B7-mediated costimulation and the immune response

Lightweight UHPC-FRP Composite Deck System

Downloaded from ascelibrary.org by University of California, San Diego on 04/13/17. Copyright ASCE. For personal use only; all rights reserved.

Munaf A. Al-Ramahee, S.M.ASCE1; Titchenda Chan, S.M.ASCE2; Kevin R. Mackie, M.ASCE3; Sahar Ghasemi, S.M.ASCE4; and Amir Mirmiran, F.ASCE5 Abstract: A new lightweight, low-profile composite deck system is presented that was fabricated using vacuum-assisted resin transfer molding. The deck in this system is composed of ultrahigh-performance concrete (UHPC) for high compressive strength and durability, carbon-fiber-reinforced polymer (CFRP) for high tensile strength, and glass-fiber-reinforced polymer (GFRP) for shear resistance. The highperformance modular deck panels are beneficial for applications such as movable bridges, where spans may be mechanically lifted, and deck repair or replacement, where transportation and installation assist in accelerated bridge construction. Seven single-unit deck specimens were fabricated with two different spans (1,220 and 864 mm) and tested under two different flexural load configurations (wheel-tire load and fourpoint load). The experimental results demonstrated that the target recommended strength demand for this deck was satisfied prior to failure as a result of delamination of the top UHPC plate from the core. Analysis of the results and corresponding finite-element models (FEMs) showed that the load-transfer mechanism was limited by the crushing of the UHPC plate and interfacial bond with the core. DOI: 10.1061/(ASCE) BE.1943-5592.0001049. © 2017 American Society of Civil Engineers. Author keywords: Bridge; Bond; Composite; Finite-element analysis; Resin infusion; Ultrahigh-performance concrete.

Introduction According to the Federal Highway Administration (FHWA 2015), of the over 611,000 bridges in the United States, approximately 10% are structurally deficient, and another 14% are functionally obsolete. Three out of four of the structurally deficient bridges have major problems with their decks. In addition to structural deficiency, the geometry and weight of the deck often limit both the load rating and the bridge’s functionality. The service life of these bridges could be extended by replacing their decks (Mirmiran et al. 2015). However, traditional construction methods and deck systems (cast-in-place concrete bridge deck) are usually time-consuming and lead to major traffic delays (Keller et al. 2014; Manalo et al. 2016) that may limit the interest in replacing or widening a bridge or the ability to do so, especially in urban areas. Therefore, there is a need for deck systems that satisfy functionality requirements, can be used with existing substructures, and support accelerated bridge construction (ABC) techniques.

1 Ph.D. Candidate, Civil, Environmental, and Construction Engineering, Univ. of Central Florida, Orlando, FL 32816-2450; Lecturer, Civil Dept., College of Engineering, Univ. of Al-Qadisiyah, Al-Qadisiyah 58002, Iraq. E-mail: [email protected] 2 Graduate Student, Civil, Environmental, and Construction Engineering, Univ. of Central Florida, Orlando, FL 32816-2450. E-mail: titchenda.chan@ knights.ucf.edu 3 Associate Professor, Civil, Environmental, and Construction Engineering, Univ. of Central Florida, Orlando, FL 32816-2450 (corresponding author). E-mail: [email protected] 4 Ph.D. Candidate, Civil and Environmental Engineering, Florida International Univ., Miami, FL 33174. E-mail: sghas006@fiu.edu 5 Professor, Provost, and Vice President for Academic Affairs, Univ. of Texas at Tyler, Tyler, TX 75799. E-mail: [email protected] Note. This manuscript was submitted on August 29, 2016; approved on January 20, 2017; published online on April 12, 2017. Discussion period open until September 12, 2017; separate discussions must be submitted for individual papers. This paper is part of the Journal of Bridge Engineering, © ASCE, ISSN 1084-0702.

© ASCE

Ultrahigh-performance concrete (UHPC) is an advanced structural composite material with high strength, ductility, and durability. The recent use of UHPC in bridge applications has proved efficient and economical (Aaleti and Sritharan 2014). The durability and wearing resistance of UHPC make it ideal to serve the dual purpose of the structural slab and integral riding surface (Shann 2012). One such application was a new lightweight UHPC waffle slab deck with high-strength steel (HSS) rebars that included a solid riding surface developed by Saleem et al. (2011). They proved the efficiency of a 127-mm-thick UHPC waffle deck with No. 22 HSS bars through a series of experimental tests with different configurations. This deck system was further improved and optimized by Ghasemi et al. (2016b) by reducing its overall depth to 114 mm with No. 16 HSS bars while still meeting the strength and serviceability requirements. Fiber-reinforced polymer (FRP) composite materials have also found increasing application in the repair and rehabilitation of existing bridges and to some extent in new bridge construction. FRP composites typically exhibit such desirable properties as high strength, high stiffness-to-weight ratio, long-term durability, fatigue resistance, and good environmental resistance (Zhang et al. 2006), depending on the fiber and matrix materials employed. Several all-FRP bridge decks constructed in the United States have been studied (Lopez-Anido et al. 1997; Karbhari and Seible 2000; Lee et al. 2007). Ehlen (1999) examined the lifecycle cost-effectiveness of FRP bridge decks compared with conventional ones. One of the three FRP decks was found to lower lifecycle costs in high-traffic overpasses by lowering construction time to offset the higher initial cost. In addition, many FRP sandwich-panel and FRP hybrid decks have been constructed, and their use as lightweight decks in widening projects or new construction has been studied (Alagusundaramoorthy et al. 2006; Keller et al. 2007; Robinson and Kosmatka 2008; Camata and Shing 2010; Tuwair et al. 2016). Different failure modes have been observed in previously tested FRP composite and sandwich-deck systems. Delamination of the top and bottom faces (Alagusundaramoorthy et al. 2006; Tuwair et al. 2016; Camata and Shing 2010; Lee et al. 2007), web shear failure (Robinson and Kosmatka 2008; Lee et al. 2007), and web buckling (Robinson and Kosmatka 2008; Alagusundaramoorthy et al. 2006)

04017022-1

J. Bridge Eng., -1--1

J. Bridge Eng.

Downloaded from ascelibrary.org by University of California, San Diego on 04/13/17. Copyright ASCE. For personal use only; all rights reserved.

Fig. 1. Strain profile and stress distribution of the composite section

Table 1. Testing Matrix Deck F1-13-1 F1-19-2 F2-13-1 F2-13-2 F2-13-3 F2-13-4 F2-13-5 C-13-1 C-13-2 a

Total length (mm)

Total depth H (mm)

UHPC plate thickness t (mm)

CFRP plies

B-GFRP plies

U-GFRP plies

Web orientation u

Load case

1220 1220 864 864 864 864 864 254 254

127 127 102 102 102 102 102 102 102

13 19 13 13 13 13 13 13 13

4 5 5 5 5 5 5 5 5

3 4a 5 5 5 6 6 6 6

3 3 5 5 5 6 6 6 6

60 60 63 63 63 63 63 63 63

Tire load Tire load Four-point Four-point Four-point Four-point Four-point Compression Compression

No chopped mat was used with this specimen.

were the major failure modes observed. Many techniques have been employed to enhance the bond between the deck components. The sandwich-panel system of Tuwair et al. (2016) consisted of glassfiber-reinforced polymer (GFRP) facing layers separated by a polyurethane foam core. The bond between the components of the system was enhanced by the corrugated shear layers that connected the top and bottom facing layers. These shear layers increased both the core shear stiffness and the global flexural stiffness in the longitudinal direction. Alternative FRP–concrete connections have been studied in terms of their ability to enhance delamination resistance, such as FRP dowel and shear key, coated sand layer, steel stud (Kim et al. 2015), and perforated FRP rib (PFR). Zou et al. (2016) experimentally studied the efficiency of using PFR connections in the FRP–concrete hybrid beam using push-out tests. They showed that the PFR ultimate capacity was almost 2.5 times the ultimate capacity of steel bolts, and the slip resistance was 10 times more than steel-bolt slip resistance.

Fig. 2. Number of grooves for resin flow in F1-13-4 and F1-13-5 decks

Research Objectives and Plan Although FRP sandwich panels and composites decks are promising options for bridge deck replacement and new construction, certain problems arise with existing FRP decks. The flexibility of the GFRP deck as a result of the low elastic modulus of GFRP (Mirmiran et al. 2012) makes satisfying serviceability criteria difficult. The delamination of the top and bottom faces from the core often limits the capacity of the system. Moreover, the additional wearing surface can add more self-weight to the deck and create an additional interface (top face and the overlay) that may debond. The objective of © ASCE

Table 2. Laminate Properties Fiber type CFRP U-GFRP B-GFRP

04017022-2

J. Bridge Eng., -1--1

Width (mm)

Thickness (mm)

Tensile modulus (GPa)

Shear modulus (GPa)

Tensile ultimate strain (mm/mm)

25.4 25.4 25.4

2.54 1.8 3.8

124 38 12.4

5.1 3.8 4.1

0.015 0.012 0.014

J. Bridge Eng.

Downloaded from ascelibrary.org by University of California, San Diego on 04/13/17. Copyright ASCE. For personal use only; all rights reserved.

this study was to develop and evaluate a new UHPC-FRP composite deck system with reduced weight and high performance that serves to increase the load rating of existing bridges and improve

functionality and service life. This lightweight bridge deck would also allow widening of existing bridges without placing additional deadweight on their substructures. The use of UHPC with FRP in

Fig. 3. Instrumentation plan for strain gauges for tire-wheel load test

Fig. 4. Instrumentation plan for strain gauges for four-point load test

© ASCE

04017022-3

J. Bridge Eng., -1--1

J. Bridge Eng.

80

•

A1

70

•

C2

60

•

A2

Downloaded from ascelibrary.org by University of California, San Diego on 04/13/17. Copyright ASCE. For personal use only; all rights reserved.

Load (kN)

50

•

40

B2

•

C1

30

•

B1

20

10 C-13-1 C-13-2

0 0

5

10

(a)

15

20

25

Displacement (mm) 80 C-13-1 C-13-2

•

A1

70

•

C2

60

•

A2

Load (kN)

50

•

40

B2

C1

30

•

B1

20

10

0 -3500

(b)

-3000

-2500

-2000

-1500

-1000

Strain (

-500

0

500

1000

)

Fig. 5. Responses of the compression test specimens: (a) load–displacement response; (b) web strain

© ASCE

04017022-4

J. Bridge Eng., -1--1

J. Bridge Eng.

Downloaded from ascelibrary.org by University of California, San Diego on 04/13/17. Copyright ASCE. For personal use only; all rights reserved.

Fig. 6. Failure mode of C-13-2

90

80

70

Load (kN)

60

Ultimate demand

50

40

Service demand

30

20

10 F1-13-1 F1-19-2

0 0 (a)

2

4

6

8

10

12

14

16

18

20

Displacement (mm)

Fig. 7. Load–displacement responses for flexural specimens: (a) tire-wheel load configuration; (b) four-point load configuration

© ASCE

04017022-5

J. Bridge Eng., -1--1

J. Bridge Eng.

90

80

70

Load (kN)

Downloaded from ascelibrary.org by University of California, San Diego on 04/13/17. Copyright ASCE. For personal use only; all rights reserved.

60

50

Ultimate demand

40

30 Service demand 20 F2-13-1 F2-13-2 F2-13-3 F2-13-4 F2-13-5

10

0 0 (b)

2

4

6

8

10

12

Displacement (mm) Fig. 7. (Continued.)

bridge deck application increases the deck stiffness and strength, minimizes the possibility of FRP buckling failure, and acts as an integral wearing surface, so no additional overlay is needed. Seven UHPC-FRP composite deck units were fabricated using the vacuum-assisted resin transfer molding (VARTM) process for monotonic flexural testing. Two deck segments were fabricated and tested in compression to determine the vertical load path and failure modes. Tensile coupon tests for material characterization were also conducted. The experimental results were compared with strength and serviceability requirements per AASHTO (2013). Finite-element models of the deck units were found to demonstrate good agreement with experimental results, enabling further investigation of the loadtransfer mechanisms.

process while preserving the ratio of shear span over depth for all decks. The UHPC-FRP composite deck unit (after fabrication) consisted of a top UHPC plate for compression resistance and as an integral wearing surface. A unidirectional GRFP (U-GFRP) laminate was placed underneath the UHPC plate to promote bonding. The tension resistance was derived from a unidirectional carbon-fiber-reinforced polymer (CFRP) laminate on the bottom face of the deck. The inclined web consisted of a bidirectional GFRP (B-GFRP) laminate. Foam was used as filler material to minimize deck weight and increase the buckling capacity of the web.

Materials and Properties

Design and Materials A cross section of the typical deck unit proposed is shown in Fig. 1. The overall depth of the deck (H) was selected to be in the range of 102–127 mm to be compatible with the decks in previous research (Saleem et al. 2011; Ghasemi et al. 2016a, b), and the unit width was set to be 254 mm. A typical panel width for a bridge deck application is 1,524 mm, so six of these deck units would form one typical panel. The decks were prepared with two different sizes. The first size was 1,220 mm long and 127 mm deep, and the second size was 864 mm long and 102 mm deep. The length of the second set of decks was reduced to enable better quality control in the VARTM © ASCE

The commercial UHPC material Ductal (Lafarge North America, Chicago, Illinois) was used for the top plate. The premix cement, superplasticizer, and high-strength-steel fibers were supplied by Lafarge North America. The mechanical properties of UHPC have been investigated by previous researchers (Graybeal 2006; Chanvillard and Rigaud 2003). The compressive strength of UHPC according to the manufacturer’s specifications can range between 150 and 200 MPa for Ductal with metallic fibers if a heat-treatment process is applied. The postcracking direct tensile strength at 0.3-mm crack width can reach 10 MPa. The heat-treatment process was not applied to any of the specimens presented in this paper. Two batches were used for the casting of the top UHPC plates. The average 28-day compressive strengths were 166 and 160 MPa for

04017022-6

J. Bridge Eng., -1--1

J. Bridge Eng.

Experimental Specimen Design A sectional, buckling, and preliminary finite-element analysis (FEA) were performed for a series of deck cross sections to select the design dimensions. The performance criteria were minimizing the selfweight below 1.0 kN·m2, maintaining a maximum depth of 125 mm, and satisfying the AASHTO load requirements. The live-load demand for the single-unit, single-span deck was calculated to be 47.5 kN based on loading configuration and deck width (AASHTO 2013), which represents an upper bound on the demand because multiple units will resist the load in full-size deck panels. The corresponding midspan moment demands were 13.1 and 7.8 kN·m for the 1,220- and 864-mm-long decks, respectively. The design goal from sectional analysis was crushing of the top UHPC plate prior to U-CFRP rupture. However, the achieved strength in the preliminary FEA was always limited by the assumed value of interfacial strength between the UHPC and GFRP. The CFRP laminate thickness (number of plies) was determined by sectional analysis, whereas the B-GFRP laminate thickness (number of plies) was determined by buckling analysis. The top UHPC plate thickness was selected to be either 13 or 19 mm for weight limitations and to avoid punching failure. The web was constructed using multiple plies of bidirectional glass fiber with orientations between 60 and 63°. The web orientation range was chosen based on preliminary FEA. Moment-curvature analysis was used to determine the sectional flexural capacity and effective stiffness of the composite deck. The

70 F1-13-1 F1-19-2

60

Ultimate demand

50

Load (kN)

Downloaded from ascelibrary.org by University of California, San Diego on 04/13/17. Copyright ASCE. For personal use only; all rights reserved.

the first and second batches, respectively, using 102
205 mm cylinder tests. The carbon fiber used was 410-g/m2, 12-K unidirectional fabric from iLLSTREET Composites (Charleston, South Carolina), and it is compatible with polyester, epoxy, and vinyl ester resins. The unidirectional glass fiber that was placed under the UHPC plate was 440-g/m2 fabric manufactured by AGY (Aiken, South Carolina). The bidirectional glass fiber used in the web was 610-g/m2, 0.55mm-thick E-Glass fabric from US Composites (West Palm Beach, Florida). A chopped mat was used with the bidirectional glass in the web laminate to build the thickness (increase buckling resistance) and enhance resin transfer. Chopped mat contains randomly oriented long fiberglass strands that are linked together with a styrene-soluble binder that works like glue in connecting the fibers. Vinyl ester typically has low viscosity, which makes it a good choice for resin infusion applications. Vinyl ester #1110 resin from Fibre Glast Developments (Brookville, Ohio) was used. The manufacturer-specified tensile strength is 82 MPa, and the modulus of elasticity is 3,720 MPa. It has a pot life of 15–30 min, 275-cps viscosity, and good corrosion resistance, and it is fully cured in 24– 48 hours. The foam core was 32 kg/m3 closed-cell polyisocyanurate foam from Fibre Glast Developments. The manufacturer-specified tensile modulus is 8,440 kPa, and the compressive modulus is 4,823 kPa. The foam was easily customized to the desired shapes by cutting to size and gluing several sheets together.

40

30

Service demand

20

10

0 -700 (a)

-600

-500

-400

Strain (

-300

-200

-100

0

)

Fig. 8. Load–strain responses for tire-wheel load flexural test specimens: (a) top UHPC strain; (b) bottom CFRP strain

© ASCE

04017022-7

J. Bridge Eng., -1--1

J. Bridge Eng.

70

60

Ultimate demand

Load (kN)

Downloaded from ascelibrary.org by University of California, San Diego on 04/13/17. Copyright ASCE. For personal use only; all rights reserved.

50

40

Service demand

30

20

10 F1-13-1 F1-19-2

0 0 (b)

500

1000

1500

Strain (

2000

2500

)

Fig. 8. (Continued.)

deck unit cross section was subdivided into layers for each material, as shown in Fig. 1. It was assumed that the linear variation of strain through depth was valid (constant curvature f and perfect bond between composite materials) and section forces could be computed through the equilibrium of the normal stresses. The axial strains of each layer e ðyÞ ¼ e a  y f were used to generate layer stresses s (y) based on material constitutive models, where y is the distance from the middepth of the composite section to the center of gravity (CG) of each layer; and e a is the strain at the middepth. Both the strain and stress profiles are shown in Fig. 1. The stress–strain behaviors of CFRP and GFRP were modeled as linear brittle materials with fracture strains of 0.015 and 0.014, respectively. The elastic moduli in preliminary design were assumed to be 45, 75, 25, and 10 GPa for UHPC, CFRP, U-GFRP, and B-GFRP, respectively. A single layer was used for CFRP, U-GFRP, and horizontal B-GFRP. UHPC in compression was modeled using a linear stress–strain curve with strain at peak stress of –0.0032 (Aaleti et al. 2013). The UHPC plate was divided into five layers. The contribution of the foam to the flexural strength and stiffness was neglected. The membrane action of the inclined GFRP webs was assumed minimal, and therefore the B-GFRP stresses were only included for the top and bottom horizontal layers (s BGc and s BGt). Because of the linear behavior of the materials prior to failure, the slope of the moment-curvature plot corresponds to the effective flexural stiffness EIeff. The sectional analysis results show that within the constrained parameter space, the average nominal moment capacity and flexural stiffness were 36 and 345 kN·m2, respectively, defined by crushing © ASCE

of the top UHPC plate. It was observed that nearly 95% of the flexural capacity was contributed from the UHPC plate and CFRP and only 5% from U-GFRP and B-GFRP. Given linear material models for all components of the composite deck, an equivalent section is an alternative approach to determining the flexural capacity and effective stiffness. Disregarding the small contribution of GFRP, the effective stiffness for preliminary design can be obtained from UHPC and CFRP as Eq. (1). An estimate of the section moment capacity can be obtained by multiplying EIeff by the limiting curvature, corresponding to the ultimate strain of either UHPC f ¼ e c =ðy  yc Þ or CFRP f ¼ e cf =ðy  ycf Þ. Ec Ac yc þ Ecf Acf ycf Ec Ac þ Ecf Acf h i h i ¼ Ec Ic þ Ac ðyc  y Þ2 þ Ecf Icf þ Acf ðycf  y Þ2

y ¼ EIeff

(1)

where y , yc, and ycf = distance from the middepth to the CG of the composite deck, UHPC, and CFRP, respectively; Ec, Ic, and Ac = modulus of elasticity, moment of inertia (about CG of UHPC), and area of UHPC, respectively; and Ecf, Icf, and Acf = modulus of elasticity, moment of inertia (about CG of CFRP), and area of CFRP, respectively. For buckling analysis, the FRP web was treated as an elastic beam, whereas the surrounding foam was treated as elastic foundations on both sides of the web. Past studies indicated that the foam increased the buckling capacity of the web (Robinson and Kosmatka 2008). In addition, the foam provides extra support to local bending

04017022-8

J. Bridge Eng., -1--1

J. Bridge Eng.

and punching of the UHPC plate in the proposed deck. The web buckling load for a fixed-end boundary condition (on both ends) with a finite-length beam has been given by Hetenyi (1946) as pffiffiffiffiffiffiffiffiffi p EI þ 2 kf EI l2

(2)

where Pcr = critical buckling load; E = web modulus; I = web moment of inertia; l = length of the supported beam; and kf = stiffness of the elastic foundation. The second term of Eq. (2) is related to the foam contribution, and the value of kf was set to double the foam elastic modulus (Robinson and Kosmatka 2008). The critical buckling load based on beam on elastic foundation (BEF) theory for the selected web thicknesses and orientations was found to be within a range of 70–120 kN, and therefore was not the limiting mode. The final CFRP and U-GFRP laminate thicknesses required were found to be in the range of 3–4 mm and 2–2.5 mm, respectively, based on the sectional analysis. The final thickness of B-GFRP required was found to be in the range of 5.1–6.5 mm based on the buckling analysis. A chopped mat was used between each two B-GFRP plies in all 13mm-thickness UHPC specimens. The final laminate thickness was achieved by using the numbers of plies indicated for each specimen in Table 1. These values represent the lower bound that satisfies the demand load requirement based on sectional and buckling analyses. Increasing the number of plies or the thickness of each material will increase the deck stiffness and weight and overestimate the capacity. However, the failure mode of the deck was shown to be related to the interfacial bond stress rather than the dimensions of the decks.

In total, seven deck units were tested in flexure, as shown in Table 1, in which the decks are grouped according to test, loading type, and UHPC thickness. The specimen nomenclature uses F1 or F2 to refer to flexural specimens and loading configuration, F1 for wheel-tire load and F2 for four-point load. The numbers 13 and 19 represent the UHPC thickness, followed by the specimen number. Two deck segments were tested in compression, with both specimens having the same cross section as the F2-13-4 and F2-13-5 specimens. The dimensions of each deck segment were 254 mm long, 254 mm wide, and 102 mm thick. The specimen nomenclature uses the prefix C for the compression tests, followed by the UHPC thickness and specimen number. Composite Manufacturing Process The composite, or hybrid, specimens were prepared using VARTM infusion to get a high-quality system with a better adhesion and fiber volume fraction than would be achieved using typical wet layup. This method is typically suitable for manufacturing of carbon- and glass-fiber composites. After casting the UHPC plates, several longitudinal and transverse grooves were drilled to promote resin flow and bond between the UHPC plate and top U-GFRP laminates. The grid spacing was 51 mm for all specimens, except for F1-13-4 and F1-13-5, which had a decreased spacing of 25.4 mm to improve resin transferability, as shown in Fig. 2. The two deck units for wheel-tire load testing (F1-13-1 and F1-19-2) were infused and cured in one step, including the UHPC

90

80

70

60

Load (kN)

Downloaded from ascelibrary.org by University of California, San Diego on 04/13/17. Copyright ASCE. For personal use only; all rights reserved.

Pcr ¼ 4

2

Experimental Methods

50

Ultimate demand

40

30 Service demand 20

10

0 -2500 (a)

F2-13-1 F2-13-2 F2-13-3 F2-13-4 F2-13-5

-2000

-1500

-1000

Strain (

-500

0

)

Fig. 9. Load–strain responses for four-point load flexural test specimens: (a) top UHPC strain; (b) bottom CFRP strain

© ASCE

04017022-9

J. Bridge Eng., -1--1

J. Bridge Eng.

90 F2-13-1 F2-13-2 F2-13-3 F2-13-4 F2-13-5

80

70

Load (kN)

Downloaded from ascelibrary.org by University of California, San Diego on 04/13/17. Copyright ASCE. For personal use only; all rights reserved.

60

50 Ultimate demand 40

30 Service demand 20

10

0 0

500

(b)

1000

1500

Strain (

2000

2500

3000

)

Fig. 9. (Continued.)

plate. All of the decks for four-point loading were infused in two steps to better enable resin transfer and fiber wet-out along the length. The foam core and web fiber reinforcement were infused and cured first, upon which they were combined with UHPC and CFRP to achieve the final infused specimen. Laminate Characterization Tensile tests were performed to obtain the mechanical properties of FRP laminates. To mimic the flexural specimen preparation, all coupons were prepared using VARTM and the same number of plies utilized in F2-13-4 and F2-13-5 specimens. Twenty-three coupons were prepared with 305-mm length, 25.4-mm width, and different thicknesses depending on the number of FRP plies. Natural G10 FR4 fiberglass epoxy sheets from ePlastics (San Diego, California) were used in the end tabs, with beveled edges to prevent gripping damage. The coupon tests were conducted using an Instron (Norwood, Massachusetts) SATEC universal testing machine (UTM) with a loading rate of 1.27 mm/min as recommended by the ASTM (2014). One 120-X strain gauge produced by Kyowa (Novi, MI) was attached to the middle of the coupon to record the longitudinal strain during the test. A summary of the test results is shown in Table 2. Test Setup and Instrumentation Plans The compression tests were performed using a UTM under displacement control with a loading rate of 2.54 mm/min. Steel plates were used at the top and bottom of each specimen along with a © ASCE

25.4-mm-thick high-grade neoprene sheet under the top steel plates to uniformly distribute the applied load over the whole surface area. The middle part of foam was removed to provide enough space to attach the strain gauges on the web and U-GFRP laminates. A polyurethane foam was poured later to substitute the removed parts. In addition to the strain gauges, load and displacement from the machine were recorded during the test. In the flexural tests, all decks were loaded monotonically until a significant drop in the load was observed. Both F1-13-1 and F1-192 were loaded at the middle of the span. A 508
254 mm steel plate on top, with the longer side parallel to the length of the specimen, was used to simulate the HS20 truck dual-tire wheel (AASHTO 2013). The instrumentation plan and loading arrangement are shown in Fig. 3. Four strain gauges were attached to the bottom surface (CFRP laminate on the tension side), and one strain gauge was attached to the top UHPC plate (compression side). In addition to the strain gauges, two string pots were used at the center of each specimen to record the maximum deflection. After preparing the setup, they were tested using a 1,024-kN-capacity hydraulic actuator with a loading rate of 0.76 mm/min. All of the 102-mm-thick decks were tested using a four-point load configuration with a UTM under a displacement control rate of 0.76 mm/min. Three strain gauges were attached to the bottom CFRP surface, and one strain gauge was attached to the top UHPC plate between the loading noses. A LVDT was used to record the displacement at the midspan. Also, one LVDT was placed at each support to calculate the relative displacement. The instrumentation plan and loading arrangement of this system are shown in Fig. 4.

04017022-10

J. Bridge Eng., -1--1

J. Bridge Eng.

All instruments for all tests were connected to a data-acquisition system with a sampling frequency of 1 Hz.

Experimental Results Compression Tests

Downloaded from ascelibrary.org by University of California, San Diego on 04/13/17. Copyright ASCE. For personal use only; all rights reserved.

Figs. 5(a and b) show the load–deflection and load–strain responses of the two specimens. Both specimens ultimately experienced

bearing failure rather than buckling failure of the webs. The web strain showed unloading in C-13-1 (B1) as a result of bending of the web as the UHPC plate developed a longitudinal crack at the midpoint between the webs. Failure occurred (A1) as a result of excessive movement of the webs/foam transversely outward. However, there was a small gap under the top plate in C-13-2 that caused transverse UHPC bending and punching at the faces of the webs (A2), as shown in Fig. 6. The load increased again (B2) as a result of the webs, with the ultimate failure (C2) occurring in a similar manner to C-13-1.

Table 3. Flexural Test Results Deck F1-13-1 F1-19-2 F2-13-1 F2-13-2 F2-13-3 F2-13-4 F2-13-5

Total depth (mm)

Ultimate load (kN)

Normalized ultimate moment (kN·m/m)

Deflection at service load (mm)

Deflection at ultimate load (mm)

Capacity/demand

127 127 102 102 102 102 102

63.9 53.1 60.5 66.5 47.3 88.3 73.2

1.77 1.47 1.00 1.10 0.78 1.46 1.21

4.67 8.128 1.795 2.12 2.02 1.836 1.878

12.2 16.8 6.6 6.9 9.9 6.86 7.0

1.35 1.12 1.27 1.4 1 1.86 1.5

Fig. 10. FEM of four-point load test: (a) loading and boundary conditions; (b) element configurations

© ASCE

04017022-11

J. Bridge Eng., -1--1

J. Bridge Eng.

With the geometry of the tested specimens, the total vertical component of predicted Pcr according to Eq. (2) is 75.0 kN (37.5 kN for a single web), which exceeded the ultimate experimental capacity for the two specimens of the compression test. The contribution of the foam in Eq. (2) was found to be 52% of the critical load for a web thickness equal to 5.1 mm and 44% for a web thickness equal to 6.1 mm.

Downloaded from ascelibrary.org by University of California, San Diego on 04/13/17. Copyright ASCE. For personal use only; all rights reserved.

Flexural Tests Figs. 7(a and b) show the load–deflection curves for the wheel-tire and four-point load flexural tests, respectively. The interface strength between the UHPC and the top GFRP layers plays a significant role in the performance of the proposed deck and controls the

behavior of the system. An audible indication of delamination of the top UHPC plate was noticed for all specimens during the test. Delamination of the top UHPC plate started at the supports. As the load increased, delamination propagated toward the two point loads until the peak load was reached, similar to the observations of Tuwair et al. (2016). For the wheel-tire load test, F1-13-1 (with chopped mat) showed a stiffer behavior and higher load capacity than F1-19-2 (without chopped mat). Forensic investigation on the tested sections showed that specimen F1-13-1 was fully infused, whereas specimen F1-192 had some regions of dry fiber. The lack of resin in some regions of F1-19-2 made the debonding failure even lower. There was a significant increase in the strength of F2-13-4 and a slight improvement for F2-13-5 as a result of the bond enhancement because of the

Table 4. Comparison of Experimental and FEA Results Peak load Deck F2-13-2 F2-13-4

Deflection at peak load

CFRP strain at peak load

Experimental (kN)

FEA (kN)

Experimental (mm)

FEA (mm)

Experimental (mm/mm)

FEA (mm/mm)

66.5 88.3

71.7 97.6

6.6 6.9

5.36 6.41

0.002331 0.002173

0.001926 0.002189

Fig. 11. Failure modes and contour lines of interfacial shear stress in FEM: (a) interface failure (test); (b) interface failure (FEM)

© ASCE

04017022-12

J. Bridge Eng., -1--1

J. Bridge Eng.

to satisfy the serviceability criteria for both the F1 and F2 specimens. However, only a single unit for the F2 specimens is necessary to satisfy the serviceability limit of L/360 used by other researchers (Williams et al. 2003; GangaRao et al. 1999) in interpreting experimental results. Williams et al. (2003) demonstrated that the L/800 limit is for bridge girders, not the decks. Moreover, the limit of L/800 is considered only for steel, aluminum, and/or concrete vehicular bridges, not for FRP bridges (AASHTO 2013).

Analytical and Numerical Results The flexural composite deck specimen results were investigated using FEA, sectional analysis, and analytical expressions based on first-order shear deformation theory (FSDT). Finite-Element Analysis Three-dimensional (3D) FEA was performed using Marc for comparison with measured deformations from the four-point bending tests and for better understanding of the structural behavior of the deck. Eight-node hexahedral elements were used to simulate the UHPC and foam with isotropic material properties. Four-node quadrilateral composite laminate membrane elements were used to model the web GFRP, and shell elements were used to model the rest of the FRP laminates. The foam element nodes were connected directly to the web (i.e., the edges of the foam and

80

70

60

Ultimate demand

50

Load (kN)

Downloaded from ascelibrary.org by University of California, San Diego on 04/13/17. Copyright ASCE. For personal use only; all rights reserved.

increased number of grooves. The normalized midspan moment for F2-13-4 was 14.6 kN·m/m, which is approximately 95% of the normalized moment capacity of a deck of the same depth that was tested by Ghasemi et al. (2016a). Because both decks were of the same depth, the midspan moment was only normalized by the width. At the end of the test, some minor cracks in the UHPC plate were observed under and between the loading points in the fourpoint loading tests. Also, a UHPC punching failure was observed at the supports of F1-19-2 as a result of high local shear force. Figs. 8 and 9 show the load-versus-strain curves for both test configurations at the top UHPC plate and bottom CFRP laminate, respectively. The peak compressive strains in the UHPC plate were −0.0006 and −0.0022 for the wheel-tire and four-point load tests, respectively, or 20% and 70% of the crushing strain of the concrete (ɛcr ¼ 0:0032). The maximum tensile strains at the bottom CFRP laminate for the wheel-tire and four-point load test of the deck were 0.0022 and 0.0025, respectively, which are lower than the ultimate CFRP strain (ɛu ¼ 0:01). The intermediate drops shown in the responses of both the F2-13-1 and F2-13-3 decks appear to be a result of slip during debonding propagation, as evident in Fig. 9(a). Table 3 summarizes the experimental load, normalized moment, and deflection results for all deck specimens along with the capacity/demand ratio achieved. The strength demand was met for all deck specimens. The midspan deflection limits corresponding to L/800, where L is the span of the deck (AASHTO 2013), are 1.38 mm and 0.95 mm for the F1 and F2 specimens, respectively. Considering only one-way bending, two units would be necessary

40

Service demand

30

20 Experimental Results FEM Results FSDT Results Sectional Analysis Results

10

0 0 (a)

1

2

3

4

5

6

7

8

9

Displacement (mm)

Fig. 12. Experimental, numerical, and analytical results for F2-13-2 deck: (a) load-displacement relation; (b) load–strain relation

© ASCE

04017022-13

J. Bridge Eng., -1--1

J. Bridge Eng.

80

70

Ultimate demand

50

Load (kN)

Downloaded from ascelibrary.org by University of California, San Diego on 04/13/17. Copyright ASCE. For personal use only; all rights reserved.

60

40

Service demand

30

20

10 Experimental Results FEM Results

0 0 (b)

500

1000

1500

Strain (

2000

2500

)

Fig. 12. (Continued.)

the web were sharing the same nodes, with no interface or contact elements). The configuration of the element types is shown in Fig. 10(b). Eight-node hexahedral elements were used to simulate the bond between the UHPC and top GFRP. An interface/cohesive model was used to simulate the onset and progress of delamination. The constitutive behavior of these elements is defined in terms of traction versus the relative displacements between the top and bottom edge/surface of the elements. A bilinear bond-slip relation was used in the model, with parameters of cohesive fracture energy Gf, critical opening displacement (displacement at peak stress) vc, and maximum displacement opening vm. The parameters of this cohesive zone model were calibrated using single- and double-lap shear tests previously performed on specimens utilizing the same composite system as this study (Mirmiran et al. 2015). To obtain bond-slip curve parameters, the analytical strain distribution equations along the laminate derived by Yuan et al. (2004) and Wu et al. (2002) were fitted to the experimental strain data using nonlinear least squares. The calibrated parameter values were cohesive energy Gf = 0.62 N/mm, critical opening displacement vc = 0.01 mm, and maximum displacement opening vm = 0.5 mm. The multiaxial behaviors of UHPC were defined through a uniaxial principal stress–strain relation. The tension response was assumed to be bilinear with tensile strength s cr = 8.3 MPa, modulus of elasticity Ec = 48.8 GPa, ultimate crushing strain ɛcr ¼ 0:0032, and tension-softening modulus Es = 345 MPa. Cracking was defined by comparing the principal stress to the critical cracking stress in © ASCE

tension. The stresses in the principal direction decrease with increasing strain after cracking based on the softening modulus Es. The UHPC behavior in compression was modeled using a trilinear curve. The maximum stress value occurs at the crushing strain. The crushing behavior is described in a multiaxial stress state by a crushing surface having the same shape as the yield surface. However, the UHPC response was found to be within the elastic range only. The F2-13-2 and F2-13-4 deck specimens were analyzed with the same FEM by using the corresponding FRP thickness and the interface parameters for each deck. The typical mesh, boundary conditions, and applied load for the generic FEM are shown in Fig. 10(a). The two point loads were modeled as line loads at locations consistent with the experiments, as shown in Fig. 10(a). An arc-length-based load-stepping scheme was used to control the load step. The load–displacement and load–strain plots comparing the simulated results with the experimental results for F2-13-2 and F2-13-4 are shown in Figs. 12 and 13, respectively. The FEM for both decks captured the trends in initial stiffness and failure mode. Table 4 shows the comparison between simulated and experimental results. The model successfully captured the interfacial shear failure, as shown in Fig. 11. The drop in the peak load occurred when the slip reached its maximum effective opening displacement (vm) at the critical location. Similar to the experimental observations, the softening of the interface started at the support at the end of the elastic stage, as shown in the contours of Fig. 11(a). As the load increased, the delamination propagated to the other locations of the deck.

04017022-14

J. Bridge Eng., -1--1

J. Bridge Eng.

100

90

80

60

Load (kN)

Downloaded from ascelibrary.org by University of California, San Diego on 04/13/17. Copyright ASCE. For personal use only; all rights reserved.

70

Ultimate demand

50

40

Service demand

30

20

Experimental Results FEM Results (with foam) FEM Results (without foam) FSDT Results Sectional Analysis Results

10

0 0 (a)

1

2

3

4

5

6

7

8

9

Displacement (mm)

Fig. 13. Experimental, numerical, and analytical results for F2-13-4 deck: (a) load–displacement relation; (b) load–strain relation

The FE model showed that the foam had a minor effect on the flexural behavior of the deck. As shown in Fig. 13(a), the ultimate load for F2-13-4 when the foam was removed was 91.5 kN, which represents 94% of the ultimate load for the original case. The corresponding displacement for the no-foam case at the ultimate load level was 6.05 mm. The vertical load was shared between the UHPC and inclined webs. The percentage of the shear force that the webs resisted was found to be 65% at the beginning of the analysis and then dropped to 52% at the end of the elastic stage. This is consistent with the compression test results where the full contribution of the foam to buckling resistance was not mobilized. The model was extended to better understand the limiting failure modes. Mechanical shear connectors were introduced into the F2-134 model. Vertical FRP dowels were modeled using circular solid elements with 3.2-mm diameter and 13-mm length. These connectors were distributed along the length of the deck at 51-mm spacing. The compressive modulus of elasticity was set to be 13 GPa. The results indicate that arresting interfacial shear failure shifted the failure mode to crushing of the UHPC top plate under the two point loads. Fig. 14 shows the comparison with the experimental results and the contours of the stress distribution, where the FEM was stopped once the strain at the top plate reached the ultimate crushing strain. Sectional Analysis Sectional analysis results overpredicted the ultimate load for F2-132 and F2-13-4 because the strain was assumed to vary linearly © ASCE

through the depth and not vary in the direction transverse to the axis of the deck. The FEM demonstrated that the peak UHPC strain was at the point above the top end of the web where negative transverse bending and punching shear occurred simultaneously. By comparing the limiting concrete strains at the center line of the FEM with those from sectional analysis, the predicted loads were 139 and 153 kN. The failure criteria in the sectional analysis were the UHPC compressive failure or the CFRP tensile failure, whereas interfacial failure governed the failure load in the FEA. The maximum interfacial stress t f (difference between the UHPC stress and GFRP stress at the interface layer) was reached at a load level of 55.7 kN in the sectional analysis for F2-13-4. However, this value corresponds to the end of the elastic stage, not to the failure load resulting from delamination. The linear variation of strain prevented modeling of slip beyond this point; therefore, the sectional analysis could only be used to bound the strength (here, between 55.7 and 153 kN). Additionally, if the perfect bond was modeled in FEA (i.e., no interface elements), the predicted strength of the deck was 149 kN, which confirms the sectional analysis results. Load–Displacement Prediction The analytical load–deflection behavior of the composite decks was obtained considering both the flexural (DM) and shear (DS) deflections. Shear deformations in a composite section with low shear modulus contribute significantly to the total deformation. At a given location xi, the total deflection D can be obtained using the principle of virtual forces, as in Eq. (3). The virtual unit load was placed at xi.

04017022-15

J. Bridge Eng., -1--1

J. Bridge Eng.

100

90

80

60

Load (kN)

Downloaded from ascelibrary.org by University of California, San Diego on 04/13/17. Copyright ASCE. For personal use only; all rights reserved.

70

Ultimate demand

50

40

Service demand

30

20 Experimental Results FEM Results(with foam) FEM Results(without foam)

10

0 0

500

1000

1500

Strain (

(b)

2000

2500

)

Fig. 13. (Continued.)

Dðxi Þ ¼

ðL

½d MðxÞ f ðxÞ þ d VðxÞ g ðxÞdx

(3)

0

where d MðxÞ and f ðxÞ = virtual moment and real curvature, respectively; and d VðxÞ and g ðxÞ = virtual shear and real shear strain, respectively. For analysis of the composite decks, the curvature was obtained from the real moment normalized by EIeff. The effective flexural stiffness was obtained from moment-curvature analysis considering contributions from all components of the composite or using Eq. (1) assuming linear materials and considering only contributions from UHPC and CFRP. The shear strain was obtained by the real shear multiplied by k/GAeff, which is the effective shear stiffness of the inclined B-GFRP web (GAeff = 4,306 kN) with a shear correction factor (k ¼ 5=6). Considering the deck configuration shown in Fig. 4, the solution of Eq. (3) for the midspan deflection is shown in Eq. (4). The results were compared with both the experimental and numerical results, as shown in Figs. 12 and 13.   L 61PL3 21:6PL D x¼ ¼ þ (4) ðmmÞ 2 3000EIeff 100kGAeff

Conclusions and Recommendations This paper presents the characterization of an innovative lightweight low-profile UHPC-FRP composite deck system. This deck combines a UHPC plate for compression resistance, CFRP laminates for © ASCE

tension resistance, and GFRP for shear resistance. The reduced selfweight compared with existing decks allows an increase in the load rating of existing bridges and accordingly improves their functionality and service life. The following conclusions can be drawn from this research: 1. The proposed deck weighs 0.5–0.6 kN·m2, which makes it a good option for cases when the weight is a concern. Additionally, the UHPC top layer acts as an integral wearing surface, so an additional overlay is not needed. 2. AASHTO demand load requirements were met by the singleunit, single-simple-span specimens. Multiple units (e.g., two units for 864-mm-span specimens) are necessary to satisfy the AASHTO and FHWA serviceability guidelines (L/800). 3. The bond strength between the UHPC and the top U-GFRP plays a significant role in the performance of the deck and limits the strength. Increasing the number of grooves in the top UHPC plate effectively enhanced the bond strength between the top UHPC plate and the GFRP layer (Specimens F2-13-4 and F2-13-5). Further improvement in the bond may be achieved with mechanical/FRP connectors, resin beads, and more grooves/holes in the UHPC. FEA results confirmed that as the interface was improved, the failure mode was the crushing of the UHPC top plate. 4. The compression test results show that the vertical load path was limited by bearing failure as a result of transverse bending and punching of the UHPC plate. Comparing the results of flexural and compression tests, the ultimate bearing capacity was higher than the load in the flexural test except in Specimen F2-13-4. The predicted buckling load from BEF

04017022-16

J. Bridge Eng., -1--1

J. Bridge Eng.

Downloaded from ascelibrary.org by University of California, San Diego on 04/13/17. Copyright ASCE. For personal use only; all rights reserved.

Fig. 14. FEM results of case with additional mechanical connectors: (a) results comparison; (b) UHPC longitudinal stress distribution contours

theory was higher than the ultimate load for the compressive test but only because of the contribution of the foam. Therefore, the foam serves the important role of arresting local bending of the UHPC plate and preventing the buckling failure mode from occurring, but it does not contribute substantially to the flexural strength. © ASCE

5. The load–deflection response using FSDT slightly overestimated the deck stiffness from the experimental results because of the assumption in the underlying section equilibrium calculations. The shear deformation contribution was estimated at the ultimate load as 57% of the total deformation of the deck because of the low shear rigidity compared with

04017022-17

J. Bridge Eng., -1--1

J. Bridge Eng.

Downloaded from ascelibrary.org by University of California, San Diego on 04/13/17. Copyright ASCE. For personal use only; all rights reserved.

the flexural rigidity. The analytical and numerical results were in agreement. 6. Despite the promising results, additional investigations are required before the deck is ready for field implementation; load distribution between adjacent units, effects of continuity, and negative moment effects need to be studied by manufacturing and testing multiple-unit, multiple-span decks. The deck-togirder/deck-to-deck connection and fatigue performance also need to be investigated.

Acknowledgments This study was sponsored by the Florida Dept. of Transportation under Contract BDV29-977-11 with Mr. Sam Fallaha and Mr. William Potter as project managers. Additional support was provided by the National Center for Transportation Systems Productivity and Management (NCTSPM) University Transportation Center (UTC) at the Georgia Institute of Technology (Contract DTRT12GUTC12) to broaden the applications of the proposed lightweight deck system for accelerated bridge construction. The support of Lafarge North America for providing its UHPC (Ductal) is gratefully acknowledged. The views and findings reported here are those of the writers alone and not necessarily the views of the sponsoring agencies. All experiments were conducted at the Structural Laboratory of the University of Central Florida and at the Titan America Structures and Construction Testing Laboratory of Florida International University.

References Aaleti, S., and Sritharan, S. (2014). “Design of ultrahigh-performance concrete waffle deck for accelerated bridge construction.” Transport. Res. Rec., 2406, 12–22. Aaleti, S., Petersen, B., and Sritharan, S. (2013). “Design guide for precast UHPC waffle deck panel system, including connections.” Rep. No. FHWA-HIF-13-032, Federal Highway Administration, Washington, DC. AASHTO. (2013). AASHTO LRFD bridge design specifications, 6th edition with 2013 interim revisions, U.S. customary units, Washington, DC. Alagusundaramoorthy, P., Harik, I., and Choo, C. (2006). “Structural behavior of FRP composite bridge deck panels.” J. Bridge Eng., 10 .1061/(ASCE)1084-0702(2006)11:4(384), 384–393. ASTM. (2014). “Standard test method for tensile properties of polymer matrix composite materials.” ASTM D3039/D3039M-14, West Conshohoken, PA. Camata, G., and Shing, P. B. (2010). “Static and fatigue load performance of a GFRP honeycomb bridge deck.” Composites Part B, 41(4), 299–307. Chanvillard, G., and Rigaud, S. (2003). “Complete characterization of tensile properties of Ductal UHPFRC according to the French recommendations.” Proc., 4th Int. RILEM Workshop High Performance Fiber Reinforced Cementitious Composites, A. E. Naaman and H. W. Reinhardt, eds., RILEM Publications, Bagneux, France, 21–34. Ehlen, M. A. (1999). “Life-cycle costs of fiber-reinforced-polymer bridge decks.” J. Mater. Civ. Eng., 10.1061/(ASCE)0899-1561(1999)11:3(224), 224–230. FHWA (U.S. Federal Highway Administration). (2015). “Deficient bridges by highway system 2015.” hwww.fhwa.dot.gov/bridge/nbi/no10 /defbr15.cfmi (Mar. 23, 2017). GangaRao, H. V. S., Thippeswamy, H. K., Shekar, V., and Craigo, C. (1999). “Development of glass fiber reinforced polymer composite bridge deck.” SAMPE J., 35(4), 12–24. Ghasemi, S., Mirmiran, A., Xiao, Y., and Mackie, K. (2016a). “Novel UHPC-CFRP waffle deck panel system for accelerated bridge construction.” J. Compos. Constr., 10.1061/(ASCE)CC.1943-5614.0000607.

© ASCE

Ghasemi, S., Zohrevand, P., Mirmiran, A., Xiao, Y., and Mackie, K. (2016b). “A super lightweight UHPC-HSS deck panel for movable bridges.” Eng. Struct., 113, 186–193. Graybeal, B. (2006). “Material property characterization of ultra-high performance concrete.” Rep. No. FHWA-HRT-06-103, Federal Highway Administration, Washington, DC. Hetenyi, M. (1946). Beams on elastic foundation: Theory with applications in the fields of civil and mechanical engineering, University of Michigan Press, Ann Arbor, MI. Karbhari, V. M., and Seible, F. (2000). “Fiber reinforced composites— Advanced materials for the renewal of civil infrastructure.” Appl. Compos. Mater., 7(2), 95–124. Keller, T., Rothe, J., de Castro, J., and Osei-Antwi, M. (2014). “GFRP-balsa sandwich bridge deck: Concept, design, and experimental validation.” J. Compos. Constr., 10.1061/(ASCE)CC.1943-5614.0000423, 04013043. Keller, T., Schaumann, E., and Vallee, T. (2007). “Flexural behavior of a hybrid FRP and lightweight concrete sandwich bridge deck.” Composites Part A, 38(3), 879–889. Kim, J.-S., Kwark, J., Joh, C., Yoo, S.-W., and Lee, K.-C. (2015). “Headed stud shear connector for thin ultrahigh-performance concrete bridge deck.” J. Constr. Steel Res., 108, 23–30. Lee, J., Kim, Y., Jung, J., and Kosmatka, J. (2007). “Experimental characterization of a pultruded GFRP bridge deck for light-weight vehicles.” Compos. Struct., 80(1), 141–151. Lopez-Anido, R., GangaRao, H. V., Vedam, V., and Overby, N. (1997). “Design and evaluation of a modular FRP bridge deck.” Proc., Int. Composites Expo ’97, SPI, Washington, DC, 1–6. Manalo, A., Aravinthan, T., Fam, A., and Benmokrane, B. (2016). “Stateof-the-art review on FRP sandwich systems for lightweight civil infrastructure.” J. Compos. Constr., 10.1061/(ASCE)CC.1943-5614.0000729, 04016068. Mirmiran, A., et al. (2015). “Innovative modular high performance lightweight decks for accelerated bridge construction (ABC).” Rep. No. DTRT12GUTC12, National Center for Transportation Systems Productivity and Management (NCTSPM), Atlanta. Mirmiran, A., Mackie, K., Saleem, M. A., Xia, J., Zohrevand, P., and Xiao, Y. (2012). “Alternatives to steel grid decks—Phase II.” Rep. No. BDK80 977-06, Florida Dept. of Transportation (FDOT), Tallahassee, FL. Marc [Computer software]. MSC Software Corporation, Newport Beach, CA. Robinson, M., and Kosmatka, J. (2008). “Light-weight fiber-reinforced polymer composite deck panels for extreme applications.” J. Compos. Constr., 10.1061/(ASCE)1090-0268(2008)12:3(344), 344–354. Saleem, M. A., Mirmiran, A., Xia, J., and Mackie, K. (2011). “Ultra-highperformance concrete bridge deck reinforced with high-strength steel.” ACI Struct. J., 108(5), 601–609. Shann, S. V. (2012). “Application of ultra high performance concrete (UHPC) as a thin-bonded overlay for concrete bridge decks.” M.S. thesis, Michigan Technological Univ., Houghton, MI. Tuwair, H., Volz, J., ElGawady, M. A., Mohamed, M., Chandrashekhara, K., and Birman, V. (2016). “Testing and evaluation of polyurethane-based GFRP sandwich bridge deck panels with polyurethane foam core.” J. Bridge Eng., 10.1061/(ASCE)BE.1943-5592.0000773, 04015033. Williams, B., Shehata, E., and Rizkalla, S. H. (2003). “Filament-wound glass fiber reinforced polymer bridge deck modules.” J. Compos. Constr., 10.1061/(ASCE)1090-0268(2003)7:3(266), 266–273. Wu, Z., Yuan, H., and Niu, H. (2002). “Stress transfer and fracture propagation in different kinds of adhesive joints.” J. Eng. Mech., 10.1061 /(ASCE)0733-9399(2002)128:5(562), 562–573. Yuan, H., Teng, J. G., Seracino, R., Wu, Z. S., and Yao, J. (2004). “Fullrange behavior of FRP-to-concrete bonded joints.” Eng. Struct., 26(5), 553–565. Zhang, Y., Cai, C. S., Shi, X., and Wang, C. (2006). “Vehicle-induced dynamic performance of FRP versus concrete slab bridge.” J. Bridge Eng., 10.1061/(ASCE)1084-0702(2006)11:4(410), 410–419. Zou, X., Feng, P., and Wang, J. (2016). “Perforated FRP ribs for shear connecting of FRP-concrete hybrid beams/decks.” Compos. Struct., 152, 267–276.

04017022-18

J. Bridge Eng., -1--1

J. Bridge Eng.