Академический Документы
Профессиональный Документы
Культура Документы
Presentation 18.03.2004
STATIC BEHAVIOR OF CURVED GIRDERS.....................................................................3
ABSTRACT...............................................................................................................................3
INTRODUCTION......................................................................................................................3
1. TORSIONAL WARPING STRESS ...................................................................................3
2.2 Variation in the distortional warping stresses due to various parameters ...14
ABSTRACT
This is the seminar presentation of the Seminar in Structural Engineering in spring 2004.
The course is arranged in the Helsinki University of Technology by Laboratory of Bridge
Engineering, Laboratory of Structural Mechanics and Steel Structures and is for under- and
postgraduate students.
In this paper the static behaviour of curved bridges is clarified by investigating actual bridge
cases. Basis of torsional warping is shortly explained and the behaviour of three types of
cross sections are studied and compared. The effects of diaphragm spacing, central angle
and cross sectional quantities to distorsional warping stresses are presented with four typical
box girder bridges. Based on the examinations practical design guidelines are derived and
explained.
Finally the theory of deflection of curved girder bridge is formulated. Also in this part
monobox, twin-box and multiple I-girders are compared and practical design instructions are
presented.
INTRODUCTION
Curved bridges are often constructed in multi-level junctions. Analysing of torsional stresses
of the girders is the most challenging and interesting part of the design process.
In the design of curved girder bridges, the engineer is faced with a complex stress situation,
since these types of bridges are subjected to both bending and torsional forces. In general,
the torsional forces consists of two parts, i.e., St. Venant’s and warping. Thus the procedure
for determining the induced stresses of a curved girder is difficult.
In order to clarify the magnitude of the torsional warping stress, the following preliminary
analysis is conducted.
The coverning differential equation for the twisting angle θ of a curved beam subjected to
torque mT is
(1)
Page 3
The bimoment Mω is given by the well-known formula
(2)
In this formula one should note the analogy between warping and bending.
The differential equation (1) can be rewritten with respect to the bimoment as
(3)
(4)
This parameter can be nondimensionalized by multiplying by the central angle of curved
girder Φ, which yields
(5)
1.2 Values of parameter κ for actual bridges
In the following torsional parameters κ for various curved girder bridges with cross sections
as illustrated in Fig.1 were investigated by using the actual dimensions of the bridges. The
investigated cross sections are open multi-I-girder, twin-box-girder and monobox girder.
Figure 1. Investigated curved girder bridges: a) multiple-I girder, (b) twin-box girder, and (c)
monobox girder.
Page 4
arbitrary point B is chosen as the
origin, as shown In Fig. 2.
If we assume horizontal
and vertical axes ξ and η,
respectively, the location of the
shear center S for the multiple
girder bridge, idealized as a
single unit, can be determined
from the equations.
(6a)
(6b)
(7)
(8)
Also the centroid C for the system of curved beams can be determined from
Page 5
(9a)
(9b)
where Ai is the cross-sectional area of the ith girder. The corresponding moment and
product of inertia, which will be required in the stress analysis, can approximated by
(10a)
(10b)
(10c)
Torsional parameter κ
Page 6
1.3 Relationships between the stress ratio σω/σb and κ.
The design of curved girder bridges is related to the dead and live loads. In this section the
most severe loading conditions that will induce the largest bending stress σb and warping
stress σω will be determined. These loading conditions can be idealized by a concentrated
load P or the uniformly distributed load q, as shown in Fig. 4.
For a concentrated load P and a uniform load q, the induced midspan bending moments Mx
are:
or (12)
The corresponding bimoments Mω can be obtained by solving Eqs. (1) and (2), which
results in
or (13)
Next, the ratio of warping stress σω to bending stress σb, can be estimated. By applying
values Ro/np = 1, Ixy = 0 and n = 1 in equations we get:
(14)
where Y is the fiber distance and ω is the warping function of the cross section.
Figure 4. Load conditions to estimate bending stress σb and warping stress σω.
Ix = 2IH (15)
The maximum fiber distance Y1 to point 1 located on the lower flange of the I girder, is
(16)
The warping constant Iω of a single curved girder bridge can be calculated by utilizing Eq.
(8), or
(17)
where B is the spacing of the web plates. The warping function ω1, also at point 1, can be
evaluated easily from the well-known formula
(18)
(19)
(20)
and assuming that this parameter Ψ can be applied to twin-box and monobox curved girder
bridges, we obtain a generalized form of Eq. (14):
(21)
Numerical values for the parameter Ψ, given by Eq. (20), have been determined for actual
bridges. This parameter can be related to the cross-sectional shape of curved girder and is
categorized as follows:
Also, the maximum ratio of the span L to the girder width B, as indicated in Fig. 1, has the
limitation
(23)
This limitation is required because negative reactions will occur at the inner side of the
supports, and thus the bearing shoes must be designed for uplift.
It is assumed that there is a critical value of the torsional parameter κcr at which the warping
stress σω cannot be determined exactly. This value will occur between the twin-box and
monobox curved firder configuration. Therefore, eo estimate the stress ratio, let ε (%) =
100σω/σb. Now assume that ψ = 2.5, which is the upper value for a twin-box section, as
given in Eq. (22), and let L/B = 10, as shown in Eq. (23). The by applying Eq. (21).
(24)
Setting equal to various values gives the following ε values:
If the analysis of the warping stress σω is not important in comparison with bending stress
σb when ε<4 percent, then , as shown in Eq. (25b). Therefore, the critical torsional
parameter κcr r can be rewritten by using Eq. (5).
(26)
This equation has been plotted, as shown in Fig. 3. Examination of this figure shows that
the value of κcr increases as the value of the central angle Ф increases. For Ф≥0.5, however,
a constant value of κcr =30 may be assumed.
Under these considerations, a more convenient formula, for practical design purposes, can
be proposed:
Page 9
10 + 40Φ for 0≤Φ≤0.5 (27a)
κcr =
30 for Φ≥0.5 (27b)
Note that κcr=10 when Φ=0, which represents the critical value for straight plate girders.
The proposed equations (27) are plotted in Fig. 3. From this figure we conclude that the
evaluation of the warping stress σω is not important for monobox girder bridges, whereas
the evaluation of the warping stress σω is required for twin-box or multicell curved girder
bridges.
The warping stress σω in a curved box girder is small enough that the following approximate
method can be applied. The warping and shear stresses are as follows:
Warping stress :
(28)
Shearing stress:
(29)
Page 10
{
And the parameters are
ξ=
{ 2.0
4.0
pattern
pattern
(30c)
(30d)
λ=
{ 0.8
-0.6
pattern
pattern
(30e)
(30f)
Χ=
{ 2.0
5.0
pattern
pattern
Note that both patterns A and B coexist where b/h = 1.5 to 2.5.
(30g)
(30h)
Furthermore, the torsional warping normal stress at points 1 and 2 in Fig. 6 can be found
modifying σω as follows:
(31a)
(31b)
Page 11
where the additional parameters γ1 and γ2 are given by
(32a)
(32b)
The fundamental differential equation for the distortion of curved box girders can be written
as
(33)
where
(34)
Page 12
2.1 Parameters of distortion.
The distortional warping parameter λ, the parameter β , and the distortional warping
constant IDω occur within the following ranges (the data are result of a parametric survey of
actual bridges):
unit: m6 (36)
As shown in Fig. 7.
Although there may be many different combinations of these distortional parameters, the
parametric analyses were performed with the actual data limiting the following four box-
girder bridges, as indicated in Table 1.
Bridge h b a tu tw tl L Ix Idω λ β Ψ
cm cm cm cm cm cm m m4 m6 X10-
2
/m
B-1 200 410 80 1.98 1.0 1.29 60.0 0.162 0.065 3.49 2.46 0.508
B-2 250 480 110 1.82 1.0 1.39 90.0 0.306 0.306 2.57 2.44 0.504
B-3 199 594 199 1.89 1.3 2.97 120.0 0.377 0.377 2.73 2.36 0.504
B-4 400 550 225 1.86 1.1 2.77 150.0 1.471 0.665 1.71 2.56 0.512
Parametric studies were performed to determine the variations in the distortional warping
stresses σDω due to the diaphragm spacing LD, central angle Φ, cross-sectional
quantities L/b, and the rigidity parameter of the diaphragm, γ. In these analyses, the
Page 13
transverse bending stresses σDb in the curved box girders due to the distortion are ignored
as being very small in comparison with the distortional warping stresses σDω.
Moreover, the loading conditions are as follows: a uniformly distributed load w, a line load
p in the direction of the girder axis, and a concentrated load P. The p and P loads were
applied on the inner web of the curved box girder bridges to make the distortional warping
stresses as large as possible.
Finally, the distortional warping stresses σDω are calculated at the junction point 3, shown in
Fig. 8, and are taken as the extreme values in the direction of the bridge axis.
Figure 8. Warping function due to torsion.
These values are also nondimensionalized by the flexural stresses σb due to the bending
moment Mx:
(38)
where Wl is the section modulus at the junction point of the web and bottom plates. And the
approximate formulas for the bending moment of the curved bridge are
Page 14
for concentrated P (39c)
a. Effects of diaphragm spacing. The variations of σDω/σb due to the diaphragm spacing
– LD/L=1/20, 1/10, 1/8, 1/5 and 1⁄4 - can be plotted as in Fig. 9 by assuming that Φ=1/3 and
KD= ∞. From these figures the variations in σDω/σb are parabolic forms for the distributed
load w and line load p, in accordance with the increase in LD/L. Also the values of σDω/σb
vary linearly with LD/L for the concentrated load P. Conclusion; the longer is the distance
between diaphragms, the bigger is the value of σDω/σb.
Figure 9. Variations of σDω/σb with LD/L: (a) uniformly distributed load (value at section
on diaphragm), (b) line load along bridge axis (value at section on diaphragm), and (c)
concentrated load (loaded and calculated at section on middiaphragm).
b. Effects of the central angle. The influences of the central angle Φ were examined by
varying Φ from 0 to 1/5, 1/3 and 2/3 radian under the conditions of LD/L = 1/10 and KD = ∞.
Figure 10 shows the results.
The variations in σDω/σb due to Φ are linear for a distributed load w and a line load q but
nearly constant for a concentrated load P.
Figure 10.
Variations of σDω/σb with L/R: (a) uniformly distributed load (value at section on diaphragm),
(b) line load along bridge axis (value at section on diaphragm), and (c) concentrated load
(loaded and calculated at section on middiaphragm).
Page 15
c. Effects of the cross-sectional quantities. The preliminary analyses revealed that the
effects of the thicknesses tu, tw and tl and of the dimensions a and h on σDω/σb are negligible,
but the effect of the spacing of the web plate b is significant.
Therefore, the variations in σDω/σb were examined by altering the spacing b and by setting L/
b equal to 10, 30 and 40. Figure 11 shows the results where LD/L = 10, Φ =2/3, and KD =∞.
For distributed and line loads p, the influences of L/b on the distortional warping stress are
positive. This tendency is, however, reversed for a concentrated load.
From these analyses, the approximate formulas to evaluate σDω/σb can be summarized as in
Table 2 for KD =∞.
Figure 11 Variations of σDω/σb with L/b: (a) uniformly distributed load (value at section
on diaphragm), (b) line load along bridge axus (value at section on diaphragm), and (c)
concentrated load (loaded and calculated at section on middiaphragm).
Table 2 Approximation of
Viewpoint
Load At section on middle of diaphragm At section on diaphragm
Uniformly distributed
Load w
Concentrated load
P
In the above discussion, the rigidities of the diaphragm were assumed to be infinitely large.
To determine the effects of the rigidity of the diaphragm on the distortional warping stress,
the rigidity KD is expressed by a dimensionless parameter as
Page 16
(40)
Figure 12. Variations of σDω along girder axis: (a) line load p on a web plate of box girder,
(b) concentrated load P on a web plate of box girder at x =9l/20.
Next, the effects of the central angle Φ corresponding to various rigidity parameters γ
were examined. These results are plotted in Fig. 13, where LD/L = 10 and viewpoints are
fixed at the sections s = L/20 and s=9L/20 for line and concentrated loads, respectively.
These figures show that the effects of the rigidity parameter γ decrease in accordance with
increases in the central angle Φ, so that the distortional warping stress σDω of the curved
box-girder bridges can safely be evaluated by setting the central angle Φ=0.
Page 17
Figure 13 Variation of σDω with Φ and γ : (a) Line
load along bridge axis. Viewpoint: S= 1/2L. (b) Concentrated load. Viewpoint: S= 9/20L.
For the above reasons, the stress ratio was examined for Φ=0 and Lp/
L=10. These results are plotted in Fig. 14. Observing the relationships between the stress
ratio and the rigidity parameter γ of Fig. 14a, we see that the absolute
value of the stress σDω at the middiaphragm section is almost equal to the stress σDω at the
diaphragm (s=L/10) when the rigidity parameter of the diaphragm is
γ≥ 1500 (41)
for the line load along the bridge axis. Within these conditions, the distortional warping
stress σDω is given by the approximate formula at the diaphragm (indicated on the right-hand
side of Table 2) instead of the formula for the section at middiaphragm. Unless the condition
of Eq.(41) is satisfied, the distortional warping stress at the section at middiaphragm will be
greater than that of the value in Table 2, and an additional exact analysis will be required to
determine the distortional stress.
For the concentrated load of Fig. 14b, however, the stress ratio is reduced to
(42)
Figure 14 Variation of σDω with γ : (a) line load along bridge axis and (b) concentrated load.
Page 18
Finally, the results of the parametric survey based on actual bridges concerned with the
rigidity parameter γ can plotted, as shown in Fig. 15.
(43)
(44)
(45)
The corresponding loading conditions for live loads are assumed to be one-half the width
of the deck, as illustrated in Fig. 16. Moreover, the influences of curvature in Eqs. (39a) to
(39c) are small enough to be ignored, and the coefficient 2.2 in Eq. (45) is derived from Eq.
(42).
Note that the magnitude of σDω is specified at point 3 in Fig. 8, and this value should be
modified corresponding to the ordinate of the warping function of Fig. 8.
(46)
From the equation, a condition which makes the distortional warping stress σDω less than 5
percent of the bending stress σb can e derived numerically, i.e.,
(47)
For example, taking the live load intensities as
(48a)
and (48b)
and setting b = 2,0 m (the minimum width of the actual bridge data), we can plot the
relationships among the span L, central angle Φ, and diaphragm spacing LD (in meters) can
be found from the following equations:
Page 20
lD <6 l< 60 m (49a)
lD≤0.14 l -2.4 60 m≤ l ≤160 m (49b)
lD = 20 l> 160 m (49c)
for the straight box-girder bridges, where l (m) is the span length of the straight box-girder
bridges. For the curved box-girder bridges, however, the spacing of the diaphragm should
be smaller than that of Eq. (49) in accordance with the increases in the central angle Φ.
The approximate formula for the spacing of the diaphragm LD (m) in the curved box girder
with the span length L (m) is governed by the following equation:
(50)
c. Required rigidity of diaphragms. The approximate formulas, Eqs. (43) to (45), should
be adopted under the condition that
d. Stress checks for diaphragms. The diaphragm should be designed to provide not only
enough rigidity but also sufficient strength against the distortional stresses. Conservatively,
the distortional warping moment can be expressed approximately by
Page 21
(53)
Considering the loading condition of Fig. 17 and setting Ψ = 0.525 [see Eq. (37)], we get
(54a)
(54b)
and (55)
(56)
In addition to designing a bridge for strength or stress, the structure must have sufficient
stiffness. Stiffness is needed to ensure against dynamic loads, overall and local buckling,
and any intolerable human response.
(57a)
(57b)
which satisfies the boundary conditions for a simply supported curved girder.
(58)
where U denotes the strain energy stored in the curved girder. This strain energy can be
evaluated by applying Eq. (9a) with R = Rs and Eqs. (23) and (28), which leads to
(59a)
(59b)
(59c)
The unknown coefficients wi and bi, as given in Eq. (57), can be evaluated by applying the
principle of least work such that ∂Π/∂ωi = 0 and ∂Π/∂b1 = 0. This gives
(60a)
(60b)
(61a)
(61b)
Page 23
(61c)
(62)
Substituting wi and bi from Eq. (60) into Eq. (57), we find that
(63a)
(63b)
where (64a)
(64b)
When a concentrated load P acts at the midspan of a curved girder bridge, the vertical
deflection w at s=a =L/2 can be written as
(65)
For a straight girder bridge, the parameters Φ=0 and kw,i = kβ,i =0 and the deflection of a
straight bridge wΦ=0 reduce to
(66)
In general, this equation is accurate for i<3. Therefore, by setting i=1, the deflection
increment factor for curved girders in comparison to straight girders of identical span and
cross-sectional geometry can be expressed by
(67)
Page 24
(68)
to the flexural rigidity EIx, or
(69)
The deflection incremental factor has also been determined for the uniformly distributed
load q. However, this factor seems to depend on the torsional-flexural rigidities ratio γ and
the central angle Φ of curved girder bridges.
Therefore, let us now investigate the value of γ for various types of curved girder bridges.
The values of torsional-flexural rigidity ratio γ were determined from actual bridges. These
results are plotted in Fig. 19. From this figure, the relationships between γ and Φ are given
by the following inequalities:
The variation in the displacement of a curved girder relative to a straight girder is given by ν.
The rate of increase of the relative displacements can be expressed as
(71)
Page 25
A limiting value of the central angle Φ of the bridge can now be determined if we assume
that the greatest variation between the curved and straight girder deformations is 5 percent.
The angle Φ will be a function of γ and of the type of system. The results are given by the
following inequalities:
<0.2 (72a)
These relationships give the minimum angle Φ for which the bridge can be designed as a
straight girder. However, when ε≥5 percent for central angles Φ greater than those predicted
in Eq. (72), the design should be governed by curved beam theory. However, many
remaining problems need to be solved in connection with the buckling strength of curved
girders when Φ takes large values. So, it is better to set a limitation on the central angle Φ,
to avoid the considerable reduction of stiffness in the curved girder bridges. For instance,
setting ε=25 percent results in the following inequalities:
(73a)
Page 26
4. SUMMARY AND CONCLUSION
The comparison analysis of warping stress σω and bending stress σb revealed that the
evaluation the warping stress σω is not important for monobox girder bridges, whereas
the evaluation of the warping stress σω is required for twin-box or multicell curved girder
bridges.
While investigating the effect of diaphragm spacing of the box girder it was concluded
that the longer is the distance between diaphragms, the bigger is the value of σDω/σb. The
analysis revealed also, that the effects of the plate thicknesses of the cross section tu, tw
and tl and of the dimensions a and h on σDω/σb are negligible, but the effect of the spacing of
the web plate b is significant.
Practical and effective formulas for design work can be derived from the complex stress
situation of warping by analysing actual bridge cases.
References:
1. Nakai, H. & Yoo,C.H.: Analysis and Design of Curved Steel Bridges. McGraw-Hill 1988.
Chapter 3.3, pages 198 – 224.
Page 27