1 BERNOULLI BEAM THEORY (EXCLUDES SHEAR DEFORMATIONS)

Shear correction factors for beams, plates and shells

by

Fredy Andrés Mercado Navarro

NOTES

30 de julio de 2016
Structural elements are called degenerate isoparametric elements because in their for-
mulation, the displacements u, v, w are interpolated in terms of midsurface displacements
and rotations and because there is the major assumption that the stress normal to the
midsurface is zero. Continuum elements, on the other hand, have their displacements
interpolated in terms of nodal point displacements of the same kind.
Beam, plate and shell elements can be formulated using Bernoulli beam and Kirch-
hoff plate theory, in which shear deformations are neglected (taken from Finite Element
Procedures, Bathe, p. 397).

Problem Displacem. Stress vec- Material matrix C Strain

component tor τ T vector T
Beam w [Mxx ] EI [κxx ]
 
1 ν 0
 
Eh3
Plate bending w [Mxx Myy Mxy ] 12(1−ν 2 ) ν 1 [κxx κyy κxy ]
 
0 
 
1−ν
0 0 2

Tabla 1: Kinematic and static variables in beam and plate bending problems.

With h being the thickness of the plate, I the moment of inertia and,

∂ 2w ∂ 2w ∂ 2w
κxx = κyy = κxy = 2 (1)
∂x2 ∂y 2 ∂x∂y

1. Bernoulli beam theory (excludes shear deforma-

tions)
Also called Clasical beam theory. The basic assumption in beam bending analysis
exluding shear deformations is that a normal to the midsurface (neutral axis) of the
beam remains straight during deformation and that its angular rotation is equal to the
slope of the beam midsurface (see Figure 1).

1
 1 BERNOULLI BEAM THEORY (EXCLUDES SHEAR DEFORMATIONS)

Figura 1: Beam deformations excluding shear effect.

This kinematic assumption leads to the well-known beam-bending governing diffe-

rential equation in which the transverse displacement w is the only variable. Therefore,
using beam elements formulated with this theory, displacement continuity between ele-
ments requires that w and dw/dx be continuous.
dw
βBernoulli = β(x) = (2)
dx
NOTE: don’t forget that w is the displacement in z direction.

Following the assumptions, the displacement field is written as,

dw
u = −zβ(x) = −z v=0 w = w(x) (3)
dx
The procedure to derive u is (see Fig. 2 and Oñate, vol. 2),
u u
sin β = β→0 β= u = −zβ (4)
z z

2
 1 BERNOULLI BEAM THEORY (EXCLUDES SHEAR DEFORMATIONS)

Figura 2: Derivation of u.

Rotation is equal to the slope of the beam axis. Strain and stress fields are,

du d2 w
εx = = −z 2 = −zκ εy = εz = γxy = γxz = γyz = 0 (5)
dx dx
i.e. the strain is under a pure axial strain (εx ) state. The axial stress σx is related to εx
by Hook law as,

d2 w
σx = Eεx = −zE (6)
dx2
E is the Young modulus. The bending moment for a cross section is defined as (Oñate
Vol. 2, pág. 3),

d2 w d2 w
Z Z Z Z 
2
M =− zσx dA = z dA E 2 = EIy 2 = EIy κ (7)
A A dx dx
where,
Z Z 
2
Iy = z dA (8)
A

is the moment of inertia of the section with respect to the y axis and,

d2 w
κ= (9)
dx2
is the curvature of the beam axis.

3
 2 TIMOSHENKO BEAM THEORY (INCLUDES SHEAR DEFORMATIONS)

Figura 3: Sign criteria for axial stress σx , bending moment M and shear force Q.

2. Timoshenko beam theory (includes shear defor-

mations)
This theory is also called Thick beam theory. Here we retain the assumption that a
plane section originally normal to the neutral axis remains plane, but because of shear
deformations this section does not remain normal to the neutral axis (see Figure 4).
Timoshenko hypothesis is equivalent to assuming an average rotation for the deformed
cross section which is kept plane (Oñate, Vol. 2, p. 38).

Figura 4: Beam deformations including shear effect.

The total rotation of the plane originally normal to the neutral axis of the beam is
given by the rotation of the tangent to the neutral axis and the shear deformation,

4
 2 TIMOSHENKO BEAM THEORY (INCLUDES SHEAR DEFORMATIONS)

dw
βTimoshenko = −γ (10)
dx
So, β(x) is the angle with respect to the z axis that the shear strains produce in the
material, dw/dx is the slope of the beam axis and γ is a constant shearing strain
across the section (or an additional rotation due to the distortion of the cross-section).
The displacement field is written as,
 
dw
u = −zβ(x) = −z −γ v=0 w = w(x) (11)
dx

The strain and stress fields are (use Eq. 11 to derive the expressions),

du dβ dw du dw(x)
εx = = −z γxz (x) = + = − β(x) = γ (12)
dx dx dx dz dx
The other strain components are zero. Hence, Timoshenko theory introduces a transverse
shear deformation γxz , which value coincides with the rotation γ (see Fig. 4). The axial
and shear stresses σx and τxz at a point of the beam cross section are related to the
corresponding strains by,
 
dβ(x) dw(x)
σx (x, z) = Eεx = −zE τxz (x) = Gγxz (x) = G − β(x) (13)
dx dx

where G is the shear modulus,

E
G= (14)
2(1 + ν)
and ν is the Poisson ratio. Timoshenko beam theory accounts for the effect of transverse
shear deformation. Timoshenko beam elements are therefore applicable for “thick” beams
(L/h < 10) where transverse shear deformation has an influence in the solution, as well
as for slender (esbeltas) beams (L/h > 100) where the influence is irrelevant (Oñate Vol.
2, pág. 37).

2.1. Timoshenko: resultant stresses and generalized strains

The bending moment M and the shear force Q are defined with the sign criterion of
Figure 5 as,

5
 2 TIMOSHENKO BEAM THEORY (INCLUDES SHEAR DEFORMATIONS)

Figura 5: Timoshenko beam. Normal and tangential stresses.

Z Z Z Z
M =− zσx dA Q= τxz dA (15)
A A

Remember that σx = σx (x, z) and τxz = τxz (x). This means that if we fix the x
coordinate the stress σx will vary linearly with z. On the other hand τxz is just a function
of the x coordinate, so it is constant through the thickness of the beam. Substituting Eq.
13 in Eq. 15 we get,

Z Z   Z Z
dβ(x)
M =− z −zE dA Q= (Gγxz (x)) dA (16)
A dx A

Z Z  Z Z 
2 dβ(x)
M= Ez dA · Q= GdA · γxz (x) (17)
A dx A

dβ(x)
M = D̂bending · Q = Ĝ · γxz (x) (18)
dx
κ = dβ(x)/dx is the bending strain (not curvature). D̂bending and Ĝ are resultant
values over the section. For homogeneous materials,

6
 2 TIMOSHENKO BEAM THEORY (INCLUDES SHEAR DEFORMATIONS)

Z Z Z Z 
2 2
D̂bending = Ez dA = E z dA = EIy Ĝ = GA (19)
A A

σx varies linearly through the thickness, which is considered “exact” according to

classical beam theory, but τxz is constant across the thickness, which is in contradiction
with the exact quadratic distribution for a rectangular beam.

2.2. Shear stress distribution in rectangular beam cross section

The shear stress analysis presented here was developed by D. J. Jourawski (russian
engineer, 1821-1891. See Gere). The Eq. for the shear stress over the cross section that
comes from force equilibrium equations is (see Strength of materials by Timoshenko, Vol.
1, page 114),

Figura 6: Figure taken from book of Timoshenko, Vol. 1.

Z h/2
V
τxz = zdA (20)
Iy b z1

V is the shear force, Iy is the moment of inertia around y axis and b is the width of
the cross section. The following integral is the moment of the shaded portion of the cross
section with respect to the neutral axis y (with dA = bdz for rectangular cross section).

7
 2 TIMOSHENKO BEAM THEORY (INCLUDES SHEAR DEFORMATIONS)

h/2
b h2
Z  
2
zdA = − z1 (21)
z1 2 4
the same result can be obtained by multiplying the area of the shaded portion
   
h 1 h
b − y1 by the distance + z1 (22)
2 2 2
Which is the distance of its centroid to the neutral axis. Substituting Eq. 21 into Eq.
20 we get,

V h2
 
2
τxz = − z1 (23)
2Iy 4
Shearing stresses are not uniformly distributed from top to bottom of the beam. The
maximum value of τxz occurs for z1 = 0 (on the neutral axis). If the moment of inertia
around y is,

bh3
Iy = (24)
12
Then Eq. 23 becomes,
  2 
h
− z12 
z12
 
3V 
 2  3V h h
τxz (z1 ) =   2  = 1−4 2 for − ≤ z1 ≤ (25)
2A  h  2A h 2 2
2
Maximum shear stress occurs for z1 = 0,

3V
(τxz )máx = (26)
2A
For the bottom and for the top of the cross section z1 = ±h/2 and τxz = 0.

2.3. Shear strain energy

The strain energy in a general state of stress and strain is,
1
dU = (σxx εxx + ... + τxy γxy + ... + τxz γxz ) dV (27)
2
Integrating over the entire volume,
Z
1
U= (σxx εxx + ... + τxy γxy + ... + τxz γxz ) dV (28)
2 V
Z
1
U= σ T εdV (29)
2 V
The shear strain energy for the beam is,
Z
1 τxz
Us = τxz γxz dV with γxz = (30)
2 V Gxz

8
 2 TIMOSHENKO BEAM THEORY (INCLUDES SHEAR DEFORMATIONS)

then,
Z
1 2
Us = τxz dV (31)
2Gxz V
Using τxz found before for a rectangular cross section where shear stresses have a
parabolic distribution we have,
h/2 2 Z b Z L
z2
Z  
1 3V
Us = 1−4 2 dxdydz (32)
2Gxz −h/2 2bh h 0 0
Z h/2  2
z2

bL 3V
Us = 1−4 2 dz (33)
2Gxz −h/2 2bh h
Z h/2 
1 9 bLV 2

8 2 16 4
Us = · 1 − 2 z + 4 z dz (34)
2 4 h2 b2 Gxz −h/2 h h
"Z #
h/2 Z h/2 Z h/2
1 9 bLV 2 8 16
Us = · dz − 2 z 2 dz + 4 z 4 dz (35)
2 4 h2 b2 Gxz −h/2 h −h/2 h −h/2

1 9 bLV 2
 
2 1
Us = · h− h+ h (36)
2 4 h2 b2 Gxz 3 5
1 9 bLV 2 1 3 LV 2 2
   
8
Us = · h = · (37)
2 4 h2 b2 Gxz 15 2 1 hbGxz 5
1 LV 2
 
6
Us,parabolic = · · (38)
2 hbGxz 5
Now, assuming a constant shear distribution we have τxz = V /A, then based in Eq.
31,
Z Z  2
1 2 1 V
Us,constant = τxz dV = dV (39)
2Gxz V 2Gxz V A
h/2 Z b Z L
V2 hbLV 2
Z
Us,constant = 2 2 dz dy dx = (40)
2h b Gxz −h/2 0 0 2h2 b2 Gxz
LV 2
Us,constant = (41)
2hbGxz
Strain energies are made equal and a shear correction factor κxz is introduced in the
expression for the strain energy given by the constant shear stress distribution,

Us,constant = Us,parabolic (42)

LV 2 1 LV 2
 
6
= · · (43)
2hbκxz Gxz 2 hbGxz 5
LV 2 1 LV 2
= ·   (44)
2hb (κxz Gxz ) 2 5
hb Gxz
6

9
 3 PLATE BENDING

Finally,
5
κxz = (45)
6
See http://people.duke.edu/ hpgavin/cee201/strain-energy.pdf for more interesting
information.
Since the actual shearing stress and strain vary over the section, the shearing strain
γ in Eq. 10 is an equivalent constant strain on a corresponding shear area As ,

V τ As
τ= γ= κ= (46)
As G A
where V is the shear force at the section being considered. Different assumptions
may be used to evaluate a reasonable factor κ. One simple procedure is to evaluate
the SCF using the condition that when action on As , the constant shear stress in Eq.
46 must yield the same shear strain energy as the actual shearing stress (evaluated from
beam theory) acting on the actual cross-sectional area A of the beam.

3. Plate bending
To calculate the generalized stress-strain matrix for plate bending analysis the stress-
strain matrix corresponding to plane stress conditions is used. The strains at a distance
z measured upward from the midsurface of the plate are,

∂ 2w ∂ 2w ∂ 2w
 
−z 2 −z 2 −2z (47)
∂x ∂y ∂x∂y
In plate bending analysis it is assumed that each layer of the plate acts in plane stress
condition and positive curvatures correspond to positive moments. Hence, integrating the
normal stresses in the plate to obtain moments per unit length, the generalized stress-
strain matrix is,
 
Z +h/2 1 ν 0
E  
Cb = z2 0  dz (48)
 
1 − ν 2 ν 1

−h/2 
0 0 1−ν
2

We have that,
+h/2
h3
Z
z 2 dz = (49)
−h/2 12
Then,  
1 ν 0
Eh3  
Cb = (50)
 
ν 1 0 
12(1 − ν 2 )  
1−ν
0 0 2

The basic proposition in plate bending and shell analyses is that the structure is thin
in one dimension, and therefore the following assumptions can be made:

10
5 REISSNER-MINDLIN PLATE THEORY (INCLUDES SHEAR DEFORMATIONS)

1. The stress through the thickness (i.e., perpendicular to the midsurface) of the pla-
te/shell is zero.

2. Material particles that are originally on a straight line perpendicullar to the mid-
surface of the plate/shell remain on a straight line during deformations.

KIRCHHOFF THEORY: shear deformations are neglected (the straight line re-
mains perpendicular to the midsurface during deformations).

REISSNER/MINDLIN THEORY: shear deformations are included (the straight

line in general does not remain perpendicular to the midsurface during deformations).

4. Kirchhoff plate theory (excludes shear deforma-

tions)
For the small displacement bending theory. Rotations in the x,z and y,z planes:

∂w ∂w
βx = βy = (51)
∂x ∂y

5. Reissner-Mindlin plate theory (includes shear de-

formations)
For the small displacement bending theory. Displacements are:

u = −zβx (x, y) v = −zβy (x, y) w = w(x, y) (52)

with rotations as,

∂w ∂w
βx = βy = (53)
∂x ∂y
See Figure 7. For this case bending strains εxx , εyy and εxy vary linearly through the
plate thickness and are given by the curvatures of the plate using Ec. 52,

11
5 REISSNER-MINDLIN PLATE THEORY (INCLUDES SHEAR DEFORMATIONS)

Figura 7: Deformation assumptions in analysis of plates including shear deformations.

∂u ∂ ∂βx (x, y)
εxx = = [−zβx (x, y)] = −z (54)
∂x ∂x ∂x
∂v ∂ ∂βy (x, y)
εyy = = [−zβy (x, y)] = −z (55)
∂y ∂y ∂y
 
∂u ∂v ∂ ∂ ∂βx ∂βy
γxy = 2εxy = + = (−zβx ) + (−zβy ) = −z + (56)
∂y ∂x ∂y ∂x ∂y ∂x
then, in matrix form,
∂βx
 
 
∂x
εxx
 
 
   ∂β y

 εyy  = −z  (57)
   
∂y

   
 
γxy  ∂βx ∂βy 
+
∂y ∂x
whereas the transverse shear strains are assumed to be constant through the thickness
of the plate,
 ∂w 
− βx
 
εxz  ∂x 
 =  (58)
 ∂w
εyz

− βy
∂y
Each transverse shear strain component is of the form used in the description of the
beam deformations.

12