Beams and Frames

beam theory can be used to solve simple

beams
complex beams with many cross section
changes are solvable but lengthy
many 2-d and 3-d frame structures are
better modeled by beam theory

One Dimensional Problems

The geometry of the problem is three dimensional, but the
variation of variable is one dimensional
y
u(x)
Contract

Stretch

kx du + P
(x=l)=0
dx
v(x)

4v
d
kx
+ P(x)=0
4
dx

u(x)
px

y
px

x v(x)

Variable can be scalar field like

temperature, or vector field like
displacement.
For dynamic loading, the variable can
be time dependent

Element Formulation
assume the displacement w is a cubic polynomial in

v(x) a1 a 2 x a 3x 2 a 4 x 3
a1, a2, a3, a4 are the undetermined coefficients
L = Length
I = Moment of Inertia of
the cross sectional area
E = Modulus of Elsaticity
v = v(x) deflection of the
neutral axis
= dv/dx slope of the
elastic curve (rotation of
the section
F = F(x) = shear force
M= M(x) = Bending
moment about Z-axis

j
i

(x)

j
vi

Fi, vi

v(x)
x

Fj,vj

vj
Mj,j
x

Mi,i

j
L, EI

a1
a1
a
a
2
2
2
3
v(x) 1 x x x ; (x) 0 1 2x 3x 2
a3
a3
a 4
a 4

x 0, v(0) v1;
x L, v(L) v 2 ;
v i

i
v
j
j

dv
dx

1
x 0

dv
dx

1 0 0 0 a1
0 1 0 0
a2

2 3
1
L
L
a3

2
0 1 2L 3L a 4

2
x L

Applying these boundary conditions, we get

{d} [P(x)]{a}
{a} [P(x)]1{d}
a 1 v1 ;
a 2 1
1
a 3 2 ( 3v1 2L1 3v 2 L2 )
L

Substituting

coefficients ai back into the original equation

for v(x) and rearranging terms gives

v(x) 1 x x 2

a1
a
2
x3
a 3
a 4

The interpolation function or shape function is given by

3x 2 2x 3
2x 2 x 3
v(x) (1 2 3 )v1 (x
2 )1
L
L
L
L
3x 2 2x 3
x2 x3
( 2 3 )v 2 ( 2 )2
L
L
L L
dN1
(x=0) = 0
dx

v1
L

1
v N1 (x) N 2 (x) N 3 (x) N 4 (x)
[N]{d}
v
2
L 2

N1=1

N1(x=L) =1
dN1
(x=L) = 0
dx

N2=1

strain for a beam in bending is defined by the curvature, so

Hence

du
d 2 v d 2 [N]

y 2
{d} y[B]{d}
2
dx
dx
dx

u(x) = y

y
N

12x 6
2
3
L
L

[B]

6x 4
2
3
L L

6x 2
2
3
L h

6 12x
3
2
L
L

dv
dx
x

Internal virtual energy U e =

v

dv

substitute E in above eqn.

U e =
v

E dv

= y B d
U e = d
ve

dv
dx

B T E B d y 2dv

External virtual workdue to body force

w = d(x)
e
b

ve

b dv d N b y dv
T

ve

External virtual work due to surface force

w = d(x)
e
s

p dv d N T p y ds
T

External virtual work due to nodal forces

w d
e
c

P , P = P , M , P ,....
e

e T

yi

yj

From virtual work principle U e W e

d T ( B T E B y 2dv d d T N T b dv N T p dv Pe
y
y
e

e
s
v
v

K Ue F
e
e
where
K B T D B y 2dv Element stiffness matrix
e
ve
F N T b dv N T p ds Pe Total nodal force vector
e
y
y
e
s
v

the

stiffness matrix [k] is defined

[k] [B] E[B]dV dAy E [B]T [B]dx

T

6L 12 6L
12
2
2

EI 6L 4L 6L 2L
3
L 12 6L 12 6L

2
2
6L 2L 6L 4L

dA
y

To compute equivalent nodal force vector

for the loading shown

Fe N p y ds
T

From similar triangles

py

w
w
; p y x; ds = 1 dx
x
L
L

3x 2 2x 3
2x 2 x 3
3x 2 2x 3
x2 x3
2 ) ( 2 3 ) ( 2 )
N (1 2 3 ) (x
L
L
L
L
L
L
L L

py
L

 Fe N p y ds
T

Fe

3x 2 2x 3

(1 2 3 )

L
L

2x 2 x 3

(x
2)

L
L wx

dx

2
3
L

3x
2x

( 2 3 )

L
L

x2 x3

( 2 )

L L

3wL

20

wL2

30

7wL

20
wL 2
20

vi
i

vj
j

+ve directions

Equivalent nodal force due to

Uniformly distributed load w

v1
1

v2
2

v3
v3
v4

Member end forces

For element 1
V1

1
1555.6

V2

M 2

12 18 -12 18 0


18 36 -18 18 0

-12 -18 12 -18 0

18 18 -18 36 0.00249

V1

1
1555.6

V2

M 2

12 18 -12 18
18 36 -18 18

-12 -18 12 -18

18 18 -18 36

0.00249

0.01744

0.007475

70
70

70

139.6

46.53

139.6

46.53

70
70

70

139.6

46.53

139.6

46.53
0

v1
1

v2

v3
3

v1
v2

1
2

v2 2
v3 3

 V1
12
M
6
1

V2
5 -12

8x10

M
2

6
V3

3
Boundary condition

6
4
-6
2

v1

-6
2
1
12+12 -6+6 -12 6 v 2

-6+6
4+4
-6 2 2
-12
-6
12 -6 v 3

6
2
-6
4 3

-12

v1 , 1 , v 2 , v 3 0
Loading Condition
M 2 1000; M 3 1000
1000
8 2 2

1000.0
2
4

8x105

2.679x10 4
2
4 -2 1000
1

4
5

-2
8
1000.0
28*8x10

4.464x10
3

Final member end forces

f k d {FEMS}

For element 1
V1
M
1

V2
M 2

0
5
8X10

0

12
6
-12
6

6
4
-6
2

-12
-6

12
-6

6 0
1285.92

2 0
428.64

-6 0
1285.92

857.28
4 2.679x10

1285.92
428.64

V1
M
1

V2
M 2

6000

1000

5
8X10

6000

1000

12
6
-12
6

6
4
-6
2

-12
-6

12
-6

6 0
6856.8

4
856.96
2 2.679x10

-6 0
51
43.2

4 4.464x10

1285.92

857.28

5143.2

6856.8

856.96

Find slope at joint 2 and deflection

at joint 3. Also find member end
forces

20kN/m

20 KN

EI

EI
2m
6m
EI=2 x 10 4kN-m2

2m
v

Global coordinates

20kN/m

20 KN

Fixed end reactions (FERs)

m
m

Action/loads at global
coordinates

Guided Support

For element 1
f1
m
1X104
1

43
f2
m 2

12 24
24 64

-12 -24

24 32

-12
-24
12
-24

24
32

-24

64

v1

1

v2
2

v1

v 2 2

12
36

-12

36

36
144
-36
72

-12
-36
12
-36

36
72

-36

144

v2

v3

For element 2
f1
m
1X104
1

3
f
6
2

m 2

v1

1
v2
2

3750
-1875
3750
F1
1875

M
3750 10000

-3750
5000
1

-1875 -3750 1875+555.56 -3750+1666.67 -555.56 1666.67
F2

M
3750
5000
-3750+1666.67
10000+6666.67
-1666.67
3333.33
2

F3

-555.56
-1666.67
555.56 -1666.67

1666.67
3333.332
-1666.67 6666.67
M 3

Boundary condition

v1
1

v 2

2
v3

v1 , 1 , v 2 , 3 0
Loading Condition
M 2 50; F3 60
50
16666.67 -1666.67 2

-1666.67 555.56 v
60

3
0.019714
2

555.56 1666.67 50
1

0.16714
6481481.5 1666.67 16666.67 60

v 3
Final member end forces

f k d {FEMS}
04/09/15

26

For element 1
f1
m
1

f2
m 2

10

1X104
10

3
10
4

10

24

32

-24

64

12 24
24 64

-12 -24

24 32

-12
-24
12
-24

12
36

-12

36

-12
-36
12
-36

63.93
88.57
0

0
83.93

207.14
0.019714

For element 2
f1
m
1

f2
m 2

60

1X104
60

3
60
6

60

36
144
-36
72

36

72

-36


144

120

0.019714 207.14

0.16714 0
152.85

q2

q5

q1

q4
q3

displacement in local coordinates

q6

 AE
L

AE
L

[k]

12EI
L3
6EI
L3

6EI
L3
4EI
L3

12EI
3
L
6EI
L3

6EI
3
L
2EI
L3

AE
L
0
0

AE
L
0
0

12EI
3
L
6EI
3
L

6EI
L3
2EI

L3

12EI
L3
6EI
3
L

6EI
3
L

4EI
L3

If f ' member end forces in local coordinates then

f' k ' q '

 q {q1,q 2 ,q 3 ,q 4 ,q 5 ,q 6}
are forces in global coordinate direction

At node i
q1' q1 cos q 2 sin
q '2 q1 sin q 2 cos
q '3 q 3
l cos ;

m sin

using conditions q' [L]{q};

and

f' [L]{f}

Stiffness matrix for an arbitrarily oriented beam element is given by

k L T k ' L

Grid Elements

f = GJ/l
a
a

qxi fi

qxj fj

JG
L

JG
L

JG
q
L xi

JG q xj
L

fi

fj

 GJ
L

GJ
L

12EI
L3
6EI
L3

6EI
L3
4EI
L3

12EI
3
L
6EI
L3

6EI
3
L
2EI
L3

GJ
L
0
0

GJ
L
0
0

12EI
3
L
6EI
3
L

6EI
L3
2EI

L3

12EI
L3
6EI
3
L

6EI
3
L

4EI
L3

If f ' member end forces in local coordinates then

f' k ' q '

 C 0 -s
0 1 0

0 0 0
-s 0 c
L

0
0
0
c
0
-s

0 0 0 0 1 0

0 0 0 --s 0 c

k L

k ' L

Beam element for 3D analysis

q8

q2

q11
q5
q7

q1

q3
z

q10

q6

q4

displacement in local coordinates

q12
q9

if axial load is tensile, results from beam

elements are higher than actual results
are conservative
if axial load is compressive, results are less
than actual

size of error is small until load is about 25% of

Euler buckling load

for 2-d, can use rotation matrices to get

stiffness matrix for beams in any
orientation
to develop 3-d beam elements, must also
add capability for torsional loads about the
axis of the element, and flexural loading in
x-z plane

to derive the 3-d beam element, set up the

beam with the x axis along its length, and y
and z axes as lateral directions
torsion behavior is added by superposition
of simple strength of materials solution
JG JG

xi
i
L
JG JG T
j

xj
L
L

J = torsional moment about x axis

G = shear modulus
L = length
xi, xj are nodal degrees of freedom of
angle of twist at each end
Ti, Tj are torques about the x axis at each
end

flexure in x-z plane adds another stiffness

matrix like the first one derived
superposition of all these matrices gives a
12 12 stiffness matrix
to orient a beam element in 3-d, use 3-d
rotation matrices

for beams long compared to their cross

section, displacement is almost all due to
flexure of beam
for short beams there is an additional lateral
displacement due to transverse shear
some FE programs take this into account,
but you then need to input a shear
deformation constant (value associated with
geometry of cross section)

limitations:
same assumptions as in conventional beam and
torsion theories
no better than beam analysis
axial load capability allows frame analysis, but
formulation does not couple axial and lateral
loading which are coupled nonlinearly

 analysis does not account for

stress concentration at cross section changes
where point loads are applied
where the beam frame components are connected

Finite Element Model

Element formulation exact for beam spans

with no intermediate loads
need only 1 element to model any such
member that has constant cross section

for distributed load, subdivide into several

elements
need a node everywhere a point load is
applied

need nodes where frame members connect,

where they change direction, or where the
cross section properties change
for each member at a common node, all
have the same linear and rotational
displacement
boundary conditions can be restraints on
linear displacements or rotation

simple supports restrain only linear

displacements
built in supports restrain rotation also

 restrain vertical and horizontal displacements

of nodes 1 and 3
no restraint on rotation of nodes 1 and 3
need a restraint in x direction to prevent rigid
body motion, even if all forces are in y
direction

cantilever beam

has x and y linear displacements and rotation of node 1

fixed

point loads are idealized loads

structure away from area of application
behaves as though point loads are applied

only an exact formulation when there are

no loads along the span
for distributed loads, can get exact solution
everywhere else by replacing the distributed
load by equivalent loads and moments at the
nodes

Computer Input Assistance

preprocessor will usually have the same
capabilities as for trusses
a beam element consists of two node
numbers and associated material and
physical properties

material properties:
modulus of elasticity
if dynamic or thermal analysis, mass density
and thermal coefficient of expansion

physical properties:

cross sectional area

2 area moments of inertia
torsion constant
location of stress calculation point

boundary conditions:
specify node numbers and displacement
components that are restrained

loads:
specify by node number and load components
most commercial FE programs allows
application of distributed loads but they use
and equivalent load/moment set internally

Analysis Step
small models and carefully planned element
and node numbering will save you from
bandwidth or wavefront minimization
potential for ill conditioned stiffness matrix
due to axial stiffness >> flexural stiffness
(case of long slender beams)

Output Processing and Evaluation

graphical output of deformed shape usually
uses only straight lines to represent
members
you do not see the effect of rotational
constraints on the deformed shape of each
member
to check these, subdivide each member and
redo the analysis

most FE codes do not make graphical

presentations of beam stress results
user must calculate some of these from the stress
values returned

for 2-d beams, you get a normal stress normal to

the cross section and a transverse shear acting on
the face of the cross section
normal stress has 2 components
axial stress
bending stress due to moment

 expect the maximum normal stress to be at the

top or bottom of the cross section
transverse shear is usually the average
transverse load/area
does not take into account any variation across the
section

3-d beams
normal stress is combination of axial stress,
flexural stress from local y- and z- moments
stress due to moment is linear across a section,
the combination is usually highest at the
extreme corners of the cross section
may also have to include the effects of torsion
get a 2-d stress state which must be evaluated

also need to check for column buckling