International Journal of Steel Structures

December 2013, Vol 13, No 4, 707-721

DOI 10.1007/s13296-013-4012-z

www.springer.com/journal/13296

Fatigue Strength Evaluation of Welded Structural Details

in Corrugated Steel Web Girders
Zhiyu Wang1,2*, Lianfei Tan1,3, and Qingyuan Wang1,2
1
Department of Civil Engineering & Mechanics, Institute of Architecture and Environment, Sichuan University,
Chengdu, Sichuan, 610065, China
2
Key Laboratory of Energy Engineering Safety and Disaster Mechanics, Ministry of Education, Sichuan University,
Chengdu, Sichuan, 610065, China
3
Yunnan Electric Power Design Institute, Kunming, Yunnan, 650051, China

Abstract

Fatigue strength evaluation of welded structural details is of practical significance in the design of corrugated steel web
girders in highway bridges and industrial structures. In this paper, the fatigue strengths of corrugated steel web girders with
several welded structural details and welding methods are analytically examined by fatigue tests of corrugated web beams and
small-size welded joints. The stress concentration & distribution characteristics of corrugated web beams were analysed using
finite element analysis. The beam test results showed that the structure with scallops or notches in the flange has lower fatigue
strength and that with butt joints is prone to suffer from shear crack on the corrugated web. Within the inside range of the
scallop, the stress concentration becomes greater with the increase of scallop radius and the tension flange contributes
significantly to the bending capacity of corrugated steel web beams. The tests for small-size welded joints indicated their
applicability in the analysis and prediction of S-N relationship of the test corrugated steel web beams. Finally, the fatigue crack
propagation lives of weld joints were compared with those of test corrugated steel web beams with respect to related design
Categories of the AASHTO LRFD Bridge Design Specifications.

Keywords: fatigue strength, corrugating, bridge, stress concentration, life prediction

1. Introduction conventional girders with welding transverse stiffeners

(Fisher and Wright, 2001). Experimental investigations
Corrugated steel web girders have been paid increasing on the fatigue behaviour of corrugated steel web beams
attention in recent projects of highway bridges and have been conducted by previous researchers (Harrison,
industrial structures (e.g. crane runway). In contrast to the 1965; Korashy and Varga, 1979; Kamuki et al., 2001;
girders with conventional stiffened webs, the corrugated Kotaki et al., 2004; Ibrahim et al., 2006; Sause et al.,
steel webs with adequate out-of-plane stiffness and buckling 2006; Kövesdi and Dunai, 2008; Li et al., 2012). It has
resistance contribute beneficially to the structural mechanical been confirmed in these studies that the corrugated steel
performance. Their merits in economic design are also web beams have improved fatigue performance in contrast
indicated from several previous investigations, in which it to conventional stiffened beams. Based on the design curves
was reported that the weight of corrugated web girders of the AASHTO LRFD Bridge Design Specifications
can be 30% less than that of conventional flat web girders (2004) and limited number of tests, the fatigue design
(Roberts and Heywood, 1994). curve of the corrugated web girders was suggested
Apart from being economic, the corrugated steel web between categories B and C (Ibrahim et al., 2006) or
girders have also been regarded as an alterative to category B’ (Sause et al., 2006).
Depending on the geometric configurations, however,
Note.-Discussion open until May 1, 2014. This manuscript for this the fatigue failures at critical points in the corrugated steel
paper was submitted for review and possible publication on February web girders may be practically different from those reported
7, 2013; approved on October 29, 2013. in the literature. For example, the welding methods and
© KSSC and Springer 2013 scallop & lap joint details are introduced in the fabrication
*Corresponding author
process; or the corrugation angles are varied for wide flange
Tel: +86-28-85406919; Fax: +86-28-85406919 section. Most reported experiments were based on the
E-mail: zywang@scu.edu.cn beam tests with limited numbers and specific configurations,
708 Zhiyu Wang et al. / International Journal of Steel Structures, 13(4), 707-721, 2013

e.g., a fixed corrugation angle, which evidently could not GB/T8110-2008 (2009) and GB50017-2003 (2003). The
provide inclusive understanding of structural performance chemical composition and mechanical properties are
for fatigue design. Compared with related studies for listed in Table 2.
conventional flat web girders, very few experimental
studies have been conducted regarding these issues for 2.1.2. Fatigue test procedure
corrugated steel web girders so far. An overview of test set-up is shown in Fig. 1. All
In this study, experimental tests were conducted to fatigue tests were performed with three-point bending.
investigate the fatigue strength of corrugated steel web The smooth beam surfaces at support and loading point
girders including several welded structural details and were ensured to provide simply supported conditions and
welding methods. Finite element (FE) modelling was avoid any eccentricity in loading. Lateral braces were
performed for the analysis of stress concentration & also installed at the support and loading point to avoid
distribution of corrugated web beams and their comparison excessive lateral distortion. The fatigue test was carried
with counterpart conventional flat web beams. As a out using a 200 kN capacity Shimadzu 4890 fatigue
supplement for the beam tests, fatigue tests of tensile testing machine and the test data were monitored and
plates were also carried out on the small-size welded recorded using Windows-based Gluon test execution
joints to further examine the influences of configuration version 2.50c software which offers comprehensive
details on the fatigue strength. Based on the comparative customisation wizards. During fatigue loading process for
study of experimental and analytical results, comments all tests, the load was applied at the centre of span and
are provided for the assessment of fatigue strength of controlled using a form of sinusoidal wave with constant
corrugated steel web girders with the AASHTO LRFD frequency of 3 Hz. The stress ratio was set at 0.1. The
Bridge Design Specifications. minimum load was applied at 8 kN within 3,000,000
cycles, and subsequently increased to 10 kN up to failure.
2. Fatigue Test Program and Fe Analysis of Monotonic loading and full cycle reading were applied as
Test Beams repetitive loading at following intervals: 10,000th cycle;
100,000th cycle; 200,000th cycle; 500,000th cycle;
2.1. Experimental tests 1,000,000th cycle; 2,000,000th cycle; 3,000,000th cycle.
2.1.1. Description of test specimens During the test, the vertical deflection of test beams at
The dimension and configuration of four test beam mid-span was measured using Linear Variable Differential
specimens are shown in Fig. 1. These specimens are 1760 Transformers (LVDTs). To measure the local longitudinal
mm long, and their flange plates are 6 mm thick & strain variations, four single element strain gauges (SG)
88 mm wide. The longitudinal fold and the inclined fold were placed on the top surface of tension flange
are of the same widths of 82 mm and connected with the (connecting corrugated web) at the centre of span and
transition curvature in between. The corrugated steel web near the welds corresponding to the scallop & lap joint
are varied under design consideration of the influences of details, as illustrated in Fig. 3. The fatigue cracks in the
corrugation angle, scallop and lap joint details. As listed vicinity of fillet welds were inspected using magnifying
in Table 1, Specimens B1-30o-n and B4-45o-n are normal glass.
corrugated steel web beams with web corrugation angles
of 30o and 45o respectively. Specimen B2-30o-c has four 2.1.3. Test results
scallops with radii of 20 mm interrupting fillet welds at Due to different configurations, each test beam was
lower and upper flange plates respectively. Specimen B3- supposed to have a varied range of strain and deflection
30o-l has lap joints and butt joints near and away from during fatigue testing although the stress ratio and
mid-span respectively. To avoid transverse bending of minimum load were the same. The fatigue crack initiation
compressed flange, the local stability of beam flange was lives (Ni) when the appearance of crack was detected and
checked in accordance with the Chinese technical the final lives (Nf) when test was terminated are listed in
specification CECS 291-2011 (2011). Table 1. Figure 2 shows a comparison of the defection
CO2 shielded semiautomatic Gas Metal Arc Welding range (DR) versus the number of cycles of the test
(GMAW) were performed for all fillet welds jointing specimens. It can be seen that Specimen B2-30o-c with
flange plates and corrugated webs. The welding stop- scallops and Specimen B3-30o-l with thicker web and lap
starts are avoided during welding process. After fabrication, joints have highest and lowest deflections respectively
the specimens were finished by grinding in accordance throughout the full fatigue loading range. This observation
with related specifications. As listed in Table 2, the parent can be expected as the contribution of the corrugated web
metal used in the beam tests are PM1 which conforms to stiffness on the beam deflection. To compare the influence
Q345 steel of the Chinese national standard GB/T700. of the corrugation angle, a web flexibility ratio (FR)
The welding wire material (Type: ER50-6) and fillet welds proposed by Abbas et al. (2006) is also introduced as
meet the requirement of the Chinese national standards follows.
 Fatigue Strength Evaluation of Welded Structural Details in Corrugated Steel Web Girders 709

Figure 1. Dimension of test specimens and overview of test arrangement.

Table 1. Summary of test specimens

Reference Corrugation angle Web thickness Configuration Ni Nf
Stress ratio
Name (degree) (mm) type ×103 ×103
B1-30o-n 30.6 4 w/o cope hole - 1048.3
B2-30o-c 30.6 4 w cope hole 600 1288.4
w/o cope hole 0.1
B3-30o-l 30.6 6 500 3330.0
web lap jointing
o
B4-45 -n 45 4 w/o cope hole 200 3200.7
710 Zhiyu Wang et al. / International Journal of Steel Structures, 13(4), 707-721, 2013

Table 2. Chemical composition and mechanical properties of parent metal (PM), welding wire & rode (WW) materials
Chemical composition (%) Mechanical properties
Type Yield stress Ultimate stress Elongation
C Si Mn P S
(MPa) (MPa) (%)
PM1:Q345 0.16 0.33 1.36 0.035 0.022 455*/480 555*/565 33*/25
PM2:Q345 0.16 0.19 0.71 0.013 0.013 370 510 29
WW:ER50-6 0.15 1.15 1.85 0.025 0.025 420 500 22
WW:CHE42 0.14 0.32 0.48 0.045 0.04 410 495 22
Note: * denotes the composition & properties of 4mm thick steel plate.

Figure 2. Comparison of deflection range versus number of cycles curves.

Specimens B1-30o-n & B4-45o-n: These specimens are

h 2 3β + 2 ⎞ varied with the corrugation angles. Specimen B1-30o-n
FR = 2⎛----r⎞ ⎛------------------
- (1)
⎝ tw⎠ ⎝β + cosα⎠ failed earlier from a fatigue crack initiation from a notch
in the flame cut edge of the flange plate, while the failure
where hr and tw are the corrugation depth and the web in Specimen B4-45o-n was seen initiating from the end of
thickness respectively, β is the ratio of the longitudinal the inclined fold intersecting with the transition curvature.
fold width to the inclined fold width, and α is the At the final stage of crack propagation, the fatigue cracks
corrugation angle. As hr, tw and β are unchanged variables are almost perpendicular to the longitudinal stress direction.
for test specimens, Eq. (1) can be reduced to 1/(β+cosα) From the readings of the strain measurement, it can be
for the purpose of comparing the flexibility of different seen that the strain range distribution in Specimen B1-
corrugated web profiles. As shown in Fig. 2, the average 30o-n is wider than that in Specimen B4-45o-n across the
DR value for B4-45o-n is very close to the prediction that tension flange. Compared with a four-point bending test
multiplying average DR value of B1-30o-n by (1+cos30o)/ reported early by Harrison (1965), the failure observation
(1+cos45o) in the range between 150,000 cycles and in this test is similar and, however, the fatigue lives of test
220,000 cycles. beams are slightly lower. This can be attributed to that
Since the fatigue strength of the corrugated steel web most ranges of the beam are in combined bending &
girders are dependent on the welded structural details, the shearing and the corrugations are not in sinusoidal form.
failure mechanism and related influential parameters were Specimen B2-30o-c: The half round cut-outs on the
investigated in the test program. Typical fatigue damage corrugated web were supposed to induce weakening effect
was recorded for the test specimens corresponding to the on the corrugated web and interrupting the longitudinal
fatigue crack initiation and the end of final life, as shown fillet welding on the tension flange. As shown in Table 3,
in Table 3. The strain range measurements corresponding the crack initiated at the end of scallops on the top surface
to each strain gauge (SGM) at the top surface of tension of the tension flange which was visible earlier than that
flange at mid-span are also indicated in Fig. 3. Based on on the surface of corrugated web. At the final stage, the
the test results, the following failure modes can be cracks propagates on the beam tension flange perpendicular
identified: to the longitudinal direction of the beam. Also, a visible
 Fatigue Strength Evaluation of Welded Structural Details in Corrugated Steel Web Girders 711

Table 3. Typical graphs of fatigue crack and failure of test beams

crack turns into the corrugated web around the scallop 2.2. Finite element analysis
about the top surface of tension flange by approximately 2.2.1. Description of finite element models
50o. The influence of scallop details on the strain range Experimental observations have indicated that fatigue
can also be identified. It is shown in Fig. 3 that a cracks and failures took place at aforementioned welded
fluctuation and decrease in strain range to nearly 600,000 structural details in the corrugated steel web beams. Given
cycles which is followed by an increased trend as the irregular structural configuration of corrugated webs, it is
results of local deterioration. unlikely that all the data needed for the fatigue strength
Specimen B3-30o-l: The initiation and propagation of analysis are available from experimental instrumentation.
fatigue cracks are seen at the web side and its related Finite element modelling, due to its efficient applicability
weld toes of flange-to-web, stiffener-to-web and lap to structural analysis, was carried out to evaluate the
joints. It is noted that the welded joints between the web stress distributions characteristics and structural fatigue
and the intersecting plate can be taken as a load-carrying vulnerable points of corrugated steel web beams. The
type (Miki and Tateishi, 1997). This means it is more software package ANSYS was adopted for this analysis.
susceptible to suffer fatigue cracks than that for non-load- Three dimensional finite element models were developed
carrying type, such as welded joints between web and for the corrugated steel web beams tested in the experiment
intersection plate. The final failure of this specimen was and conventional flat web beams with similar geometry
due to the shear crack at the butt joints on the corrugated for comparison. For the ease of visualization, only one-
web and fracture in the tension flange. In addition, the half of the finite element model is shown in Fig. 4. The
strain range variation recorded at the mid-span for this varied geometric configurations were considered as
case is relatively stable in contrast to other specimens, as corrugation angle and scallop details. Shell elements,
shown in Fig. 3. which have four nodes with six degrees of freedom at
712 Zhiyu Wang et al. / International Journal of Steel Structures, 13(4), 707-721, 2013

Figure 3. Strain range versus number of cycles curves.

Figure 4. Illustration of FE models.

each node, were used for finite element model construction. measurement. Stress contours in the finite element
Relative fine mesh was used in the vicinity of scallops analysis are also compared in relation with the failure
and the transition curvature areas between the longitudinal observation in the experiment. A typical stress contours
folds and the inclined folds of the corrugated webs. To of FE modelling for specimen B2-30o-c is shown in Fig.
explicitly include the stiffness of the fillet welds in the 6. It can be seen that relative high stress occurred at the
corrugated steel web beam, inclined elements were created end of a scallop on the top surface of tension flange in the
connecting the corrugated web and the flange plate. The modelling which agrees well with the fatigue cracking
finite element modelling results were compared with the location observed in the tests.
monotonic loading test results after above mentioned
loading intervals. Figure 5 shows a comparison of moment 2.2.2. Stress concentration analysis
versus middle span deflection from test measurement and Given the fatigue crack growth from the weld toe as an
finite element modelling for Specimen B4-45o-n. It can essential failure mode of tested corrugated steel web
be seen that the modelling result agrees well with the test beams, the structural stress approach was introduced to
 Fatigue Strength Evaluation of Welded Structural Details in Corrugated Steel Web Girders 713

Figure 5. Comparison of moment-deflection data for specimen B4-45o-n.

surface (Niemi, 1995). The first and second extrapolation

points herein were selected at distances at 0.4tp and 1.0tp
away from the weld toe, in which tp is the plate thickness.
It is noted that this plate thickness dependent stress
distribution is not applicable to the case that the weld is
located on a plate edge (Niemi and Marquis, 2002), such
as the edge of the scallop in the web plate in this study.
Accordingly, only the former hot spot is considered
herein. In the calculation of stress concentration, the
values of stresses at two points, which lie in 2.4 mm and
6 mm from the weld toe, are obtained from nodal stress
results in FE analysis firstly. The hot spot stress is then
counted through a linear extrapolation from these two
points to the weld toe on the top surface of the tension
flange. For the purpose of simplicity, the stress concentration
Figure 6. Typical stress contour of FE modelling for
along the weld toe can be identified from a normalized
specimen B2-30o-c.
parameter, i.e. stress concentration factor (SCF), which
can be written as:
evaluate the stress concentration characteristics herein.
As illustrated in Fig. 7, the hot spot stress is obtained by σhs
SCF = ------
- (2)
extrapolation from stresses located at distance in relation σn
to plate thickness away from the weld toe on a plate

Figure 7. Illustration of hot spot stress calculation through extrapolation.

714 Zhiyu Wang et al. / International Journal of Steel Structures, 13(4), 707-721, 2013

Figure 8. Comparison of SCF for specimens with varying structural details.

Figure 9. Comparison of bending moment effect on the stress distribution in the gap range.

where σhs and σn are the hot spot stress and nominal web, the longitudinal stress state of flange are assumed to
stress respectively. Using the SCF defined in this manner, be only influenced by the bending moment. Accordingly,
the stress concentration was further evaluated for the the shear force was kept constant while the bending
specimens with expected crack initiation at the end of the moment was varied for a comparative study of the
inclined fold intersecting with the transition curvature or bending moment effect on the stress distribution in the
at the gap range of the scallop. scallop gap range on the tension flange. The applied
The analytical SCF results with varying geometric stress magnitude (S) is calculated as the longitudinal
parameters related to corrugated web and scallop are stress on the top surface of the tension flange exclusively
plotted in Fig. 8. It can be seen that the SCF values from bending moment in accordance with the simple
increase significantly with the increase of the corrugation beam theory. The analytical results at a shear force level
angle. This is owing to the abrupt change of stress state of 37.5 kN are compared in Fig. 9. The end of gap range
at the end of the inclined fold intersecting with the away from the centre of span exhibits higher longitudinal
transition curvature of corrugated web which is greatly stress with respect to that close to the mid-span when
influenced by the corrugation angle. An improvement of S>100 Mpa. The increase of bending moment indicates a
SCF values can also be observed for the test specimens significant increase of longitudinal stress on the tension
with increased scallop radius. The stress concentrations at flange. The longitudinal stress distributed in the gap range
welds relating to the scallops for corrugated steel web is relatively higher for the flat web beams than that for
beams are slightly lower than these for the conventional the corrugated steel web beams, especially at the location
flat web beams with varying scallop radius. near the centre of span. The peak stress differences
between both ends of the gap range for the corrugated
2.2.3. Stress distribution analysis web beams are larger than these for the conventional flat
As the shear force is primarily carried by the corrugated web beams which suggests a larger contribution of
 Fatigue Strength Evaluation of Welded Structural Details in Corrugated Steel Web Girders 715

Figure 10. Comparison of transverse stress distribution at selected cross-sections.

tension flange in the gap range to the bending capacity of corrugated web. From the stress distribution shown in
corrugated web beams. With the increase of the bending Fig. 10(b), it can be seen that the increase of corrugation
moment induced stress, this variation decreases gradually. angle from 30o to 45o amplifies the transverse stress at the
To study the influence of geometric parameters on the toe of welds near the end of the inclined fold intersecting
stress state of the tension flange perpendicular to the with the transition curvature. The maximum amplification
longitudinal stress direction, a comparative study was of this change is 15%~18%. Similarly, the transverse stresses
also conducted on the transverse stress distribution at the of the flat web beam with similar geometric parameters
weld toes near the end of the inclined fold intersecting distribute linearly along beam width and higher than
with the transition curvature and of the gap range of the those of the corrugated web beams in comparison.
scallop. When a load of 72 kN was applied at the centre The web locally weakened by scallop details may be
of span, the modelling results are compared in Fig. 10. subjected to high stress and fatigue crack. Since the
The transverse stress distributions exhibit almost full corrugated web is mostly influenced by the shear force
symmetric pattern along the middle of beam width for the rather than bending moment, the structural stress analysis
conventional flat web beams and partly symmetric form was conducted by changing the beam span so as to vary
along the centreline of corrugation for corrugated web shear force and keep bending moment constant. The non-
beams. As the scallop radius increases, transverse stress dimensional equivalent stress distributions in the
increases significantly on the tension flange away from circumferential direction along the periphery of the
the weld toes at the end of the gap range of the scallop. scallop adjacent to the mid-span are compared in Fig. 11.
In contrast, such an increase seems to be almost The circumferential stress increases significantly with the
negligible for flat web beams as shown in Fig. 10(a). This increase of shear force. The peak stresses on the scallop
observation can be expected as the configuration of edge of the corrugated web occur at the 45o~65o angle to
corrugated web is essentially irregular and producing the horizontal close to the centre of the span which agrees
such higher transverse stress when the end of the gap with the experimental observation of fatigue cracking
range of the scallop is approaching the intersection points mode. In contrast, the flat web beam has greater stress
between the longitudinal fold and the inclined fold of the distributed along the most range of periphery of the
716 Zhiyu Wang et al. / International Journal of Steel Structures, 13(4), 707-721, 2013

range versus fatigue life were also considered for

comparison.

3.1.1. Description of test specimens

As shown in Fig. 12, the specimen with an attachment
in corrugated shape is fillet welded to a tensile plate to
represent similar web-to-flange fillet welded details in the
corrugated web beam. The designed specimens incorporate
the local stress concentration factors induced by the
geometry of the weld and local features of the corrugated
web. The width and thickness of main tensile plates are
similar to those of flange plate in corrugated web beam
specimens. The corrugation angle of 30o and 45o were
used for all test specimens. The design fillet weld leg
Figure 11. Comparison of stress distribution around cope length is 3 mm for all welds between the attachment and
hole.
the tensile plate. For the sake of a comparative study,
several specimens with scallops of radius of 8 mm were
scallop and its peak stress point has larger angle from the considered on the longitudinal folds of corrugated web
horizontal. This observation can be expected as the stress and manual arc welding (MAW) using type CHE42
on the flat web is contributed by the combination of shear welding material. The parent metal is PM2 which
force and bending moment which produces higher stress conforms to Q345 steel of the Chinese national standard
at local web around the scallop details. Additionally, it is GB/T700, as described previously in Table 2.
noted that the distinct stress distribution on the corrugated
web around the scallop details is also affected by the 3.1.2. Fatigue test procedure
web-to-flange fillet welds on the flange plate. A typical set-up of fatigue tests for small-size welded
joints is shown in Fig. 13. The tension load was applied
3. Fatigue Test Program of Small-size using aforementioned Shimadzu 4890 fatigue testing
Welded Joints machine in the beam tests. The affixture for bending used
formerly was replaced by a set of tensile accessories. The
3.1. Experimental tests one end of the main plate of the small-size welded joint
To further the understanding of the fatigue strength of was gripped in the upper head of the test machine with
welded details indicated in the previous section, a small- special clamp splices to ensure that only the main plate is
size welded joint test was conducted as a supplement to involved with directly applied tensile stress. Meanwhile,
aforementioned beam tests. The design of test specimens the other end of the main plate was gripped in the fixed
for this part of study is in agreement with the basis of bottom head of the test machine to provide reaction load.
fatigue design rules in which the fatigue strength of a In this way, the weld is not carrying the appreciable part
welded detail in a small-size specimen can be used to of the tensile load but strained by the load in its joined
estimate the performance of the same details in a structure. stressed main plate. This was considered to replicate the
Accordingly, the welded joint specimens were designed loading condition of the web-to-flange fillet welds in the
and manufactured with similar corrugation angles, welding test beams in which, as noted previously in the beam
methods and scallop details. The relationships of stress tests, the bending moment induced longitudinal stress is

Figure 12. Typical configurations of test welded joints (Unit: mm).

 Fatigue Strength Evaluation of Welded Structural Details in Corrugated Steel Web Girders 717

tested specimens are schematically shown in Fig. 14.

Apart from one specimen cracks from an original notch in
the plate edge, the rest tested specimens without scallops
all failed at the location where the transition curvature
and the inclined fold of the attachment plate intersects.
Comparing the specimens with corrugation angles at 30o
and 45o (Types I & II), it seems that the location of
fatigue crack initiation moves from the end of the inclined
fold with the increase of corrugation angle. This can be
attributed to the change of local stress concentration
manner with the corrugation angle. The specimens with
scallops (Types III) in the test failed with the fatigue
cracks arising from the ends of scallops close to the
intersection points between the transition curvature and
the inclined fold of the attachment plate. A typical
Figure 13. Fatigue test set-up for small-size welded joints. fracture surface of welded joints under stress range of
160 Mpa is shown in Fig. 15. The fatigue cracks were
initiated from the weld toes at the end of the inclined fold
primarily carried by the flange plate. Constant amplitude intersecting with the transition curvature, perpendicular to
sinusoidal stress cycles with a frequency of 8 Hz were the longitudinal stress. The welded joint failed as the
conducted during the fatigue test. The levels of stress cracks propagated through the thickness of the tensile
range between 120 Mpa and 200 Mpa were selected and plate.
the stress ratio was set at 0.1 for the test. The fatigue life The comparison of fatigue life relating to the joint
was determined as the test specimen was tested to configuration and welding method is shown in Fig. 16.
rupture. The actual measured cross-sectional dimensions The test data corresponding to the stress range of 140 Mpa
were used for the calculation of nominal stress of testing and 160 Mpa were selected for this comparison. For
specimens. welding method, it can be seen that the fatigue life of
specimens with MAW welding (Mat type: CHE42) is
3.1.3. Failure modes and fatigue life comparison slightly longer than those with GMAW welding (Mat
The typical modes of fatigue cracks at fillet welds of type: ER50-6). This difference seems more evident for

Figure 14. Illustration of typical fatigue cracks on welded joints.

Figure 15. Typical fracture surface of welded joints under ∆S=160 Mpa.
718 Zhiyu Wang et al. / International Journal of Steel Structures, 13(4), 707-721, 2013

Figure 16. Comparison of cycle number with varying joint configuration and welding method.

the specimens with increased corrugation angle. For the Fig. 17 with the same scale and data range for comparison.
scallop details, it can be seen that the fatigue life of the The stress range of referred specimens with the corrugation
welded joint is decreased by 33.8% which is due to the angle of 30o is closer to the mean regression line
fact that the fatigue cracks initiated early at the ends of expression of Eq. (3). In contrast, the stress range of test
scallops. specimens with corrugation angle at 36.8o (a slope of 3:4)
reported by Ibrahim et al. (2006) lies between mean
3.2. Test data comparison between welded joints and regression line expressions of Eq. (3) and Eq. (4)
beams corresponding to the corrugation angles of 30o and 45o
As the small-size welded joints under repetitive primary respectively. As such, the test results of corrugated web
stress are designed similar to welded areas of the corrugated beams with the corrugation angle (36.9o) conducted by
web beams subjected to load-induced fatigue, the fatigue Sause et al. (2006) are added for comparison. In this
test data of both models are considered comparable with referred study, a best fit equation for the test data from the
each other and the fatigue test results of welded joints and remaining life analysis was suggested as:
corrugated steel web beams in this study and reported in
Log (N)=12.68−3 Log (∆S) (5)
the literature are compared in this section. The fatigue test
results of welded joints in Log (N) versus Log (∆S) As can be seen from Fig. 17, the referred Eq. (5)
relationship are summarised in Fig. 17, in which N and represents almost the mean value between Eq. (3) and Eq.
∆S denote fatigue life and stress range in Mpa respectively. (4), which corresponds to the calculated stress ranges of
The method of least squares was used in the test result 133.76, 148.19, and 116.95 Mpa respectively at 2 million
post-processing to produce the best fit mean Log (N) cycles. This demonstrates the equations from the regression
versus Log (∆S) curve. The mean regression line for the analysis of the fatigue test results of welded joints are
welded joints with the corrugation angle at 30º can be applicable to the case of beam tests provided that the
given by: stress concentration is dependent on the parameter of the
corrugation angle. In addition, it is noted that different
Log (N)=12.942−3.059 Log (∆S) (3)
fatigue crack propagation modes and structural details,
o
and, those with corrugation angle at 45 can be expressed e.g. crack initiating from a notch in the flame cut edge of
as: the flange plate, the presence of scallops, would result in
lower fatigue strengths than those predicted from linear
Log (N)=11.773−2.646 Log (∆S) (4)
regression analytical expressions of welded joints.
The standard deviations (SD) were also calculated to The comparison of S-N relationship of the welded joint
account for the scatter in fatigue lives obtained in the tests for the AASHTO fatigue details are plotted in Fig.
experiment. A Log (N)-Log (∆S) curve corresponding to 18. Generally, the detail of web-to-flange weld parallel to
two standard deviations below the mean (M-2SD) was the flange longitudinal stress direction in conventional
plotted for each group of specimens to represent 95% unstiffened flat web girder is considered as Category B in
confidence limit. This transformation can also be regarded the AASHTO LRFD Bridge Design Specifications. In
as a lower bound curve for the purpose of fatigue design. contrast, the test data for the welded joints with the
Apart from the beam tests mentioned in the Section 2.1, corrugation angle of 30o distribute between Categories A
the results of related beam test reported by Kotaki et al. and B, while those of 45o fall between Categories B and
(2004) and Ibrahim et al. (2006) are added and plotted in B’. The lower bound design curve (M-2SD) for the welded
 Fatigue Strength Evaluation of Welded Structural Details in Corrugated Steel Web Girders 719

Figure 17. Comparison of log (N)-Log (∆S) curves for welded joints and corrugated beams.

Figure 18. Comparison of welded joint fatigue data for AASHTO fatigue detail categories.

joint with corrugation angle of 30o lies above Categories where FS, FE and FT are correction factors for free
B, and that of 45o is slightly higher than the mean for surface, crack shape and finite thickness or width
Category B’. Consequently, the fatigue life of the respectively, which are associated with the parameters of
Categories B’ curve can be regarded as a lower bound depth and/or half-length of elliptical surface crack. FG is
reference for the welded joints with corrugation angles a geometry correction factor which is related to stress
between 30o and 45o. gradient. Since the cracks occur at geometric discontinuities
Based on the fracture mechanics theory, the relationship and the forces are distributed symmetry with respect to
between fatigue crack propagation rate, da/dN and stress the centre of the crack, the calculation of FG can be
intensity factor, ∆K, can be expressed as follows (Albrecht assumed under discrete stress condition si over the
and Yamada, 1977; Maddox, 1991): element width from xi to xi+1 (Albrecht and Yamada,
1977). Then the stress concentration produced by
da- m structural details can be given by the ratio of the effect for
------ = C(∆K) (6)
dN a non-uniform stress distribution along the line of the
crack to that for an uniformly distributed mean stress, sm,
where C and m are material constants determined as:
experimentally. In this study, m is assumed as 3.0. ∆K can
n
σ
FG = --2- ∑ -----i- arcsin⎛--------
xi + 1⎞ x
– arcsin⎛ ---i⎞
be obtained as:
(8)
π σm ⎝ a ⎠ ⎝ a⎠
i=1
∆K = Y∆S πa = FSFEFTFG∆S πa (7)
where a is the surface-half-length for a semi-circular
720 Zhiyu Wang et al. / International Journal of Steel Structures, 13(4), 707-721, 2013

Figure 19. Comparison of fatigue crack propagation life ratio from welded joint test and FE modelling.

crack in a semi-infinite plate subjected to uniform stress. 4. Summary and Conclusion

This equation can be recast in the form of SCF when σi
is replaced by σhs over the width ranging from 0 to a. This paper presents a study to investigate the fatigue
By integrating Eq. (6), the fatigue crack propagation strength of welded structural details in corrugated steel
life from an initial crack length, ai, to a final crack length, web girders. Firstly, the experimental tests and finite
af, can be obtained as: element analyses were conducted on the typical failure
modes of varied welded details in the corrugated steel
1
1- af -------------- 1
1- af ------------------------------------------------ web beam specimens. The stress concentration & distribution
N = --- ∫ -da = --- - da
C∫ai (F F F F ∆S πa)m
a
(9)
C i (∆K) m characteristics were then examined and compared between
S E T G
the conventional flat web beams and corrugated web
The ratio of fatigue crack propagation life between beams with scallop and lap joint details. The small-size
welded joint with corrugation angle of 30 degree (N300) welded joint tests with similar configuration were also
and that of 45 degree (N450) can be written in Eq. (10). performed to examine the relationship between fatigue
Assuming the fatigue crack propagation life of welded life and stress range with respect to the AASHTO LRFD
joints with different corrugation angle is primarily Bridge Design Specifications. The conclusion can be
determined by FG and its related stress concentration drawn as follows:
factor, this equation can be further simplified as: (1) The fatigue crack initiation of corrugated web beams
mostly occurs at the end of the inclined fold intersecting
af 1 with transition curvature and the stress concentration at
1 -da
∫a ------------------------------------------------m- da ∫
af
------------- this location increases with the increase of corrugation
N45o
i
(FSFEFTFG∆S πa) o
ai
( FG )
m
o
angle. However, a notch in the flame cut edge of the
45 45
--------- = -----------------------------------------------------------------------
- = ------------------------------------
N30o 1 af 1 flange plate would alter the failure mode and shorten the
af
∫ai ------------------------------------------------m- da ∫ai -------------
( F )
m
-da
o
structural fatigue life.
(FSFEFTFG∆S πa) o
G 30
(2) The stress concentration inside the scallop of
30
corrugated web beams becomes greater as the scallop
(10)
radius increases and its magnitude is slightly lower than
Substituting aforementioned numerical SCF results of that of counterpart conventional flat web beams.
corrugated web beam in the Section 2.2 into Eq. (10), (3) The peak stress variations between both ends of the
N450/N300=0.64 can be calculated and compared with test gap range in the scallops for the corrugated web beams
welded joint results within stress range between 100Mpa are larger than those for the conventional flat web beams.
and 200 Mpa. As shown in Fig. 19, the test fatigue crack This suggests a larger contribution of tension flange in the
propagation life ratio of test welded joints is slightly gap range of the scallops to the bending capacity of
lower than that predicted from Eq. (10) based on the corrugated web beams.
numerical SCF in the beam tests, which indicates a slightly (4) The corrugated web beam with butt joint suffered
underestimated prediction of fatigue crack propagation from shear crack on the corrugated web which indicates
life with the increase of stress range. This also suggests special attention should be paid to the effect of shear
the welded joint test results would provide a safe estimation force in the corrugated web beam for this detail.
of fatigue crack propagation life when determined from (5) The basic failure modes of small-size welded joints
the test data of welded joints with lower corrugation under fatigue loading are similar to those in the experimental
angle and numerical SCF ratio. tests of corrugated web beams. The fatigue test of welded
 Fatigue Strength Evaluation of Welded Structural Details in Corrugated Steel Web Girders 721

joints showed that the location of fatigue crack initiation (in Chinese).
moves from the end of the inclined fold with the increase Harrison, J. D. (1965). “Exploratory fatigue tests on two
of corrugation angle. girders with corrugated webs.” British Welding Journal,
(6) The fatigue life of welded joint fabricating by 12(3), pp. 121-125.
MAW welding method is shorter than its counterpart with Ibrahim, S. A., El-Dakhakhni, W. W., and Elgaaly, M.
GMAW welding method which becomes more significant (2006). “Behavior of bridge girders with corrugated webs
under monotonic and cyclic loading.” Engineering
with the increase of the corrugation angle.
Structures, 28, pp. 1941-1955.
(7) As for S-N relationship, the equations from regression
Kamuki, M., Mori, T., Tanaka, K., Tanaka, M., and Irube, T.
analysis of the fatigue test results of welded joints are (2001). “Stress and fatigue properties of steel or
applicable to the case of beam tests when the corrugation prestressed concrete girder with corrugated steel webs.”
angle has significant effect on the stress concentration. Proc. International Symposium of the Japan Welding
The fatigue life of the AASHTO Categories B’ design Society, 10(69), pp. 460-461 (in Japanese).
curve can be regarded as a lower bound reference for the Korashy, M. and Varga, J. (1979). “Comparative evaluation
welded joints with corrugation angles between 30o and of fatigue strength of beams with web plate stiffened in
45o. the traditional way and by corrugation.” Acta Tech. Acad.
(8) Compared with fatigue life prediction based on the Sci. Hung, 89, pp. 309-346.
numerical SCF in the beam tests, the test results of Kotaki, N., Ichikawa, A., and Sasaki, E. (2004). “A proposal
welded joints showed a slightly underestimated prediction of steel girder bridges with ripple web and their fatigue
of fatigue crack propagation life with the increase of performance.” Journal of Structural Mechanics and
stress range. Earthquake Engineering, JSCE, 68(766), pp. 233-244 (in
Japanese).
Kövesdi, B. and Dunai, L. (2008). “Experimental fatigue
Acknowledgment analysis of girders with corrugated web.” Proc. 25th
Danubia-Adria Symposium on Advances in Experimental
The support from the National Natural Science Mechanics, Ceske Budejovice (Budweis) and Cesky
Foundation of PR China (No. 51308363 and No. Krumlov, Czech Republic, pp. 127-128.
50978174) is appreciated. Thanks are also given to Mr Li, G., Luo, X., Sun, F., and Fan, X. (2012). “Experimental
Liwei Tang for his assistance in the experimental work. investigation on fatigue performance of welded H-beam
with corrugated webs.” Journal of Building Structures, 1,
References pp. 96-103 (in Chinese).
Maddox, S. J. (1991). Fatigue strength of welded structures.
AASHTO (2004). AASHTO LRFD bridge design 2nd Ed., Abington Publishing, Cambridge, England.
specifications. 3rd Ed., AASHTO, Washington, D.C. Miki, C. and Tateishi, K. (1997). “Fatigue strength of cope
Abbas, H. H., Sause, R., and Driver, R. G. (2006). “Behavior hole details in steel bridges.” International Journal of
of corrugated web I-girders under In-plane loads.” Fatigue, 19, pp. 445-455.
Journal of Structural Engineering, ASCE, 132(8), pp. Niemi, E. (1995). Stress determination for fatigue analysis of
806-814. welded components. Technical Report IIS/IIW-1221-93,
Albrecht, P. and Yamada, K. (1977). “Rapid calculation of International Institute of Welding, Cambridge, UK.
stress intensity factors.” Journal of the Structural Niemi, E. and Marquis, G. (2002). “Introduction to the
Division, ASCE, 103(ST2), pp. 377-389. structural stress approach to fatigue analysis of plate
CECS 291-2011 (2011). Technical specification for steel structures.” Proc. IIW Fatigue Seminar, IIW-Commission
structures with corrugated webs. China Planning Press, XIII, pp. 73-90.
Beijing (in Chinese). Roberts, W. S. and Heywood, P. (1994). “An investigation to
Fisher, J. W. and Wright, W. J. (2001). “High performance increase the competitiveness of short span steel concrete
steel enhances the fatigue and fracture resistance of steel composite bridges.” Proc. 4th Int. Conf. on Short and
bridge structures.” International Journal of Steel Structures, Medium Span Bridges, Halifax, Nova Scotia, Canadian
1, pp. 1-7. Society for Civil Engineering, pp. 1161-1171.
GB50017-2003 (2003). Code for design of steel structures. Sause, R., Abbas, H. H., Driver, R. G., Anami, K., Fisher,
China Planning Press, Beijing (in Chinese). and J. W. (2006). “Fatigue life of girders with trapezoidal
GB/T8110-2008 (2009). Welding electrodes and rods for gas corrugated webs.” Journal of Structural Engineering,
shielding arc welding of carbon and low alloy steel. ASCE, 132(7), pp. 1070-1078.
Standardization Administration of China (SAC), Beijing