You are on page 1of 21

Available online at www.sciencedirect.

com

Finite Elements in Analysis and Design 40 (2004) 463 – 483


www.elsevier.com/locate/ nel

Transient analysis of laminated composite plates with


embedded smart-material layers
S.J. Leea , J.N. Reddyb;∗ , F. Rostam-Abadic
a
Department of Civil Engineering, Texas A&M University, College Station, TX 77843-3136, USA
b
Department of Mechanical Engineering, Texas A&M University, ENPH Building, Room 210, College Station,
TX 77843-3123, USA
c
U.S. Army Tank-Automotive Tank and Armaments Command, AMSTA-TR-R, Warren, MI 48397, USA

Received 1 September 2002; accepted 23 December 2002

Abstract

The transient response of laminated composite plates with embedded smart material layers is studied using
a uni ed plate theory that includes the classical, rst-order, and third-order plate theories. The simple velocity
feedback control is used. The nite element method is used for numerical simulations, and numerical results
are presented to study the e1ects of the lamination scheme, boundary conditions, and loading. As a speci c
example, Terfenol D magnetostrictive material layers are used to control the vibration suppression. The e1ect
of material properties, smart layer position, and smart layer thickness on the vibration suppression is also
investigated.
? 2003 Elsevier B.V. All rights reserved.

Keywords: Composite plates; Vibration control; Finite element model; Shear deformation theory; Transient analysis

1. Introduction

The study of smart materials and structures has received considerable attentions in recent years.
The advantage of incorporating these special types of materials into the structure is that the sens-
ing and actuating mechanism becomes part of the structure so that one can monitor the structural
integrity/health of the structure. There are a number of materials that have the capability to be used
as a sensor or an actuator or both. Piezoelectric materials, magnetostrictive materials, electrostrictive
materials, shape memory alloys, and electrorheological 8uids provide examples of such materials.


Corresponding author. Tel.: +1-979-862-2417; fax: +1-979-862-3989.
E-mail address: jnreddy@shakti.tamu.edu (J.N. Reddy).

0168-874X/$ - see front matter ? 2003 Elsevier B.V. All rights reserved.
doi:10.1016/S0168-874X(03)00073-8
464 S.J. Lee et al. / Finite Elements in Analysis and Design 40 (2004) 463 – 483

Among these, piezoelectric and magnetostrictive materials have the capability to serve as both sen-
sors and actuators. Piezoelectricity [1] is a phenomenon in which some materials develop polarization
upon application of strains. Examples of piezoelectric materials are Rochelle salt, quartz, and lead
zirconate titanate or PZT (Pb (Zr,Ti) O3 ). Piezoelectric materials exhibit a linear relationship be-
tween the electric eld and strains for low eld values (up to 100 V=mm); and it exhibits nonlinear
behavior for large elds, and the material exhibits hysteresis [2]. Furthermore, piezoelectric materials
show dielectric aging and hence lack reproducibility of strains, i.e., a drift from zero state of strain
is observed under cyclic electric eld conditions. Terfenol-D, a magnetostrictive material [3], has
the characteristics of being able to produce strains up to 2500 m and energy density as high as
25000 J=m3 in response to a magnetic eld.
Vibration and shape control of 8exible structures is achieved with the help of actuators and a
feedback control law. Many modern techniques have been developed in recent years to meet the
challenge of designing controllers that suit the function under some required conditions. There have
been a number of studies on vibration control of 8exible structures using smart materials. Anjanappa
and Bi [4,5] investigated the feasibility of using embedded magnetostrictive mini actuators for smart
structure applications, such as vibration suppression of beams. A self-sensing magnetostrictive actu-
ator design based on a linear model of magnetostrictive transduction for Terfenol-D was developed
and analyzed by Pratt and Flatau [6]. Eda et al. [7] and Krishna Murty et al. [8,9] proposed mag-
netostrictive actuators that take advantage of the ease with which the actuators can be embedded
and remote excitation capability of magnetostrictive particles as actuators for smart structures. In
addition Pulliam et al. [10] provided very recent magnetostrictive particulate technology in damp-
ing applications. Reddy and Barbosa [11] presented a general formulation and analytical solution
for simply supported boundary conditions of laminated composite beams with embedded magne-
tostrictive layers. Using a combination of magnetostrictive and ferro-magnetic alloys, the combined
passive and active damping strategy was proposed by Bhattacharya et al. [12]. Beneddou [13] sur-
veyed more than 100 papers and discussed the research activity trends in piezoelectric nite element
modeling.
In the present study, control of the transient response of laminated composite plates with integrated
smart material layers, used as sensors and actuators, is studied using a uni ed plate theory that
includes the classical, rst-order and third-order plate theories as special cases. A simple negative
velocity feedback control is used to actively control the dynamic response of the structure through
a close-loop control. A displacement nite element model of the equations of motion governing the
uni ed theory is developed. The e1ects of material properties, lamination scheme, and placement of
the smart material layer on de8ection suppression are investigated.

2. Theoretical formulation

2.1. Introduction

The simplest equivalent single-layer theories are the classical laminate plate theory (CLPT) and
the rst-order shear deformation theory (FSDT). These theories describe the kinematics of most
laminated plates adequately. The third-order shear deformation theory (TSDT) represents the plate
kinematics better and yields better inter-laminar stress distributions. Quadratic variations of the
S.J. Lee et al. / Finite Elements in Analysis and Design 40 (2004) 463 – 483 465

transverse shear strains and stresses through the layer avoid the need for shear correction coeL-
cients as required in the rst-order theory.

2.2. Displacement =eld

The displacement eld for the third-order shear deformation theory (TSDT) can be expressed in
the form
 
3 9w0
u(x; y; z; t) = u0 (x; y; t) + zx (x; y; t) − c1 z x + ; (1)
9x
 
3 9w0
v(x; y; z; t) = v0 (x; y; t) + zy (x; y; t) − c1 z y + ; (2)
9y

w(x; y; z; t) = w0 (x; y; t); (3)


where (u0 ; v0 ; w0 ) and (x ; y ) have the same physical meaning as in the rst-order theory; they
denote the displacements and rotations of a transverse normal on the plane z = 0, respectively. Then
the displacement eld of FSDT is obtained by setting c1 = 0, and for the Reddy third-order theory
it is equal to c1 = 4=3h2 .

2.3. Equations of motion

The equations of motion of the third-order shear deformation theory are derived using the dynamic
version of the principle of virtual displacements [17].
 
9Nxx 9Nxy 92 u0 92 x 92 9w0
+ = I0 2 + J 1 − c 1 I3 2 ; (4)
9x 9y 9t 9t 2 9t 9x
 
9Nxy 9Nyy 9 2 v0 92  y 92 9w0
+ = I0 2 + J1 − c 1 I3 2 ; (5)
9x 9y 9t 9t 2 9t 9y
 2 
9QN x 9QN y 9 Pxx 92 Pxy 92 Pyy
+ + c1 +2 + +q
9x 9y 9x2 9x9y 9y2
 
92 w 0 2 92 9 2 w 0 9 2 w 0
= I0 − c 1 I6 2 +
9t 2 9t 9x2 9y2
    
92 9u0 9v0 92 9x 9y
+ c1 I3 2 + + J4 2 + ; (6)
9t 9x 9y 9t 9x 9y
 
9MN xx 9MN xy N 92 9w0
+ − Q x = 2 J 1 u0 + K 2  x − c 1 J4 ; (7)
9x 9y 9t 9x
 
9MN xy 9MN yy 9 2
9w 0
+ − QN y = 2 J1 v0 + K2 y − c1 J4 ; (8)
9x 9y 9t 9y
466 S.J. Lee et al. / Finite Elements in Analysis and Design 40 (2004) 463 – 483

where

MN  = M − c1 P ; QN  = Q − 3c1 R (;  = x; y);

N 
 k+1
Ii = (k) (z)i d z (i = 0; 1; 2; : : : ; 6); Ji = Ii − c1 Ii+2 (i = 1; 4);
k=1 k

4
K2 = I2 − 2c1 I4 + c12 I6 ; c1 = ; c2 = 3c1 (9)
3h2
and (Nxx ; Nyy ; Nxy ) denote the total in-plane force resultants, (Mxx ; Myy ; Mxy ) the moment resultants,
and (Pxx ; Pyy ; Pxy ) and (Rx ; Ry ) denote the higher-order stress resultants.
       

 Nxx  
  
 
 M 
 
  

   h=2  xx
  xx
  h=2  xx

Nyy = yy d z; Myy = yy z d z;

  −h=2     −h=2  
N   
   
M   
  
xy xy xy xy
   

 Pxx 
  
 xx 

  h=2  
Pyy = yy z 3 d z;

  −h=2  
P   
  
xy xy

         
Ry h=2 xz Qy h=2 xz
= z 2 d z; = d z: (10)
Rx −h=2 yz Qx −h=2 yz

The force and moment resultants are related to the strains by


     (0)   M 



{N } 

 
[A] [B] [E]   {# }     {N }  
  (1)   
{M } =   [B] [D] [F]   {# } − {M M
} ;

 
  
 
 

 {P}  [E] [F] [H ]  {#(3) }   {P M } 
     (0) 
{Q} [A] [D] {$ }
= ; (11)
{R} [D] [F] {$(2) }

N 
 zk+1
(Aij ; Bij ; Dij ; Eij ; Fij ; Hij ) = QN ij(k) (1; z; z 2 ; z 3 ; z 4 ; z 6 ) d z; (12)
k=1 zk

where the sti1nesses Aij ; Dij , and Fij are de ned for i; j = 1; 2; 6 as well as i; j = 4; 5. The sti1nesses
Bij ; Eij , and Hij are de ned only for i; j = 1; 2; 6. The coeLcients of Aij ; Bij ; Dij ; Eij ; Fij , and Hij are
given in terms of the layer sti1nesses QN ij and layer coordinates zk+1 and zk .
S.J. Lee et al. / Finite Elements in Analysis and Design 40 (2004) 463 – 483 467

The stress resultants associated with the magnetostrictive materials, {N M }; {M M }, and {P M } are
de ned by
 M  (k)  M  (k)

 N 
 Nm  zk+1   eN 
 
 M 
  eN 
Nm  zk+1  
xx 31 xx 31
         
M M
Nyy = eN 32 Hz d z; M yy = eN 32 Hz z d z;

       
 M  k=1 zk   eN    M
  k=1 zk   eN  
Nxy 36 Mxy 36

 M   (k)

 Pxx 
  
 eN 31 

  
N m zk+1  
M
Pyy = eN 32 Hz z 3 d z; (13)

 
 z 
 

 M  k=1 k  eN 
Pxy 36

where eN ij is the transformed moduli of the actuating/sensing material, which in the present study
is taken to be a magnetostrictive material, and Hz is the magnetic eld intensity which should be
excluded in the constitutive relations for the structural part of the composite structures.

2.4. Velocity feedback control

Considering velocity proportional closed-loop feedback control, the magnetic eld intensity H can
be expressed in terms of coil constant kc and coil current I (x; y; t) as

H (x; y; t) = kc I (x; y; t) (14)

and
nc 9w0
kc =  ; I (x; y; t) = c(t) ; (15)
bc + 4rc2
2 9t

where bc is the coil width, rc is coil radius, nc is number of turns in the coil, and c(t) is the control
gain which is assumed to be a constant in this study.

3. Finite element model

3.1. Virtual work statements

The weak forms of the equilibrium equations are


   2  
9+u0 9+u0 9 u0 92  x 92 9w0
0= Nxx + Nxy + +u0 I0 2 + J1 2 − c1 I3 2 d x dy
*e 9x 9y 9t 9t 9t 9x

− {+u0 (Nxx nx + Nxy ny )} ds; (16)
,
468 S.J. Lee et al. / Finite Elements in Analysis and Design 40 (2004) 463 – 483

   2  
9+v0 9+v0 9 v0 92 y 92 9w0
0= Nxy + Nyy + +v0 I0 2 + J1 2 − c1 I3 2 d x dy
*e 9x 9y 9t 9t 9t 9y

− {+v0 (Nxy nx + Nyy ny )} ds; (17)
,

   2 
9+w0 N 9+w0 N 9 +w0 92 +w0 92 +w0
0= Qx + Q y − c1 Pxx + 2 Pxy + pyy − +w0 q
*e 9x 9y 9x2 9x9y 9y2
  
92 w 0 2 9+w0 93 w0 9+w0 93 w0
+ +w0 I0 + c 1 I6 +
9t 2 9x 9x9t 2 9y 9y9t 2
    
9+w0 92 u0 9+w0 92 v0 9+w0 92 x 9+w0 92 y
− c1 I3 + + J4 + d x dy
9x 9t 2 9y 9t 2 9x 9t 2 9y 9t 2
 
9+wo
− +wo VN n ds − Pnn ds; (18)
9n
, ,

   2  
9+x N 9+x N 9 9w 0
0= M xx + M xy + +x QN x + +x J 1 u0 + K 2  x − c 1 J4 d x dy
*e 9x 9y 9t 2 9x

− {+x (MN xx nx + MN xy ny )} ds; (19)
,

   2  
9+y N 9+y N N 9 9w0
0= M xy + M yy + +y Qy + +y J 1 v 0 + K 2  y − c 1 J4 d x dy
*e 9x 9y 9t 2 9y

− {+y (MN xy nx + MN yy ny )} ds; (20)
,

where VN n is de ned as
    
N 9Pxx 9Pyy 9Pxx 9Pyy
V n = c1 + nx + + ny
9x 9y 9x 9y
    
9wO 0 9wO 0
− c1 I3 uO 0 + J4 ’O x − c1 I6 nx + I3 vO0 + J4 ’O y − c1 I6 ny
9x 9y
9Pns
+ (QN x nx + QN y ny ) + c1 : (21)
9s
The primary variables of the third-order theory are un ; us ; w0 ; 9w0 =9n; n ; s , where (un ; us ) denotes
in-plane normal and tangential displacements, and (n ; s ) are the rotations of a transverse line about
the in-plane normal and tangent. Lagrange interpolation functions of (un ; us ; n ; s ) and Hermite
S.J. Lee et al. / Finite Elements in Analysis and Design 40 (2004) 463 – 483 469

interpolation function for w0 are used for the formulation of the displacement nite element model.
A conforming element that has eight degrees of freedom (u0 ; v0 ; w0 ; w0; x ; w0; y ; w0; xy ; x ; y ) is used
in this study.

3.2. Semidiscrete =nite element model

The generalized displacements are approximated over an element *e by the expressions


m

ux (x; y; t) = uie (t) i
e
(x; y); (22)
i=1

m

vx (x; y; t) = vie (t) i
e
(x; y); (23)
i=1

m

w0 (x; y; t) = 4N ei (t)’ei (x; y); (24)
i=1

m

x (x; y; t) = Xie (t) i
e
(x; y); (25)
i=1

m

y (x; y; t) = Yie (t) i
e
(x; y); (26)
i=1

where ie denotes the Lagrange interpolation functions and ’ei are the Hermite interpolation functions.
Here we chose the same approximation for the in-plane displacements (u0 ; v0 ) and rotations (x ; y ),
although one could use di1erent approximations for these two pairs. In the case of the conforming
element, the four nodal values associated with w0 are
9w0 9w0 92 w0
4N 1 = w0 ; 4N 2 = ; 4N 3 = ; 4N 4 = :
9x 9y 9x9y
The nite element model is of the compact form
n
5 

(Kij 4j + Cij 4̇j + Mij 4O j ) − Fi = 0; i = 1; 2; : : : ; n; (27)
=1 j=1

where  = 1; 2; 3; 4; 5; n1 = n2 = n4 = n5 = 4 and n3 = 16.


The nodal values 4j are 41j = uj ; 42j = vj ; 43j = 4N j ; 44j = Xj ; 45j = Yj , and damping coeLcients are
de ned by
  N
m
 
13 9 i
Cij = dc(t) (Zk+1 − Zk ) ’j d x dy; (28)
*e 9x
k=1
470 S.J. Lee et al. / Finite Elements in Analysis and Design 40 (2004) 463 – 483

  N
 m
 
9 i
Cij23 = dc(t) (Zk+1 − Zk ) ’j d x dy; (29)
*e 9y
k=1

  N
 m

2 2
 
1 9 ’ i 9 ’ i
Cij33 = dc(t)(−c1 ) 4
(Zk+1 − Zk4 ) + ’j d x dy; (30)
* e 4 9x 2 9y 2
k=1

   N
 m
 N
 m
 
1 2 1 4 9 i
Cij43 = dc(t) 2
(Z − Zk ) − c1 4
(Z − Zk ) ’j d x dy; (31)
*e 2 k+1 4 k+1 9x
k=1 k=1

   N
 m
 N
 m
 
1 1 9 i
Cij53 = dc(t) (Z 2 − Zk2 ) − c1 (Z 4 − Zk4 ) ’j d x dy; (32)
*e 2 k+1 4 k+1 9y
k=1 k=1

where dc(t)=em dm kc c(t) and em is modulus of the magnetostrictive layer, dm the magneto-mechanical
coupling coeLcient, and kc and c(t) are de ned in the previous section. For the details of the sti1ness
and mass coeLcients, see Reddy [14,15].
This completes the general nite element model development of the third-order shear deformation
plate theory.

3.3. Transient analysis

The equations of motion can be solved exactly using analytical methods, but those are algebraically
complicated and require the determination of eigenvalues and eigenfunctions, as in the state-space
approach. Newmark method that takes advantage of the static solution form for spatial variation and
uses a numerical method to solve the resulting di1erential equations in time was used to determine
the transient response of composite laminates in this study.
The second-order di1erential equation (27) can be expressed in matrix form as

[M e ]{uO e } + [C e ]{u̇ e } + [K e ]{ue } = {F e }: (33)

Using the Newmark’s scheme, Eq. (33) can be reduced to the form

[K̂]s+1 {4}s+1 = {F̂}s;s+1 ; (34)

where

[K̂]s+1 = [K]s+1 + a3 [M ]s+1 + a6 [C]s+1 ; (35)

{F̂}s; s+1 = {F}s+1 + [M ]s+1 {A}s + [C]s+1 {B}s ; (36)

{A}s = a3 {u}s + a4 {u̇}s + a5 {u}


O s; {B}s = a6 {u}s + a7 {u̇}s + a8 {u}
O s; (37)
S.J. Lee et al. / Finite Elements in Analysis and Design 40 (2004) 463 – 483 471

2 (1 − $)
a1 = (1 − )Tt; a2 = Tt; a3 = ; a4 = a3 Tt; a5 = ;
$(Tt)2 $
 
2 2 
a6 = ; a7 = − 1; a8 = Tt −1 (38)
$Tt $ $

and  and $ are parameters that determine the stability and accuracy of the scheme.
In this numerical method, the time derivatives are approximated using di1erence approximations,
and therefore solution is obtained only for discrete times and not as a continuous function of time.

4. Numerical results and discussion

Numerical studies are carried out to analyze the de8ection suppression characteristics for di1erent
lamination schemes and boundary conditions using the developed TSDT nite element models. The
baseline of the simulations is the simply supported square laminate (a=b = 1; a=h = 10; see Fig. 1)

Fig. 1. Laminated composite plate con gurations and nite element meshes.
472 S.J. Lee et al. / Finite Elements in Analysis and Design 40 (2004) 463 – 483

Table 1
Material properties of magnetostrictive and elastic composite materials

Material E1 (GPa) E2 (GPa) G13 (GPa) G23 (GPa) G12 (GPa) ;12  (Kg m−3 ) Dk (10−8 mA−1 )

Terfenol-D 26.5 26.5 13.25 13.25 13.25 0 9250 1.67


CFRP 138.6 8.27 4.96 4.12 4.96 0.26 1824 —
Gr-Ep(AS) 137.9 8.96 7.20 6.21 7.20 0.30 1450 —
Gl-Ep 53.78 17.93 8.96 3.45 8.96 0.25 1900 —
Br-Ep 206.9 20.69 6.9 4.14 6.9 0.30 1950 —

Fig. 2. Two types of simply supported boundary conditions, SS-1and SS-2, for laminated plates.

under sinusoidal distributions of the initial velocity eld


9w <x <y
(x; y; t = 0) = sin sin :
9t a b
The time step selected in the present study is Tt = 0:0005 s. The notation for lamination scheme
(=1 ; =2 ; =3 ; =4 ; m)s means that there are 10 layers symmetrically placed about the midplane with the
ber orientations being (=1 ; =2 ; =3 ; =4 ; m; m; =4 ; =3 ; =2 ; =1 ), where m stands for the magnetostrictive layer
and subscript s stands for symmetric. The material properties of Terfenol-D and the elastic composite
materials are listed in Table 1.
In nite element analysis, solution symmetries should be taken advantage of by identifying the
computational domain to reduce computational e1ort. For a laminated composite plate with all edges
simply supported or clamped, a quadrant of the plate may be used as the computational domain.
Fig. 2 shows two types of simply supported boundary conditions for the third-order shear deformation
theory. Fig. 3 shows the e1ects of the nite element results for the di1erent laminations and boundary
S.J. Lee et al. / Finite Elements in Analysis and Design 40 (2004) 463 – 483 473

conditions. Quarter plate models with proper boundary conditions can be used in the antisymmetric
laminates with simply supported boundary condition, but not for laminated plates with the clamped
edges. For symmetric laminates, the simply supported cross-ply laminates can be modelled as a
quarter plate. The boundary conditions along a line of symmetry should be correctly identi ed and
0.0006 SS1 Cross-ply Quarter
SS1 Cross-ply Full
0.0004 SS2 Angle-ply Quarter
SS2 Angle-ply Full
SS2 General angle-ply Quarter
Displacement (m)

0.0002 SS2 General angle-ply Full

0.0000

-0.0002

-0.0004

-0.0006
0.010 0.012 0.014 0.016 0.018 0.020
(a) Time (sec)
0.0006 Quarter Cross-ply
Full Cross-ply
0.0004 Quarter Angle-ply
Full Angle-ply
Quarter General angle-ply
Displacement (m)

0.0002 Full General angle-ply

0.0000

-0.0002

-0.0004

-0.0006
0.010 0.012 0.014 0.016 0.018 0.020
(b) Time (sec)
0.0006 SS Cross-ply Quarter
SS Cross-ply Full
0.0004 SS Angle-ply Quarter
SS Angle-ply Full
SS General angle-ply Quarter
Displacement (m)

0.0002 SS General angle-ply Full

0.0000

-0.0002

-0.0004

-0.0006
0.010 0.012 0.014 0.016 0.018 0.020
(c) Time (sec)

Fig. 3. E1ects of using full and quarter plate models in nite element modelling of CFRP composite plates; cross-ply
(m; 90; 0; 90; 0)s or anti-s , angle-ply (m; 30; −30; 30−30)s or anti-s , and general angle-ply (m; 45; −45; 90; 0)s or anti-s : (a) simply
supported antisymmetric laminate, (b) clamped antisymmetric laminates, (c) simply supported symmetric laminates, and
(d) clamped symmetric laminates.
474 S.J. Lee et al. / Finite Elements in Analysis and Design 40 (2004) 463 – 483

0.0006 Quarter Cross-ply


Full Cross-ply
0.0004 Quarter Angle-ply
Full Angle-ply
Quarter General angle-ply
Displacement (m)

0.0002 Full General angle-ply

0.0000

-0.0002

-0.0004

-0.0006
0.010 0.012 0.014 0.016 0.018 0.020
(d) Time (sec)

Fig. 3. Continued.

0.010
Analytic solution FEA solution (dt=0.0005)
FEA solution (dt=0.0001) Uncontrolled motion
0.005
Displacement (m)

0.000

-0.005

-0.010
0.00 0.05 0.10 0.15 0.20
Time (sec)

Fig. 4. Comparison of the center de8ection predicted by the analytical and nite element methods for the case of symmetric
cross-ply CFRP laminates (m; 90; 0; 90; 0)s .

imposed in the nite element model. When one is not sure of the solution symmetry, it is advised
that the whole plate be modelled.

4.1. Simply supported composite laminates

To compare with the analytical results, the SS-1 boundary conditions and quarter plate model
have been used for symmetric cross-ply laminates. The de8ections predicted from the analytical
(eigenvalue analysis) [16] and transient nite element analysis are within the reasonable agreement,
as shown in Fig. 4. Fig. 5 shows the central displacements using the di1erent plate theories (CLPT,
FSDT, and TSDT) for two di1erent lamination schemes. It is observed that CLPT shows higher
de8ection suppression capacity in both cases. This is expected because the CLPT renders plate
sti1er compared to the other theories (also see Table 2).
S.J. Lee et al. / Finite Elements in Analysis and Design 40 (2004) 463 – 483 475

0.0006
TSDT FSDT (K=5/6) CLPT
0.0004
Displacement (m)

0.0002

0.0000

-0.0002

-0.0004

-0.0006
0.00 0.01 0.02 0.03 0.04 0.05
(a) Time (sec)

0.0006
TSDT FSDT (K=5/6) CLPT
0.0004

0.0002
Displacement (m)

0.0000

-0.0002

-0.0004

-0.0006
0.00 0.01 0.02 0.03
(b) Time (sec)

Fig. 5. Comparison of the center displacements by the di1erent plate theories for simply supported cross-ply CFRP
laminates: (a) laminate (m; 90; 0; 90; 0)s , and (b) laminate (0; 90; 0; 90; m)s .

After studying the in8uence of lamina material properties on the amplitude of vibration and
vibration suppression times, it is observed that materials having the almost same E1 =E2 ratio have
similar vibration suppression characteristics. Fig. 6 shows the de8ection damping characteristics for
the di1erent laminate materials.
The de8ection suppression time is de ned as the time required to reduce the uncontrolled center
de8ection to one-tenth of its magnitude. The de8ection suppression time ratio (suppression time
divided by the maximum suppression time) can be shown to be Ts = hm =2zm , where hm is the
thickness of the magnetostrictive layer and zm is the positive distance between the midplane of
the magnetostrictive layer and the midplane of the plate. The maximum de8ections (Wmax ) of the
composite plate and the suppression times for the di1erent position of smart layers are presented in
Table 3. The e1ect of the smart layer positions on the vibration suppression can be shown in Fig. 7.
It is observed that as the smart material layer is moved farther from the midplane the suppression
time decreases, as may be expected because of the moment e1ect by smart layer actuations.
The e1ect of the thickness of smart-material layer on de8ection damping characteristics is studied
next. It is observed that thicker smart material layers result in better attenuation of the de8ection.
476 S.J. Lee et al. / Finite Elements in Analysis and Design 40 (2004) 463 – 483

Table 2
Selected center displacement values on the symmetric cross-ply CFRP laminates for the di1erent laminate theories

t Center displacement (m)

(m; 90; 0; 90; 0)s laminate (90; 0; 90; 0; m)s laminate

CLPT FSDT TSDT CLPT FSDT TSDT

0.0005 3.93E-04 3.92E-04 3.89E-04 3.94E-04 4.00E-04 3.89E-04


0.0010 5.21E-04 5.06E-04 4.87E-04 4.87E-04 4.92E-04 4.57E-04
0.0015 3.32E-04 2.96E-04 2.57E-04 2.11E-04 2.09E-04 1.51E-04
0.0020 −3:72E-05 −8:25E-05 −1:20E-04 −2:21E-04 −2:30E-04 −2:74E-04
0.0025 −3:52E-04 −3:75E-04 −3:85E-04 −4:82E-04 −4:90E-04 −4:72E-04
0.0030 −4:36E-04 −4:11E-04 −3:73E-04 −3:78E-04 −3:75E-04 −2:83E-04
0.0035 −2:54E-04 −1:86E-04 −1:18E-04 1:18E-05 2:37E-05 1:35E-04
0.0040 6.08E-05 1.34E-04 1.91E-04 3.88E-04 4.01E-04 4.38E-04
0.0045 3.15E-04 3.46E-04 3.47E-04 4.68E-04 4.70E-04 3.81E-04
0.0050 3.61E-04 3.22E-04 2.61E-04 1.94E-04 1.81E-04 1.35E-05

0.0006
CFRP Gr-Ep
0.0004
Gl-Ep Br-Ep
Displacement (m)

0.0002

0.0000

-0.0002

-0.0004

-0.0006
0.00 0.02 0.04 0.06
Time (sec)

Fig. 6. The e1ect of the lamina material properties on the damping of de8ection in symmetric cross-ply laminates
(m; 90; 0; 90; 0)s .

Table 3
Vibration suppression time for the di1erent smart layer positions on the symmetric cross-ply CFRP laminate (m; 90; 0; 90; 0)s

Lamination scheme zm (m) Ts Wmax (10−4 m) t at Wmax =10

(m; 90; 0; 90; 0)s 0.045 0.111 5.21 0.0285


(90; m; 90; 0; 90)s 0.035 0.143 5.09 0.0350
(0; 90; m; 90; 0)s 0.025 0.200 4.90 0.0480
(90; 0; 90; m; 90)s 0.015 0.333 4.85 0.0850
(0; 90; 0; 90; m)s 0.005 1.000 4.87 0.2560
S.J. Lee et al. / Finite Elements in Analysis and Design 40 (2004) 463 – 483 477

0.0006
(m ,90,0,90, 0) s (90,m ,90, 0,90)s
0.0004 (0,90,m ,90,0 )s (90,0,90, m ,90)s
Displacement (m)

0.0002

0.0000

-0.0002

-0.0004

-0.0006
0.00 0.02 0.04 0.06 0.08 0.10
Time (sec)

Fig. 7. The e1ect of the smart material layer position on the de8ection for the symmetric cross-ply CFRP laminates.

Table 4
Suppression times for the di1erent smart layer thicknesses in symmetric cross-ply laminates (m; 90; 0; 90; 0)s

Lamina thickness (mm) zm (m) Ts Wmax (10−4 m) t at Wmax =10

he = 10; hm = 2 0.0410 0.0244 4.77 0.0780


he = 10; hm = 4 0.0420 0.0476 4.98 0.0485
he = 10; hm = 5 0.0425 0.0588 5.05 0.0420
he = 10; hm = 6 0.0430 0.0698 5.10 0.0350
he = 10; hm = 8 0.0440 0.0909 5.17 0.0320
he = 10; hm = 10 0.0450 0.1111 5.21 0.0285
he = 5; hm = 5 0.0225 0.1111 9.12 0.0310
he = 5; hm = 10 0.0250 0.2 9.45 0.0400

This is due to a larger mass inertia that is caused by the large increase in the moment of inertia of the
system when thickness of the smart material layer is increased. We note that the smart material layer
has a density of ve times that of the composite material. The suppression times and characteristics
for di1erent smart layer thicknesses are shown in Table 4 and Fig. 8.
Fig. 9 shows the e1ect of the feedback coeLcient c(t)kc on the vibration suppression character-
istics. Two di1erent values of the feedback coeLcient are used; 104 and 103 . It can be seen that
the suppression time increases when the value of the feedback coeLcient decrease. This is expected
because the coeLcients of the damping matrix decrease, thereby resulting in less damping.
The de8ection damping characteristics of symmetric angle-ply and general angle-ply composite
laminates are studied using full plate FE models. Observations made earlier on various characteristics
such as the e1ects of smart layer position, its thickness, and magnitude of the feedback coeLcient
are also valid for these laminates, as shown in Figs. 10 and 11.
478 S.J. Lee et al. / Finite Elements in Analysis and Design 40 (2004) 463 – 483

0.0010
he=10, hm=10 he=10, hm=5
he=5, hm=10 he=5, hm=5
0.0006
Displacement (m)

0.0002

-0.0002

-0.0006

-0.0010
0.00 0.01 0.02 0.03
(a) Time (sec)

0.0006
he=10, hm=2 he=10, hm=4
he=10, hm=6 he=10, hm=8
0.0004 he=10, hm=10
Displacement (m)

0.0002

0.0000

-0.0002

-0.0004

-0.0006
0.00 0.01 0.02 0.03
(b) Time (sec)

Fig. 8. (a,b) The e1ect of the thickness of smart material layers on the de8ection damping characteristics of symmetric
cross-ply laminates (m; 90; 0; 90; 0)s .

3
c(t)Kc=1000
2.5
Suppression Time (sec)

c(t)Kc=10000
2

1.5

0.5

0
0.005 0.015 0.025 0.035 0.045
Distance from the mid-plane (m)

Fig. 9. E1ects of the magnitude of the feedback coeLcients on the suppression time for symmetric cross-ply CFRP
laminates (m; 90; 0; 90; 0)s .
S.J. Lee et al. / Finite Elements in Analysis and Design 40 (2004) 463 – 483 479

0.0006
(m,45,-45,45,-45)s
0.0004 (45,-45,m,45,-45)s
Displacement (m)

0.0002

0.0000

-0.0002

-0.0004

-0.0006
0.00 0.02 0.04 0.06 0.08 0.10
(a) Time (sec)
0.0006
(m,30,-30,30,-30)s
0.0004 (m,45,-45,45,-45)s
(m,60,-60,60,-60)s
Displacement (m)

0.0002

0.0000

-0.0002

-0.0004

-0.0006
0.00 0.01 0.02 0.03 0.04 0.05
(b) Time (sec)

Fig. 10. (a,b) Center displacement versus time for symmetric angle-ply CFRP laminates.

4.2. Clamped composite laminates

Next, fully clamped laminated plates are analyzed using 8 × 8 mesh in a full plate. The e1ect
of the boundary conditions on the de8ection is shown in Fig. 12. The maximum displacements of
the simply supported plate are greater than those of the clamped case, which is expected. Simply
supported laminates, which have larger displacements, take less suppression time compared to the
clamped laminates.

4.3. EAects of the mechanical loading

Numerical studies are also carried out to analyze smart composite laminates under uniformly dis-
tributed load q0 instead of speci ed initial velocity eld. Fig. 13 shows the center de8ection for
selected simply supported and clamped laminates under continuously applied uniformly distributed
loading, while Fig. 14 shows the case under suddenly applied step loading. The e1ect of sinusoidal
loading on the central displacement has been studied. The results of symmetric cross-ply laminates
with simply supported boundary conditions and subjected to sinusoidal and uniformly distributed
loads are shown in Fig. 15. The transverse displacements in Figs. 13–15 are plotted in the nondi-
mensionalized forms as w0 (1 0 0) E2 h3 =b4 q0 .
480 S.J. Lee et al. / Finite Elements in Analysis and Design 40 (2004) 463 – 483

0.0006
(m,45,-45,0,90)s
Displ a cement (m) 0.0004 (45,m,-45,0, 90)s
(45,-45,m,90,0)s
0.0002

0.0000

-0.0002

-0.0004

-0.0006
0.00 0.02 0.04 0.06 0.08 0.10
(a) Time (sec)

0.0004
(45,-45,0,90,30,-30,m,60)s
(45,-45,0,90,30,m,-30,60)s
0.0002 (45,-45,0,90,m,30,-30,60)s
Displacement (m)

0.0000

-0.0002

-0.0004
0.0 0.1 0.2 0.3 0.4 0.5
(b) Time (sec)

0.0004
(45,-45,0,m,90,30,-30,60)s
(45,-45,m,0,90,30,-30,60)s
0.0002 (45,m,-45,0,90,30,-30,60)s
Displacement (m)

(m,45,-45,0,90,30,-30,60)s

0.0000

-0.0002

-0.0004
0.00 0.02 0.04 0.06 0.08 0.10
(c) Time (sec)

Fig. 11. (a–c) The e1ect of the position of the smart layer on the de8ection of symmetric general angle-ply CFRP
laminates.

5. Conclusions

A uni ed third-order plate theory is used to develop a displacement nite element model of
laminated composite plates with smart material layers for transient response. The uni ed formulation
S.J. Lee et al. / Finite Elements in Analysis and Design 40 (2004) 463 – 483 481

0.0010
C(m,90,0,90,0)s C(90,m,90,0,90)s
C(0,90,m,90,0)s C(90,0,90,m,90)s
S(m,90,0,90,0)s S(90,m,90,0,90)s
0.0005 S(0,90,m,90,0)s S(90,0,90,m,90)s
Displacement (m)

0.0000

-0.0005

-0.0010
0.000 0.002 0.004 0.006 0.008 0.010
(a) Time (sec)

0.0004
C(m,90,0,90,0)s
C(90,m,90,0,90)s
0.0002 C(0,90,m,90,0)s
Displacement (m)

C(90,0,90,m,90)s

0.0000

-0.0002

-0.0004
0.00 0.05 0.10 0.15 0.20
(b) Time (sec)

Fig. 12. The e1ect of boundary conditions on the center displacements for the di1erent laminates: (a) comparison of
simply supported and clamped laminates, and (b) results for di1erent smart layer position.

2.50
Clamped plate (m,90,0,90,0)s
Nondimensionalized Displacement

Simply supported plate (m,90,0,90,0)s


2.00
Clapmed plate (90,m,90,0,90)s
Simply supported plate (90,m,90,0,90)s
1.50

1.00

0.50

0.00
0.00 0.04 0.08 0.12 0.16 0.20
Time (sec)

Fig. 13. Nondimensionalized center de8ection versus time for simply supported and clamped laminated plates under
uniform load.
482 S.J. Lee et al. / Finite Elements in Analysis and Design 40 (2004) 463 – 483

1.00
Nondimensionalized Displacement Clamped plate (m,90,0,90,0)s
Clamped plate (90,m,90,0,90)s
0.50 Simply supported plate (m,90,0,90,0)s
Simply supported plate (90,m,90,0,90)s

0.00

-0.50

-1.00
0.00 0.02 0.04 0.06 0.08 0.10
Time (sec)

Fig. 14. Nondimensionalized center de8ection versus time for simply supported and clamped laminated plates under
suddenly applied uniform load.

2.50
Uniform loading (m,90,0,90,0)s
Nondimensionalized Displacement

Uniform loading (90,m,90,0,90)s


2.00
Sinusoidal loading (m,90,0,90,0)s
Sinusoidal loading (90,m,90,0,90)s
1.50

1.00

0.50

0.00
0.00 0.04 0.08 0.12 0.16 0.20
Time (sec)

Fig. 15. Comparison of center displacement versus time for simply supported laminated plates under sinusoidal and
uniformly distributed loads.

includes the classical, rst-order, and third-order shear deformation plate theories as special cases.
The smart material used in this study to achieve damping of transverse de8ection is the Terfenol-D
magnetostrictive material, although in principle any other actuating material can be used. The negative
velocity feedback control is used to damp the transient response. A number of parametric studies
were carried out to understand the damping characteristics of various laminates with embedded
smart-material layers. Some of the observations are summarized below.
Use of quarter plate models in place of full plate models was studied rst. It is found that
for antisymmetric cross-ply, angle-ply, and general angle-ply laminates and symmetric cross-ply
laminates with simply supported boundary conditions, a quadrant of the plate with proper symmetry
boundary conditions may be used to reduce the computational e1ort.
The maximum damping of de8ection occurs when the smart material layers are placed farthest from
the midplane because the bending moments generated by the smart material layer are maximum when
S.J. Lee et al. / Finite Elements in Analysis and Design 40 (2004) 463 – 483 483

they are away from the midplane. It is observed that for a lower value of the feedback coeLcient the
time taken to damp de8ection is longer. The damping tendencies of symmetric cross-ply, angle-ply
and general angle-ply composite laminates have been found to be similar. The damping characteristics
of a fully clamped composite plate are similar to those of a simply supported composite plate.

Acknowledgements

The rst two authors acknowledge the support of the work by the Army Research OLce (ARO)
through Grant DAAD19-01-1-0483.

References

[1] G.A. Maugin, Continuum Mechanics of Electromagnetic Solids, North-Holland, Amsterdam, 1988.
[2] K. Uchino, Electrostrictive actuators: materials and applications, Cer. Bull. 65 (1986) 647–652.
[3] M.J. Goodfriend, K.M. Shoop, Adaptive characteristics of the magnetostrictive alloy, Terfenol-D, for active vibration
control, J. Intell. Mater. Systems Struct. 3 (1992) 245–254.
[4] M. Anjanappa, J. Bi, A theoretical and experimental study of magnetostrictive mini actuators, Smart Mater. Struct.
3 (1994) 83–91.
[5] M. Anjanappa, J. Bi, Magnetostrictive mini actuators for smart structural application, Smart Mater. Struct. 3 (1994)
383–390.
[6] J.R. Pratt, A.B. Flatau, Development and analysis of self-sensing magnetostrictive actuator design, J. Intell. Mater.
Systems Struct. 6 (1995) 639–648.
[7] H. Eda, T. Kobayashi, H. Nakamura, T. Akiyama, Giant magnetostriction compounds with structure textured by
resin bound on giant magnetostriction ne powder in magnetic eld and its actuator, Trans. Japanese Soc. Mech.
Eng. Ser. C 61 (1995) 168–170.
[8] A.V. Krishna Murty, M. Anjanappa, Y.-F. Wu, The use of magnetostrictive particle actuators for vibration attenuation
of 8exible beams, J. Sound Vib. 206 (1997) 133–149.
[9] A.V. Krishna Murty, M. Anjanappa, Y.-F. Wu, B. Bhattacharya, M.S. Bhat, Vibration suppression of laminated
composite beams using embedded magneto-strictive layers, J. A-S 78 (1998) 38–44.
[10] W. Pulliam, G. Mcknight, G. Carman, Recent advances in magnetostrictive particulate composite technology, Proc.
SPIE, Smart Struct. Mater. 4698 (2002) 271–281.
[11] J.N. Reddy, J.I. Barbosa, Vibration suppression of laminated composite beams, Smart Mater. Struct. 9 (2000)
49–58.
[12] B. Bhattacharya, B.R. Vidyashankar, S. Patsias, G.R. Tomlinson, Active and passive vibration control of 8exible
structures using a combination of magnetostrictive and ferro-magnetic alloys, Proc. SPIE, Smart Struct. Mater. 4073
(2000) 204–214.
[13] A. Benjeddou, Advances in piezoelectric nite element modeling of adaptive structural elements: a survey, Comput.
& Struct. 76 (2000) 347–363.
[14] J.N. Reddy, On laminated composite plates with integrated sensors and actuators, Eng. Struct. 21 (1999) 568–593.
[15] J.N. Reddy, Mechanics of Laminated Composite Plates: Theory and Analysis, CRC Press, Boca Raton, FL, 1997.
[16] F. Rostam-Abadi, J.N. Reddy, S.J. Lee, Vibration suppression of cross-ply laminated plates with magnetostrictive
layers, Proceedings of SECTAM XXI, May 2002.
[17] J.N. Reddy, Energy Principles and Variational Methods in Applied Mechanics, Wiley, New York, 2002.