability amplitude and the wider instability regions, the so- where matrices B1, B2, and B3 are given in “Appendix 3”.
lutions with a periodic 2T have a greater practical significance than the Eq. (44) is a group of linear homogeneous equations and it has non-
solutions with a periodic T. Therefore, in this paper, the solutions with a trivial solutions for X only when the determinant of its coefficient
periodic 2T of dynamic instability regions for a plate with concentrated matrix vanishes as
masses will be focused in this investigation. The standard Mathieu–Hill
equation given by Eq. (32) can be solved to determine the critical fre- B3 0 0 B3
=0
quency of the load for the out-of-plane dynamic instability of the plate. 0 B1 2 kl B3 B 2 (45)
The solutions are herein derived in two steps: the effects of the non-
linear effects are first ignored and then the nonlinear effects are in- from which the maximum and minimum critical excitation frequencies
cluded. When the nonlinear effects are ignored, the solution of Eq. (32) for the transverse parametric resonance of the simply supported rec-
corresponding to period 2T can be expressed as tangular plate with arbitrary concentrated masses can be obtained.
Consequently, the regions of the kl mode with respect to the critical
Wkl (t ) = diag[W11 (t ), W12 (t ), ...,Wkl (t )] (35) excitation frequencies and amplitudes of the periodic load for dynamic
where transverse instability of the plate can be determined.

n n 3.2. Effects of nonlinear terms

Wkl (t ) = akl, n sin t + bkl, n cos t
n = 1,3,5,...
2 2 (36)
The solutions presented in the preceding section do not consider the
in which akl,n and bkl,n (n = (2i − 1), i = 1, 2, 3,…) are coefficients of
effects of the nonlinear terms. The effects of the nonlinear terms are
the Fourier series of the kl mode for the solutions. The solution in Eq.
therefore illustrated in this section using the principal parametric re-
(35) can be further expressed in terms of the Fourier series matrix S and
sonance. In this case, the effects of the rest harmonics can be neglected
harmonic coefficient matrix A of various modes as
and the solutions of Eq. (32) can be expressed as
Wkl (t ) = SA (37)
Wkl (t ) = diag[W11 (t ), W12 (t ), ...,Wkl (t )] (46a)
where
with
Cs 0 . . . 0 .. . 0
0 Cs . . . 0 .. . 0 Wkl (t ) = akl,1 sin t + bkl,1 cos t = C kl A kl
S= ... ... ... ... .. . ... 2 2 (46b)
0 0 . . . Cs .. . 0
... ... ... ... .. . ... where C′s and A′kl are the Fourier series and harmonic coefficient
0 0 ... 0 .. . Cs (38)
k×l vectors given by

5
Z. Zhong, et al. Engineering Structures 196 (2019) 109288

C s = sin t , cos t
2 2 (47)
and
A kl = [akl,1, bkl,1 ]T (48)
Similarly to Eq. (37), the solution of the response including the ef-
fects of nonlinearity for various modes given by Eq. (46a) can be re-
presented as
Wkl (t ) = S A (49)
where

Cs 0 ... 0 .. . 0
0 Cs ... 0 .. . 0
S = ... ... ... ... .. . ... ,
0 0 ... Cs .. . 0
... ... ... ... .. . ...
0 0 ... 0 .. . Cs k×l (50)
Fig. 2. The distribution of concentrated masses.
A = [A 11, A 12, ...,A 21, ...,A kl ]T (51)
By substituting Eq. (49) into Eq. (32) and applying the harmonic used in the investigation. Young’s modulus, density and Poisson ratio of
balance method to set the terms having common factors sin(θt/2) and the plate are assumed to be E = 6.9 × 1010 Pa, ρ = 2700 kg/m3, and
cos(θt/2) to vanish, a group of linear homogeneous equations regarding μ = 0.33 respectively. The analytical solutions of the instability modal
to the coefficient akl,1 (akl,1 = [a11,1, a12,1, …, akl,1]T) and bkl,1 displacements obtained from Eqs. (13) and (14) show that the first
(bkl,1 = [b11,1, b12,1, …, bkl,1]T) can be obtained. Dividing the linear mode instability has only one half-wave in both x and y directions,
homogeneous equations by Ω2kl and letting the determinant of its while the second mode instability has two half-waves in the x direction
coefficient matrix to vanish lead to the following equations and one half-wave in the y direction (Fig. 2). The largest first modal
displacement is located at the point P1(a/2, b/2), while largest second
2
Ikl + kl i kl
Ckl Dkl 0 modal displacements are located at points P2 (a/4, b/2) and (3a/4, b/
Akl2 =0
Ckl Ikl j 2 0 Dkl 2). For convenience of discussion, the regions of non-zero displace-
kl kl (52)
ments of the first and second modes are defined as the region DA, while
where the regions of the zero displacements of the second mode are defined as
the region DB.
3 V pq
To investigate the effects of the sizes and distributions of con-
kl
kl = /2 kl Ikl , Akl2 = akl2 + bkl2 , Ckl = kl , Dkl = 2
Fkl
4 kl E pq (53) centrated masses on dynamic instability of the plate, a mass located at
where Dkl are the nonlinear terms, and Akl is the amplitude of the kl the position P1 (a/2, b/2), which may influence the first mode in-
mode of the plate under excitation of the external in-plane load. From stability, and masses located at the positions P2 (a/4, b/2) and (3a/4, b/
Eq. (52) the stable solution (i = 0, j = 1) and unstable solution (i = 1, 2), which may influence the second mode instability, are also con-
j = 0) of nonlinear responses of the plate with arbitrary concentrated sidered (Fig. 2). In addition, two concentrated masses located at (a/3,
masses can be obtained. However, it is note that the excitation fre- b/3) and (3a/4, 3b/4) corresponding to non-zero modal displacements
quency intervals for the nonlinear instability of the plate is symmetric and two concentrated masses located at (a/2, b/2) and (a/2, 5b/8) on
about θ/2Ωkl = 1. Subsequently, ηkl in Eqs. (52) and (53) reduces to a the symmetry axis (x = a/2) (Fig. 2) are also considered. The analytical
unit matrix Ikj and this makes Eq. (52) not convenient to investigate the solutions for the first mode dynamic instability regions of the plate with
influences of the concentrated mass on the excitation frequencies and a concentrated mass located at different positions obtained from Eq.
their intervals for nonlinear dynamic instability of the plate. For the (45) are shown in Fig. 3 where θ is the excitation frequency, Ω11 is the
convenience of calculation, the ratio of the natural frequencies of the first modal vibration frequency of the plate loaded by the constant
plate with and without concentrated masses is introduced into the di- component of the in-plane load from Eq. (31d) with α0 = 0, β being the
mensionless excitation frequency θ/2Ωkl of Eq. (52) as (ωklθ/2Ωklωkl0) dimensionless excitation coefficient defined by β = β0Pcr/Pmax in which
and Eq. (52) can then be expressed as Pcr = 2106 N, Pmax = 100 N are the maximum loads, and
β0 = 0–0.0475.
It can be observed from Fig. 3 that as the size of the concentrated
2
Ikl + kl i kl Ckl Dkl 0
Akl2 =0
Ckl Ikl kl j 2
kl
0 Dkl
(54) mass increases from 0 kg to 0.204 kg, the width of instability regions
reduces and the positions of instability regions move gradually to the
where κkl=(ωklθ/2Ωklωkl0)Ikl, ωkl is the natural frequency of kl mode lower value of θ/2Ω11 (i.e. excitation frequencies decrease). This in-
for the plate with concentrated masses, ωkl0 is the natural frequency of dicates that an increase of the size of the concentrated mass leads to a
kl mode for the plate without concentrated masses. From Eq. (54) the decrease of the natural frequency of the plate and subsequently to a
stable solution (i = 0, j = 1) and unstable solution (i = 1, j = 0) for the decrease of boundaries of the critical frequencies and an increase of the
comparison of nonlinear response results of the plate with and without critical excitation amplitude that constitute the instability region. For
concentrated masses for various modes can be obtained. example, when the mass is located at the position (a/2, b/2) (Fig. 3(d)),
the minimum excitation amplitude and the corresponding frequency
3.3. Effects of the sizes and distributions of concentrated masses on dynamic causing the first mode instability of the plate without concentrated
instability mass is β = 0.218 and θ/2Ω11 = 1, β = 0.26 and θ/2Ω11 = 0.981 with
a mass of 0.068 kg, β = 0.3 and θ/2 Ω11 = 0.96 with a mass of 0.136 kg
The effects of the sizes and distributions of concentrated masses on and β = 0.34 and θ/2 Ω11 = 0.936 with a mass of 0.204 kg. Hence, the
dynamic instability of a plate are herein investigated. A plate having the concentrated mass can play a stabilized role and mitigate the possibility
length a = 1.2 m, the width b = 1.0 m and the thickness h = 0.002 m is

6
Z. Zhong, et al. Engineering Structures 196 (2019) 109288

2 2

2 2

Fig. 3. First mode dynamic instability regions of the plate: (a) mass at (3a/4, 3b/4), (b) mass at (a/3, b/3), (c) mass at (a/2, 5b/8), (d) mass at (a/2, b/2).

for dynamic instability of the plate. It can also be seen from and dimensionless excitation amplitudes β for the first mode instability
Fig. 3(a)–(d) that the mass located at the point P1 (a/2, b/2) has the of the plate are the lowest and highest ones respectively, and the first
largest influence on the instability regions and that the closer the po- mode instability region of the plate has the smallest width. With the
sition of the mass to the point P1 is, the larger the influence of the mass mass moving away from the center point, the critical frequency of ex-
on the first mode instability of the plate is. citation θ/2Ω11 for instability increases while the excitation amplitude
The analytical solutions for the second-mode dynamic instability β for instability decreases. For example, because the distance of the
regions of the plate obtained from Eq. (45) are presented in position (3a/4, 3b/4) from the center point (a/2, b/2) is larger than that
Fig. 4(a)–(d), where Ω21 is the second modal vibration frequency of the of the position (a/2, 5b/8) from the center point (Fig. 2), the width of
plate loaded by the constant component of in-plane force from Eq. the first mode instability region of the plate with a mass at (3a/4, 3b/4)
(31d) with α0 = 0 and β = β0Pcr/Pmax where Pcr = 2616.6 N, is larger than that of the plate with a mass at (a/2, 5b/8) and so the
Pmax = 100 N, and β0 = 0–0.0382. plate with mass at (3a/4, 3b/4) is more prone to the first mode dynamic
Fig. 4(a) and (b) show that the second mode dynamic instability instability.
regions of the plate with a concentrated mass located in the region DB For the second mode instability regions, the points P2 (a/4, b/2) and
for example at the points (a/2, b/2) and (a/2, 5b/8) (Fig. 2) are almost (3a/4, b/2) correspond to the maximum negative and positive second
the same as those of the plate without a concentrated mass. However, modal displacements (Fig. 2). It can be observed from Fig. 5(d)–(f) that
when the concentrated masses are located in the region DA for example the location of the mass for the smallest width and largest critical ex-
at points (3a/4, 3b/4) and (a/3,b/3) (Fig. 2) as shown in Fig. 4(c) and citation amplitude β of the second mode instability region changes to
(d), the mass influences the second mode instability regions of the plate points P2 (a/4, b/2) and (3a/4, b/2), which is different from that for the
significantly. As a result, the excitation coefficient β for the second first mode instability. In addition, with the increase of the distance of
mode instability of the plate with a concentrated mass in the region DB the mass location from the point P2 (a/4, b/2) or P2 (3a/4, b/2) within
is smaller than those with a mass in the region DA. Hence, for the the corresponding half-wave region, the width of the second mode in-
second mode dynamic instability, when a concentrated mass is in the stability region increases and so the plate becomes more unstable. For
region DB (zero second modal displacements), the plate is more prone to example, the distance of the mass at the points (a/3, b/3) and (3a/4,
dynamic instability than that when a concentrated mass is in the region 3b/4) in the region DA from the points P2 (a/4, b/2) and (3a/4, b/2) are
DA (non-zero second modal displacements). 0.1947 m and 0.25 m respectively. Hence, the plate with a mass at the
The effects of the distribution of concentrated masses on the first point (3a/4, 3b/4) has a larger width of the second mode instability
mode instability regions are shown in Fig. 5(a), (b) and (c) and the region than that with a mass at the point (a/3, b/3). However, a mass
effects on the second mode instability regions are shown in Fig. 5(d), located in the region DB for example at the point (a/2, 5b/8) or the
(e), and (f), where the instability regions are obtained from Eq. (45). point (a/2, b/2) has the same effect on the second mode instability of
It can be seen from Fig. 5(a)–(c) that when mass is at the center the plate. For example, the distance of a mass at points (a/2, b/2) and
point (a/2, b/2) of the plate (Fig. 2), the excitation frequencies θ/2Ω11 (a/2, 5b/8) of the region DB from the points P2 (a/4, b/2) and (3a/4, b/

7
Z. Zhong, et al. Engineering Structures 196 (2019) 109288

2 2

2 2

Fig. 4. Second mode dynamic instability regions of the plate: (a) mass at (a/2, b/2), (b) mass at (a/2, 5b/8), (c) mass at (3a/4, 3b/4), (d) mass at (a/3, b/3).

2) are 0.3 m and 0.325 m respectively, but the width of the second Figs. 7(b) and 5(d), it can be seen that as the aspect ratio a/b decreases
mode instability region is almost the same. The reason is that the region from a/b = 1.2 to a/b = 0.5, the bandwidth of the second mode in-
DB corresponds to the zero second modal displacements, where the stability region and the range of the excitation amplitude coefficient β
effects of concentrated mass on the second mode dynamic instability of also decrease. However, the positions of the second mode instability
the plate vanish. region changes dramatically from the plates with the aspect ratio a/
Three plates having an aspect ratio a/b = 1.2, 0.8 and 0.5 are used b = 1.2 to the plates with a/b = 0.5. The second mode instability region
to investigate the effects of aspect ratios on the first and second modes of the plate having the aspect ratio a/b = 1.2 with the mass at (a/2, b/
dynamic instability of the plates based on the analytical solutions. The 2) is almost the same as that with the mass at (a/2, 5b/8) as shown in
width and thickness of plates are assumed to be b = 1.0 m and Fig. 5(d), while the bandwidth and range of the excitation amplitude
h = 0.002 m. A concentrated mass of 0.068 kg is located at (a/2, b/2) coefficient β and the critical excitation frequency for instability of the
and (a/4, b/2) of each plate. The variations of the critical excitation plate having the aspect ratio a/b = 0.5 with the mass at (a/2, b/2) is
coefficient β with the dimensionless excitation frequency θ/2Ω11 or θ/ larger than those with the mass at (a/2, 5b/8) as shown in Fig. 7(b).
2Ω21 for instabilities of the three plates are shown in Fig. 6, where the However, when a/b = 0.5, the second mode instability region of the
maximum excitation load Pmax = 200 N. It can be observed from Fig. 6 plate with the mass at (a/2, b/2) is almost the same as those with the
that for both first and second modes, the critical excitation coefficient β mass at (3a/4, b/2) and (3a/4, b/2) also as shown in Fig. 7(b). This
of the instability region increases with a decrease of the aspect ratio a/ indicates that the second modal displacements vanish along the line
b, indicating a larger excitation amplitude is needed to produce re- y = b/2 for the plate having an aspect ratio a/b = 0.5. This is different
sonance instability of the plate with a smaller aspect ratio. It can be also from the plate having a/b = 1.2, for which the second modal dis-
found that the range of the critical excitation coefficient β decreases placements vanish along the line x = a/2. It is known that the excita-
with a decrease of the aspect ratio. tion frequency ratio of the instability region decrease when the con-
It can also be seen from Fig. 6 that the influence of the mass position centrated mass moves to the position with a larger modal displacement.
on the dynamic stability of plates is also related to their aspect ratio. For the plate a/b = 0.5, therefore, when the mass is located at positions
The dynamic instability regions of the first and second modes for plates (a/2, 5b/8), (3a/4, 3b/4), or (a/3, b/3), the excitation frequency for the
having an aspect ratio a/b = 0.5 with a concentrated mass of 0.068 kg second dynamic instability of the plate decreases, which leads to the
at different positions are shown in Fig. 7. The plates used in Fig. 5(a) instability region moves towards the direction of the lower frequency
and (d) have the same concentrated mass of 0.068 kg but a different ratio θ/2Ω21.
aspect ratio a/b = 1.2. By comparing Fig. 7(a) with Fig. 5(a), it can be
found that as the aspect ratio a/b decrease from a/b = 1.2 to a/b = 0.5, 3.4. Effects of nonlinearity on dynamic instability
the bandwidth of the first mode instability region, the range of the
excitation amplitude coefficient β, and the excitation frequency ratio θ/ In order to investigate the effects of nonlinearity on dynamic in-
2Ω11 decrease. Hence, the instability regions move towards to the di- stability, the amplitudes of vibration of the plate with a concentrated
rection of lower excitation frequency ratios. From the comparison of mass having a size of 0.204 kg located at (a/3, b/3) (Fig. 2) under the

8
Z. Zhong, et al. Engineering Structures 196 (2019) 109288

Fig. 5. Dynamic instability regions of the plate. First mode instability: (a) mass of 0.068 kg, (b) 0.136 kg, (c) 0.204 kg; second mode instability: (d) mass of 0.068 kg,
(e) 0.136 kg, (f) 0.204 kg.

excitation with the dimensionless parameter β = 0.4, 0.6, 0.8 (i.e. obtained from the values of lowest and highest excitation frequencies
Pcr = 2106 N, Pmax = 100 N, and β0 = 0.019, 0.0285, 0.038 for the first shown in Fig. 8(a) and (b) that the excitation frequencies within the
mode instability and Pcr = 2616.6 N, Pmax = 100 N, and β0 = 0.0153, instability frequency intervals Df are about twice of the corresponding
0.0229, 0.0306 for the second mode instability) are shown in Fig. 8(a) natural frequency of the plate. Under the excitation having a frequency
against the dimensionless excitation frequency θ/2Ω11 for the first in the intervals Df, the plate will have the transverse parametric re-
mode instability, and in Fig. 8(b) against the dimensionless excitation sonance leading to its transverse dynamic instability. When the ex-
frequency θ/2Ω21 for the second mode instability, where the solid lines citation frequency is smaller than θ/2Ω11 = 0.961, 0.958, 0.956 for the
and dashed lines denote the analytical stable and unstable solutions first mode or θ/2Ω21 = 0.962, 0.960, 0.958 for the second mode, the
obtained from Eq. (54). It can be observed from Fig. 8(a) that the lowest plate under the excitation with the dimensionless parameter β = 0.4,
and highest excitation frequencies of the dynamic instability intervals 0.6, 0.8 will have steady vibration. When the excitation frequency is
Df for the first mode instability of the plate areDf L0.4 0.8 = θ/ initially in the interval Df and then increases in a stepwise fashion to a
2Ω11 = 0.961, 0.958, 0.956 andDf H 0.4 0.8 = θ/2Ω11 = 0.970, 0.973, value larger than θ/2Ω11 = 0.970, 0.973, 0.974 for the first mode or θ/
0.974. It can also be observed from Fig. 8(b) that lowest and highest 2Ω21 = 0.969, 0.971, 0.973 for the second mode, the “amplitudes” of
excitation frequencies of the dynamic instability intervals Df for the the plate vibration in the intervals Df exert perturbation to the plate,
second mode instability of the plate are Df L0.4 0.8 = θ/2Ω21 = 0.962, causing traction dynamic instability of the plate as shown in Fig. 8(a)
0.960, 0.958 and Df H0.4 0.8 = θ/2Ω21 = 0.969, 0.971, 0.973. It can be and (b).

9
Z. Zhong, et al. Engineering Structures 196 (2019) 109288

Fig. 6. Dynamic instability regions with the variation of aspect ratios of the plate: (a) mass at (a/2, b/2), first mode instability, (b) mass at (a/2, b/2), first mode
instability, (c) mass at (a/4, b/2), second mode instability, (d) mass at (a/4, b/2), second mode instability.

For comparisons, the linear solutions for the first and second mode to the nonlinear results obtained from Eq. (54). Hence, the linear so-
dynamic instability regions of the plate with a concentrated mass of lutions of Eq. (45) are sufficiently accurate for determining the dynamic
located at (a/3, b/3) were obtained from Eq. (45) and are shown in instability regions of the plate with arbitrary concentrated masses.
Fig. 8(c) and (d). It can be seen from Fig. 8(c) and (d) that under ex- However, to determine the traction dynamic instability of the plate, the
citation amplitude parameter β = 0.4, 0.6, 0.8, the lowest and highest geometric nonlinearity of a plate should be included and Eq. (54)
boundary frequency ratios θ/2Ω11 of the first mode dynamic instability should be used as shown in Fig. 8(a) and (b).
region are 0.962, 0.959, 0.958 and 0.969, 0.973, 0.975 respectively, To investigate the effects of the mass size on the nonlinear in-
while the lowest and highest boundary frequency ratios θ/2Ω21 of the stability of the plate, the curves of the vibration amplitudes of the plate
second mode dynamic instability region are 0.962, 0.959, 0.956 and having a mass at the position (a/3,b/3) of the non-zero modal dis-
0.968, 0.970, 0.972 respectively. The linear results for the lowest and placement obtained from Eq. (54) against the dimensionless excitation
highest boundary frequency ratios obtained from Eq. (45) are very close frequencies are shown in Fig. 9(a) and (b) for the first and second

Fig. 7. Dynamic instability regions with the location variation of concentrated mass 0.068 kg of the plate, a/b = 0.5, Pmax = 100 N: (a) first mode instability, (b)
second mode instability.

10
Z. Zhong, et al. Engineering Structures 196 (2019) 109288

Fig. 8. Dynamic instability of the plate with concentrated mass of 0.204 kg at (a/3,b/3): (a) nonlinear first mode instability, β = 0.4, 0.6, 0.8; (b) nonlinear second
mode instability, β = 0.4, 0.6, 0.8; (c)linear first mode instability region; and (d) the linear second instability region.

modes respectively, where the size of the mass changes from 0 kg to dynamic instability of the plate are shown Fig. 9(c) and (d). It can be
0.204 kg and the excitation amplitude parameter has a constant value seen that when the mass is located at (a/2, b/2) and (a/2, 5b/8), the
of β = 0.8 (Pcr = 2106 N, Pmax = 100 N, and β0 = 0.038 for the first widths of excitation frequency intervals of the dynamic instability of
mode instability and Pcr = 2616.6 N, Pmax = 100 N, and β0 = 0.0306 the plate are almost the same as those of the plate without a con-
for the second mode instability). centrated mass. Therefore, under the same excitation amplitude, both
It can be observed from Fig. 9(a) and (b) that as the size of the mass the excitation frequencies and their interval causing nonlinear dynamic
increases from 0 kg to 0.204 kg, the width of instability excitation fre- instability of the plate decrease with an increase of the size of the mass
quency intervals Df reduces and the positions of nonlinear response located at positions corresponding to non-zero modal displacements.
curves move gradually toward the lower value of θ/2Ω11 and θ/2Ω21 However, when the mass is located at a position corresponding to zero
(i.e. excitation frequencies decrease). For example, for the first-mode modal displacements, the size of the mass almost does not affect the
nonlinear response curves (Fig. 9(a)), the lowest and highest excitation critical excitation frequencies and their width for the nonlinear dy-
frequencies and the highest excitation frequencies of the dynamic in- namic instability of the plate.
stability intervals Df are respectivelyDf L0.8 = θ/2Ω11 = 0.986 The effects of the position of the concentrated mass on the excita-
andDf H 0.8 = θ/2Ω11 = 1.008 for no concentrated mass, Df 0.8 = 0.978
L
tion frequencies for the first and second mode instabilities of the plate
and Df H 0.8 = 0.997 and for a mass having a size of 0.068 kg, are shown in Fig. 10(a) and (b) respectively, where the amplitudes of
Df L0.8 = 0.9675 and Df H0.8 = 0.986 for a mass having a size of 0.136 kg, nonlinear response curves are obtained from Eq. (54). It can be seen
and Df L0.8 = 0.9562 and Df H 0.8 = 0.974 for a mass having a size of from Fig. 10(a) that when the mass is at the center point (a/2, b/2) of
0.204 kg. That is, the widths of the excitation frequency interval for the the plate (Fig. 2), the lowest and highest excitation frequencies θ/2Ω11
first-mode instability of the plate with the mass having sizes of 0, and their interval for the first mode instability of the plate are lower
0.068 kg, 0.136 kg and 0.204 kg are Df = 0.022, 0.019, 0.0185, 0.0178 than those when the mass is located at other positions. With the mass
respectively (Fig. 9(a)). Similarly, the widths of the excitation fre- moving away from the center point (a/2, b/2), the critical excitation
quency interval for the second-mode instability of the plate are frequencies θ/2Ω11 and their interval for instability of the plate in-
Df = 0.017, 0.0157, 0.0146, 0.014 respectively (Fig. 9(b)). The reduc- crease. For example, because the distance of the position (3a/4, 3b/4)
tion of the widths of excitation frequency intervals for dynamic in- from the center point (a/2, b/2) is larger than that of the position (a/2,
stability with an increase of the size of the mass is due to that fact that 5b/8) from the center point (Fig. 2), the width of the excitation fre-
an increase of the size of the mass leads to an increase of the damping quency interval for first mode instability of the plate with a mass at (3a/
ratio and a decrease of the natural frequency of the plate and subse- 4, 3b/4) is larger than that of the plate with a mass at (a/2, 5b/8) and so
quently to a decrease of width of the instability intervals. the plate with mass at (3a/4, 3b/4) is more prone to the first mode
When the mass is located at position (a/2, b/2) and (a/2, 5b/8) dynamic instability.
(Fig. 2) corresponding to zero modal displacements, the nonlinear re- For the second mode instability of the plate, the positions P2 (a/4, b/
sponse curves of the amplitude vs. θ/2Ω21 for the second-mode 2) and (3a/4, b/2) correspond to the maximum negative and positive

11
Z. Zhong, et al. Engineering Structures 196 (2019) 109288

Fig. 9. Nonlinear dynamic instability of a plate with concentrated mass, β = 0.8: (a) first mode: mass at (a/3,b/3); (b) second mode: mass at (a/3,b/3) (c) second
mode: mass at (a/2,b/2); (d) second mode: mass at (a/2,5b/8).

second mode displacements (Fig. 2). It can be observed from Fig. 10(b) for example at the point (a/2, 5b/8) or (a/2, b/2), the width of the
that as the distance of the mass position from the point P2 (a/4, b/2) or excitation frequency interval is almost the same for the second mode
(3a/4, b/2) increases within the corresponding half-wave region, the instability of the plate although the distance of the points (a/2, b/2) and
width of the excitation frequency interval for the second mode in- (a/2, 5b/8) from the points P2 (a/4, b/2) and (3a/4, b/2) has different
stability of the plate increases. For example, the distance of the mass at values 0.3 m and 0.325 m. The reason is that the mass located at the
the points (a/3, b/3) and (3a/4, 3b/4) corresponding to non-zero modal zero second natural modal displacements of the plate has no effects on
displacements from the points P2 (a/4, b/2) and (3a/4, b/2) are the second mode dynamic instability of the plate.
0.1947 m and 0.25 m respectively. Hence, when the mass is at the point The same three plates as those in Fig. 6 are herein used to in-
(3a/4, 3b/4), the width of the excitation frequency interval for second vestigate the nonlinear dynamic instability of plates based on the
mode instability of the plate is larger than that when the mass is at the analytical solutions. In the calculation, the maximum excitation load
point (a/3, b/3). However, when a mass is located at positions of zero applied to the plates is Pmax = 100 N. The variations of the critical ex-
modal displacements for the second natural vibration mode of the plate citation amplitude with the dimensionless excitation frequency θ/2Ω11

Fig. 10. Dynamic instability of a plate with a concentrated mass: (a) nonlinear response of first mode instability: mass of 0.136 kg, β = 0.8; (b) nonlinear response of
second mode instability: mass of 0.136 kg, β = 0.8.

12
Z. Zhong, et al. Engineering Structures 196 (2019) 109288

Fig. 11. Nonlinear dynamic instability of a plate with a concentrated mass of 0.068 kg, β = 0.8: (a) first mode instability, mass at (a/2, b/2); (b) second mode
instability, mass at (a/2, b/2); (c) first mode instability, mass at (a/4, b/2); (d) second mode instability, mass at (a/4, b/2).

or θ/2Ω21 for the three plates are plotted in Fig. 11. It can be seen that and second modes for plates having an aspect ratio a/b = 0.5 with a
as the aspect ratio decrease, the critical excitation frequency and the concentrated mass of 0.068 kg at different positions are shown in
curvature of first and second mode nonlinear instability curves in- Fig. 12.
creases, which indicates that the nonlinear traction vibration of a plate By comparing Fig. 12(a) with Fig. 10(a), it can be seen that the
with a smaller aspect ratio will be towards to the direction of larger bandwidth of excitation frequency intervals for the first mode nonlinear
excitation frequencies. Hence, larger excitation frequency of the peri- instability of the plate having an aspect ratio a/b = 0.5 are smaller than
odic load is needed to induce the nonlinear dynamic instability of plates that of the plate having a/b = 1.2. For example, for the plate having a/
having a smaller aspect ratio. b = 0.5, the first instability frequency intervals of the plate with a mass
It can also be seen from Fig. 11 that the influence of the mass po- at positions (a/2, b/2) and (3a/4, 3b/4) are 0.009 and 0.012
sition on the nonlinear dynamic instability of plates is also related to (Fig. 12(a)), while for the plate having a/b = 1.2, the corresponding
their aspect ratio. The dynamic nonlinear instability regions of the first instability frequency intervals are 0.021 and 0.026 (Fig. 10(a)). By

Fig. 12. Nonlinear dynamic instability with the location variation of concentrated mass 0.068 kg of the plate, β = 0.8, Pmax = 100 N, a/b = 0.5: (a) first mode
instability (b) second mode instability.

13
Z. Zhong, et al. Engineering Structures 196 (2019) 109288

comparing Fig. 12(b) with Fig. 10(b), it is found that the aspect ratio of both ends of the U-shape space of each of the blocks U1-U4 have the
a plate has a significant influence on its second nonlinear dynamic in- bilateral back-up blocks with bearing groove and a steel circular bar
stability. For the plate with a/b = 1.2, the excitation frequency ratio for with a “V”-shaped groove is mounted with two bearings at its ends and
the second mode nonlinear instability of the plate with a mass at (a/2, the bearings are embedded in the bearing grooves on the back-up block
b/2), (a/2, 5b/8), (a/3, b/3), (3a/4, 3b/4), (3a/4, b/2) and (a/4, b/2) at the ends of the U-shape space. The four edges of the test plate spe-
are about θ/2Ω21 = 1, 1, 0.981, 0.979, 0.962, and 0.962, while for the cimen are inserted to the V-shape grooves of steel circular bars. The
plate with a/b = 0.5, the corresponding excitation frequency ratio are simply supported boundary conditions of the plate are then established
about θ/2Ω21 = 1, 0.976, 0.973, 0.975, 1, and 1. Similar to the second as shown in Fig. 15.
mode linear instability region, the zero second modal displacements for The preload force applied to the edges of the test plate can be ad-
the plate for the aspect ratio a/b = 0.5 are located along the line y = b/ justed by tightening and relaxing the screw in the middle of the circular
2, while they are located along the line x = a/2 for the plate with a/ steel bar, so that the reliability of test results can be guaranteed. The
b = 1.2. As the mass moves to the position with a larger modal dis- blocks U1, U2 and U4 are fixed in the sliding device and the block U1 is
placement, the second mode nonlinear instability interval and the ex- fixed beside the distributive girder which has large rigidity sufficiently
citation frequency ratio decrease. Hence, for the plate having a/ to maintain its shape and evenly distributes excitation load to the plate
b = 0.5, the second nonlinear instability interval of the plate with a during the process of loading. Three bearings are fixed below the block
mass at (a/2, 5b/8) is smaller than that with a mass at (a/2, b/2) and U3 which is fastened to the right of the baffle bar. In order to transfer
the excitation frequency ratio decreases from 1 to 0.976 as shown in the vibrating forces from the vibrator uniformly to the block U1 and
Fig. 12(b). then to the plate, the distributive girder with very larger stiffness is
used. The vibrator exerts the excitation forces to the distributive girder,
4. Dynamic instability tests and analyses which transfers the excitation forces uniformly to the block U1. The
block U1 then transfer the uniform forces to the plate through steel
4.1. Test setup circular bar which is located at the middle of block U1. Since the block
U1 is fixed in the sliding device, when uniformly distributed load is
Vibration tests were performed in the laboratory to verify the ana- applied to the block U1, the sliding device moves along the negative
lytical solutions of the critical excitation frequencies of instability re- direction of the x-axis. Because the block U3 is blocked by the baffle
gions and the responses of the plate with arbitrary concentrated masses. bar, the sliding device is prevented moving along the negative direction
The experimental setup for the simply supported rectangular plate of the x-axis. Consequently, the steel circular bar connected with the
with concentrated masses is shown in Fig. 13, and a schematic diagram block U3 will be subjected to the corresponding uniformly distributed
for the vibration test is illustrated in Fig. 14. An aluminum plate spe- load and the other edge of the plate which is inserted to the V-shape
cimen was used in the test and the geometrical and mechanical prop- grooves of steel circular bars will be also subjected to the same uniform
erties of the plate are: the length, width and thickness are a = 1.2 m, excitation forces. In order to eliminate the initial imperfection of the
b = 1.0 m and h = 0.002 m, respectively, Young’s modulus aluminum plate under the influence of self-weight and concentrated
E = 6.9 × 1010 Pa, density ρ = 2700 kg/m3, and Poisson’s ratio masses, an aluminum plate with corresponding pre-camber was used in
μ = 0.33. this test. Brüel and Kjær 4381 sensor having a size of 0.034 kg is used to
The test setup (Fig. 13) is mainly composed of a plate specimen and collect the vibration signals of the plate and at the same time to act as
its fixture, a vibration test system of TIRA from the TIRA GmbH Com- concentrated masses as well. In view of the position of the coordinate
pany, and a PULSE vibration measuring system from the Brüel and Kjær system in the middle of the plate from Fig. 1, these sensors were at-
company, which are introduced as follows. tached to the plate on each sides of the plate, i.e. a concentrated mass is
composed of two sensors, as shown in Fig. 16.
(1) Plate specimen and its fixture
(2) Vibration test system
A foundation table is firmly bolted on the ground to prevent it from
any movement. The sliding device consisting of four rectangular beams The vibration test system is composed of a vibrator, power ampli-
fastened together by bolts with each rectangular beam mounted with fier, an exhaust fan and a vibration controller. The relevant parameters
three bearings is put on the foundation table to allow the movement of this experiment are set up to generate digital signals that are trans-
during the dynamic loading process. A frame consisting of four blocks mitted to a vibration controller whose control signals are displayed in
U1-U4 with each block having a “U”-shaped hollow space is used to the computer-control system. The amplifier receives the input signals
establish simply supported boundary conditions for the test plate. The from the regulating system, and then the excitation force is generated

Baffle Steel circular Aluminium Exhaust Front-end

bar bar plate fan module

U3 Sensor U4 U1
z
o x
y

U2 Vibrator
Shaped Foundatio Sliding Distributive Adjustable Charge Vibration Power
steel n table device girder support amplifier controller amplifier

Fig. 13. Experimental setup for the rectangular plate with concentrated masses.

14
Z. Zhong, et al. Engineering Structures 196 (2019) 109288

Fig. 14. Schematic diagram for lab vibration tests of the plate.

Screw

Circular
Block-up Steel
block bar

Sensor
Pulley (concentrated
masses)

Fig. 15. Simply-supported boundaries. Fig. 16. The setup of concentrated masses.

and the fixed frequency and sweep sine tests are simulated by the vi- processing function of the front-end module, the signals are changed to
brating coil supply and field power supply. The harmonic load gener- digital signals that can be identified by the vibration signal analysis
ated by the vibrator was evenly distributed to the edge block U1 software, and thus data collection, data management and analysis can
through the distributive girder and the sliding device can also slide back be conducted.
and forth so that the edge block U3 was subjected to the same reactive
force as the load. During the vibration test, the “U”-type edge blocks
4.2. Modal testing
and circular steel bar were sufficiently rigid to maintain their shapes
without deformation.
To perform the test for transverse dynamic instability, the modal
shape, natural frequencies and damping ratio of transverse free vibra-
(3) Vibration measuring system
tion of the plate with the different sizes and distributions of con-
centrated masses need to be known first. The transverse free vibration
The PULSE vibration measuring system includes sensors, a front-end
of the plate can be induced by using Brüel and Kjær hammer with a
module, a charge amplifier and a vibration signal analysis software-
rubber head to impact the position (a/4, b/2) or (3a/4, b/2) of the plate
Labshop (Fig. 14). The vibration signals of the plate are collected by the
in the z-direction, and then the free decay responses were collected by
sensors and are transmitted to the charge amplifier which converts vi-
sensors. For the measurement of shapes of the first and second modes,
bration signals to voltage signals. Through the storage capabilities and
sensors were placed at a/4, a/2 and 3a/4 (a is the length of plate) in the

15
Z. Zhong, et al. Engineering Structures 196 (2019) 109288

Fig. 17. Modal shapes of the plate with mass of 0.068 kg at (a/3, b/3): (a) first mode; (b) second mode.

Fig. 18. The frequency response function curve of the plate with concentrated masses of 0.068 kg: (a) Located at (a/3, b/3); (b) Located at (a/2, b/2).

x direction as well as at b/4 and 3b/4 (b is the width of plate) in the y increase when the concentrated mass is located at the point (a/3, b/3)
direction, as shown in Fig. 13. It should be pointed out that in order to as shown in Fig. 19(a). The maximum displacements of the first and
obtain the measured natural frequencies and damping ratio of the plate second instability modes of the plate with the mass at the positions
without concentrated mass, the sensors can be placed beside the P1(a/2, b/2) and P2(a/4, b/2) or P1(a/2, b/2) and P2(3a/4, b/2) (Fig. 2)
boundary condition where the natural frequencies and damping ratio of are 0.6 m and 0.5 m, respectively. Fig. 19(b) shows that the frequencies
the plate are not affected by the concentrated mass according to the increase with an increase of the distance of the mass from the points P1
theoretical result. The impact signal of the hammer was picked up by and P2 of the corresponding modes. However, the plate with a con-
the force transducer and the response signals of transverse free vibra- centrated mass of 0.068 kg at the positions (a/2, b/2) and (a/2, 5b/8)
tion of the plate with concentrated mass were obtained by sensors, along the axis a/2 has the same second modal frequencies although
which were converted to voltage signals and entered to the PULSE they have different distances from the point P2. The results indicate that
system through the amplifier. According to the response signals for each the second modal frequency of the plate is not affected by the con-
channel, the modal shape of transverse free vibration of the plate can be centrated mass which is located along the axis a/2.
determined by the operational modal analysis program in the PULSE Although the mode shapes and the damping ratio can be obtained
system of Brüel and Kjæ [35]. The fundamental frequency of the plate by the operational modal analysis, the value of damping ratio is influ-
with arbitrary concentrated masses can be determined by the frequency enced by the weights of sensors distributed on the plate. Therefore, the
response function of PULSE system. For example, the modal shapes and logarithmic decrement method is used to calculate the damping ratio in
natural frequency of the plate having a concentrated mass of 0.068 kg this test, which has been described and confirmed in [34,35], and so the
at (a/3, b/3) and (a/2, b/2) for the first and second modes can be ob- damping ratio can be calculated as
tained from the operational modal analysis and frequency response
function curve as shown in Figs. 17 and 18. 1 Ati 1
= ln
It can be seen that there is only one half-wave in the x and y di- j At (i + j ) 2 (55)
rections for the first mode (which is defined as W11 with k = 1 and
l = 1)), while there are two half-waves in the x direction and one half- where Ati and At(i+j) are the amplitudes of free vibration of the plate
wave in the y direction for the second mode (which is defined as W21 with different concentrated masses at times ti and t(i + j). The results of
with k = 2 and l = 1). Ati and At(i+j) can be obtained by the decay records as shown in Fig. 20.
The analytical frequencies of transverse free vibration of the plate The calculated damping ratios for the plate with concentrated
for the first and second modes are compared with the test results in masses are listed in Table 1, which shows that the sizes and distribu-
Fig. 19(a) and (b), where good agreements between the test and theo- tions of the masses have a great influence on the damping ratio. The
retical results can be found. The frequency decreases with a size damping ratio of the second mode is smaller than that of the first mode.

16
Z. Zhong, et al. Engineering Structures 196 (2019) 109288

Fig. 19. The natural frequencies of the plate for first and second mode: (a) frequency versus mass at (a/3,b/3); (b) frequency versus distance of the concentrated mass
0.068 kg from the positions P1 and P2 of the instability modes.

The damping ratio increases with the increase of the size and distance To collect the first mode instability responses of the plate, a sensor
of the mass from the positions P1 and P2 for the first and second in- of 0.204 kg was used as the concentrated mass and was placed at the
stability modes. However, the damping ratios of the second mode of the position P1(a/2, b/2) and at the same time play the role of collecting the
plate are almost identical when the concentrated mass was located instability responses of the plate, another sensor with a different size of
along the axis a/2 such as at the positions (a/2, 5b/8) and (a/2, b/2). 0.017 kg (Brüel and Kjær 4382) was placed at the position P2(a/4, b/2)
or (3a/4, b/2) to collect the second mode instability responses of the
4.3. Instability regions of excitation frequencies and nonlinear responses plate because the second mode instability responses at the position
P1(a/2, b/2) (corresponding to zero modal displacement) are almost
In order to verify the analytical solutions of critical excitation fre- equal to zero. Because the mass of the sensor (Brüel and Kjær 4382) is
quencies and responses of the plate, the upward and downward swept very small, it do not affect the natural frequency and damping ratio of
frequency vibration tests were conducted. In these tests, the static the plate. Similar method of measurement was adopted for other cases
component of the periodic load was assumed to be equal to zero, i.e. of sweep tests.
α0 = 0 (Eq. (31d)) and the in-plane uniformly-distributed periodical The swept test results for the first and second instability modes are
load was applied to the plate. The frequencies of excitation in the swept shown in Figs. 21(a)–(d) and 22(a)–(d) for different excitation ampli-
frequency vibration test were taken about twice of the natural fre- tude parameters β respectively, where the excitation amplitude para-
quencies of the plate. The amplitudes of the excitation were increased meter β is defined as β = β0Pcr/Pmax and Pmax = 100 N is the maximum
in a stepwise fashion with each step of 10 N, and the vibration re- load in the test. A time domain analysis method is presented to de-
sponses of the plate were recorded and analyzed by the system of termine the critical excitation frequencies of dynamic instability region
PLUSE measuring system of Brüel and Kjær [36]. of the plate with concentrated mass.
For example, the natural frequencies of the first and second modes It can be seen that when the excitation amplitudes is very small
of the plate with a concentrated mass of 0.204 kg at the point (a/2, b/2) (β = 0.2) (Figs. 21(a) and 22(a)), the displacement time histories of
can be obtained by hammer tests as 7.875 Hz and 18.225 Hz, respec- upward and downward swept frequency vibration tests indicate that
tively. The plate will lose its stability owing to parametric resonance there is no sudden divergence in vibration of the plate and the plate
when the excitation frequencies are about twice of the natural fre- vibrates within bounded amplitudes. However, when the excitation
quencies, thus, the swept frequency tests of the dynamic stability can be amplitude increases to β = 0.4, 0.6 and 0.8, the displacement time
set in the range of 14.9–16.9 Hz for the first instability mode and histories may suddenly become quite large. For the upward swept fre-
35.1–37.1 Hz for the second instability mode with a swept rate 1 Hz per quency vibration test, when the excitation frequency of the dynamic
minute. load reaches around the critical excitation frequency, i.e. 15.52 Hz for

Fig. 20. The damping decay curves of free vibration of the plate with concentrated mass 0.068 kg at (a/3,b/3): (a) the first mode ; (b) the second mode.

17
Z. Zhong, et al. Engineering Structures 196 (2019) 109288

Table 1 22(b)–(d)), the vibration amplitude of the plate suddenly increases

Damping ratios (%) calculated based on damping decay curves. significantly when the excitation frequency is reaches the highest cri-
Size of 0.068 kg 0.136 kg 0.204 kg tical frequency and then gradually decreases until the excitation fre-
concentrated quency is lower than the lowest critical frequency and the vibration of
mass the plate returns to the steady state. Therefore, the lowest and highest
Position of First Second First Second First Second critical frequencies of the dynamic instability region of the plate under
concentrated mode mode mode mode mode mode
mass
different excitation amplitudes can be obtained.
It can also be observed from test results shown in Figs. 21(a)– (d)
(a/3,b/3) 0.278 0.231 0.321 0.257 0.358 0.293 and 22(a)–(d) that there are three cases for the instability state of the
(a/2,b/2) 0.302 0.217 0.348 0.223 0.396 0.228 simply supported rectangular plates with concentrated masses under
(3a/4,3b/4) 0.262 0.221 0.292 0.246 0.335 0.275
the in-plane uniformly-distributed periodic load:
(a/2,5b/8) 0.289 0.213 0.332 0.223 0.374 0.226
(a/4,b/2) 0.265 0.253 0.306 0.281 0.345 0.322
(3a/4,b/2) 0.268 0.248 0.310 0.289 0.339 0.319 (1) When the excitation frequencies are lower than the critical fre-
quencies, the vibration frequencies of the plate are the same as the
excitation frequencies of the dynamic load, and the vibration of the
plate is in a stable state with bounded amplitudes.
the first instability mode (Fig. 21(b)) and 36.23 Hz for the second in- (2) When the excitation frequencies reach the critical frequencies, the
stability mode (Fig. 22(b)) when β = 0.4; 15.50 Hz for the first in- rate of work done by the vibration load is greater than the rate of
stability mode (Fig. 21(c)) and 36.13 Hz for the second instability mode energy dissipation of the plate. When the excitation frequency is
(Fig. 22(c)) when β = 0.6; and 15.43 Hz for the first instability mode about twice of the natural frequency of the plate, the plate will lose
(Fig. 21(d)) and 35.98 Hz for the second instability mode (Fig. 22(d)) its stability owing to parametric resonance and the vibration am-
when β = 0.8, the vibration amplitude of the plate increases con- plitude will increase significantly. However, the vibration ampli-
siderably until the highest critical frequencies are reached (15.82, tude of the plate is limited to a certain value due to the hardening
15.87 and 15.95 Hz for the first instability mode when β = 0.4, 0.6, 0.8, effect of the plate which is equivalent to the increase in the rigidity
and 36.46, 36.55 and 36.63 Hz for the second instability mode when of the plate. The rate of energy dissipation will also increase with
β = 0.4, 0.6, 0.8). When the excitation frequencies exceed the corre- the increase of the vibration amplitude of the plate.
sponding highest critical frequencies, the vibration amplitude of the (3) When the frequencies of the dynamic excitation are higher than the
plate decreases and the vibration of the plate returns to a steady state. highest critical frequencies of the plate, the rate of work done by
In the downward swept frequency vibration test (Figs. 21(b)–(d) and the vibration load is less than the rate of energy dissipation of the

Fig. 21. Swept frequency tests of first instability mode of the plate with a concentrated mass of 0.204 kg at point (a/2,b/2): (a) β = 0.2, (b) β = 0.4, (c) β = 0.6, (d)
β = 0.8.

18
Z. Zhong, et al. Engineering Structures 196 (2019) 109288

Fig. 22. Swept frequency tests of the second instability mode of the plate with concentrated mass of 0.204 kg at point (a/2,b/2): (a) β = 0.2, (b) β = 0.4, (c) β = 0.6,
(d) β = 0.8.

plate. The plate vibrates with the excitation frequencies, and the areas around the theoretical point. Hence, the experimental results in-
vibration amplitude of the plate attenuates until the plate returns to clude the influences of the modal displacements of other points around
a stable state. the theoretical point. For example, the analytical second modal dis-
placement of the plate vanishes at the point (a/2, b/2) and so theore-
The experimental results of critical excitation frequencies are com- tically, the mass located at the point should have little influence on the
pared with analytical results in Fig. 23(a) and (c) for the first and second mode instability of the plate. In the experiment, however, the
second instability modes, respectively. It can be seen that they agree sensor is fixed to the area around (a/2, b/2) where the modal dis-
with each other quite well. The greatest difference between analytical placements are not equal to zero except at the point (a/2, b/2). Hence,
and experimental results occurs when the concentrated mass is placed the mass fixed to a small area around (a/2, b/2) has some influence on
at the point (a/2, b/2). the second nonlinear instability of the plate leading the natural fre-
The analytical and test results for the nonlinear response curves quency of the plate smaller than the analytical one. Consequently, the
under the excitation amplitude parameter β = 0.8 are obtained shown measured nonlinear instability curves move to the lower excitation
in Fig. 23(b) and (d) for the first and second instability modes, re- frequency ratio and experimental results for the bandwidth of fre-
spectively, which also exhibit good agreements between the analytical quency interval and the amplitude of the nonlinear dynamic instability
and test results. The vibration amplitude of the plate with the same of the plate would have discrepancies with the corresponding analytical
mass increases with a decrease of the distance of the mass from the ones.
positions P1 and P2 (Fig. 2) for the first and second instability modes. Table 2 lists the analytical and experimental results for the lowest
The width of instability intervals Df of nonlinear response increases and highest critical excitation frequencies, which shows that the max-
when masses move away from the position of the largest modal dis- imum difference is 2.05% for the first instability mode and 1.21% for
placement of the plate. The experimental results show that the geo- the second instability mode.
metric nonlinearity of a plate can induce traction vibration toward to
the direction of higher frequencies as shown in Fig. 23(b) and (d). 5. Conclusions
It can be found from Fig. 23(b) and (d) that when the concentrated
mass is located at the position (3a/4, 3b/4), the agreement between the The transverse dynamic instability of a simply supported rectan-
analytical and the experimental results of the nonlinear dynamic in- gular plate with arbitrary concentrated masses under an in-plane uni-
stability responses are quite good, while their agreements for the other formly-distributed periodic load along two opposite edges was studied
two positions are not so good, particularly, for the second mode in- analytically and experimentally in this paper. Based on the von-Kármán
stability when the concentrated mass is at (a/2, b/2) as shown in large deflection theory of thin plate, differential equations governing
Fig. 23(d). The main reason for this is that the masses are considered to the transverse motion of simply supported rectangular plates with ar-
be located at points in the analytical solutions, while in the experi- bitrary concentrated masses were derived and the equations of in-
mental investigation, the masses (i.e. sensor) were fixed over small stability regions and nonlinear response associated with various

19
Z. Zhong, et al. Engineering Structures 196 (2019) 109288

Fig. 23. (a) Dynamic instability regions for first mode, (b) Nonlinear responses under β = 0.8 for first mode, (c) Dynamic instability regions for second mode, (d)
Nonlinear responses under β = 0.8 for second mode.

Table 2
Analytical and experimental critical frequencies of the plate with mass of 0.204 kg at the point (a/2, b/2).
Dimensionless excitation amplitude β The first mode instability in Fig. 17(a)

Lowest critical frequencies (Hz) Highest critical frequencies (Hz)

Analytical Experimental Error (%) Analytical Experimental Error (%)

0.9 15.33 15.38 0.33 15.65 15.97 2.04

0.8 15.36 15.45 0.59 15.63 15.95 2.05
0.7 15.38 15.47 0.59 15.61 15.86 1.60
0.6 15.41 15.49 0.52 15.59 15.87 1.79
0.5 15.43 15.53 0.65 15.56 15.81 1.61
0.4 15.45 15.59 0.91 15.53 15.82 1.87

Dimensionless excitation amplitude β The second mode instability in Fig. 17(c)

0.9 36.37 35.93 1.21 36.98 36.78 0.54
0.8 36.43 36.05 1.04 36.95 36.63 0.87
0.7 36.45 36.03 1.15 36.91 36.65 0.70
0.6 36.49 36.13 0.99 36.87 36.55 0.87
0.5 36.53 36.22 0.85 36.83 36.58 0.68
0.4 36.57 36.25 0.88 36.79 36.46 0.90

instability modes were obtained. The effects of the sizes and positions of the plate with a concentrated mass needs larger excitation amplitude to
concentrated masses on the excitation frequency regions for linear dy- develop dynamic instability than that without a concentrated mass.
namic instability of the plate and on the excitation frequency intervals However, when a mass is located at a location corresponding to the zero
for nonlinear linear dynamic instability of the plate were investigated. second modal displacement, the mass does not influence the second
The variations of nonlinear responses with the increase of excitation mode dynamic instability of the plate.
amplitude for different excitation frequency ratios were also explored. The effects of concentrated masses at different positions on the ex-
It has been found that concentrated masses increase the required citation frequency regions for the dynamic instability of the plate were
amplitude of excitation for the dynamic instability of the plate. Hence, explored. A plate with a concentrated mass at the position

20
Z. Zhong, et al. Engineering Structures 196 (2019) 109288

corresponding to the largest modal displacement has the smallest width the direction of higher frequencies caused by the geometric non-line-
of the dynamic instability region. The width of the instability regions of arity of a plate was found.
the plate with the same mass were found to increase with an increase of The characteristics of these three cases were identified: (1) when the
the distance of the mass position to the largest modal displacement. excitation frequencies are lower than the critical frequencies, the vi-
However, if the mass is located at the position corresponding to the zero bration frequencies of the plate are the same as the excitation fre-
modal displacement, the width of the dynamic instability regions of the quencies of the dynamic load, and the vibration of the plate is in a
plate does not increase regardless the distance to the largest modal stable state with bounded amplitudes; (2) when the excitation fre-
displacement. Effects of the sizes and positions of concentrated masses quencies reach the critical frequencies (about twice of the corre-
on the width of excitation frequency intervals for nonlinear dynamic sponding natural frequencies of the plate), the plate will lose its sta-
instability of the plate are found to be the same as those on the width of bility in the transverse direction owing to parametric resonance and the
excitation frequency regions for linear dynamic instability of the plate. transverse vibration amplitude will increase significantly if the excita-
The effects of aspect ratio on the dynamic instability region and tion amplitude is sufficient large; (3) when the frequencies of the dy-
nonlinear instability response of the plate with concentrated masses namic excitation are higher than the highest critical frequencies of the
were also analyzed. The critical excitation coefficient β of instability plate, the plate vibrates with the excitation frequencies, and the vi-
region increases with a decrease of aspect ratio, however, the band- bration amplitude of the plate attenuates until the plate returns to a
width of instability region decreases with the decrease of aspect ratio. stable state.
As the decrease of aspect ratio, the excitation amplitude for the non- In addition, it was found that the rate of energy dissipation of the
linear instability of the plate increases and the traction vibration moves plate and the critical values of excitation amplitude for the dynamic
towards to the larger excitation frequency. The position of the con- instability region of the plate increase when the concentrated mass is
centrated mass has a significant influence on the dynamic instability of located in the region corresponding to the non-zero modal displace-
the plate owing to the decrease of aspect ratio for the second mode ments. However, at the same time, the natural frequencies of the plate
instability. and the critical values of excitation frequencies of the dynamic in-
The swept frequency tests were carried out to verify the theoretical stability region decrease. It worth pointing out that the proposed
results of the transverse dynamic instability region and responses of the method can be extended to dynamic instability of plates having con-
plate and the test results have shown good agreements with the ana- centrated masses with other boundary conditions.
lytical counterparts. There are three cases for the stability states of the The results obtained in this paper are useful for a better under-
simply supported rectangular plates with concentrated masses under an standing of the dynamic stability mechanism of rectangular plates with
in-plane uniformly-distributed periodic load. Compared with those re- concentrated masses and have potential applications in the vibration
ported in literature, the experimental work mainly has the following control of these plates, such as bridge decks in turbulent winds.
contributions: (1) a time domain analysis method was presented to
determine the critical excitation frequency of dynamic instability re-
gion of the plate with concentrated masses; (2) the out-of-plate dynamic Acknowledgement
instability of the plate with concentrated masses under an in-plane
periodic load was verified by several groups of experiments, and the The research of this paper has been financially supported by the
influence mechanism of concentrated mass on dynamic instability of National Natural Science Foundation of China (No. 51578188),
the plate was revealed; (3) the traction vibration instability toward to Technology Planning Project of Guangzhou City (No. 201807010021).

Appendix 1. Coefficients Gn and Hn (n = 1–7)

Coefficients Gn and Hn (n = 1–7) are given by

a b
0 i k dx 0 j l dy
G1ik = , H1jl =
k2 2 l2 2
2a 2b (56)
a b
i Xp k dx 0 j Yq l dy
G2ipk = 0
, H2jql =
k2 2 l2 2
2a 2b (57)
a b
i X p k dx 0 j Yq l dy
G3ipk = 0
, H3jql =
a
2 ( )
a
p 2 b
2 ( ) q 2
a (58)
a b
i X p k dx 0 j Yq l dy
G4ipk = 0
pk
, H4jql = ql
2 a 2 b (59)
a b
Xm X p dx Xn Yq dy
G5mp = 0
, H5nq = 0

a ( )
p 2
a
b ( ) q 2
b (60)
a b
0
Xm k k dx 0
Yn l l dy
G6mkk = , H6nll =
k2 2 l2 2
a b (61)
a b
0
Xm k k dx 0
Yn l l dy
G7mkk = , H7nll =
k2 2 l2 2
a b (62)

21
Z. Zhong, et al. Engineering Structures 196 (2019) 109288

Appendix 2. Matrices Bkl,11, Bkl,12, Bkl,21 and Bkl,22

Matrices Bkl,11, Bkl,12, Bkl,21 and Bkl,22 in Eq. (42) are given by
... ... ... ... ...
. . . Ikl Ei 2 . . . 0 0
... ... ... ... ...
Bkl,11 =
... 0 . . . Ikl E 2 2 kl
... 0 ... kl Ikl + kl E1 2 (63)
... ... .. . ... .. .
0 0 .. . Fi .. .
... ... .. . ... .. .
Bkl,12 =
0 F2 . . . 0 .. .
F1 0 .. . 0 .. . (64)

.. . 0 .. . 0 F1
.. . 0 .. . F2 0
Bkl,21 = . . . ... .. . ... ...
.. . Fi .. . 0 0
.. . ... .. . ... ... (65)

Ikl E1 2 .. . 0 .. .
kl kl
Ikl E2 2 .. . 0 .. .
kl
Bkl,22 = ... ... .. . ... .. .
0 0 .. . Ikl Ei 2 .. .
... ... .. . ... .. . (66)
where Ei = Ikl (2i 1) 2 /4 2
kl , Fi = kl (2i 1)/2 kl , 0 is the null matrix and Ikl is the identity matrix of kl mode.

Appendix 3. Matrices B1, B2, B3

Matrices B1, B2, B3 in Eq. (43) are given by

... ... ... ... ... ... ... .. . ... .. .
. . . Ikl . . . 0 0 0 0 .. . 0 .. .
... ... ... ... ... ... ... .. . ... .. .
... 0 ... Ikl kl 0 0 .. . 0 .. .
... 0 ... kl Ikl + kl 0 0 .. . 0 .. .
B1 =
... 0 ... 0 0 Ikl kl kl .. . 0 .. .
... 0 ... 0 0 kl Ikl .. . 0 .. .
... ... ... ... ... ... ... .. . ... .. .
... 0 ... 0 0 0 0 .. . Ikl .. .
... ... ... ... ... ... ... .. . ... .. . (67)
.. . ... .. . ... ... ... ... .. . ... .. .
.. . 0 .. . 0 0 0 0 .. . Hi .. .
.. . ... .. . ... ... ... ... .. . ... .. .
.. . 0 .. . 0 0 0 H2 .. . 0 .. .
.. . 0 .. . 0 0 H1 0 .. . 0 .. .
B2 =
.. . 0 .. . 0 H1 0 0 .. . 0 .. .
.. . 0 .. . H2 0 0 0 .. . 0 .. .
.. . ... .. . ... ... ... ... .. . ... .. .
.. . Hi .. . 0 0 0 0 .. . 0 .. .
.. . ... .. . ... ... ... ... .. . ... .. . (68)
.. . ... .. . ... ... ... ... .. . ... .. .
.. . Gi .. . 0 0 0 0 .. . 0 .. .
.. . ... .. . ... ... ... ... .. . ... .. .
.. . 0 .. . G2 0 0 0 .. . 0 .. .
.. . 0 .. . 0 G1 0 0 .. . 0 .. .
B3 =
.. . 0 .. . 0 0 G1 0 .. . 0 .. .
.. . 0 .. . 0 0 0 G2 .. . 0 .. .
.. . ... .. . ... ... ... ... .. . ... .. .
.. . 0 .. . 0 0 0 0 .. . Gi .. .
.. . ... .. . ... ... ... ... .. . ... .. . (69)
where Gi = (2i 1) 2Ikl , Hi = (2i 1) kl / . 0 is the null matrix and Ikl is the identity matrix of kl mode.

References [5] Cha PD. free vibration of a rectangular plate with a concentrated mass. J Sound Vib
1997;207(4):593–6.
[6] Li QS. An exact approach for free vibration analysis of rectangular plates with line-
[1] Chakraverty S. Vibration of plates, 1st ed. New York: USA; 2009. concentrated mass and elastic line-support. Int J Mech Sci 2003;45(4):669–85.
[2] Ostiguy GL, Evan-Iwanowski RM. Influence of the aspect ratio on the dynamic [7] Chai GB. Free vibration of rectangular isotropic plates with and without a con-
stability and non-linear response of rectangular plates. J Mech Des centrated mass. Comput Struct 1993;48(3):529–33.
1982;104(2):417–25. [8] Gutiérrez RH, Laura PAA. Effect of a concentrated mass on large amplitude, free
[3] Deolasi PJ, Datta PK. Experiments on the parametric vibration response of plates flexural vibrations of elastic plates and beams. Appl Acoust 1984;17(2):135–51.
under tensile loading. Exp Mech 1997;37(1):56–61. [9] Cheng G, Cheng Q, Wang WD. An integral equation method for free vibration of
[4] Yeh MK, Liu CS, Chen CC. Dynamic instability of rectangular composite plates under circular plate with concentrated mass at arbitrary positions. Adv Mater Res
parametric excitation. Cmc-Comput Mater Con 2014;39(1):3–20. 2011;255(5):1830–5.

22
Z. Zhong, et al. Engineering Structures 196 (2019) 109288

[10] Ciancio PM, Rossit CA, Laura PAA. Approximate study of the free vibrations of a material plates on elastic foundation. Defence Sci J 2015;65(3):245–51.
cantilever anisotropic plate with a concentrated mass. J Sound Vib [26] Bondarev PA. Oscillations of a plate with concentrated masses lying on a nonlinear
2007;302(3):621–8. elastic base. Ukr Math J 1974;26(1):48–52.
[11] Wu J, Luo S. Use of the analytical-and-numerical-combined method in the free vi- [27] Eshmatov BK, Khodzhaev DA. Dynamic stability of a viscoelastic plate with con-
bration analysis of a rectangular plate with any number of point masses and centrated masses. Int Appl Mech 2008;44(2):208–16.
translational springs. J Sound Vib 1997;200(2):179–94. [28] Eshmatov BK, Eshmatov K, Khodzhaev DA. Nonlinear flutter of viscoelastic rec-
[12] Wu J, Luo S. Free vibration analysis of a rectangular plate carrying any number of tangular plates and cylindrical panels of a composite with a concentrated masses. J
point masses and translational springs by using the modified and quasi-analytical Appl Mech Tech Ph 2013;54(4):578–87.
and numerical combined methods. Int J Numer Meth Eng 2015;40(12):2171–93. [29] Azarboni HR, Darvizeh M, Darvizeh A, Ansari R. Nonlinear dynamic buckling of
[13] Sadiku PDS. On the flexural vibration of an elastic plate with a concentrated mass. imperfect rectangular plates with different boundary conditions subjected to var-
Arch Appl Mech 1989;60(2):92–8. ious pulse functions using the Galerkin method. Thin-Wall Struct 2015;94:577–84.
[14] Doherty SM, Dowell EH. Experimental study of asymptotic modal analysis applied [30] Gao K, Gao W, Wu D, Song C. Nonlinear dynamic characteristics and stability of
to a rectangular plate with concentrated masses. J Sound Vib 1994;170(5):671–81. composite orthotropic plate on elastic foundation under thermal environment.
[15] Lin RM, Lim MK. Natural frequencies of plates with arbitrary concentrated mass and Compos Struct 2017;168:619–32.
stiffness modifications. Comput Struct 1995;57(4):721–9. [31] Levy S. Large deflection theory for rectangular plates. Proc Symposia Appl Math
[16] Ostachowicz W, Krawczuk M, Cartmell M. The location of a concentrated mass on 1949;1(1):197–210.
rectangular plates from measurements of natural vibrations. Comput Struct [32] Nguyen H, Ostiguy GL. Effect of boundary conditions on the dynamic instability and
2002;80(16):1419–28. non-linear response of rectangular plates, Part I: Theory. J Sound Vib
[17] Maiz S, Rossit CA, Bambill DV, Susca A. Transverse vibrations of a clamped ellip- 1989;133(3):381–400.
tical plate with a concentrated mass at an arbitrary position. J Sound Vib [33] Nguyen H, Ostiguy GL, Samson LP. Effect of boundary conditions on the dynamic
2009;320(4):1146–63. instability and non-linear response of rectangular plates, Part II: Experiment. J
[18] Bolotin VV. The dynamic stability of elastic systems. In: Weingarten VI, Greszcuzuk Sound Vib 1989;133(3):401–22.
LB, Trirogoff KN, Gallegos KD, editors. Determination of regions of dynamic in- [34] Liu AR, Yang ZZ, Bradford MA, Pi YL. Nonlinear dynamic buckling of fixed shallow
stability. London: Holden-Day Inc; 1964. p. 18–29. arches under an arbitrary step radial point load. J Eng Mech
[19] Badalov FB, Akbarov UI, Éshmatov K. Stability of viscoelastic plate under dynamic 2018;144(4):04018012.
loading. Soviet Appl Mech 1991; 27(9): 892–9. V.I. Weingarten, L.B. Greszcuzuk, K. [35] Yao MH, Zhang W. Shilnikov-type multipulse orbits and chaotic dynamics of a
N. Trirogoff, K.D. Gallegos. parametrically and externally excited rectangular thin plate. Int J Bifurcat Chaos
[20] Pierre C, Dowell EH. A study of dynamic instability of plates by an extended in- 2007;17(03):851–75.
cremental harmonic balance method. J Appl Mech 1985;52(3):693–8. [36] Liu AR, Lu HW, Fu JY, Pi YL, Huang YQ, Li J, et al. Analytical and experimental
[21] Wu HL, Yang J, Kitipornchai S. Parametric instability of thermo-mechanically studies on out-of-plane dynamic instability of shallow circular arch based on
loaded functionally graded graphene reinforced nanocomposite plates. Int J Mech parametric resonance. Nonlinear Dyn 2017;87(1):677–94.
Sci 2017;135:431–40. [37] Liu AR, Yang ZC, Lu HW, Pi YL. Experimental and analytical investigation on the in-
[22] Yang J, Shen HS. Free vibration and parametric resonance of shear deformable plane dynamic instability of arches owing to parametric resonance. J Vib Control
functionally graded cylindrical panels. J Sound Vib 2003;261(5):871–93. 2017;24(19):4419–32.
[23] Yang J, Liew KM, Kitipornchai S. Dynamic stability of laminated FGM plates based [38] Brüel, Kjær. PULSE-getting started: an introduction to PULSE. Brüel & Kjær Sound
on higher-order shear deformation theory. Comput Mech 2004;33(4):305–15. and Vibration Measurement A/S, Nærum; 2003.
[24] Darabi M, Ganesan R. Nonlinear dynamic instability analysis of laminated com- [39] Huang YQ, Liu AR, Pi YL, Lu HW, Gao W. Assessment of lateral dynamic instability
posite thin plates subjected to periodic in-plane loads. Nonlinear Dyn of columns under an arbitrary periodic axial load owing to parametric resonance. J
2018;91(1):187–215. Sound Vib 2017;395:272–93.
[25] Ramu I, Mohanty SC. Free vibration and dynamic stability of functionally graded

23