Journal of Constructional Steel Research 119 (2016) 113–122

Contents lists available at ScienceDirect

Journal of Constructional Steel Research

Experiment on fatigue behavior of rib-to-deck weld root in orthotropic

steel decks
Shigenobu Kainuma a, Muye Yang a, Young-Soo Jeong a,⁎, Susumu Inokuchi b,
Atsunori Kawabata b, Daisuke Uchida b
a
Department of Civil Engineering, Kyushu University, Fukuoka 8190395, Japan
b
Technical Research Laboratory, Japan Bridge Association, Tokyo 1050003, Japan

a r t i c l e i n f o a b s t r a c t

Article history: This study experimentally investigated the fatigue behavior of the weld root in orthotropic steel decks stiffened
Received 21 August 2015 with U-ribs in relation to the loading conditions and welding details. To examine the structural response and local
Received in revised form 24 November 2015 stress near the welded joint, field loading tests, measurement of residual stress, and fatigue tests were carried out.
Accepted 26 November 2015
A total of 12 specimens were manufactured, and the fatigue tests were performed by simulating the double tire
Available online 17 December 2015
loading of an actual vehicle. The welding residual stress distribution at the root was examined to understand the
Keywords:
mechanism of root crack initiation and propagation behavior. Thus, fatigue behavior of the root crack was inves-
Orthotropic bridge deck tigated and evaluated by considering different stress ratios and weld penetration rates. Based on the fatigue test
Root crack results and crack patterns, it was revealed that both tensile stress and stress range could affect the root crack ini-
Fatigue test tiation. However, the tensile stress rather than the magnitude of stress range would be the effective stress after a
Residual stress crack initiated, and it was seen to be an important factor for root crack propagation. In addition, a penetration rate
Weld penetration in the range of 0% to 75% was beneficial for fatigue durability of this structural detail.
© 2016 Elsevier Ltd. All rights reserved.

1. Introduction
Orthotropic steel bridge decks were widely applied to long-span
bridges because of their characteristics such as light weight, high
strength and durability, and rapid construction [1,2]. In Japan,
many orthotropic steel bridges were extensively constructed in the
1960s–80s. Recently, various types of damage damages to the steel
decks were reported owing to long-term operation and the changed op-
erating environment including the emergence of heavy-duty vehicles
and increasing traffic volume.
Fatigue cracks in orthotropic steel bridges significantly occurred at
partial-penetration fillet-welded connections owing to the cyclic load
stress caused by a high number of vehicles. The following typical fatigue
cracks have been reported in recent years [3], as shown in Fig. 1. A rib-
to-deck crack is usually initiated at the weld root, and it propagates
into the deck plate or through the weld bead. It has been reported
that root cracks in U-rib to deck connections account for about 18.9%
of the total damage types in orthotropic steel decks [4]. The weld toe
crack of the deck plate, which is located at the rib-to-cross beam con-
nection with a partially welded joint, and the crack occurring at the fillet
weld toe near field-welded splice plates fixed to the webs were also
common damage types in orthotropic steel bridges. In addition, there
⁎ Corresponding author.
were many other types of cracks. However, crack initiation at the weld
root of the longitudinal fillet welding between deck plates and closed-
rib cannot be observed by visual inspection, which might have serious
implications to bridge safety. Moreover, at present, U-ribs account for
about 60% of rib stiffeners in the orthotropic steel deck construction.
The rib-to-deck structure includes a large number of single-fillet weld
joints because of the wide use of U-ribs [5]. Therefore, fatigue cracks
of the weld root between the deck plate and U-rib are one of the most
common and serious cracks, because of the inadequate welding details
in conjunction with the residual stress induced by the welding process
and unfavorable weld shape.
The root cracks have been investigated by several experimental and
analytical studies. Analytical studies usually focused on the local stress
around the rib-to-deck weld joint, and the fatigue durability of the
structure was evaluated by finite-element simulation [6]. Past experi-
mental studies were conducted through full-scale fatigue tests under
fixed-point loading [7–9] and rib-to-deck small specimen tests [10,11].
Full-scale fatigue tests of orthotropic steel decks are usually conducted
to evaluate the fatigue performance. Bending fatigue tests of small spec-
imens can be conducted to evaluate the fatigue life of crack initiation
and propagation within a certain length of the specimens, and it is
possible to assess the fatigue crack growth and the fatigue strength.
However, the boundary condition is different between the actual
orthotropic steel decks and small specimens. It is difficult to compare
the fatigue behavior of small specimens with that of actual orthotropic
steel decks. This is because the tests are performed under constant

http://dx.doi.org/10.1016/j.jcsr.2015.11.014
0143-974X/© 2016 Elsevier Ltd. All rights reserved.
114 S. Kainuma et al. / Journal of Constructional Steel Research 119 (2016) 113–122
Fig. 1. Typical fatigue cracks in orthotropic steel bridges.
strain range and without restraint on the rib. Moreover, the secondary
stress caused by deformation and the effect of release of the welding
residual stress after crack initiation could not be properly simulated in
these tests.
For penetration rate, most researchers considered that a higher pen-
etration rate could lead to a higher strength in the weld joint. However,
the welding process of full penetration weld is more complex. Its reheat
cycle would reduce the performance of the welded joint and easily
cause weld defects. Recently, full-scale fatigue tests of orthotropic
steel decks conducted by some researchers have proved that a
shallower weld penetration at the partial joint penetration (PJP) joint
appeared to have a positive effect on enhancing the fatigue resistance.
Rib-to-deck fatigue tests of small specimens also showed that the spec-
imens with weld melt-through seemed to have slightly lower fatigue
strengths than those with 80% PJP [11,12]. Although most researchers
agree that a penetration rate below 75 ~ 80% is beneficial to fatigue du-
rability, it is still inconclusive whether full penetration could enhance
the fatigue strength of a rib-to-deck weld joint.
In this study, the fatigue behavior at different penetration rates at the
PJP weld joint was experimentally evaluated. The structural responses of
the reference points near the weld joints were examined using field mea-
surements and fatigue tests. The residual stress at the weld root of the
deck plate to U-rib connection was measured using a cutting method.
The test results also explained the cause identification of the crack initia-
tion of this structural detail. In addition, based on the measurement of the
welding residual stress and fatigue test results, the fatigue behavior and
cracking mechanism of the weld root between the deck plate and U-rib
were investigated by considering the effective stress.
2. Loading test of orthotropic deck bridge
A1 Bridge is located in the heavy traffic route of the Tokyo Metro bay
area in Japan. A1 Bridge consists of two simple box girder bridges and a
three-span continuous box girder bridge with orthotropic steel decks.
Large vehicles account for 45.2% of the daily traffic volume. The dimen-
sions of the orthotropic steel deck are given in Table 1. Asphalt pave-
ment damage in A1 Bridge was visually inspected in 2004, and severe
root cracks were observed during the re-paving construction process
in 2005. The root cracks occurred around the intersections of the U-rib
at the weld joints between diaphragms, and the maximum length of
cracks was over 400 mm [13].
Table 1
Dimensions of the orthotropic steel decks of A1 Bridge (unit: mm).
Thickness of asphalt pavement 75
Thickness of deck plate 12
Dimension of U-rib 340 × 284 × 8
Spacing of the transverse rib 3000
Spacing of U-rib 680
Spacing of web of girder 2800
Distance between diaphragms 5600
Based on A1 Bridge, this study focused on the rib-to-deck weld joint
where the root cracks had been observed. A field loading test was con-
ducted in August (2005) when the influence of the asphalt pavement
was most negligible at high temperatures. The mid-span of the U-rib
was set as the target section, and the measured transverse stresses
near the target used as “reference stress”, which were 5 mm away
from the weld toe at the bottom of the deck plate. The measured refer-
ence stress range was about − 40 MPa to 6 MPa under the rear tires,
as shown in Fig. 2. Although the tensile stress was much lower than
the compressive stress, it has the potential to open the weld root gap,
and lead to the initiation and propagation of the root cracks [9]. In
this study, a stress ratio σt,max/σc,max equal to 0.13 was obtained
from field measurement data. The stress range of A1 Bridge was deter-
mined by a stress histogram measurement, which was carried out for
3 days by using the rain-flow method, and the maximum stress range
was 170 MPa [14,15]. These results were used in the following fatigue
tests.
3. Fatigue test
3.1. Test specimen
A total of 12 fatigue tests were conducted on full-scale orthotropic
steel deck specimens. Each specimen was 1400 mm wide and consisted
of two U-ribs of 2000 mm span between two cross beams. Steel grade of
JIS G3106 SM400A [16] was used for the tested specimens, whose
chemical composition and mechanical properties are listed in
Table 2(a). Each specimen consisted of a 12-mm-thick deck plate and
6-mm-thick U-rib, and was welded by a semi-automatic CO2 welding
method. Three types of specimens were fabricated with 0%, 75%, and
100% penetration rates at the U-rib to deck plate connection. The di-
mensions and welding conditions of these specimens are listed in
Table 3, where D and U are the thickness of the deck plate and U-rib, re-
spectively, and SP is the penetration rate. Thus, D12U6SP0, for example,
represents a 12-mm-thick deck plate and 6-mm-thick U-rib with a weld
joint of 0% penetration rate. Photographs of the weld joints of the spec-
imens in the etching test are shown in Fig. 3; in the figure, the heat af-
fected zone caused by the welding process can be observed. The
welding area varies between the specimens with different penetration
rates.
The dimensions of the specimens and the loading positions are shown
in Fig. 4. The dynamic loading area at the mid-span was based on the ac-
tual double tires. As the weld root tip stress could not be measured during
the test, the transverse stress of points that were 5 mm away from the
weld toe in this weld line was considered as the reference stress. The
stress at the weld root tips and the reference points tend to have a positive
correlation. The tested primary reference stress at the mid-span was con-
sidered as the stress condition for each specimen.
3.2. Test set-up
The experimental set-up consisted of two hydraulic jacks that con-
trolled static loading Ps and dynamic loading Pd. The loading positions
are shown in Fig. 5. Ps could be changed to vary the stress ratio so that
both the tensile stress and compressive stress could be simulated. Pd
 S. Kainuma et al. / Journal of Constructional Steel Research 119 (2016) 113–122
Fig. 2. Field measurements in loading test of orthotropic deck bridge.
provided a cyclic load, and its hydraulic jack applied force to the loading
jig. The loading jig consisted of a loading brick, steel plate, and rubber
plate, as shown in Fig. 6. Two steel plates of 25 mm thickness were
used as the double tires, whose size depended on the pressure area of
the actual double tires [14]. The fatigue tests were conducted using rub-
ber plates instead of asphalt pavement. The properties of the rubber
plates are summarized in Table 2(b). This study used rubber plates of
80 mm thickness so that load dispersion could be considered effectively
[15].
Based on field measurement results of A1 Bridge, the test primary
stress range of the mid-span reference point of basic specimens was
controlled at 180 MPa. The basic fatigue test set the maximum compres-
sive stress at −160 MPa and the maximum tensile stress at 20 MPa be-
cause the measured stress ratio of A1 Bridge was 0.13. The stress ratio R
is defined as the algebraic ratio of absolute value of minimum stress to
maximum stress. The other stress range conditions and different stress
ratios were used for comparison tests. The maximum value of tensile
stress was considered as an important factor for crack initiation, and
the effects of stress ratio for a given stress range were explained,
e.g., the stress range of −140 MPa to 40 MPa.
Test coordinates were set at the mid-span of longitudinal U-rib as
x = 0, and the static load side was regarded as the positive x direction.
Uniaxial strain gages were attached at the reference points at a distance
of 150 mm to 600 mm away from the mid-span. The test primary stress
distribution in the longitudinal (x) direction under dynamic loading is
shown in Fig. 7.
Three million cycles were applied to all specimens in the first loading
stage. If there was no crack initiation, another 1.5 million cycles were
applied to continue the testing. An additional single tire loading was ap-
plied to two D12U6SP100 specimens, which was controlled by one hy-
draulic jack that provided Pd at the mid-span. Both the dynamic loading
Table 2
Chemical composition and mechanical properties.
(a) Specimen of fatigue test and measurement of residual stress
Yield point Tensile strength Elongation C Si Mn
(MPa) (MPa) (%) (%) (%) (%)
302 423 30 0.16 0.05 0.78
(b) Rubber plate
Yield point Tensile strength
(MPa) (MPa)
1.2 3.2
115
and tire area of the single test were set at half the corresponding values
of the double tire test. A single loading plate was placed between two U-
ribs at the mid-span.
3.3. Test results
The conditions of test stress, crack length, and crack shapes are given
in Table 4. The surface cracks were detected by the magnetic particle
testing (MPT) method after all the tests were completed. The cracks
mainly existed at a distance of ± 200 mm from the mid-span. It was
thought that the cracks were mostly initiated at the mid-span and
extended in both depth and length directions. Therefore, the specimens
were cut into 25-mm-thick pieces in the longitudinal direction, so that
the crack shape and details in the cut sections could be acquired.
The photos of the crack patterns are shown in Fig. 8. The root crack
fracture surface of D12U6SP0-4 in the longitudinal direction is shown
in Fig. 8(a). The fracture surface consists of some discontinuous
fractures that could be multi-cracks initiated by the root crack and
eventually forming a large crack. The corresponding crack patterns
of D12U6SP75 and D12U6SP100 by MPT inspection are shown in
Fig. 8(b, c, d).
The test results could be divided into three crack patterns as follows:
(1) Only root crack existed; (2) Both toe crack and root crack existed,
and at N ar; (3) No crack initiated. The crack patterns are described in
Table 5.
An analysis of the stress condition of category (1) revealed that the
tensile stress and crack length of the SP0 specimen were larger than
those of the SP75 and SP100 specimens for the same stress range.
However, the stress range and crack depth of the SP75 specimens did
not show a correlation because of the differences in the geometry of
the root gap. The root crack length of SP100 and SP0 were almost half
of deck thickness under the stress condition of − 140 to 40 MPa.
SP100-2 and SP100-3 were under the same stress conditions of
first stage loading, and then, SP100-2 continued 3 million cycles of
single tire loading while SP100-3 continued 6 million cycles. Crack
depth comparison of two specimens showed their propagation lengths
in the vertical direction were similar. Due to the limit of test number,
P S the effect of single tire load on the root crack propagation seems not
(%) (%) obvious so far. In addition, the case of SP100 showed a different
0.014 0.004 mechanical mechanism because of its special geometry of the root
gap at the weld joint. Stress concentration caused by the geometry
mutation and the crack initiation point were always located at the
Elongation root tip of PJP, while stress distribution and stress concentration of the
(%)
full penetration weld joint were different because of the weld shape.
An alternative idea is that the SP100 specimen might have created a
250
geometrical discontinuity at the weld root that led to a lower fatigue
116 S. Kainuma et al. / Journal of Constructional Steel Research 119 (2016) 113–122
Table 3
Specimen dimensions and welding conditions.
Thickness (mm) Penetration
Specimen
Deck td U-rib tu Pw (mm)
D12U6SP0 12 6 0
D12U6SP75 4.5
D12U6SP100 6 100
resistance. Related research can be found in the fatigue tests conducted
by Sim et al. [8].
An analysis of the toe and root crack patterns showed that the crack
was easily initiated at the weld toe under a high stress ratio. Some past
full-scale orthotropic steel deck fatigue tests on 12- and 14-mm-thick
deck plates also showed both root and toe cracking. However, there
has been no report of toe cracking in actual orthotropic steel decks
with 12 mm deck plates in Japan. The past full-scale orthotropic steel
deck fatigue tests proved that toe cracking would also occur with a
high tensile stress [9,17].
An analysis of category (3) showed that compressive stress would
not cause fatigue cracking even under the basic stress range of
180 MPa. At the same time, a stress condition with half or 2/3 of the
basic stress range may lead to an improvement in fatigue life. This
means that both stress ratio and stress range could affect the fatigue
life before crack initiation.
4. Measurement of residual stress
The welding process creates high residual stress in the weld joint.
The magnitude of the peak residual stress may correspond to the yield
point. Thus, it would seem that the stress range, and not the magnitude
of the structural stress, determines the fatigue life. Even if the structural
stress range is entirely compressive, the effect of the welding residual
stress makes the structure susceptible to fatigue failure [18]. As shown
in Fig. 9, the residual stress can increase the range of effective stress in-
tensity factor above the threshold value of K [19].
The residual stress distribution near the weld root tip could be mea-
sured by the cutting method [20,21]. To evaluate the welding residual
stress in the deck vertical direction, the residual stress of D12U6SP0-1
was measured. The location of the cut out pieces of the specimen
is shown in Fig. 4. In this study, to understand the distribution of resid-
ual stress, residual stress tests using the cutting method were carried
out for three-scale large models of D36U18 with penetration rates of
0% and 75%. The residual stress was then compared with that of
D12U6SP75.
The stress distribution of the tested three-scale specimens was sim-
ilar to the results from the full-scale fatigue test [20], as shown in Fig. 10.
Although residual stress values were different for each specimen, it was
obvious that stress trends were similar in the vertical direction of the
Fig. 3. Weld joint photos of specimens in etching test.
Penetration rate Current Voltage Speed
Number of passes
Pw/tu (%) (A) (V) (cm/min)
0 220–280 28–30 30 2
75 280 32 45 2
200–280 25–30 35–45 2
deck, even for specimens with different penetration rates. In particular,
residual tensile stress always existed at about 10% of the thickness of the
deck plate from the root tip. Based on the residual stress test results
shown in Fig. 10, the values of residual stress near the root tip were con-
fined to 60 to 100 MPa, while the actual residual stress of the root tip
should be higher. Superposition of the tensile stress and residual stress
would make the structure susceptible to fatigue failure [18]. However,
the residual tensile stress would be released after crack initiation and
propagation.
The crack depth of each specimen was usually smaller than half of
deck thickness, which agrees with the results of a previous study
[7–9]. The full-scale tests with the 12-mm-thick deck plate showed
that the root-to-deck crack was likely to take a long time before coales-
cence and failure, even though it nearly reached the surface. The full-
scale tests of the 16-mm-thick deck plate showed that the fatigue cracks
seemed to be concentrated at around half of the deck plate thickness.
Past tests with a 12- or 14-mm-thick deck plate also reported that the
root-to-deck cracks mostly stopped propagating after they reached
the neutral axis of the deck plate thickness. As suggested before, all
these fatigue tests showed that the residual tensile stress would be re-
leased after the root crack was initiated. The root crack appeared to
close after the residual stress was released because of crack growth,
which finally stopped the crack propagation.
Therefore, the tensile stress instead of the stress range tends to affect
crack propagation. However, the tensile stress was small for continued
crack propagation. A past fatigue test on fillet-welded cruciform joint
under compressive cyclic stress also concluded that the fatigue root
cracks propagated to a length of 4 mm (1/3 of plate thickness) and
then ceased to grow [22]. It was suggested that either the tensile stress
around the crack tip was not large enough to reach the stress intensity
factor or the stress range provided only by the tensile stress was smaller
than the constant amplitude of the fatigue limit stress range, which
stopped the fatigue crack propagation [24].
5. Fatigue behavior of weld root
5.1. Crack initiation and propagation process
In order to investigate the fatigue behavior of root cracks, cracking
position, depth, and propagation angles were analyzed. Figs. 11 and 12
 S. Kainuma et al. / Journal of Constructional Steel Research 119 (2016) 113–122 117

Fig. 4. Fatigue loading area and dimensions of specimen.

show the measured reference stress and relative reference stress range
of the SP75 specimens. These figures show that the maximum stress
range and maximum compressive stress of all specimens were located
in the mid-span, and the maximum tensile stress point should be be-
tween the dynamic and static loading positions. According to the resid-
ual stress measurement and analysis, the theoretical crack initiation
point should obey the following formulas:
σ res ≤σ max;c Maximum tensile point ð1Þ
σ max;c ≤σ res ≤σ y −σ max;t Maximum tensile point ð2Þ
σ y −σ max;t;m ≤σ res Mid‐span ðMaximum stress range pointÞ ð3Þ
where σres is the residual stress of the weld root tip, σy is the yield
strength of base metal, σmax, t and σmax, c are actual maximum tensile/
compressive stress at the root tip, respectively, and σmax,t,m is the
maximum tensile stress at the mid-span.
Crack images illustrated in Table 4 show that most of the root cracks
were initiated at the mid-span, which was consistent with the case of
(3). The crack depth distribution of the D12U6SP0 specimens was dif-
ferent from others. The D12U6SP0-1 specimen had a crack depth in
the range of 0 to 175 mm in the longitudinal direction. However, the
D12U6SP0-4 specimen had a crack depth in the range of − 200 to
200 mm, because the origin of fatigue failure changed with the magni-
tude of the stress range owing to the welding residual stresses in the
weld root [23].
The residual stress of the D12U6SP0-1 specimen seemed to be
smaller than those of the SP75 and SP100 specimens. Thus, the

Fig. 5. Test set-up.

Fig. 6. Loading jig. Fig. 7. Reference stress range of D12U6SP0.

118 S. Kainuma et al. / Journal of Constructional Steel Research 119 (2016) 113–122

Table 4
Test stress and results.

Stress (MPa)
Specimen ID Cycle (×104) Maximum crack depth, a (mm) Image of crack depth (mm) and shape
σmax σmin Δσ

D12U6SP0 1 20 −160 180 300 ar = 3.2

2 20 −100 120 450 No crack

3 20 −70 90 300 No crack

4 40 −140 180 300 ar = 5.4

D12U6SP75 1 20 −160 180 300 ar = 1.5

2 20 −100 120 300 ar = 2.7

3 40 −140 180 300 ar = 3.2

at = 4.5

4 0 −180 180 450 No crack

5 40 −140 180 300 ar = 4.6

at = 5.7

D12U6SP100 1 40 −140 180 300 ar = 2.5

at = 3.5

2 40 −140 180 300 ar = 6

+300Single

3 40 −140 180 300 ar = 6.1

+600Single
 S. Kainuma et al. / Journal of Constructional Steel Research 119 (2016) 113–122 119

Fig. 8. Typical crack patterns of test results.

crack was initiated at the maximum tensile point and not at the
mid-span. In contrast, the long cracks of D12U6SP0-1 were ob-
served in the range x = 50 to 150 mm. To explain the effect of stress
distribution according to the loading condition, the crack depth was
measured in the y and z directions in each cut section as shown in
Fig. 13. The red and blue lines indicate that the trends of root
crack propagation were opposite, and the crack depth in the z direc-
tion varied notably even for a given y value. The width of root cracks
in the transverse direction for different propagation angles is illus-
trated in Fig. 14. It is clear from Fig. 7 that crack angles at x = 125
were very different from those at other sections because the tensile
stress was larger than those for the other sections. The fatigue
cracks were usually observed as small semi-elliptical cracks at mul-
tiple points along the weld line, and then the small fatigue cracks
propagated and coalesced into a larger crack. Thus, the x = 125 sec-
tion was probably near the location of the confluence of two small
cracks.
In the case of SP0-4, the root crack had a relatively small depth at
the location where the deck crack propagated rapidly. The crack
occurred on the upper surface of the deck plate because the steel
reached the yield strength during the fatigue test. This crack was per-
pendicular to the deck plate, and a semi-elliptical crack was observed
as shown in Fig. 15. It was thought that this type of deck crack tends
to propagates through the root crack. From the perspective of bridge
structural safety, this case would be worse than when only root cracking
occurs.
5.2. Crack behavior and parametric analysis
In order to investigate the effect of penetration rate, crack behaviors
of specimens SP0-5, SP75-3, SP75-5, and SP100-3 were compared.
These four specimens were under the same stress range (− 140 to
40 MPa) and same fatigue life of 3 million cycles. Typical macro sections
of the root crack of 80% PJP and weld melt-through specimens in

Table 5
Relationship between stress situation and crack length.

Crack types Category No. Tensile stress Stress range Crack length
(MPa) (MPa)
ar (mm) at (mm)

Root crack only SP0-1 20 180 3.2 –

SP0-4 40 180 5.4
SP75-1 20 180 1.5
SP75-2 20 120 2.7
SP100-2 40 180 6
(+300Single)
SP100-3 40 180 6.1
(+600Single)

Toe and root SP75-3 40 180 3.2 4.5

cracks SP75-5 40 180 4.6 5.7
SP100-1 40 180 2.5 3.5

No crack SP0-2 20 120 – –

SP0-3 20 90
SP75-4 0 180
120 S. Kainuma et al. / Journal of Constructional Steel Research 119 (2016) 113–122
Fig. 9. Effective stress at the weld joint.
reference [24] showed that the fatigue crack paths are nearly perpendic-
ular to the deck plate. However, in this study, the root cracks were not
entirely perpendicular to the deck plate, which might have been due
to the different deck plate thickness and boundary condition. The root
cracks of all these tested specimens initiated at the root tip. The propa-
gation direction tends to approach the vertical direction of the deck be-
cause of the maximum principal stress.
Fig. 16 shows the origin coordinates of the root tip and the location
of all crack tips. In Fig. 16(a), the SP0 specimens have the smallest
crack angles, and the average cracking angles of the SP75 and SP100
specimens are nearly the same. The crack depth of SP75 ranged from
1.6 mm to 4.6 mm, while SP100 had a crack depth of around
4 ~ 6 mm in the deck vertical direction, as shown in Fig. 16(b). SP100
clearly had the largest crack length in both y and z directions. In order
to eliminate the interference effect of small cracks, the three longest
cracks of every specimen were compared, which facilitated a compari-
son of the maximum crack depth within a segment of 50 mm in the lon-
gitudinal direction. As shown in Fig. 17, the SP75 specimens had the
smallest crack depths and largest propagation angles. The SP0
Fig. 10. Residual stress distribution comparison.
Fig. 11. Measured reference stress of D12U6SP75.
specimens had more oblique cracks, which might have been caused
by the geometric shape of the root gap.
Similar results were obtained by related experiments in past studies.
Small-scale specimens of a rib-to-deck welded joint with no restraint on
the ribs were tested. It reported that the fatigue resistance of a 100%
penetration weld was slightly lower than that of an 80% PJP weld,
which is consistent with the findings that a single-sided welded speci-
men tends to have a higher fatigue resistance as the weld penetration
rate is increased from 0% to 75%. An analytical study also suggested
that an increase in the penetration rate (up to 75%) would produce a
lower stress at the weld root, which results in a higher fatigue resistance
[24,25]. We considered that a penetration rate in the range of 0% to 75%
is desirable for fatigue durability of this structural detail. However,
Fig. 12. Measured reference stress range of D12U6SP75.
 S. Kainuma et al. / Journal of Constructional Steel Research 119 (2016) 113–122 121

Fig. 15. Depth distribution of deck/root cracks of D12U6SP0-4.

Fig. 13. Crack depths comparison of D12U6SP0-1 cross sections.

increasing the penetration rates above 75% might have an opposite

effect.

6. Conclusion

This study experimentally investigated the fatigue behavior of the

weld root in an orthotropic steel deck stiffened with U-ribs in relation
to the loading conditions and welding details. Field measurements, re-
sidual stress tests, and fatigue tests were carried out on a total of 12
full-scale specimens, which had 0%, 75%, and 100% penetration rates.
The fatigue behaviors were compared with those of other recent

Fig. 14. Comparison of crack propagation angles of D12U6SP0-1 cross sections. Fig. 16. Average values of all crack tips.
122 S. Kainuma et al. / Journal of Constructional Steel Research 119 (2016) 113–122
Fig. 17. Comparison of angles and depths of crack tips.
studies. Based on the experimental data and parametric analysis, the
main conclusions of the study are as follows:
- Root cracks tend to initiate easily in the basic stress range − 160 to
20 MPa under double tire loading. The tested specimens could be
divided into three types: 1) Specimens having only root crack;
2) Specimens having both toe crack and root crack and usually
at N ar under the condition of high stress ratio; 3) Specimens having
no cracks.
- The tensile residual stress existed around the weld root tip. The basic
trends of residual stress of specimens with different penetration
rates were similar in the vertical direction of the deck, and the 0%
penetration specimens had the smallest residual stress according
to a theoretical analysis of cracking initiation point.
- Based on the fatigue test results, both the tensile stress and stress
range could affect the root crack initiation. Based on the residual
stress test results, the tensile stress would be the effective stress
for root crack propagation because the residual stress of the crack
tip would be released after crack initiation. The root crack appeared
to close after the residual stress was released, which finally stopped
the crack propagation.
- Specimens with a 0% penetration rate had more oblique root cracks
based on a crack morphology comparison. In general, a penetration
rate in the range of 0% to 75% was beneficial for fatigue durability
of this structural detail.
References
[1] Z.H. Qian, D. Abruzzese, Fatigue failure of welded connections at orthotropic bridges,
Giornata IGF Forni di Sopra (UD), 9 2009, p. 105.
[2] S. Sakai, Innovative Technologies for the Steel Bridges of Expressway in Japan, 2010.
[3] S. Abdou, W. Zhang, J.W. Fisher, Orthotropic deck fatigue investigation at Triborough
bridge, New York. Transportation research board 2003 meeting.
[4] Japan Society of Civil Engineers, Committee on steel structures. Fatigue of
orthotropic steel bridge deck, Steel Struct. (19) (2010) (In Japanese).
[5] F.B.P. DeJong, Renovation Techniques for Fatigue Cracked Orthotropic Steel Bridge
Decks, Delft University of Technology, 2007 (Doctoral dissertation).
[6] X. Ju, S.M. Choi, K. Tateishi, Analytical study on crack propagation at rib-to-deck
welded joints in orthotropic steel decks, Steel Constr. Eng. Jpn. Soc. Steel Constr.
19 (73) (2012) 85–94.
[7] C.V. Dung, E. Sasaki, K. Tajima, T. Suzuki, Investigations on the effect of weld pene-
tration on fatigue strength of rib-to-deck welded joints in orthotropic steel decks,
Int. J. Steel Struct. 12 (2014) 1–12.
[8] H.B. Sim, C.M. Uang, C. Sikorsky, Effects of fabrication procedures on fatigue resis-
tance of welded joints in steel orthotropic decks, J. Bridg. Eng. 14 (5) (2009)
366–373.
[9] S. Inokuchi, Wheel trucking test for weld of U-shaped rib and deck plate in the
orthotropic deck, Proceedings of the 23th U.S.-Japan Bridge Engineering Workshop,
11 2007, pp. 325–336.
[10] S. Ya, K. Yamada, T. Ishikawa, Fatigue durability of trough rib to deck plate welded
detail of some orthotropic steel decks, J. Struct. Eng. Jpn. Soc. Civ. Eng. 56 (A)
(2010) 77–90.
[11] S. Ya, K. Yamada, Fatigue durability evaluation of trough to deck plate welded joint
of orthotropic steel deck, J. Jpn .Soc. Civ. Eng A 64 (3) (2008) 603–616.
[12] H.B. Sim, C.M. Uang, Stress analyses and parametric study on full-scale fatigue tests
of rib-to-deck welded joints in steel orthotropic decks, J. Bridg. Eng. 17 (2012) 765-
733.
[13] S. Inokuchi, D. Uchida, A. Kawabata, T. Tamakoshi, Influence of asphalt pavement
failure on local stress of orthotropic steel decks, Steel Constr. Eng. Jpn. Soc. Steel
Constr. 15 (59) (2008) 75–86 (In Japanese).
[14] S. Kainuma, S. Onoue, K. Miura, S. Inokuchi, A. Kawabata, D. Uchida, Development of
an experimental system for fatigue-cracking from weld roots between deck plate
and u-rib in orthotropic steel decks, J. Jpn. Soc. Civ. Eng. A 64 (2) (2008) 297–302
(In Japanese).
[15] S. Inokuchi, S. Kainuma, A. Kawabata, D. Uchida, Field measurement and develop-
ment of an experimental system for fatigue-cracking from weld roots between
deck plate and U-rib in orthotropic steel decks, Proceeding of 2nd International
Orthotropic Bridge Conference, Sacramento Section, American Society of Civil
Engineers, Sacramento, California, USA 2008, pp. 345–357.
[16] Japanese Standards Association, Japanese Industrial Standard G3106: Rolled steels
for welded structure, 2015.
[17] C. Miki, Fatigue damage in orthotropic steel bridge decks and retrofit works, Int. J.
Steel Struct. 6 (4) (2006) 255–267.
[18] S.J. Maddox, Fatigue Strength of Welded Structures, Woodhead publishing, 1991.
[19] C. Acevedo, A. Nussbaumer, J.M. Drezet, Evaluation of residual welding stresses and
fatigue crack behavior in tubular K-joints in compression, Stahlbau 80 (7) (2011)
483–491.
[20] S. Inokuchi, S. Kainuma, D. Uchida, D. Shiro, Influence of press reforming in fabrica-
tion process on stress properties of welded joint between deck plate and U-shaped
rib in orthotropic steel decks, Steel Constr. Eng. Jpn. Soc. Steel Constr. 19 (73) (2013)
1–8 (In Japanese).
[21] N. Tebedge, G. Alpsten, L. Tall, Residual-stress measurement by the sectioning meth-
od, Exp. Mech 13 (2) (1973) 88–96.
[22] S. Kainuma, D. Takamatsu, Fatigue behavior of fillet welded cruciform joint under
compressive cyclic stress, J. Constr. Steel Jpn. Soc. Steel Constr. 8 (2000) 723–730
(In Japanese).
[23] S. Kainuma, T. Mori, A study on fatigue crack initiation point of load-carrying fillet
welded cruciform joints, Int. J. Fatigue 30 (9) (2008) 1669–1677.
[24] S. Ya, K. Yamada, T. Ishikawa, Fatigue evaluation of rib-to-deck welded joints of
orthotropic steel bridge deck, J. Bridg. Eng. 16 (4) (2011) 492–499.
[25] T. Mori, Influence of weld penetration of fatigue strength of single-sided fillet
welded joints, Steel Constr. Eng. Jpn. Soc. Steel Constr. 10 (40) (2003) 9–15
(In Japanese).