You are on page 1of 11

Home Search Collections Journals About Contact us My IOPscience

A consistent beam element formulation considering shear lag effect

This article has been downloaded from IOPscience. Please scroll down to see the full text article.

2010 IOP Conf. Ser.: Mater. Sci. Eng. 10 012211

(http://iopscience.iop.org/1757-899X/10/1/012211)

View the table of contents for this issue, or go to the journal homepage for more

Download details:
IP Address: 89.122.75.79
The article was downloaded on 06/09/2010 at 17:01

Please note that terms and conditions apply.


WCCM/APCOM 2010 IOP Publishing
IOP Conf. Series: Materials Science and Engineering 10 (2010) 012211 doi:10.1088/1757-899X/10/1/012211

A Consistent Beam Element Formulation Considering Shear


Lag Effect

E Nouchi1, M Kurata1, Buntara S Gan1,3and K Sugiyama2


1
Department of Architecture, College of Engineering, Nihon University, 1-
Nakagawara, Koriyama, Fukushima, 963-8642, Japan.
2
Department of Architecture, Graduate School of Engineering, Nihon University, 1-
Nakagawara, Koriyama, Fukushima, 963-8642, Japan.

E-mail: buntara@arch.ce.nihon-u.ac.jp

Abstract. This paper presents general solutions for beam elements which are derived
analytically by considering the effect of shear lag phenomenon. The mass and stiffness
matrices which can be obtained from the general solution of displacements can be assembled
for general frame analysis by using finite element procedure.

1. Introduction
In studying and evaluating the ultimate strength of steel frame structures, it is apparent that the
behavior of steel frame structures at failure is fully controlled by consecutive local buckling failures of
its members. Hence, to evaluate the local buckling of each member element, especially thin-walled
beam, the stress and deformation at section should be considered properly by taking into account the
effect of shear lag phenomenon.
The classical theory of beam and thin-walled members was unable to reflect the shear lag
phenomenon since it was based on the assumption of cross section remains plane after deformation.
The effect of shear lag results in a distribution of direct stresses in the cross section which is different
from that predicted by the classical theory of beam. Therefore, the formulations of most analytical
and/or semi-empirical methods involved many simplified assumptions. As a result, they can only
handle shear lag problems for a particular simple geometry and cannot be easily extended to structures
with complex geometry.
The energy approach has been proven to be a simple and practical method in shear lag analysis [1].
Later, the shear lag phenomenon was investigated in box-girders using finite elements [2]. A
combination of finite element and transfer matrix techniques are proposed by [3]. Introduction of
basis warping function in the finite element was used to explain the shear lag phenomenon [4].
However, all of these approaches are mainly suitable for a particular type of cross section and static
analysis, thus it is not applicable for general applications. When the shear lag phenomenon is to be
considered for solving dynamic analysis, the mass property has to be introduced and/or approximated.
This is due to the short of fundamental theories which can include the effect of shear lag directly in the
formulation [5]. Present paper addresses a consistent beam element formulation considering shear lag
phenomenon to deal with the difficulties.

3
To whom any correspondence should be addressed.


c 2010 Published under licence by IOP Publishing Ltd 1
WCCM/APCOM 2010 IOP Publishing
IOP Conf. Series: Materials Science and Engineering 10 (2010) 012211 doi:10.1088/1757-899X/10/1/012211

2. General solution for beam with solid cross section considering shear lag
In this section, some fundamental equations of an elastic beam with solid cross section lying on a
right-handed axes coordinates convention considering shear lag will be derived.
Displacement at any arbitrary point on the cross section beam with member axis in the x direction
is given by U. The transversal displacements of the arbitrary point in the y and z direction are
expressed by V and W, respectively. The following assumptions are made for the arbitrary point at the
cross section of a plane beam in deriving equations.
U = u + yβ + f (2.1a)
V = V ( x, y ) = v ( x ) = v , W = W ( x , y ) = 0 (2.1b)
where, U = U ( x, y ) , u = u (x ) and β = β (x) in which, f = f ( x, y ) which is defined as a 2nd
order or higher function in y direction.
Hence, strain and displacement relationships can be given as,
∂U ∂f ∂V ∂U ∂f
εx = = ε + yκ + , γ xy = + = γ0 + (2.2)
∂x ∂x ∂x ∂y ∂y
∂V ∂W ∂W ∂V ∂W ∂U
εy = =0 , εz = =0 , γ yz = + = 0 , γ xz = + =0
∂y ∂z ∂y ∂z ∂x ∂z
which are obtained from the assumption that the cross section remain straight after deformation

where, ε =
du
dx
,κ =
dx
[ ]
and γ xy y =0 = γ 0 =
dv
dx
+ β . Here, ε x , ε y , ε z are normal strains in x, y and
z directions and γ xy , γ yz , γ xz are engineering shear strains on x-y, y-z and x-z planes. It can be noted
from equations (2.1) that the displacement U and V are independent of z axis and y-z plane,
respectively.
Assuming elastic material following the Hooke’s law, the shear strain γ xy can be related to the
shear stress τ xy from the relationship τ xy = Gγ xy , hence the function f from equation (2.2) can be
obtained as,
1 y
(τ xy − τ 0 ) dy
G ∫0
f = (2.3)

τ0
where, G is the shear modulus of elasticity and γ 0 is given by γ 0 = .
G
Substituting the function f into equation (2.1a) and (2.2), the normal strain and stress equations are
then given as,
∂U 1 y ∂
εx = = ε + yκ + ∫ (τ xy − τ 0 ) dy (2.4a)
∂x G 0 ∂x
E y ∂
σ x = Eε + Eyκ + ∫ (τ xy − τ 0 ) dy (2.4b)
G 0 ∂x
where, E is the normal modulus of elasticity
By integrating the normal stress σ x for the entire cross sectional area, the following axial force and
bending moment can be obtained.
y ∂
E
N = ∫ σ x dA = EAε + ∫ ∫ (τ xy − τ 0 ) dydA (2.5a)
A G A 0 ∂x
y ∂
E
M = ∫ yσ x dA = EIκ + ∫ y ∫ (τ xy − τ 0 ) dydA (2.5b)
A G A 0 ∂x
where, A = ∫A
dA , S = ∫ ydA = 0 I = ∫ y 2 dA .
A A

2
WCCM/APCOM 2010 IOP Publishing
IOP Conf. Series: Materials Science and Engineering 10 (2010) 012211 doi:10.1088/1757-899X/10/1/012211

By substitution of ε and κ from equations (2.5a) and (2.5b) into (2.4b), results in the following
normal stress equation.
N M
σx = + y + στ (2.6)
A I
Here, σ τ is the effect of shear lag in the normal stress which is given as follow.
y ∂ ∂
στ = −
E
∫ ∫ (τ xy − τ 0 ) dydA − y E ∫ y∫
y
(τ xy − τ 0 ) dydA + E ∫0y ∂ (τ xy − τ 0 ) dy (2.7)
GA A 0 ∂x GI A 0 ∂x G ∂x
The dynamic equilibrium equation at an arbitrary point can be given as follow.
∂σ x ∂τ xy ∂τ xz
+ + + X − ρU&& = 0 with τ xz = Gγ xz = 0 (2.8)
∂x ∂y ∂z
By integration, the shear stress τ xy in the equation (2.8) can be given as,
1 e ⎛ ∂σ x ⎞
τ xy = ∫ ⎜
b y ⎝ ∂x
+ X − ρU&& ⎟ bdy

(2.9)

∂ 2U
where, X is the body force, ρ is the material density, U&& = is the acceleration with t in time,
∂t 2
b = b( y ) is width of cross section as a function of y, e is defined as a distance to the outermost fiber of
[ ]
cross section where τ xy y =e = 0 .
Dynamic governing differential equations for the analysis of a beam by considering first order
solution of normal stress σ x which correspond to displacement u,v and deflection β are given by,
N ′ + q x − ρAu&& = 0 , M ′ − Q + m − ρIβ&& = 0 , Q′ + q y − ρAv&& = 0 (2.10)

where, ( )′ = d ( ) , Q = ∫Aτ xy dA , q x = ∫AXdA = XA , q y = ∫AYdA and m = ∫A yXdA = 0 .


dx
The solution of member forces can be obtained by integrating the equations (2.10) to reach the
following equations.
N = N (0) − ∫ (q x − ρAu&&) dx , Q = Q (0) − ∫ (q y − ρAv&&) dx
x x

0 0
(2.11)
∫ (q − ρAv&&) dxdx + ∫ ρIβ&&dx
x x x
M = M (0) + Q (0) x − ∫ y
0 0 0
Defining difference of shear stresses in the following form,

= (S − S0 ) = −
QS QS0 Q QS
τ xy − τ 0 = − (2.12)
bI bI bI bI
QS QS 0
where, τ xy = and τ 0 = . Here, first moment of areas from 0 and or y to e of the cross
bI bI
e e
∫ ybdy , S = ∫ ybdy and S = ∫ ybdy .
e
section are defined by S = 0
y 0 y

Substituting equation (2.12) into equations (2.5a) and (2.5b), resulted in,
N = EAε −
E
(− q y + ρAv&&)∫A ∫0y SdydA
GbI
(2.13)
M = EIκ −
E
(− q y + ρAv&&)∫A y ∫0y SdydA
GbI

3
WCCM/APCOM 2010 IOP Publishing
IOP Conf. Series: Materials Science and Engineering 10 (2010) 012211 doi:10.1088/1757-899X/10/1/012211

from where, ε and κ can be obtained as follow.

ε=
du N
= +
1
dx EA GAbI
(− q y + ρAv&&)∫ ∫ SdydA
A 0
y

(2.14)
dβ M
κ= = +
1
2
(− q y + ρAv&&)∫A y ∫0y SdydA
dx EI GbI
Finally, the displacements u, β and v of a beam considering shear lag phenomenon can be
obtained as,
N
EA
k x
dx + 0 ∫0 − q y + ρAv&& dx + u (0)
GA
u = ∫0
x
(2.15a) ( )
xM k x
β = ∫0 dx + 1 ∫0 − q y + ρAv&& dx + β (0)
EI GA
(2.15b) ( )
dxdx + 0 Q(0) x + 1 ∫ ∫ (− q y + ρAv&&)dxdx − β (0) x + v(0)
x x M k k x x
v = −∫ ∫ (2.15c)
0 0 EI GA GA 0 0
1 y A y S A S A
where, k 0 = ∫A ∫0
SdydA , k1 = 2 ∫A y ∫0 SdydA , k0 = 0 and k1 = 0 − k1 .
bI bI bI bI
It can be shown that for a solid rectangular beam section with dimension of b × h , the values of
normal and bending shear lag coefficients k0 = 0.0 , k1 = 0.3 , k0 = 1.5 and k1 = 1.2 computed are
agree with the common shear lag coefficients reported elsewhere.
By using the equations in (2.11) which are based on the linear first order approximation, the
dynamic governing equations, i.e. mass matrix, stiffness matrix and loading vector, of a beam element
considering shear lag effect which connecting beam’s nodal displacements with displacements along
the member axis can be derived explicitly by using the displacements given in equations (2.15).

2.1. Cantilever beam example


Cantilever beams, shown in figure 1, are subjected to concentrated load P at the free end and subjected
to distributed load q are considered. It is noted that the transverse vertical displacement v result for
both cases are different with the classical beam theory. It is also shown that the difference between
stresses computed by the present method and the classical beam theory is quite significant, i.e. the
effect of shear lag must be considered in engineering design.
E = 205000
N/mm2
100 mm

P = 1000 N q = 10 N/mm

50 mm
500 mm 500 mm

-50 -50
y ( mm )

y ( mm )

x ( mm ) x ( mm )
0 100 200 300 400 500 -25 -25
0 100 200 300 400 500
0.00 0.00
-20 0 20 -0.1 -0.05 0 0.05 0.1
0.01 0.02
0 σx ( MPa ) 0
0.02 0.04 f(×10-5 )
v ( mm )

v ( mm )

0.03 0.06

0.04 0.08 25 25
Considering shear lag effect Considering shear lag effect
0.05 0.10
Classical beam theory Classical beam theory
0.06 0.12
50 50
Rectangular beam deflection ( concentrated load ) Rectangular beam deflection ( distributed load )
Normal stress distribution Function f(y)

Figure 1. Rectangular section cantilever beam example.

4
WCCM/APCOM 2010 IOP Publishing
IOP Conf. Series: Materials Science and Engineering 10 (2010) 012211 doi:10.1088/1757-899X/10/1/012211

3. General solution for thin-walled beam considering shear lag


In this section, the fundamental equations of an elastic plane thin-walled beam either open or close
section which lying on a right-handed x, y, z Cartesian and an auxiliary ς , s coordinate systems by
considering shear lag will be derived.
In figure 2, point O is the centroid of a thin-walled beam cross section and lying along the x
coordinate axis. This point O will become the origin of the y and z coordinates axes. Axis s is defined
along the center line of the thickness t of the curvature thin-walled beam, from where axis n which is
drawn normal to this curvature line is defined as t ( s ) which is a function of s.
Displacements at any arbitrary point P ( x, y, z ) on the cross section of thin-walled beam are given
) ) )
by U ,V ,W in the x, y, z directions and U ,V ,W in the x, s, ς directions. Here, the displacement u in
the x direction is referred to a point where s=0. The transversal displacements vc , wc in
the y, z directions are originated at point C which is the shear center of the cross section.

z
t
ς
o x,U
z
θ yc
rs s
y
) )
c wc p, U V
zc W

vc V p′
)
s=0 −W
s

y
Figure 2. Cross section of a thin-walled beam.

The strain and displacement relationships are given as follow.


) ) )
∂U ∂V ∂U
εx = and γ xs = + (3.1)
∂x ∂x (1 − ς r )∂s
where, the radius curvature of the thin-walled beam cross section is given by r = r (s ) .
The displacements of the point P ( x, y , z ) can be assumed as follow.
U = u + yβ z + zβ y + f , V = vc − ( z − zc )φ and W = wc + ( y − yc )φ (3.2)
From figure 2, the displacements can then be expressed in terms of x, s, ς coordinates system,
)
U = U = u + yβ z + zβ y + f (3.3a)
)
V = V cos θ + W sin θ = vc cos θ + wc sin θ + rς φ (3.3b)
)
W = −V sin θ + W cosθ = −vc sin θ + wc cosθ + rsφ (3.3c)
where, rς = ( z − z c )sin θ + ( y − y c )cosθ and rS = ( z − z c )cosθ − ( y − y c )sin θ .
The following assumptions can be assumed for any arbitrary point at the cross section of the thin-
walled beam.
) )
∂V ∂U
γ xs = + (3.4)
∂x ∂s

5
WCCM/APCOM 2010 IOP Publishing
IOP Conf. Series: Materials Science and Engineering 10 (2010) 012211 doi:10.1088/1757-899X/10/1/012211

By substituting equation (3.4), the function f in equation (3.3a) can be solved as follow.
f ( s ) = ∫ (γ xs − (γ y cosθ + γ z sin θ ))(1 − ς r )ds + ωχ + f (0)
s
(3.5)
0
dvc dw dφ
In which, γ y = + βzc , γ z = c + β y , χ = − and ω = ∫0 rς (1 − ς r )ds are defined.
s

dx dx dx
Accordingly, the average normal displacement of the thin-walled beam which can be expressed by
1 )
A ∫A
using u = UdA is given as,
)
U = U = u + yβ z + zβ y + ωχ + F (3.6)
1 s
(γ xs − (γ y cosθ + γ z sin θ ))(1 − ς r )dsdA .
A ∫A ∫0
where, F = −

The normal strain is then given by,


ε x = u′ + yβ z′ + zβ y′ + ωχ ′ + F ′ (3.7)
Hence, the normal stress σ x for an elastic material following Hooke’s law can be obtained by,
σ x = Eε x = Eu′ + Eyβ z′ + Ezβ y′ + Eωχ ′ + EF ′ (3.8)
The first order solution of normal stress σ x in terms of axial force and moments of the beam can be
expressed by,
N I y M z − I yz M y I M − I yz M z M
σx = + y+ z y z+ ωω (3.9)
A I y I z − I yz
2
I y I z − I yz
2

where, the internal forces and section properties are given as follow: N = ∫ σ dA ;
A
x

M z = ∫ yσ x dA ; M y = ∫ zσ x dA ; M ω = ∫ ωσ x dA ; A = ∫ dA ; S z = ∫ ydA = 0 ; S y = ∫ zdA = 0 ;
A A A A A A

S ω = ∫ ωdA = 0 ; I z = ∫ y dA ; I y = ∫ z dA ; I yz = ∫ yzdA ; I yω = ∫ yωdA = 0 and


2 2
A A A A A

I zω = ∫ zωdA = 0 .
A
The dynamic equilibrium equation of any arbitrary point on the cross section is given as below.
∂σ x ∂τ xs ∂τ xς ) )
+ + + X − ρU&& = 0 (3.10)
∂x (1 − ς r )∂s ∂ς
Substitution of equation (3.9) into equation (3.10), the shear stress τ xs can be solved as,
∂σ x s )
(1 − ς r )ds − ∫ X (1 − ς r )ds + ρ ∫ U&& (1 − ς r )ds + τ xs ( x,0, ς )
s s
τ xs = − ∫ (3.11)
0 ∂x 0 0

where, τ xs ( x,0, ς ) is the indeterminate shear stress.


) )
The body force X = X ( x) and applied moment per unit length terms are m x , m y , m z = 0 , the
first order approximation of shear stress can be given as,
4
τ xs = ∑ g~i Qi (3.12)
i =1
~ = g~, g~ = g~ (θ ), g~ = g~ (θ ), g~ = g~ (θ ) .
where, Q1′ = − χ ′, Q2′ = Q′y , Q3′ = Qz′ , Q4′ = Tω′ , g1 2 y 3 z 4 ω

6
WCCM/APCOM 2010 IOP Publishing
IOP Conf. Series: Materials Science and Engineering 10 (2010) 012211 doi:10.1088/1757-899X/10/1/012211

Herewith, all the functions in equation (3.12) are given explicitly as:
) ) ) )
2Ω I y S z − I yz S y I z S y − I yz S z )
g~ = G ; g 2 (θ ) = ; g 3 (θ ) = ; g 4 (θ ) = Sω ;
∫ (1 − ς r )rdθ I y I z − I yz I y I z − I yz
2 2

τ y = [τ xs ]dφ =0, df =0,θ =0 = g~y (0)Qy + g~z (0)Qz ; τ z = [τ xs ]dφ =0, df =0,θ =π = g~y (π 2)Qy + g~z (π 2)Qz ;
dx dx dx dx 2
) θe ) θe ) 1 θ
S z (θ ) = ∫ y (1 − ς r )rdθ S y (θ ) = ∫ z (1 − ς r )rdθ Sω (θ ) = ∫θ ω (1 − ς r )rdθ
e
; ; ;
θ θ Iω
2Ω = ∫ rτ (1 − ς r )rdθ ; τ z = Gγ z = G (wc′ + β y ) and τ y = Gγ y = G (vc′ + β z ) .
Furthermore, F ′ in the equation (3.8) can be solved as,
1 4
F′ = ∑ aiQi′
G i=1
(3.13)

Herewith, all the functions in equation (3.13) are given explicitly as: a1 = a (θ ) ; a2 = a y (θ ) ;
a3 = az (θ ) ; ( ) ( ) (
a4 = aω (θ ) ; ai = Gi (θ ) − Gˆ i (θ ) − g~i 2 (0) C (θ ) − Cˆ (θ ) − g~i 3 (π 2) S (θ ) − Sˆ (θ ) ; )
θ 1
g~12 (0) = g~13 (π 2) = g~42 (0) = g~43 (π 2) = 0 ; Gi (θ ) = ∫ g~i (1 − ς r )rdθ ; Gˆ i (θ ) = ∫ Gi (θ )dA ;
0 A A
θ 1 θ
C (θ ) = ∫ cos θ (1 − ς r )rdθ ; Cˆ (θ ) = ∫ C (θ )dA ; S (θ ) = ∫ sin θ (1 − ς r )rdθ ;
0 A A 0

1
Sˆ (θ ) = ∫ S (θ )dA .
A A
The differential governing equations of x-derivative displacements can also be given as,
u′ =
N

1
(k12Q′y + k13Qz′ + k14Tω′ − k11χ ′) (3.14a)
EA GA
Mˆ I − Mˆ y I yz Mˆ y I z − Mˆ z I yz
β z′ = z y , β ′ = (3.14b)
I y I z − I yz2 I y I z − I yz2
y

χ′ =
(


) (
1
(k42Q′y + k43Qz′ + k44Tω′ )
) (3.14c)
EI ω 1 − k 41 GIω GIω − k 41
1
(k11Qy + k21Qz )
vc′ = − β z + (3.14d)
GA
wc′ = − β y +
1
(k12Qy + k22Qz ) (3.14e)
GA
are defined by Mˆ z = M z − (k22Q′y + k 23Qz′ + k 24Tω′ − k 21 χ ′) and
E
where, M̂ y and M̂ z
G
Mˆ y = M y − (k32Q′y + k33Qz′ + k34Tω′ − k31 χ ′) .
E
G
The shear lag coefficients are then can be given as follow.
~
k jk = λ j ak dA ( j ,k =1, 2,3, 4 ) and k mn = Ag~mn
∫ ( m ,n=1, 2 ) (3.15)
A

here, λ1 = 1 , λ2 = y , λ3 = z , λ4 = ω , a1 = a (θ ) , a2 = a y (θ ) , a3 = a z (θ ) , a4 = aω (θ ) ,
g11 = g~y (0) , g~21 = g~z (0) , g~12 = g~y (π 2) , g~22 = g~z (π 2) .
~

7
WCCM/APCOM 2010 IOP Publishing
IOP Conf. Series: Materials Science and Engineering 10 (2010) 012211 doi:10.1088/1757-899X/10/1/012211

Since the internal forces N , Qy , Qz , M z , M y , M ω can be determined from the first order
approximation of the dynamic governing equation, the bimoment M ω can be solved from,
⎛ GJ ⎞ d 2M ω GJ
⎜⎜1 + kmωω ⎟⎟ − Mω
⎝ (GIω − kmω ) ⎠ dx 2
EIω (1 − kmω GIω )
⎛ ⎞
=−
GJ
(GIω − kmω )
(kmω yQ′y + kmω z Qz′ ) − ⎜⎜1 + kmωω
GJ
⎟(m′ + ρIω χ&&′)
(GIω − kmω ) ⎟⎠ ω
(3.16)

x (
− m + ρ S v&& + S w
yc c
&& + I φ&& zc c pc )
where, S yc = ∫ (− rς sin θ − r
A
s cos θ ) dA , S zc = ∫ (rς cos θ − r sin θ ) dA , I
A
s pc = ∫ (rς
A
2
)
+ rs2 dA φ&& .
From the above equation, the right hand side of the second order differential equation are known

values, thus the general solution of the x-derivative twist χ , which is defined by φ ′ = = − χ , can
dx
be solved. The general solution of the χ can then be used to derive the element mass and stiffness
matrices in the finite element procedure.

4. Pipe beam example considering shear lag


To examine a close section example of thin-walled beam, a solution for pipe beam shown in figure 3 is
derived by following the same procedures for obtaining the general solution of thin-walled beam,
ς
) s
) V
U =U W

x ς r P z

) z θ
θ ) V s
W V
dz
) ) θ
U =U s P ( x, s , ς ) −W
dy
(r − ς )dθ
y y y

Figure 3. Thin-walled circular beam

The displacements of the centroid cross section of the pipe beam and the x, y coordinate system
can be expressed in terms of rotational angles β x = φ , β y , β z as follow.
U = u + yβ + f , V = v and W = 0 (4.1)
where, β z = β , β y = 0 . By assuming w = 0 , φ = 0 , y = (r − ς ) sin θ and z = −(r − ς ) cosθ , the
displacements become U = U , V = v and W = 0 . Hence, equation (4.1) yields to the following
equations.
) ) )
U = U = u + yβ + f , V = v cosθ and W = −v sin θ (4.2)
The strain and displacement relationships for pipe beam can further be expressed as,
) ) ) )
∂U 1 ⎛ ∂V ) ⎞ ∂V 1 ∂U
εx = , εθ = ⎜ − W ⎟⎟ and γ xθ = + (4.3)
∂x (r − ς ) ⎜⎝ ∂θ ⎠ ∂x (r − ς ) ∂θ
Hence the function f from equation (4.1) can be solved for,
θ
f (θ ) = ∫ (γ xθ − γ cosθ )(r − ς )dθ (4.4)
0

8
WCCM/APCOM 2010 IOP Publishing
IOP Conf. Series: Materials Science and Engineering 10 (2010) 012211 doi:10.1088/1757-899X/10/1/012211

Substitution of equation (4.2) into equation (4.3) results in the following normal strain and stress
equations.
∂U 1 θ
εx = = u′ + yβ ′ + ∫ (τ ′xθ − τ 0′ cosθ )(r − ς )dθ (4.5a)
∂x G 0
E θ
σ x = Eε x = Eu′ + Eyβ ′ + ∫ (τ ′xθ − τ 0′ cos θ )(r − ς )dθ (4.5b)
G 0
By integrating the normal stress σ x for the entire cross sectional area, the following axial force and
bending moment can be obtained.
E θ
N = ∫ σ x dA = EAu′ + (τ ′xθ − τ 0′ cosθ )(r − ς )dθ dA
G ∫A ∫0
(4.6a)
A

θ
M = ∫ yσ x dA = EIβ ′ + ∫ (r − ς )sin θ ⎛⎜ ∫ (τ ′xθ − τ 0′ cos θ )(r − ς )dθ ⎞⎟dA
E
(4.6b)
A G A ⎝ 0 ⎠
For pipe beam section, σ τ = 0 , the first order solution of normal stress σ x becomes,
N M
σx = + y (4.7)
A I
The dynamic equilibrium equation at an arbitrary point can be given as follow.
∂σ x ∂τ xθ ∂τ
+ + xς + X − ρU&& = 0 (4.8)
∂x (r − ς )∂θ ∂ς
By integration, the shear stress τ xθ in the equation (4.8) yields to,
Q θ
yt (r − ς )dθ + τ xθ (0)
tI ∫0
τ xθ = − (4.9)

where, due to symmetric geometry of the cross section in y axis, from τ xθ (π 2) = 0 condition, the

at [τ xθ ]θ =0 in which, S = ∫ yt (r − ς )dθ and


QS QS 0 π 2
equation (4.9) resulted in τ xθ = and τ 0 =
tI tI θ
π 2
S0 = ∫ yt (r − ς )dθ are denoted.
0
From which, the following equations can be obtained as,
E
N = ∫ σ x dA = EAu ′ +rk n Q′
A G
(4.10)
EQ ′ EQ′I
M = ∫ yσ x dA =EIβ ′ +
Gt ∫A
yg (θ )dA = EIβ ′ + km
A GA
where, g (θ ) = ∫ (S − S 0 cos θ )(r − ς )dθ , k n = ∫ g (θ ) dA and k m = ∫ yg (θ ) dA are given.
1 θ 1 A
I 0 tr A tI A
Finally, the displacements u, β and v of the pipe beam considering shear lag phenomenon can be
obtained as,
dx + n ∫ (q y − ρAv&&)dx + u (0)
N x rk x
u=∫ (4.11a)
0 EA GA 0
dx + m ∫ (q y − ρAv&&)dx + β (0)
x M k x
β =∫ (4.11b)
0 EI GA 0
dxdx + τ Q (0) x − m ∫ ∫ (q y − ρAv&&)dxdx − β (0) x + v(0)
x x M k k x x
v = −∫ ∫ (4.11c)
0 0 EI GA GA 0 0

9
WCCM/APCOM 2010 IOP Publishing
IOP Conf. Series: Materials Science and Engineering 10 (2010) 012211 doi:10.1088/1757-899X/10/1/012211

AS 0
where, kτ = and k m = k m + kτ .
tI
Thus, the shear lag coefficients in the displacement solutions can be obtained as k n = 0 , k m = 0 ,
kτ ≈ 2.0 and k m ≈ 2.0 .

4.1. Cantilever beam example


Cantilever beams, shown in figure 4, are subjected to concentrated load P at the free end and subjected
to distributed load q are considered. It is noted that the transverse vertical displacement v results for
both cases are different with the classical beam theory. Thus, the effect of shear lag must be
considered in engineering design.

t x ( mm ) x ( mm )
0 50 100 150 200 250 0 50 100 150 200 250
0.00 0.00
0.05 0.05

r 0.10 0.10
v ( mm )

v ( mm )

0.15 0.15

0.20 0.20
Considering shear lag effect Considering shear lag effect
0.25 Classical beam theory 0.25 Classical beam theory
0.30 0.30

Pipe beam deflection ( concentrated load ) Pipe beam deflection ( distributed load )

Figure 4. Pipe beam section cantilever beam example.

5. Conclusions
Based on the classical beam theory where the effect of shear lag is included in the formulations of
displacements, strains and stresses of solid, thin-walled and pipe beam consistently without using any
kind of stress approximation functions. The shear lag phenomenon is now can be considered directly
for solving dynamic problems in which the mass and stiffness matrices can be derived consistently.

References
[1] Hadji-Argyris J and Cox H L 1944 Diffusion of load into stiffened panels of varying section Br.
Aeron. Res. Council Reports. Mem. 1969
Reissner E 1946 Analysis of shear lag in box beams by the principle of minimum potential
energy Q. Appl. Math. 4 268-78
[2] Malcolm D J and Redwood R G 1970 Sehar Lag in stiffened box-girders J. Struct. Div. ASCE
96 ST7 1403-15
Moffatt K R and Dowling P J 1975 Shear lag in steel box-girder bridges Struct. Engineer 53
439-48
[3] Tesar A 1996 Shear lag in the behavior of thinwalled box bridges Comp. and Struct. 4 607-12
[4] Prokić A 1996 New warping function for thin-walled beams I: Theory J. Struct. Eng. ASCE 122
ST12 1437-42
Prokić A 1996 New warping function for thin-walled beams II: Finite element method and
applications J. Struct. Eng. ASCE 122 ST12 1443-52
Prokić A 2002 New finite element for analysis of shear lag Comp. and Struct. 80 1011-24
[5] Kurata M 2009 General solution of beam considering shear lag phenomenon (in Japanese)
Summaries of Tech. Papers of Research Annual Meeting Nihon Univ. College of Eng. 2009
1-2

10