You are on page 1of 10

Composite Structures 55 (2002) 147156

www.elsevier.com/locate/compstruct

Analysis of shear lag and shear deformation eects in laminated


composite box beams under bending loads
Wu Yaping
a

a,b,*

, Zhu Yuanlin a, Lai Yuanming a, Pan Weideng

State Key Laboratory of Frozen Soil Engineering, Cold and Arid Regions Environmental and Engineering Research Institute,
Chinese Academy of Sciences, Lanzhou 730000, People's Republic of China
b
College of Civil Engineering and Architecture, Lanzhou Railway University, Lanzhou 730070, People's Republic of China
c
Department of Geology, Lanzhou University, Lanzhou 730000, People's Republic of China

Abstract
A method for analyzing the shear lag and shear deformation eects on symmetrically laminated thin-walled composite box beams
under bending load is presented. The method is based on the theory of composite laminated plates and is deduced by means of the
principle of minimum potential energy, which makes the procedure simple and practical. The formulas given by this method not
only satisfy the equilibrium conditions on the cross-section of the thin-walled composite box beams but can be simplied to the
known formulas of thin-walled box beams with isotropic materials as well. Finally, an example is given and its numerical results are
analyzed and discussed. The values of vertical displacements and ply normal stresses obtained by this paper are compared with those
obtained by the nite element method (FEM) or previously published experimental and numerical results, respectively. A comprehensive analysis on the eects of shear lag and shear deformation is given for a simply supported thin-walled composite box beam
subjected to a centralized load at mid-span. 2002 Elsevier Science Ltd. All rights reserved.

1. Introduction
With the features of high eciency, low consumption
and good mechanical characteristics, thin-walled composite box beams have been broadly adopted in structural engineering. To meet the needs of the construction,
mechanical and aeronautic industries, the analysis on
mechanical behavior of thin-walled composite box beam
has became an active research area. The shell nite element analysis for composite box beams is very eective
but requires specialized training and is always dicult
for design engineers to analyze the contributions of
various factors (e.g. shear lag and shear deformation,
etc.) to the numerical results. Therefore, to develop a
comprehensive analysis approach for the engineering
design of composite box beams is desirable. Such a
procedure should allow designers to eciently perform
deep consideration of strength and stiness so as to
optimize the layer design of composite box beams.
Up to now, many analysis models have been formulated to analyze composite box beams with varying
levels of assumptions. Cheng et al. [1] presented a
*

Corresponding author.
E-mail address: wypsw@263.net (W. Yaping).

method of calculating the stiness of bending and torsion for thin-walled carbonepoxy box beams, in which
the structural couplings and transverse shear eect were
neglected. Chandra et al. [2] discussed the eects of
structural couplings for symmetric and anti-symmetric
box beams under bending, torsional, and extensional
loads. Song Min Jeon et al. [3] developed an analysis
model of large deection for the static and dynamic
analysis of composite box beams. In the above-mentioned researches, the ply stress of composite box beams
is not dealt with. In fact, because the composite box
beam consists of four composite panels in which the ply
stresses vary with the ply angle [4,5], the analysis of ply
stress is of particular importance for the strength design
of composite box beams. Davalos and Qiao [6] proposed
a computational approach for the analysis and optimal
design of thin-walled composite beams under bending
load, in which the structural coupling is eliminated by
restricting the o-axis plies to be balanced symmetrically, and the Timoshenko beam theory was used to
evaluate the beam deections [6]. In [6], the mid-surface
strains and resultant forces acting on the panels are
determined rstly and then ply strains and stress are
obtained through a post-process which cannot be expressed explicitly.

0263-8223/02/$ - see front matter 2002 Elsevier Science Ltd. All rights reserved.
PII: S 0 2 6 3 - 8 2 2 3 ( 0 1 ) 0 0 1 3 8 - 6

148

W. Yaping et al. / Composite Structures 55 (2002) 147156

The achievements above mentioned undoubtedly are


very constructive, but none of them concerned the shear
lag eect which should not be ignored for the analysis of
mechanical behavior of thin-walled box beams under
bending load (especially for the beams with wide-anges). As the shear lag eect can bring an uneven normal
stress distribution on the anges, it would remarkably
aect the strength design of thin-walled beams. To analyze the shear lag eect of thin-walled beams with
isotropic materials, Reissner [7] developed a variation
procedure based on the principle of minimum potential
energy in which the longitudinal displacement of anges
was assumed to be a quadratic parabolic curve, thereafter, many analogous research papers with the assumption of longitudinal displacement of anges in the
form of various polynomial curves have been published
over the past several decades [811]. Combining Reissner's method with the Timoshenko beam theory, Wu
Yaping [12] presented an analysis model in which both
the shear lag and shear deformation eects can be considered for thin-walled box beams with isotropic materials.
Some investigations on the shear lag eect in composite thin-walled beams can be found too. Takayanagi
et al. [13] studied the shear lag eect in CFRP I-beams
under three-point bending. Roberto and GangaRao [14]
presented a warping solution for shear lag in prismatic
thin-walled orthotropic composite beams. Both Takayanagi and Roberto did not deal with the ply stress of
panels of composite box beams.
From above, it can be seen that the analytic
procedures of investigating simultaneously shear lag,
shear deformation eects and ply stress of panels in
composite thin-walled box beams may still be lacking.
In the present research, the theory of composite
laminated plates is associated with Reissner's method
and Timoshenko beam theory to form an applied
analysis model for thin-walled composite box beams
under bending loads. In this model, shear lag eect;
shear deformation eect, and ply stresses; strains of the
anges in thin-walled composite box beams can be
investigated and expressed explicitly. The related formula is very concise and easy to use for the engineering
designer performing strength analysis. To validate the
theory of this paper, a comprehensive investigation on
the eects of shear lag and shear deformation, displacement, ply normal stresses is given for a simply
supported symmetric carbonepoxy box beam subjected a concentrated load P at the mid-span. The
numerical results of this paper are correlated with the
predictions of the nite element method (FEM) or
previously published experimental results. The results
obtained from the analyses of this paper can provide
reference for the design of related engineering structures.

2. Elemental assumption and stressstrain relationship


Fig. 1 shows the cross-sectional contour of a symmetric single-cell thin-walled composite box beam (the
ply laminas on opposite panels are mirror images with
respect to the mid-axis of the beam cross-section). We
dene a coordinate system (oxyz) with the x-axis as
longitudinal direction of the beam. The basic assumptions and the stressstrain relationships for structure
analysis of the beam are presented as follows.
1. For the thin-walled box beam, the vertical shear
force S is borne mainly by webs (side panels). Because
the thickness of the ange is very small, and stresses on
the surface of top or bottom ange are zero (e.g.
sxz 0; rz 0), the shear stress along the thickness of
the anges can be neglected (i.e. sxz  0), which means
the anges are in a state of plane stress. According to the
theory of composite laminated plates, the stressstrain
relationship of the ith lamina of a top or bottom ange
with n laminas can be dened in the form [4]
8 9
2
38 9
>
Q11 Q12 Q16 >
=
=
< rx >
< ex >
6
7
ry
4 Q21 Q22 Q26 5 ey
;
1
>
>
;
;
: >
: >
sxy i
Q16 Q26 Q66 i cxy i
where Qkl (k and l 1; 2; 6) denotes the o-axis stiness
of the ith lamina and can be determined by elastic
constants E1 ; E2 ; G12 ; V12 .
Owing to the assumption that there is no conguration distortion for cross-section of box beams, we set
ey 0 in the top and bottom anges, and ez 0 in the
webs. Eq. (1) can be simplied to
8 9
2
3
>
Q11 Q16 ( )
=
< rx >
6
7 ex
ry
4 Q21 Q26 5
:
2
>
>
cxy
: ;
i
sxy i
Q16 Q66 i
2. Similarly, the stressstrain relationship of the jth
lamina of a web with m laminas can be expressed as
follows:

Fig. 1. Cross-section of composite box beam.

W. Yaping et al. / Composite Structures 55 (2002) 147156

8 9
2
>
Q11
=
< rx >
6
rz
4 Q21
>
;
: >
sxz j
Q16

Q16

 
7 ex
:
Q26 5
cxz j
Q66 j

3. When the shear lag eect-induced nonuniform


distribution function of axial displacement along the
ange width is expressed by a cubic parabola [10,12],
with consideration of shear deformation, the axial displacement at any point of the beam section (see Fig. 1)
under vertical bending symmetric load can be given by
[12]




y3
U x; y; z ux z /y x
1
nx
uT ; 4
b3
where ux is the longitudinal translation of the crosssection; uT uT x; y; z denotes the longitudinal displacement due to bending-twisting structural coupling;
/y x is the angular rotation of the cross-section about
y-axis; and nx is the maximum angular rotation of the
top and bottom anges due to shear lag.
Apparent that twist angular of the cross-section
about the x-axis (shear center) due to bendingtwisting
structural coupling is /x x, thus, the horizontal and
vertical displacements at any point of the panels can be,
respectively, given as
V v

z/x x;

W w y/x x;

ogT
cTxs
ox

oU x; y; zi oV
3y 2
dy

3 zi nx cTxs ;
oy
ox
ds
b

oU x; y; zi
ox

u0 x zi /0y x

ex i


1



y3 0
ouT
;
x

n
3
b
ox

11

12

where dy=ds 1 for the top ange and dy=ds 1 for


the bottom ange.
The geometrical equations of webs can be obtained
from Eq. (4) by designating y b:

where r n for anges, r m for webs. Eq. (15) gives

q/0x x;

cxy i

where q h=2 for anges, q b for webs.


Applying s as the circumferential coordinate, which
starts from the middle of the top ange, of beam conguration and considering Eq. (7), we have
ouT
cTxs
os

On the mid-surface of the ith lamina of a ange, we


set z zi so the normal strain and shear strain of the ith
lamina are

oU x; b; z
ouT
u0 x z/0y
;
13
ox
ox
oW oU x; b; z
dz

W 0 x /y x cTxs ;
cxz
14
ox
oz
ds
where dz=ds 1 for the right web and dz=ds 1 for
the left web.
For the beam under vertical symmetric bending
load, Bredt shear stress current qT T =X 0, so we
have
r
r
X
X

qT
sTxsk tk
Q16 eTx Q66 cTxs k tk 0;
15

where v; w are the translations of the beam cross-section


in the y- and z-directions, respectively, for the problem
of this paper v 0.
The /x x-induced circumferential displacement of
the panels is
gT q/x x;

149

where 0 d=dx; s y for the top ange, s y for


the bottom ange, s z for the right web, s z for the
left web. Applying a closed integral to Eq. (8) yields
I
1
cTxs ds;
/0x x
9
X
H
where X q ds. Integrating Eq. (8) and considering
Eq. (7) give
H T
Z
Z S
cxs ds S
T
T
T
cxs ds
q ds;
10
u u0
X
0
0
where uT0 0 for the box beam shown in Fig. 1, when s
starts from the middle of the top ange [15].

ex

k1

cTxs

b16 eTx ;

where

eTx

k1

16

is the normal strain due to


,
r
r
X
X
b16
Q16k tk
Q66k tk :
k1

cTxs ,

k1

In this paper, to study the eect of shear lag denitely, the structural couplings of bendingtwisting and
shearingtwisting are eliminated by restricting the oaxis plies of the panels to be balanced symmetrically [6],
in fact, if only we set b16 0 in all panels of the box
beam we would have cTxs 0, uT 0, and duT =dx 0.
Noticing that Q16 h Q16 h and Q16 0
Q16 90 0, so we can set equal lamina thickness and
restrict the ply angle in every panel of the beam to be 0,
90, or h degree to meet the condition of b16 0.

3. Analysis
The strain energy of the top and bottom anges Us
and strain energy of the webs Ub can be given, respectively, as
Us

2n Z Z Z
1X
rx ex ry ey sxy cxy i dx dy dz;
2 i1

17

150

Ub

W. Yaping et al. / Composite Structures 55 (2002) 147156


2m Z Z Z
1X
rx ex rz ez sxz cxz j dx dy dz:
2 j1

18

D11 ;

A11

i1

Based on the foresaid assumptions, we have ey 0 in


the top and bottom anges, and ez 0 in the webs.
Substituting Eq. (2) into Eq. (17), and Eq. (3) into
Eq. (18), respectively, we have
2n Z Z Z 
1X
Us
Q11 e2x Q66 c2xy
2 i1

2Q16 ex cxy dx dy dz;
19

2n
X
i1

Ds11 ;

Ds66 ;

The external loading-induced potential energy Uq can


be given by
Z x2
Uq
qW dx M/ SW jxx21 ;
21
x1

where M Mx and S Sx denote, respectively,


bending moment about the y-axis and vertical shear
force, q is the vertical distributed loading (in z-direction)
applied along the x-axis of the beam.
The total potential energy of the beam is given as
P Us Ub Uq .
We combine Eqs. (11)(14) with Eqs. (19)(21) under
the condition of b16 0. Then implementing the integral
over the cross-section under the symmetric conditions
Qkl z Qkl z and Qkl y Qkl y, the total potential energy of the beam can be written as
Z 
1 x2
2
A11 u0 2Ab16 u0 W 0 /y
P
2 x1


Ab66 0
3 0 0 9 0 2
2
0 2
s
n / n
W /y D11 /y D11

2 y 14
a

9Ds 2 Ds16
66
2/0y n0 n qW dx
n
5b2
b
x
M/y SW x2 ;
22
1

in which D11 ; A11 are the bending stiness and axial


stiness of the beam, respectively; Ds11 ; Ds66 ; Ds16 are the
bending, shear lag, shear lagbending stiness of the top
and bottom anges; Ab66 ; Ab16 are the shear stiness, shearextension coupling stiness of webs (in factor, when
restricting the o-axis plies to be balanced symmetrically, we have Ab16 0); Ii ; Ai ; ti are the moment of inertia, area, thickness of the ith lamina of the top and
bottom anges, respectively; Ij ; Aj ; tj are the moment of
inertia about y-axes, area, thickness of jth lamina of the
webs; and a is the shear correction factor for cross-section of the webs, appreciatively, a 1. Some of the
constants are expressed as

2m
X

Q11 i Ai

Ds16

j1

Q11 j Ij ;

2m
X
j1

Q11 j Aj ;

2n
X
Q11 i Ii ;

2n
X

i1

i1

23

Q66 i Ii ;
24

i1

Ab66 ;
20

Q11 i Ii

2n
X
Q16 i Ii ;

2m Z Z Z 
1X
Ub
Q11 e2x Q66 c2xy
2 j1

2Q16 ex cxz dx dy dz:

2n
X

Ab16

2m
X
j1

Q66 j Aj ;

2m
X
Q16 j Aj :

25

j1

According to the principle of minimum potential


energy dP 0, and considering that if b16 is zero, Ab16 is
zero too, the governing dierential equations and natural boundary conditions of the system are given by
A11 u0 0;
D11 /00y

26
Ab66

3 s 00
D n
4 11

W 0 /

Ds16
b

n0 0;

Ab66
W 00 /0y q 0;
a
9Ds66
3 00
9 00
Ds16 0
/
n
n

/ 0;
4 y 14
5b2 Ds11
bDs11 y
 x2


3 s 0 Ds16
0
D11 n
n M d/y 0;
D11 /y
4
b
x1
 x2
 b

A66
W 0 /y S dW 0;
a
x1
 x2

s

D11
3 0 9 0
Ds
/y n 16s n dn 0:
2
7
2
bD11
x1

27
28
29
30
31
32

Integrating Eq. (28) under the natural boundary


conditions, Eq. (31), and dS=dx q, then
Ab66
W 0 /y S:
a

33

Substituting Eq. (33) into Eq. (27) then implementing


integration along the x-axis under natural boundary
conditions, Eq. (30), and dM=dx S gives
Ds16
n
b

3 s 0
D n D11 /0y M:
4 11

34

In fact, Eqs. (26), (33), (34) can also be obtained by


substituting Eqs. (2), (3) and Eqs. (11)(14) into equilibrium conditions of cross-section
Z
Z
Z
rx dA 0;
sxz dA S;
rx z dA M:
A

Combining Eqs. (29) and (34), the following governing equation for the shear lag analysis can be obtained:

W. Yaping et al. / Composite Structures 55 (2002) 147156

n00

k2n

7n
S
6D11

where

n1
a16

7Ds11

8D11

a16 M;

35

4Ds16
:
3bDs11

36

Substituting Eq. (34) into boundary condition Eq.


(32) gives

 x
7nM 2
n0 An
dn 0;
37
6D11 x1
where
7nDs16
A
6bD11


2D11
1 :
3Ds11

Function n nx is determined by Eqs. (35)(37),


the general solution of Eq. (35) can be expressed as
nx

7n
C1 shkx C2 chkx n ;
6D11

38

where n is the special solution related to Sx and Mx.


Integral constants C1 ; C2 are determined by boundary
conditions. Substituting Eq. (34) into Eq. (33), one has
W0

/y

aV
Wb0 Ws0 ;
Ab66

39

where Wb0 ; Ws0 are the contributions made by bending


(with the consideration of shear lag) and shear deformation to W 0 , respectively. Using Wb0 /y and integrating Eq. (34) two times yield

Z Z 
Mx 3Ds11 0
Ds16
Wb x
n x
nx dx dx

D11
4D11
bD11
C3 x C4 ;

40

in which integral constants C3 ; C4 are determined by


boundary conditions of Wb x. Likewise, integrating Ws0
in Eq. (39) gives
Ws x

rxi ri x; y; z ris rio k

v
"
#
u
s
s 2
u
1 14D66 n
5D16
;
k t
1
b 5bDs11
9D11 Ds66

a
Mx C5 ;
Ab66

41

where the integral constants C5 are determined by


boundary conditions of Ws x. The total vertical translation of the beam is
W x Wb x Ws x:

151

3Q16 i zi
Q zi M
yn 11 i
k;
3
D11
b

43

where ris , rio , and k are the coupling stress of shear laginduced transverse shear deformation in anges, traditional bending stress, and shear lag coecient [1012],
respectively. The shear lag coecient is expressed as
k kx; y
1

D11
M


1

y3
b3



3Ds11 0
Ds16
n :
n
4D11
bD11

44

In which function n nx is determined by Eqs. (37)


and (38).
For some simple composite beams and cantilever
composite beams, with length l, subjected to a vertical
concentrated load P or a uniform distributed vertical
load q along the x-axis, respectively, the parameters in
the function n nx (see Eq. (38)) are listed in Table 1.
In Table 1, we have


p k 2 la16
B1 3
A a16 ;
45
2k
2shkl=2



q kl
kl a16
a16 k A
B2 2
k
2
2
k



a16 l
l a16
A
1

2
2 k2

 Ashkl kchkl ;
46


p k 1 la16
A

a
;
16
k3
shkl

 2 2

q
k l
2
B4 3
1 a16 k l k
k
2
B3

47


Aa16
shkl :
k
48

From above it can be seen that the eects of various


coupling items are related to the o-axis stiness Q16 and
its corresponding coupling stiness Ab16 ; Ds16 .
For thin-walled box beams with isotropic materials,
we have Q16 Ab16 Ds16 0, D11 EI, Ds11 EIs and
Ds66 GIs (I; Is are the moment of inertia about the yaxes of beam section and that of the top and bottom
ange sections, respectively), so it is easy to prove that
formulas of this paper can be simplied to the ones of
thin-walled box beam with isotropic materials considering shear deformation eect a 1 [12] or the ones
ignoring the shear deformation eect a 0 [10], respectively.

42

Obviously, Eqs. (38)(42) compose the solution of


governing dierential equations. The normal stress of
the ith lamina of top or bottom ange can be deduced
from Eq. (2), (11), (12), (26), (34) and expressed as

4. Numerical results
Consider a 0=  h2 =0T symmetric carbonepoxy
box beam, as shown in Fig. 1, with geometrical pa-

152

W. Yaping et al. / Composite Structures 55 (2002) 147156

Table 1
The parameters in the function n nx
Count diagram

Boundary conditions

C1

x 0, n0 An 0;
x l=2, n 0

A2k 2 C2
2K 3

0 6 x 6 l=2
x 0, n0 An 0;
x l, n0 An 0

Ak 2 C2
K3

x 0, n0 An 0;
x l, n 0

A
c2
k

rameters l 210 mm, h 17:5 mm, t1 1:5 mm, t2 1


mm, and elasticity constants EL 114576 MPa,
ET 9981:7 MPa, GLT 4664:8 MPa, lLT 0:325, the
o-axis stiness Qkl of the laminas vs o-axis ply angle h
is given in Fig. 2.
When the beam is supported simply and subjected to
a concentrated load P 100 N at the mid-span, the
vertical displacement, ply normal stress at the rst layer
(bottom layer, 0) and second layer (h) of the bottom
ange, and shear lag coecient are evaluated in the
analysis of this paper (Figs. 315). Most of the results

Fig. 2. O-axis stiness Qkl vs ply angle h.

2PDs16
3k 3 bDs11



q
la16
1
3
k
2


Aq l a16
3
2
k
2 k

x 0, n0 An 0;
x l, n 0

n

C2
B1
cthkl=2 A=k

k 2

B2
A2 shkl

AC2
k

qa16  qA
3
k
k4

B3
cthkl

A=k

B4
kchkl

Ashkl


p
1
2
2k

4Ds16 x
3bDs11


l a16
2
2 k


q
la16
x
2 1
k
2

q
k2


p
1
k2

4Ds16 x
3bDs11

q
x
k2

a16 2
x
2

qa16 2
x
2k 2

a16 
k2

are compared with values obtained from a FEM analysis with ANSYS [16], which employs an 8-node isoparametric laminated shell element (SHELL99), nay, for
mid-span displacement vs concentrated load P (Fig. 4),
the present analysis results are compared with the model
test results and analysis results obtained from [1]. It can
be observed that analysis predictions of this paper agree
closely with the ANSYS analysis results and model test
results.
In Figs. 39, the results are given on the condition of
h 45, 2b 17 mm. In Figs. 1013, ply angle h is
variable, and 2b 17 mm. In Figs. 14 and 15, the ratio

Fig. 3. Vertical displacement vs spanwise coordinate (h 45,


2b 17, P 100 N). Simplied analysisanalysis ignoring shear lag
and shear deformation.

W. Yaping et al. / Composite Structures 55 (2002) 147156

153

Fig. 7. Shear lag coecient k along the bottom ange (measured at


mid-span, and h 45, 2b 17 mm, P 100 N).

Fig. 4. Mid-span displacement vs load P (h 45, 2b 17, P 100


N).

Fig. 5. Ply normal stress rx1 along the width of the bottom ange at
the rst layer (0) (measured at mid-span, and h 45, 2b 17 mm,
P 100 N).

Fig. 8. Ply normal stress of bottom ange along spanwise (h 45,


2b 17 mm, P 100 N).

Fig. 6. Ply normal stress rx2 along the width of the bottom ange at
the second layer (h) (measured at mid-span, and h 45, 2b 17 mm,
P 100 N).

Fig. 9. Shear lag coecient k along spanwise (h 45, 2b 17,


P 100 N).

of span length to width of the ange l=2b is variable,


and h 45.
In Figs. 3, 10, and 14, to illustrate the shear deformation and shear lag eects, the calculated results of

displacement ignoring the shear deformation and shear


lag eects (simplied analysis) are given, from which it
can be seen that the dierence between the results of
analysis and those of simplied analysis is obvious

154

W. Yaping et al. / Composite Structures 55 (2002) 147156

Fig. 13. Shear lag coecient k vs o-axis ply angle h (measured at midspan, and 2b 17 mm, P 100 N).
Fig. 10. Mid-span displacement vs o-axis ply angle h (2b 17,
P 100 N).

Fig. 14. Mid-span displacement vs l=2b (h 45, P 100 N).


Fig. 11. Normal stress rx1 at the rst layer (0) of bottom ange vs oaxis ply angle h (measured at mid-span, and 2b 17, P 100 N).

Fig. 15. Mid-span shear lag coecient k vs l=2b (h 45, P 100 N).

Fig. 12. Normal stress rx2 at the second layer (h) of bottom ange vs
o-axis ply angle h (measured at mid-span, and 2b 17, P 100 N ).

(especially, at the mid-span of the beam). This dierence


is caused mainly by the shear deformation eects (i.e.
dierence Ws , see Table 2 and Eqs. (40)(42)).

W. Yaping et al. / Composite Structures 55 (2002) 147156

155

Table 2
Analysis of shear lag and deection at mid-span of carbon-epoxy box beams
2b
(mm)

l=2b

17

12.35

34

6.176

kl=2; b or
W l=2
kl=2; b
kl=2; 0
W l=2
(10 2 mm)
Contribution
rate to
deection (%)
kl=2; b
kl=2; 0
W l=2
(10 2 mm)
Contribution
rate to
deection (%)

Bending
eect (1)
1.000
1.000
6.388
89.49

1.000
1.000
3.450
78.61

Shear deformation
eect (2)

0.7661
10.68

0.7661
16.63

The contribution of Ws to the displacement increases


with a lower value of the aspect ratio l=2b (see Table 2
and Fig. 14), so the shear deformation eect on displacement cannot be ignored. From Fig. 10 we can see
that the dierence (Ws ) is on the smaller side within the
ply angle h 3060, and larger when h tends to 0 or
90, it is because that o-axis stiness Q66 has larger
values within the ply angle h 3060 and less values
when h approaches 90 (see Fig. 2). So the eect of shear
deformation on composite box beam can be adjusted by
changing the ply angle, and when the ply angle h is set
within 3060, the eect of shear deformation on
composite box beam can be on the smaller side.
Fig. 9 shows that at the mid-span of the beam, the
eect of shear lag comes to a head. Due to the shear lag,
the ply normal stress is no longer distributing evenly
along the width of the ange. This result can be shown
in Figs. 5 and 6. According to Figs. 5 and 6, the shear
lag coecient k along the width of the ange is shown in
Fig. 7 (for this example, at mid-span ris 0, so
rxi rio k). When h 45, P 100 N, the inscape of the
shear lag coecient k is listed in Table 2.
From Table 2, it can be seen that for shear lag coecient k, the value of bend-shear coupling item is small
but it makes k close to 1, which means that bend-shear
coupling reduces the shear lag eect; on the other hand,
the total shear lag eect increase with lower value of the
aspect ratio l=2b (see Fig. 15).
Fig. 13 shows that the shear lag eect is on the
smaller side (k comes close to 1) within the ply angle
h 4060, and larger (k departs from 1) when h tends
to 0 or 90, which means that eects of shear lag on
composite box beam can be adjusted by changing the
ply angle too, and when h 4060, shear lag eect can
be reduced eectively.
As shown in Fig. 8, due to the coupling stress ris , ply
normal stress at the second layer (h) of the bottom

Shear lag eect

Theoretical data from


this paper
(1)+(2)+(3)+(4)

Basic
item (3)

Coupling
item (4)

0.115
)0.070
0.0041

)0.005
0.008
)0.0152

1.1
0.938
7.170

0.057

)0.207

1.00

0.283
)0.137
0.2107

)0.017
0.008
0

1.266
0.871
4.427

4.76

1.00

ange does not equal to zero at the hinge (x 0) of the


beam.
From Figs. 11 and 12, we can see that as the ply angle
h approaches 90, the normal stress at the rst layer (0)
of bottom ange rx1 increases. Meanwhile the normal
stress at the second layer (h) rx2 decreases, because
when h approaches 90, o-axis stiness Q11 diminishes
(see Fig. 2), which leads to rx2 reducing. To balance the
curtailment of normal stress at h layer, the normal
stress at the 0 layer must enlarge. On the other hand,
Fig. 12 shows that when h approaches 90, although the
dierence between rx2 at y b and rx2 at y 0 is small,
the shear lag eect is still large (see Fig. 13), it is for the
sake of r20 being synchronously small (see Eq. (43)).
From the above, for a simply supported 0=
h2 =0T symmetric carbonepoxy box beam subjected a
concentrated load P at the mid-span, the following
conclusions can be drawn:
1. The eects of shear lag and shear deformation cause
ply normal stress distributing nonuniformly along the
width of the ange and vertical displacement increases, respectively, which can inuence the design
of strength and stiness for composite box beam, so
it cannot be ignored.
2. To reduce the eects of shear lag and shear deformation we can set o-axis ply angle h within 4060
and, and if possible, increase the aspect ratio l=2b.
3. Ply normal stress of layer varies with the ply angle, so
it is very important to calculate exactly stresses of every lamina of the panel in the strength analysis of
composite box beams.
5. Conclusions
An analytical model capable of analyzing the static
behavior (including eects of shear lag and shear de-

156

W. Yaping et al. / Composite Structures 55 (2002) 147156

formation, ply normal stress, etc.) of composite box


beams under bending load, in which the structural
couplings of bendingtwisting and shearingtwisting are
eliminated by restricting the o-axis plies of the panels
to be balanced symmetric is presented. The formulas
required for analyzing the above-mentioned static behavior characteristics are given which can be easily
adopted by practising engineers for the analysis and
design of composite box beams. For the analysis of
shear lag, the solutions of controlling variable n suit
some simply supported composite beams and cantilever
composite beams subjected to a vertical concentrated
load P or an even distributed vertical load q along the xaxis are also presented. A calculated example is worked
out and a comprehensive numerical analysis of the effects of shear lag and shear deformation, ply normal
stress, displacement is accomplished. The numerical result of this paper for the composite box beams correlates
very well with the results of nite element (ANSYS)
analysis and results of the model test. It is concluded
that the shear lag and shear deformation signicantly
aect the strength and stiness characteristics of composite box beams, but it can be adjusted by changing the
ply angle of the layer in the analysis and design of
composite box beams.
Acknowledgements
The helpful comments made by the reviewers of this
paper and nancial supports from Science Foundation
of Gansu Province, China (ZS991-A22-021-G); Foundation of ``Hundred People Plan'' (to Dr. Y.M. Lai) and
Foundation of Innovation Engineering, Cold and Arid
Regions Environmental and Engineering Research Institute (210053) of Chinese Academy of Sciences; and
Special Science Foundation of Chinese Railway Ministry (J99Z097) are gratefully acknowledged.

References
[1] Cheng C et al. Researches on bending and torsional stiness of
thin-walled carbonepoxy box beam. In: Mechanics and practice.
Beijing: Beijing University Press; 1985. p. 1115.
[2] Chandra R, Stemple AD, Chopra I. Thin-walled composite beams
under bending, torsional, and extensional loads. J Aircraft
1990;27(7):61936.
[3] Jeon Song Min, Cho Maeng Hyo, Lee In. Static and dynamic
analysis of composite box beams using large deection theory.
Comput Struct 1995;57(4):63542.
[4] Vinson JR, Sierakowski RL. The behavior of structures composed
of composite materials. Dordrecht: Martinus Nijho; 1986.
[5] Yuanming Lai, Yaping Wu. An exact method for dynamic
analysis of rectangular laminated orthotropic plates under heating
load. J Compos Technol Res 1996;18(2):13540.
[6] Davalos JF, Qiao P. A computational approach for analysis and
optimal design of FRP beams. Comput Struct 1999;70:16983.
[7] Reissner E. Analysis of shear lag in box beams by the principle of
minimum potential energy. Q Appl Math 1946;6(3):26878.
[8] Moatt KR, Dowling PJ. Shear lag in steel box girder bridges. J
Struct Eng, ASCE 1975;53:43948.
[9] Foutch DA, Chang PC. A shear lag anomaly. J Struct Eng, ASCE
1982;108(7):16538.
[10] Jinqiong Guo, Zhenzheng Fang, Xiaodeng Luo. Analysis of shear
lag eect in box gird bridges. China Civil Eng J, CSCE
1983;16(1):113.
[11] Luo QZ, Li QS. Shear lag of thin-walled curved box girder
bridges. J Eng Mech, ASCE 2000;126(10):11114.
[12] Yaping Wu. Analysis of shear lag and shear deformation eects in
thin-walled box beam. Chin Eng Mech CSME 1994; Supplementary Issue: pp. 42528.
[13] Takayanagi H, Kemmochi K, Sembokuya H, Hojo M, Maki H.
Shear lag eect in CFRP I-beams under three-point bending. Exp
Mech 1994;34(2):1007.
[14] Roberto L-A, GangaRao HVS. Warping solution for shear lag in
thin-walled orthotropic composite beams. J Eng Mech, ASCE
1996;122(5):44957.
[15] Gjelsvik A. The theory of thin-walled bars. New York: Wiley/
Interscience; 1981.
[16] ANSYS, Inc. ANSYS elements reference. 9th ed. SAS, IP Inc.,
1997.

You might also like