Chapter 3

ONE-DIMENSIONAL ELEMENTS
A 3-D beam element of uniform cross section will be studied in this chapter. It is capable
of resisting three force componets and three moment components, as can be seen from
Figure 3.1. Fx is an axial force acting in the direction of centroidal axis. Fy and Fz, are the
shear forces acting in the cross-section plane of beam. Mx is a twisting moment about the
centroidal axis. My and Mz are the bending moments about the two principal axis in the
plane of its cross section. Figure 3.1 shows the displacement and rotation componets
resulted from the forces and moments. Here u, v, w are the translations in the x, y and z
directions, respectively. θx, θy and θz are the rotations about these axis. Therefore, the
stiffness matrix of a frame element will be order of 12 x 12.
My
Mx
Fy
Fx

y Fz
Mz
Mz
x
Fz

Fx Fy z
Mx
My

Figure 3.1 Forces acting on a space frame element.

According to the engineering theory of bending and torsion of beams, the axial
displacements u1 and u2 depend only on the axial forces, and the torsional displacements
θx1 and θx2 depend only on the torsional moments. If the local axes (xyz system) are
chosen to coincide with the principle axes of the cross section, the bending displacements
and forces in the two planes can be considered to be independent of each other.
Displacements for the bending in the xy plane are v1, θz1, v2, θz2, Displacements for the
bending in the xz plane are w1, θy1, w2, θy2
 32

θy2
θx2
v2
u2

w2
θz2
x
θy1 y

θx1
v1 u1

w1
θz1 z

Figure 3.2 The displacements of the space frame element.

Here, we shall chose the local xyz coordinate system to coincide with the principle axes of
the cross section with x-axis representing the centroidal axis of the frame element. Thus
the displacements can be separated into four groups each of which can be considered
independently of others. Then, the finite element formulations can be derived separately
for the groups of displacements given below.

• Axial displacements : axial element (bar or spar element)

• Torsional displacements : torsional element
• Bending displacements in the plane xy : flexural element (beam element)
• Bending displacements in the plane xz : flexural element (beam element)

At the next of this chapter, the stiffness matrices will be derived for the different
independent sets of displacements. Then, the total stiffness matrix of a 3-D beam element
will be obtained by superposition.

Since the procedures for the FE derivations are the same for two- and three-dimensional
problems discussed later in the book, the reader have to make an effort to comprehend the
formulations for these simple cases.

3.1 AXIAL ELEMENT

The axially loaded members in the structures can be modeled by the axial elements. Let us
consider an axial element with two nodes as shown in Fig. 3.3. A slender bar under the
axial load can be considered a one-dimensional problem and the stress, strain,
displacement, and loading depend only on the x variable. That is, the vectors {δ}, {σ}, {ε},
{b}, and {Τ} are now reduced to
 33

q1 1
u(x) 2 q2
x
Figure 3.3 An axial element with two nodes.

{δ} = u(x) {σ} = σx(x) {ε} = εx(x) {b} = b(x) {Τ} = Τ(x) (3.1)

The strain-displacement relations given by Eqs. (2.1) are reduced to

du
εx = (3.2)
dx

In this case, the constitutive relations for a linear elastic material are given by

1
σ x = Eε x εx = σx (3.3)
E

Nodal displacements q1 and q2 consist of translations in the x direction at nodes 1 and 2

(see Fig. 3.3).

{q} = ⎧⎨
q1 ⎫ ⎧u1 ⎫
⎬=⎨ ⎬ (3.4)
⎩q 2 ⎭ ⎩u 2 ⎭

Corresponding nodal forces at points 1 and 2 are

⎧ f1 ⎫ ⎧ f x 1 ⎫
{f } = ⎨ ⎬=⎨ ⎬ (3.5)
⎩ f 2 ⎭ ⎩ f x2 ⎭

We shall assume that the displacement u at any point within the element varies linearly
with x, as follows:

u = c1 + c2x (3.6)

This expression is called a displacement function. Conditions for nodal compatibility are

u = q1 at x = 0 and u = q2 at x = L (3.7)

Applying these conditions we obtain the displacement function in terms of the nodal
displacements as:

q 2 − q1
u = q1 + x (3.8)
L
 34

Equation can be rewritten as

⎢ x x ⎥ ⎧ q1 ⎫
u = ⎢1 − ⎨ ⎬ = ⎣⎢ N ⎦⎥ {q} (3.9)
⎣ L L ⎥⎦ ⎩q2 ⎭

Strain-displacement relationships for the axial element consist of merely one derivative:

du d
εx = = ⎢⎣ N ⎥⎦ {q} = ⎢⎣B ⎥⎦ {q} (3.10)
dx dx

Hence,

d 1
⎢⎣B ⎥⎦ = ⎢⎣ N ⎥⎦ = ⎢⎣ −1 1⎥⎦ (3.11)
dx L

Constitutive (stress-strain) relationships become

σ x = Eε x = E ⎣⎢ B ⎦⎥ {q} (3.12)

Element stiffness matrix can be evaluated as follows

E ⎧− 1⎫ EA ⎡ 1 − 1⎤
L
[k ] = ∫ ⎣B⎦T E ⎣B⎦dV = 2 ⎨ ⎬⎣− 1 1⎦ ∫ ∫ dAdx = (3.13)
V L ⎩1⎭ 0A
L ⎢⎣− 1 1 ⎥⎦

assuming that the cross-sectional area A is constant.

fb1 1 bx(x) 2 fb2 x

L
bx b2
b1

x dx
Figure 3.4 Linearly distributed load on an axial element.
 35

Figure 3.4 shows a linearly varying distributed load bx (force per unit length) that is defined to be:

b2 − b1
bx = b1 + x (3.14)
L

Due to this body force, Eq (2.20) gives equivalent nodal loads as:

⎧ x⎫
L ⎪1 − ⎪ L ⎧2b + b ⎫
L
b −b
{fb } = ∫ ⎣⎢ N ⎦⎥T bx dx = ∫ ⎪⎨ L ⎪⎬ ⎡⎢b1 + 2 1 x ⎤⎥ dx = ⎨ 1 2 ⎬ (3.15)
0 0⎪
x ⎪⎣ L ⎦ 6 ⎩b1 + 2b2 ⎭
⎩⎪ L ⎪⎭

Note that the integration is taken over the beam length since the distributed load is given in
the dimension of force per unit length.

Similarly, equivalent nodal forces due to linearly varying traction force are as follows

⎧ x⎫
L ⎪1 − ⎪ L ⎧2Φ + Φ ⎫
L
Φ −Φ
{fs } = ∫ ⎢⎣ N ⎥⎦T Φ x dx = ∫ ⎪⎨ L ⎪⎬ ⎡⎢Φ1 + 2 1 x ⎤⎥ dx = ⎨ 1 2 ⎬ (3.16)
0 0⎪
x ⎪⎣ L ⎦ 6 ⎩Φ1 + 2Φ 2 ⎭
⎩⎪ L ⎪⎭

On the other hand, if the element were subjected to a uniform temperature change ΔT, the
initial strain due to this influence would be:

{ε0 } = {εT } = α ( ΔT ) (3.17)

Equivalent nodal forces for temperature change are given by

⎧ 1⎫
L ⎪−
L ⎪⎪ E α ΔT Adx = EAα ΔT ⎧−1⎫
L
{fT } = ∫ ∫ ⎣⎢B ⎦⎥T E {ε0 } dAdx = ∫ ⎨⎪ ⎬ { ( )} ( )⎨ ⎬ (3.18)
0A 0⎪
1 ⎪ ⎩1⎭
⎩⎪ L ⎭⎪

If the temperature increases, these equivalent nodal loads act in the negative direction at
node 1 and in the positive direction at node 2.

Example 3.1 The bar shown in Figure 3.1 is fixed from its one end and carries a 1 kN
load at its free end in addition its own weight. E = 200 GPa, ρ = 7800 kg/m3.
(a) Model the bar with two axial elements.
(b) Determine the nodal displacements
(c) Calculate the stress for each element
(d) Determine the reaction force at the support.
 36

160 mm

1 300 mm
140 mm

600 mm

2 300 mm
100 mm
80 mm

1 kN 1 kN

Figure E3.1
Solution:
(a) We can model the bar with two axial elements of equal width, 300 mm. The average
diameter of element one is 140 mm, and the average diameter of element two is 100 mm.

1 1 2 2 3 1 kN
Q2 Q3
Q1 = 0

(b) The element stiffness matrices can be written as

3 1 −1⎤
⎡k 1 ⎤ = 200 x10 x15394 ⎡⎢ ⎡ 1.96 −1.96 ⎤
⎥ = 5236 x103 ⎢ ⎥
⎣ ⎦ 300 ⎣ −1 1 ⎦ ⎣ −1.96 1.96 ⎦

3 1 −1⎤
⎡k 2 ⎤ = 200 x10 x7854 ⎢⎡ ⎡ 1 −1⎤
⎥ = 5236 x103 ⎢ ⎥
⎣ ⎦ 300 ⎣ −1 1 ⎦ ⎣ −1 1 ⎦

The body force, or specific weight for this problem, is b = ρg = 7800x10-9x9.81 =

7.652x10-5 N/mm3. Hence, the element nodal force vectors due to gravity can be calculated
as

7.652 x10−5 x15394 x300 ⎧1⎫ ⎧176.7 ⎫

{ }
f1 =
2
⎨ ⎬=⎨ ⎬ N
⎩1⎭ ⎩176.7 ⎭

7.652 x10−5 x7854 x300 ⎧1⎫ ⎧90.2 ⎫

{f }
2
=
2
⎨ ⎬=⎨ ⎬ N
⎩1⎭ ⎩90.2 ⎭
 37

Now, we can assemble the global stiffness matrix and load vector as

⎡ 1.96 −1.96 0 ⎤
[K ] = 5236 x103 ⎢ −1.96 1.96 + 1 −1⎥⎥
⎢
⎢⎣ 0 −1 1 ⎥⎦

⎧ 176.7 ⎫
{F} = ⎪⎨176.7 + 90.2⎪⎬
⎪ 90.2 + 1000 ⎪
⎩ ⎭

In the elimination approach, the stiffness matrix [K] is modified by deleting rows and
columns corresponding to fixed nodes. In this problem, dof 1 is fixed. Thus, we delete the
first row and column of the stiffness matrix, and the first component of the force vector.
Now, the reduced equation system can be solved.

⎡ 2.96 −1⎤ ⎧Q2 ⎫ ⎧ 266.9 ⎫

5236 x103 ⎢ ⎥⎨ ⎬= ⎨ ⎬
⎣ −1 1 ⎦ ⎩Q3 ⎭ ⎩1090.2 ⎭

Solution of the above equations yields

Q2 = 1.3224x10-4 mm

Q3 = 3.4045x10-4 mm

(c) The stresses at each element can be obtained

⎧ 0 ⎫⎪
[ −1 1] ⎪⎨
1
σ1 = 200 x103 −4 ⎬ = 0.0882 N / mm
2
300 ⎩⎪1.3224 x10 ⎭⎪

1 ⎧⎪1.3224 x10−4 ⎫⎪
σ 2 = 200 x103 [ − 1 1]⎨ −
⎬ = 0.1388 N / mm
2
300 4
⎩⎪3.4045 x10 ⎭⎪

(d) In order to obtain the reaction force we use the eliminated equation.

⎧ 0 ⎫
⎪ −4 ⎪
R1 = 5236 x10 [1.96 −1.96 0] ⎨1.3224 x10 ⎬ − 176.7 = −1533.8 N
3

⎪ −4 ⎪
⎩3.4045 x10 ⎭

Example 3.2 A composite rod are made from aluminum and steel parts. There is a 2mm
gap between the free end of the rod and the rigid wall at 20°C . Then, the temperature of
the rod raised 50 °C and an axial load P = 50 kN is applied at the connection section of the
two parts.
 38

Properties Aluminum Steel

Young Modulus, E (Gpa) 70 200
Thermal Expansion Coeff. α 23x10-6 per °C 11.7x10-6 per °C
Cross-sectional area, A (mm2) 1200 800

ALUMINUM STEEL
P

150 mm
0.12 mm (before applying the
300 mm 200 mm

Figure E3.2a

Solution
Consider a three elements finite element model in Fig. E3.2b.

1 1 2 2 3 3 4

Q1 = 0 Q2 Q3 Q4

Figure E3.2b

Table of nodal coordinates is given below

Node number Coordinate, x (mm)

1 0
2 150
3 300
4 500

Element connectivity table is

Element number Node, 1 Node, 2

1 1 2
2 2 3
3 3 4

We can write down the element stiffness matrices for three elements as

1 2
− 1⎤ 1
[k ] = 70x10150x1200 ⎡⎢−11
3
1
N/mm
⎣ 1 ⎥⎦ 2

2 3
 39

− 1⎤ 2
[k ] = 70x10150x1200 ⎡⎢−11
3
2
N/mm
⎣ 1 ⎥⎦ 3

3 4
− 1⎤ 3
[k ] = 200x10 x800 ⎡ 1
3
3
⎢− 1 N/mm
200 ⎣ 1 ⎥⎦ 4

The global stiffness matrix [K] is assembled from the element stiffness matrices

1 2 3 4
⎡ 56 −56 0 0 ⎤ 1 ⎡ 56 −56 0 0 ⎤
⎢ −56 56 + 56 −56 0 ⎥⎥ ⎢ 0 ⎥⎥
2 4 ⎢ −56 112 −56
[K ] = 10 ⎢
4⎢
= 10
0 −56 56 + 80 −80 ⎥ 3 ⎢ 0 −56 136 −80 ⎥
⎢ ⎥ ⎢ ⎥
⎣ 0 0 −80 80 ⎦ 4 ⎣ 0 0 −80 80 ⎦

The element temperature forces due to ΔT = 30 are obtained as

⎧− 1⎫ ⎧− 57960⎫1
{f } = 70x10
1
T
3
x1200x 23x10 −6 x30⎨ ⎬ = ⎨ ⎬ N
⎩ 1 ⎭ ⎩ 57960 ⎭2

⎧− 1⎫ ⎧− 57960⎫ 2
{f } = 70x10
2
T
3
x1200 x 23x10 −6 x 40⎨ ⎬ = ⎨ ⎬ N
⎩ 1 ⎭ ⎩ 57960 ⎭ 3

⎧− 1⎫ ⎧− 56160⎫ 3
{f } = 200x10
3
T
3
x800 x11.7 x10 −6 x 40⎨ ⎬ = ⎨ ⎬ N
⎩ 1 ⎭ ⎩ 56160 ⎭ 4

Global force vector is assembled from the element force vectors and the concentrated load.

⎧ − 57960 ⎫ ⎧− 57960⎫
⎪57960 − 57960 − 50000⎪ ⎪− 50000⎪
{F} = ⎪⎨ ⎪ ⎪
⎬=⎨
⎪
⎬ N
⎪ 57960 − 56160 ⎪ ⎪ 1800 ⎪
⎪⎩ 56160 ⎪⎭ ⎪⎩ 56160 ⎪⎭

In this problem, we should firs determine whether contact occurs between the bar and the
wall. Hence, let us assume that there is no rigid wall at the right hand side of the beam and
solve the problem.

⎡ 56 − 56 0 0 ⎤⎧Q1 ⎫ ⎧− 57960⎫
⎢− 56 112 − 56 0 ⎥⎪Q ⎪ ⎪− 50000⎪
104 ⎢ ⎥⎪⎨ 2 ⎪⎬ = ⎪⎨ ⎪
⎬
⎢ 0 − 56 136 − 80⎥⎪Q3 ⎪ ⎪ 1800 ⎪
⎢ ⎥
⎣ 0 0 − 80 80 ⎦⎪⎩Q4 ⎪⎭ ⎪⎩ 56160 ⎪⎭
 40

Now, the boundary condition, Q1 = 0, is applied on the matrix equation by means of the
elimination approach. We cancel the first equation of the linear algebraic equation system
and the first column of the stiffness matrix. The system of retained equations can be
written as

⎡112 − 56 0 ⎤⎧Q2 ⎫ ⎧− 50000⎫

⎪ ⎪ ⎪ ⎪
10 ⎢⎢− 56 136 − 80⎥⎥⎨Q3 ⎬ = ⎨ 1800 ⎬
4

⎢⎣ 0 − 80 80 ⎥⎦⎪⎩Q4 ⎪⎭ ⎪⎩ 56160 ⎪⎭

Q4 = 0.1879 mm

From this result, we see that contact does occur. The problem has to be re-solved, since the
boundary conditions are now different:

Q1 = 0, Q2 = 0.12 mm

⎡ 56 − 56 0 0 ⎤⎧Q1 ⎫ ⎧− 57960⎫
⎢− 56 112 − 56 0 ⎥⎪Q ⎪ ⎪− 50000⎪
104 ⎢ ⎥⎪⎨ 2 ⎪⎬ = ⎪⎨ ⎪
⎬
⎢ 0 − 56 136 − 80⎥⎪Q3 ⎪ ⎪ 1800 ⎪
⎢ ⎥
⎣ 0 0 − 80 80 ⎦⎪⎩Q4 ⎪⎭ ⎪⎩ 56160 ⎪⎭

Let’s apply the elimination method to account the boundary conditions. The first equation
is eliminated and the first column of the stiffness matrix is cancelled since Q1 = 0. The last
equation is also eliminated and the last column of the stiffness matrix is cancelled since Q4
is prescribed as 1.2 mm. However, the contributions from the last column have to be
accounted since the corresponding displacement is different from zero.

⎡112 −56⎤ ⎧Q2 ⎫ ⎧ −50000 − 0.12x0 ⎫

104 ⎢
−56 136 ⎥ ⎨Q ⎬ = ⎨1800 − 0.12x(−80x104 )⎬
⎣ ⎦⎩ 3⎭ ⎩ ⎭

The solution is

Q2 = -0.01094 mm and Q3 = 0.0.06741 mm

Now we can obtain the stress in each element

{σ} = E({ε} − {ε 0 }) = E(⎣B⎦{q} − αΔT )

1⎛ ⎧ 0 ⎫ ⎞
{σ } = 70 x10 ⎜ 150
1 3
⎢⎣ −1 1⎥⎦ ⎨ −6
⎬ − 23x10 x30 ⎟ = −53.4 MPa
⎝ ⎩−0.01094 ⎭ ⎠

1⎛ ⎧−0.01094 ⎫ ⎞
{σ } = 70 x10 ⎜ 150
2 3
⎣⎢ −1 1⎦⎥ ⎨ −6
⎬ − 23x10 x30 ⎟ = −11.7 MPa
⎝ ⎩ 0.06741 ⎭ ⎠
 41

1 ⎛ ⎧0.06741⎫ ⎞
{σ } = 200 x10 ⎜ 200
3 3
⎢⎣ −1 1⎥⎦ ⎨ −6
⎬ − 11.7 x10 x30 ⎟ = −17.6 MPa
⎝ ⎩ 0.12 ⎭ ⎠

The reaction forces are

⎧ 0 ⎫
⎧ R1 ⎫ 4 ⎡56 −56 0 0 ⎤ ⎪−0.01094⎪⎪ ⎧−57960⎫ ⎧ 64086 ⎫
⎪
⎨ ⎬ = 10 ⎢0 ⎨ ⎬−⎨ ⎬=⎨ ⎬ N
⎩ R4 ⎭ ⎣ 0 −80 80⎥⎦ ⎪ 0.06741 ⎪ ⎩ 56160 ⎭ ⎩−14088⎭
⎪⎩ 0.12 ⎪⎭

3.2 TORSIONAL ELEMENT

Figure 3.5 depicts a torsional element that could be the shaft in a machine or a member in a
grid. It has a single generic displacement θx, which is a small rotation about x axis.

{δ} = θx(x) {σ} = τ(x) {ε} = γ(x) {b} = m(x) (3.19)

q1 θx(x) q2
x
Figure 3.5 A torsional element with two nodes.

Furthermore, the stress-strain and strain-displacement relations are

dθ x
τ = Gγ γ = r (3.20)
dx

Nodal displacements q1 and q2 consist of small axial rotations about the x axis at nodes 1
and 2 (see Fig. 1.1a).

q 1 ⎫ ⎧ θ x1 ⎫
{q} = ⎧⎨ ⎬=⎨ ⎬ (3.21)
⎩q 2 ⎭ ⎩θ x 2 ⎭

Corresponding nodal actions at points 1 and 2 are

⎧ f 1 ⎫ ⎧ M x1 ⎫
{f } = ⎨ ⎬=⎨ ⎬ (3.22)
⎩f 2 ⎭ ⎩M x 2 ⎭
 42

We shall assume that the displacement θx at any point within the element varies linearly
with x, as follows:

θx = c1 + c2x (3.23)

This expression is called a displacement function. Conditions for nodal compatibility are

θx = q1 at x = 0
θx = q2 at x = L

Applying these conditions we obtain the displacement function in terms of the nodal
displacements as:

q 2 − q1
θ x = q1 + x (3.24)
L

Equation can be rewritten as

⎢ x x ⎥ ⎧ q1 ⎫
θ x = ⎢1 − ⎨ ⎬ = ⎢⎣ N ⎥⎦ {q} (3.25)
⎣ L L ⎥⎦ ⎩q2 ⎭

Strain-displacement relationships for the axial element consist of merely one derivative:

d d
γ =r θ x = r ⎢⎣ N ⎥⎦ {q} = ⎢⎣B ⎥⎦ {q} (3.26)
dx dx

Hence,
d ⎣N ⎦ r
⎣B⎦ = [d ]⎣N ⎦ = r = ⎣− 1 1⎦ (3.27)
dx L

Constitutive (stress-strain) relationships become

τ = Gγ = G ⎣⎢ B ⎦⎥ {q} (3.28)

Element stiffness matrix can be evaluated as follows

⎡− 1⎤ GJ ⎡ 1 − 1⎤
L 2πR 2
[k ] = ∫ [B]T [E][B]dV = ∫ ∫ ∫ Gr2 ⎢ 1 ⎥[− 1 1]rdrdθdx = L ⎢− 1 1 ⎥ (3.29)
V 0 0 0 L ⎣ ⎦ ⎣ ⎦

where GJ is constant. The polar moment of inertia J is defined as:

2πR
J= ∫ ∫ r drdθ
3
(3.30)
0 0

For noncircular cross sections the polar moment of inertia is replaced by a torsion constant.
 43

Equivalent nodal loads due to a linearly distributed axial torque (per unit length), defined
as:

m x 2 − m x1
m x = m x1 + x (3.31)
L

Due to this influence, Eq. (2.20) gives equivalent nodal loads as:

⎧ x⎫
L
⎪1 − L ⎪⎡
L
m −m ⎤ L ⎧2 m x 1 + m x 2 ⎫
{f b } = ∫ [N] m x dx = ∫ ⎨ x ⎬⎢m x 1 + x 2 x 1
T
x ⎥ dx = ⎨ ⎬ (3.32)
0 0⎪ ⎪⎣ L ⎦ 6 ⎩m x 1 + 2 m x 2 ⎭
⎩ L ⎭

Equivalent nodal loads due to a concentrated axial torque Mx at the location x = L/6
become

⎧ 1⎫
⎪1 − 6 ⎪ M x ⎧5⎫
{f b } = M x [N T ]x =L / 6 = Mx ⎨
1 ⎬
= ⎨ ⎬ (3.33)
⎪ ⎪ 6 ⎩1⎭
⎩ 6 ⎭

In this case the result is obtained by evaluating the displacement shape functions at the
point x = L/6 and no integration is required.

3.3 FLEXURAL ELEMENTS

3.3.1 Bending Displacements in the xy plane

Figure 3.6 shows a straight flexural element, for which the x-y plane is a principal plane of
bending. Indicated in the figure is a single generic displacement v, which is a translation in
the y direction.

{δ} = v(x) {σ} = σx(x) {ε}= εx(x) {b} = by(x) (3.34)

y
q1 v(x) q3
2 x
1

q2 x
q4
z
 44

Figure 3.6 A flexural element with two nodes for bending in the xy plane.

Furthermore, the stress-strain and strain-displacement relations are

du
σx = Eεx εx = (3.35)
dx

Strain-displacement relationships can be developed for the flexural element if we assume

that plane sections remain plane during deformation, as indicated in Fig. 3.7. The
translation u in the x direction at any point on the cross section is:

dv
u = −y (3.36)
dx

Using this relationship, we obtain the following expression for flexural strain

d 2v
εx = −y = − yψ = [d ]{δ} (3.37)
dx 2

d2
in which ψ represents the curvature. [d ] = − y and {δ} = v
dx 2

B
y x
A

dv
y, v dx x, u
dv
u = −y
dx
dv
B′ dx

A′
v(x)

Figure 3.7 Deflection of a beam.

 45

At node 1 nodal displacements q1 and q2 are a translation in the y direction and a small
rotation in the z sense. Similarly, at node 2 the displacements 3 and 4 are a translation and
a small rotation respectively. Therefore, the vector of nodal displacements becomes:

⎧ q 1 ⎫ ⎧ v1 ⎫
⎪q ⎪ ⎪ θ ⎪
{q} = ⎪⎨ 2 ⎪⎬ = ⎪⎨ z1 ⎪⎬ in which θ z1 = dv1 θz2 =
dv 2
(3.38)
⎪q 3 ⎪ ⎪ v 2 ⎪ dx dx
⎪⎩q 4 ⎪⎭ ⎪⎩θ z 2 ⎪⎭

Corresponding nodal actions at points 1 and 2 are:

⎧ f1 ⎫ ⎧ f x1 ⎫
⎪ f ⎪ ⎪m ⎪
{ f } = ⎪⎨ 2 ⎪⎬ = ⎪⎨ z1 ⎪⎬ (3.39)
⎪ f 3 ⎪ ⎪ f x2 ⎪
⎪⎩ f 4 ⎪⎭ ⎪⎩m z 2 ⎪⎭

The terms fx1 and fx2 denote forces in the y direction at nodes 1 and 2, and the symbols mx1
and mx2 represents moments in the z sense at those points.

Because there are four nodal displacements, a complete cubic displacement function may
be assumed for the flexural element, as follows:

v = c1 + c2x + c3x2 + c4x3 (3.40)

Conditions for nodal compatibility are

dv
v = q1 and = q2 at x = 0
dx
dv
v = q2 and = q4 at x = L
dx

Applying these conditions we obtain the displacement function in terms of the nodal
displacements as:

⎧q1 ⎫
⎪q ⎪
⎪ ⎪
{δ} = v = ⎣N1 N2 N3 N 4 ⎦⎨ 2 ⎬ = ⎣N ⎦{q} (3.41)
⎪q 3 ⎪
⎪⎩q 4 ⎪⎭

where Ni, i =1,2,3,4 are the shape functions.

N1 =
1
L3
[
2 x 3 − 3 x 2 L + L3 ]
 46

N2 =
1 3
L3
[
x L − 2 x 2 L2 + xL3 ] (3.42)
1
N 3 = 3 ⎡⎣ −2 x 3 + 3x 2 L ⎤⎦
L
1
[
N 4 = 3 x 3 L − x 2 L2
L
]
Strain-displacement matrix [B] can be obtained as

y
3 ⎣
⎣B⎦ = ⎣d ⎦⎣N ⎦ = − 12 x − 6L 6 xL − 4L2 − 12 x + 6L 6 xl − 2L2 ⎦ (3.43)
L

Constitutive (stress-strain) relationships become

σ x = Eε x = EBq (3.44)

Element stiffness matrix can be evaluated as follows

∫ ⎣⎢B ⎦⎥
T
k= E ⎣⎢B ⎦⎥ dV
Ve
⎧ 12 x − 6 L ⎫
⎪ 2⎪ (3.45)
Ey 2 ⎪ 6 xL − 4 L ⎪ ⎢
L
2⎥
=∫∫ 6 ⎨ 2
⎬ ⎣12 x − 6 L 6 xL − 4 L −12 x + 6 L 6 xl − 2 L ⎦ dAdx
0A L ⎪ −12 x + 6 L ⎪
⎪ 6 xL − 2 L2 ⎪
⎩ ⎭
Assuming that the bending rigidity EI is constant, multiplication and integration yield

⎡ 12 6L − 12 6L ⎤
⎢ 6L 4L2 − 6L 2L2 ⎥
[k ] = EI3z ⎢ ⎥ (3.46)
L ⎢− 12 − 6L 12 − 6L⎥
⎢ ⎥
⎣ 6L 2L
2
− 6L 4L2 ⎦

where I z = ∫ y 2 dA represents the moment of inertia of the cross section with respect to
A
the neutral axis.

Equivalent nodal loads due to a uniformly distributed force by per unit length may be
calculated as

⎧ N1 ⎫ ⎧ 6 ⎫
L L ⎪ ⎪ b y L ⎪⎪ L ⎪⎪
⎪N 2 ⎪
f b = ∫ ⎣N ⎦ b y dx = ∫ ⎨ ⎬b y dx =
T
⎨ ⎬ (3.47)
0
N
0 ⎪ 3⎪
12 ⎪ 6 ⎪
⎪⎩ N 4 ⎪⎭ ⎪⎩− L⎪⎭
 47

y b2
b(x) = b2 x/L
fb1
fb3 x
1
fb2 x 2
fb4
z
Figure 3.8 A linearly varying distributed load.

Figure 3.8 shows a linearly varying distributed load bx (force per unit length) that is
defined to be:

b2
bx = x (3.48)
L
Due to this body force, Eq. (2.30a) gives equivalent nodal loads as:

⎧ 9 ⎫
⎪ ⎪
b2 L ⎪ 2 L ⎪
L
T b2
{fb } = ∫ ⎢⎣ N ⎥⎦ xdx = ⎨ ⎬ (3.49)
0
L 60 ⎪ 21 ⎪
⎪⎩−3L ⎪⎭

Assume that the flexural element is subjected to a linear variation of temperature from ΔT1
at its lower surface to ΔT2 at its upper surface, as indicated in Fig 3.9. If ΔT1 > ΔT2 and
the depth of the element is d, the temperature change at any point in the element is:

1
ΔT = (ΔT1 + ΔT2 ) − y (ΔT1 − ΔT2 ) (3.50)
2 d

ΔT2
d

1 e 2

ΔT1
Figure 3.9 A flexural element subjected to a linear variation of temperature.
 48

The first term in Eq. (3.50) causes axial strains that are not to be considered here.
However, the second term produces flexural strains because it varies linearly with y. Thus,

y
{ε 0 } = {ε T } = ε xT = −α (ΔT1 − ΔT2 ) (3.51)
d

Using this equation in Eq. (2.41), we find:

L
{f T } = − ∫ ∫ ⎣B⎦T Eα y (ΔT1 − ΔT2 )dAdx (3.52)
0A
d

After the matrix B from Eq. (3.43) is substituted and the integrations are performed for
constant EI, this expression yield:

⎧0⎫
⎪− 1⎪
{f T } = αEI (ΔT1 − ΔT2 )⎪⎨ ⎪⎬ (3.53)
d ⎪0⎪
⎪⎩ 1 ⎪⎭

The equivalent nodal loads in Eq. (3.53) consist of zero forces and equal and opposite
moments at the two ends.

Example 3.3 A rolled steel beam and loading shown in the Figure E3.3.
a) Model the beam with two finite elements.
b) Solve the nodal displacements.
c) Determine the vertical deflection at the midpoint of the distributed load.
d) Recover the maximum strain and stress at this point.
e) Calculate the reaction forces and moment.

Take E = 200 GPa.

40 kN/m 60 kN

2m 1m
W 250x28
Iz = 39.2 x10-6 m4
Iy = 1.77 x10-6 m4
d = 260 mm

Figure E3.3
 49

Solution:
(a) Finite element model without loading.
Q1 Q5
Q3
Q2 1 2
Q6
Q4
3
1 2

Element stiffness matrices are

⎡ 12 6 x2 − 12 6 x2 ⎤ ⎡3 3 −3 3 ⎤
⎢ 6 x2 4 x2 2 − 6 x2 2 x2 2 ⎥ ⎢3 4 − 3 2 ⎥⎥
[k ] = 200 x10
9 −6
1 x39.2 x10 ⎢ ⎥ = 39.2 x10 5 ⎢
23 ⎢− 12 − 6 x 2 12 − 6 x 2⎥ ⎢− 3 − 3 3 − 3⎥
⎢ 2 ⎥ ⎢ ⎥
⎣ 6 x2 2 x2
2
− 6 x2 4 x2 ⎦ ⎣3 2 −3 4 ⎦

⎡ 12 6 x1 − 12 6 x1 ⎤ ⎡ 24 12 − 24 12 ⎤
⎢ 6 x1 4 x12 − 6 x1 2 x12 ⎥ ⎢ 12 − 12 4 ⎥⎥
[k ] = 200 x10 x39.2 x10 −6 8
9
2 ⎢ ⎥ = 39.2 x10 5 ⎢
13 ⎢− 12 − 6 x1 12 − 6 x1⎥ ⎢− 24 − 12 24 − 12⎥
⎢ 2 ⎥ ⎢ ⎥
⎣ 6 x1 2 x1 − 6 x1 4 x1 ⎦ − 12
2
⎣ 12 4 8 ⎦

If the assembly process is applied, the structural stiffness matrix can be obtained as

⎡3 3 −3 3 0 0 ⎤
⎢3
⎢ 4 −3 2 0 0 ⎥⎥
⎢− 3 − 3 27 11 − 24 12 ⎥
[K ] = 39.2 x10 5 ⎢ ⎥
⎢3 2 11 12 − 12 4 ⎥
⎢0 0 − 24 − 12 24 − 12⎥
⎢ ⎥
⎣⎢ 0 0 12 4 − 12 8 ⎥⎦

The element load vector can be calculated as

⎧ 6 ⎫ ⎧6⎫ ⎧3⎫
b y L ⎪ L ⎪ − 40 x10 3 x 2 ⎪⎪ 2 ⎪⎪ − 40 x10 3
⎪ ⎪ ⎪1⎪
⎪ ⎪
{f b } = ⎨ ⎬= ⎨ ⎬= ⎨ ⎬
12 ⎪ 6 ⎪ 12 ⎪6⎪ 3 ⎪3⎪
⎪⎩− L ⎪⎭ ⎪⎩− 2⎪⎭ ⎪⎩− 1⎪⎭

The system of linear algebraic equations is given by

 50

⎡3 3 −3 3 0 0 ⎤ ⎧Q1 ⎫ ⎧ − 40 ⎫
⎢3
⎢ 4 −3 2 0 0 ⎥⎥ ⎪⎪Q2 ⎪⎪ ⎪⎪− 40 / 3⎪⎪
⎢− 3 − 3 27 11 − 24 12 ⎥ ⎪⎪Q3 ⎪⎪ ⎪⎪ − 40 ⎪⎪ 3
39.2 x10 5 ⎢ ⎥⎨ ⎬ = ⎨ ⎬ x10
⎢3 2 11 12 − 12 4 ⎥ ⎪Q4 ⎪ ⎪ 40 / 3 ⎪
⎢0 0 − 24 − 12 24 − 12⎥ ⎪Q5 ⎪ ⎪ − 60 ⎪
⎢ ⎥⎪ ⎪ ⎪ ⎪
⎢⎣ 0 0 12 4 − 12 8 ⎦⎥ ⎪⎩Q6 ⎪⎭ ⎪⎩ 0 ⎪⎭

Boundary conditions can be stated as

Q1 = Q2 = Q3 = 0

After alimination the global stiffness matrix can be obtained as

⎡ 12 − 12 4 ⎤ ⎧Q4 ⎫ ⎧40 / 3⎫
⎪ ⎪ ⎪ ⎪
39.2 x10 ⎢− 12 24 − 12⎥⎥ ⎨Q5 ⎬ = ⎨ − 60 ⎬ x10 3
⎢5

⎢⎣ 4 − 12 8 ⎥⎦ ⎪⎩Q6 ⎪⎭ ⎪⎩ 0 ⎪⎭

The solution is

Q4 = -0.0029762 rad, Q5 = -0.0055272 m, Q6 = -0.0068027 rad

The deflection at the midpoint of element 2 can be determined by using displacement

function.

⎧ q1 ⎫
⎪q ⎪
⎪ ⎪
{δ} = v = ⎣⎢ N1 N2 N3 N 4 ⎦⎥ ⎨ 2 ⎬
⎪ q3 ⎪
⎪⎩q4 ⎪⎭
⎧ q1 ⎫
⎪ ⎪
1 ⎢ 3 2 2 ⎥ ⎪ q2 ⎪
= 3⎣
2 x − 3 x 2 L + L3 x3 L − 2 x 2 L2 + xL3 −2 x 3 − 3 x 2 L 3
x L− x L ⎨ ⎬
⎦ q3
L ⎪ ⎪
⎪⎩q4 ⎪⎭
⎧ 0 ⎫
⎪ 0 ⎪
1 ⎢ 3 2 2⎥⎪ ⎪
= 3 ⎣
2.1 − 3.12.2 + 23 13.2 − 2.12.22 + 1.23 3 2 3
−2.1 − 3.1 .2 1 .2 − 1 .2 ⎨ ⎬
2 ⎦ 0
⎪ ⎪
⎪⎩−0.0029762 ⎪⎭

= 7.44 x10−4 m

Maximum strain and stress occur at the maximum distance from the neutral surface.
 51

⎧ q1 ⎫
⎪q ⎪
y ⎪ ⎪
{ε } = ε x = ⎢⎣B ⎥⎦ {q} = − 3 ⎢⎣12 x − 6 L 6 xL − 4 L2 −12 x + 6 L 6 xL − 2 L2 ⎥⎦ ⎨ 2 ⎬
L ⎪ q3 ⎪
⎪⎩q4 ⎪⎭
⎧ 0 ⎫
⎪ 0 ⎪
y ⎢ 2 2⎥⎪ ⎪
= − 3 12.1 − 6.2 6.1.2 − 4.2 −12.1 + 6.2 6.1.2 − 2.2 ⎨ ⎬
2 ⎣ ⎦ 0
⎪ ⎪
⎪⎩−0.0029762 ⎪⎭

= 1.4881 x 10−3 y
(ε x )max = 1.4881x10−3 ymax = 1.4881 x 10−3 x 0.13 = 1.93453 x 10−4

(σ x )max = E (ε x )max = 200 x109 x1.93453 x 10−4 = 38.69 MPa

The support reactions can determined as

⎧ 0 ⎫
⎪ 0 ⎪
⎧ 1⎫
R ⎡ 3 3 − 3 3 0 0 ⎤ ⎪ ⎪ ⎧ − 40 ⎫ ⎧ 5000 ⎫
⎪ ⎪ 5⎢ ⎥ ⎪⎪ 0 ⎪⎪ ⎪ ⎪ 3 ⎪ ⎪
⎨ R2 ⎬ = 39.2 x10 ⎢ 3 4 −3 2 0 0 ⎥⎨ ⎬ − ⎨− 40 / 3⎬ x10 = ⎨− 10000⎬
− 0.0029762⎪ ⎪
⎪R ⎪ ⎢⎣− 3 − 3 27 11 − 24 12⎥⎦ ⎪ − 40 ⎪⎭ ⎪135000 ⎪
⎩ 3⎭ ⎪− 0.0055272⎪ ⎩ ⎩ ⎭
⎪ ⎪
⎪⎩− 0.0068027⎪⎭

3.3.2 Bending Displacements in the xz plane

Here bending of the element takes place in the xz plane instead of xy plane. Figure (3.10)
shows a straight flexural element, for which the x-z plane is a principal plane of bending.

y q2
q4

2 x
1
q1 w(x)
q3
z x
Figure 3.10 A flexural element with two nodes for bending in the xz plane.
 52

Indicated in the figure is a single generic displacement w, which is a translation in the z

direction.

{δ} = w(x) {σ} = σx(x) {ε}= εx(x) {b} = bz(x) (3.54)

Furthermore, the stress-strain and strain-displacement relations are

σx = Eεx

du
εx =
dx

Strain-displacement relationships can be developed for the flexural element if we assume

that plane sections remain plane during deformation, as indicated in Fig. (3.7) The
translation u in the x direction at any point on the cross section is:

dw
u = −z (3.55)
dx

Using this relationship, we obtain the following expression for flexural strain

d 2w
ε x = −z = − zκ = [d ]{δ} (3.56)
dx 2
d2
in which κ represents the curvature. [d ] = − z and {δ} = w
dx 2
At node 1 nodal displacements q1 and q2 are a translation in the z direction and a small
rotation in the y sense. Similarly, at node 2 the displacements 3 and 4 are a translation and
a small rotation respectively. Therefore, the vector of nodal displacements becomes:

⎧ q1 ⎫ ⎧ w1 ⎫
⎪q ⎪ ⎪θ ⎪
{q} = ⎪⎨ 2 ⎪⎬ = ⎪⎨ y1 ⎪⎬ in which θ y1 = dw 1 θz2 =
dw 2
(3.57)
⎪ q 3 ⎪ ⎪ w2 ⎪ dx dx
⎪⎩q 4 ⎪⎭ ⎪⎩θ y 2 ⎪⎭

Corresponding nodal actions at points 1 and 2 are:

⎧ f1 ⎫ ⎧ f z1 ⎫
⎪f ⎪ ⎪ m ⎪
{f } = ⎪⎨ 2 ⎪⎬ = ⎪⎨ y1 ⎪⎬ (3.58)
⎪f 3 ⎪ ⎪ f z 2 ⎪
⎪⎩f 4 ⎪⎭ ⎪⎩m y 2 ⎪⎭

The terms fz1 and fz2 denote forces in the z direction at nodes 1 and 2, and the symbols my1
and my2 represents moments about y axis at those points.

By proceeding as in the case of bending in the plane xy, we can derive the stiffness matrix
as
 53

⎡ 12 6L − 12 6L ⎤
EI 6L 4L2 − 6L 2L2 ⎥
⎢
[k ] = 3y ⎢ ⎥ (3.59)
L ⎢− 12 − 6L 12 − 6L⎥
⎢ 2 ⎥
⎣ 6L 2L − 6L 4L ⎦
2

where Iy denotes the area moment of inertia of the cross section of the element about y
axis.

3.4 TOTAL ELEMENT STIFFNESS MATRIX

The stiffness matrices derived for different sets of independent displacements can now be
assembled to obtain the overall stiffness matrix of the frame element.
q5 q11
q2 q8
q4 q1 q10
q7
q6 q12
q6 q12 q8
q2
q1 q7 ≡
q6 q12
q5 q11
q4 q10
B

q3 q9
 54

q1 q2 q3 q4 q5 q6 q7 q8 q9 q10 q 11 q12
⎡ EA ⎤
⎢ L ⎥
⎢ 12EI z ⎥
⎢ 0 ⎥
⎢ L3 ⎥
⎢ 12EI y ⎥
⎢ 0 0
L3 ⎥
⎢ GJ ⎥
⎢ 0 0 0 ⎥
⎢ L ⎥
⎢ − 6EI y 4EI y ⎥
⎢ 0 0 0 ⎥
⎢ L2 L ⎥
⎢ 0 6EI z 4EI z ⎥
0 0 0
[k ] = ⎢⎢ EA L2 L
EA
⎥
⎥
(3.60)

⎢− L 0 0 0 0 0
L ⎥
⎢ − 12EI z − 6EI z 12EI z ⎥
⎢ 0 0 0 0 0 ⎥
⎢ L3 L2 L3 ⎥
⎢ − 12EI y 6EI y 12EI y ⎥
⎢ 0 0 0 0 0 0 ⎥
⎢ L3 L2 L3 ⎥
⎢ 0 − GJ GJ ⎥
0 0 0 0 0 0 0
⎢ L L ⎥
⎢ − 6EI y 2EI y 6EI y 4EI y ⎥
⎢ 0 0 0 0 0 0 0 ⎥
⎢ L2 L L2 L ⎥
⎢ 0 6EI z 2EI z − 6EI z 4EI z ⎥
0 0 0 0 0 0 0
⎢⎣ L2 L L2 L ⎥⎦