Lecture notes on Mechanics of composite

materials
Tomas Mares
January 4, 2016

Composite materials

What composite materials are and how they are made see
http://en.wikipedia.org/wiki/Composite material.

Mechanics

Mechanics (in Greek M ) is a branch of physics dealing with the movement of bodies and its causes. Mechanics is based on two sets of axioms. They
are either Newtons laws of motion or the principle of least action. Starting with
Newtons laws we can, using variational methods, easily obtain the principle of
least action and vice versa.
Just a reminder:
Newtons law of inertia (it defines inertial frame)
F =0a=0
Newtons law of force and acceleration
F = ma

Newtons law of action and reaction

F =- R
and on the other hand the principle of least action says the path taken
Z by a
t2

body (or a system) minimizes the action

S=

Ldt

t1

where the Lagrangian

L=T V
In our lectures we are interested only in statics (a = 0) of deformable bodies of
a special kind, namely bodies obeying the Hookes law
ab = E abcd cd

Hookes law

The last expression of Hookes law is writen in tensor notation nevertheless we

will use the Voigts notation1
= E
1 http://en.wikipedia.org/wiki/Voigt

notation

E11 E12 E13
11
22 E21 E22 E23

33 E31 E32 E33

i.e.
23 = E41 E42 E43

31 E51 E52 E53
E61 E62 E63
12
It seems there are 36 independent entries in E
As strain energy

E14
E24
E34
E44
E54
E64

E15
E25
E35
E45
E55
E65

E16
E26
E36
E46
E56
E66

11
22
33
223
231
212

1 0

2
0
0 0
2u = = E
u=

and
and at the same time 2u = 0 = 0 E
it follows
E = E0
and that there are
entries
21 independent

11
11
E11 E12 E13 E14 E15 E16
22 E12 E22 E23 E24 E25 E26 22

33 E13 E23 E33 E34 E35 E36 33

namely
23 = E14 E24 E34 E44 E45 E46 223

31 E15 E25 E35 E45 E55 E56 231

12
212
E16 E26 E36 E46 E56 E66
It holds true for every lineary elastic material. We call it Hookes law for
anisostropic material. In what follows we will study material symmetries.

Monoclinic material

In mechanics of composite materials we study symmetry in other way than in

crystallography. What we call monoclinic material is a material that have one
plane of material symmetry in point like sense. What I meen is the fact that
Hookes law in the stated form is point like and to state material symmetry it
is sufficient to study this Hookes law. We call a plane of material symmetry
such a plane with respect to which both stress and strain is either symetric or
anisotropic (both the same).

4.1

Monoclinic material with the plane of symmetry being

plane 12

Let us say, in 123 coordinate system, the plane 12 is the plane of symmetry.
Then to insure the material symmetry the entries of E that bind the entries of
symmetric stress and antisymmetric strain and vice versa should be equal zero.
And so the stiffness matrix must be like

11
E11 E12 E13
0
0
E16
11
22 E12 E22 E23

0
0
E26

22
33 E13 E23 E33

0
0
E36 33

23 = 0

0
0
E44 E45
0

223
31 0

0
0
E45 E55
0
231
12
E16 E26 E36
0
0
E66
212
2

4.2

Monoclinic material with the plane of symmetry being

plane 23

If the plane 23

11
22

33

23

31
12

4.3

is the plane of symmetry

E11 E12 E13
E12 E22 E23

E13 E23 E33
=
E14 E24 E34

0
0
0
0
0
0

then
E14
E24
E34
E44
0
0

0
0
0
0
E55
E56

0
0
0
0
E56
E66

11
22
33
223
231
212

Monoclinic material with the plane of symmetry being

plane 31

If the plane 31

11
22

33

23

31
12

is the plane of symmetry then

E11 E12 E13
0
E15
E12 E22 E23
0
E25

E13 E23 E33
0
E35
=
0
0
0
E44
0

E15 E25 E35
0
E55
0
0
0
E46
0

0
0
0
E46
0
E66

11
22
33
223
231
212

A monoclinic material has 13 independent material characteristics.

Orthotropic materal

An orthotropic material is a material that have three mutually perpendicular

planes of symmetry, let us say 12,23,31. As every one of the three above mentioned monoclinic cases holds there is just one way

11
E11 E12 E13
0
0
0
11
22 E12 E22 E23

0
0
0

22
33 E13 E23 E33
33
0
0
0

23 = 0

0
0
E44
0
0

223
31 0
0
0
0
E55
0 231
12
0
0
0
0
0
E66
212
An orthotropic material thus has 9 independent material characteristics.

Transverse isotropic material

If there is an axis such that every plane containing this axis is a plane of material symmetry then this material is called transverse isotropic material. This
material has 5 independent characteristics as may be shown using rotational
transformation about the axis of symmetry.
3

Isotropic material

It is symetric with respect to every plane and there are only 2 independent
material characteristics.

Orthotropic material in more detail

Elasticity tensor E abcd and compliance tensor Cabcd
Isotropic material (, Lame coefficients)
E abcd = g ab g cd + g ac g bd + g ad g bc

Ciarlet, P. G. (2005)
Mare
s, T. (2006)

3
2

2
before deformation

Orthotropic block

after deformation

Young modulus in the direction 1

11

11

12 =

11

12
1
, 13 = 31
11
1

22
,
22
33
= 3,
3

E22 =
and of 3

E33

21
,
22
3
= 13 ,
3
21

23
22
3
= 23
3

21 =

23 =

31

32

Pure shear 2
From the definition

12 = 21

Strain in the 1 excited by all normal stresses

11 = 11 + 21 + 31

11

12 1

11
22
33
=
21
31
E11
E22
E33

Similarly in the other directions (G23 , G31 )

12

the equilibrium equation

12 = 21

11

Similarly in the direction of 2

12
2

Poisson ratios

the stress

E11 =

21

12 = 12 = G12 (12 + 21 )
At the tensor notation. . . 15 34

Compliance tensor Cabcd

Ciarlet, P. G. (2005)
Mare
s, T. (2006)

in Cartesian coordinate system a

alined with the principal material axes of the orthotropic material

E11

= E12
11

E13
11

0
1
4G12

1
4G12

0
0
0
0
0

0
0
1
4G13

0
0
0

E21
22
0
0
0

0
1
4G12

1
4G12

1
E22

0
0
0
0
0

1
4G13

0
0

0
0
0
0
0

0
0
0

1
4G23

E23
22

1
4G23

0
0
1
4G13

0
0
0

1
4G13

0
0

0
0
0
0
0
1
4G23

1
4G23

E31
33
0
0
0
E32
33
0
0
0
1
E33

11
12
13
21
22
23
31
32
33

ab = Cabcd cd
Cabcd = Ccdab = Cbacd
Energy

11
12
13
21
22
23
31
32
33

Equilibrium
Elasticity tensor E abcd . . . 16 34

Plane stress of an orthotropic material

Plane stress is a stress state where 3a = 0. Then we can, in the main coordinate
system of orthotropy, write
= C , i.e.

1
ETTL
0
11
11
EL
1
22
22 = LT
0
EL
ET
1
12
212
0
0
GLT

or the inverse relation

11
EL
1
22 =
T L EL
1 LT T L
12
0

LT ET
ET
0

0
11
22
0
GLT (1 LT T L )
212

Symbolically
= E
The matrix C is called Compliance matrix and matrix E is called Stiffness
matrix.

10
2
BM

2D vector Cartesian transformation

x2
6
B



B
B












1 1





 











B






 


B


 









B
 



 




 
B















B b
 

v 
 




 


B 6





7



B    












B














B  



 




B
- 1
a
x
Let us have a vector v and two vectors a and b such that
v = a +b
a
The coordinates of vectors v , a and b in thecoordinate
system
x
are
respec




x
x
x
a
a
0
tively
v=
, a=
and b =
b
0
b



a
cos

a
In the coordinate system the vector a has coordinates
a=
a sin



b sin
vector b has coordinates
b=
b cos 


a cos + b sin
and vector v
v =a + b =
a sin + b cos

x
Thus we have coordinate transformation of a vector in the form
v = T x v
6

where the transformation matrix

T x =

cos
sin

T 1
x

cos
sin

For the inverse transformation

T x =

the transformation matrix is inverse

11


sin
cos
x

v = T x 
v
sin
cos

Transformation of Voigt stress vector

As the stress is second order tensor we must at first look at second order tensor
transformation. Direct multiplication of two first order tensor may be represen T

vv

teted as matrix multiplication of components

Using transformation rules stated above
For the stress tensor then




11 12
cos sin
11
=
sin cos
21 22
21

12
22

 
x

11
21

12
22

x
x
x
x
x
x
11
cos2 + 212
sin cos + 22
sin2
(22
11
) sin cos + 12
(cos2 sin2 )
2
2
x
x
x
2
x
x
x
(22 11 ) sin cos + 12 (cos sin )
11 sin 212 sin cos + 22
cos2

Rearranging

cos2
11

22

=
sin2

12
sin cos
Symbolically
Inverse transformation
can be obtained both
and
which leads to

T x

12

sin
cos

cos
sin

Executing multiplication on the right site gives (using 12 = 21 )

x xT

v v = T x v v T Tx

cos2

=
sin2
sin cos

sin2
cos2
sin cos

x
2 sin cos
11
x
2 sin cos 22
2
x
2
12
cos sin

= T x
x

= T x
T x )1
T x = (T

T x () = T x ()
sin2
cos2
sin cos

2 sin cos
2 sin cos
cos2 sin2

Transformation of Voigt strain vector

Strain tensor has the same structure as stress tensor and so the transformation
of Voigt strain vector would by the same as the transformation of Voigt stress
vector as long as the structure of the vectors is the same. But it is not. There

is 212 instead of 12 in the last entry. This factor of 2 must be incorporated in

the transformation matrix which leads to the transformation matrices

cos2
sin2
sin cos
T x =
sin2
cos2
sin cos
2 sin cos 2 sin cos cos2 sin2
T x

13

cos2
=
sin2
2 sin cos

sin2
cos2
2 sin cos

sin cos
T x )T
sin cos = (T
2
2
cos sin

Stiffness matrix transformation

x

xx

=E =E T x

As

=E

= T x = T x E T x
x

and

14

Compliance matrix transformation

x

= ...

Similarly

15

T x )T
E = T x E T x = T x E (T

it holds
For inverse transformation...

Composite micromechanics

Given the micromechanical geometry and the material properties of each constituent, it is possible to estimate the effective composite material properties
and the micromechanical stress/strain state of a composite material.
Thus, for fibre composite we can estimate...

16

Strength theories for filamentary composite

materials

17

Composite laminate layup nomenclature

A laminate is an organized stack of uni-directional composite plies (uni-directional

meaning the plies have a single fiber direction rather than a weave pattern). The
stack is defined by the fiber directions of each ply like this:

z0
z1

= 1
= 2
= 3

-x

= N 1
= N

zN 1
zN

z
?
1 < 0 -x

2 > 0

= 1

= 2

y
?
= 3

Such laminates are often described by an orientation code

[1 /2 /3 /4 ]
For example
[0/-45/90/45/0/0/45/90/-45/0]
Short hand
[0/-45/90/45/0]s
Other examples of short hand
[0/90]4
or
[0/45/90], [0/452 /30]
etc.

18

Equilibrium equation of a laminated plate (a

laminate)

= 1
= 2
= 3

6
y


= N 1
= N

?
x

z0
z1

-x

= 1
= 2
= 3

= N 1
= N

zN 1
zN

- dx 

dy
-

?
1 < 0 -x

dy

?
Cut out the element dxdy
and apply the internal forces and moments
as resultants of applied stresses

x
y
?

- dx 

z0
z1

-y
zN 1
zN

x

y



-yx

yy

?yz

- xx (x + dx, z, y)


xy
?
xz

z
?

As h is finite the stresses are unknown functions of z. On the other hand the
dimensions dx and dy are infinitesimally small and we may approximate the
functions according to Taylor series
ab (x + dx, y, z) = ab + ab,x dx
and
ab (x, y + dy, z) = ab + ab,y dy
where by ab we understand
ab = ab (x, y, z)
Now, to write equilibrium equations we need forces and moments acting upon
the element. The acting generalized forces are the resultants of the stresses
Z h2
Nxx =
xx dz
h
2

Nyy =
Nxy =

h
2

yy dz

h
2

h
2

xy dz

h
2

Nyx =

h
2

yx dz

h
2

Qxz =

h
2

xz dz

h
2

Qyz =
Mxx =

h
2

yz dz

h
2

h
2

zxx dz

h
2

Myy =

h
2

h
2

10

zyy dz

Mxy =

h
2

zxy dz

h
2

Myx =

h
2

zyx dz

h
2

From the definition of these quantities we see that they are not forces or moments
but in fact linear densities of these forces and moments. To get a real forces we
need to multiply them by the width of the appropriate area of the element.

11

The forces acting in the x-direction and the equilibrium equation

Nxx,x + Nyx,y = 0

 Nyx

Nxx

x
- Nxx + Nxx,x dx

Nyx+Nyx,y dy
-


y



(x)

z
?

The forces acting in the y-direction and the equilibrium equation

Nxy,x + Nyy,y = 0

Nyy
*

Nxy
*

x
-


Nxy + Nxy,x dx
y




+ Nyy,y dy
Nyy

z
?

12

(y)

The forces acting in the z-direction and the equilibrium equation

p = p(x, y)
? ? ? ? ?
? ? ? ? ?
? ? ? ? ?
? ? ? ? ?
? ? ? ? ?

Qyz 6

(z)

x
-

6
Qxz

y



p + Qyz,y + Qxz,x = 0

Q + Qyz,y dy
? yz

?
Qxz + Qxz,x dx

z
?
The moments acting in the x-direction and the equilibrium equation
+resultant moment
of the couple Qyz -Qyz

Myy
Mxy
y



M +Myy,y dy
 yy

Mxy,x + Myy,y Qyz = 0

x
Mxy + Mxy,x dx

z
?

13

(mx)

The moments acting in the y-direction and the equilibrium equation

+resultant moment
of the couple Qxz -Qxz

Myx
*

Mxx
*

Mxx,x + Myx,y Qxz = 0

(y)

x
-


Mxx + Mxx,x dx

y



M
yx + Myx,y dy

z
?
Puting these equilibrium equations together we get
Mab,ab = p
and
Nab,a = 0 where a, b = x, y
There are three equations for six unknown. We need a compatibility equation.
The most common one is Kirchhoff hypothesis resulting in Classical lamination
theory.

19

Classical lamination theory

In Classical lamination theory we assume Kirchhoff hypothesis that says that

points on a normal to an undeformed middle plane stay on a normal to the
deformed middle plane.
Following the Kirchhoff hypothesis shown on the figure below
uo = uo (x, y)
vo = vo (x, y)
w = wo = w(x, y)
u = uo zw,x
v = vo zw,y

14

z
?

rAo
z
6
? rBo

-x

w(x, y)
w,x
Kirchhoff hypothesis
?
uo
rA
u
w,x
rB
z

rAo
z
6
? rBo

w(x, y)
w,y
Kirchhoff hypothesis
?
vo
rA
v
w,y
rB

z


z
?

1
ab = (ua,b + ub,a )
2
xx = u,x = uo,x zw,xx
yy = v,y = vo,y zw,yy
1
xy = (uo,y + vo,x ) zw,xy
2
zx = 0
yz = 0
zz = 0
The last expression is in contrariety with the assumption of plane stress. . .
Now, we are to express the stresses using the Hookes law for plane stress state.
Why plane stress when the Kirchhoff hypothesis leads to plane strain we will

From Cauchys strain tensor formula

we have

xx

=E
xx
x
= yy
xy

discuss later. According to (..) we have

where

uo,x
xx
w,xx
z w,yy = o + z

vo,y
and
= yy =
1
w
xy
(u
+
v
,xy
o,y
o,x
2
Note the change in the sign due to the definition of the curvature vector .
Z h2 x
Z h2
x
)dz
For the generalized forces
N =
dz =
E (o + z
x

or, as o and do not depend on z,

Z
where
A=

h
2

h
2

E dz =

h
2

15

h
2

N Z
X

=1

z1

-y

N = A o + B

T x ( ) E T x ( )dz

B=

and

h
2

z E dz =

h
2

N Z
X

=1

B=

i.e.

z1

N
2
X
z2 z1

=1

M =

Similarly for the moments

T x ( ) E T x ( )dz
zT

h
2

z dz =

h
2

or, as o and do not depend on z,

Z h2
N Z
X
x
where
D=
z 2 E dz =
h
2

D=

i.e.

20

h
2

h
2

)dz
z E (o + z
M = B o + D

z 2T x ( ) E T x ( )dz

=1 z1
N
3
X
z3 z1
=1

T x ( ) E T x ( )

T x ( ) E T x ( )

Symmetric laminate

Symmetric laminate is a laminate for which for every there is a such

that
= and z = z1
Then
B =0
and
N = A o
M = D
Using (here) and (there) we get
Dabcd wabcd = p
and
Aabcd uoc,ad = 0
Add comment on coupling...

21

Balanced laminate

22

Solved problems not only on B = 0 case

23

Buckling analysis of laminated plates

Let us consider symmetric laminate B = 0. For this case we have from above
Dabcd w,abcd = 0
Nab,a = 0
These equations of equilibrium have been derived under the undeformed geometry configuration. As in the case of column buckling we need to look at the
case of deformed shape.

16

-x

-y

w(x, y)

z
?

w(x, y)

w,y
yNxq

yNyq

w,x

w,x + w,xx dx

w,y + w,yy dy

q
Nx + Nx,x dx
The contribution to the z-direction equilibrium equation

q
Ny + Ny,y dy

Nx dy w,x + (Nx + Nx,x dx)(w,x + w,xx dx)dy Ny dx w,x + Ny + Ny,y dy)(w,y + w,yy dy)dx

y








x
-



Nxy w,y dy + (Nxy + Nxy,x dx)(w,y + w,yx dx)dy

Nxy w,x dx + (Nxy + Nxy,y dy)(w,x + w,xy dy)dx
z?



w ,y 
Nxy

w,x

i
Nxy

w,x + w,xy dy
Nxy +
Nxy
,y dy q

w,y + w,yx dx
Nxy + Nxy,x dx

The Figures above show forces whose components in the z-direction are zero

17

if the element is in undeformed position. Nevertheless, if deformed, as on the

Figures, there are nonzero components in the direction. That means the equilibrium equation in the z-direction
Dabcd w,abcd dxdy = pdxdy
has the following additional terms (after using Nab,a = 0 and O(3) = 0) on its
right side:
Dabcd w,abcd = p + Nab wab

24

Buckling of platessolved example

For one layered and orthotropic plate with k x loaded as shown in the figure
and simply supported we have the following.

x
-








?
y
?

In the Lame equation of equilibrium

we have

EL
0
0
LT ET

T L EL

ET,L
0
, ET,L =

0
1 LT T L
ET
further
Nx = F, Ny = 0, Nxy = 0, p = 0
Thus we get
D1 w,xxxx + D12 w,xxyy + D2 w,yyyy = F w,xx
E h3
T L EL h3
LT ET h3
4GLT h3
E h3
where
D1 = L , D12 =
+
+
, D2 = T
12
12
12
12
12
Let as look for the solution using Fourier series expansion

X
nx
ky
w=
wnk sin
sin
a
b
x

where Eabcd =Eabcd =

0
GLT
GLT
0

0
GLT
GLT
0

Dabcd w,abcd = p + Nab wab

Z h2
x
Eabcd z 2 dz
Dabcd =
h
2

n,k=1

Using in our Lame equation

18

wnk

n,k=1

 4
 n 2  k 2
 n 2
 n 4
k
+D12
+D2
F
D1
a
a
b
b
a

As functions sin
tions

sin

nx
ky
sin
=0
a
b

nx
ky
sin
are linearly independent, there are possible solua
b

4

2 k
2

4
+D12 n
D1 n
+D2 k
a
a
b
b
Fnk =

2
n
a

The corresponding eigenmodes are

sin

nx
ky
sin
a
b

25

Sandwich beam theory

26

Thermal deformation of simple composite beams

26.1

BimetalA beam made of two materials

Consider a beam made of two different materials unloaded by any force or

external moment but undergoing a change in temperature (see the Fig.)
b

1st material: E1 , 1

Neutral axis

2nd material: E2 , 2

y1 < 0
r

rC1

h1

y2 > 0

rC2

h2

y
The material properties are described by the Youngs modulus (E1 , E2 ) and
coefficient of thermal expansion (1 , 2 ).
As the beam is unloaded by external forces the overall internal normal force, N ,
and bending moment, Mb , are zero:
N = 0, Mb = 0
Let us suppose that the Bernoullis hypothesis holds:
= ky
where k is the curvature and y the coordinate.
The strain can be decomposed into its elastic and thermal parts:
= elastic + thermal =

19

+ T
E

That gives
= Eky ET
For the normal force we then have
Z
Z
N=
dA =
A

N=

A1

E1 ky dA

1 dA +

A1

E1 1 T dA +

A1

2 dA

A2

A2

E2 ky dA

E2 2 T dA

A2

i.e.
N = E1 kQ1 E1 1 T A1 + E2 kQ2 E2 2 T A2
where Q1 and Q2 are the first moment of area of the cross-section of the 1st
and 2nd material with respect to the Neutral axis, respectively.
For the bending moment we can write
Z
Z
Z
Mb =
y dA =
y1 dA +
y2 dA
A

Mb =

A1

E1 ky 2 dA

A1

yE1 1 T dA +

A2

A2

A1

E2 ky 2 dA

yE2 2 T dA

A2

i.e.
Mb = E1 kI1 E1 1 T Q1 + E2 kI2 E2 2 T Q2
where I1 and I2 are second moment of area with respect to the Neutral axis of
the respective areas.
As N = 0 and Mb = 0 we have the conditions fixing the position of the Neutral
axis and the curvature, k, which in the case of the rectangular cross-section
gives
k=

26.2

E12 h41

6E1 E2 (h1 + h2 )h1 h2 (1 2 )T

+ 4E1 E2 h31 h2 + 6E1 E2 h21 h22 + 4E1 E2 h1 h32 + E22 h42

A two material beam with doubly symmetric crosssection

Let us study the thermal deformation of a two material beam with doubly
symmetric cross section with the aim to design a beam without thermal change
in its length.

20

Material 1: E1 , 1
r

Material 2: E2 , 2

The hypothesis is that the displacement, , is constant across the cross-section

and consequently the strain, , is constant along the whole body:
= elastic + thermal =

+ T = a constant
E

The internal normal force, N , is zero as there are not external forces applied:
Z
Z
Z
N=
dA =
1 dA +
2 dA
A

N=

A1

A1

E1 ( 1 T ) dA +

A2

A2

E2 ( 2 T ) dA = 0

Consequently,
=

1 E1 A1 + 2 E2 A2
T
E1 A1 + E2 A2

As there are carbon fibres with a negative coefficient of thermal expansion it is

possible to arrange the dimensions and composition of the beam in such a way
that the fraction vanishes and the beam has a zero thermal expansion.

27

Deformation of loaded beams made of two

parallel parts

Let us consider a beam composed of two parallel beams as shown at the Figure:

21

q(x)

1st beam: E1 , I1
2nd beam: E2 , I2
x

r
v

27.1

Unbound case

First, consider the case of free conection, i.e. the case when the two parts can
freely slice on each orther surface. In this case we can regard it as two Bernoulli
beams with an identical displacements and an additional distributed load as a
result of action and reaction as seen in the following figure.

q(x)

1st beam: E1 , I1

w(x)

w(x)
2nd beam: E2 , I2
r

v
For the two beams we have two equilibrium equations (valid for Bernoullis

22

hypothesis and constant EI along the length of the beam)

E1 I1 v1IV = q w

(1)

E2 I2 v2IV = w

(2)

v1 = v2 = v and, consequently, v1IV = v2IV

(3)

and compatibility conditions

where v1 and v2 are the displacements, E1 and E2 are the Youngs moduli,
and I1 and I2 are the second moments of area of the upper and lower beam,
respectively.

rC1

a1

rC2

a2

C1 is the centre of the cross-sectional area of the

first beam
I1 is the second moment of the first beams area
with respect to the axis a1
C2 is the centre of the cross-sectional area of the
first beam
I2 is the second moment of the first beams area
with respect to the axis a2

Using the equations of equilibrium (5 and 7) in the compatibility condition (3)

gives
qw
w
=
E1 I1
E2 I2
i.e.
w=q

E2 I2
E1 I1 + E2 I2

(4)

As q > 0 w > 0 there is not a gap between the two beams.

Inserting w given by (4) into either (5) or (7) leads to
v IV =

q
(EI)eq

where
(EI)eq = E1 I1 + E2 I2

27.2

Ideally bound case

Now, let us consider the same beam but ideally bounded together. Once more
we assume the Bernoulli hypothesis, = ky, only this time for the whole beam
with one common Neutral axis that is not generally passing through the centroid
of the cross-section.

23

q(x)
b

1st material: E1 , I1 , A1

Neutral axis

2nd material: E2 , I2 , A2

y1 < 0
r

rC1

h1

y2 > 0

rC2

h2

strain

Neutral axis

y1 < 0
r

rC1

h1

y2 > 0

rC2

h2

stress

Neutral axis
1 = E1 = E1 ky
2 = E2 = E2 ky

y1 < 0
r

rC1

h1

y2 > 0

rC2

h2

y
The position of the neutral axis is given by the fact that the resultant axial
force, N , is zero due to the chosen supports:
Z
Z
N=
1 dA +
2 dA = 0
A1

A2

where A1 and A2 is the cross-sectional area of material 1 and 2, respectively.

Thus
Z
Z
E1
y dA + E2
y dA = 0
A1

A2

24

i.e.
E1 Q1 + E2 Q2 = 0
where Q1 and Q2 is the first moment of the cross-sectional area 1 and 2 with
respect to the Neutral axis, respectively. The last equation gives the origin of
coordinate y.
Moment-curvature relationship is based on the expression of the bending
moment as an integral of elementary bending moments
Z
Z
Z
Z
Mb =
y1 dA +
y2 dA = kE1
y 2 dA + kE2
y 2 dA
A1

A2

A1

A2

that is
Mb = k(E1 J1 + E2 J2 )
where
J1 = I1 + y12 A1
is the second moment of the cross-sectional area A1 with respect to the Neutral
axis and
J2 = I2 + y22 A2
is the second moment of the cross-sectional area A2 with respect to the same
Neutral axis.
In the case of small deformations the curvature can be approximated by k = v 00
and the differential equation for deflection is
v 00 =

Mb
(EJ)eq

where the equivalent stiffness

(EJ)eq = E1 J1 + E2 J2 = E1 (I1 + y12 A1 ) + E2 (I2 + y22 A2 )
Using the curvature k in the stress formulas gives
1 = Mb

E1
y
(EJ)eq

2 = Mb

E2
y
(EJ)eq

25

27.3

Elastically bound case

b
M1
NA1
E1 , I1 , A1

N1
NA
NA2

E2 , I2 , A2
r

N2

M2

r
yC1
y1
r
yC2

rC1

h1

rC2

h2

y2

dx

Let us assume that both parts of the beam obey the Bernoullis hypothesis
1 = ky1 and 2 = ky2
where the curvature, k, is the same for both parts as for small deformations
k = v 00 and the deflection, v, is supposed to be the same for both parts. The
coordinates y1 and y2 originates at the two respective neutral axes, NA1 and
NA2, as seen at the Figure above.
The connection between the two parts is assumed to be elastic with the shear
stress at the interface given by
= g(interface
interface
)
1
2
where g is a spring constant and interface
(a = 1, 2) are the respective normal
a
strains at the two parts at the point of the interface.
Let us cut two elements of the length dx. First, from the upper part of the
beam.

dx

N1

N1 + dN1

The equilibrium equation

dN1 bdx = 0

E1 , I1 , A1

where b is a width of the beam at

the interface

The internal normal force, N1 is given by integrating the stress, 1 = E1 1 , over

26

the cross-sectional area of the upper part of the beam, A1

Z
Z
N1 =
1 dA = E1 k
y1 dA = E1 kQ1
A1

A1

where Q1 = yC1 A1 is the first moment of the area A1 with respect to the neutral
axis of the upper part of the beam, NA1 .
Using N1 in the equilibrium equation leads to
d
(kyC1 )E1 A1 bgk(y1interface y2interface ) = 0
dx

(5)

where note that yC1 (as well as yC2 and NA = y1interface y2interface ) is a function
of x, i.e., the neutral axis is not, generally, at the same place at every crosssection.
Now, cut the element of the lower part:

dx

N2

N2 + dN2

The equilibrium equation

E2 , I2 , A2

dN2 + bdx = 0

The two equilibrium equations imply N1 = N2 that is

E1 A1 yC1 + E2 A2 yC2 = 0
Also using
N2 =

2 dA = E2 k

A2

(6)

y2 dA = E2 kQ2

A2

where Q2 = yC2 A2 , in the last equilibrium equation gives

d
(kyC2 )E2 A2 + bgkNA = 0
dx

(7)

As N1 = N2 there is a force couple, N2 NA , adding to the resulting bending

moment
Mb = Mb1 + Mb2 + N2 NA
Using the stress expression above leads to
Z
Z
Mb = E1 k
y12 dA + E2 k
y22 dA + E2 kyC2 A2 NA
A1

A2

27

and
Mb = k(E1 J1 + E2 J2 + E2 yC2 A2 NA )

(8)

where
2
2
J1 = I1 + yC1
A1 and J2 = I2 + yC2
A2

I1 and I2 being the second moments of area with respect to the two principal
central axis.
There is also a geometric condition (as seen in the Figure)
yC1 + C yC2 = NA

(9)

There are five equations, four of them linearly independent, (59) for four unknown functions, k, yC1 , yC2 and NA .

28