You are on page 1of 44

Chapter 2 MATHEMATICAL MODEL

1. INTRODUCTION

In this chapter, a hierarchical finite element model for geometrically non-linear vibration in unsymmetrical laminated plates made of composite materials is developed. The model will be presented in the time domain. 2. MATHEMATICAL MODEL

2.1 - Field Equations In this section, a hierarchical finite element method for asymmetric composite plates is presented. The plate with constant thickness h, width a and length b, is composed of orthotropic layers oriented at different angles . The origin of the co-ordinate system is located at the middle plane with the z-axis being normal to the mid-plane. Using the first order shear deformation theory (FSDT), Kirchhoffs hypothesis is relaxed by removing the third part, i.e., the transverse normals do not remain perpendicular to the midsurface after deformation. The inextensibility of transverse normals requires that w is not a function of the thickness coordinate, z. The displacement field (Figure 1) of the first-order theory is of the form [2.1] u(x, y, z, t ) = u0(x, y, t )+zy (x, y, t ) v(x, y, z, t ) = v0(x, y, t) -zx (x ,y ,t ) w(x, y, z, t) = w0(x, y, t) (2. 1) (2. 2) (2. 3)

Chapter 2 Mathematical Model


z, w0 y, , v0

27

y x, , u0 x

Figure 1 - Plate Element, coordinates, mid-plane displacements and rotations

where (u0, v0, w0) are the displacements along the coordinate lines of a material point on the xy plane, therefore in the mid-plane (z=0), and where x and y are

u v = y , = x z z
the rotations of a transverse normal about the x-axis and the y-axis, respectively. The functions (u0, v0, w0, x, y) are unknown and are to be determined. For thin plates, i.e., when the plate in-plane characteristic dimension to thickness ratio is on the order 50 or higher (a/h50) [2.2], the rotation functions x and y should approach the respective slopes of the transverse deflection: x =

w w0 , y = 0 x y

In this case, the first order shear deformation theory becomes identical to the classical plate theory where Kirchhoffs hypothesis is followed [2.5]. The non-linear strains associated with the displacement field in Figure 1 are

x x0 zx 0 y y z y 0 xy = xy z xy 0 0 zx zx 0 0 yz yz

(2. 4)

Chapter 2 Mathematical Model

28

where x 0 , y 0 and xy 0 are the in-plane strain components at z = 0 defined by the von Krmn non-linear strain-displacement relationships[2.3]:

1 1 x 0 = u,0x + ( w,0x ) 2 , y 0 = v,0y + ( w,0y ) 2 , xy 0 = u,0y + v,0x + w,0x w,0y 2 2 where u,0x represents the partial derivate Figure 2 shows these displacements.
y
Deformed Shape

(2. 5)

u0 . x

y
Deformed Shape

v y0 x0 u z y x

xy xy
v x Shear Strains

Normal displacement

w x Transverse displacement Figure 2 Displacements of a plate

The terms x , y and xy in equation (2. 4) are the curvatures or bending strains, which are given by:
y 0 0 x x + , y = and xy = x = y y x x 0 y
0

(2. 6)

The transverse shear strains are


0 0 0 0 0 0 zx = w, x + y , yz = w, y x

(2. 7)

Chapter 2 Mathematical Model

29

For each element, the middle plane in-plane displacements and the rotations are expressed in the form:

qu u0 q 0 v v 0 q w = [N ] w 0 q y y 0 x q x
where {qu } ,

(2. 8)

{qv }

and

{qw }

are the vectors of generalised in and out-of-plane


x

displacements, and q y

{ } and {q }
0 0 0 0 0

are the vectors of generalised rotations. The

complete matrix of shape functions

{ N u }T 0 [N ] = 0 0 0

{N }
0 0 0

u T

{N }
0 0

w T

{N }
y

0 0 0 x T N
0

(2. 9)

is constituted by the row vectors of bi-dimensional in-plane, out-of-plane and rotational shape functions, which are, respectively,

{ N } = { g ( ) g ( ) , g ( ) g ( ) ,..., g ( ) g ( )}
u T 1 1 1 2 pi pi

(2. 10) (2. 11)

{ N } = { f ( ) f ( ) , f ( ) f ( ) ,..., f ( ) f ( )}
w T 1 1 1 2 po po

{N }
y

= y1 ( ) y1 ( ) , y1 ( ) y2 ( ) , , y p ( ) y p ( )
y y

(2. 12)
(2. 13)

{N }
x

= x1 ( ) x1 ( ) , x1 ( ) x2 ( ) ,..., xp

( ) x ( )
p x

The vectors {g}, {f}, { y } and { x } are the vectors of in-plane, transverse, and rotational one dimensional displacement shape functions; po, pi, p y and p x are the

Chapter 2 Mathematical Model

30

numbers of respective transverse, middle plane, rotation about y and rotation about x, displacement shape functions employed; and are the local coordinates, which are given by:
= 2x a , = 2 y b

(2. 14)

In the hierarchical finite element method, one is free to choose the number and set of displacement shape functions to be applied in the definition of the element. Increasing the order of the shape functions that represent the displacements within the element increases the accuracy of the element. In a plate element, there is a set of shape functions for the transverse displacements, a set of shape functions for the inplane displacements and one set of shape functions for each rotation. An element of a layer is now considered and the plate geometric axes are x and y, as in Figure 3. The principal material axes are labelled 1 and 2, that is, the 1 direction is parallel to the fibbers and the 2 direction is normal to them.
y y yx

2
dx
x dy

xy
1
x

xy

yx y

Figure 3 Laminae Coordinate System

Mohrs circle analysis in basic strength of materials can be applied to establish that
1 2 = [T ]CL 6

x y xy

(2. 15)

Chapter 2 Mathematical Model

31

Analogously, the strain relationship is given by


1 2 = [T ]CL 12

x y , xy

(2. 16)

where
m2 n2 +2mn 2 2 2mn [T ]CL = n m mn mn (m 2 n 2 )

(2. 17)

where m = cos ( ) , n = sin ( ) , and is the positive angle defined, and CL refer to the 1-2 plane only. The effects of transverse shear deformation are shown in the inclusion of the relations 4 4 and 5 5 in composite materials [2.4], therefore the equations (2. 15)-(2. 17) are modified to:
x x 1 1 y y 2 2 4 = [T ] yz and 4 = [T ] yz 5 5 zx xz xy 6 6 xy

(2. 18)

Where
m2 2mn n2 0 0 2 2 0 0 m 2mn n 0 [T ] = 0 0 m n 0 n m 0 0 mn mn 0 0 (m 2 n 2 )

(2. 19)

Multiplying (2. 18) by [T ] , the relationships can be written as


1

x x 1 1 y y 2 2 1 1 yz = [T ] 4 and yz = [T ] 4 zx xz 5 5 6 6 xy xy

(2. 20)

Chapter 2 Mathematical Model

32

The [T ] matrix is obtained by replacing by - in [T ] , and is given by


1

m2 0 0 n2 2mn 2 2 0 0 2mn n m 1 [T ] = 0 0 m n 0 0 0 n m 0 mn mn 0 0 (m 2 n 2 )

(2. 21)

Using Hookes law that relates stresses and strains, the general equations for a lamina of composite material in terms of the principal material directions (1, 2, 3), where 3 would represent the z-axis, are given by:
1 2 4 = 5 6 Q11 Q 12 0 0 0 Q12 Q22 0 0 0 0 0 2Q44 0 0 0 0 0 2Q55 0 0 1 0 2 0 . 4 0 5 2Q66 6

(2. 22)

Qij are defined in reference [2.4] and are given by:


Q11 = E1 (1 23 32 ) / Q = E (1 ) / 2 31 13 22 Q44 = G23 Q55 = G13 Q = G 12 66 = Q E ( 12 1 21 + 31 23 ) / = E2 ( 12 + 32 13 ) /

(2. 23)

where = 1 12 21 23 32 3113 2 21 32 13 . Displacements in the z-direction are not considered, therefore 23 = 32 = 31 = 13 =0 and (2. 23) comes

Chapter 2 Mathematical Model

33

E1 Q11 = 1 12 21 E2 Q22 = 1 12 21 Q = 21 E1 12 1 12 21 Q44 = G23 Q = G 13 55 Q66 = G12

(2. 24)

Here, E1 and E2 are the major and minor Youngs moduli; 12 and 21 are the Poissons ratios; G12 is the shear modulus. [2.4] 1 and 2 denote the principal directions of the plate layer. A shear correction factor, =5/6, that accounts for the fact that the shear stresses are not constant across the section, was introduced in (2. 24). Multiplying equation (2. 20) by [T ] and substituting in equation (2. 22), we have
x Q11 Q12 y [T ] yz = 0 0 zx 0 xy
x Q11 Q y 12 1 yz = [T ] 0 0 zx 0 xy

Q12 Q22 0 0 0

0 0 2Q44 0 0

0 0 0 2Q55 0

0 x 0 y 0 [T ] yz 0 xz 2Q66 xy
0 x 0 y 0 [T ] yz 0 xz 2Q66 xy

which is equivalent to
Q12 Q22 0 0 0 0 0 2Q44 0 0 0 0 0 2Q55 0
1

by multiplying both members by [T ] . This is equivalent to

Chapter 2 Mathematical Model

34

0 0 Q16 x x Q11 Q12 Q 0 0 Q 26 y Q 22 12 y 2 (2. 25) = 0 0 0 Q Q yz 44 45 yz , 0 Q 45 Q 55 0 2 xz zx 0 2 xy xy 0 0 Q Q Q 26 66 16 1 1 1 Considering yz = yz , xz = xz , and xy = xy , equation (2. 25) comes 2 2 2

0 0 Q16 x x Q11 Q12 Q 0 0 Q 26 y Q 22 12 y = 0 0 0 Q Q yz 44 45 yz 0 Q 45 Q 55 0 xz zx 0 xy xy 0 0 Q Q Q 26 66 16 1 where Q = [T ] [Q ] [T ] is given by


Q11 = Q11m 4 + 2(Q12 + 2Q66 )m 2 n 2 + Q22 n 4 Q12 = (Q11 + Q22 4Q66 )m 2 n 2 + Q12 (m 4 + n 4 ) Q16 = mn3Q22 + m3 nQ11 mn(m 2 n 2 )(Q12 + 2Q66 ) Q 22 = Q11n 4 + 2(Q12 + 2Q66 )m 2 n 2 + Q22 m 4 Q 26 = m3 nQ22 + mn3Q11 + mn(m 2 n 2 )(Q12 + 2Q66 ) Q 44 = Q44 m 2 + Q55 n 2 Q 45 = (Q55 Q44 )mn Q 55 = Q55 m 2 + Q44 n 2 Q 66 = Q11 + Q22 2Q12 )m 2 n 2 + Q66 (m 2 n 2 ) 2

(2. 26)

(2. 27)

Equation (2. 26) can be written as

Chapter 2 Mathematical Model

35

x Q11 Q12 y = Q12 Q 22 xy Q16 Q 26 yz Q 44 = zx Q 45

Q16 x Q 26 y Q 66 xy

(2. 28)

Q 45 yz Q 55 zx

At this point the quantities Qij and Q ij can be determined relating stresses and strains in either coordinate system. Equations (2. 28) are the constitutive equations of laminae. In the following, N is the number of laminae that composes the laminate. For the kth laminae, equation (2. 26) can be written as
x y yz = Q k zx xy
k

x y yz xz xy

(2. 29)

where all matrices must have the subscript k due to the orientation of the particular
lamina with respect to the plate x-y coordinates and its unique Q .

Consider equations (2. 1) to (2. 13), and a laminated plate, elastic, with uniform thickness h, and each layer of the plate is homogeneous and orthotropic; hk is the vectorial distance from the mid-plane, to the upper surface of the kth lamina.

2.2 - Moment-Curvature and stress relations The in-plane stress resultants {Tx , Ty , Txy } and the moments {M x , M y , M xy } ,
per unit length are defined by:

{T ,
x

Ty , Txy } =

{ ,
x h 2

h 2

y , xy }dz ,

(2. 30)

Chapter 2 Mathematical Model


h 2

36

{M

, M y , M xy } =

{ ,
x

y , xy }zdz

(2. 31)

h 2

The shear stress resultants are

{Q ,
x

Qy } =

h 2

xz

, yz }dz

(2. 32)

h 2

For a laminated plate, the stress components can be integrated across each lamina and added together as follows, employing equations (2. 29), (2. 5) and (2. 6),

{Tx , Ty , Txy } =
N k =1

hk

hk 1

{ ,
x

y , xy } dz =
h

{
N k =1

k 0 Q {0 0 k y , kxy } zdz x , y , xy }k dz + h Q {k x , k hk 1 k k k 1

hk

(2. 33)

Since the derivatives of u0 and v0 (mid-surface displacements) and the Q s are not functions of z, (2. 33) can be rewritten as:

{T ,
x
N

Ty , Txy } =
0 x 0 0 y , xy }

{ Q { ,
k =1 k

hk

hk 1

dz + Q {kx , k y , kxy }
k

hk

hk 1

zdz

(2. 34)

or, in reduced form,

[T ] = [ A][ 0 ] + [ B ][ k ]
where
Aij = (Q ij ) k [ hk hk 1 ] , [i,j = 1,2,6]
k =1 N

(2. 35)

(2. 36) (2. 37)

Bij =

1 2

(Q
k =1

ij k

2 2 ) hk hk 1 , [i,j = 1,2,6]

From equation (2. 31),

Chapter 2 Mathematical Model

37 y , xy } zdz =
h

{M x , M y , M xy } =
N k =1

hk

hk 1

{ ,
x

=
k =1
N

{ = { Q { ,
N hk
k =1 k 0 x

k 0 0 2 0 hk1 Q k { x , y , xy }k zdz + hk 1 Q k { x , y , xy }k z dz

}
(2. 38)

0 0 y, xy }

hk

hk 1

zdz + Q

{ ,
x

y , xy }

hk

hk 1

z 2 dz

or, in reduced form1,

[ M ] = [ B ][ 0 ] + [ D ][ ]
Where 3 3 Dij = 1 3 (Q ij ) k hk hk 1 ,[i,j = 1,2,6]
k =1 N

(2. 39)

(2. 40)

From (2. 22), (2. 5), and (2. 30) to (2. 32), xzk =2 Q 55k xz + 2Q 45k yz yzk = 2Q 45k xz + 2Q 44k yz hence
Qx = 2 ( A55 xz + A45 yz ) Qy = 2 ( A45 xz + A44 yz )

(2. 41) (2. 42)

Combining (2.32) and (2.36),


[T ] = [ A][ 0 ] + [ B ][ k ] , M B D k = + [ ] [ ][ ] [ ][ ] 0

(2. 43)

we have
T [ A] = M [ B ]

[ B] = E { } [ ]{ } [ D ]

(2. 44)

which is equivalent to
1

The notation here used for the reduced form is equal to the representation of the mass matrix defined in section 2.3.

Chapter 2 Mathematical Model

38
B12 B22 B26 D12 D22 D26 B16 0 x 0 B26 y B66 0 xy D16 x D26 y D66 xy

Tx A11 T y A12 Txy A16 = M x B11 M y B12 B16 M xy

A12 A22 A26 B12 B22 B26

A16 A26 A66 B16 B26 B66

B11 B12 B16 D11 D12 D16

(2. 45)

where [A ] is the extensional stiffness matrix relating in-plane stress resultants (Ns) to the mid-surface strains ( 0 s); [ D ] is the flexural stiffness matrix relating the stress couples (Ms) to the curvatures (ks), and [ B ] is the bending-stretching matrix and relates Ms to 0 s and Ns to ks. The model presented is valid for laminated plates which may be not symmetrical about their mid-surface plane. 2.3 - Equations of Motion The equations of motion of the plate are derived by equating the sum of the virtual work of the inertia forces, of the elastic restoring forces, and of external forces to zero. In-plane and transverse external forces are considered. Combining equations (2. 4), (2. 5) and (2. 6), the strains are expressed as
x 1 0 0 z 0 0 y = 0 1 0 0 z 0 {} 0 0 1 0 0 z xy

(2. 46)

Where
p p 0 L {} = b + 0 0

(2. 47)

p The linear membrane and bending strains, {0 } and {b0 } , and the geometrically
p non-linear membrane strain, { L } , are defined as

Chapter 2 Mathematical Model

39

0 y ( w )2 2 x u, x ,x 0 2 p b p x {0 } = v, y , {0 } = y , { L } = ( w, x ) 2 u + v , y ,x 0 0 w, x w, y y x + x y

(2. 48)

Or, in terms of shape functions and generalized displacements, using relation (2. 8),
{ N u }T ,x {0p } = 0 u T { N , y } N y ,x {b0 } = 0 y N , y q u T u { N, y } q v u T { N, x } 0
T

(2. 49)

{ } {

0
x ,y x ,x

{N } } {N }
T

T q y q x T

(2. 50)

1 {q }T { N w }{ N w }T {q } ,x ,x w 2 w T T 1 p w w { L } = 2 {qw } { N, y }{ N, y } {qw } T T w w {qw } { N , x }{ N , y } {qw }

(2. 51)

The transverse shear strains, { zx } and { yz } are defined as


{ N w }T 0 ,y w yz , y x = 0 = T w, x + y { N ,w zx x}

N x
T

{ }
N
y

qw q y q x

(2. 52)

The principle of virtual work states that:


Win + Wv + Wex = 0

(2. 53)

Chapter 2 Mathematical Model

40

where Win , Wv and Wex are, respectively, the work done by the inertia, internal and external forces due to virtual displacements {d } . {d } is given by
u v w {d } = = [ N ]{q} x y

(2. 54)

Making use of dAlembert principle, we obtain the following expression for the virtual work of the inertia forces:
Win = h {d }
T

}d = {q} h [ N ] [ N ]d {q } ) = {q} [ M ]{q } {d (


T T T

(2. 55) Where [ M ] is the mass matrix:

[ M ] = h [ N ] [ N ]d =
T

h { Nu}T { Nu}T d 0 0 0 0 u T u T 0 0 0 0 h { N } { N } d T T 0 0 0 0 h { Nw} { Nw} d = = h3 y T y T 0 0 0 N N d 0 12 3 h x T x T 0 0 0 0 N N d 12

{ }{ }

{ }{ }

(2. 56)
} = where is the mass density of the material that constitutes the plate, {q
d 2 {q} dt 2

and represents the area of the plate. This is equivalent to, in simplified notation,

Chapter 2 Mathematical Model


M p 0 0 M p [M ] = 0 0 0 0 0 0 0 0 0 0 0 0 0 0 M Ry 0 [ M Rx ] 0

41

[Mb ]
0 0

(2. 57)

where [Mp] and [Mb] are the in-plane and out-of-plane inertia matrices, [MRy] and [MRx] are due to the rotatory inertia. The variation Wv may be expressed as:
T T T Wv = {} d {} {Q} d M

(2. 58)

Substituting equations (2. 47) and (2. 44) in (2. 58),


p T p T p p T 0 L L d {} {Q} d Wv = 0 + [ E] b + b 0 0 0 0

(2. 59)

p T p T p p L T 0 = b [ E ] + [ E ] 0 + L d {} {Q} d b 0 0 0 0

p p T p p p T p T p p T 0 0 0 L L L L 0 = b [ E ] b + b [ E ] + [ E ] b + [ E ] d 0 0 0 0 0 0 0 0

{} {Q} d =
T

u v w = x y

[ N ]

Pu ( x, y, t ) P ( x, y , t ) v Pw ( x, y, t ) d 0 0

(2. 60)

Where Pu ( x, y, t ) , Pv ( x, y, t ) and Pw ( x, y, t ) are the distributed forces (N/m2) applied to the plate in the x, y, z directions, respectively. The linear stiffness matrix [ K1 ] , non-

Chapter 2 Mathematical Model

42

linear stiffness matrices, [ K 2 ] , [ K 3 ] and [ K 4 ] and the vector of external forces { P} are defined as follows:
p p T 0 T d {} {Q} d 0 E [ ] b b 0 0

p T [ A] = 0 b B 0 [ ]
= {q}
T

[ B ] 0p T b d {} {Q} d [ D ] 0

[ K1 ]{q} ,
[ B ] Lp d [ D ] 0

(2. 61)

p p T p T [ A] L 0 b [ E ] d = 0 b B 0 0 0 [ ]

= {q}

[ K 2 ]{q} , [ B ] 0p b [ D ] 0
T

(2. 62) d

p T p p [ A] 0 L [ E ] b d = L 0 0 [ B] 0

= {q}

[ K3 ]{q}
[ B ] Lp d [ D ] 0

(2. 63)

p p T p T [ A] L L [ E ] d = L 0 0 [ B] 0

= {q}

[ K 4 ]{q}

(2. 64)

{q} [ N ]
T

Pu ( x, y, t ) P ( x, y , t ) v T P ( x , y , t ) w d = {q} { P} , 0 0

(2. 65)

Chapter 2 Mathematical Model

43

Pu ( x, y, t ) P ( x, y , t ) v where { P} = Pw ( x, y, t ) . 0 0 Therefore, equation (2. 59) may be expressed as Wv =- {q}


T

([ K ] + [ K ] + [ K ] + [ K ]) {q}
1 2 3 4

(2. 66)

The generalized excitation forces Wex can be expressed in terms of the actual forces and the shape functions of the HFEM, by means of the virtual work executed by these forces. For example, if Pj ( t ) represents a transverse concentrated force acting at the point x= x j and Pd ( x, y, t ) represents a transverse distributed force, the virtual work of the external forces is given by:
Wex = P ( t ) ( x x j )( y y j ) + Pd ( x, y , t ) w( x, t ) d = {qw } j
T

{P ( t )}
w

(2. 67)

where ( x x j ) represents a spatial Dirac delta functions given by


( x x j )
L

x xj

( x x j )dx = 1
0

(2. 68)

So that Pj ( t ) ( x x j )( y y j ) has units of distributed force (N/m2). Substituting equations (2. 61)-(2. 65) into equation (2. 60) and allowing the virtual generalized displacements to be arbitrary, gives the time domain equations of motion in generalized coordinates:
} {q} {q} [ M ]{q
T T

([ K ] + [ K ] + [ K ] + [ K ]){q} + {q} {P ( t )} = 0
T 1 2 3 4

(2. 69)

This is equivalent to,

Chapter 2 Mathematical Model

44

{q}

} ([ K ] + [ K ] + [ K ] + [ K ]) {q} + { P ( t )}) = 0 ( [ M ] {q


1 2 3 4

} + [ K1 ]{q} + ([ K 2 ] + [ K 3 ] + [ K 4 ]) {q} = { P ( t )} [ M ] {q


} + [ Kl ]{q} + [ Knl ]{q} = { P} [ M ] {q

(2. 70)

In a more detailed form, equation (2. 70) may be written as follows: M p 0 0 M p 0 0 0 0 0 0 0 0 0 q u v 0 0 q 0 0 qw + q 0 y M Ry q 0 [ M Rx ] x 0

[Mb ]
0 0

pb11 pb12 K1p11 K1p12 0 K1 K1 qu pb 21 pb 22 K1p 21 K1p 22 q 0 K1 K1 v 11 12 13 0 K1 K1 K1 + 0 qw K1bp11 K1bp12 K1 21 K1b11 + K1 22 K1b12 + K1 23 q y q 21 22 31 b b bp bp 21 32 22 33 K1 K1 K1 K1 + K1 K1 + K1 x

11 0 0 K2 21 0 K2 0 + K 11 K 12 K 31 + K 13 + K 3 3 2 3 [ 4] 0 0 0 0 0 0

0 0 qu P { u} 0 0 qv {Pv } q , w = { P } w 0 0 q y 0 0 0 q x 0 0 0

(2. 71)

Where K1p11 K1p12 K1b11 K1b12 K1pb11 K1pb12 , , , K1p 21 K1p 22 K1b 21 K1b 22 K1pb 21 K1pb 22 K1bp11 K1bp12 , K1bp 21 K1bp 22

Chapter 2 Mathematical Model

45

K111 K1 21 31 K1

12 K1 22 K1

13 K1 23 K1 , 32 33 K1 K1

11 K2 21 11 12 13 K2 K3 and K3 K3 indicate the area 31 K2

p b pb bp which K1 , K1 , K1 , [ K 2 ] and [ K 3 ] occupy, respectively. K1 , K1 , p b K1 is the out-plane linear stiffness K1 is the in-plane linear stiffness matrix,
pb bp matrix, K1 and K1 are the in-plane/bending coupling matrices, and K1 is

the shear linear stiffness matrix; they form the linear [ K1 ] matrix. [ K 2 ] , [ K 3 ] and

[ K 4 ] represent

the nonzero part of the non-linear stiffness matrices and 0 0} is the vector of generalized external forces. These
T

{P} = {{Pu } {Pv } {Pw }

matrices are defined in the following sections, with the exception of [ K 3 ] . As is demonstrated in reference [2.1], by comparing the form of

[ K3 ]

and [ K 2 ] ,

[ K3 ] =2 [ K 2 ]

From equation (2. 71), the system can be split in two parts: u q q v M p 0 0 0 0 w + q 0 M p 0 0 0 q y x q
qu q K1p11 K1p12 0 K1pb11 K1pb12 v q w + p 21 p 22 pb 21 pb 22 K1 K1 0 K1 K1 q y q x

Chapter 2 Mathematical Model

46

qu q 11 0 0 K 2 v { P } 0 0 qw + = u 21 { Pv } 0 0 K 2 0 0 q y q x

(2. 72)

And
[ M b ] 0 0 M Ry 0 0

0 q w y + 0 q x [ M Rx ] q
11 1 12 1 13 1

qu 0 q 0 K K K v q K1bp11 K1bp12 K1 21 K1b11 K1 22 K1b12 K1 23 + + + w + q K1bp 21 K1bp 22 K1 31 K1b 21 + K1 32 K1b 22 + K1 33 y q x K K K + K + [ K4 ] + 0 0 0 0 0 0


11 3 12 3 31 2 13 3

qu 0 0 qv { Pw } 0 0 qw = 0 q 0 0 y 0 q x

(2. 73)

In not too thick plates and if the in-plane displacements are much smaller than the transverse displacements, the in-plane inertia can be neglected. Therefore, from (2.72), comes
qu q K1p11 K1p12 0 K1pb11 K1pb12 v q w + p 21 p 22 pb 21 pb 22 K1 K1 0 K1 K1 q y q x

Chapter 2 Mathematical Model

47

qu q 11 0 0 K 2 v { P } 0 0 u + qw = 21 Pv } { 0 0 K 2 0 0 q y q x q Solving (2.74) for u , comes qv


K1p11 K1p12 qu = p 21 p 22 qv K1 K1
1

(2. 74)

K1p11 K1p12 { Pu } p 21 p 22 P { } K K v 1 1

K1pb11 K1pb12 q y pb 21 pb 22 K1 K1 qx

K1p11 K1p12 p 21 p 22 K1 K1 From (2.73),

11 K2 {q w } 21 K2

(2. 75)

[ M b ] 0 0

qu qv 0 0 q 0 0 0 0 w 0 bp11 bp12 q y + 0 q 0 0 0 w + K1 K1 M Ry q bp 21 bp 22 0 q y [ M Rx ] x K1 K1 0 0 0 q x


11 1 12 1 13 1

qu 0 0 K q K K v b11 b12 21 22 23 K1 K1 K1 K1 K1 + 0 0 + + qw + 0 0 K1 31 K1b 21 + K1 32 K1b 22 + K1 33 q y q x K K K + K + [ K4 ] + 0 0 0 0 0 0


11 3 12 3 31 2 13 3

qu 0 0 qv { Pw } 0 0 qw = 0 q 0 0 0 y q x

Chapter 2 Mathematical Model

48

[ M b ] 0 0 M Ry 0 0
11 1

0 0 q w K bp11 K bp12 y + 1 1 qu + 0 q bp 21 bp 22 q K K q [ M Rx ] x 1 1 v


12 1 13 1

qu 0 0 K q K K v 21 22 23 b11 b12 K1 K1 K1 K1 K1 + 0 0 + + qw + 0 0 K1 31 K1b 21 + K1 32 K1b 22 + K1 33 q y q x qu K K 0 0 0 qv q + 0 0 0 0 0 w + 0 q 0 0 0 0 y q x


11 3 12 3

0 0 K + K + [ K 4 ] + 0 0 0 0 0 0
31 2 13 3

qu 0 0 qv { Pw } 0 0 qw = 0 q 0 0 0 y q x

[ M b ] 0 0 M Ry 0 0

0 q w y + 0 q x [ M Rx ] q

K111 K112 K113 qw K1 21 K1b11 + K1 22 K1b12 + K1 23 q y + q K1 31 K1b 21 + K1 32 K1b 22 + K1 33 x


31 13 11 K2 K K 0 0 + + [ ] { Pw } K 3 qw 3 4 0 0 0 q y = 0 - K1bp11 0 0 0 0 q x K bp 21 1 q 12 K3 u q v bp12 K1 qu bp 22 q K1 v

Chapter 2 Mathematical Model

49 (2. 76)

Replacing (2.75) in (2. 76) the reduced equations of motion are obtained. From (2. 70), the mass matrix [ M ] , the linear stiffness matrix [ Kl ] and the non-linear stiffness matrix [ Knl ] are given by:
[ M b ] 0 [M ] = 0 M Ry 0 0 0 0 , [ M Rx ]

K111 K112 K113 b11 b12 21 22 23 + + K K K K K [ Kl ] = 1 1 1 1 1 , K1 31 K1b 21 + K1 32 K1b 22 + K1 33


31 13 0 0 K 2 + K3 + [ K 4 ] 0 0 0 0 0 , [ Knl ] = 0 0 0 0 0 0 0

the latter with terms which are either linear or quadratic functions of {qw } . The vector of external forces { P} is given by

11 12 { Pw } K 3 K3 {qu } {qv } bp11 bp12 {P} = K1 {qu } K1 {qv } bp 21 bp 22 K q K q { } { } 1 u 1 v

(2. 77)

If a distributed force that impinges on the plates surface in the z direction is considered, then { Pu } = { Pv } = 0 . Therefore, from equation (2.72), the new force vector is given by

Chapter 2 Mathematical Model


K p11 { P } + K 11 K 12 1 w 3 3 K p 21 1 {P} = K1bp11 K1bp12 K1p11 + bp bp p 21 22 21 K1 K1 K1
p12 K1 p 22 K1 1

50
11 K1pb11 K1pb12 q K 2 y {q } + K pb 21 K pb 22 q K 21 w 1 1 x 2 1 p12 pb11 pb12 11 K K K K q 1 1 1 y 2 {qw } p 22 pb 21 pb 22 q 21 K K K K 1 1 1 x 2

(2. 78) 2.4 - Derivation of the Matrices Used in the Equations of Motion 2.4.1 - Linear stiffness matrix [ K1 ] From equation (2. 61),
p p T p T [ A] 0 0 [ E ] b d = 0 b b 0 0 0 [ B ]

[ B ] 0p b d = [ D ] 0
T T

p = { 0 } [ A]{0p } + {b0 } [ B ]{0p } + {0p } [ B ]{b0 } + {b0 } [ D ]{b0 } d T T

(2. 79) The linear [K 1 ] matrix consists of four independent parts. The in plane stiffness
p b matrix, K1 , the bending stiffness matrix, K1 , and the coupling in-plane/bending pb bp matrices K1 , K1 which are obtained from (2. 79).

Therefore,

Nu T { , x } p T p {0 } [ A ]{0 }d = 0 u T { N , y }

{ N u }T ,x q T u { N,uy } q [A] 0 v u T u T { N, x } {N } ,y
0

q u T u { N, y } q d = v u T { N, x }
0

Chapter 2 Mathematical Model

51

{ N u }T ,x T q = u 0 qv u T {N } ,y
q = u qv
T

0 u T { N, y } T { N,ux }
0

{ N,ux } 0

{N }
u ,y

{ N u }T 0 A11 A12 A16 , x qu A A A 0 u T N d { } = y 12 22 26 , qv A16 A 26 A 66 { N u }T { N u }T ,x ,y T { N u } 0 ,x A A A u 11 12 16 { N , y } q u T A12 A 22 A 26 0 N, y } d u = { qv { N,ux } T T A A A u u 26 66 16 { N } { N, x } ,y


T

q = u qv

T u u T u u T A Nu Nu T + A + A 66 { N ,uy }{ N ,uy } 11 { , x }{ , x } 16 { N , y }{ N , x } + { N , x }{ N , y } u u T u u T u u T u u T A12 { N , y }{ N , x } + A16 { N , x }{ N , x } + A 26 { N , y }{ N , y } + A 66 { N , x }{ N , y }

T T T T A12 { N ,ux }{ N ,uy } + A 26 { N ,uy }{ N ,uy } + A16 { N ,ux }{ N ,ux } + A 66 { N ,uy }{ N ,ux } qu d = T T T T qv A 22 { N ,uy }{ N ,uy } + A 26 { N ,ux }{ N ,uy } + { N ,uy }{ N ,ux } + A 66 { N ,ux }{ N ,ux }

p = {q} K1 {q} T

(2. 80)
b K1

The bending stiffness matrix

is derived considering the virtual

N y ,x T work {b D ]{b d . Replacing {b 0 [ } } 0 } by 0 0 y N , y

{ } {

0
x ,y

{N } } {N }
T

x ,x

T q y in the q x T

integral, comes

Chapter 2 Mathematical Model


N y ,x D16 D26 0 D66 y N, y
y ,x T y ,x

52

qy qx

N 0
T

{ }
y ,x

0
x ,y

{ }
y ,y x ,x
T

{N } {N }
y ,x

D11 D12 D D 12 22 D16 D26


y ,y

{ }

0
x ,y

{N } { } {N }
T x ,x
T y ,y x ,y y ,y

q y = q x D12

{ }{N } + D ({N }{N } + { N }{ N } ) + D {N }{N } { } { N } D { N }{ N } D { N }{ N } D { N } { N } { N }{ N } D { N }{ N } D { N }{ N } D { N }{ N } q


D N y 11 , x x D12 N , y
x T ,y x ,y 16 66 y ,y y ,y T y ,x T 16 x ,x y ,x T T 26 66 x ,x y ,y y ,x 26

T qy d = qx T

x D22 N , y

}{ N }
T

+ D26

({N

y ,y

x T ,y x ,y

16

y ,x

x T ,x T

x ,x

}{ N } + { N }{ N }
T x ,y x ,x

)+ D

66

y ,y

x T ,x T

66

{N }{ N }
x ,x x ,x

d q = x

q q y b y = K 1 q q x x

(2. 81)
pb Matrix K1

is obtained from the integral { 0p } [ B ]{b Replacing 0 }d .


T

{ }

p T 0

and {b 0 } in the integral, comes


q u T u { N, y } q v u T { N, x } 0
T

Nu T { , x } 0 u T { N , y }

B11 B12 B B 22 12 B16 B26

N y ,x B16 B26 0 B66 y N , y

{ } {

0
x ,y

{N } } {N }
T x ,x
T

T q y d = q x T
q y T d = q x T
y ,y

q = u qv

{ N ,ux } 0

B { N } B { N } { N } B 0
u ,y u ,x u ,y

11

12 16

B12 B22 B26


T

Nu B + Nu B N, x y { { , x } 11 , y } 16 q = u qv { N u } B + { N u } B N, x y ,y 12 ,x 16 T

) ( { } ) + ({ N } B + { N } B ) ( { N
T u ,y 26 u ,x 66

) ( { } ) + ({ N } B + { N } B ) ( { N }
u ,x 16 u ,y 66 y ,y

N y ,x B16 B26 0 B66 N y ,y

{ } {

0
x ,y

{N } } {N }
T x ,x

) })
T T

Chapter 2 Mathematical Model

53
T u ,x u ,y x T ,x x ,x

( { N } B + { N } B ) { N } + ({ N } B + { N } B ) { N } q d q ( { N } B + { N } B ) { N } + ({ N } B + { N } B ) { N }
u ,x 12 u ,y 26 x ,y x ,y 16 66 u ,y 22 u ,x T 26 u ,y 26 u ,x T 66

= x

q = u qv

y T T T B N u N y T B u + { N ,ux } N , y y B66 { N ,uy } N , y y 11 { , x } ,x 16 { N , y } N , x T T T T y B12 { N ,uy } N , B16 { N ,ux } N , x y B26 { N ,uy } N , y y B66 { N ,ux } N , y y x T T T T x x x x + B26 { N ,uy } N , + B16 { N ,ux } N , + B66 { N ,uy } N , B12 { N ,ux } N , y y x x q y d = T T T T q x + B66 { N ,ux } N , x x B22 { N ,uy } N , y x + B26 { N ,ux } N , y x + { N ,uy } N , x x T

{ } { }
{

{ } { }
{ {

{ }) { }

{ } { }

{ } { }

{ } { }

T qu q y pb K = 1 qv q x

(2. 82)
T

bp Matrix K1

p is obtained from the integral {b Replacing 0 } [ B ]{ 0 }d .

{ }

b T 0

and { 0p } in the integral, comes

y N, x 0 y N, y

{ } {

0
x ,y

{N } } {N }
T x ,x

q y T q x T

B11 B12 B B 22 12 B B 26 16

{ N u }T B16 , x 0 B26 B66 { N u }T ,y

T qu { N,uy } d = q v u T { N, x } 0
q u T { N, y } d qu = v T { N,ux } 0

q y = q x

N 0

{ }
y ,x

0
x ,y

{ }
y ,y x ,x

{N } {N }

B11 B12 B B 22 12 B B 26 16

qy = qx

N y B N y B { N u }T + N y B N y B { N u }T ,x 11 ,y 16 ,x ,x 16 ,y 66 ,y x T T x x x B16 { N,ux } + N, B26 + N, B66 { N,uy } N, y B12 + N, x y x

({ }
({ }

{ } ) { } )

({ }

{ N u }T B16 , x 0 B26 B66 { N u }T ,y

({ }

{ } ) { } )

Chapter 2 Mathematical Model

54
u T ,y u ,y T y ,x y ,y u T ,x T

( {N
({ N

y ,x x ,y

} B {N }
12 y ,y 22 x ,x

B26 B26

} B + {N }
T

qy = qx

B12 N,x y

{ }{ N }

B N y N u T B N y N u T + N y N u T B N y N u T 16 , y { ,x} ,x { , y} 66 ,y { ,y} 11 ,x { , x } x x x x u T u T u T u T + + + N N N N N N N N B B B B { } { } { } { y x x x y y x y} 12 , , 16 , , 26 , , 66 , ,

{ }

} B {N } B ){N } q d = ){N } + ({N } B + {N } B ){N } q


16 66 u v x ,y 26 x ,x 66 u ,x

){N } + ( {N
{ }

({ }

{ }

) { }
u T ,x u T ,x

{ }

{ }
66

{ }

u T ,y u T ,y

B26 N, y y

{ }{ N }

x B22 N, y

{ }{ N }

x + B26 N, x

({

} { N } + { N }{ N } ) + B { N } { N }
u T ,y x ,y u T ,x x ,x

u T ,y

B16 N, x y

{ }{ N }

u T ,x

B66 N, y y

{ }{ N }

qu d q = v
(2. 83)

q y bp = K1 q x

qu qv
T

bp pb One verifies, as expected, that K1 = K1 .

2.4.2 Non-linear stiffness matrix [ K 2 ] and [ K 3 ] In this section the non-linear stiffness matrix [ K 2 ] is derived. [ K 3 ] is proved to be twice the transpose of [ K 2 ] , i.e., [ K 3 ] = 2 [ K 2 ] [2.1]. Both [ K 2 ] and [ K 3 ] are linear
T

functions of the generalised displacements. Considering equation (2. 62),


p p T L T 0 E d = {q} [ K 2 ]{q} . Therefore, b [ ] 0 0 p p T p T [ A ] L 0 b [ E ] d = 0 0 b 0 0 [ B ]
p = {0 } [ A ]{ Lp } + {b0 } [ B]{ Lp } d T T

[ B ] Lp d = [ D ] 0
(2. 84)

p From the definition of { 0 } , {b0 } and {Lp } , equations (2. 49), (2. 50) and (2. 51),

results that

Chapter 2 Mathematical Model

55
T

u T { N , x } 0 u T { N , y } N y ,x 0 + N y ,y

q u T u { N, y } q v T { N,ux }
0

A11 A12 A A 12 22 A16 A26


T

1 {q }T { N w }{ N w }T {q } w ,x ,x A16 2 w T T 1 w w A26 2 {qw } { N , y }{ N , y } {qw } + A66 {q }T { N w }{ N w }T {q } w ,x ,y w 1 {q }T { N w }{ N w }T {q } w ,x ,x B16 2 w T 1 w w T B26 2 {qw } { N , y }{ N , y } {qw } d B66 {q }T { N w }{ N w }T {q } w ,x ,y w

{ } {

{ } {
T

0 q T y x N , y q x x T N, x

} }

B11 B12 B B 12 22 B16 B26

T qu q y bpl K 2pl { qw } + K2 {qw } = q q v x T

(2. 85)

Evaluating each term separately,

qu pl K2 {q w } = q v
T

({q } {N } A
T u u ,x T u ,y 1 16 2 T u u ,x 1 12 2 T u ,y u ,x 1 26 2

{N } A + {q } { N } A + {q } { N } A + {q } { N } A + {q }{ N } A
+ {qu }
u T u

16

u ,y

66

{ N }{ N } {q } + {q } { N } A {q } { N }{ N } {q } + {q } { N }{ N } {q } + {q }{ N } A {q } { N }{ N } {q } + {q } { N }{ N } {q } + {q } { N } A {q } { N }{ N } {q } + {q } { N }{ N } {q } + {q } { N } A {q } { N }{ N } {q } + {q } { N }{ N } {q } + {q } { N } A {q } { N }{ N } {q } + {q } { N }{ N } {q } + {q }{ N } A {q } { N }{ N } {q }) d =
1 11 2

{q w }
T T

w ,x

w T ,x

u ,y

1 12 2

w ,x

w T ,x

w ,x

w T ,x

u T ,x u ,y

1 16 2

w ,x

w T ,x

w ,y

w T ,y

1 22 2

w ,y

w T ,y

w ,y

w T ,y

u ,x

1 26 2

w ,y

w T ,y

w ,x

w T ,y

u ,y

26

w ,x

w T ,y

w ,x

w T ,y

u ,x

66

w ,x

w T ,y

Separating the terms with {qu } and {qv } ,


T T

= {qu }

({ N } A {q } { N }{ N } + { N } A {q } { N }{ N }
1 2 u ,x T 11 w w ,x w T ,x u ,y T 16 w w ,x T w ,y w T ,y u ,y T 26 w w ,y w T ,y T 16 w w ,x w T ,y u ,y T 66 w w ,x w T ,y

w T ,x

{ N }{ N } + { N } A {q } { N }{ N } + + 2 { N } A {q } { N }{ N } + 2 { N } A {q } { N }{ N }
u ,x

+ { N ,ux } A12 {qw }

) d {q } +
w

Chapter 2 Mathematical Model

56
T w ,x w T ,x u ,x T w ,x w T ,x

+ {qv }

1 2

({ N
T w

u ,y

} A {q } { N }{ N } + { N } A {q } { N }{ N }
12 w 16 w w ,y w T ,y u ,x T 26 w w ,y w T ,y w ,x w T ,y u ,x T 66 w ,x w T ,y

+ { N ,uy } A22 {q w }
u ,y 26

{ N }{ N } + { N } A {q } { N }{ N } + + 2 { N } A {q } { N }{ N } + 2 { N } A {q } { N }{ N } ) d {q }
T w w
T T = {qu } U {qw} + {qv } V {qw}

(2. 86)
w T ,x w T ,x

Where
U = 1 2

({N } A {q } {N }{N } + {N } A {q } {N }{N } +


u ,x T 11 w w ,x u ,y T 16 w w ,x T w ,y w T ,y u ,y T 26 w w ,y w T ,y T w w ,x w T ,y u ,y T 66 w w ,x w T ,y u ,x u ,y T w ,x w T ,x

+ { N ,ux } A12 {qw }


u ,x 16

{ N }{ N } + { N } A {q } { N }{ N } + +2 { N } A {q } { N }{ N } +2 { N } A {q } { N }{ N } ) d =
=
1 2

(({N } A + {N } A ){q } {N }{N } +


11 16 w T w w ,y w T ,y u ,x 16 u ,y T 66 w w ,x w T ,y

){q } {N }{N } + + 2 ({ N } A + { N } A ) {q } { N }{ N } ) d
+ { N ,ux } A12 + { N ,uy } A26

(2. 87) V = 1 2

({N } A {q } {N }{N } + {N } A {q } {N }{N } +


u ,y T 12 w w ,x w T ,x u ,x T 16 w w ,x w T ,x T w ,y w T ,y u ,x T 26 w w ,y w T ,y T 26 w w ,x w T ,y u ,x T 66 w w ,x w T ,y

+ { N ,uy } A22 {qw }


u ,y

{ N }{ N } + { N } A {q } { N }{ N } + +2 { N } A {q } { N }{ N } + 2 { N } A {q } { N }{ N } ) d =
=
1 2 u ,y

( ({ N } A
u ,y 22 u ,y 26

12 u ,x

+ { N,ux } A16 {qw }


T 26 w w ,y

{ N }{ N }
w ,x w T ,y w ,x w T ,y

w T ,x

({N } A + { N } A ){q } {N }{N } + +2 ({ N } A + { N } A ) {q } { N }{ N } ) d


u ,x T 66 w

(2. 88)
q y bpl K2 The term {qw } , comes q x
T

Chapter 2 Mathematical Model

57

q y bpl K2 { qw } = q x T T w T = q y N , x y B11 1 qw } { N ,w { x }{ N , x } {qw } + 2


T

{ }{ }
T x ,y 1 12 2 T y ,y 1 16 2

+ q x q y + q x q y + q x q y + q x q y + q x q y + q x

{ { { { { {

{ { { { {

} { N } B {q } {N }{ N } {q } } {N } B {q } {N }{N } {q } + } { N } B {q } {N }{ N } {q } } {N } B {q } {N }{ N } {q } + } { N } B {q } { N }{ N } {q } } {N } B {q } {N }{N } {q } + } { N } B {q } { N }{ N } {q } } {N } B {q } {N }{N } {q } + } { N } B {q } {N }{ N } {q } } {N } B {q } {N }{N } {q } + } { N } B {q } {N }{ N } {q }) d


T w w ,x w T ,x w T w w ,x w T ,x w T x ,x 1 16 2 T w w ,x w T ,x w T y ,x 1 12 2 T w w ,y w T ,y w T x ,y 1 22 2 T w w ,y w T ,y w T y ,y 1 26 2 T w w ,y w T ,y w T x ,x 1 26 2 T w w ,y w T ,y w T y ,x T 16 w w ,x w T ,y w T x ,y T 26 w w ,x w T ,y w T y ,y T 66 w w ,x w T ,y w T x ,x T 66 w w ,x w T ,y w

Separating the terms with q y


= q y
y ,x T 1 2 y ,x T 11 w

{ } and {q } ,
T T x
w ,x w T ,x y ,y T 16 w w ,x w T ,x y ,y T 26 w w ,y w T ,y

{ } ({ N } B {q } { N }{ N } + { N } B {q } { N }{ N } + + { N } B {q } { N }{ N } + { N } B {q } { N }{ N } + +2 { N } B {q } { N }{ N } + 2 { N } B {q } { N }{ N } ) d {q } =
+ q x
x ,y T 1 2 x ,y T 12 w w ,x w T ,x x ,x T 16 w w ,x w T ,x T 22 w w ,y w T ,y x ,x T 26 w w ,y w T ,y x ,y T 26 w w ,x w T ,y x ,x T 66 w w ,x w T ,y w

{ } ( {N } B {q } {N }{N } {N } B {q } {N }{N } { N } B {q } { N }{ N } { N } B {q } { N }{ N } 2 { N } B {q } { N }{ N } 2 { N } B {q } { N }{ N } ) d {q } +
T 12 w w ,y w T ,y y ,x T 16 w w ,x w T ,y y ,y 1 66 2 T w w ,x w T ,y w

Chapter 2 Mathematical Model

58

= q y

{ }

Y {qw } + q x

Z {qw }

(2. 89)

where
T T y y w w Y = 1 N B q N N N B q { } { } { }{ } { N,wx }{ N,wx } x w x x y w , 11 , , , 16 2 T T

({ }
T T 16 w y ,x

{ }
T w w

{ N }{ N } { N } B {q } { N }{ N } { } 2 { N } B {q } { N }{ N } 2 { N } B {q } { N }{ N } ) d = = ( ( { N } B { N } B ) {q } { N }{ N } + + ( { N } B { N } B ) {q } { N }{ N } + 2 ( { N } B { N } B ) {q } { N }{ N } ) d
N , xy B12 {qw }
y ,x w ,y w T ,y y ,y 26 w ,y w T ,y w ,x w T ,y y ,y T 66 w ,x w T ,y 1 2 11 y ,y T 16 w w ,x w T ,x y ,x 12 y ,y T 26 w w ,y w T ,y y ,x 16 y ,y T 66 w w ,x w T ,y

(2. 90)

} B {q } { N }{ N } + { N } B {q } { N }{N } + + { N } B {q } { N }{ N } + { N } B {q } { N }{ N } + +2 { N } B {q } { N }{ N } + 2 { N } B {q } { N }{ N } ) d =
Z = 1 2
x ,y x ,y

({ N
26

12

w ,x

w T ,x

x ,x

16

w ,x

w T ,x

22

w ,y

w T ,y

x ,x

26

w ,y

w T ,y

x ,y

w ,x

w T ,y

x ,x

66

w ,x

w T ,y

} B + { N } B ){q } { N }{ N } + + ({ N } B + { N } B ) {q } { N }{ N } + + 2 ({ N } B + { N } B ) {q } { N }{ N } ) d
=
1 2 x ,y 12 x ,x T 16 w w ,x w T ,x x ,y 22 x ,x T 26 w w ,y w T ,y x ,y 26 x ,x T 66 w w ,x w T ,y

( ({ N

(2. 91) Summing the two integrals obtained, from equation (2. 84) results:

{q} [ K 2 ]{q} =
T T T = {qu } U {qw } + {qv } V {qw } + q y

{ }

Y {qw } + q x

Z {qw } =

Chapter 2 Mathematical Model

59

U T qu V qv {qw } = q [ K 2 ]{qw } Y y q x Z The non-zero part of the non-linear stiffness matrix [K 2 ] is: qu q v = q y q x
T

U V [ K 2 ] = Y Z

(2. 92)

2.4.3 Non-linear stiffness matrix [ K 4 ] In this section the non-linear stiffness matrix, [ K 4 ] is derived. The [ K 4 ] matrix is a quadratic function of the transverse displacement w and depends only upon the outof-plane shape functions and corresponding generalised displacements {qw } . Within the three non-linear stiffness matrices, [ K 4 ] is the main source of geometric nonlinearity. From equation (2. 64),
p p T L T L [ E ] d = {q} [ K 4 ]{q} = 0 0

p T [ A] = L 0 [ B]

[ B ] Lp p T p d = { L } [ A]{ L } d D [ ] 0

(2. 93)

Chapter 2 Mathematical Model


p From the definition of { L } in equation (2. 51), results that

60

({ } [ A]{ })d =

p T L p L
T T w w T q N N { } { }{ } ,x ,x w T w w T {qw } { N , y }{ N , y } T w T 2 {qw } { N ,w x }{ N , y }

1 2

{qw }

A11 A12 A A 12 22 A16 A26

1 {q }T { N w }{ N w }T {q } ,x ,x w A16 2 w T T 1 w w A26 q N N q { } { } { }{ } , , w y y w 2 d A66 {q }T { N w }{ N w }T {q } ,x ,y w w


T

1 2

{qw }

{q }T N w N w T A + {q }T N w N w T A + 2 {q }T N w N w T A { , y }{ , y } 12 { , x }{ , y } 16 w w w { , x }{ , x } 11 T T T w w T w w T w w T {qw } { N, x }{ N, x } A12 + {qw } { N, y }{ N, y } A22 + 2 {qw } { N, x }{ N, y } A26 T T T w w T w w T w w T {qw } { N , x }{ N , x } A16 + {qw } { N , y }{ N, y } A26 + 2 {qw } { N , x }{ N , y } A66

1 {q }T { N w }{ N w }T {q } ,x ,x w 2 w T 1 w w T 2 {qw } { N , y }{ N , y } {qw } d T w T qw } { N ,w { x }{ N , y } {qw }


=
1 2

{qw } {qw }
T T w ,x

1 2 {q w } +2 {q w }
T w

{ N }{ N } {q } + ({q } { N }{ N }
w T ,x T w w w ,x w ,x w T ,y

((

{ N }{ N }
w ,x

w T ,x

A11 + {qw }

{ N }{ N }
w ,y

w T ,y

A12 + 2 {qw }
T w ,y

{ N }{ N }
w ,x

w T ,y

A16

w T ,x

A12 + {qw }
w

{ N }{ N }
w ,x w T ,y

w T ,y

A22 +

{ N }{ N }
w ,x w T ,x w T ,y w ,x

A26 1 2 {q w }
T

({q } {N }{N }
{ qw }
T

{ N }{ N } {q } +
w ,y w T ,y w T ,y

A16 + {qw }
w

{ N }{ N }
w ,y

A26 + 2 {qw }

{ N }{ N }

A66

{ N } { N } {q } ) d

Chapter 2 Mathematical Model


=
1 2

61
w T ,x

{qw } {qw }
T T w ,x

((

{ N }{ N }
w ,x

A11 + {qw }
w ,x T w T ,x

{ N }{ N }
w ,y T

w T ,y

A12 + 2 {qw }
w ,y w T ,y

{ N }{ N }
w ,x

w T ,y

A16

1 2 {q w } +2 {q w }
T w

{ N }{ N } { N }{ N }
w ,x w ,x w T ,x w T ,y w ,x

w T ,x

+ {q w }

{ N }{ N }
w ,y w ,y

A12 + {qw }
w T ,y

{ N }{ N }
T w ,x

A22 +

w T ,y

A26 1 2 {q w }
T

({q } {N }{N }
{ qw }
T

{ N }{ N }
w T ,y

A16 + {qw }

{ N }{ N }

A26 + 2 {qw }

{ N }{ N }

w T ,y

A66

{ N }{ N }

) d {q }
w

= {q} [ K 4 ]{q}
T

(2. 94)

2.4.4 - Shear Linear Stiffness Matrix K1

Qx Q From equation (2. 59), and considering = 44 Qy Q 45

Q 45 yz , Q 55 zx

zx Q 44 Q 45 yz Q d = { } { } d yz zx Q 45 Q 55 T qw w T N x 0 yz { N , y } q = From , T y T y zx { N ,w 0 N x} q x becomes

{ }

the

integral

Chapter 2 Mathematical Model


w T { N , y } { N w }T ,x q w = qy qx
T

62
qw q y q x
T

N x
T

{N }
y

{ N w }T ,y w T N { , x }

N x
T

{N }
y

qw q d = y q x

{ N w} Q { N w}T + { N w} Q { N w}T + { N w} Q { N w}T + { N w} Q { N w}T ,x ,y ,y ,x ,x ,x 45 45 55 , y 44 , y T T y N y Q45 { N,w Q55 { N,w y} + N x} T T x Q55 { N,w N x Q45 { N,w y} N x} T T T T {N,wy } Q45 N y + {N,wx } Q55 N y {N,wy } Q44 N x {N,wx } Q45 N x q w T y y y x T N Q55 N N Q45 N d qy T y T x x x qx N Q45 N N Q55 N

{ }

{ }

{ }

{ }

{ } { } { } { } { } { }
T

{ }

{ }

{ } { }

{ } { }

= {q} K1 {q}

(2. 95)

3. DISPLACEMENT SHAPE FUNCTIONS

As referred in section 2.1, a matrix of shape functions is considered, and four sets of shape functions are required: one set of in-plane shape functions, one set of out-ofplane shape functions, one set of rotation about x shape functions and one set of rotation about y shape functions (2. 10)-(2. 13). The HFEM relies on the utilization of high order shape functions and ill-conditioning is common in high order polynomials. For this reason, trigonometric shape functions are suggested in [2.7]. Nevertheless, with the Rodrigues form of Legendre polynomials no ill-condition problems were met and it was decided to continue using them. Thus, the in-plane shape functions are given by
INT ( r 2 )

gr 2 =

n=0

( 1) ( 2r 2n 5 )!! r 2 n1 , r > 2
n

2n n !(r 2n 1)!

(2. 96)

Chapter 2 Mathematical Model


And the out-of-plane shape functions are given by
INT ( r 2 )

63

fr =

n=0

( 1) ( 2r 2n 7 )!! r 2 n1 , r > 4
n

2n n !(r 2n 1)!

(2. 97)

where r!! = r(r-2)(2 or 1), 0!! = (-1)!!= 1 and INT ( r 2 ) denotes the integer part of r 2 . The rotation shape functions about x and y are equal to the in-plane shape functions. In APPENDIX A, the first out-of-plane and in-plane shape functions are plotted; the shape functions f r , 1 r 4 are cubic polynomials not represented. It can be seen from the plots that:

(i) the in-plane shape functions have zero displacements at = 1 and = 1 ; the Legendre

( r > 4 ) out-of-plane

shape functions have both zero displacements and

slope at these points. These shape functions satisfy fully clamped boundary conditions; (ii) the odd number in-plane and Legendre out-of-plane shape functions are symmetric, while the even number are anti-symmetric. The adequate shape functions regarding the symmetries of the problem under study can be chosen, thus reducing the number of degrees of freedom.

4. NEWMARK METHOD

Chapter 2 Mathematical Model

64

In the previous sections, a finite element model was derived and the undamped equations of motion were obtained. Introducing damping matrix [C ] , these equations may be written as:
} + [C ]{q } + [ K ]{q} = { P} [ M ] {q

(2. 98)

Where

[M ]

and

[ K ] are

the mass and stiffness matrices;

{P} is

the vector of

externally applied loads. The equations of motion are integrated in time domain, directly, that is without transformation of co-ordinates [2.6]. In essence, direct numerical integration is based in two ideas. First, instead of trying to satisfy (2. 98) at any time t, it is aimed to satisfy (2. 98) only at discrete time intervals t apart. The second idea in which a direct integration method is based is that a variation of displacements and accelerations within each time interval t is assumed. In the following, displacement and acceleration vectors at time 0 are denoted by
} {q
0

} , respectively, are known, and let the solution to the differential and {q
0

equation (2. 98) be required from time 0 to time t f , in intervals of t . To determine the solution of the displacements and accelerations at time t + t , the equations (2. 98) at time t + t are also considered:
} [ M ] {q
t +t

} + [ C ] {q

t +t

+ [ K ]{q}

t +t

= { P}

t +t

(2.99)

It is assumed that the accelerations within each time interval, t are given by

} = {q } {q

1 t

} ( {q

t +t

} {q
t

, 0 t

(2.100)

Integrating (2.100) gives

} = {q } + {q } {q
t

1 2t

} ({ q

t +t

} 2 {q
t

(2.101)

} = {q } when = 0 . Integrating again gives since {q


t

Chapter 2 Mathematical Model


} {q} = {q} + {q
t t

65

1 1 t } 2 + {q 2 6t

} ( {q

t +t

} 3 {q
t

(2.102)

since {q} = {q} when = 0 .


t

Evaluating (2.101) and (2.102) at = t gives


} {q
t +t

} + = {q
t

t 2

} + {q } ) ( {q
t t +t

(2.103)

and

{q}

t +t

} = {q} + {q
t

( t ) t +
6

} + {q } ) ( 2 {q
t t +t

(2.104)

In the Newmark method, equations (2.103) and (2.104) are assumed to take the form [2.6]
} {q
t +t

} + t (1 ){q } + {q } = {q


t t

t +t

(2.105)

and

{q}

t +t

t t 2 1 t +t t } t + ( t ) {q } + {q } = {q} + {q 2

(2.106)

Taking =

1 1 and = corresponds to assuming that the acceleration is constant and 2 4


} + {q } ) ( {q
t t +t

equal to the average value

2 within the interval ( t , t + t ) .

The response at time t + t is obtained by evaluating the equation of motion at time


t + t , that is

} [ M ] {q

t +t

} + [ C ] {q

t +t

+ [ K ]{q}

t +t

= { P}
t +t

t +t

(2.107)
t +t

In order to obtain an equation for {q} gives

} , equation (2.106) is solved for {q

which

Chapter 2 Mathematical Model


} {q
t +t

66

1
( t )
2
t +t

({q}

t +t

{q}
t

1 t } {q t

1 t } 1 {q 2

(2.108)

} Replacing {q } {q
t +t

, given in (2.108) in (2.105) gives t t t } + t 1 {q } {q} + 1 {q 2

({q}

t +t

(2.109)

Substituting (2.108) and (2.109) into (2.107),

[M ]

1
2

( t )

({q}

t +t

{q}
t

1 t } {q t

1 t } + 1 {q 2

+ [C ] t + [ K ]{q} +
1

({q}

t +t

t t t } + t 1 {q } + {q} + 1 {q 2
t +t

t +t

= { P}

{q}
t

( t )

[ M ] ({q}

t +t

1 t } [ M ] {q t

t 1 } + 1 [ M ] {q 2

t +t t t t } + t 1 [ C ] { q } + 1 [ C ] {q [C ] {q} {q} + t 2


t +t

(2.110)

+ [ K ]{q}

= { P}
t +t

t +t

Solving for {q}

1 t +t M C K + + {q} = [ ] [ ] [ ] 2 ( t )2 ( t ) 1 1 t +t t t t 1 } + 1 [ M ]{q } + M ]{q} + {P} + [ M ]{q 2 [ t 2 ( t ) + t t t } t 1 [C ]{q } [C ]{q} 1 [ C ] {q t 2


t t t

} , {q } and {q} , in the second member, Placing together terms in {q

Chapter 2 Mathematical Model


1 t +t M ]+ C ] + [ K ] {q} = [ 2 [ t ( t ) 1 t t +t = { P} + M ]+ C ] + [ K ] {q} + [ 2 [ t ( t ) 1 t 1 } t [ M ] 1 [C ] {q} + 2 1 [ M ] t 1 2 [C ] {q

67

(2.111)

If {q} ,
t

} {q

and

} are {q
t

known, then

{q}

t +t

can be calculated using (2.111).


t +t

} Equations (2.108) and (2.109) can be used to determine {q

} and {q

t +t

. The time

history of the response is obtained from time 0 to time t f , in intervals of t . Therefore, from equations (2.108), (2.109) and (2.111) the Newmark Method is given by

( a [ M ] + a [C ] + [ K ]) {q} = {P} + ( a [ M ] + a [C ]){q} + } + + ( a [ M ] a [C ]) {q } + ( a [ M ] a [ C ] ) {q


t +t t +t t 1 2 1 2 t t 3 4 5 6

} {q } {q

t +t

t +t

( = a ({q}
= a1 {q}
2

t +t

} a5 {q } {q} a3 {q
t t t t 4 6

t +t

) } + a {q } {q} ) + a {q
t

(2.112)

where

a1 = a3 =

( t )
1 , t

a2 =

, t
(2.113)

1 a5 = 1 , 2

a4 = 1 , a6 = t 1 2

The set of linear equations given in (2.112) is solved in order to determine {q}

t +t

In reference [2.6] the stability of the method is investigated and it is stated that the method is unconditionally stable if

Chapter 2 Mathematical Model

68
2

0,50 and 0, 25 ( 0,5 + )


Unless is taken to be
1 2

(2.114)

, the method introduces artificial damping, which can be

negative (when < 1 2 ). In the Newmark integration scheme, as an unconditionally stable method (choosing parameters verifying (2.114), the time step t is based upon the period corresponding to the highest frequency likely to contribute to the response,

n . According to reference [2.6], good accuracy is obtained with a time step given
by n t = 50 . The Newmark method can be extended to non-linear dynamic analysis. This requires that iteration must be performed at each time step in order to satisfy equilibrium. Also, the non-linear stiffness matrix must be formed and triangularized at each iteration or at selective points in time.

5. CLOSING COMMENTS

The mathematical model of the p-version HEFM for asymmetrically laminated rectangular plates is given in this chapter. The equations of motion are obtained and the matrices in the equations of motion are derived for the asymmetrically laminated rectangular plates, with a detailed description of the matrices involved. All these matrices are finally expressed as the integration of shape functions and their derivatives. In the implementation of the model, the differentiation and the integration of the polynomials will be calculated symbolically to find out the exact values via symbolic computation method using MAPLE. The model is derived in time domain by applying the finite element method, the principle of virtual work and the dAlemberts principle. Because the problems to be analysed do not involve singularities and the geometry of the plate is very regular, only one element was used. The model developed in this chapter is applied to the study of the geometrical non-linear vibrations of plates in following chapters.

Chapter 2 Mathematical Model

69