You are on page 1of 10

Experimental Study on Fatigue Resistance of Rib-to-Deck

Joint in Orthotropic Steel Bridge Deck


Ming Li1; Yasuo Suzuki2; Kunitaro Hashimoto3; and Kunitomo Sugiura4

Abstract: The stress state around a rib-to-deck (RD) joint in an orthotropic steel bridge deck is extremely sensitive to wheel-load location in
Downloaded from ascelibrary.org by Glasgow University Library on 11/17/17. Copyright ASCE. For personal use only; all rights reserved.

the transverse direction. Fatigue cracks of various types can occur around the RD joint. In this study, a bending fatigue-test system was used to
simulate an RD joint under the critical-load case. The fatigue tests for two groups of RD joint specimens with weld-penetration ratios of 15
and 75% were conducted to assess the influence of the penetration ratio on the fatigue-crack type and its fatigue resistance. Fatigue-test results
suggest that the fatigue resistances of the RD joint specimens with 75 and 15% penetration ratios were approximately Class B and between
Class C and Class D, respectively, per Japanese standards. Furthermore, results show that increasing the weld-penetration ratio can prevent
root-to-throat fatigue cracks and enhance the fatigue resistance RD joints. DOI: 10.1061/(ASCE)BE.1943-5592.0001175. © 2017 American
Society of Civil Engineers.
Author keywords: Orthotropic steel bridge deck; Fatigue crack; Fatigue test; Rib-to-deck joint.

Introduction the bending moment in the deck plate inclines to cracks in the deck
plate, as Type 1 and Type 2.
Orthotropic steel bridge decks (OSBDs) are used widely in long- In fatigue design specifications, S-N curves are generally
span bridges, urban viaducts, and moveable bridges because of employed to evaluate the fatigue life of various weld joints (JSSC
their lower dead weight and shorter onsite construction time than 2012; Hobbacher 2009; AASHTO 2012). This method, which is
those of concrete decks (Gurney 1992). Moreover, a bridge deck based on categorizing various weld joints, requires massive fatigue
system with lower weight causes less seismic damage to bridge tests for each weld joint type. However, the RD joints are directly
piers. Therefore the OSBD is a good option for bridges located in subjected to heavy axle wheel loads, so the stress state around the
areas with high earthquake risk. A typical OSBD, as presented in RD joint is sensitive to the transversal location of the wheel loads.
Fig. 1, consists of a steel deck plate supported by longitudinal ribs Even for the RD joints with identical weld detail, various
and transverse cross beams. An asphalt surfacing is normally transverse-load cases might cause different crack types. Moreover,
applied onto the steel deck plate to provide a vehicle riding under the same transverse-load case, the RD joints with various
surface. weld details might have different crack types. Therefore, the fatigue
However, because the fatigue design was not considered in the evaluation for RD joints is more complicated than that for weld
design stage of most existing orthotropic steel deck bridges, over- joints of most other kinds.
loading and increases in traffic volume have produced many fatigue The RD joint is not yet covered by the latest fatigue design rec-
cracks in some OSBDs with long service time (Inokuchi et al. 2010; ommendation of Japan (JSSC 2012) and International Institute of
Sim and Uang 2012; Miki 2006; Wolchuk 1990; Kainuma et al. Welding (IIW) guidelines (Hobbacher 2009). In AASHTO LRFD
2016). Fig. 2 presents some common locations of fatigue cracks in bridge design specifications (AASHTO 2012), only crack Types 1
OSBDs. Among these fatigue cracks, those around the rib-to-deck and 3 of the RD joint are included and categorized in Class C for
(RD) joint are most critical because they may lead to asphalt dam- load-induced fatigue, as depicted in Fig. 3.
age or water leaking. Fatigue cracks of four kinds can occur around Because the fatigue data are far from sufficient to evaluate the fa-
RD joints, as presented in Fig. 2: the bending moment in the rib web tigue resistance of RD joints, an economical and high-speed bending
inclines to cracks in the weld throat or rib web, as Type 3 or Type 4; fatigue-test system was developed for use in this study. Two kinds of
RD joint specimens, having weld-penetration ratios of 15 and 75%,
were tested to ascertain the influence of weld penetration on the
1 fatigue-crack type around the RD joint and its fatigue resistance.
Assistant Professor, Dept. of Civil Engineering, Suzhou Univ. of
Science and Technology, Jiangsu 215011, China; Former Researcher,
Dept. of Civil and Earth Resources Engineering, Kyoto Univ., Kyoto 606-
8501, Japan (corresponding author). E-mail: seuliming@gmail.com Fatigue-Test Specimen and Gauge Layout
2
Assistant Professor, Dept. of Civil and Earth Resources Engineering,
Kyoto Univ., Kyoto 606-8501, Japan. In this study, the RD joints were modeled as 150-mm-wide speci-
3
Associate Professor, Dept. of Civil Engineering, Kobe Univ., Kobe mens subjected to three-point bending. Each specimen consisted of
657-0013, Japan. a main plate and a rib box. The main plate refers to the deck plate.
4
Professor, Dept. of Civil and Earth Resources Engineering, Kyoto The rib box refers to the closed longitudinal rib. The test specimen
Univ., Kyoto, 606-8501, Japan.
Note. This manuscript was submitted on November 7, 2016; approved
shapes and main dimensions are shown in Fig. 4. The specimen
on July 27, 2017; published online on November 17, 2017. Discussion pe- dimensions were taken from a bridge of Hanshin Expressway
riod open until April 17, 2018; separate discussions must be submitted for located in Sakurajima Konohana District, Osaka City, Japan. Each
individual papers. This paper is part of the Journal of Bridge specimen was fabricated in the same way: A 6-mm-thick rib was
Engineering, © ASCE, ISSN 1084-0702. welded to a 12-mm-thick steel plate. All ribs and deck plates were

© ASCE 04017128-1 J. Bridge Eng.

J. Bridge Eng., 2018, 23(2): 04017128


Downloaded from ascelibrary.org by Glasgow University Library on 11/17/17. Copyright ASCE. For personal use only; all rights reserved.

Fig. 1. Components of a typical orthotropic steel bridge deck


Fig. 3. Fatigue category of RD joint for load-induced fatigue
(AASHTO 2012)

are the wheel loads applied at the adjacent RD joints. Therefore, the
case with wheel loads riding on the in-box side adjacent RD joint
was chosen as the load case to investigate the crack type and the cor-
responding fatigue resistance of the objective RD joint in this study.
Furthermore, the objective RD joint was located in the midspan in
the longitudinal direction. It is apparent from the transversal influ-
ence lines of stresses around the objective RD joint that the ratio of
stress on the outside surface of the rib web to the stress on the under-
Fig. 2. Potential fatigue-crack types around RD joint surface of deck plate was nearly 1.
In this study, through varying the space of the load point and the
objective joint, C, as shown in Fig. 8, the ratio of the stress on the
made of SM400 steel. The mechanical properties and chemical outer surface of the rib web to the stress on the undersurface of
compositions (Ya 2009) are presented in Table 1. the deck plate around the objective RD joint in the specimens was
The fillet welding was performed in a horizontal position with adjusted to be nearly 1 to simulate the critical-load case for Type 3
carbon dioxide shielded arc welding by self-propelled robot arms and Type 4 fatigue cracks.
controlled by the computer. The key welding parameters were as
follows: tracking accuracy of the electrode = 1 mm; range of voltage
for the electrode = 30–32 V; range of current intensity for the elec- Fatigue-Test Machine and Setup
trode = 300–340 A. The intended levels of weld penetration were 15
For this study, an electric rotation vibrator was used in the fatigue
and 75%, as presented in Fig. 5. For the 15% weld-penetration RD
tests to induce the fatigue load. The fatigue-test setup is portrayed
joints, the self-propelled welding robot arms welded and moved
schematically in Fig. 9. The magnitude and frequency of the cyclic
along the weld line once (moving speed = 350 mm/min). In this
load induced by the rotation vibrator were controlled by adjusting
way, the weld penetration ratio could reach 15%. For the 75% weld-
the built-in electric-mass rotation speed. The magnitude also could
penetration RD joints, the rib web was prepared to make the groove
be changed by adjusting the distance between the rotation axle and
angle between the weld web and the deck plate reach 55°, and then the center of the electric mass. The electric-mass rotation produces
the self-propelled welding robot arms welded and moved along the centrifugal force in all directions. However, only the vertical com-
weld line twice (moving speed = 350 mm/min). In this way, the ponent force is needed for fatigue tests. Therefore, to cancel the
weld-penetration ratio could reach 75%. Finally, the welding seams influence of the horizontal component force on the fatigue-test
were checked by an ultrasonic phased array system with a meas- result, a rigid beam was used to connect the vibrator and the
uring accuracy of 0.2 mm. If the weld-penetration ratio met the reaction-force wall. A pin was used as a joint of one end of the rigid
requirement, the RD specimen was accepted as a qualified speci- beam and the reaction-force wall. The vibrator was bolted to the
men. The test specimens were fabricated and grouped into two se- other end of the rigid beam. In this way, only the cyclic sinusoidal
ries: (a) FD specimens (15% weld penetration) and (b) ED speci- load in the vertical direction was induced for the RD specimens.
mens (75% weld penetration). Compared with a fatigue-test system using a hydraulic jack, the
merits of the fatigue-test system with an electric rotation vibrator
are that the test cost is lower and that its loading frequency is
Boundary Condition and Load Case
higher.
The two ends of the specimen were connected to the steel frame
The location of the objective RD joint in this study is depicted in
using a special holder consisting of two beams and two round bars.
Fig. 6. Based on the results of finite-element (FE) model analysis
A center holder was installed to the load point of the specimen.
(Li et al. 2014), the transversal influence lines of stresses around the
objective RD joint are portrayed in Fig. 7. The FE model was made
using the general-purpose FE software Abaqus. The element type Summary of Fatigue-Test Results
of the FE model was the shell element, and the mesh grid size
around the objective RD joint was approximately 4 mm. It is appa- An outline of fatigue-test results is presented in Table 2 (Ds corre-
rent that the most important load cases for Type 3 and Type 4 cracks sponds to the stress range at the crack-initiation position). Specimen

© ASCE 04017128-2 J. Bridge Eng.

J. Bridge Eng., 2018, 23(2): 04017128


Downloaded from ascelibrary.org by Glasgow University Library on 11/17/17. Copyright ASCE. For personal use only; all rights reserved.

Fig. 4. Specimen dimensions and strain-gauge layout for fatigue testing

Table 1. Properties of SM400 Steel


X
Item Subitem Data Objective RD Joint
Yield strength (MPa) — 281
Ultimate tensile strength (MPa) — 424
Elongation (%) — 31
Chemical composition (%) C 0.16 Fig. 6. Location of objective RD joint
Chemical composition (%) Si 0.14
Chemical composition (%) Mn 0.5
Chemical composition (%) P 0.0011 objective RD joint by the fatigue-test system. It can be concluded
Chemical composition (%) S 0.004 from the test results that the stress ratio can be adjusted to a speci-
fied value during fatigue testing. It can be inferred from the fatigue
tests for the first two specimens that the setup and vibrator were reli-
able and suitable for bending fatigue tests. However, because the
first two fatigue tests were stopped many times for adjustment dur-
ing cyclic-loading processes, the fatigue-test data were unsuitable
for fatigue evaluation and were therefore not used for this study.
Fig. 10 presents a period of the strain history of Gauge 2 on
Specimen ED-4-1.

Initiation Point and Propagation Directions of


Fatigue Cracks
In this study, five specimens with 15% weld-penetration ratio were
all fractured because of root-to-weld-throat cracks (Type 4). The
other five specimens with 75% weld-penetration ratio were all frac-
tured because of toe-to-deck cracks (Type 1) except for one speci-
men that ran out. Fig. 11 presents the fatigue-test system and speci-
mens. It is noteworthy that in this fatigue-test program, a specimen
was defined as having run out if it was not fractured after 107 load
cycles.
Fig. 5. Weld-penetration ratios of specimens
Variation of Stress-Range Ratio and
Fatigue-Crack Propagation
FD-1 was first tested to verify the performance and stability of the
fatigue-test system. Test results show that the fatigue-test system The stress ranges, measured through strain gauges (length of gage =
could consistently induce loading cycles with sinusoidal stress vari- 1 mm; effective temperature range = –196 to 150°C; electric resist-
ation around the objective RD joint. Finally, the specimen fractured ance = 120 X) glued to the bottom surface of the deck plate of the
because of a fatigue crack that initiated from the weld root and specimens with 75% weld-penetration ratio, varied during fatigue
propagated throughout the weld throat. Specimen FD-2-1 was tests. Cyanoacrylate (CN) adhesives were used to attach the strain
tested to verify the adjustability of the stress ratio around the gauges on the specimens. CN adhesive is a single-component

© ASCE 04017128-3 J. Bridge Eng.

J. Bridge Eng., 2018, 23(2): 04017128


Downloaded from ascelibrary.org by Glasgow University Library on 11/17/17. Copyright ASCE. For personal use only; all rights reserved.

Fig. 7. Stress influence lines around the objective RD joint: (a) deck plate; (b) rib web

expressed as the following, can indirectly reflect the location of the


fatigue crack and its propagation:

Ds i
stress  range ratio ¼ (1)
Ds 0

where Ds i = stress range after i cycles of fatigue load; and Ds 0 =


stress range at the beginning of the fatigue test.
An example of variation in the stress-range ratio is shown in
Fig. 12. The stress-range ratio of Gauge 2 first decreased because
the crack gradually cut the flow of the stress in the deck plate, mean-
Fig. 8. Boundary condition for specimen setup ing that the fatigue crack initiated near Gauge 2, as presented in
Fig. 13(a). With the crack growing, the stress-range ratio of Gauge
3 increased because the crack tip was approaching the location of
adhesive for strain gauges. The time required to bond the gauge is Gauge 3, and stress concentration existed near the crack tip, as pre-
extremely short, and handling is very easy. The thin bonding layer sented in Fig. 13(b). After the crack tip passed the location of
allows adhesion to plastic objects and metal. Curing time under nor- Gauge 3, the stress range decreased. The stress-range ratio of Gauge
mal conditions is 20–60 s. Variation of the stress-range ratio, 1 increased because the crack growth consumed the effective

© ASCE 04017128-4 J. Bridge Eng.

J. Bridge Eng., 2018, 23(2): 04017128


thickness of the deck plate. It then decreased because a new crack and the AASHTO S-N curves are depicted in Fig. 15. It is apparent
initiated near Gauge 1. that the fatigue resistances of the specimens with 75% weld penetra-
tion and the specimen with 15% weld penetration were around
AASHTO Class A and Class B, respectively.
Fatigue-Test Results in the S-N Diagram
Determination of the Reference Stress Range and Number of Regression Analysis and S-N Diagram
Load Cycles for Fracture Regression analyses conducted for the test specimens and the
The stress range employed in the S-N diagram was taken at the weld regression analysis are expressed as follows:
toe on the deck plate for specimens with 75% weld-penetration ratio
and weld root for specimens with 15% weld penetration. The meth- logN ¼ c  m  logDs (2)
ods used to determine the stresses for the S-N diagram are presented
in Fig. 14. For the toe-to-deck cracks of specimens with a weld- where N = number of cycles at failure; Ds = stress range in MPa at
Downloaded from ascelibrary.org by Glasgow University Library on 11/17/17. Copyright ASCE. For personal use only; all rights reserved.

penetration ratio of 75%, the calculation was conducted by extrapo- weld toe or root; and m = slope of S-N curves.
lation of the strain gauges glued 5 and 20 mm from the weld toe on Therein, m was set to 3. The runout data were not included in the
the deck plate. For the root-to-throat cracks of specimens with a regression analysis. The regression equations for the specimens
15% weld-penetration ratio, the calculation was also conducted by with 75% weld-penetration ratio and specimens with 15% weld-
extrapolation of the strain gauges glued 5 and 20 mm from the weld penetration ratio are expressed, respectively, as Eqs. (3) and (4):
toe on the rib web. The fatigue life was defined as the number of
load cycles at which the specimen fractured. The fatigue-test results logN ¼ 12:888  3  logDs (3)

logN ¼ 12:492  3  logDs (4)

The regression analysis and fatigue-test results are presented in


the S-N diagram shown in Fig. 16. It can be concluded that the mean
fatigue resistance of specimens with 15% weld penetration was
between JSSC Class C and Class D; the mean fatigue resistance of
specimens with 75% weld penetration was around JSSC Class B.

Comparison with Previous Fatigue Tests

Janss (1988) conducted some fatigue testes at stress ratios of −1 to


−2. Sketches of the specimen and the fatigue-test setup are pre-
sented in Fig. 17. The specimen parameters were the lack of pene-
tration p of the welds and the fit e between the longitudinal edge of
the rib and the deck plate, as presented in Fig. 17. The stress range
Ds at the weld toe on the rib web was used in the S-N diagram. To
compare Janss’s fatigue-test data with those in this study, the stress
ranges Ds at the weld toe on the rib web of the specimens with 15%
weld penetration in this study were extrapolated by the strain
gauges 5 and 20 mm from the weld toe. With m set to 3, the regres-
sion equations for the specimens with 15% weld penetration of this
study, horizontal welded specimens with the fit e = 0–0.5 mm, and
Fig. 9. Sketch of fatigue-test system horizontal welded specimens with the fit e = 2 mm of Janss’s fatigue
tests are shown, respectively, in Eqs. (5)–(7):

Table 2. Summary of Fatigue-Test Results

Specimen PR ratio (%) R ¼ s min =s max Frequency (Hz) Crack type Data Ds (MPa) Number of cycles to failure
FD-1 15 — — Throat crack from root Setup — —
FD-2-1 15 –1 — Throat crack from root Stress ratio — —
FD-5-1 15 –1 12.043 Throat crack from root  128.3 1,682,000
ED-2-1 75 –1 12.122 Deck crack from toe  136.3 2,467,300
ED-5-1 75 –1 — —  95.6 (10,000,000)
ED-5-2 75 –1 12.918 Deck crack from toe  160.9 1,705,700
FD-3-1 15 -1 11.936 Throat crack from root  166.5 989,200
FD-2-2 15 –1 12.016 Throat crack from root  214.2 160,800
FD-3-2 15 –1 12.110 Throat crack from root  153.5 961,200
FD-2-3 15 –1 12.029 Throat crack from root  138.7 1,211,700
ED-4-1 75 –1 12.419 Deck crack from toe  168 1,669,700
ED-2-2 75 –1 12.573 Deck crack from toe  182.5 1,661,800

© ASCE 04017128-5 J. Bridge Eng.

J. Bridge Eng., 2018, 23(2): 04017128


Downloaded from ascelibrary.org by Glasgow University Library on 11/17/17. Copyright ASCE. For personal use only; all rights reserved.

Fig. 10. Strain of Gauge 2 on Specimen ED-4-1

Fig. 11. Fatigue-test system and specimens

logN ¼ 12:462  3  logDs (5) fatigue resistance of the specimens with 15% weld-penetration ratio
in this study was higher than that of the horizontal welded speci-
logN ¼ 12:168  3  logDs (6) mens in Janss’s (1988) fatigue test.
Samol also conducted some fatigue tests, as reported by Ya
logN ¼ 11:910  3  logDs (7) (2009). In his study, the RD joint was fundamentally modeled as a
300-mm-wide test specimen subjected to the cyclic bending stress
The regression analysis and fatigue-test results are presented in the deck plate. Each specimen consisted of a main plate and a rib
in the S-N diagram shown in Fig. 18. The fatigue-test results show web. The specimen dimensions, the fatigue-test setup, and the rota-
that the scatter of the data of specimens with the fit e = 2 mm was tional vibrator are presented schematically in Fig. 19. The speci-
wider than those with the fit e = 0–0.5 mm, perhaps because the uni- mens and the setup differ from those used for this study. Each speci-
formity of the welding quality of the specimens with the fit e = men was welded with a rib web in Samol’s fatigue-test program.
2 mm was lower than the specimens with the fit e = 0–0.5 mm. However, the specimens used in this study, having full closed rib
Moreover, the regression analysis results clarify that the mean boxes, can simulate the influence of the closed rib box on the RD

© ASCE 04017128-6 J. Bridge Eng.

J. Bridge Eng., 2018, 23(2): 04017128


Δσ /Δσ
Downloaded from ascelibrary.org by Glasgow University Library on 11/17/17. Copyright ASCE. For personal use only; all rights reserved.

Fig. 12. Relation between the ratio of stress-range variation and load cycles

Data from the fatigue tests of Samol (Ya 2009) and the fatigue
tests in this study are depicted in the S-N diagram presented in
Fig. 20. In Samol’s fatigue tests, the fatigue resistance of most
specimens was between JSSC Class B and Class C. The fatigue re-
sistance of most specimens tested with the stress ratio R = –1 was in
the upside range. The fatigue resistance of most specimens tested
(a) with the stress ratio R = 0 was in the downside range of the area
between the fatigue resistance Classes B and C in JSSC.
The reason might be the following: Because of the higher resid-
ual stress around the weld line in the specimens, the effective stress
ratio, as measured by strain gauges, was less than R = –1. This low
ratio would be expected to increase the fatigue life of these speci-
mens. Moreover, the fatigue resistance of the specimens in this
study was slightly higher than that the fatigue-test program of
(b) Samol (Ya 2009). Two reasons can be inferred: (1) the fatigue re-
sistance of the weld toe on the deck plate was slightly higher than
that of the weld root on the deck plate; (2) compared with the
fatigue-test setup of Samol’s fatigue-test program, the transverse-
cycle load was eliminated in the fatigue-test setup of this study,
which might have extended the fatigue life.

(c) Discussion

Fig. 13. Fatigue-crack propagation phases in the deck plate: (a) Phase In Janss’s (1988) fatigue-test program, the RD joint specimen con-
1; (b) Phase 2; (c) Phase 3 sisted of a steel plate and a fully closed rib, as in the specimen tested
in this study. The test setup also used a three-point bending system
to simulate the stress state around the RD joint under the critical-
joint in real bridges. In Samol’s fatigue-test setup, one side of the load case in a real orthotropic bridge deck. The fatigue-test results
specimen was fixed to the frame to simulate the cantilever boundary show that the mean fatigue resistance of all specimens with the fit
condition. Therefore, the weld root was closer to the fixed end than e = 0–0.5 mm and a horizontal weld were between the JSSC Class C
the weld toe. For that reason, the setup is ineffective for inducing and Class D. The mean fatigue resistance of all the specimens with
the fatigue crack that initiates from the weld toe and propagates the fit e = 2 mm and a horizontal weld were lower than the JSSC
to the deck plate. However, in this study, the two transversal edges Class E. Therefore, one can infer that decreasing the fit-up between
of the RD joint specimen were supported by roller bars. The fatigue the rib and deck plate can improve the fatigue resistance. All speci-
load was applied near the objective RD joint to simulate the actual mens with a horizontal weld were fractured because of root-to-
stress state around the objective RD joint while the wheel loads throat cracks. All specimens with an overhead weld were fractured
were riding on the adjacent RD joint. Therefore, the type of fatigue because of deck cracks. The likely reason is that the weld-throat
crack and its fatigue resistance could be assessed in this study. thickness of specimens with an overhead weld was larger than that

© ASCE 04017128-7 J. Bridge Eng.

J. Bridge Eng., 2018, 23(2): 04017128


Fig. 14. Linear extrapolation for the reference stress
Downloaded from ascelibrary.org by Glasgow University Library on 11/17/17. Copyright ASCE. For personal use only; all rights reserved.

Fig. 15. Fatigue-test results in S-N curve (AASHTO 2012)

Fig. 16. Fatigue-test results in S-N curve (JSSC 2012)

© ASCE 04017128-8 J. Bridge Eng.

J. Bridge Eng., 2018, 23(2): 04017128


Downloaded from ascelibrary.org by Glasgow University Library on 11/17/17. Copyright ASCE. For personal use only; all rights reserved.

Fig. 17. Specimen and boundary condition of Janss’s fatigue test

Fig. 18. Comparison with Janss’s fatigue-test results

Fig. 19. Specimen and boundary condition of Samol’s fatigue test

© ASCE 04017128-9 J. Bridge Eng.

J. Bridge Eng., 2018, 23(2): 04017128


Downloaded from ascelibrary.org by Glasgow University Library on 11/17/17. Copyright ASCE. For personal use only; all rights reserved.

Fig. 20. Comparison with Samol’s fatigue-test results

of specimens with a horizontal weld. Moreover, the mean fatigue Acknowledgments


resistance of all specimens with a horizontal weld was lower than
the mean fatigue resistance of specimens with 15% weld penetra- The authors appreciate the advice and help of Associate Professor
tion in this study. This might have been caused by different weld Toshiyuki Ishikawa, from the Dept. of Environmental and Urban
methods and materials. Engineering, Kansai University, Japan.
In the fatigue-test program of Samol (Ya 2009), the specimen
consisted of a steel plate and a rib web. The mean fatigue resistance
of these specimens was slightly lower than the mean fatigue resist-
References
ance of the specimens with 75% weld penetration in this study. The
AASHTO. (2012). AASHTO LRFD bridge design specifications,
reason might be that the fatigue resistance of the weld root on the Washington, DC.
deck plate was slightly lower than that of the weld toe on the deck Abaqus [Computer software]. SIMULIA, Providence, RI.
plate and the cycle transverse-load force, which might have Gurney, T. R. (1992). Fatigue of steel bridge decks, HMSO, London.
decreased the fatigue life. Hobbacher, A. (2009). Recommendations for fatigue design of welded joints
and components, Welding Research Council, New York.
Inokuchi, S., Ishii, H., Ishigaki, T., Maneno, H., Sumi, T., and Yamada, K.
Conclusion (2010). “Fatigue assessment for weld between deck plate and u-rib in
orthotropic steel decks with consideration of pavement properties.” J.
Struct. Earthquake Eng., 66(1), 79–91.
It can be inferred from fatigue-test results of this study that for
Janss, J. (1988). “Fatigue of welds in orthotropic bridge deck panels with
the rib-to-deck joint with a 12-mm-thick deck plate and 6-mm- trapezoidal stiffeners.” J. Constr. Steel Res., 9(2), 147–154.
thick rib web, the fatigue resistance and the type of fatigue crack JSSC (Japanese Society of Steel Construction). (2012). Fatigue design rec-
were strongly influenced by the weld-penetration ratio. The mean ommendations for steel structures, Gihodo Shuppan, Tokyo.
fatigue resistance of the 15% weld-penetration specimen was Kainuma, S., Yang, M., Jeong, Y.-S., Inokuchi, S., Kawabata, A., and
between JSSC Class C and Class D. The mean fatigue resistance Uchida, D. (2016). “Experiment on fatigue behavior of rib-to-deck weld
of the 75% weld-penetration specimens was around JSSC B root in orthotropic steel decks.” J. Constr. Steel Res., 119, 113–122.
class. Li, M., Hashimoto, K., and Sugiura, K. (2014). “Influence of asphalt surfac-
ing on fatigue evaluation of rib-to-deck joints in orthotropic steel
In this study, although the RD specimens were tested under the
bridge decks.” J. Bridge Eng., 10.1061/(ASCE)BE.1943-5592.0000610,
critical-load case for Type 3 and 4 cracks, all RD specimens with 04014038.
75% weld penetration fractured because of Type 1 cracks. Miki, C. (2006). “Fatigue damage in orthotropic steel bridge decks and ret-
Therefore, one might conclude that improving the penetration ra- rofit works.” Int. J. Steel Struct., 6(4), 255–267.
tio might prevent Type 3 and 4 cracks. In on-site OSBDs in Japan Sim, H.-B., and Uang, C.-M. (2012). “Stress analyses and parametric study
and various laboratory fatigue experiments, only cracks of Types on full-scale fatigue tests of rib-to-deck welded joints in steel orthotropic
1, 2, and 4 were reported. However, in the AASHTO (2012) decks.” J. Bridge Eng., 10.1061/(ASCE)BE.1943-5592.0000307,
design specification, only Type 1 and 3 cracks are included. The 765–773.
Wolchuk, R. (1990). “Lessons from weld cracks in orthotropic decks on
RD weld joint is categorized as AASHTO Class C. Moreover, the
three European bridges.” J. Struct. Eng., 10.1061/(ASCE)0733
fatigue resistances of the RD specimens in this study were higher -9445(1990)116:1(75), 75–84.
than those of AASHTO Class C. Therefore, more RD specimens Ya, S. (2009). “Fatigue durability evaluations of trough to deck plate
should be tested in future work to assess their fatigue resistance welded details of orthotropic steel deck.” Ph.D. thesis, Nagoya Univ.,
and crack types. Nagoya, Japan.

© ASCE 04017128-10 J. Bridge Eng.

J. Bridge Eng., 2018, 23(2): 04017128

You might also like