Acta Polytechnica Vol. 44 No.

5–6/2004

A Shear Lag Analysis for Composite Box

Girders with Deformable Connectors
V. Křístek
A method is proposed for shear lag analysis which can be applied to steel-concrete composite box girders. The proposed method uses harmonic
analysis and allows the determination of shear lag effects from simple calculations so that the method is regarded as a design aid. The
character of the method can illustrate the influence of certain key parameters upon the extent of the shear lag effect.

Keywords: shear lag, composite girder, stress distribution, harmonic analysis.

obvious way of strengthening these regions is to weld on more

1 Introduction longitudinal stiffeners in the compression zone of the bottom
The effects of shear lag can cause a significant increase flange. This will increase the buckling resistance of the flange,
in the longitudinal stresses developed in steel box girders. but it will also accentuate the shear lag problem. An alter-
Previous investigations have shown that the extent of shear native method of strengthening an existing bridge girder is to
lag within a flange plate is dependent on the ratio between add a concrete layer to the compression flange so that it acts
the axial stiffness and the shear stiffness of the plate. The compositely with the steel (Fig. 1b), which will increase the
introduction of longitudinal stiffeners increases the axial stiff- buckling resistance while also controlling the shear lag effect.
ness without changing the shear stiffness so that there is a Although the method is applicable primarily for strength-
consequent increase in shear lag. Stiffeners are, of course, ening an existing bridge, it may well provide an economic
introduced to increase the resistance of the compressed flange alternative in the design of a new box girder.
to buckling. It has been proven in [1] that it is far more A perfect connection between the steel flange and the
advantageous, from the point of view of shear lag, if the flange concrete layer exists, however, only theoretically. Although
plate is stiffened with a layer of concrete that is made to act there certainly will be an intention to benefit from full
compositely with the steel plate (Fig.1a). The necessary composite interaction, the studs placed at regular distances,
composite action can be achieved by means of shear studs which are commonly used as connectors at the present time,
welded to the steel plate. exhibit some unavoidable deformability.
Among many applications of composite arrangements,
the case of increasing the load carrying capacity of an exist-
ing steel box girder bridge may be mentioned as a special 2 Governing equations
example. The bottom flange plate in the hogging moment Shear flows sq and cq, and normal forces snx and cnx per
regions over the internal supports of a continuous girder is unit width act on a typical element of the steel flange sheet or
particularly susceptible to the effects of shear lag. The most the concrete layer, respectively (see Fig. 2).
The equations governing the equilibrium in the longitudi-
concrete nal direction are:
for the steel sheet (Fig. 2a)
¶ s nx ¶ s q
steel + + f = 0, (1)
¶x ¶y
for the concrete layer (Fig. 2b)
(a) ¶ c nx ¶ c q
+ - f = 0, (2)
¶x ¶y
in which f is the shear acting in the longitudinal direction at
the interface between the steel flange and the concrete layer.
If the contribution of small traverse forces to the strains is
neglected, it may be written:
¶ su n
s ex = = s x , (3)
¶ x ts E s
¶ cu n
c ex = = c x , (4)
¶ x tc Ec
s nx
sey = -n s s e x = -n s , (5)
(b) ts E s
Fig. 1: (a) Steel-concrete composite girder – simplified form of
c nx ,
cross section, (b) concrete layer added to the compression c ey = -n c c e x = -n c (6)
bottom flange plate in the hogging moment regions tc E s

© Czech Technical University Publishing House http://ctn.cvut.cz/ap/ 125

Acta Polytechnica Vol. 44 No.5–6/2004

It may be assumed that the shear f acting between the steel

sheet and the concrete layer, being provided by deformable
connectors, is proportional to the mutual longitudinal slip
which occurs at the interface between the two components,
i.e.

f = k ( su - c u ) (14)

where k is the connector stiffness.

Eq. (14) may be written in the form:

¶f ¶ u ¶ u æ n ö
= k æç s - c ö÷ = k çç s x - c x
n
÷. (15)
¶x è ¶x ¶x ø t E t ÷
è s s c Ec ø
The following Fourier series may express the searched
functions:
jp x
s nx =å s Nj ( y) sin
L
, (16)

jp x
c nx =å c Nj ( y) sin , (17)
L
¥
jp x
f = å F ( y) cos
j =1
j
L
, (18)

Fig. 2: Equilibrium conditions in the longitudinal direction

where L is the effective span-length.

sq Eqs. (12), (13) and (15) can be written in the form:

gs = , (7)
ts G s
2 2
cq
&& - j p (2 + n ) N - 2 jp (1 + n ) F = 0 ,
N (19)
gc = , (8) s j
L2
s s j
L
s j
tc Gc
2 2
where ts and tc are the thicknesses of the steel flange and the && - j p (2 + n ) N - 2 jp (1 + n ) F = 0 ,
N (20)
c j c c j c j
concrete layer, respectively, ex and ey are the direct strains in L2 L
the longitudinal and transverse directions, respectively, and g jp æ s Nj c Nj ö
are the shear strains. E, G and n represent Young’s moduli, Fj + k çç - ÷ = 0,
÷ (21)
L è t s E s tc Ec ø
the shear moduli and Poisson’s ratios, respectively; u are the
longitudinal displacements. ¶2 s Nj
&& =
in which s N , etc.
The general form of the condition of compatibility is as j
¶y2
follows:
These relations represent a set of three equations for the
¶2 e x ¶2 e y ¶2 g
+ = . (9) unknown functions s Nj ( y), c Nj ( y) and Fj ( y), which can be
¶y2 ¶x2 ¶x ¶y adjusted to the following system of two differential equations
&&
Substituting the strains from Eqs. (3)–(8) it is obtained: s Nj + a j s Nj - c j c Nj = 0 , (22)
for the steel: && + b N - d N = 0,
N (23)
c j j c j j s j
¶2 s nx ¶2 s nx ¶2 s q
- ns = 2 (1 + n s ) , (10) where
¶y2 ¶x2 ¶ x¶ y
j2 p2 2k(1 + n s )
for the concrete layer: aj = - 2
(2 + n s ) + ,
L ts E s
¶2 c nx ¶2 c nx ¶2 c q 2 2
- nc = 2 (1 + n c ) . (11) j p 2k(1 + n c )
¶ y2 ¶ x2 ¶ x¶ y bj = - (2 + n c ) + ,
L2 tc Ec (24)
Substituting for the shear flows sq and cq from equations 2k(1 + n s )
(1) and (2): cj = ,
tc Ec
¶2 s nx ¶2 s nx ¶f 2k(1 + n c )
+ (2 + n s ) + 2 (1 + n s ) = 0, (12) dj = .
¶y 2
¶x 2 ¶x ts E s
It follows from Eq. (22) that
¶2 c nx ¶2 c nx ¶f 1 &&
+ (2 + n c ) - 2 (1 + n c ) = 0. (13) c Nj = ( s Nj + a j s Nj ), (25)
¶y2 ¶x2 ¶x cj

126 © Czech Technical University Publishing House http://ctn.cvut.cz/ap/

 Acta Polytechnica Vol. 44 No.5–6/2004

which substituted into Eq. (23) allows to obtain a differential The amplitude function governing the distribution of the
equation of the fourth order shear at the interface between the steel and concrete can be
d 4 s Nj ( y) d2 s Nj ( y) expressed from equation (21) as
+ Aj + B2j s Nj ( y) = 0, (26)
dy4 dy2 kL ìï 1
Fj ( y) = í C1, j [ rj P1, j ( y) + s j P2 , j ( y)] -
whose coefficients are jp ïî cj tc Ec
j2 p2 é(1 + n s ) (1 + n c ) ù P1, j ( y)
Aj = - 2
( 4 + n s + n c ) + 2k ê + ú, - +
L ë ts E s tc Ec û ts E s
(35)
j p ì j2 p2
2 2
B2j = 2 í 2 (2 + n s )(2 + n c ) + (27) +C2 , j
1
[ rj P2 , j ( y) - s j P1, j ( y)] -
L î L cj tc Ec
é(1 + n s )(2 + n c ) (1 + n c )(2 + n s ) ù ü P2 , j ( y) ü
+ 2k ê + úý.
ts E s tc Ec - ý.
ë ûþ ts E s þ
The general solution of (26), if the case of complex roots
of the characteristic equation is assumed, is It is seen that also this distribution is symmetrical about
s Nj ( y) = C1, j P1, j ( y) + C2 , j P2 , j ( y) + the flange mid-width.
(28)
+ C3 , j P3 , j ( y) + C4 , j P4, j ( y), The values of the remaining constants C1, j and C2, j can be
determined from the shear loading conditions at the edges of
where the steel flange and the concrete layer.
P1, j ( y) = sinh x j y sin h j y Combining equations (1), (16), (18):
P2 , j ( y) = cosh x j y cos h j y ¥
¶ sq ¶ n jp jp x
P3 , j ( y) = cosh x j y sin h j y
(29)
¶y
=- s x -f =
¶x å éêë L
j =1
s Nj ( y)
ù
+ Fj ( y) cos
úû L
=

P4 , j ( y) = sinh x j y cos h j y
¥ ì é jp kL æ r j 1 ö÷ù
and = å ï
í C1, j
ï
ê
êë L
+ ç - ú P1, j ( y) +
jp çè c j tc Ec t s E s ÷øú
Bj Aj Bj Aj j =1 î û
xj = - , hj = + .
2 4 2 4 kLs j
+ P2 , j ( y) + (36)
The amplitude function c Nj ( y), according to Eq. (25), is jp c j tc Ec
determined as
é jp kL æ r j 1 ö÷ù
c Nj ( y) =
1
{
C1, j [ rj P1, j ( y) + s j P2 , j ( y)] + + C2 , j ê + ç - ú P2 , j ( y) -
cj
êë L jp çè c j tc Ec t s E s ÷øú
û
+C2 , j [ rj P2 , j ( y) - s j P1, j ( y)] + (30)
kLs j üï jp x
+C3 , j [ rj P3 , j ( y) + s j P4 , j ( y)] + - P1, j ( y) ý cos
jp c j tc Ec ïþ L
}
+C4 , j [ rj P4 , j ( y) - s j P3 , j ( y)] ,
in which so that (by integrating with respect to y) the shear flow in the
rj = x2j - h2j + ai , steel flange at any point may be expressed as:
(31)
s j = 2 x j hj .
ì é jp kL 1 ö÷ ù
¥ æ rj
s q ( x, y ) = åx 2
j =1 j + h j ï
1
2
ï
íC 1, j ê
êë L
+
jp
ç - ú´
ç c j tc E c t s E s ÷ ú
è øû
3 Boundary and loading conditions î
Shear lag analysis is carried out for loads, placed symmet- ´[ xj P3 , j ( y) - h j P4 , j ( y)] +
rically on the girder cross-section. Thus, assuming the origin kL s j
of the traverse co-ordinate y to be taken at the mid-width of + [ x j P4 , j ( y) + h j P3 , j ( y)] +
jp c j tc E c
the flange, i. e. at the axis of symmetry, then, because of the
symmetry é jp kL æ r j
ç 1 ö÷ ù
C3 , j = C4 , j = 0 (32) + C2 , j ê + - ú´ (37)
êë L jp çè c j tc E c t s E s ÷ø ú
so that from equations (28) and (30), the distributions across û
the flange width of the normal forces in the steel flange and in ´[ x j P4 , j ( y) + h j P3 , j ( y)] -
the concrete layer are governed by:
kLs j üï jp x
s Nj ( y) = C1, j P1, j ( y) + C2 , j P2 , j ( y) (33) - [ x j P3 , j ( y) - h j P4 , j ( y)] ý cos =
jp c j tc E c ïþ L
c Nj ( y) =
1
cj
{
C1, j [ rj P1, j ( y) + s j P2 , j ( y)] +
(34) ¥
jp x
}
+ C2 , j [ rj P2 , j ( y) - s j P1, j ( y)] .
=- å [C
j =1
1, j Z1, j ( y ) + C2 , j Z2 , j ( y )]cos
L
.

© Czech Technical University Publishing House http://ctn.cvut.cz/ap/ 127

Acta Polytechnica Vol. 44 No.5–6/2004

Similarly, the shear flow in the concrete layer, combining From simple beam theory, the shear flow qe(x) transmitted
equations (2), (17) and (18), is governed by the following from the web to the edge of the steel flange can be approxi-
relation: mated as
æ E ö
¶ cq ¶ n
¥
é jp jp x çç t s + tc c ÷÷ be
¶y
= - c x + f =-
¶x å
j =1
ù
êë L c Nj ( y ) - Fj ( y ) úû cos L = qe ( x) = V ( x) è
Es ø
, (40)
2I
¥ ì æ jp ö
åïíîïC ç kL ÷ [ r P ( y ) + s P ( y )] + where V(x) is the total shear force acting on the beam cross-
= - +
1, j ç Lcj jpcj tc Ec ÷ø
j 1,j j 2 ,j
-section at position x; I is the second moment of area of the
j =1 è
composite cross-section (the contribution of the concrete
kL layer being reduced by the ratio Ec /Es), e is the distance from
+ P ( y) + (38)
j p t s E s 1 ,j the cross-sectional neutral axis to the centroid of the compos-
æ jp kL ö ite flange.
+ C2, j ç + ÷ [ r P ( y ) - s P ( y )] +
ç Lcj j p c t E ÷ j 2 ,j j 1 ,j The shear flow transmitted at the edge of the flange can
è j c cø
also be expressed in the form of the Fourier series. For the
kLsj üï jp x case of simply supported ends, the series takes the form
+ P2 ,j ( y ) ý cos .
jp t s E s ïþ L
¥
jp x
Thus, the shear flow in the concrete layer at any point is
qe ( x) = åQ
j =1
e, j cos
L
, (41)
expressed as:
where
¥ ìï æ jp ö L
c q ( x, y) = - å
1
x2 + h2j ï
í C1, j ç
ç Lc j
è
-
j p c
kL
j tc Ec
÷´
÷
ø
Q e, j =
2
ò q ( x) cos
e
jp x
dx =
j =1 j î L L
0
(
´ rj [ x j P3 , j ( y) - h j P4 , j ( y)] + æ E ö
(42)
çç t s + tc c ÷÷ be
)
L
+ s j [ x j P4 , j ( y) + h j P3 , j ( y)] + Eb ø jp x
=è
kL IL ò V ( x) w
0
L
dx .
+ [ x j P3 , j ( y) - h j P4 , j ( y)] +
jp t s E s Values of the coefficients Q e, j , evaluated according to this
æ jp ö formula, are listed in Table 1 for a few typical cases.
kL
+ C2 , j ç - ÷´ It must hold at the joint of the web and the steel flange
ç Lcj j p c t E ÷
è j c c ø that
(
´ rj [ x j P4 , j ( y) + h j P3 , j ( y)] - æ bö
s q ç x, ÷ = qe ( x) (43)
è 2ø
- s j [ x j P3 , j ( y) - h j P4 , j ( y)] + )
and at the edge of the concrete layer that
kL ü jp x
+ [ x j P4 , j ( y) + h j P3 , j ( y)] ý cos æ bö
jp t s E s L s q ç x, ÷ = 0 (44)
þ è 2ø
since the concrete layer is not directly connected to the web.
which may be written in the form
Combining relations (37), (39), (41), (43) and (44), it is
possible to form two equations to determine the constants
¥ ìï C1, j and C2, j , which can be written in matrix form as:
1 æ jp ö
c q ( x, y) = - åx 2
1
+ h2j
í C1, j
c
ç
kL
ç L - jp t E
÷´
÷ é æbö æ b öù
j =1 j îï j è c c ø
ê Z1, j çè 2 ÷ø Z2 , j çè 2 ÷ø ú ì C1, j ü ì Q e, j ü
=
ê b ú í C ý í 0 ýþ
´ [( rj x j + s j h j ) P3 , j ( y) + ( s j x j - rj h j ) P4 , j ( y)] + (45)
êZ3 , j æç ö÷ Z4 , j æç ö÷ú î 2 , j þ î
b
kL ë è2ø è 2 øû
+ [ x j P3 , j ( y) - h j P4 , j ( y)] +
jp t s E s

1 æ jp kL ö 4 Completion of shear lag analysis

+ C2 , j ç ÷´
cj ç L - jp t E ÷ (39)
Having thus determined the two constants, the ampli-
è c c ø
tudes of the normal longitudinal forces s Nj a c Nj for any
´ [( rj h j - s j x j ) P3 , j ( y) + ( rj x j + s j h j ) P4 , j ( y)] +
particular harmonic can be obtained from equations (33) and
kL ü jp x (34). The magnitudes of these forces varies across the width of
+ [ h j P3 , j ( y) + x j P4 , j ( y)] ý cos = the flange; the peak value of the normal longitudinal force
jp t s E s þ L
s Nj ( y) acting in the steel sheet occurs at the edge, i. e. where
¥ y = b/2. The distribution of the normal longitudinal force
jp x
=- å [C
j =1
1, j Z3 , j ( y) + C2 , j Z4 , j ( y)]cos
L
.
c Nj ( y) across the width of the concrete layer may have a more
general character.

128 © Czech Technical University Publishing House http://ctn.cvut.cz/ap/


Table 1: Values of the coefficient Qe, j for different tapes of loading
Knowing the amplitudes s Nj and c Nj , the values of
the longitudinal normal forces per unit width s nx ( x, y) and
c n x ( x, y) may be determined from equations (16) and (17) for
any position on the flange. Also the shear flows s q( x, y) and
c q( x, y) at any point may be determined from equations (37)
and (39).
To evaluate the forces taken by the studs, the shear f acting
at the interface between the steel flange and the concrete
layer is to be determined according to equation (18). The am-
plitude function Fj ( y), describing the distribution of the shear
across the flange width, is determined – knowing constants
C1, j and C2, j – by equation (35).
5 Summary of calculations
For any particular girder with composite flanges, the first
step in the calculation of the shear lag effect is to determine
the value of coefficient Q e, j from equation (42). The value
Q e, j is then substituted into the right-hand side of equations
(45) to give the values of constants C1, j and C2, j , and, finally,
for any harmonic the amplitudes of all the functions are
required. These, in turn, are substituted into equations (16),
(17), (18), (37) and (39) to give the normal forces per unit
width, the shear acting at the interface between the steel and
concrete components, and the shear flows at any position on
the composite flange.
The corresponding value of the longitudinal stress in the
steel component of the flange is then given by:
n ( x, y)
s sx ( x, y) = s x (46)
ts
and the longitudinal stress in the concrete layer is obtained
as:
c n x ( x, y)
scx ( x, y) = . (47)
tc
© Czech Technical University Publishing House http://ctn.cvut.cz/ap/
Acta Polytechnica Vol. 44 No.5–6/2004
Should the shear stress values also be required, then, hav-
ing evaluated the shear flows s q ( x, y) and c q ( x, y) at any point,
the shearing stress in the steel is obtained as:
s q( x, y)
t sx ( x, y) = (48)
ts
and the shear stress in the concrete layer is given by:
c q( x, y)
tcx ( x, y) = . (49)
tc
6 Conclusions
This paper has described the development of an approxi-
mate analytical method for analysing the stress distribution in
the flanges of composite steel-concrete beams with deform-
able connectors. Its primary advantage is the closed form of
the results obtained and its ease of application. The method is
also very suitable for parametric studies investigating the
influences of various arrangements, and for optimisation
studies.
To conclude, it should be noted that – besides the
mechanical effects – the thermal effects can also play an im-
portant role in the structural performance of steel-concrete
composite beams, see, e.g., [2].
7 Acknowledgment
Support for this research through grants 103/02/.1005
and 103/02/.0020 from the Grant Agency of the Czech Re-
public is gratefully acknowledged.
129
Acta Polytechnica Vol. 44 No.5–6/2004

(Chap. in Composite Steel-Concrete Structures),

References Applied Science Publishers, Elsevier, London 1988,
[1] Křístek V., Evans H. R.: “A hand calculation of the shear Edited by R. Narayanan.
lag effect in unstiffened flanges and in flanges with [6] Evans H. R., Křístek V., Škaloud M.: “Strengthen-
closely spaced stiffeners.” Civil Engng Practising Design ing steel box girder bridges by controlling the effects
Engrs, Vol. 4 (1985) No. 2. of shear lag.” Proc. Instn. Civ. Engrs. Structs. and Bldgs.,
[2] Římal J.: “Measurement of Temperature Fields in Com- Vol. 104 (Nov. 1994).
posite Steel and Concrete Bridges.” CTU Publishing
House, Prague, 2003.
Prof. Ing. Vladimír Křístek, DrSc.
[3] Křístek V. Evans H. R., Ahmad M. K. M.: “A Shear Lag phone: +420 224 353 875
Analysis for Composite Box Girders.” J. Construct. Steel e-mail: kristek@fsv.cvut.cz
Research, Vol. 16 (1990).
[4] Evans H. R., Ahmad M. K. M., Křístek V.: “Shear lag in Department of Concrete Structures and Bridges
composite box girders of complex cross sections”. J. Con- Czech Technical University in Prague
struct. Steel Research, Vol. 24 (1993). Faculty of Civil Engineering
[5] Křístek V., Studnička J.: “Composite girders with de- Thákurova 7
formable connection between steel and concrete.” Praha 6, Czech Republic

130 © Czech Technical University Publishing House http://ctn.cvut.cz/ap/