Basic Theory of Plate Bending

(Prepared by Dr. K.K. Wijesundara)

Thin plate with small deflection in which the membrane stresses are very small compared to
stresses due to bending deformation under a transverse loading. This class may be taken to
comprise plates for which the ratio of span to thickness (h) does not exceed 10 and maximum
deflection w is less than h/10 to h/5.

1. Thin Plate Theory (Kirchhoff and Love Theory)

Classical thin plate theory is based upon the assumptions initiated for beams by Eular and
Bernoulli but first applied to plates and shells by Love and Kirchhoff. The assumptions are as
follows:
1. A plane section of the plate normal to the mid surface remains in plane and normal to
the deformed plate mid-surface on the application of transverse loads as shown in
Figure 1 and hence plate undergoes at most a vertical translation and rotations about
the in-plane axes. (γ!"   =  𝛾!"   = 0  )
2. The stresses normal to the plate can be neglected. (σzz=0)

Figure 1: Thin plate bending in XZ plane

Considering the geometric relationship referring the bending in XZ plane;
∂w
u = u0 − z
∂x
∂w
Assuming that u0=0 u = −z
∂x
Similarly, referring the bending in YZ plane;
∂w
v = −z
∂y

Prepared by K.K. Wijesundara 1

2. Compatibility Relationships
Considering the compatibility relationships for small displacements:

∂u ∂2 w
ε xx = = −z 2
∂x ∂x
∂u ∂2 w
ε yy = = −z 2 Eq (1)
∂y ∂y
∂u ∂v ∂2 w
γ xy = + = −2
∂y ∂x ∂x∂y
and 𝜀!! = γ!"   =  𝛾!"   = 0  . Then, the constitutive relationship is reduced to a plane stress
problem.

3. Constitutive Relationship
Stress-strain relationship for plate bending can be found by using constitutive relationship for
the plane stress problem as:
) ,
! % + 1 υ .! ε %
## σ xx ## 0 .## xx ##
E +
" σ yy &= 2+
υ 1 0 ." ε yy &
# # 1− ν + 1− υ .# #
#$ τ xy #' + 0 0 .#$ γ xy #'
* 2 -
Then combining with Eq (1), the resultant stresses can be expressed in the function of
transverse displacement w as:
! 2 %
∂ w
) ,# −z 2 #
! % + 1 υ .# ∂x #
## σ xx ## 0 .## #
E + ∂2 w #
" σ yy &= 2+
υ 1 0 ." −z 2 &
# # 1− ν ∂y
+ 1− υ .# #
#$ τ xy #' + 0 0 .# 2 #
* 2 -# −2z ∂ w #
#$ ∂x∂y #'

Prepared by K.K. Wijesundara 2

 −zE # ∂2 w ∂2 w &
σx = % + υ (
1− υ 2 $ ∂x 2 ∂y 2 '
−zE # ∂2 w ∂2 w &
σy = %υ + ( Eq (2)
1− υ 2 $ ∂x 2 ∂y 2 '
−2zE ) 1− υ ,# ∂2 w &
τ xy = + .% (
1− υ 2 * 2 -$ ∂x∂y '
4. Equilibrium Equations
An infinitesimal element of dx and dy extracted from a plate subjected to a distributed
transverse load w is shown in Figure 2. q(x,y)
Y dM yx
M yx + dy
dy
dM y
My + dy
Z dy
dM xy
dy w Qy +
dQy
dy
M xy +
dx
dx
dy
dM x
Mx + dx dQ
dx Qx + x dx
dx
X

dx

Figure 2: Internal forces on infinitesimal element

Where Mx (bending moment about y) My (bending moment about x) are bending moments per
unit width along X and Y directions respectively. Qx (shear force in z dir.) Qy (shear force in z
dir.) are shear forces per unit width along X and Y directions, respectively. Mxy (about y) Myx
(about x) are twisting moments per unit width along X and Y directions, respectively.
Considering vertical force equilibrium of the infinitesimal element:
# # ∂Qx && # # ∂Qy &&
% Qx − % Qx + (( dxdy + % Qy − % Qy + (( dydx − qdxdy = 0
$ $ ∂x '' $ $ ∂y ''
∂Qx ∂Qy
+ +q = 0
∂x ∂y
Considering moment equilibrium about X-axis of the infinitesimal element:

Prepared by K.K. Wijesundara 3

 # # ∂M x && # # ∂M yx &&
% Mx − % Mx + dx (( dy + % M yx − % M yx + dy (( dx
$ $ ∂x '' $ $ ∂y ''
# ∂Q & dx 2
% Qx + x dx ( dydx + q dy = 0
$ ∂x ' 2
∂M x ∂M yx
+ − Qx = 0
∂x ∂y
Similarly,
∂M y ∂M xy
+ − Qy = 0
∂y ∂x
dQx dQy
+ +q = 0
dx dy
dM x dM yx
+ − Qx = 0 Eq (3)
dx dy
dM y dM xy
+ − Qy = 0
dy dx

by substituting the Qx and Qy into the first equation in the equation set (3):

∂2 M x ∂2 M xy ∂2 M y
+ 2 + +q = 0 Eq (4)
∂x 2 ∂x∂y ∂y 2

5. Internal Stresses due to Plate Bending

Y

Z
τ xz

σ xx
X
τ xy
σ yy τ yz
Figure 3: Resultant internal stresses

Prepared by K.K. Wijesundara 4

 h/2
−Eh 3 $ ∂2 w ∂2 w '
Mx = ∫ σ xx z dz = &
12 (1− υ 2 ) % ∂x 2
+ υ )
∂y 2 (
−h/2

h/2
−Eh 3 $ ∂2 w ∂2 w ' Eq (5)
My = ∫ σ yy z dz = &υ + )
12 (1− υ 2 ) % ∂x 2 ∂y 2 (
−h/2

h/2
−2Eh 3 * 1− υ -$ ∂2 w '
M xy = M yx = ∫ τ xy z dz = , /& )
12 (1− υ 2 ) + 2 .% ∂x∂y (
−h/2

By combining Eq (1), Eq (4 ) and Eq (5), the governing partial differential equation for
isotropic, thin-plate bending may be derived as:

Eh3 # ∂ 4w ∂ 4w ∂ 4w &
% 4 +2 2 2 + 4 (= q
(
12 1− υ 2 ) $ ∂x ∂x ∂y ∂y '

Prepared by K.K. Wijesundara 5

6. Finite Element Formulation of Thin Plate Four-Node Rectangular Element
Displacement based Finite Element (FE) method is used to formulate an element stiffness
matrix following the four steps as shown in Figure 4. In the FE Method, the displacement
field inside an element is typically approximated through a polynomial function.

Figure 4: Steps in displacement based finite element procedure

In general, there are three classes of displacement functions, which can be chosen for a plate
element.
1. Class C2: The assumed function w(x,y) has continuous second derivatives
(hence curvature) at element corners and inside the element.
2. Class C1: The assumed function w(x, y) has continuous first derivatives but
may have discontinuous corner curvatures.
3. Class C0: The assumed function w(x, y) has only continuous and independent
functions are assumed to present the variation of w and slopes.
6.1 Selection of Displacement Functions
Degrees of freedom at a node in a four-node plate element are a transverse displacement wi
and normal rotations about each axis of X and Y i.e. θxi and θyi, respectively as illustrated in
Figure 5. Therefore, altogether, there is twelve degrees of freedom per a 4-node plate
element.

Figure 5: Degrees of freedom in a four-node plate element

The 12 degrees of freedom are arranged in the {q} vector as:

Prepared by K.K. Wijesundara 6

 {q}T = {w1 θ x1 θ y1 w2 θ x2 θ y 2 w3 θ x3 θ y3 w4 θ x4 θ y4}T
It is important to note that θxi and θyi are function of wi. θ xi = ∂wi / ∂y and θ yi = ∂wi / ∂x

In the following section, it is discussed the formulating the element stiffness matrix of a four
node rectangular element with dimensions of 2a x 2b using the displacement based finite
element formulation. xc and yc are the coordinate at the center of the element referring the
global coordinate system of X and Y as shown in Figure 6(a).

y η
(-1,+1) (+1,+1)
(xc,yc)
b

b ξ

a a (-1,-1) (+1,-1)
x

Figure 6: Plate element

In deriving the element stiffness matrix, it is expedient to transform the element geometry
shown in Figure 6(a) into a normalized geometry shown in Figure 6(b) referring the
coordinates ξ and η such that ξ=x/a and η=y/b. In the normalized (ξ,η) coordinate system,
the plate coordinates are given by −1 ≤ ξ ≤ 1 and −1 ≤ η ≤ 1.
As all the 12 degrees of freedom are function of vertical translation, w (4 DOFs are directly
function of w and two sets of four DOFs for x-axis and y-axis rotations are function of first
derivatives of w), the assumed displacement function of w can be expressed in a forth order
in-complete polynomial function as:

w ( x , y ) = a1 + a 2ξ + a 3η + a 4ξ 2 + a5ξη + a 6η 2 + a 7ξ 3 + a8ξ 2η + a 9ξη 2 + a10η 3 + a11η 3ξ + a12ξη 3

The terms of the polynomial are selected based on the Pascal’s triangle as shown below.
From the 5 quartic terms, the two terms, which are symmetric (x3y and xy3), are selected for
the displacement function. The choice of x3y and xy3 ensure that there will be a continuity in
the displacement among the inter element boundaries. The terms x4 and y4 would yield
discontinuities along the inter element boundaries. The final term x2y2 cannot be paired with
any other term so it is also rejected.

Prepared by K.K. Wijesundara 7

It is important to note that the displacement function above is one of the earliest plate
bending element formulations investigated. FEM developers have utilized many other
displacement functions and some turn out to be much better than the above. Some involve
sub-dividing the element into three triangular regions and patching a displacement field over
each triangular element. Most of other works are attempted to satisfy stress continuity
conditions along the edges and improve the convergence characteristics of the element for
stresses.
1

𝜉 η

ξ2 ξ𝜂              η2

ξ3 ξ2η ξη2 η3

ξ4 ξ3η ξ2η2 ξη3 η4

Figure 7. Pascal’s triangle

Then “w” displacement function can be rewritten as below in terms of non-dimensional
coordinates (ξ ,η ) using the transformation ξ = x / a and η = y / b by substituting the nodal
coordinates of normalized geometry in each degrees of freedom.

w ( x , y ) = N 1w1 + N 2θ x 1 + N 3θ y 1 + N 4w2 + N 5θ x 2 + N 6θ y 2
+ N 7w3 + N 8θ x 3 + N 9θ y 3 + N 10w4 + N 11θ x 4 + N 12θ y 4

where N1 to N12 are shape functions of the 4-node plate element.

N1 (ξ ,η ) = 18 (1 − ξ )(1 − η ) 2 − ξ − η − ξ 2 − η 2
( )
Similarly,

N 2 (ξ ,η ) = (b / 8)(1 − ξ )(1 − η ) 2 (1 + η )
N3 (ξ ,η ) = (a / 8)(1 − ξ ) 2 (1 + ξ ) 2 (1 − η )

and so forth through N12 (ξ ,η) . Note that {N} is a (12x1) column matrix (since we have 12
degrees of freedom).

Prepared by K.K. Wijesundara 8

6.2. Constitutive Relationship

As εzz=γxz=γyz=0, the constitutive relationship for plate bending is reduced to a plane stress
problem. Therefore, Stress-strain relationship can be expressed as:
{σ } = [ D ]{ε }
where

" 1 ν 0 %
E $ '
[D ] = $ ν 1 0 '
1− ν 2 $ 0 0 (1− ν ) / 2 '
# &

and the strain-displacement relations as discussed before are given by (only have the bending
part)

! % ! 2 2 %
# ε xx # # − z ∂ w / ∂x #
# # # 2 2 #
{ε} = " ε yy & = " − z ∂ w / ∂y &
# # # 2 #
#$ γ xy #' #$ −2 z ∂ w / ∂x ∂y #'
The assumed displacement function for w(x,y) may be substituted into the strain expressions
to obtain:

{ε } = [ B]{q}
where

⎡ − z ∂ 2 N1 / ∂x 2 − z ∂ 2 N 2 / ∂x 2 ... − z ∂ 2 N12 / ∂x 2 ⎤
⎢ ⎥
[ B] = ⎢ − z ∂ 2 N1 / ∂y 2 − z ∂ 2 N 2 / ∂y 2 ... − z ∂ 2 N12 / ∂y 2 ⎥
⎢ 2 2 2 ⎥
⎢⎣ −2 z ∂ N1 / ∂x∂y −2 z ∂ N1 / ∂x∂y ... −2 z ∂ N1 / ∂x∂y ⎥⎦

In the above, it should be noted that Ni=Ni (ξ,η) are function of non-dimensional coordinates
(ξ ,η ) and [B] is a (3x12) matrix. It should be clear that we need to transform these to non-
dimensional coordinates (ξ ,η ) using chain rule. For example, we obtain

3
B11 = − zξ (1 − η )
4a 2
B12 = 0
3
B21 = − zη (1 − ξ )
4b 2

Prepared by K.K. Wijesundara 9

6.3. Element Stiffness Matrix

The element stiffness matrix is developed using the general relation given below as derived
using the principle of virtual displacement.

[k ] = ∫ [ B]T [ D][ B]dV

V

In this case, the differential volume element dV=hdxdy and the integral is over the range x=-a
to +a and y=-b to +b. However, since [B] is written in terms of the non-dimensional
coordinates, we need to transform the volume integral from (x,y) to (ξ ,η ) coordinates.
Hence, in non-dimensional coordinates, dV = abt dξ dη and integration is over the range
ξ = −1 to + 1 and η = −1 to + 1 . The stiffness matrix then becomes:
+1 +1
[k ] = abt ∫ ∫ [ B]T [ D][ B]dξ dη
−1 −1

Note that the element stiffness matrix, [k], is a (12x12) matrix (recall that [B] is (3x12) and
[D] is (3x3)). It should be noted that the purpose of transforming from dimensional (x,y) to
non-dimensional (ξ ,η ) Cartesian coordinates was two-fold: 1) the form of the displacement
function w(x,y) and resulting [N] matrix is simpler in form, and 2) the integration is simpler
because of non-dimensional limits. In addition, because the limits are -1 to +1, Gaussian
quadrature may be easily used to do the integration numerically.

5. Element Force Matrix

The element force matrix {F} is obtained for the plate element in the same manner as is done
for elements like the CST. In this case, the applied load is a distributed pressure acting
normal to the surface (in +z direction). Recall the external potential expression,
V = − ∫ p z wdA where the integral is over the area, A, which the normal load pz acts
A

(normally 2a x 2b). Substituting the displacement function into V gives:

+b + a T
V = −∫ pz wdA = −∫ p {N } {q}dxdy
A −b ∫− a z

Transforming to (ξ ,η ) and noting that the element dof {q} may be taken outside the integral
gives
+1 +1
V = − ⎛⎜ ab ∫ ∫ p z {N }T dξ dη ⎞⎟ {q} = −{F }T {q}
⎝ −1 −1 ⎠

Prepared by K.K. Wijesundara 10

The quantity in the parenthesis can be identified as the generalized nodal force matrix {F}.
Hence, we have
+1 +1
{F} = ab ∫ ∫ pz {N }dξ dη
−1 −1

Note that {F} is a (12x1) column matrix. Note also that {F} contains both forces and
moments, i.e., F1 is a force in the z direction at node 1, F2 is a moment about the x-axis at
node 1, F3 is a moment about the y-axis at node 1 and so on. If thermal effects are being

considered, this will add another term to the element force matrix. See any finite element text
for including this.

Prepared by K.K. Wijesundara 11