Engineering Beam Theory for the First Order Analysis

with Finite Element Method

1998 Winter, Kikuchi

Slender structures whose length is much larger than the size of the cross section, are
called beams. In such structures, deformation may be decomposed into

1. axial deformation
2. bending deformation in two directions
3. torsional deformation

In order to describe the engineering beam theory, we shall introduce the Cartesian
coordinate system (x,y,z), where the z axis coincides with the beam axis define by the
line formed by the centroid of the cross section, while the x and y axes are the
principal axes of the cross section. The origin is set up at the centroid of the left edge
cross section of the beam. Based on

a) the cross section does not deform, i.e., x = y = xy = 0

b) Bernoulli-Euler assumption that no shearing strain is generated by pure
bending
c) Saint-Venant torsion theory

the displacement of an arbitrary point (x,y,z) of the beam is approximated by

U(x, y, z) = us ( z) (y y s )(z )
V (x,y,z ) = v s (z ) + (x xs )(z )
W (x, y, z ) = w( z) xus ' ( z) yv s ' (z) + ns (x, y)' ( z)

where g is the derivative of g in z,

(us(z),vs (s)) transverse deflections of the shear center (xs , ys ) in the x and y
direction, respectively,

w( z ) average axial displacement

( z) angle of twist of the cross section at z

ns (x, y) normalized Saint-Venant warping function when torque is

applied about the shear center axis ( that is the line passing through the
shear center (xs , ys ) ), and is defined a property attached to the cross
section such that
 ns (x, y) = n (x, y ) ys x + xs y

A
ns
dA = A
ns
xdA =
A
ns
ydA = 0

n (x, y) normalized warping function when torque is applied at the

centroid axis ( that is the beam axis, the line passing through the
centroid
of the cross section )

The beam theory based on the displacement approximation stated in above is called
the engineering beam theory.

Using the assumed displacement field, strains are calculated as

x = y = xy = 0

z = w' xus '' yvs '' + ns' '

ns
zx = (y ys ) '
x

zy = ns + (x x s ) '
y

Assuming Hooks low

z = E z , zx = G zx , zy = G zy ,

the total strain energy Ue stored in the beam is given by

( )
1
Ue = z z + zx zx + zy zy dV
2 V

( + zx zx + zy zy )dAdz
1 l
= z z
2 0 A

{EAw' ( )}
1 l
= 2
+EI xx us' ' 2 +EI yy vs' ' 2 + EI (s )'' 2 +GK' 2 dz
2 0

where

A = dA A
cross sectional area

Ixx = x 2 dA
A
moment of inertia about the y axis

Iyy = y 2 dA
A
moment of inertia about the x axis

I(s) =
2
ns
dA warping moment
A

2
ns
2

K= ns
(y ys ) + + (x xs ) dA Saint-Venant
A x y

torsion constant.

Here the x and y axes are the principal axes of the cross section such that

xdA = ydA = xydA = 0 .

A A A

As an example, we shall consider the cross section shown in Fig. 1 that is a typical
configuration of the side member of the car body frame structure. In this example,
(x,y) is a coordinate ststem to define geometry of the cross section of the beam, and is
not the principal axes of the cross section as in above. Section properties are
computed by AISI CARS96, GAS Program1. Geometry of the cross section is given
in the output of CARS96-GAS as follows :

1
AISI/CARS96 for Window : First Order Analysis in Automotive Steel Design, Auto/Steel
Partnership, 2000 Town Center, 19th Floor, Southfield, MI 48075, (248) 351-2664. CARS96 consists
of four modules : Key ( The Key to Automotive Steel Design), GAS ( Geometric Analysis of
Sections ), MAP ( Material Archive Program ), and ASDM ( Automotive Steel Design Manual ).
GAS - CARS Geometric Analysis of Sections
Version 5.0

Date: Jan 20 1998 Time: 11:12:13

Units : N, mm, MPa
Database : C:\CARS96\USER\SIDEBAR
Section Name: sidemember
Description :

Cross Section Geometry:

Point No. X Y
1 0 0
2 2 0
3 2 7
4 33 10
5 36 12
6 47 27
7 57 47
8 57 50
9 52 52
10 47 53
11 47 57
12 40 58
13 40 64
14 13 67
15 13 73
16 6 68
17 2 65
 Line Start End Material Line
No. Pt. Pt. Length Thickness Archive No. Type
1 1 2 2 1 asdm. 1 Segment
2 2 3 7 2 asdm. 1 Segment
3 3 4 31.1448 1 asdm. 1 Segment
4 4 5 3.60555 1 asdm. 1 Segment
5 5 6 18.6011 1 asdm. 1 Segment
6 6 7 22.3607 1 asdm. 1 Segment
7 7 8 3 1 asdm. 1 Segment
8 8 9 5.38516 1 asdm. 1 Segment
9 9 10 5.09902 1 asdm. 1 Segment
10 10 11 4 1 asdm. 1 Segment
11 11 12 7.07107 1 asdm. 1 Segment
12 12 13 6 1 asdm. 1 Segment
13 13 14 27.1662 1 asdm. 1 Segment
14 14 15 6 1 asdm. 1 Segment
15 14 16 7.07107 1 asdm. 1 Segment
16 16 17 5 1 asdm. 1 Segment
17 17 3 58 1 asdm. 1 Segment

Material Description:

Archive Mat'l No. E Fy Archive Location

asdm. 1 203000 234.422 c:\cars96\

*** Results ***

Nominal Properties:

Area = 225.5
cx = 22.561
cy = 36.9
Ixx = 1.1938E+05
Iyy = 87317
Ixy = 18950
Sx+ = 3306.9
Sy+ = 2499.2
Sx- = -3192
Sy- = -3870.2
Theta = -24.884
Iuu = 1.2817E+05
Ivv = 78528
Su+ = 3956.3
Sv+ = 2856.5
Su- = -2951.8
Sv- = -2538.9
rx = 23.009
ry = 19.678
J = 1.2743E+05
Cw = 0
ex = 24.561
ey = 37.234
Cuu = 2.6574
Cvv = 2.1266
Jopen = 21.333
Jc = 1.2741E+05
tomax = 2
tcmin = 1
Ao = 2546

In this example, the principal axes are defined by the u and v axes, and then
 Ixx (= Iuu ) = 1.2817E + 05mm 4
Iyy (= Ivv ) = 7.8528E + 04mm 4
K( = J ) = 1.2743E + 05mm 4

and others.

In the above theory,

axial deformation ( w )
bending about the y axis ( us )
bending about the x axis ( vs )
torsional deformation ( )

are independent each other. Thus we can consider these separately.

(1) Axial Bar ( Truss ) Element

In this case, we assume

U(x, y, z) = V (x,y, z) = 0 , W (x,y, z ) = w (z )

and the average axial displacement w varies linearly in the bar element in z :

c1
w( z) = c1 + c2 z = {1 z} = b T c
c2

Pi, w i Pj , w j
EA

node i
node j

Figure 2 Axial Bar Element ( Truss Element )

Noting that

wi 1 0 c1 c1 1 l 0 wi
w = =
j 1 l c2 c2 l 1 1 w j
we have

1 l 0 wi z z wi
w( z) = {1 z} = 1 = Ndw
l 1 1 w j l l w j

and

w 1 wi
z = = {1 1} = Bdw
z l w j

This yields the total strain energy

EA(Bd
1 l 1 l 1 T
Ue = EA z 2 dz = w )T Bdw dz = d k d
2 0 2 0 2 w w w

where

EA 1 1

l
kw = BT EABdz = .
0 l 1 1

The corresponding generalized force vector is defined by the axial forces Pi and Pj :

Pi
fw =
Pj

and the discrete form of equilibrium becomes

kw dw = fw .

(2) Bending Element

For the bending about the y axis, we consider the displacement field

U(x, y, z) = us ( z) , V (x, y,z ) = 0 , W (x, y,z ) = xus ' ( z)

and we shall assume the deflection us(z) of the shear center on the z cross section in
the x direction by a third degree polynomial in z :

c1
c2
{
us ( z ) = 1 z z2 }
z3 .
c
3
c4
 x

My i My j
u s' i u s' j
V xi, usi EI x x V xj ,usj

z
node i node j

Figure 3 Beam Element ( Moment Applied about the y axis )

If the generalized displacement vector of the bending element is given by

usi
us ' i
du =
usj

us ' j

where they are the deflection and slop at the beam end points i and j, respectively, we
have the following relation :

usi 1 0 0 0 c1 c1 1 0 0 0 1
us ' i 0 1 0 0 c2 c2 0 1 0 0
du = = =
usj 1 l d
l2 l 3 c3 c3 1 l l 2
l3 u

us ' j 0 1 2l 3l 2 c4 c4 0 1 2l 3l
2

and

1 0 0 0 1
0 1 0
{ }
0
us ( z ) = 1 z z2 z3 d = Ndu .
1 l l2 l3 u
0 1 2l 3l 2

Using this the normal strain is given by

 1 0 0 0 1
W d 2 us 0 1 0 0
z = = x = x {0 0 2 6z} d = xBdu
z dz 2 1 l l2 l3 u
0 1 2l 3l 2

where

B = 2 + 3
6 12z 4 6z 6 12z 2 6z
+ 3 +
l l l l2 l2 l l l2

and the strain energy stored in the beam element becomes

l
1 T l T
EI xx (us '' ) dz =
1 2 1
Ue = du B EI xx Bdzdu = du T ku du
2 0 2 0 2

where

12 sym
6l 4l 2

l EI xx
ku = BT EIxx Bdz = .
0 l3 12 6l 12
6l 2l 2 6l 4l 2

The corresponding generalized force vector is defined by the transverse forces Vxi, Vxj
and bending moments Myi, Myj about the y axis at the two end points I and j of the
beam element, respectively :

Vxi
Myi
=
fu .
Vxj

Myj

Then the discrete form of equilibrium becomes

ku du = fu .

Next we shall consider the bending vs about the x axis :

U(x, y, z) = , V (x,y,z ) = vs ( z ) , W(x, y,z ) = yvs ' ( z)

where vs ( z) is the deflection of the shear center of the z cross section in the y
direction.
 y

Mx i Mx j
- v s' i - v s' j
V yi, v si EIy y V y j , v sj

z
node i node j

Figure 4 Beam Element ( Moment Applied About the x Axis )

Noting that the positive bending moment about the x axis coincides with the negative
slope of the beam axis, the generalized displacement vector d v must be defined by

vsi
vs ' i
dv =
vsj

vs ' j

which is associated with the generalized force vector fv :

Vyi
Mxi
fv =
Vyj
.

M xj

Thus, the element stiffness matrix kv can be written by the similar form of ku after
exchanging the sign of the second and fourth columns and rows with the moment of
inertia Iyy :

12 sym
EIyy 6l 4l 2
kv = 3 .
l 12 6l 12
6l 2l 2 6l 4l 2

(3) Bending Element with Shear Deformation

When the effect of shear deformation is considered, the deflection u is decomposed
into ub and us :

u=u +u
b s

where ub is the deflection due to pure bending, and ub is that due to shear,
respectively. Noting that equilibrium of shear forces and bending moments in the
beam element yields

Mi + M j
Vi = V j =
l
(i.e. M j = Mi + Vi l . )
Because of the shearing force, we have

U W du s zx V du s V
zx = + = = = i i.e. = i
z x dz G GAs dz GAs

where G is the shear modulus, As is the effective area of shear deformation defined as
a section property of the beam cross section. For example, if the cross section of the
5 3
beam is rectangular, then As = A , while it becomes As = A if the cross section is
6 4
circular. Integrating the differential equation, we have

Vi l
u si = c and us j = c
GAs

where c is a constant.

Now noting that the matrix equation of the beam for bending may be written by

ku b du b = fu that is ( )
ku b du du s = fu

where

u bi c Vi
u i
b' 0 Mi
du b = b , du s = c Vil , and fu =
u Vj
b' j GAs
u j 0 Mj

Here we have used the fact that u s ' = 0 since the shear deformation is independent of
bending moment. Since
 c c 0 0 0 0 Vi
0 0 0 0 0 0 Mi
du s = c Vi l = + l
c 0 0 0 Vj
GAs GAs
0 0 0 0 0 0 M j

we have

c 0 0 0 0 Vi
0 0 0 0 0 Mi
ku b du + l 0 0 0 Vj = fu
c
GAs
0 0 0 0 0 M j

and then

0 0 0 0
0 0 0 0
ku b du ku b l 0 fu = fu
0 0
GAs
0 0 0 0

Therefore, we have
1
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
ku b du = I + ku b l 0 0 0 fu I + ku l
b
0 0 0 ku du = fu
b

GAs GAs
0 0 0 0 0 0 0 0

that is, the element stiffness matrix becomes

1
0 0 0 0
0 0 0 0
ku b+ s = I + ku b l b
0 ku .
0 0
GAs
0 0 0 0

Evaluating this by using the matrix obtained in the previous section, we have

12 sym
EI 6l (4 + x )l 2

ku b+ s = 3 xx
l (1+ x ) 12 6l 12
6l (2 x )l 2 2
6l (4 + x )l

where

12EI xx
x =
GAsx l 2

is the shear deformation parameter such that

12EI x A rx 2 Ix
x = = 24(1 + ) with rx =
GAsx l 2 Asx l A

rx
, is the ratios of radius of gyration to beam element length. If the beam is slender,
l
it becomes zero. Algebra involved in above may be evaluated by, e.g.
MATHEMATICA.

The following is a typical algebra by MATHEMATICA to determine the element stiffness

matrix including the shear deformation :

KB=(EI/L^3)*{{12,6*L,-12,6*L},
{6*L,4*L^2,-6*L,2*L^2},
{-12,-6*L,12,-6*L},
{6*L,2*L^2,-6*L,4*L^2}};
CS=Table[0,{i,1,4},{j,1,4}];
CS[[3,1]]=-L/GAs;
ID=Table[If[i==j,1,0],{i,1,4},{j,1,4}];
KB.CS
IS=Inverse[ID+KB.CS];
KS=Simplify[IS.KB]
Out[49]=
12 EI 6 EI -12 EI
{{------, 0, 0, 0}, {-----, 0, 0, 0}, {------, 0, 0, 0},
2 GAs L 2
GAs L GAs L

6 EI
{-----, 0, 0, 0}}
GAs L

Out[51]=
12 EI 6 EI -12 EI 6 EI
{{------------, ----------, ------------, ----------},
12 EI L 3 12 EI 2 12 EI L 3 12 EI 2
------- + L ----- + L ------- + L ----- + L
GAs GAs GAs GAs

2
6 EI GAs 4 EI 36 EI -6 EI GAs
{--------------, ---- - ----------------, --------------,
2 L 3 2
12 EI + GAs L 12 EI L + GAs L 12 EI + GAs L
 2
2 EI 36 EI
---- - ----------------},
L 3
12 EI L + GAs L

-12 EI GAs -6 EI GAs 12 EI GAs

{----------------, --------------, ----------------,
3 2 3
12 EI L + GAs L 12 EI + GAs L 12 EI L + GAs L

-6 EI GAs 6 EI GAs
--------------}, {--------------,
2 2
12 EI + GAs L 12 EI + GAs L

2
2 EI 36 EI -6 EI GAs
---- - ----------------, --------------,
L 3 2
12 EI L + GAs L 12 EI + GAs L

2
4 EI 36 EI
---- - ----------------}}
L 3
12 EI L + GAs L

Similarly, the stiffness matrix of the bending beam with shear deformation about the y
axis becomes

12 sym

kv b+ s =
EI yy 6l (4 + y )l 2

,
(
l3 1 + y ) 12 6l 12
6l
(2 y )l 2
6l ( )
4 + y l 2

where

A ry
2
12EI y Iy
y = 2 = 24(1 + ) with ry =
GAsy l Asy l A

ry
is the ratios of radius of gyration to beam element length. If the beam is slender, it
l
becomes zero.

(4) Saint-Venant Torsion Element

Torsional deformation is governed by

 U(x, y, z) = (y y s )( z) , V (x, y, z ) = + (x xs )(z ) , W (x,y,z ) = ns (x,y )' (z )

For the Saint-Venant torsion theory, the angle of twist is assumed to be

' = = constant

in the bar element, and then we assume

1 l 0 i
( z ) = c1 + c2 z = {1 z} .
l 1 1 j

Figure 5 Torsion Bar Element

The shear strains are then obtained by

W U ns n n
zx x + z x (y ys ) y 1 y i
{1 1}
x x
= W V = ' = ' =
y + z y + (x xs )
zy ns
y + x
n l
y + x
n j

and the strain energy becomes

G 0 zx
V ( zx zx + zy zy)dV = 2 V { zx }
1 1
Ue = zy dV
2 0 G zy

n
y
1 1 n n G 0 1 x

1 1 T
= dt T l y + x {1 1}dAd = d kd
A l 1 x y 0 G l n + x
t
2 2 t t t
y

where
 i
dt =
j

GK 1 1
kt =
l 1 1

and

2
n
2

K= n
y + + x dA .
A x y

The corresponding generalized force vector is defined by the applied torque Mzi and
Mzj at the two end points, i.e.,

Mzi
ft =
Mzj

and the matrix form of the equilibrium becomes

kt dt = ft .

The value of K for a typical cross sections we can see in many applications are
summerized as follows :

a 3b3
a) elliptic cross section with two radii a and b K=
a 2 + b2

bt 3 192 b
b) rectangular cross section with b and t K 1 tanh
3 b
5
2t

0.025A4
c) prismatic cross sections ( Saint-Venant ) K
Ip

where A is the cross sectional area and Ip is the polar moment of inertia. This
approximation is good enough except for the cross sections having one dimension
which is much larger than the rest.
n

b t
1
d) thin-walled open sections K ii
3
3
i=1

where ti is the thickness of the ith thin-walled open section, bi is the length of the ith
thin-walled open section, and n is the total number of sections
 4 2
e) thin-walled single-cell tubes K
ds t
where is the area enclosed by the center line of the tube wall, t is the thickness, and
s is the coordinate along the center line of the tube wall
n

b t
4 2 1
f) closed tube with fins K + 3

i i
ds t 3 i =1

(5) Element Stiffness Matrix of the Beam Element

Combining all the independent components :

axial deformation
bending deformations in the two orthogonal directions with shear effect
torsional deformation

we have obtained the equilibrium in discrete form :

kw 0 0 0 dw fw
0 ku 0 0 du fu
0 = .
0 kv 0 dv fv
0
0 0 kt dt ft

Since this form is not convenient in coordinate transformation that is required at the
assembling of all the beam elements of a structure, we shall define the element
generalized displacement and force vectors as follows :

{
d T = ui vi wi vs ' i us ' i i uj vj wj v s ' j us ' j j }
and

{
f T = Vxi Vyi Pi Mxi Myi Mzi Vxj Vyj Pj Mxj Myj }
Mzi .

Then the element stiffness matrix k becomes

 12EI xx
(1+ x )l 3
12EI yy
(1 + y )l 3
0 sym

EA
0 0
l


6EI yy (4 + y )EI yy
( ) (1 + y )l
0 0
1+ y l
2

6EIxx (4 + x )EI xx
0 0 0
(1 + x )l
2
(1 + x )l

GK
0 0 0 0 0
k= l
12EI xx 6EIxx 12EI xx
0 0 0 0
(1+ x )l 3 (1 + x )l 2 (1+ x )l 3
12EIyy 6EI yy 12EI yy

(1 + y )l 3 (1 + y )l 2 (1 + y )l 3
0 0 0 0 0

EA EA
0 0 0 0 0 0 0
l l


6EI yy (2 y )EI yy 6EI yy (4 + y )EIyy

(1 + y )l 2 (1 + y )l (1 + y )l 2 (1 + y )l
0 0 0 0 0 0


6EIxx
0 0 0
(2 x )EI xx 0
6EI xx
0 0 0
(4 + x )EI xx
(1 + x )l
2
(1 + x )l (1+ x )l2 (1 + x )l
GK GK
0 0 0 0 0 0 0 0 0 0
l l

Now, if we define the three orthonormal basis vectors of the local coordinate system
(x,y,z) and the global coordinate system (X,Y,Z) by

T
e = ix [ iy iz ] and E = [i X
T
iY iZ ]

respectively, they are related to the coordinate transformation matrix T by

e = TE

where

l xX l xY l xZ
T = l yX l yY lyZ ,
l l zY lzZ
zX

{l xX lxY l xZ } is the directional cosine of the local coordinate x with respect to the
global coordinate system (X,Y,Z), and similarly, lyX lyY lyZ and {lzX lzY lzZ } { }
are the directional cosines of the local coordinates y and z with respect to the global
coordinate system (X,Y,Z), respectively. Therefore, the matrix equation in the local
coordinate system is transformed into the one in the global coordinate system
(X,Y,Z) :

k Gd G = f G

where
 T 0 0 0 T T 0 0 0
0 T 0 0 0 T 0 0
kG = k
0 0 T 0 0 0 T 0
0 0 0 T 0 0 0 T

d G = uiG { viG wiG GXi G

Yi
G
Zi uGj v Gj wG
j
G
Xj Yj
G
Zj
G T
}
and

{ }.
T
f G = PXi PYi PZi MXi MYi MZi PXj PYj PZj M Xj MYj MZj

Finite Element Analysis of Plane Frame Structures

For plane beams, in the local coordinate system (x,y,z) we have elementwise
equilibrium relation from the principle of minimum potential energy, or equivalently
from the first Castigliano theorem :

ku = f

where

12EI yy
sym
(
1+ y l 3 )
EA
0
l


6EI yy
0
(4 + y )EI yy

k=
(
1+ l 2
y ) (1 + y )l

12EIyy 6EI yy 12EI yy
0
(
1 + y l3 ) (1 + y )l 2 (1 + y )l 3
EA EA
0 0 0
l l


6EI yy
0
(2 y )EI yy 6EI yy 0
( )
4 + y EI yy

(
1+ y l
2
) (1 + y )l (1 + y )l 2 (
1 + y l )
 vsi Vyi
wi Pi
vs ' i Mxi
d= f =
Vj
and
v sj

wj Pj

vs ' i Mxj

Here vs is the transverse deflection of the shear center of the beam in the y direction, w
is the axial displacement of the beam that is the average displacement of the cross
section in the z direction, -vs is the rotation ( that is the slope of the transverse
deflection vs ) about the x axis that is orthogonal to the yz plane in the right hand
coordinate system, while Vy is the transverse shear force in the y direction, P is the
axial force in the z direction, and Mx is the bending moment about the x axis.
Similarly, E is Youngs modulus, A is the area of the cross section, Iyy is the moment
of inertia of the cross section about the x axis, i.e. Iyy = Ay 2dA , y is the shear
constant in the y axis such that

A ry
2
12EI yy Iyy
y = = 24 (1+ ) with ry = ,
GAsy l 2 Asy l A

Asy is the effective area of the cross section for the transverse shear in the y direction,
is Poissons ratio, ry is the radius of gyration, and l is the beam length.
 y

z
(Yj,Z j)

i beam element
Y
(Yi,Z i) local coordinate system y

Z
global coordinate system
x

cross section
Figure X Global and Local Coordinate Systems of a Plane Beam Element
for Side Frame Analysis

Noting that

vs = uY cos uZ sin and w = uY sin + uZ cos

we have the transformation matrix

cos sin 0
T = sin cos 0
0 0 1

and the relation between the displacements and rotation in the local and global
coordinate systems :

vs cos sin 0 uY uY
w = sin cos 0 uZ = T uZ .
v ' 0 1 X
s 0 X

Defining the transformation matrix for the element stiffness matrix

 cos sin 0 0 0 0
sin cos 0 0 0 0
0 0 1 0 0 0
TG =
0 0 0 cos sin 0
0 0 0 sin cos 0

0 0 0 0 0 1

where

Z j Zi Yj Yi
cos = and sin =
(Yj Yi )2 + (Z j Zi ) (Yj Yi )2 + (Z j Zi )
the element stiffness matrix for the global coodinate system (Y,Z) setting up at the
whole structure that can be assembled is given by

T
k G = T G kT G

and the generalized displacement and force vectors for the global coordinate system
are defined by

uYi PYi
uZi PZi
Xi M Xi
G
d = fG =
PYj
and
uYj

u PZj
Zj
Xj Xj

where uY and uZ are the displacement in the Y and Z directions, respectively, X is the
rotation about the X axis, PX and PY are the forces in the Y and Z direction,
respectively, and MX is the moment about the X axis.

We shall consider an example of a plane beam structure consisting of three beam

elements and three nodal points shown in
 3

(2)
P
(3)

1 (1) 2

Here a vertical point force P is applied at node 1, while node 2 is fixed in the
horizontal and vertical directions and node 3 is supported by hinge roller that can
move vertically.

Noting the element connectivities of the structure that describe i and j nodes of a beam
element are given by

element i j
(1) 1 2
(2) 1 3
(3) 2 3

the global stiffness matrix of the whole structure becomes 9 x 9 matrix, while the
global generalized displacement and force vectors are 9 component vectors. If the
element stiffness matrices for the global coordinate system are expressed by kG1,kG2,
and kG3, they are assembled to the global stiffness matrix K for the structure by the
following algorithm :

for element=1:totalnumberofelement
% determine the location of the global stiffness matrix of the structure where
% element stiffness matrix is assembled ( or placed )
numberofnode=2
numberofdegreepernode=3
for nodel=1:numberofnode
nodeg=ijk(nodel,element)
for degree=1:numberofdegreepernode
il= numberofdegreepernode*(nodel-1)+degree
ig=numberofdegreepernode*(nodeg-1)+degree
location(il)=ig
end
end
%
% form element stiffness matrix ske : 6 x 6 matrix for plane beam elements
 %
% assembling of ske to the global one sk
totaldegree=numberofnode*numberofdegreepernode
for i=1:totaldegree
ig=location(i)
for j=1:totaldegree
jg=location(j)
sk(ig,jg)=sk(ig,jg)+ske(i,j)
end
end
end

Here ske is representating each element stiffness matrix kG in the global coordinate
system, and sk represents the global stiffness matrix K of the whole structure. Array
ijk is the list table of the element connectivities. If we perform the above algorithm for
assembling to form the global stiffness matrix K of the whole structure, we have

k (1) + k (2 ) k (1) + k (2 ) ()
k13 + k13
1 ( )
2
k14
(1)
k15
(1)
k16
(1) (2 )
k14
(2 )
k15
(2 )
k16
11 11 12 12
(1) + (2 )
k 22 k22 (1 ) + ( 2 )
k23 k23 (1)
k24 (1)
k25 (1)
k26 (2 )
k 24 (2 )
k25 (2 )
k26
(1 ) (2) (1) (1) (1) (2 ) (2 ) (2 )
k33 + k33 k34 k35 k36 k 34 k35 k36
() ( ) () ( ) () () (3 ) (3 ) (3 )
k44 + k11 k 45 + k12 k46 + k13
1 3 1 3 1 3
k14 k15 k16
(1) (3 ) (1) (3) (3 ) (3 ) (3 )
k 55 + k22 k56 + k23 k24 k25 k26
(1) + (3) (3 ) (3 ) (3 )
k66 k33 k34 k35 k36
(2 ) (3 ) (2 ) (3 ) (2 ) (3 )
+ +
k44 k44 k45 k45 k46 k46 +
(2 ) (3 ) (2 ) (3 )
k55 + k55 k56 + k56
sym k66 + k66
(2 ) (3 )

( e)
Here kij is the ij component of kG of beam element (e). Since displacement is
constrained at node 2 and node 3, the global displacement vector d must be
constrained to satisfy

d4 = d5 = d7 = 0 .

Therefore the global stiffness matrix K must be modified by KM, for example,
k (1) + k (2 ) k (1) + k (2 ) (1 )
k13 ( 2)
+ k13 0 0 k16(1)
0 k15(2) (2 )
k16
11 11 12 12
(1)
k 22 + 22
k (2 ) (1 ) ( 2)
k23 + k23 0 0 k (26
1)
0 k (252) (2 )
k 26
(1 ) ( 2) (1) (2) (2 )
k33 + k33 0 0 k36 0 k35 k 36

1 0 0 0 0 0

1 0 0 0 0
(1 ) (3 ) (3) (3 )
k66 + k33 0 k35 k36
1 0 0
(2 ) (3) (2 ) (3 )
k55 + k55 k56 + k56
sym k (2 ) + k (3 )
66 66

In this modification, we replace the colums and rows related to the degrees of freedom
to be constrained by zeros except the unit diagonal terms. The global generalized
force f of the whole structure becomes

f T = {0 P 0 0 0 0 0 0 0},

and then the equilibrium of the whole structure is written by

KM d = f .

Since the modified stiffness matrix KM is not singular, we can solve the matrix
equation by

( M)
1
d= K f.

Once the global generalized displacement d is computed, we have displacement

components uY and uZ and the rotation about the X axis at the end points i and j of
each beam element, we can transform these into the components in the local
coordinate system (x,y,z) so that strains, stresses, and strain energy of each beam
element are caluculated in the local coordinate system.

A MATLAB Program for Plane Frame Analysis

We shall develop a MATLAB program for plane beam analysis with possibly flexible
joints in order to make finite element (FE) study on side frame analysis and
examination of topology of a car body structure. For simplicity, we assume that a
structure can be modeled as a plane frame with flexible joints, although most of frame
structures in real automotive bodies behave essentially three dimensionally.

A FE program for plane frame structures is written in MATLAB, and only the beam
element and flexible joint element are used to model a side frame of a automotive
body. The beam element is defined by two end nodes i and j, and three degrees of
freedom { uY, uZ, X} are assumed at each node, where uY and uZ are the displacement
components in the Y and Z global coordinate system for the structure, respectively,
and X is the rotation about the X axis perpendicular to the Y and Z axes that is the
same with the rotation about the x axis orthogonal to the local coordinate system y and
z, where the z axis is parallel to the beam axis passing through the shear center of the
beam cross section. The possibly flexible joint element consists of two nodes i and j
whose nodal coordinates are the same with the three degrees of freedom { uY, uZ, X}
which are the same with the beam element, and its stiffness matrix kG in the global
coordinate system (Y,Z) is given by

kY 0 0 kY 0 0
0 kZ 0 0 k Z 0
0 0 kX 0 0 kX
kG =
kY 0 0 kY 0 0
0 k Z 0 0 kZ 0

0 0 kX 0 0 kX

where kY, kZ, and kX are the stiffness of the flexible joint in the Y and Z displacement
components, respectively, and of the rotation about the X axis. In this sense, we may
consider kY and kZ are the spring constants of the springs inserted to the two nodes i
and j, while kX is the rotational spring constant of the torsional spring attached at the
two nodes i and j. In the MATLAB program, we are specifying these stiffness by the
percent of the average stiffness of the beam elements connected at the flexible joint.

For simplicity, sequence of the node numbers is made from 1 with no skipping, that
is, the node numbers are started from 1 and ended at the total number of nodes nx.

The following is the list of the MATLAB program which is more or less self-
explained. It is strongly recommended to be modified for your study so that you
develop your own analysis tool for side frame analysis and topology consideration.
The program consists of three parts : pre-processing, FE analysis, and post-
processing.

% MEAM 599- 02 / 1998 Winter

% FEM for Side Frame Analysis using plane beam elements
% with possibly flexible joints
% N. Kikuchi
% ______________________________________________________________
%
% Pre-Processing / set up an analysis model
%
nfile=input('Is data in the datafile ? [y/n] = ','s');
if nfile~='y'
% read nodal coordinates ( Z, Y )
nx=input('number of nodes of the whole structure = ');
for i=1:nx
i
Z(i)=input('Z coordinate of node = ');
Y(i)=input('Y coordinate of node = ');
end
% plot the nodes of the plane beam structure
plot(Z,Y,'+')
% read element connectivity and section type of beam elements
% ijk(1,nel)=node i of beam element nel
% ijk(2,nel)=node j of beam element nel
% ijk(3,nel)=section type of beam element nel
nelx=input('number of beam elements = ');
for nel=1:nelx
nel
ijk(1,nel)=input('node i = ');
ijk(2,nel)=input('node j = ');
ijk(3,nel)=input('section type = ');
end
% plot the beam elements
for nel=1:nelx
Ze(2*nel-1)=Z(ijk(1,nel));
Ze(2*nel)=Z(ijk(2,nel));
Ye(2*nel-1)=Y(ijk(1,nel));
Ye(2*nel)=Y(ijk(2,nel));
end
plot(Z,Y,'+',Ze,Ye)
% read section type ( properties )
nsecx=input('total number of section type = ');
for i=1:nsecx
i
E(i)=input('Young,s modulus / E = ');
A(i)=input('crossectional area / A = ');
Iyy(i)=input('moment of inertia about the x axis / Iyy = ');
Fy(i)=input('shear constant / Fy = ');
end
% read data for flexible joints
nfjx=input('number of flexible joints = ');
if nfjx>0
for i=1:nfjx
i
fjoint(1,i)=input('node i of the flexible joint = ');
fjoint(2,i)=input('node j of the flexible joint = ');
kZ(i)=input('stiffness percent in the Z direction / kZ = ');
kY(i)=input('stiffness percent in the Y direction / kY = ');
kqX(i)=input('stiffness percent about the X axis rotation / kqX
= ');
end
end
% plot the flexible joints
if nfjx>0
for i=1:nfjx
Zfj(i)=Z(fjoint(1,i));
Yfj(i)=Y(fjoint(1,i));
end
plot(Z,Y,Ze,Ye,'+',Zfj,Yfj,'o')
end
% displacement constraints
spc=[];
nspc=input('number of single point constraints = ');
for i=1:nspc
i
spc(1,i)=input('node number = ');
spc(2,i)=input('degree of freedom for spc = ');
spc(3,i)=input('constrained value = ');
end
% applied forces and moments at the nodes
afm=[];
nafm=input('number of applied forces and moment = ');
for i=1:nafm
i
afm(1,i)=input('node number = ');
afm(2,i)=input('degrees of freedom = ');
afm(3,i)=input('applied forces or moment = ');
end
%
save input.mat nx Y Z nelx ijk nsecx E A Iyy Fy nfjx fjoint kY kZ kqX nspc
spc nafm afm
end
if nfile=='y'
load input.mat
% load input.mat nx Y Z nelx ijk nsecx E A Iyy Fy nfjx fjoint kY kZ kqX
nspc spc nafm afm
end
%
% FE-Processing / forming the global stiffness matrix
%
% beam elements
%
sk=zeros(3*nx);
f=zeros(3*nx,1);
sksize=3*nx;
%
for nel=1:nelx
%
Zji=Z(ijk(2,nel))-Z(ijk(1,nel));
Yji=Y(ijk(2,nel))-Y(ijk(1,nel));
Lji=sqrt(Zji^2+Yji^2);
cji=Zji/Lji;
sji=Yji/Lji;
TG=zeros(6);
TG(1,1)= cji;
TG(1,2)=-sji;
TG(2,1)= sji;
TG(2,2)= cji;
TG(3,3)= 1;
TG(4,4)= cji;
TG(4,5)=-sji;
TG(5,4)= sji;
TG(5,5)= cji;
TG(6,6)= 1;
%
skel=zeros(6);
ijk3=ijk(3,nel);
EIyy=E(ijk3)*Iyy(ijk3);
Fy1=(1+Fy(ijk3));
skel(1,1)=12*E(ijk3)*Iyy(ijk3)/((1+Fy(ijk3))*Lji^3);
skel(2,2)=E(ijk3)*A(ijk3)/Lji;
skel(3,1)=-6*E(ijk3)*Iyy(ijk3)/((1+Fy(ijk3))*Lji^2);
skel(3,3)=(4+Fy(ijk3))*E(ijk3)*Iyy(ijk3)/((1+Fy(ijk3))*Lji);
skel(4,1)=-skel(1,1);
skel(4,3)=-skel(3,1);
skel(4,4)= skel(1,1);
skel(5,2)=-skel(2,2);
skel(5,5)= skel(2,2);
skel(6,1)= skel(3,1);
skel(6,3)=(2-Fy(ijk3))*E(ijk3)*Iyy(ijk3)/((1+Fy(ijk3))*Lji);
skel(6,4)=-skel(3,1);
skel(6,6)= skel(3,3);
for i=1:5
for j=1+1:6
skel(i,j)=skel(j,i);
end
 end
%
skeg=TG'*skel*TG;
%
ijk1=ijk(1,nel);
ijk2=ijk(2,nel);
ndg=[3*ijk1-2,3*ijk1-1,3*ijk1,3*ijk2-2,3*ijk2-1,3*ijk2];
for i=1:6
for j=1:6
sk(ndg(i),ndg(j))=sk(ndg(i),ndg(j))+skeg(i,j);
end
end
%
end
%
% flexible joint elements
%
if nfjx>0
for nfj=1:nfjx
skeg=zeros(6);
ijk1=fjoint(1,nfj);
ijk2=fjoint(2,nfj);
kYb=(sk(3*ijk1-2,3*ijk1-2)+sk(3*ijk2-2,3*ijk2-2))/2;
kZb=(sk(3*ijk1-1,3*ijk1-1)+sk(3*ijk2-1,3*ijk2-1))/2;
kqXb=(sk(3*ijk1,3*ijk1)+sk(3*ijk2,3*ijk2))/2;
refrencestiffness=[nfj,kYb,kZb,kqXb]
skeg(1,1)= kY(nfj)*kYb;
skeg(4,4)= kY(nfj)*kYb;
skeg(1,4)=-kY(nfj)*kYb;
skeg(4,1)=-kY(nfj)*kYb;
skeg(2,2)= kZ(nfj)*kZb;
skeg(5,5)= kZ(nfj)*kZb;
skeg(2,5)=-kZ(nfj)*kZb;
skeg(5,2)=-kZ(nfj)*kZb;
skeg(3,3)= kqX(nfj)*kqXb;
skeg(6,6)= kqX(nfj)*kqXb;
skeg(3,6)=-kqX(nfj)*kqXb;
skeg(6,3)=-kqX(nfj)*kqXb;
ndg=[3*ijk1-2,3*ijk1-1,3*ijk1,3*ijk2-2,3*ijk2-1,3*ijk2];
for i=1:6
for j=1:6
sk(ndg(i),ndg(j))=sk(ndg(i),ndg(j))+skeg(i,j);
end
end
end
end
%
% single point constraints
%
for i=1:nspc
spc1=spc(1,i);
spc2=spc(2,i);
spc3=spc(3,i);
ndof=3*(spc1-1)+spc2;
f=f-sk(:,ndof)*spc3;
sk(:,ndof)=zeros(sksize,1);
sk(ndof,:)=zeros(1,sksize);
sk(ndof,ndof)=1;
end
%
% applied for#es and moments
%
for i=1:nafm
afm1=afm(1,i);
 afm2=afm(2,i);
afm3=afm(3,i);
ndof=3*(afm1-1)+afm2;
f(ndof)=f(ndof)+afm3;
end
%
% solving the matrix equation
%
d=sk\f;
d'
%
% Post-Processing of the computed results
%
strainenergy=[];
Zp=[];
Yp=[];
uZ=[];
uY=[];
for nel=1:nelx
Zji=Z(ijk(2,nel))-Z(ijk(1,nel));
Yji=Y(ijk(2,nel))-Y(ijk(1,nel));
Lji=sqrt(Zji^2+Yji^2);
cji=Zji/Lji;
sji=Yji/Lji;
TG=zeros(6);
TG(1,1)= cji;
TG(1,2)=-sji;
TG(2,1)= sji;
TG(2,2)= cji;
TG(3,3)= 1;
TG(4,4)= cji;
TG(4,5)=-sji;
TG(5,4)= sji;
TG(5,5)= cji;
TG(6,6)= 1;
deg=zeros(6,1);
for i=1:2
ijki=ijk(i,nel);
for j=1:3
deg(3*(i-1)+j)=d(3*(ijki-1)+j);
end
end
del=TG*deg;
skel=zeros(6);
ijk3=ijk(3,nel);
EIyy=E(ijk3)*Iyy(ijk3);
Fy1=(1+Fy(ijk3));
skel(1,1)=12*E(ijk3)*Iyy(ijk3)/((1+Fy(ijk3))*Lji^3);
skel(2,2)=E(ijk3)*A(ijk3)/Lji;
skel(3,1)=-6*E(ijk3)*Iyy(ijk3)/((1+Fy(ijk3))*Lji^2);
skel(3,3)=(4+Fy(ijk3))*E(ijk3)*Iyy(ijk3)/((1+Fy(ijk3))*Lji);
skel(4,1)=-skel(1,1);
skel(4,3)=-skel(3,1);
skel(4,4)= skel(1,1);
skel(5,2)=-skel(2,2);
skel(5,5)= skel(2,2);
skel(6,1)= skel(3,1);
skel(6,3)=(2-Fy(ijk3))*E(ijk3)*Iyy(ijk3)/((1+Fy(ijk3))*Lji);
skel(6,4)=-skel(3,1);
skel(6,6)= skel(3,3);
for i=1:5
for j=1+1:6
skel(i,j)=skel(j,i);
end
 end
nel
strainenergynel=(del'*skel*del)/2
strainenergy(nel)=strainenergynel;
% axial strain ez = ez0 + y ( ezi*(1-z/l) + ezj*(z/l) )
ez0=(del(5)-del(2))/Lji;
ezi= 6*(del(1)-del(4))/Lji^2-(4*del(3)+2*del(6))/Lji;
ezj=-6*(del(1)-del(4))/Lji^2+(2*del(3)+4*del(6))/Lji;
axialstrain=[ez0,ezi,ezj]
% axial stress sz = sz0 + y ( szi*(1-z/l) + szj*(z/l) )
sz0=E(ijk3)*ez0;
szi=E(ijk3)*ezi;
szj=E(ijk3)*ezj;
axialstress=[sz0,szi,szj]
% displacement of the beam axis
ipx=11;
for ip=1:ipx
zpi=(ip-1)/(ipx-1);
nv=[1-3*zpi^2+2*zpi^3,Lji*(zpi-2*zpi^2+zpi^3),3*zpi^2-
2*zpi^3,Lji*(-zpi^2+zpi^3)];
nw=[1-zpi,zpi];
uyi=nv*[del(1),-del(3),del(4),-del(6)]';
uzi=nw*[del(2),del(5)]';
ipnel=ipx*(nel-1)+ip;
Yp(ipnel)=nw*[Y(ijk(1,nel)),Y(ijk(2,nel))]';
Zp(ipnel)=nw*[Z(ijk(1,nel)),Z(ijk(2,nel))]';
uY(ipnel)= uyi*cji+uzi*sji;
uZ(ipnel)=-uyi*sji+uzi*cji;
end
end
if nfjx>0
for nfj=1:nfjx
skeg=zeros(6);
ijk1=fjoint(1,nfj);
ijk2=fjoint(2,nfj);
kYb=(sk(3*ijk1-2,3*ijk1-2)+sk(3*ijk2-2,3*ijk2-2))/2;
kZb=(sk(3*ijk1-1,3*ijk1-1)+sk(3*ijk2-1,3*ijk2-1))/2;
kqXb=(sk(3*ijk1,3*ijk1)+sk(3*ijk2,3*ijk2))/2;
skeg(1,1)= kY(nfj)*kYb;
skeg(4,4)= kY(nfj)*kYb;
skeg(1,4)=-kY(nfj)*kYb;
skeg(4,1)=-kY(nfj)*kYb;
skeg(2,2)= kZ(nfj)*kZb;
skeg(5,5)= kZ(nfj)*kZb;
skeg(2,5)=-kZ(nfj)*kZb;
skeg(5,2)=-kZ(nfj)*kZb;
skeg(3,3)= kqX(nfj)*kqXb;
skeg(6,6)= kqX(nfj)*kqXb;
skeg(3,6)=-kqX(nfj)*kqXb;
skeg(6,3)=-kqX(nfj)*kqXb;
for i=1:2
ijki=fjoint(i,nfj);
for j=1:3
deg(3*(i-1)+j)=d(3*(ijki-1)+j);
end
end
nfj
strainenergynfj=(deg'*skeg*deg)/2
strainenergy(nelx+nfj)=strainenergynfj;
end
end
bar(strainenergy)
xlabel('beam elements and flexible joints')
ylabel('strain energy')
title('strain energy distribution')
% plot the deformed configuration
mag=input('magnification factor of the fisplacement = ');
Zup=Zp+mag*uZ;
Yup=Yp+mag*uY;
plot(Zp,Yp,Zup,Yup)
title('deformed configuration')
xlabel('Z')
ylabel('Y')

Example 1 We shall consider an idealized model of the portion of the joint of the
center pillar and the rocker frame as shown in Figure X, whose idealized dimensions
are given as in the figure. Assuming standard structural steel, whose Youngs
modulus is 200 Gpa ( i.e. 200 kN/mm2 ), we shall consider the idealized 50 mm x 100
mm rectangular cross section of the rocker and center pillar with 1 mm thickness of a
thin walled box beam. Then this structure is modeled by 6 beam elements and 1
flexible joint defined by node 3 and 6.

1 mm

P = 1 kN

400 mm 100 mm

50 mm
1000 mmm
Figure X Schematic View of a Plane Frame Structure

(6)

(5)
6

Flexible Joint ( 3-6)

(1) (2) (3) (4)

1 2 3 4 5
 Figure X A Finite Element Model with a Flexible Joint

Execution of the MATLAB program yields the following result :

Is data in the datafile ? [y/n] = n

number of nodes of the whole structure = 8

i = 1
Z coordinate of node = -500
Y coordinate of node = 0

i = 2
Z coordinate of node = -250
Y coordinate of node = 0

i = 3
Z coordinate of node = 0
Y coordinate of node = 0

i = 4
Z coordinate of node = 250
Y coordinate of node = 0

i = 5
Z coordinate of node = 500
Y coordinate of node = 0

i = 6
Z coordinate of node = 0
Y coordinate of node = 0

i = 7
Z coordinate of node = 0
Y coordinate of node = 200

i = 8
Z coordinate of node = 0
Y coordinate of node = 400
 400

350

300

250

200

150

100

50

0
-500 0 500

number of beam elements = 6

nel = 1
node i = 1
node j = 2
section type = 1

nel = 2
node i = 2
node j = 3
section type = 1

nel = 3
node i = 3
node j = 4
section type = 1

nel = 4
node i = 4
node j = 5
section type = 1

nel = 5
node i = 6
node j = 7
section type = 1

nel = 6
node i = 7
node j = 8
section type = 1

400

350

300

250

200

150

100

50

0
-500 0 500

total number of section type = 1

i= 1
Young,s modulus / E = 200
crossectional area / A = 300
moment of inertia about the x axis / Iyy = 401900
shear constant / Fy = 0

number of flexible joints = 1

i= 1
node i of the flexible joint = 3
node j of the flexible j&int = 6
stiffness percent in the Z direction / kZ = 100
stiffness percent in the Y direction / kY = 100
stiffness percent about the X axis rotation / kqX = 0.7

400

350

300

250

200

150

100

50

0
-500 0 500

number of single point constraints = 3

i= 1
node number = 1
degree of freedom for spc = 1
constrained value = 0

i= 2
node number = 1
degree of freedom for spc = 2
constrained value = 0

i= 3
node number = 5
degree of freedom for spc = 1
constrained value = 0

number of applied forces and moment = 1

i= 1
node number = 8
degrees of freedom = 2
applied forces or moment = 1

refrencestiffness = 1.0e+06 * 0.0000 0.0002 0.0003 2.0899

ans =

Columns 1 through 7
0 0 -0.0002 0.0389 0.0042 -0.0001 0.0000

Columns 8 through 14
0.0083 0.0004 -0.0389 0.0083 -0.0001 0 0.0083

Columns 15 through 21
-0.0002 0.0000 0.0084 0.0007 0.0000 0.2289 0.0014

Columns 22 through 24
0.0000 0.5490 0.0017

nel = 1
strainenergynel = 0.0073
axialstrain = 1.0e-04 * 0.1667 -0.0000 0.0124
axialstress = 0.0033 -0.0000 0.0002

nel = 2
strainenergynel = 0.0384
axialstrain = 1.0e-04 * 0.1667 0.0124 0.0249
axialstress = 0.0033 0.0002 0.0005

nel = 3
strainenergynel = 0.0363
axialstrain = 1.0e-05 * 0.0000 -0.2488 -0.1244
axialstress = 1.0e-03 * 0.0000 -0.4976 -0.2488

nel = 4
strainenergynel = 0.0052
axialstrain = 1.0e-05 * 0 -0.1244 0.0000
axialstress = 1.0e-03 * 0 -0.2488 0.0000

nel = 5
strainenergynel = 0.1161
axialstrain = 1.0e-05 * 0 0.4976 0.2488
axialstress = 1.0e-03 * 0 0.9953 0.4976

nel = 6
strainenergynel = 0.0166
axialstrain = 1.0e-05 * 0.0000 0.2488 0
axialstress = 1.0e-03 * 0.0000 0.4976 0

nfj = 1
strainenergynfj = 0.0946

strain energy distribution

0.12

0.1

0.08
strain energy

0.06

0.04

0.02

0
0 1 2 3 4 5 6 7 8
beam elements and flexible joints

magnification factor of the fisplacement = 300

 deformed configuration
450

400

350

300

250
Y

200

150

100

50

-50
-600 -400 -200 0 200 400 600
Z

Example 2 We shall consider another plane frame structure consisting of beam

and flexible joint elements shown in the following figure :
 Y

a a a a
b

800 mm

1200 mm

a - a cross section b - b cross section

Z
Y

40 mm
X
100 mm X

80 mm

60 mm
thickness 1 mm

Assuming Youngs modulus E = 200 kN/mm2 and zero shear constant Fy = 0, we

have

a-a cross section

 A = 236 mm2
Iyy = 69999 mmm4

b-b cross section

A = 316 mm2
Iyy = 450910 mm4

Assuming two flexible joints with 70% and 80% rigidity for rotation, while full
rigidity is asigned in the axial and transverse displacements, we shall make up a finite
element model for the first order analysis by using four beam elements and two
flexible joint elements.

4 5
(3)

(2) (4)

3 6
<1> <2>

1
(1) 2

For the shear loading at the top beam element, we apply 5 kN horizontal forces at the
end points ( that is, node 4 and node 5 ). Input data to the MATLAB program
becomes as follows :

Is data in the datafile ? [y/n] = n

number of nodes of the whole structure = 6

i= 1
Z coordinate of node = 0
Y coordinate of node = 0

i= 2
Z coordinate of node = 1200
Y coordinate of node = 0

i= 3
Z coordinate of node = 0
Y coordinate of node = 0

i= 4
Z coordinate of node = 0
Y coordinate of node = 800

i= 5
Z coordinate of node = 1200
Y coordinate of node = 800

i= 6
Z coordinate of node = 1200
Y coordinate of node = 0

number of beam elements = 4

nel = 1
node i = 1
node j = 2
section type = 1

nel = 2
node i = 3
node j = 4
section type = 2

nel = 3
node i = 4
node j = 5
section type = 1

nel = 4
node i = 5
node j = 6
section type = 2

total number of section type = 2

i= 1
Young,s modulus / E = 200
crossectional area / A = 316
moment of inertia about the x axis / Iyy = 450910
shear constant / Fy = 0

i= 2
Young,s modulus / E = 200
crossectional area / A = 236
moment of inertia about the x axis / Iyy = 69999
shear constant / Fy = 0

number of flexible joints = 2

i= 1
node i of the flexible joint = 1
node j of the flexible joint = 3
stiffness percent in the Z direction / kZ = 10
stiffness percent in the Y direction / kY = 10
stiffness percent about the X axis rotation / kqX = 0.7

i= 2
node i of the flexible joint = 2
node j of the flexible joint = 6
stiffness percent in the Z direction / kZ = 10
stiffness percent in the Y direction / kY = 10
stiffness percent about the X axis rotation / kqX = 0.8

number of single point constraints = 3

i= 1
node number = 1
degree of freedom for spc = 1
constrained value = 0

i= 2
node number = 1
degree of freedom for spc = 2
constrained value = 0

i= 3
node number = 2
degree of freedom for spc = 1
constrained value = 0

number of applied forces and moment = 2

i= 1
node number = 4
degrees of freedom = 2
applied forces or moment = 5

i= 2
node number = 5
degrees of freedom = 2
applied forces or moment = 5
refrencestiffness =

1.0e+05 *

0.0000 0.0003 0.0003 1.8530

refrencestiffness =

1.0e+05 *

0.0000 0.0003 0.0003 1.8530

ans =

Columns 1 through 7

0 0 0.0038 0 0.0957 0.0042 0.0123

Columns 8 through 14

0.0187 0.0174 0.0747 24.0790 0.0050 -0.0747 24.0782

Columns 15 through 18

0.0051 -0.0123 0.1148 0.0164

nel = 1
strainenergynel = 7.3671
axialstrain = 1.0e-04 * 0.7979 -0.1958 0.2016
axialstress = 0.0160 -0.0039 0.0040

nel = 2
strainenergynel = 38.9183
axialstrain = 1.0e-03 * 0.0780 0.1261 -0.1572
axialstress = 0.0156 0.0252 -0.0314

nel = 3
strainenergynel = 10.8133
axialstrain = 1.0e-04 * -0.0067 -0.2440 0.2457
axialstress = -0.0001 -0.0049 0.0049
nel = 4
strainenergynel = 39.9879
axialstrain = 1.0e-03 * -0.0780 0.1583 -0.1299
axialstress = -0.0156 0.0317 -0.0260

nfj = 1
strainenergynfj = 20.7920

nfj = 2
strainenergynfj = 20.7402

strain energy distribution

40

35

30

25
strain energy

20

15

10

0
0 1 2 3 4 5 6 7
beam elements and flexible joints
 deformed configuration
900

800

700

600

500
Y

400

300

200

100

-100
0 500 1000 1500
Z

References

1. Gallagher, R.H., Finite Element Analysis : Fundamentals, Prentice-Hall, 1976

2. Martin, H.C., Introduction to Matrix Methods of Structural Analysis, McGraw-Hill,

1966