Experimental Study on Fatigue Strength of Corroded

Bridge Wires
Shunichi Nakamura, M.ASCE1; and Keita Suzumura2

Abstract: Fatigue tests were conducted for corroded galvanized steel wires on three corrosion levels, showing that fatigue strength of corroded
wires lowers as corrosion progresses. Corrosion pits were measured on the corroded specimens, showing severer corrosion produced deeper pits
Downloaded from ascelibrary.org by CASA Institution Identity on 09/01/21. Copyright ASCE. For personal use only; all rights reserved.

in more condensed areas. Fatigue tests were then conducted for wire specimens with artificial pits whose sizes were decided by the measured
corrosion pit data. Three different pit shapes were assumed: round, triangle, and triangle with a notch. The wire specimens with round pits did
not break until 1 million cycles in the stress range of 400 MPa. The fatigue strength of wires with the triangular pit was lower than that with
a round shape. Triangular pit specimens broke at fewer cycles for shorter pit length. The fatigue strength of wires with a notched triangle further
decreased, and critical cycles did not depend on pit length. As the S-N relation of the wire specimens with triangular pits and notched triangular
pits has a similar tendency as those of the actually corroded wires, the pit shape seems to be a dominant factor in lowering fatigue strength. The
stress concentration factor at the sharp edge of the pits were obtained by strain gauge measurement and FEM analysis. Both methods showed that
the stress concentration is larger for sharper pit shapes, indicating that this is the major cause for the decrease of fatigue strength. DOI: 10.1061/
(ASCE)BE.1943-5592.0000366. © 2013 American Society of Civil Engineers.
CE Database subject headings: Cable-stayed bridges; Steel; Fatigue; Experimentation; Corrosion; Stress concentration; Wire products.
Author keywords: Bridge cables; Galvanized steel wires; Fatigue strength; Corrosion pits; Stress concentration.

Introduction
Bridge cables consist of high tensile galvanized steel wires. Many old
suspension bridges all over the world have deteriorated cables. Some
of the steel wires of the main cables are heavily corroded and frac-
tured. Rehabilitation work of corroded wires was carried out on the
Brooklyn Bridge, the Williamsburg Bridge, and other bridges (Betti
and Yanev 1999; Mayrbaurl 2000; Stahl and Gagnon 1996). When
water enters a suspension bridge cable, the wires that make up the
cable start to deteriorate. The protective zinc coating is the first element
that is damaged, followed by corrosion of the steel itself. Hanger ropes
of suspension bridges and stays of cable-stayed bridges have also been
severely corroded on many bridges (Stahl and Gagnon 1996).
Long-span suspension bridges are relatively new in Japan.
However, it was found in field surveys that galvanized steel wires
were partly corroded on several suspension bridges within 10 years
after the bridges were completed (Furuya et al. 2000). Wide surveys
were carried out on these bridges. Water was present inside the cable
and made the environment inside the cable highly humid, which
caused the corrosion of wires. However, the steel corrosion was not
deep, and no broken wires were observed.
Tension tests were carried out for corroded galvanized wires and
found that the actual tensile strength, breakage load divided by
1
Professor, Dept. of Civil Engineering, Tokai Univ., 4-1-1 Kitakaname,
Hiratsuka 259-1292, Japan (corresponding author). E-mail: snakamu@
keyaki.cc.u-tokai.ac.jp
2
Manager, Research Laboratory, Nippon Steel Engineering, 20-1
Shintomi, Futtu 293-0011, Japan.
Note. This manuscript was submitted on May 28, 2011; approved on
January 3, 2012; published online on January 5, 2012. Discussion period
open until August 1, 2013; separate discussions must be submitted for
individual papers. This paper is part of the Journal of Bridge Engineering,
Vol. 18, No. 3, March 1, 2013. ©ASCE, ISSN 1084-0702/2013/3-200–209/
$25.00.
reduced cross-sectional area, does not decrease with the corrosion
level (Nakamura et al. 2004). Conversely, although elongation does
not change when only the galvanized layer is corroded, it decreases
when the steel part starts to corrode.
The mechanism of fracture of corroded wires has been widely
studied in the past. Although many studies have been conducted to
clarify the mechanisms that cause wire breakage, discussions are
still in progress. Some researchers have found that the wires suffer
high tensile stresses and bending stresses and residual stresses, along
with the corrosive environment, which leads to stress corrosion
cracking or hydrogen-assisted cracking (Mayrbaurl and Camo 2001).
Others have found that the hydrogen concentration does not in-
fluence the loss of ductility of the wires, indicating that hydrogen
absorption into the steel core is not a controlling factor in the
embrittlement process (Betti et al. 2005). The amount of hydrogen
absorbed in the steel layer does not reach the level that causes
hydrogen embrittlement, and wire breakage is more susceptible to
corrosion fatigue (Nakamura et al. 2004; Nakamura and Suzumura
2009; Tarui et al. 2001).
Broken wires of an old suspension bridge were investigated. The
fracture surface was similar to that caused by fatigue rather than by
hydrogen embrittlement (Nakamura et al. 2004; Nakamura and
Suzumura 2009). It was therefore hypothesized that the wires were
fractured by the mixed effects of corrosion, cyclic stresses, and
hydrogen. Hanger ropes have a high stress fluctuation that increases
risks of wire breakage caused by corrosion fatigue.
This paper presents studies on the fatigue strength of corroded
galvanized steel wires in four steps.
In the first step, corroded wires of different corrosion levels are
produced at the laboratory, and fatigue tests are conducted with these
corroded specimens and new wires. Because actual bridge wires are
usually in a wet and warm environment, the fatigue strength of
corroded wires under wet environments may be less than that in dry
environments because of corrosion fatigue (Uhlig and Revie 1963).
The tests are therefore conducted under dry and wet environments to

200 / JOURNAL OF BRIDGE ENGINEERING © ASCE / MARCH 2013

J. Bridge Eng., 2013, 18(3): 200-209

 find the fatigue strength of corroded wires (Nakamura et al. 2004;
Nakamura and Suzumura 2009).
It is known that surface notches decrease the fatigue strength
because of stress concentration (Pilkey and Pilkey 2008). Stress
concentration depends on the shapes of the notches and can be
calculated for various notch shapes (Pilkey and Pilkey 2008). Miki
et al. (1981) produced steel plates with artificial pits and carried out
fatigue tests. They showed that sharper notches decrease fatigue
strength. However, these analyses and experiments have mainly
been applied to steel plates and bars and have not targeted gal-
vanized steel wires. Furthermore, almost no research has been
carried out to clarify the effects of sizes and distribution of pits on
the wire surface caused by corrosion. It is, therefore, not known
Downloaded from ascelibrary.org by CASA Institution Identity on 09/01/21. Copyright ASCE. For personal use only; all rights reserved.

how corrosion levels and corrosion pits relate. In the second step,
surface roughness or pits are measured on the corroded galvanized
steel wires to find the relationship between corrosion levels and
pit sizes.
In the third step, cyclic tests are conducted for the wire specimens
with artificial pits whose shapes and sizes are decided based on the
measured data. Although corrosion pits are physicochemical defects
accompanied by small cracks, mechanical defects are used to in-
vestigate the effect of the size and shape of the pits. The specimens
with nine different shapes and sizes are used. Then, the fatigue
strength of the wires with artificial pits are compared with that of
actually corroded wires to find the effect of pit shapes and sizes to
fatigue strength.
A decrease in fatigue strength may be caused by stress concen-
tration at the sharp notch of the corrosion pit. In the fourth step, the
stress concentration factor is evaluated by two methods: measure-
ment by strain gauges and FEM analysis. Four wire models with
different artificial pit shapes are tensioned, and strains are measured
at the pit bottom and at the parallel parts, from which stress con-
centration factors are obtained. The same models are analyzed by
FEM, and the maximum stresses are calculated, by which the stress
concentration factor is obtained. Then, there is a discussion of how
fatigue strength relates to stress concentration and is dependent on
pit shapes.
Fatigue Strength of Wire Specimens on Three
Corrosion Levels
Fatigue tests were conducted with corroded wire specimens on three
corrosion levels, and the results are described and discussed in this
section (Nakamura and Suzumura 2009). A fatigue test specimen
was a steel galvanized wire with a diameter of 5.0 mm. The tensile
strength was 1,570 MPa. The attached zinc mass was 350 g/m2,
which was equivalent to 50 mm in thickness. A piano wire with a
carbon content of about 0.8% is drawn into a wire, which has a
pearlite microstructure.
Galvanized steel wires were wrapped with wet gauze and kept in
a chamber at a temperature of 40°C to accelerate corrosion. The wire
covered with gauze is kept in a confined box. This box is kept in
a chamber where temperature and relative humidity are maintained
constant. These enclosure measures make the gauze wet. It was
found in previous studies (Furuya et al. 2000; Nakamura et al. 2004;
Suzumura and Nakamura 2004) that these corrosion acceleration
simulation tests produce the same corrosion phenomenon of actual
cables. The wires were kept for 3, 8, and 12 months, producing the
corroded wires on three different corrosion levels. Their appearances
are shown in Fig. 1. The galvanized steel wires after 3 months were
covered with a white zinc corrosion substance (Corrosion Level 1),
ferrous rust occurred locally in the specimens after 8 months
(Corrosion Level 2), and ferrous rust spread widely in the specimens
Fig. 1. Wire specimen on three corrosion levels
after 12 months (Corrosion Level 3). The corroded wires from which
corrosion substances were removed are also shown in Fig. 1. The
corrosion substances on the specimen surface were removed by
a cloth containing 10% H2SO4.
The fatigue strength of corroded galvanized steel wires may
decrease as corrosion progresses. Cyclic tests were therefore con-
ducted for the new wires and corroded wires on Corrosion Levels 1–
3. A specimen was held with a distance of 100 mm between the grips
and cyclically loaded with a cycle of 64 Hz. Cyclic tests were
conducted, keeping the minimum stress at 500 MPa and varying the
stress range from 200 to 600 MPa. The tests were conducted with
a relative humidity under 50%.
It is known that the fatigue strength of high-tensile-strength steel
plates further decreases when they are kept under wet conditions
(Uhlig and Revie 1963). This is called corrosion fatigue. It is
reported that there is no fatigue limit for corrosion fatigue of high-
tensile steel plates. If there is no fatigue limit, even a small stress
fluctuation could initiate a fatigue crack. Further study is required to
see whether corroded galvanized steel wires also have this corrosion
fatigue property.
The bridge wires are often exposed to wet environments while
they are cyclically stressed because of temperature change and live
loads. Fatigue tests were therefore conducted under wet environ-
ments to clarify the corrosion fatigue effect of galvanized steel wires.
The fatigue test method under wet environment is as follows. A
corroded wire is wrapped with wet gauze, inserted into a silicon
tube, sealed with vinyl tape at the upper and bottom ends, and cy-
clically loaded (Figs. 2 and 3). A specimen was held at a distance of
100 mm between the grips and cyclically loaded with a cycle of
64 Hz, which was the same as the dry cyclic tests. It is preferable to
carry out fatigue tests with slower cycles to confirm the effect of

JOURNAL OF BRIDGE ENGINEERING © ASCE / MARCH 2013 / 201

J. Bridge Eng., 2013, 18(3): 200-209

Downloaded from ascelibrary.org by CASA Institution Identity on 09/01/21. Copyright ASCE. For personal use only; all rights reserved.
Fig. 2. Fatigue test specimen under a wet environment
Fig. 3. Fatigue tests under a wet environment
corrosion fatigue. However, the high-frequency test is convenient
from a practical point of view. Fatigue tests were conducted for steel
wires with 10 and 62 Hz and both had the same results. Other re-
search has carried out fatigue tests for steel plates under various
environmental conditions with cycles of 24 and 30 Hz, and the
corrosion fatigue phenomenon was observed (Uhlig and Revie
1963). This also indicates that 62 Hz is not too fast to prove the
existence of the corrosion fatigue of steel wires.
Fig. 4 shows the S-N data of the Level 2 specimens under dry and
wet environments (Nakamura and Suzumura 2009). The S-N data of
new wires are also plotted in this figure. It can be seen that the fatigue
strength of new wires is higher than that of Level 2 wires under dry
conditions, and that of Level 2 wires under wet conditions is even
less. Other data show that there is not much difference between the
fatigue strength of the new and corrosion Level 1 specimens. It is
therefore clear that the fatigue strength does not change when only
the galvanized layer is corroded, but it significantly decreases after
steel corrosion below the galvanized layer progresses.
Fig. 5 shows the S-N data of new wires and the Level 3 wires
under dry and wet environments. The same tendency is seen more
clearly in this figure. It is therefore understood from Figs. 4 and 5 that
fatigue strength of new wires is the highest, and more severe
202 / JOURNAL OF BRIDGE ENGINEERING © ASCE / MARCH 2013
J. Bridge Eng., 2013, 18(3): 200-209
Fig. 4. S-N data of Level 2 specimens
Fig. 5. S-N data of Level 3 specimens
corrosion decreases fatigue strength further. There is also a tendency
that the fatigue strength under wet environments is less than that
under dry environments. Although the cyclic tests are conducted for
relatively short duration with high cyclic frequency, this test speed is
the same for the dry and wet conditions, and the difference in fatigue
strength is a result of the environment. When the steel layer starts to
corrode, stress concentrates more intensely at sharp corroded pits
and a crack occurs at the tip. The crack is accelerated by the hydrogen
that is produced by the chemical reaction in wet conditions. This is
why fatigue strength under wet conditions is less than that under dry
conditions.
Because the S-N data for wires under a stress range of 150 MPa
are not available, the existence of a fatigue limit cannot be confirmed
in this study. Because the stress ranges of the main cables of sus-
pension bridges are small, it is important to study how lower stress
ranges affect corrosion fatigue. This will be studied in the future.
Measurement of Corrosion Pits
Because corrosion pits decrease fatigue strength, it is important to
find their sizes and distributions. In this section, corrosion pit sizes
 Table 1. Maximum Depth of Measured Corrosion Pits
Wire Corrosion Dry or Width Length Depth
number level wet (mm) (mm) (mm)
1 2 Dry 4.00 5.00 0.21
2 2 Dry 4.50 10.50 0.22
3 2 Wet 0.50 0.40 0.18
4 2 Wet 2.30 3.60 0.18
5 2 Wet 4.40 2.10 0.20
6 2 Wet 3.20 4.80 0.15
7 2 Wet 1.50 1.60 0.15
8 2 Wet 4.00 3.80 0.20
9 2 Wet 0.70 0.90 0.10
10 3 Dry 4.40 3.10 0.21
Downloaded from ascelibrary.org by CASA Institution Identity on 09/01/21. Copyright ASCE. For personal use only; all rights reserved.

11 3 Dry 4.80 4.50 0.19

12 3 Dry 3.50 1.60 0.10
13 3 Dry 3.90 6.40 0.42
14 3 Dry 1.80 1.00 0.38
15 3 Wet 6.50 22.10 0.26
16 3 Wet 4.30 8.90 0.30
17 3 Wet 7.00 9.60 0.41
18 3 Wet 4.30 8.10 0.56
19 3 Wet 4.70 6.60 0.40
20 3 Wet 5.00 7.70 0.46

Fig. 6. Measurement of corrosion pits

and distribution were measured on the corroded Level 2 and Level 3

wire specimens used in the previous section.
Fig. 6 shows the appearance of a corroded wire from four angles
after corrosive substances were removed. The depth, length, and
width of the corrosion pits were measured with a specially tailored
gauge. The dial gauge with a sharp pit is fixed to the frame, which is
set parallel to the wire. The gauge can measure with an accuracy of
0.01 mm. Length is a longitudinal size along the wire axis, and width
is a circumferential size around a wire.
A total of 20 corroded wires were available. Ten deep pits were
chosen for each wire and carefully measured. Corrosion makes the
wire surface rough, which is a basically random property, and the
measured data scatter. The sizes of the deepest pits for each wire are
shown in Table 1. The average maximum depth of Level 2 wires is
0.18 mm and that of Level 3 is 0.34 mm. More severe corrosion
produces deeper pits. The maximum depth is 0.56 mm among all of
the data.
Fig. 7 shows the numbers of pits versus pit depths for 200 pieces Fig. 7. Measured pit depth and the number
of data (20 wires 3 10 pits/wires). The number of deeper pits is fewer
than that of shallower pits. Figs. 8–10 show the relations between pit
depth, width, and length. Most of the data are within the area with
a width of 10 mm and a length of 10 mm. However, some of the pits Fatigue Test Wires with Artificial Pits
with a depth less than 0.4 mm are outside of this area (Figs. 8 and 9).
Conversely, all of the pits with a depth of 0.4–0.6 mm are within this Galvanized steel wires with artificial pits were produced, and cyclic
area (Fig. 10). It is understood from these figures that relatively tests were conducted in this section. As stress concentration can be
shallow corrosion pits spread widely, and deeper corrosion pits a major factor that decreases fatigue strength and it depends on notch
concentrate on a narrower area. sizes and shapes, specimens with nine different shapes and sizes

JOURNAL OF BRIDGE ENGINEERING © ASCE / MARCH 2013 / 203

J. Bridge Eng., 2013, 18(3): 200-209

Downloaded from ascelibrary.org by CASA Institution Identity on 09/01/21. Copyright ASCE. For personal use only; all rights reserved.
Fig. 8. Measured pit length and width (depth , 0.2 mm)
Fig. 9. Measured pit length and width (depth 5 0.2–0.4 mm)
were used. Corrosion pits have many different shapes and sizes. The
most common shape is a round shape. However, it is similar to
a triangular shape when the pit is sharp and narrow. Therefore, round
and triangular shapes are adopted as representative of corrosion pits.
The wire specimens are shown in Fig. 11. Considering the
measured pit sizes, the wire specimens have 0.6-mm-deep and
204 / JOURNAL OF BRIDGE ENGINEERING © ASCE / MARCH 2013
J. Bridge Eng., 2013, 18(3): 200-209
Fig. 10. Measured pit length and width (depth 5 0.4–0.6 mm)
3.5-mm-wide pits. The width of 3.5 mm for the artificial pits is
chosen from the average of the width of the corroded wires with
a maximum depth of the measured corrosion pits (Table 1). The
measured maximum depth was 0.56 mm. Because stress con-
centration depends on the ratio of width to length, three lengths
of 3.5, 6.0, and 10.0 mm are adopted. Also, three pit shapes are
adopted. Group P has a round-shaped pit, Group Q has a triangular-
shaped pit, and Group R has a triangular-shaped pit with a notch
(a depth of 0.1 mm and a width of 0.3 mm) at the pit bottom. The
notch is seen in R-1, -2, and -3 of Fig. 11, and this was made by
a diamond file.
The total number of specimens is 20 so that at least two speci-
mens represent each shape and each length. Photographs of the
artificial pits are also shown in Fig. 11. The round and triangular
shapes were machine cut, and the notch was manufactured using
a diamond file.
Fatigue Rest Results of Wires with Artificial Pits
The wire specimens in Fig. 11 were cyclically loaded with a stress
fluctuation from 500 to 900 MPa (stress range, 400 MPa). The limit
cycle was set at 1 million. Fig. 12 shows the critical cycles (cycles
until breakage) of the wire specimens with artificial pits. All of the
Group P specimens with round-shaped pits did not break until the
cycle limit of 1 million. The Group Q specimens with triangular
shapes broke at fewer cycles for a shorter pit length. At a pit length of
3.5 mm, the wires broke at cycles between 4,800 and 36,000. At a pit
length of 6.0 mm, the wires broke at cycles between 49,000 and
72,000. At a pit length of 10.0 mm, the wires did not break. The
fatigue strength of Group Q specimens proportionally increases with
pit length, as shown in Fig. 12. Conversely, the critical cycles do not
depend on pit length for the Group R specimens (triangular shape
with a notch). All of the wires broke at cycles between 18,000 and
36,000, as shown in Fig. 12.
Downloaded from ascelibrary.org by CASA Institution Identity on 09/01/21. Copyright ASCE. For personal use only; all rights reserved.

Fig. 11. Fatigue test specimens with artificial corrosion pits

It is understood from these results of the Group P, Q, and R
specimens that the pit with a notched triangular shape has the lowest
fatigue strength. This is because stresses concentrate on the bottom
notch at the sharp edge of a triangle and initiate a fatigue crack from
there. The triangular-shaped pit also decreases the fatigue strength,
but this depends on the ratio of pit length to width. This ratio relates
to the stress concentration factor. The round-shaped pit is safer be-
cause stress concentration is smaller than the other two pit shapes.
The S-N data of the specimens with artificial pits are plotted and
compared with those of the Level 3 corroded wires in Fig. 13. It is
understood from this figure that the S-N data of the triangle pit
specimens (Group Q) and the notched triangle pit specimens (Group
R) are on the extended S-N line of the Level 3 corrosion wires.
Because there is a tendency that the fatigue strength of the artificial
notched wires is similar to that of actually corroded wires, it can be
judged that the pit shape is a dominant factor in lowering the fatigue
strength of corroded wires. However, the fatigue tests were con-
ducted with a stress range of 400 MPa, and further data with stress
ranges such as 200 and 300 MPa are required to confirm this. Also,
the information obtained from the fatigue tests of the notched wire
specimens can be applied to actual bridge wires. As deeper and
sharper corrosion pits decrease fatigue strength further, it is im-
portant to find corrosion in earlier stages whn the corrosion pits are
shallow and spread wider.

JOURNAL OF BRIDGE ENGINEERING © ASCE / MARCH 2013 / 205

J. Bridge Eng., 2013, 18(3): 200-209

Downloaded from ascelibrary.org by CASA Institution Identity on 09/01/21. Copyright ASCE. For personal use only; all rights reserved.

Fig. 12. Cycles until breakage of artificial pit wires (stress fluctuation,
500–900 MPa; stress range, 400 MPa)

Fig. 14. SEM photograph of a fractured section of Specimen R-2

Table 2. Models to Estimate Stress Concentration Factors

Notch Notch Radius at
Model Notch depth, length, Notch pit bottom,
number shape h (mm) L (mm) width (mm) R (mm)
Q-1 Triangle 0.6 3.5 3.5 0.1
Q-2 Triangle 0.6 6.0 3.5 0.1
Q-3 Triangle 0.6 10.0 3.5 0.1
P-3 Round 0.6 10.0 3.5 21.33
Note: h, L, and R are shown in Fig. 16.

Fig. 13. Comparison of S-N data of artificial pit wires and corroded
wires
A scanning electron microscope (SEM) photograph of the fracture
section of Specimen R-2 is shown in Fig. 14. It is seen that a fatigue
crack initiates at the notch and progresses toward the wire center.
When the fatigue crack reached about one-third of the whole section,
the wire fractured. Other wire specimens showed a similar fracture
section.
Stress Concentration Factors Estimated by Strain
Gauge Measurement
Stress is likely to concentrate at the pit. Two approaches were
adopted to find the stress concentration factors: measurements with
strain gauges and FEM analysis. Four typical models were chosen
(Table 2): triangular-shaped notches with lengths of 3.5, 6.0, and
10 mm and a round-shaped notch with a length of 10 mm.
Fig. 15. Wire specimen attached with strain gauges
A test wire was 5.0 mm in diameter and 230 mm long. A strain
gauge was attached at the pit bottom, and four strain gauges were
attached in parallel parts at 30 and 40 mm from the pit center
(Fig. 15). A compact strain gauge with a gauge length of 0.3 mm was
used to measure local stress at the pit bottom (Fig. 16).
A wire was tensioned until the measured strain at the pit bottom
reached 0.025. Fig. 17 shows the relation between the stress, the
applied tension divided by the cross-sectional area, and the strain
measured at the pit bottom. In Models Q-1 and Q-2, the inclination of
the stress and strain relation has a sharper increase compared with
Models Q-3 and P-3. Considering that the new wires have a tensile
strength of 1,600 MPa and a yield stress of 1,100 MPa at 0.7% strain,
the mechanical property of Models P-3 and Q-3 is low and of Models
Q-1 and Q-2 is significantly low.

206 / JOURNAL OF BRIDGE ENGINEERING © ASCE / MARCH 2013

J. Bridge Eng., 2013, 18(3): 200-209

 The stress concentration factors are estimated by dividing the
measured strain at the pit bottom by the average of the four measured
strains in parallel parts. Table 3 shows the estimated stress con-
centration factors when the stress reached 1,200 MPa. The stress
concentration factor is highest in the Q-1 model, the sharpest tri-
angular pit shape, with a factor of 3.48. This decreases to 2.88 in
Model Q-2 and further decreases to about 1.7 in Models Q-3 and P-3.
The triangle model with a notch is not studied in this paper because
the strain gauge cannot be applied to the notch. The main objective
on the stress concentration factor is the comparison between the
strain gauge measurement and FEM analysis.
Stress Concentration Factors Estimated by
FEM Analysis

Four models (Table 2) are analyzed by the FEM method to estimate

stress concentration factors. Fig. 18 shows the two-dimensional
FEM model with a length of 50 mm and a thickness of 5 mm. This
model is used for the specimens with four different pit shapes. One
end of the model is fixed longitudinally and pulled at the other end
Downloaded from ascelibrary.org by CASA Institution Identity on 09/01/21. Copyright ASCE. For personal use only; all rights reserved.

to produce 1.0 MPa. A typical mesh is shown in Fig. 19, with

a minimum mesh size of 0.003 mm. The analysis is carried out using
a general purpose FEM program (MSC Marc). The four-node plane-
strain element is applied.
Fig. 20 shows principal stresses along the longitudinal direction
of a wire. Model Q-1 has the sharpest peak at the pit, and Models Q-2
and Q-3 also have a sharp peak in this order. Conversely, a sharp
peak is not seen in Model P-3. The stress concentration factors are
obtained by dividing the maximum principal stress by the nominal
stress at the flat parts (Table 4). This is highest in Model Q-1 with
a factor of 4.80, and then 3.65 for Model Q-2, 2.81 for Model Q-3,
and 1.50 for Model P-3.
Both methods, strain gauge measurement and FEM analysis,
show that the stress concentration factor is higher for sharper
Fig. 16. Strain gauge attached at pit bottom (Model Q-2)
triangular-shaped pits with a shorter pit length, and it is smaller for
a round-shaped pit. Comparing Tables 3 and 4, although the ten-
dency is confirmed, the values of the FEM analysis are larger than
those of the strain measurement. This is because strain gauges cannot
measure the strain at the exact point of the pit bottom, whereas FEM
can calculate local stresses. In addition, the stress measured by the
strain gauge is one dimensional and the calculated one by FEM is
principal stress. However, the precision of FEM depends on the
meshing, the element types, etc.

Fig. 18. FEM model

Fig. 17. Relationship between measured strain and the stress of four
models

Table 3. Estimate of Stress Concentration Factor by Strain Gauge

Measurement
Strain gauge measurement
Model Strain at pit Average strain in Stress concentration
number bottom, Sp parallel parts, Sa factor, S 5 Sp=Sa
Q-1 2.09 0.60 3.48
Q-2 1.76 0.61 2.88
Q-3 1.06 0.62 1.71
P-3 1.00 0.58 1.72
Note: These strains correspond to a stress of 1,200 MPa. Sp 5 average of Fig. 19. Mesh for Model Q-2
strains in the four parallel parts.

JOURNAL OF BRIDGE ENGINEERING © ASCE / MARCH 2013 / 207

J. Bridge Eng., 2013, 18(3): 200-209

 wires with artificial pits was compared with that of actual corroded
wires. However, the fatigue tests were conducted with a stress range
of 400 MPa, and further data with stress ranges such as 200 and 300
MPa should be required to confirm this tendency.
The wire specimens were cyclically loaded with a stress range of
500–900 MPa. All of the Group P specimens with round-shaped pits
did not break until 1 million cycles. The Group Q specimens with
triangular shapes broke at fewer cycles for a shorter pit length. The
critical cycles did not depend on pit length for the Group R
specimens with a triangular shape with a notch, and the pit with
a notched triangular shape had the lowest fatigue strength. This is
because stresses concentrate on a notch at the sharp edge of a triangle
and initiate a fatigue crack from there. The triangular-shaped pit also
Downloaded from ascelibrary.org by CASA Institution Identity on 09/01/21. Copyright ASCE. For personal use only; all rights reserved.

decreased fatigue strength, but it depended on the ratio of pit length

to width. This ratio also related to the stress concentration factor. The
round-shaped pit is safer because the stress concentration is smaller
than those of the two others. The S-N data for the triangular pit
specimens (Group Q) and the notched triangular pit specimens
(Group R) were on the extended S-N line of the Level 3 corrosion
Fig. 20. Stress distributions of four models wires.
In the fourth step, the stress concentration factor was evaluated
by two methods: measurement by strain gauges and FEM analysis.
Table 4. Estimate of the Stress Concentration Factor by FEM Analysis Four wire models with different artificial pit shapes were ten-
FEM analysis sioned, and strains were measured at the pit bottom and at the
parallel parts, from which the stress concentration factors were
Max stress at pit Nominal stress Stress concentration obtained. The same models were analyzed by FEM, and the
Model number bottom, Fp at flat parts, Fn factor, F 5 Fp=Fn maximum stresses were calculated, by which the stress concen-
Q-1 5.147 1.136 4.80 tration factor was obtained. Both methods showed that the stress
Q-2 3.919 1.136 3.65 concentration factor was higher for sharper triangular-shaped pits
Q-3 3.016 1.136 2.81 with a shorter pit length, and it was smaller for the round-shaped
P-3 1.610 1.136 1.50 pits.
Note: Fn is the stress at a flat part considering the loss of cross section This study targets the bridge wires used as suspension bridge
because of a pit. main cables and hangers and stay cables of cable-stayed bridges.
These bridge wires consist of galvanized steel wires with a di-
ameter of approximately 5–7 mm and a tensile strength greater
Conclusion than 1,600 MPa. The results obtained in this study can be applied
to parallel wire strands and semiparallel wire strands. The parallel
In this paper, the fatigue strength of corroded galvanized steel wires wire strand consists of these galvanized steel wires arranged in
was studied in four steps. parallel. Today, the semiparallel wire strand, which is basically
In the first step, corroded wires on three corrosion levels (Levels a parallel strand but with wires that are slightly helically wound,
1, 2, and 3) were produced at the laboratory, and fatigue tests has been developed and used as hangers and stays. The usual
were conducted under dry and wet environments to find the effects spiral strand consists of helically wound galvanized wires and
of corrosion fatigue of corroded wires. It was found that fatigue may need more study on fretting corrosion at the contact points of
strength of new wires was the highest and more severe corrosion each wire.
decreased fatigue strength further. There is also a tendency for fa- It was found in this study that deeper and sharper corrosion pits
tigue strength under wet environments to be less than that under dry decrease fatigue strength, and therefore, it is very important to find
environments. corrosion in earlier stages when the corrosion pits are shallow.
In the second step, surface roughness or pits were measured on
the corroded galvanized wires to find the relation between corro-
sion levels and pit sizes. An average maximum depth of Level 2 References
wires was 0.18 mm and that of Level 3 was 0.34 mm, which means
more severe corrosion produces deeper pits. The maximum depth Betti, R., West, A. C., Vermaas, G., and Cao, Y. (2005). “Corrosion and
was 0.56 mm among all of the data. All of the pits with a depth of embrittlement in high-strength wires of suspension bridge cables.”
0.4–0.6 mm were within an area with a width of 10 mm and a length J. Bridge Eng., 10(2), 151–162.
of 10 mm. Conversely, some of the pits with a depth of less than Betti, R., and Yanev, B. (1999). “Conditions of suspension bridge cables:
0.4 mm were outside of this area. Relatively shallow corrosion The New York City case study.” Proc., Workshop on Safety Appraisal of
pits spread widely, and deeper corrosion pits concentrated on nar- Suspension Bridge Main Cables, National Cooperative Highway Re-
rower area. search Program, Transportation Research Board, Newark, NJ, 105–112.
Furuya, K., Kitagawa, M., Nakamura, S., and Suzumura, K. (2000).
In the third step, cyclic tests were conducted for the wire speci-
“Corrosion mechanism and new protection methods of suspension
mens with artificial pits whose sizes were decided based on the bridge cables.” J. Int. Assoc. Bridge Struct. Eng., 10(3), 189–193.
measured data. To clarify the effect of pits, specimens with three MARC [Computer software]. Santa Ana, CA, MSC.
different shapes, round (Group P), triangular (Group Q), and tri- Mayrbaurl, R. (2000). “Corrosion in suspension bridge cables.” Proc.,
angular with a notch (Group R), and with three different lengths of Congress of the International Association of Bridge and Structural
3.5, 6.0, and 10 mm were used. Then, the fatigue strength of the Engineers, Zurich, Switzerland.

208 / JOURNAL OF BRIDGE ENGINEERING © ASCE / MARCH 2013

J. Bridge Eng., 2013, 18(3): 200-209

 Mayrbaurl, R., and Camo, S. (2001). “Cracking and fracture of suspension
bridge wires.” J. Bridge Eng., 6(6), 645–650.
Miki, C., Nishimura, T., Tanabe, H., and Nishikawa, K. (1981). “Study on es-
timation of fatigue strength of notched steel members.” Proc., JSCE, Tokyo,
153–166.
Nakamura, S., and Suzumura, K. (2009). “Hydrogen embrittlement and
corrosion fatigue of corroded bridge wires.” J. Construct. Steel Res.,
65(2), 269–277.
Nakamura, S., Suzumura, K., and Tarui, T. (2004). “Mechanical properties
and remaining strength of corroded bridge wires.”J. Int. Assoc. Bridge
Struct. Eng., 14(3), 50–54.
Downloaded from ascelibrary.org by CASA Institution Identity on 09/01/21. Copyright ASCE. For personal use only; all rights reserved.
J. Bridge Eng., 2013, 18(3): 200-209
Pilkey, W., and Pilkey, D. (2008). Peterson’s stress concentration factors,
Wiley, New York.
Stahl, F. L., and Gagnon, C. P. (1996). Cable corrosion, ASCE Press,
New York.
Suzumura, K., and Nakamura, S. (2004). “Environmental factors affecting
corrosion of galvanized steel wires.” J. Mater. Civ. Eng., 16(1), 1–7.
Tarui, T., Maruyama, N., Eguchi, T., and Konno, S. (2001). “Development
of high strength galvanized steel wires for bridge cables.” Proc., IABSE
Conf. on Cable Supported Bridges, Zurich, Switzerland.
Uhlig, H. H., and Revie, R. W. (1963). Corrosion and corrosion control,
Wiley, New York.
JOURNAL OF BRIDGE ENGINEERING © ASCE / MARCH 2013 / 209