You are on page 1of 13

Composite Structures xxx (2010) xxxxxx

Contents lists available at ScienceDirect

Composite Structures
journal homepage: www.elsevier.com/locate/compstruct

Structural performance of composite sandwich bridge decks with hybrid GFRPsteel core
Hyo-Seon Ji a,, Jae-Kwan Byun b, Chang-Soo Lee b, Byung-Jik Son c, Z. John Ma d
a

Department of Civil Engineering, Daewon University College, 599 Sinwol-Dong, Jecheon-Si, Chungbuk-Do, Republic of Korea Department of Civil Engineering, University of Seoul, 90 Jeonnong-Dong, Dongdaemun-Gu, Seoul, Republic of Korea c Department of Civil Engineering, Konyang University, 26 Nae-Dong, Nonsan-Si, Chungnam-Do, Republic of Korea d Department of Civil and Environment Engineering, University of Tennessee, Knoxville, TN 37996, USA
b

a r t i c l e

i n f o

a b s t r a c t
This paper presents the static and fatigue performance of composite sandwich bridge decks with hybrid GFRPsteel core. The composite sandwich bridge deck system is comprised of wrapped hybrid core of GFRP grid and multiple steel box cells with upper and lower GFRP facings. Its structural performance under static loading and fatigue loading with a nominal frequency of 5 Hz was evaluated. The responses from laboratory testing were compared with the ANSYS nite element predictions. The failure mode of the proposed composite sandwich bridge deck was more favourable because of the yielding of the steel tube when compared with that of all-GFRP decks. The ultimate failure of the composite sandwich deck panels occurs by shear of the bonded joints between GFRP facings and steel box cells. Results from fatigue load test indicated no loss in stiffness, no signs of de-bonding and no visible signs of deterioration up to 2 million load cycles. The thickness of the composite sandwich deck retaining the similar stiffness may be decreased to some extent when compared with the all-GFRP deck. This paper also presents design of a connection between composite sandwich deck and steel girder. 2010 Elsevier Ltd. All rights reserved.

Article history: Available online xxxx Keywords: Composite sandwich bridge deck Hybrid GFRPsteel core Structural performance ANSYS

1. Introduction It was recognized that most of structural deciencies of bridges were shown in the bridge deck geometry and deck condition. The bridge deck is a structural component that distributes and transfers loads transversely to the girders that bear on piers or abutments. The most commonly used type for slab-on-girder systems is a cast-in-place concrete deck. Structural deciencies of a bridge deck are related to physical deterioration due to environmental conditions, unanticipated factors like heavy vehicle trafc, fatigue and dynamic responses. They usually cause substantial problems for structural repair and replacement of a concrete deck. For high trafc volume bridges, the speed of construction, especially for the cases of bridge deck repair and replacement, has become a critical issue more than ever. Strong momentum exists for the spread of precast construction for bridges with a push to expand the limits, especially for the use in long-span bridges. One of the promising systems for precast bridge construction has been concretedecked, precast and prestressed concrete (DPPC) girders for superstructure [16]. The DPPC girder bridge does not need a cast-inplace deck, or any wearing course. By having the concrete deck
Corresponding author. Tel.: +82 43 649 3266; fax: +82 43 649 3508.
E-mail address: hsji@mail.daewon.ac.kr (H.-S. Ji). 0263-8223/$ - see front matter 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.compstruct.2010.08.037

as an integral member, several benets are obtained, for instance, an accelerated time frame for construction, a high quality plantproduced concrete deck, and an entire cross section prestressed for enhanced section properties and durability. Despite several major benets, one of the perceived disadvantages with the DPPC girders is its increased weight in long-span bridges. The writers have been working on nding potential solutions including the use of ber-reinforced polymer (FRP) decks replacing the concrete deck in DPPC girder bridges [15]. In recent years the interest in using FRP in construction eld has signicantly increased worldwide. However, the use of FRP panels for bridge decks is still limited, due to the relatively higher cost of the FRP materials than the conventional ones [10,20]. Currently most FRP deck panel systems have the cellular-box geometries fabricated by the pultrusion process [2,9,13,17]. It is believed that innovative designs based on FRP materials could play a great role in extending the application of FRP materials in bridge decks and cutting down the cost. One such design method is to use the hybrid concept. To make the most of FRP materials, combinations of FRP and conventional materials have recently been investigated. Deskovic et al. [7] investigated a hybrid FRPconcrete beam that consists of lament-wound glass ber-reinforced polymer (GFRP) box section combined with a layer of concrete in the compression zone and a thin carbon ber-reinforced polymer (CFRP) laminate in

Please cite this article in press as: Ji H-S et al. Structural performance of composite sandwich bridge decks with hybrid GFRPsteel core. Compos Struct (2010), doi:10.1016/j.compstruct.2010.08.037

H.-S. Ji et al. / Composite Structures xxx (2010) xxxxxx

the tension zone. Burgueno et al. [6] presented a hybrid bridge system with light weight concrete lled circular carbon/epoxy shell girders and an E-glass/polyester deck. They showed that the design is stiffness-driven for the modular composite slab bridge system with the beams of CFRP materials shell type. Lopez-Anido and Han [14] presented the hybrid ber-reinforced polymer-glulam panels for bridge decks. Mufti et al. [19] developed hybrid FRP concrete bridge decks using pultrusion and lament-winding fabrication methods. Van Erp et al. [23] developed hybrid FRPconcrete beams for a bridge application in Australia. The hybrid FRPconcrete beam consists of two parts: concrete in the compression zone and GFRP box section in the tensile zone. Additional CFRP laminate was added to the tensile ange of the box section to increase stiffness of the hybrid beam. A layer of concrete is bonded on the GFRP box section by an epoxy adhesive. The weight of the hybrid FRPconcrete beam was claimed to be about 1/3 of that of a reinforced concrete beam. Aref et al. [4] investigated a hybrid FRPconcrete multi-cell bridge superstructure that consists of GFRP trapezoidal box beams combined with a layer of concrete in the compression zone. As stated above, the research on the hybrid structures has been mainly applied to the combinations of FRP and concrete. Ji et al. [11] proposed the composite sandwich deck system with a hybrid FRPsteel core that has the potential to be used with DPPC girders. The main advantages of composite sandwich panels for bridge decks are intended to not only improve stiffness and buckling response by the continuous support of core elements with the face laminates, but also be cost efcient compared to all-GFRP decks. This paper is focused on the static and fatigue performances of the composite sandwich panels with hybrid FRPsteel core for bridge decks. They were investigated by a combination of experimental and analytical programs. Research has shown that the design of the composite sandwich panels is governed by stiffness and fatigue. This paper also presents the connection design of a composite sandwich deck for a steel I-girder bridge.

2. Experimental program The experimental program was to investigate the static and fatigue performance of the proposed composite sandwich deck system [11] and to verify that the proposed deck system meets the stiffness requirement and has signicant reserve strength. Three different tests in this study were conducted on the composite sandwich deck panels as following: (1) design load test for obtaining exural strength and structural stiffness of the deck; (2) ultimate load test up to failure; and (3) fatigue load test under service load up to 2 million to inspect the reduction in stiffness or strength due to fatigue. 2.1. Specimen description The composite sandwich bridge deck system is comprised of wrapped hybrid core of GFRP grid and multiple steel box cells with upper and lower GFRP facings [11]. Fig. 1 shows the cross section of the composite sandwich deck. The design procedure of the composite sandwich deck prole was summarized in [11].The dimension details of the GFRP grid are shown in Fig. 2. The composite sandwich panel specimen used in this laboratory testing is shown in Fig. 3. The detailed manufacturing process of the composite sandwich panel specimens was described in [11]. 2.2. Properties of the composite sandwich deck system The main constituents of the composite sandwich bridge deck system are bers, resins and steel. Unidirectional E-glass roving, biaxial E-glass roving, continuous strand mat, polyester and vinyl ester resins were selected as the constituents for the GFRP laminates in the composite sandwich panels. The core of steel HSSs in the composite sandwich panel was made of steel, grade KS D3568 steel [18]. The constituents material properties used for the composite sandwich panels are shown in Table 1. Considering

UFL 25.0
6.5

Grid
5.6

25.0
4.0 4.5

CFL + Bonding layer Steel Tube Bonding layer + LFL


4.8 4.5 4.8

116

4.8

4@125 = 500

GFRPGrid
GFRP Facing

Steel Tube
GFRP Facing

Fig. 1. The proposed composite sandwich deck panel (all dimensions in millimeter).

Please cite this article in press as: Ji H-S et al. Structural performance of composite sandwich bridge decks with hybrid GFRPsteel core. Compos Struct (2010), doi:10.1016/j.compstruct.2010.08.037

H.-S. Ji et al. / Composite Structures xxx (2010) xxxxxx

25

Longitudinal

the fact that the vinyl ester resin offers a superior failure elongation which is about twice that of polyester, the vinyl ester resin was used at the bonding layers between the steel box cells and GFRP laminates as shown in Fig. 1. This selection took into account the coefcient of thermal expansion (CTE) mismatch between the GFRP and steel. The properties of the layers in the proposed deck system are shown in Tables 2 and 3. The detailed information about the ber layups and material design was summarized in [11]. 2.3. Experimental setup, instrumentation, and procedure 2.3.1. Static load test The experimental setup for the static load test is shown in Fig. 4. A servo-hydraulic universal testing system was used for the exural tests under the three-point loading. Its capacity was 1000 kN. The effective span length of the composite sandwich panel was 2.7 m, which is a little different from 2.5 m provided by the design manual [18]. The Korean Highway Specications

1 10 00 0

25

07 10

25

6. 5

25 0 10

Transverse

30 07

Table 2 Layer properties of the proposed deck. Ply name UFL UFL LFL CFL Grid core Steel core Bonding layer Ply type CSM Biaxial roving CSM Biaxial roving Uniaxial roving CSM Weight (g/m2) 450 570 450 570 570 450 Ply thickness (mm) 3.2 2.4 3.2 2.4 25.0 1.6 Plies per layer 4.0 2.0 4.0 2.0 25 2.0 Vf 0.194 0.234 0.194 0.234 0.246 0.183

Fig. 2. Plan view of the GFRP grid (all dimensions in millimeter).

Table 3 Layer stiffness properties predicted using micro-mechanics model. Ply name UFL UFL LFL CFL Grid core Steel core Bonding layer Ply orientation Random 0/90 Random 0/90 0 Isotropic Random E1 (GPa) 7.99 19.55 7.99 19.55 20.38 204.00 8.17 E2 (GPa) 7.99 4.35 7.99 4.35 4.42 204.00 8.17 G12 (GPa) 1.69 1.77 1.69 1.77 1.80 76.90 1.73

m12
0.349 0.343 0.349 0.349 0.341 0.300 0.343

Fig. 3. Composite sandwich deck panel.

Table 1 Material properties. Material E-glass ber Polyester resin Vinyl ester resin Steel E (GPa) 72.50 3.38 3.91 205.8 G (GPa) 27.60 1.38 1.43 79.15

m
0.22 0.38 0.37 0.3

q (g/cm3)
2.54 1.24 1.15 7.85

Fig. 4. Experimental setup (static load test). (a) Locations of strain gauges at top. (b) Locations of LVDTs and strain gauges at bottom. (c) Locations of strain gauges through the cross section.

Please cite this article in press as: Ji H-S et al. Structural performance of composite sandwich bridge decks with hybrid GFRPsteel core. Compos Struct (2010), doi:10.1016/j.compstruct.2010.08.037

H.-S. Ji et al. / Composite Structures xxx (2010) xxxxxx

dene a tire contact area. The tire contact area shall be assumed as a rectangle with a length-to-width ratio equal to 1/2.5. Its area shall be determined as 1.416P (mm2), where P is the design wheel load in N. As a result, the length L (in the trafc direction) and width W in mm can be calculated as follows:

p 0:566P;

p 3:541P

Therefore, the size of loading steel patch in the tests was 580 mm 230 mm as shown in Fig. 5a. The load was applied to the deck specimen through a 580 mm 230 mm loading patch to represent a design truck wheel loading DB-24 [18]. A rubber pad with the same dimensions as the steel plate was placed between the deck and the steel plate. A spreader beam was used so as to apply force from the actuator. The load was applied at the center of the span. Two composite sandwich deck specimens were tested. Each test sample was instrumented with both strain gages and displacement transducers. The vertical deections of the specimens were measured using linear variable deection transducers (LVDTs) at three points [D1, D2, D3 in Fig. 5b] at the mid-span and another point [D4 in Fig. 5b] at the quarter span from the right support. Strains were measured using strain gauges at the top face of the specimen [T1, T2, TR1, TR2, TR3, and TR4 in Fig. 5a], at the

bottom face [B1, B2, B3, BR1, BR2, BR3, and BR4 in Fig. 5b], and through a cross section [M1, M2, M3, M4, and M5 in Fig. 5c]. It should be noted that TR and BR gauges mean the 45 strain rosette. The test samples were preloaded up to 100 kN for three cycles at regular intervals. Then they were subjected to service load tests and ultimate load tests. In service load tests, they were tested up to the design wheel load 122.3 kN rst and then to 491 kN due to the 500 kN capacity of the load cell. In ultimate load tests, they were loaded to failure by the setup of 1000 kN capacity to examine the failure progress and strength. 2.3.2. Fatigue load test The experimental setup for the fatigue load test of the composite sandwich deck panel is shown in Fig. 6. A servo-hydraulic universal testing system was used for the fatigue test. A MTS 436 controller was used to control the load range, the loading frequency and the number of cycles of the hydraulic actuator. The composite deck panel in the test had two continuous spans and was supported on one steel beam and two rollers as shown in Fig. 7. The span between the steel beam on the right end and the middle roller is 0.6 m and the other span is 2.1 m. This conguration was intended to get the fatigue responses of the composite sandwich deck panel at the critical sections for both the positive bending moment and negative bending moment. It can also study the fatigue responses of the deck-to-girder connection which is similar to those in actual bridge conditions. Fig. 8 shows a side view of the deck panel before testing with the steel through-bolts attaching the deck to the steel support beam. A square steel tube 150 mm 150 mm 6 mm was additionally used for the deckto-beam connection due to upward displacement. A 600 mm

Loading

TR3

TR1 T1 0.23m

0.58m

TR 4

T2 TR 2
2.7m

(a) Locations of strain gauges at top


D1 BR1 B1 D4 D 2 B3
BR3 0.75m

BR 4 B2 D3 BR 2
2.7m

(b) Locations of LVDTs and strain gauges at bottom

Fig. 6. Experimental setup (fatigue load test).

M1 M2

M3

M4
M5

(c) Locations of strain gauges through the cross section


Fig. 5. Locations of LVDTs and strain gauges. Fig. 7. Schematic of the fatigue load test geometry.

Please cite this article in press as: Ji H-S et al. Structural performance of composite sandwich bridge decks with hybrid GFRPsteel core. Compos Struct (2010), doi:10.1016/j.compstruct.2010.08.037

H.-S. Ji et al. / Composite Structures xxx (2010) xxxxxx

Fig. 8. Side view of composite sandwich deck in fatigue test conguration. (a) Locations of strain gauges at top (units in millimeter). (b) Locations of strain gauges at bottom (units in millimeter). (c) Locations of LVDTs at bottom (units in millimeter).

500 mm box beam was utilized to apply the load at the mid-span. This spreader box beam was supported by the deck via a stack of 50 mm thick steel plates used as loading patches. These rectangular loading patches of a 580 mm 230 mm with the larger dimension transverse to the direction of trafc, were used to simulate the contact area of one DB-24s wheel. The test sample was subjected to fatigue loading for 2000,000 cycles. The load was cycled between the maximum load of 76.5 kN and the minimum load of 7.65 kN at a frequency of 5 Hz under the load control. The loading

range was determined from the rear axle load of the DB-24 truck including the impact factor 1.3, which is 122.4 kN. Since the distribution width of wheel load is 1.6 m, the maximum load for the 1.0 m wide test sample should be 76.5 kN (122.4 kN/1.6). The fatigue load test was stopped periodically for static service load tests up to 122.4 kN, and the composite sandwich deck panel was inspected for signs of damage periodically as well. The static service load tests were conducted on the test sample after 500,000, 1000,000, 1500,000 and 2000,000 cycles during the fatigue load test. The test sample was instrumented with both strain gages and displacement transducers as shown in Fig. 9. The vertical deections of the deck were measured using linear variable deection transducers (LVDTs) at three points D1, D2, D3 along the midsection of the right span, D4 at the quarter-section of the right span and D5 at the mid-section of the left span as shown in Fig. 9c. Strains were measured using strain gauges at the top face of the specimen (T1, T2, T3, T4, TL1, TL2, TN1, TN2, and TN3 in Fig. 9a) and at the bottom face (B1, B2, B3, B4, BL1, BL2, BL3, BN1, BN2, and BN3 in Fig. 9b). The BL3 point in Fig. 9b shows the location of the strain gauge mounted on the bottom surface of the steel tube. The deections and strains for the test sample were recorded during the test using a high-speed data acquisition system. 3. Results and discussion 3.1. Stiffness The displacement at the mid-span of the deck panel was 5.65 mm at the service load 122.3 kN in the rst test, which was

TN1

T1 T2 Loading

TN 2

TL1

500 T3

200 TL 2

TN 3 600 1050

T4
1050

(a) Locations of strain gauges at top (units in mm)


BN1

B1 B2
BL3Po int

BN 2

BL1

BL3 B3

BL 2

600

BN 3

1050

B4

1050

(b) Locations of strain gauges at bottom (units in mm)


D1

D5

D4

D2
D3

600

1050

1050

(c) Locations of LVDTs at bottom (units in mm)


Fig. 9. Locations of LVDTs and strain gauges.

Please cite this article in press as: Ji H-S et al. Structural performance of composite sandwich bridge decks with hybrid GFRPsteel core. Compos Struct (2010), doi:10.1016/j.compstruct.2010.08.037

H.-S. Ji et al. / Composite Structures xxx (2010) xxxxxx

about L/478 for the span of 2.7 m. This was slightly less than the deection limit of L/425 of 6.35 mm. The loaddisplacement curves of the measuring points are shown in Fig. 10. The panel exhibited a fairly linear behavior up to 122.3 kN. The stiffness of the deck panel was calculated to be 16.7 kN/mm. The predicted deection value of the steel deck alone at the service load of 122.3 kN was compared with the experimental deection value of the composite sandwich deck. The maximum deection at mid-span was calculated to be 6.77 mm, which is 19.8% higher than the tested deection of 5.65 mm. The stiffness of composite sandwich deck has slightly increased when compared to the steel deck with tubes alone. Some audible noise was detected at 150 kN. The mid-span bending strains from the top and bottom faces were compared for the panel at a load of 491 kN in Fig. 11.

For example, the top surface longitudinal strain (X direction) at Point T1, located at 135 mm away from the loading patch in Fig. 5a, was 1006le in compression when the load was 122.3 kN, and the strain at Point B1 was 911le at the same load level. These strains are less than 20% of the maximum strain from the FRP coupon test. Please note that, in Fig. 11, the notations TR1 ex, TR1 e1 and TR1 e2 are used to denote the strains measured at the strain rosette point TR1 in the span direction and two principal directions. The strains BR1 ex, BR1 e1 and BR1 e2 are similar except that they were measured at the bottom facing of the specimen. From Fig. 11, TR1 e1 and TR1 e2 show 3000le and 1250le, respectively, at the load of 400 kN, whereas BR1 e1 and BR1 e2 are 2960le and 1250le at the same load level. The calculated principal stress was 23.33 MPa in compression at the top facing and 75.08 MPa in tension at the bottom facing. The safety factor of the stresses against the ultimate strength varied from 2.8 to 8.8. 3.2. Failure mode The failure mode of the sandwich panel was studied by the ultimate load test. The loaddisplacement curves up to 632.5 kN are shown in Fig. 12. As load was increased beyond 400 kN, the loaddisplacement curves became nonlinear. A maximum load of 632.5 kN was obtained. The maximum measured displacement was 92.3 mm. When the load of 632.5 kN was reached, a very strong and loud noise was heard due to the de-bonding of GFRP laminates from the wrapped GFRP grid and steel tube core. A permanent displacement was observed beyond the load of 632.5 kN as shown in Fig. 12. The deck panel did not return to its original shape after releasing of the load unlike the cellular all-GFRP deck panels. The permanent displacement of the composite sandwich panel was due to the yielding of the steel tube core, and the failure of the bers. Fig. 13 shows the failure of the specimen. The specimen failed due to the delamination of the GFRP grid from the steel tube core instead of any local damage around the loading patch on the top facing. This failure mode was different from the local punching failure mode or other local damages of the cellular all-GFRP panels. The steel tube core yielding created the desired ductility in the

Fig. 10. Loaddisplacement response(s).

Fig. 11. The loadstrain responses from design load test. (Note: ex = strain in x-direction; e1 = the maximum principal strain; e2 = the minimum principal strain.) (a) At top facing and (b) at bottom facing.

Please cite this article in press as: Ji H-S et al. Structural performance of composite sandwich bridge decks with hybrid GFRPsteel core. Compos Struct (2010), doi:10.1016/j.compstruct.2010.08.037

H.-S. Ji et al. / Composite Structures xxx (2010) xxxxxx

Fig. 12. The loaddisplacement responses from ultimate load test.

Fig. 14. Strain gage data at a cross-section of the panel.

hybrid sandwich panel. The ultimate failure of the panel occurred because of the shear failure of the bonding layer as was investigated by Teixeira de Freitas et al. [22]. They showed that the ultimate failure of the composite bonded systems occurred by shear of the adhesive layer. A further study should be conducted to investigate the shear behavior of the bonding layer due to the inuence of different thickness in the proposed deck. Fig. 14 shows the load strain curves for the strain measuring points M1M5 indicated in Fig. 5c. From the test results shown in Fig. 15, the location of the neutral axis was found at 94.99 mm from the bottom face of the panel at the service load level. 3.3. Stiffness degradation The exural fatigue load test was conducted to evaluate the fatigue performances at the critical sections within the two spans and the deck-to-girder connection. The loaddeection curves at several deck locations and loadstrain curves during the static tests after 0, 0.5, 1.0, 1.5 and 2.0 million cycles were shown in Fig. 16 and 17. The deection and strain responses remained almost the same for all of the static load tests, and the composite sandwich deck showed no apparent changes in stiffness of the structure. The residual deection in the composite sandwich deck panel under the fatigue loading was found to be insignicant as shown in Fig. 18. Damage due to the fatigue loading was also

Fig. 15. Strain gage data at a section of neutral axis.

inspected at the area of the deck panel around the bolted connection. No visible signs of deck panel deterioration or de-bonding

Fig. 13. Composite sandwich panel at a failure load.

Please cite this article in press as: Ji H-S et al. Structural performance of composite sandwich bridge decks with hybrid GFRPsteel core. Compos Struct (2010), doi:10.1016/j.compstruct.2010.08.037

H.-S. Ji et al. / Composite Structures xxx (2010) xxxxxx

Fig. 16. Static deections after different fatigue cyclings (D2 point).

Fig. 18. Mid-span deection history during fatigue test.

Table 4 Deections and strains from interrupted fatigue cycling (load = 122.4 kN). Load cycling (cycles) Mid-span vertical deection (mm) Left centerline D1 3.198 2.651 2.689 2.689 2.670 2.708 Centerline D2 4.441 3.967 4.015 4.034 3.967 4.015 Right centerline D3 2.955 2.541 2.591 2.620 2.561 2.581 Mid-span transverse strain (microstrain) Top facing T3 571 548 556 555 554 554 Bottom facing B3 201 207 207 215 206 206

0 440,000 800,000 1200,000 1600,000 2000,000

Table 5 Comparison of strains (load = 122.4 kN). Load cycling (cycles)

and

strengths

from

interrupted

fatigue

cycling

Fig. 17. Deck strains after different fatigue cyclings at bottom facing (B3 point). 0 440,000 800,000 1200,000 1600,000 2000,000 Design limits
a b

Bottom facing centerline (FRP composites) Mid-span transverse strain (microstrain) 490 462 475 471 470 465 3930a

Bottom facing centerline (steel tube)

Mid-span transverse strain (microstrain) 383 363 371 371 383 379

Mid-span transverse stress (MPa) 78.8 74.7 76.4 76.4 78.8 77.9 130b

between the GFRP grid core and steel multi-cell tube core were observed. A comparison of deection and strain measurements between the static load test without any previous fatigue cycles and the static load test conducted after 2000,000 fatigue cycles are shown in Table 4. The deection at D2 for the test after fatigue cycling was 15.88 mm and the maximum strain in the transverse direction measured at the bottom facing was 477le. And the strain of the steel tube measured at the same location was 379le as shown in Table 5. The allowable strain of the FRP composites is 20% of the maximum strain from the FRP coupon test, 3930le and the allowable stress of the steel is 130 MPa as summarized in [11]. The limit normal strain not to cause the ber de-bonding in the laminates made with chopped strand mat (CSM), woven roving (WR) and uni-directional (UD) glass reinforcement may be estimated as 1000le after 1000,000 fatigue cycles as provided in the

The allowable strain = 20% of failure strain of the FRP coupon test. The allowable stress is provided by the Korean Highway Specications [18].

handbook [8]. The tensile strain in the FRP composites and steel members of the composite sandwich deck panel were found to be well below the design allowable strain. The value of EI calculated is 8:63 106 N m2 after the complete fatigue test. The exural rigidity after the fatigue test was reduced about 3.5% when compared to that calculated from displacements measured in the design load test.

Please cite this article in press as: Ji H-S et al. Structural performance of composite sandwich bridge decks with hybrid GFRPsteel core. Compos Struct (2010), doi:10.1016/j.compstruct.2010.08.037

H.-S. Ji et al. / Composite Structures xxx (2010) xxxxxx

4. Finite element analysis 4.1. FE model FEM procedures have been successfully employed in research studying the performance of FRP bridge decks or their components [24]. A FE analysis using software ANSYS [3] was carried out for the test specimen. The overall FE model for the static load tested specimen is shown in Fig. 19. The boundary conditions were modeled as a pin and a roller for the end supports. Displacement beyond a load of 400 kN became nonlinear. This load level was predicted when the steel tube core yielded, as shown in Table 6. The comparisons of displacement and strain results between experiment and analysis are shown in Table 8. The overall FE model for the fatigue load tested specimen is shown in Fig. 20. The boundary conditions of the deck in fatigue load test were modeled as one span with a xed support and a roller, while the adjacent other span was simply supported. The steel support girder in fatigue test was modeled by three dimensional beam elements. The interface between the deck and steel support girder was modeled by rigid elements in order to transmit the load and displacement from the deck to the main girder. The boundary conditions at the nodes of the hybrid FRPsteel decks lower facing and steel support girders top ange were modeled as constrained displacement in the vertical direction. 4.2. Material and stiffness properties and boundary conditions The layers properties as described in Section 2.2 were used for the FE analysis of the test specimens in this study. The boundary conditions in static load test were modeled as constrained

Table 6 Stresses for the composite sandwich deck (load = 400 kN). Item GFRP (Experiment) Top facing (fxa) Stress (MPa) Ultimate strength (MPa) Safety factor 65.10 255.35b 3.92 Bottom facing (fxa) 23.65 210.03c 8.88 Steel (ANSYS) Bottom facing 244 248.4d 1.02

Safety factor = stress/ultimate strength. a Longitudinal direction of deck (perpendicular to bridge axis). b Yield point of GFRP laminate (0/90). c Yield point of GFRP laminate (chopped strand mat). d Yield point of steel.

Table 7 Comparison of deections and strains (load = 122.4 kN). Load cycling (cycles) Mid-span vertical deection (mm) Left centerline D1 0 2000,000 ANSYS 3.198 2.708 0.757 Centerline D2 4.441 4.015 1.955 Right centerline D3 2.955 2.581 0.757 Mid-span transverse strain (microstrain) Top facing T3 571 554 248 Bottom facing B3 201 206 177

displacement in the vertical direction at nodes of the composite sandwich deck panel lower facing and steel support girder top ange of testing. The boundary conditions in fatigue load test were modeled as one span with a xed support and roller, while the

CFL

LFL

Steel Tube
Grid

UFL

Fig. 19. FE model of the specimen for the static load test (ANSYS).

Please cite this article in press as: Ji H-S et al. Structural performance of composite sandwich bridge decks with hybrid GFRPsteel core. Compos Struct (2010), doi:10.1016/j.compstruct.2010.08.037

10 Table 8 Data of all-GFRP deck for comparison. Deck system Superdeck [2] DuraSpan [17] Proposed deck Depth (mm) 203 203 164 Weight (kg/m2) 107 90 177 Deection reported L/530 L/450 L/478

H.-S. Ji et al. / Composite Structures xxx (2010) xxxxxx

Flexural rigiditya Nm2 106 9.83 4.28 8.95

a Normalized exural rigidity to DB-24 loading [18] for a 2.7 m center-to-center distance between supporting girders.

adjacent other span simply supported. The steel support girder in fatigue test was modeled by three dimensional beam elements. 4.3. Results of analysis The predicted deection as shown in Table 7 at the design load level of 122.3 kN was 1.955 mm, while the tested deection was 4.44 mm. These deection values are less than the design limit of 6.35 mm. The predicted deection at D2 is 55.9% lower than the tested average deection. But predicted strain at mid-span (B3) is 11.9% lower than the tested strain. The comparisons of deection and strain results from FEA and test data at the load 122.4 kN (27.4 kip) are shown in Table 8. The predicted deection at the design load level of 122.3 kN (27.4 kip) was 1.955 mm, while the tested deection is 4.44 mm (0.17 in.). These deection values are less than the design limit of 6.35 mm (0.25 in.). The predicted deection at center is 55.9% lower than the tested average deection. But predicted strain at mid-span (B3) is 11.9% lower than the tested strain. As is pointed out by Wan et al. [24], these discrepancies between the responses from laboratory testing and the FEA predictions are justied as follows. Each monitoring point on the test specimen was instrumented with both one displacement transducer and strain gages. It is not possible to place these instruments at the monitoring point of the specimen as assumed in the FE model. Displacement beyond a load of 400 kN (89.69 kip) became nonlinear. The predicted yield load was considered as 400 kN (89.6 kip). The laminated facing and the steel tubes top ange was the parts that were particularly investigated since the deck failed in these components. The normal stress contours for

the laminated facing and the steel tubes top ange are shown in Fig. 21 and 22. As shown in Fig. 21, the maximum stress in the hybrid FRPsteel deck panel is 371.309 MPa. At the load of 400 kN, The induced stresses within the steel cells from FE analysis were smaller than those of the strength limits. The proposed deck system has a safety factor of 3.928.8 against the material failure. The location of the neutral bending axis by the FE analysis was 90.36 mm (3.56 in.) from the bottom face of the panel. The location of the neutral bending axis by testing result was found at 94.99 mm from the bottom face of the panel at the service load level. Fig. 23 shows the contours of deection, strain and stress for the model of the deck in the fatigue test. As shown in Fig. 23, the curvature slope of the deection contour is very inclined in the transverse direction. As is pointed out by Zi et al. [25,26], it could be conrmed that the contours nonlinearity in the transverse direction is due to the shear deformation of the rectangular core type. An additional research of the composite sandwich deck with the foam-lled GFRP grid could be conducted in the further research to improve such a contours nonlinearity in the transverse direction. The comparisons of deection and strain results from FEA and test data at the load 122.4 kN are shown in Table 7. The predicted deection at the design load level of 122.3 kN for the span of fatigue load test was 1.955 mm, while the tested deection is 4.44 mm. These deection values are less than the design limit of 6.35 mm. The predicted deection at center is 55.9% lower than the tested average deection. But predicted strain at mid-span (B3) is 11.9% lower than the tested strain. 5. Performance comparisons with all-GFRP decks The results of three different tests conducted on the proposed composite sandwich panel were investigated. The failure load of 632.5 kN was ve times larger than the design wheel load of 122.3 kN as specied in the specication [18]. The mid-span deections of the deck sample at the design wheel load were 5.65 mm and 4.02 mm for the static test and the post-fatigue load test respectively. A deection-to-span ratio of the deck sample was about L/478, which was less than the span design limitation of L/ 425 of 6.35 mm as specied in the specication [1,18]. This indicates good structural performance of the deck as far as strength is concerned. In other words, the deections will control the design

Fig. 20. FE model of the specimen for the fatigue load test (ANSYS).

Please cite this article in press as: Ji H-S et al. Structural performance of composite sandwich bridge decks with hybrid GFRPsteel core. Compos Struct (2010), doi:10.1016/j.compstruct.2010.08.037

H.-S. Ji et al. / Composite Structures xxx (2010) xxxxxx

11

Fig. 21. Stress contour in bottom ange of steel tube of composite sandwich deck panel (load = 400 kN).

Fig. 22. Stress contour in top ange of steel tube of composite sandwich deck panel (load = 400 kN).

of this composite sandwich panel bridge deck system. From the experimental observations and failure mode analysis, it can be concluded that the failure mode of the proposed hybrid deck was more favourable because of the yielding of the steel tube when compared with that of all-GFRP decks. The structural performance of the test sample with regards to the design criteria showed that the deck had sufcient stiffness against the design wheel load. No reduction in stiffness or strength was found after 2000,000 cycles of fatigue load in the service load condition. The proposed composite sandwich deck met the service load criteria after fatigue cycling. For the comparison purpose, the exural rigidity, depth, and deection of the pultruded all-GFRP deck made by creative pultrusions (CP), and MMC were considered [2]. Table 8 shows the comparisons of the commercial decks with the proposed composite sandwich deck. Normalized exural rigidity to DB-24 loading for a 2.7 m center-to-center distance between supporting girders was calculated without considering shear deformation. Comparing

with those all-GFRP decks, the thickness of the proposed deck can be decreased by 19%. The minimum depth [tmin (ft)] of a reinforced concrete bridge deck is stated in Eq. (2), where the span is in feet [5].

tmin

1:2span 10 30

The depth of the tested deck panel was 164.4 mm, which is larger than the minimum thickness of a reinforced concrete deck used at the span length of 30 m. The required minimum depth for a RC deck from Eq. (2) in this case is 110.2 mm. The estimated cost of the available all-GFRP decks in the United States is about $700 807/m2 [5], whereas the estimated cost of the proposed deck is about $375/m2. The cost of constructing concrete decks is about $200/m2 in Korea. This deck system may be used for a deck replacement and repair projects of the bridges having a high trafc volume to save the cost.

Please cite this article in press as: Ji H-S et al. Structural performance of composite sandwich bridge decks with hybrid GFRPsteel core. Compos Struct (2010), doi:10.1016/j.compstruct.2010.08.037

12

H.-S. Ji et al. / Composite Structures xxx (2010) xxxxxx

Fig. 23. Displacement contour of composite sandwich deck panel in fatigue load test (Z-direction).

Grout Grid

Polystrene Haunch

ShearStud

CompositeSandwichPanel

SteelGirder

Fig. 24. Schematic of proposed composite sandwich deck-to-girder connection.

6. Connection between deck and girder The connection of FRP decks to steel or concrete main girders remains a challenge. The connection of FRP decks to main girders will be focused on reducing construction time and ensuring composite action between decks and girders. Currently, Keller et al. [12,21] investigated adhesively bonded joints between pultruded FRP decks and steel girders. The proposed connection system between the composite sandwich deck and a steel girder was given in Fig. 24. The components of the proposed deck connection consist of the module of all-GFRP grids and steel shear stud. After the steel girders are placed, polystyrene haunches are attached to the edge of the top anges of the steel girders with an epoxy adhesive. In a composite bridge, the haunch height will vary due to the super-elevation of the roadway geometry. The modular hybrid deck panels with longer connection grid are then placed on the foam. The surface of the longer connection grid of the modular hybrid deck panels will be coated by a resin sand to adhere the grout

or polymer mortar to the horizontal surfaces of the top anges of the steel girders. This method will be a design method for hybrid decked steel girders with exible hybrid shear connections including conventional steel studs and FRP grids. After the installation of all of the composite sandwich deck panels, the non-shrinkage grout or polymer mortar will be placed up to the bottom of the longer connection grid. These deck-to-girder connections should be capable of sustaining the required number of loading cycles with no loss in structural capacity. The modular composite sandwich deck panels are light enough for two construction workers to carry. The assessments of connection behaviours between the composite sandwich deck panels and girders will be studied in the future based on the investigations in [12,21]. 7. Conclusions The performance of the proposed hybrid deck panel for highway bridges was evaluated using an experimental approach. Based on

Please cite this article in press as: Ji H-S et al. Structural performance of composite sandwich bridge decks with hybrid GFRPsteel core. Compos Struct (2010), doi:10.1016/j.compstruct.2010.08.037

H.-S. Ji et al. / Composite Structures xxx (2010) xxxxxx

13

the results of the investigations presented in this paper, following conclusions are summarized: (1) The proposed hybrid FRPsteel core for bridge decks was able to not only improve stiffness and buckling response, but also be cost efcient compared to all-GFRP decks. Besides, it was also able to reduce the weight of all-steel decks. This paper shows a combination of FRP and conventional materials like steels that can fully take advantage of each materials strengths and characteristics. (2) The thickness of the hybrid deck may be decreased by 19% when compared with the all-GFRP decks with the same exural rigidity. A deection-to-span ratio of the hybrid deck, L/ 478 was less than the span design limitation of L/425 as specied in the specication. (3) The failure mode of the proposed hybrid deck was more favourable because of the yielding of the steel tubes when compared with that of all-GFRP decks. (4) The estimated cost of the available all-GFRP decks in the United States is about $700807/m2, whereas the estimated cost of the proposed deck is about $375/m2. In Korea, the cost of constructing concrete decks is about $200/m2. The proposed deck system may be used for a deck replacement and repair projects having a high trafc volume to save the cost. (5) No reduction in stiffness or strength was found up to 2000,000 cycles of fatigue load in the service load condition. The proposed composite sandwich deck met the service load criteria after fatigue cycling. (6) The assessments of connection behaviours between the composite sandwich deck panels and girders will be studied in the future.

Acknowledgments The research described in this paper was conducted with the support from Seong-won Construction Ltd., the University of Seoul, and the National Science Foundation under Grant No. 0550899 awarded to the fth author. The authors deeply thank for the assistance. Any opinions, ndings, and conclusions or recommendations expressed in this paper are those of the authors, and do not necessarily reect the views of the sponsors. References
[1] AASHTO. AASHTO LRFD bridge design specications, 4th ed. Washington (DC); 2007.

[2] Alagusundaramoorthy P, Harik IE, Choo CC. Structural behavior of FRP composite bridge deck panels. J Bridge Eng, ASCE 2006;11(4):38493. [3] ANSYS 10.0: documentation ANSYS Inc. distributed with ANSYS 10.0; 2005. [4] Aref AJ, Kitan Y, Lee GC. Analysis of hybrid FRPconcrete multi-cell bridge superstructures. Compos Struct 2005;69:34659. [5] Bakis CE, Bank LC, Brown VL, Cosenza E, Davalos JF, Lesko JJ, et al. Fiberreinforced polymer composites for construction state-of-the art review. J Bridge Eng, ASCE 2002;6(2):7387. [6] Burgueno R, Karbhari VM, Seible F, Kolozs RT. Experimental dynamic characterization of an FRP composite bridge superstructure assembly. Compos Struct 2001;54:42744. [7] Deskovic N, Triantallout, Meier U. Innovative design of FRP combined with concrete: short-term behavior. J Struct Eng, ASCE 1995;121(7):106978. [8] European structural polymeric composites group (EUROCOM). In: Clarke JL, editor. Structural design of polymer composites EUROCOMP design code and handbook. UK: E&FN Spon, Ltd.;1996. [9] Federal Highway Administration (FHWA). FRP decks and superstructures: current practice; 2002. <http://www.fhwa.dot.gov/bridge/frp/deckprac.htm>. [10] Ji HS, Son BJ, Chang SY. Field testing and capacity-ratings of advanced composite materials short-span bridge superstructure. Compos Struct 2007;78(2):299307. [11] Ji HS, Son BJ, Ma ZJ. Evaluation of composite sandwich bridge decks with hybrid FRPsteel core. J Bridge Eng, ASCE 2009;14(1):3644. [12] Keller T, Gurtler H. In-plane compression and shear performance of FRP bridge decks acting as top chord of bridge girders. Compos Struct 2006;72:15162. [13] Kim HY, Hwang YK, Park KT, Lee YH, Kim SM. Fiber reinforced plastic deck prole for I-girder bridges. Compos Struct 2005;67:4116. [14] Lopez-Anido R, Han X. Structural characterization of hybrid ber-reinforced polymer glulan panels for bridge decks. Journal of Composites for Construction, ASCE 2002;6:194203. [15] Ma ZJ, Choppali U, Li L. Cycling tests of a ber-reinforced polymer honeycomb sandwich deck panel at very cold temperatures. International Journal of Materials and Product Technology 2007;28(1/2):17897. [16] Ma ZJ, Chaudhury S, Millam J, Hulsey JL. Field test and 3D FE modeling of decked bulb-tee bridges. Journal of Bridge Engineering, ASCE 2007;12(3):30614. [17] Martin Marietta Composites (MMC), Ltd. DuraSpan ber-reinforced polymer bridge deck system; 2005. <http://www.martinmarietta.com/products>. [18] Ministry of Construction and Transportation (MOCT). Standards specications for highway bridge, 2nd ed., Korea; 2000. [19] Mufti AA, Labossiere P, Neale KW. Recent bridge applications of FRPs in Canada. Struct Eng Int 2002;12(2):968. [20] Nystrom HE, Watkins SE, Nanni A. Financial viability of ber-reinforced polymer bridges. J Manage Eng, ASCE 2003;19(1):28. [21] Schollmayer M, Keller T. Modeling of through-thickness stress state in adhesive joints connecting pulturded FRP bridge decks and steel girders. Compos Struct 2009;90:6775. [22] Teixeira de Freitas S, Kolstein H, Bijlaad F. Composite bonded systems for renovations of orthotropic steel bridge decks. Composite Structures 2010;92: 85362. [23] Van Erp G, Heldt T, McCormick L, Carter D, Tranberg C. An Australian approach to bre composite bridges, bre composites design and development. Toowoomba, Australia: University of Southern Queensland. <http://www.fcdd. com.au:2002>. [24] Wan B, Rizos DC, Petrou MF, Harries KA. Computer simulations and parametric studies of GFRP bridge deck systems. Compos Struct 2005;69:10315. [25] Zi G, Kim BM, Hwang YK, Lee YH. An experimental study on static behavior of a GFRP bridge deck lled with a polyurethane foam. Compos Struct 2008;82: 25768. [26] Zi G, Kim BM, Hwang YK, Lee YH. The static behavior of a modular foam lled GFRP bridge deck with a strong web-ange joint. Compos Struct 2008;85: 15563.

Please cite this article in press as: Ji H-S et al. Structural performance of composite sandwich bridge decks with hybrid GFRPsteel core. Compos Struct (2010), doi:10.1016/j.compstruct.2010.08.037

You might also like