You are on page 1of 11

International Journal of Steel Structures

December 2014, Vol 14, No 4, 831-841


DOI 10.1007/s13296-014-1214-y

www.springer.com/journal/13296

Influence of Corrosion on Load-Carrying Capacities of


Steel I-Section Main-Girder End and Steel End Cross-Girder
Eiki Yamaguchi*, Toshiaki Akagi, and Hiroyuki Tsuji
Department of Civil Engineering, Kyushu Institute of Technology, Kitakyushu, Japan

Abstract

Water leakage from an expansion joint is often found, leading to the deterioration of corrosion environment near the girder
end of a steel bridge. As a result, corrosion occurs much more often at the girder ends than in the other parts. The main-girder
end is subjected to large concentrated load at the bearing. The cross girder located at the end of a main girder is important for
seismic resistance. In the present study, the load-carrying capacities of the corroded main-girder end and the end cross-girder
are investigated by the nonlinear finite element analysis. Various patterns of local corrosion are taken into account. In both
girders, the influence of corrosion is found to vary from corrosion pattern to corrosion pattern rather considerably.

Keywords: steel bridge, corrosion, load-carrying capacity, main-girder end, end cross-girder, nonlinear finite element analysis

1. Introduction The leakage water would create corrosive environment


around the girder end easily.
The maintenance of bridges is now a very important Consequently, corrosion does not develop uniformly
issue in Japan to keep a highway network in good shape, over the entire steel girder. Kayser and Nowak (1989)
as many bridges were constructed in 1970s/1980s and stated that corrosion was usually more extensive in the
they are getting old. As for the steel bridge, corrosion is girder end. A survey in Japan revealed that about 90% of
one of the most influential phenomena in its service life: severe corrosions were observed in the girder end
the corrosion could reduce the load-carrying capacity of (Tamakoshi et al., 2006).
the steel bridge, threatens its safety and eventually terminate The main-girder end is subjected to large concentrated
its service. Due to humidity in summer, airborne salt load at the bearing. The cross girder located at the end of
attack caused by typhoon/strong winter-wind and much a main girder plays an important role for bridge safety
usage of deicing salt in winter, corrosion in steel bridges under seismic loading (JRA, 2012b): in fact, bridges
is indeed a crucial issue in Japan (Murakoshi, 2006): without end cross-girders had been constructed in Chile
Some 15% of highway steel bridges renewed and some and many of them were badly damaged in the 2010 Chile
50% of railway bridges renewed were said to be due to Earthquake (Chen et al., 2010). Therefore, corrosion at
corrosion (Huang et al., 2002; JRA, 2012a). these parts of a steel bridge is considered one of the most
An expansion joint located at the end of a bridge is one important issues for the maintenance of a steel bridge.
of the most susceptible elements of the bridge to damage From this viewpoint, some studies have been conducted
because of the impact loads of vehicles. The damage of in recent years: Liu et al. (2011) have investigated the
the expansion joint causes water leakage. A space around influence of corrosion in the web of the main-girder end
the girder end is quite congested by many bridge members on shear failure and Khurrama et al. (2012) have studied
and elements such as main girders, end cross-girders (or the influence of the thinning of a bearing stiffener due to
end cross-frames), bearings, a bridge seat and a parapet. corrosion on its load-carrying capacity. Nevertheless, this
class of research on the main girder is very limited. As for
Note.-Discussion open until May 1, 2015. This manuscript for this the end cross-girder, no study on the influence of corrosion
paper was submitted for review and possible publication on March has been conducted so far.
26, 2014; approved on December 1, 2014. Against the background of the above information, the
© KSSC and Springer 2014 load-carrying capacities of the corroded main-girder end
*Corresponding author and the end cross-girder are investigated numerically in
Tel: +81-93-884-3110; Fax: +81-93-884-3100 the present study. Various patterns of corrosion are taken
E-mail: yamaguch@civil.kyutech.ac.jp into account, and their influences are discussed.
832 Eiki Yamaguchi et al. / International Journal of Steel Structures, 14(4), 831-841, 2014

Figure 1. Main-girder model.

Figure 2. End cross-girder model.

2. Analysis Models of Steel Girders


Referring to a steel bridge model employed in the book
by Japan Bridge Association (JBA, 2005), the main girder
model and the end cross-girder model shown in Figs. 1
and 2 are employed for the present investigation. The end
cross-girder model includes part of the main girder,
100 mm long on each side of the end cross-girder. The Figure 3. Panels L and R in web of end cross-girder.
book intends to provide a standard design procedure so
that these girders are typical ones. The material properties
of Young’s modulus E, Poisson’s ratio and the yield stress it is expected to resist earthquake. To this end, static
for the steel of the girders are 2.0×105 N/mm2, 0.3, 235 N/ horizontally-distributed load is applied to the upper flange,
mm2, respectively. The uniaxial material behavior of steel as indicated by the arrows in Fig. 2, in order to investigate
is of bilinear type with the second slope of E/100. basic characteristics of the behavior of the end cross-
The corrosion in the main girder end influences the girder. The load represents an inertia force associated
load-carrying capacity under vertical load above the with the concrete slab.
bearing (Huang et al., 2002). Therefore, the vertical load Figure 2 is the side view of the cross-girder model on
is applied to the upper flange above the bearing for the the parapet side. The web of the end cross-girder model
investigation of the corroded main-girder end. has two longitudinal stiffeners and one transverse stiffener.
On the other hand, the load-carrying capacity of an end The stiffeners are indicated by the dotted lines as they are
cross-girder under seismic loading is of much interest, as installed on the opposite side of the web. The stiffeners
Influence of Corrosion on Load-Carrying Capacities of Steel I-Section Main-Girder End and Steel End Cross-Girder 833

and 10 corrosion models for the end cross-girder. The


models are shown in Figs. 4 and 5. It is assumed that
corrosion develops on one side of the girder.
In Fig. 5, each figure illustrates more than one corrosion
model, which is just for the sake of conciseness and does
not mean that the corrosions in one figure are considered
together in the analysis: the corrosion models are taken
into account in the analysis one by one.
The end cross-girder corrodes due to water leakage
from the expansion joint, so that the parapet side of the
end cross-girder tends to undergo more severe corrosion,
and the stiffeners are set up on the other side for better
maintenance. Therefore, the corrosion of the stiffeners is
not considered in the corrosion models for the end cross-
girder.
In the main-girder end, CM-S is located in the transverse
stiffener at the bearing, while the other corrosion models
are in the web; CM-C is in the web of the cantilever part
of the man girder and the other two are in the other part
of the web, right next to the transverse stiffener. In the
end cross-girder, all the corrosions are on the parapet side
of the web.
Various degrees of corrosion are considered. For the
Figure 4. Corrosion model for main-girder end.
main-girder end, the height h and the plate-thickness loss
on one side of a plate ∆t are changed; 10, 20, and 40% of
h0 are given to h while ∆t takes 2, 4, and 6 mm. For the
and the main girders form two panels, Panel L and Panel end cross-girder, the lengths h and w are 25, 50, 75% of
R, as indicated in Fig. 3. h0 and w0, respectively, while ∆t takes 2, 4, and 6 mm. In
all the corrosion models, ∆t is assumed uniform over each
3. Corrosion Models corroded region.
Including the original (non-corroded) girder, 37 main
Based on the observations available in literature (Huang girders and 91 end cross-girders are to be analyzed in the
et al., 2002; Liu et al., 2011; Tamakoshi et al., 2006), 4 present study.
corrosion models are constructed for the main-girder end,

Figure 5. Corrosion model for end cross-girder (unit: mm).


834 Eiki Yamaguchi et al. / International Journal of Steel Structures, 14(4), 831-841, 2014

Figure 6. Residual stress distribution.

4. Outline of Analysis
The load-carrying capacities of the two girders are
evaluated by the nonlinear finite element analysis with
material and geometrical nonlinearities. Initial imperfections
of the steel girder can influence the load-carrying capacity.
Initial imperfections are therefore taken into account as
well: the effects of initial deformation and residual stress
are included in the analysis. All the analyses are conducted
by ABAQUS (2008). 4-node shell elements are employed
to model the whole structures. Similar numerical study
was conducted by Khurrama et al. (2012), the results of
which compare well with experimental data.
To find the relevant initial deformation, the eigenvalue
analysis is conducted first to obtain buckling modes. The
initial deformation mode is made identical to the mode of
the smallest buckling strength. The initial deformation is
then constructed so as to have the maximum displacement
within the range of the fabrication tolerance specified in
Japanese Specifications for Highway Bridges Part 2 Steel
Bridges (JRA, 2012a).
Referring to residual stress distributions given in the
literature by Usami (2005), the distributions shown in
Fig. 6 are employed in the present analyses. Since the
residual stress distributions are in a state of self equilibrium,
the residual stresses are given to the analysis models by Figure 7. Analysis of quarter main-girder.
conducting thermal stress analysis. The analysis is done
by assuming temperature distribution so as to insert the
stress distribution desired. The relevant temperature The load-carrying capacity under the vertical load
distribution to this end is found by a trial-and-error above the bearing is the main issue for the main-girder
method. end. Since both the load and the corrosion lie in the
Both girders are simply supported. Moreover, in the neighborhood of the bearing, deformation is expected to
bridge, the upper flanges of the main girder and the end occur in a local region of the main girder. If so, it may not
cross-girder are connected to the concrete slab, which be necessary to analyze the whole girder.
restricts the torsional rotation and the lateral displacement To scrutinize the point, a quarter-length main-girder
normal to the girder axis of the upper flanges. Therefore, model is constructed (Fig. 7(a)). Finer mesh is employed
in the present girder models the constraints are imposed in the region within which the deformation is expected to
on the upper flange of each girder so that the upper flange take place. The full girder model and the quarter girder
does not rotate torsionally and does not cause out-of- model are analyzed; indeed the results are so close to
plane displacement. each other as presented in Fig. 7(b). In the following
Influence of Corrosion on Load-Carrying Capacities of Steel I-Section Main-Girder End and Steel End Cross-Girder 835

Table 1. Load-carrying capacity of main-girder end


Corrosion model ∆t (mm) h/h0 Pmax/P0 Corrosion model ∆t (mm) h/h0 Pmax/P0
0.1 0.998 0.1 1.000
2 0.2 0.952 2 0.2 0.921
0.4 0.924 0.4 0.893
0.1 0.928 0.1 0.926
CM-S 4 0.2 0.841 CM-C 4 0.2 0.848
0.4 0.787 0.4 0.806
0.1 0.804 0.1 0.865
6 0.2 0.747 6 0.2 0.772
0.4 0.714 0.4 0.728
0.1 1.000 0.1 1.000
2 0.2 0.993 2 0.2 0.987
0.4 0.962 0.4 0.941
0.1 1.000 0.1 1.000
CM-N 4 0.2 0.987 CM-W 4 0.2 0.961
0.4 0.913 0.4 0.883
0.1 0.999 0.1 0.998
6 0.2 0.970 6 0.2 0.919
0.4 0.871 0.4 0.819

analyses, hence, the quarter main-girder model is to be capacities of all the results obtained for each corrosion
used for saving computational cost. model are 0.855 for CM-S, 0.862 for CM-C, 0.966 for
CM-N and 0.945 for CM-W.
5. Numerical Results These observations indicate that CM-S is the most
influential in the load-carrying capacity, CM-C is a close
5.1. Main-girder end second, CM-W is a distant third and CM-N is the least
Numerical results are presented in Table 1 and Figs. 8- influential; while the differences between CM-S and CM-
10. Pmax is the load-carrying capacity of each corroded C and between CM-W and CM-N are small, a large gap
main-girder end and P0 is that of the original main-girder between CM-C and CM-W is noteworthy.
end. From the viewpoint of the reduction in the load-
The load-carrying capacity is reduced by the corrosion. carrying capacity, the corrosion models can be thus
The influence of the corrosion varies from corrosion grouped into two: one consists of CM-S and CM-C, and
model to corrosion model. the other of CM-W and CM-N. The former is more
In the case of h/h0 =0.1, the reductions in the load- influential in the load-carrying capacity than the latter.
carrying capacities of CM-N and CM-W are very small. The corrosion area of CM-W is twice as large as that of
Even with ∆t =6 mm, the decrease in the capacities of CM-N. Naturally, the reduction in the load-carrying
CM-N and CM-W are only 0.1 and 0.2%, respectively. capacity of CM-W is larger for the same h/h0 and ∆t . But
On the other hand, while the reductions in the load- the difference is not significant, only about 5% at most.
carrying capacities of CM-S and CM-C are little for h/h0 The corrosion area of CM-W for h/h0 =0.4 is much larger
=0.1 and ∆t =2 mm, they become 20 and 14%, respectively, than that of CM-S for h/h0 =0.2, yet for ∆t =4 and 6 mm
when ∆t increases to 6 mm. For a large h/h0, the decrease the load-carrying capacity of CM-S is smaller than that of
in the load-carrying capacity of CM-S and CM-C is CM-W. The observations indicate that the corrosion area
significant even for ∆t =2 mm: with h/h0 =0.4, for example, is not necessarily the most critical factor for the reduction
the reductions are 8 and 11%, respectively. in the load-carrying capacity.
Smallest capacity is reached at the maximum development Local buckling occurs at around Pmax in all the analyses.
of the corrosion, h/h0 =0.4 and ∆t =6 mm, at which the An example of the deformed configuration is given in
maximum reductions in the load-carrying capacity are Fig. 10. A large out-of-plane displacement is recognized
found as 29% for CM-S, 27% for CM-C, 13% for CM- in the corroded region. This suggests that the boundary
N and 18% for CM-W. condition of the corroded region can be important for the
The average load-carrying capacity over all ∆t for each reduction in the load-carrying capacity. In fact, the
corrosion model is plotted in Fig. 8(e), and the average difference between the two groups of corrosion lies in the
capacity over all h/h0 is in Fig. 9(e). The average constraints along the boundary of the corroded region: a
836 Eiki Yamaguchi et al. / International Journal of Steel Structures, 14(4), 831-841, 2014

Figure 8. Variation of load-carrying capacity with respect to h/h0 (main-girder end).

part of the boundary of the corroded region in the first model that undergoes a larger reduction in the load-
group of CM-S and CM-C, is free, while no part of the carrying capacity, is more important. For instance, for h/h0
boundary is free in the case of the second group of CM- =75% and ∆t =6 mm, Pmax/P0 of CE-LL is 0.911 while
S and CM-C. The free boundary is less resistant to out- that of CE-LR is 0.964; for the objective of the present
of-plane displacement, and indeed the capacity reduction research, CE-LL is more significant than CE-LR. From
in the former group is larger than that in the latter group. this viewpoint, even though all the 91 end cross-girders
were analyzed, only the results of CE-LL, CE-HL, CE-
5.2. End cross-girder WL, CE-NL and CE-UR are presented in Table 2 and
In the present study, pairs of the corrosion models Figs. 11-14.
located in the symmetric positions such as CE-LL and In the original girder, shear buckling takes place in both
CE-LR are considered. Monotonic horizontal load is Panels L and R. Then the diagonal tension field is formed
applied herein, but actual seismic load is cyclic. Therefore, before P0 is reached.
only one of the paired corrosion models, a corrosion In CE-LL, the diagonal tension field is formed in both
Influence of Corrosion on Load-Carrying Capacities of Steel I-Section Main-Girder End and Steel End Cross-Girder 837

Figure 9. Variation of load-carrying capacity with respect to ∆t (main-girder end).

panels. But out-of-plane displacement is larger in Panel


R. The diagonal tension field in Panel L is wider and
located in a higher position.
Except for the most corroded case of w/w0 =75% and
∆t =6 mm, the end cross-girder of CE-HL deforms in a
very similar way to that of the original girder and the
reduction in the load-carrying capacity is very limited.
However, for w/w0 =75% and ∆t =6 mm, it deforms very
differently (Fig. 11(c)): the diagonal tension field does
not develop in Panel L; two diagonal tension fields are
formed in Panel R; and out-of-plane displacement in
Panel R extends beyond the lower longitudinal stiffener.
The reduction in the load-carrying capacity amounts to Figure 10. Deformed configuration (CM-C with ∆t =6
15% for w/w0 =75% and ∆t =6 mm. mm and h/h0 =0.2; deformation magnification factor=2).
838 Eiki Yamaguchi et al. / International Journal of Steel Structures, 14(4), 831-841, 2014

Table 2. Load-carrying capacity of end coss-girder


Corrosion model ∆t (mm) h/h0 Pmax/P0 Corrosion model ∆t (mm) w/w0, h/h0 Pmax/P0
0.25 0.997 0.25 0.997
2 0.50 0.984 2 0.50 0.995
0.75 0.993 0.75 0.996
0.25 0.984 0.25 0.986
CE-LL 4 0.50 0.968 CE-HL 4 0.50 0.987
0.75 0.972 0.75 0.989
0.25 0.954 0.25 0.980
6 0.50 0.923 6 0.50 0.979
0.75 0.911 0.75 0.849
0.25 0.975 0.25 0.983
2 0.50 0.956 2 0.50 0.963
0.75 0.934 0.75 0.949
0.25 0.961 0.25 0.971
CE-WL 4 0.50 0.895 CE-NL 4 0.50 0.923
0.75 0.848 0.75 0.887
0.25 0.937 0.25 0.958
6 0.50 0.799 6 0.50 0.850
0.75 0.735 0.75 0.788
0.25 0.980
2 0.50 0.982
0.75 0.982
0.25 0.979
CE-UR 4 0.50 0.957
0.75 0.975
0.25 0.974
6 0.50 0.964
0.75 0.953

In CE-WL, the diagonal tension field is found in both


panels. In addition, there is a region of out-of-plane
displacement near the left bottom corner of the web. The
load-carrying capacity reduces as much as by 26%.
In CE-NL, the same deformation characteristics as
those of CE-WL are observed, but the influence on the
load-carrying capacity is smaller: the maximum reduction
in the load-carrying capacity is 21%.
In CE-UR, the diagonal tension field forms in both
panels. The influence on the load-carrying capacity is
rather small: the reduction of the load-carrying capacity is
about 5% at most.
An overall observation is that corrosion tends to shade
off the formation of a diagonal tension filed: as the corrosion
becomes more severe, the diagonal tension field becomes
less clear, resulting in a larger reduction in the load-
carrying capacity.
The degree of the reduction varies from corrosion
model to corrosion model. For the investigation of this
point, the average load-carrying capacity over all h/h0 or
all w/w0 for each corrosion model is plotted in Fig. 12(f), Figure 11. Deformed configuration at Pmax (h/h0 (w/w0)=
and the average capacity over all h/h0, w/w0 is in Fig. 75%, ∆t =6 mm; deformation magnification factor=2).
Influence of Corrosion on Load-Carrying Capacities of Steel I-Section Main-Girder End and Steel End Cross-Girder 839

Figure 12. Variation of load-carrying capacity with respect to ∆t (end cross-girder).

13(f). Moreover, the average capacities of all the results 6. Concluding Remarks
for each corrosion model are found as 0.965 for CE-LL,
0.973 for CE-HL, 0.893 for CE-WL, 0.919 for CE-NL The influence of corrosion on the load-carrying capacities
and 0.972 for CE-UR. Figure 14 shows the relationship of the main-girder end under vertical load and the end
between Pmax and the corroded area AC. In all those cross-girder under horizontal load was investigated by the
results, CE-WL and CE-NL give a larger reduction in the nonlinear finite element analysis. Various corrosion models
load-carrying capacity, and the difference between CE-WL constructed based on the actual corrosions encountered in
and CE-NL is not much. It can be concluded therefore steel bridges were taken into account. Several degrees of
that CE-WL and CE-NL have a greater influence on the corrosion in terms of the size of the corroded region and
load-carrying capacity under horizontal loading. the plate-thickness loss were also considered for each
840 Eiki Yamaguchi et al. / International Journal of Steel Structures, 14(4), 831-841, 2014

Figure 13. Variation of load-carrying capacity with respect to h/h0 , w/w0 (end cross-girder).

corrosion model. the second group, while the gap between the two groups
The numerical results show that with the extension of is much larger, no less than 0.090. The area of the
the corrosion, the reduction in the load-carrying capacity corroded region is not necessarily a decisive factor for the
increases, as expected. But the way the corrosion influences load-carrying capacity reduction. The constraint along the
the capacity varies from corrosion pattern to corrosion boundary of the corroded region appears to be more
pattern rather considerably in both girders. influential: the corroded region having a free-boundary
In the main-girder end, it is found that the four corrosion part reduces the load-carrying capacity greatly, as the
models studied can be grouped into two, depending on local buckling in the corroded region is the phenomenon
the degree of influence: one consists of those in the that leads to the ultimate state of the main-girder end.
vertical stiffener (CM-S) and in the cantilever part of the In the end cross-girder, the corrosion developed in a
main-girder web (CM-C); the second consists of those in triangular region at the bottom corner of the web (CE-
the web on the central side of the main girder (CM-N, WL, CE-NL) tends to deteriorate the load-carrying capacity
CM-W). On average, the difference in Pmax/P0 between more greatly: on average those corrosions lead to the
the corrosions in the first group is 0.007 and it is 0.021 in reduction of about 0.1 in Pmax/P0 while the reductions due
Influence of Corrosion on Load-Carrying Capacities of Steel I-Section Main-Girder End and Steel End Cross-Girder 841

References
ABAQUS (2008). User’s Manual, ABAQUS Ver. 6.8.
Dassault Systemes Simulia Corp.
Chen, G., Yen, P.W., Buckle, I., Allen, T., Alzamora, D., Ger,
J., and Arias, J. G. (2010). “Chile Earthquake implications
to the seismic design of bridges.” Proc. 26th US-Japan
Bridge Engineering Workshop, pp. 203-216.
Huang, V. T., Nagasawa, H., Sasaki, E., Ichikawa, A., and
Natori, T. (2002). “An experimental and analytical study
on bearing capacity of supporting point in corroded steel
bridges.” Journal of JSCE, No. 710/I-60, pp. 141-151.
JBA (2005). Design example and commentary of composite
Figure 14. Variation of load-carrying capacity with respect girder. Japan Bridge Association (in Japanese).
to AC/(w0h0).
JRA (2012a). Specifications for Highway Bridges, Part 2:
Steel Bridges. Japan Road Association (in Japanese).
to the other corrosions (CE-LL, CE--HL, CE--UR) are JRA (2012b). Specifications for Highway Bridges, Part 5:
only about 0.03. The corrosion areas of CE-WL and CE- Seismic Design. Japan Road Association (in Japanese).
NL are larger than in the other corrosion models. However, Kayser, J. R. and Nowak, A. S. (1989). “Relaiability of
the corrosion area does not to play a decisive role in the corroded steel girder bridges.” Structural Safety, 6, pp.
capacity reduction: the decreases in the load-carrying 53-63.
capacities due to CE-WL and CE-NL are larger than Khurrama, N., Sasaki, E., Kihira, H., Katsuchi, H., and
those due to the other corrosions, even when the corrosion Yamada, H. (2012). “Analytical demonstrations to assess
residual bearing capacities of steel plate girder ends with
areas are the same.
stiffeners damaged by corrosion.” Structure and
The present results would help estimate the damage
Infrastructure Engineering: Maintenance, Management,
due to corrosion. Considering the complexity in actually Life-Cycle Design and Performance, DOI:10.1080/
corrosion development, the quantitative estimate may 15732479.2012.697904.
serve only as first-order approximation, yet it could be of Liu, C., Miyashita, T., and Nagai, M. (2011). “Analytical
much help to judge the seriousness of the corrosion study on shear capacity of steel I-girder with local
problem at hand. corrosion nearby girder ends.” Journal of Structural
In any case, however, it is always important to treat the Engineering, JSCE, 57A, pp. 715-723.
corrosion as early as possible: steel can never recover Murakoshi, J. (2006). “Maintenance of steel bridges.” Proc.
from corrosion by itself; the extensive development of 15th Conference on Public Work and Development in
corrosion, regardless of its location, could threaten the Asia, National Institute for Land and Infrastructure
bridge safety, and delayed treatment would make the Management, MLIT, Japan.
maintenance cost very high. Tamakoshi, T., Nakasu, K., Ishio, M., Takeda, T., and Suizu,
N. (2006). Research on local corrosion of highway steel
bridges. Technical Note of National Institute for Land
Acknowledgments and Infrastructure Management, No. 294 (in Japanese).
Usami, T. (2005). Guidelines for stability design of steel
The financial support for the present study, Grant-in- structures, 2nd ed., JSCE (in Japanese).
Aid for Scientific Research (C) (KAKENHI, No. 23560570),
is gratefully acknowledged.

You might also like