Performance of newly Developed CFRP Precast Prestressed Decked Bulb T Beams

Nabil Grace

Dean, College of Engineering, Lawrence Technological University, Southfield, MI, U.S.A.

Tsuyoshi Enomoto

Manager, Tokyo Rope Manufacturing Co. Ltd. Tokyo, Japan

Prince Baah

Graduate Student, Center for Innovative Materials Research (CIMR), Lawrence Technological University,

Southfield, MI, U.S.A.

Mena Bebawy

Post-Doctoral Research Fellow, Center for Innovative Materials Research (CIMR), Lawrence

Technological University, Southfield, MI, U.S.A.

ABSTRACT

This study introduces an innovative scheme of bridge superstructure for expedited

construction, improved inspection, serviceability, and extended lifespan. The new bridge

superstructure is assembled from precast prestressed decked bulb T beams reinforced and prestressed with corrosion-free fiber reinforced polymer (FRP) materials. An experimental

investigation accompanied by a numerical simulation was developed to evaluate the

performance of the newly developed beams. Through the experimental investigation, three single decked bulb T beams were constructed and tested to failure. The first beam served as

a control beam and was prestressed and reinforced with conventional steel strands and

reinforcing bars. The second and third beams were prestressed and reinforced with carbon fiber composite cables (CFCC) strands and carbon fiber reinforced polymer (CFRP)

tendons, respectively. The investigation revealed that the performance of beams reinforced

with CFCC strands or CFRP tendons is comparable with the performance of the control

beam at both service and ultimate limit states. All three beams exhibited high load carrying

capacity with large corresponding deflection and fair amount of absorbed energy before

failure. The study showed that the corrosion-free FRP-reinforced decked bulb T beams can

be safely deployed in construction to enhance the performance and extend the lifespan of bridge superstructures.

KEYWORD

carbon fiber reinforced polymer, prestressed, precast concrete, flexural, T beams

F3A02

1. INTRODUCTION

The use of decked bulb T beams in the

construction of bridge superstructures has

emerged rapidly during the last few decades. Several design agencies have implemented bulb

T beams in their design guidelines with some

differences in dimensions and construction techniques. For example, Utah Department of

transportation (UDOT) categorizes bulb T beam

bridges according to construction technique into

three classes: Bulb T beams with concrete deck,

decked bulb T beams without concrete deck, and

post-tensioned bulb T beam with concrete deck

and post-tensioning strands. Likewise,

Washington State Department of Transportation

(WSDOT) provides details for both bulb T

beams with deck and decked bulb T beams without decks [1].

Examples for construction of decked bulb T

beams in the U.S.A. can be traced back to 1986

with the construction of a six-span prestressed

concrete decked bulb T beam bridge in

Minnesota [2]. The bridge superstructure

comprised five decked bulb beams with a depth

of 1000 mm and top flange width of 1800 mm. The end spans had a length of 21.3 m, while the

interior spans had a length of 25.9 m. Steel bars

of 25.4 mm in diameter were used to transversely post-tension the top flange. Another

decked bulb T beam bridge was constructed in

Kittitas County, WA to replace a deteriorated bridge in 2009. The beams were interconnected

using welded steel joints. In addition, to

overcome the problem of longitudinal joint

leakage, the new bridge was provided with a

waterproof membrane in addition to an asphalt

emulsion to hold the membrane in place.

A bridge superstructure constructed using

adjacent decked bulb T beams has several advantages over bridge superstructures

constructed using different adjacent beams. For

instance, adjacent decked bulb T beams allow adequate space between the beams for inspection

and maintenance. In addition, the top flange of

the beams is fabricated to act as a deck slab to

save time and effort and expedite the

construction of the bridge by eliminating the

need for a cast-in-place deck slab. However,

durability issues related to the corrosion of steel

reinforcement in decked bulb T beams is the

concern yet to be handled.

Through this investigation, the durability of the

decked bulb T beams is improved by replacing

the conventional steel reinforcement with CFCC

[3] or CFRP [4] reinforcement. CFCC and CFRP

are corrosion resistant and can significantly extend the lifespan of the bridge superstructure.

The use of CFRP in bridge construction has been

proven to be successful with the construction and monitoring of the ten-year old Bridge Street

Bridge in Southfield, Michigan, U.S.A. [5] and

four-year old Penobscot Narrows Cable Stayed

Bridge in Maine, U.S.A. [6].

The investigation presented in this paper

represents Phase #1 of a multi-task project

dedicated to establish comprehensive design

guidelines for Decked bulb T beam bridges

reinforced and prestressed with different kinds of FPR reinforcement for flexure and shear.

2. EXPERIMENTAL PROGRAM

To evaluate the performance of the developed

beams, three prestressed decked bulb T beams

were constructed, instrumented, and tested to

failure under vertical loads. The beams had a

span of 9,750 mm, top flange width of 457 mm, bottom flange width of 305 mm, and a total

depth of 356 mm. All three beams were

reinforced with an identical reinforcement scheme, shown in Fig.1, but with different

reinforcement materials. The first beam (steel

beam) served as a control beam and was pretensioned with four low relaxation

prestressing steel strands and reinforced with

non-prestressing steel reinforcement. The second

beam (CFCC beam) was pretensioned with four

prestressing CFCC strands and reinforced with

non-prestressing CFCC strands. The third beam

(CFRP beam) was pretensioned with four CFRP-

leadline tendons and reinforced with CFRP-DCI

tendons [7]. Details of reinforcement are provided in Table 1 and Table 2 for non-

prestressing and prestressing reinforcement,

respectively.

Table 1 Properties of non-prestressing

reinfocement Diam.

(mm)

Area

(mm2)

Yield

strength

(MPa)

Ultimate

strength

(MPa)

Tensile

modulus

(GPa)

Steel

16 200 413 620 200

CFCC

15.2 115 -- 2,590 159

CFRP 10 72 -- 2,344 157

Table 2 Properties of prestressing

reinforcement Diam.

(mm)

Area

(mm2)

Yield

strength

(MPa)

Ultimate

strength

(MPa)

Tensile

modulus

(GPa)

Steel

16 140 1585 1,861 201

CFCC

15.2 115 -- 2,590 159

CFRP 10 72 -- 2,744 163

Fig.1 Cross section of decked bulb T beams

In all beams, each strand/tendon was prestressed

with an initial prestressing force of 111 kN. The

prestressing force was applied using two bulkheads anchored to a heavily reinforced

concrete foundation with high strength bolts. In

addition, the CFCC pretensioning strands were

provided with a special mechanical anchorage

system at each end to facilitate pulling of the

strands without damaging their ends. This

anchorage system consisted of sleeve for CFCC,

wedges, joint coupler, mesh sheet, braided grip

and wedges for the steel strand. Fig.2 shows the

process of pretensioning the strands using hydraulic pump, while Fig.3 shows the newly

developed and tested anchorage system for the

CFCC strands that can be installed in field.

The transverse reinforcement for all beams was

made of steel stirrups with a diameter of 10 mm

and center-to-center spacing of 102 mm. Fig.4

shows the completed reinforcement cage in the

formwork while casting the concrete. The concrete mix was designed to achieve an

average 28-day compressive strength of 62 MPa.

However, the concrete in the CFRP beam

achieved an average 28-day compressive

strength of 50 MPa.

Fig.2 Applying pretensioning force to

longitudinal strands

Fig.3 Couplers for CFCC strands

Fig.4 Placing concrete in bulb T beam

The release of the prestressing forces took place

14 days after casting the concrete. At the time of prestress release, the concrete compressive

strength averaged 41 MPa. The release of

prestressing force in the steel beam was

performed by cutting the steel strands at the ends

of the beam using a gas/oxygen torch, while the

release of the prestressing force in the CFRP

beam was performed by saw cutting the CFRP

tendons using electric saw. On the other hand, the release of the prestressing forces in the

CFCC beam was performed by further pulling

the CFCC strands slightly beyond the prestressing force and untying the steel

anchorage couplers.

Both the steel and the CFRP beams were

designed to fail in tension by yielding of steel

strands or rupture of the CFRP tendons while the

CFCC beam was designed to fail in compression

by concrete crushing. The change in the failure

mode was necessary to evaluate both tension and

compression failure of the FRP-prestressed decked bulb T beams against the common

tension failure mode of steel-prestressed beams.

2.1 Instrumentation and Test Setup Strain gauges were attached to all prestressing

strands and non-prestressing reinforcement to

measure the strain in the strands at the mid-span

section during different stages of construction

and loading. In addition, strain gauges were mounted on the concrete surface at different

locations to measure the strain in the concrete.

Furthermore, a set of three strain gauges was attached to the concrete surface at the soffit of

the beam near the first initiated flexural crack to

predict the decompression load. Measurement of the decompression load was performed to

estimate the effective prestress in the

pretensioning tendons/strands. Calibrated load-

cells were mounted on the prestressing strands to

measure the initial prestressing forces. Linear

Motion transducers were attached to the beams

to measure the vertical deflection of each beam

at the mid-span under different load levels. All

sensors were calibrated and connected to a calibrated digital data acquisition system to

monitor the deformation of the beam specimens

during loading.

2.2 Loading Test As shown in Fig.5, the decked bulb T beams

were loaded under a four-point loading setup

over an effective span of 9,450 mm. The load

was applied through incremental cycles until

failure. Two steel reinforced neoprene bearing

pads were provided at the ends of each beam as

supports. The loading test was performed to

evaluate the flexural performance of each beam

under service limit state, post-cracking limit

state, and ultimate limit state. The performance

of the beams was examined by recording the

deflection at the mid-span, strain readings in

concrete and reinforcement, crack propagation, crack width, and crack pattern at different load

levels. The following sections present a

discussion for the results obtained from three beams.

Fig.5 Four-point Loading setup

2.3 Service Limit State For the purpose of this study, the service limit

state was defined by the state, at which the

concrete beam remained uncracked. The service

limit state ended with the initiation of the first flexural crack. The first crack was observed at a

load level between 42 and 44 kN in all beams.

All the beams exhibited nearly the same elastic

performance during the service limit state. In

addition, an estimate for the effective prestress force in each beam was made by recording the

concrete strain at the soffit of the beam. First,

the beam was loaded until the initiation of the

first crack. Second, the beam was unloaded and

stain gauges were attached to soffit of the beam

near the crack location. Third, the beam was

loaded again while the strain near the crack was recorded. During the second loading cycle, the

strain in the concrete near the first initiated

flexural crack increased nearly linearly with

increasing the load until reaching a certain load

level, at which the strain experienced no further

increase with increasing the load. This load level

was identified as the decompression load, which

defined the load level where the moment due to

dead plus applied load exceeded the moment due to prestressing force. The recoded strain value at

the decompression load represented the effect of

the prestress force and was used to backwards

calculate the effective prestressing force. The

decompression load for all the three beams

ranged between 26 and 31 kN. Based on the

elastic calculations, this level of decompression

load corresponded to an effective prestressing force of approximately 80 to 90 % of the

initially applied prestressing force.

2.4 Post-Cracking Limit State This state started with the initiation of the first

flexural crack and was marked by an apparent

change in the slope in the load-deflection curves.

Several flexural cracks developed in the beams

with increasing the load beyond the cracking

load. Consequently, the beams experienced

further reduction in their flexural stiffness with

each loading/unloading cycle. The load-

deflection curves for the CFCC and the CFRP Beams were nearly linear until failure while, the

load-deflection curve for steel beam showed a

ductile plateau near the failure. At load level of

approximately 169 kN, the steel beam exhibited

a steady increase in the deflection with a little or

no increase in the load carrying capacity.

Crack width and crack pattern were recorded

and plotted for each beam. Under a certain load level, all three beams experienced nearly similar

overall flexural crack pattern. However, the

crack width was slightly different. The CFRP beam exhibited the largest crack width followed

by the CFCC beam and then the steel beam. For

example, at load level of 80 kN, the maximum observed crack width was around 0.4, 0.3, and

0.25 mm in the CFRP, CFCC, and steel beams,

respectively. The wider crack width was

interpreted into an increase in the rotation and

deflection of the beam as the CFRP beam had

the largest mid-span deflection followed by the

CFCC beam and then the steel beam. However,

the situation changed when the bottom

reinforcement of the steel beam approached the yield. At the yield of the steel beam, the flexural

cracks progressively widened and the steel beam

exhibited rapid increase in the deflection. The CFCC beam and the CFRP beam, on the other

hand, showed a gradual increase in the crack

pattern and width since the initiation of the first

flexural crack until the failure of the beam.

2.5 Ultimate Limit State and Failure The steel beam failed at ultimate load of 191 kN

due to yielding of steel strands followed by

crushing of the concrete at the top flange (Fig.6).

The measured deflection at failure was 348 mm.

The CFCC beam failed at ultimate load of 205

kN with corresponding deflection of 329 mm.

The failure was characterized by crushing of the

concrete at the top flange near the mid-span

section (Fig.7). The CFRP beam failed at ultimate load level of 169 kN with a

corresponding mid-span deflection of 359 mm.

The failure was characterized by rupture of CFRP tendons followed by crushing of the

concrete at the top flange (Fig.8). The failure

patterns in all three beams matched their

designed and anticipated failure modes.

Fig.6 Reinfocement yield and concrete

crushing failure of the steel beam

Fig.7 Compression failure of the CFCC beam

Fig.8 Tensile failure of the CFRP beam

2.6 Strength and Energy The CFCC beam achieved the highest load

carrying capacity, followed by the steel beam,

and then the CFRP beam. The load carrying

capacity of the CFCC beam was 7 % higher than

the load carrying capacity of the steel beam,

while the load carrying capacity of the CFRP

beam was 12 % less than that of the steel beam.

On the other hand, the steel beam had the highest energy absorption capacity followed by

the CFCC beam and finally the CFRP beam. The

total energy absorbed by the steel, CFCC, and CFRP beams until failure (area under the load-

deflection curves) were approximately 49, 41,

and 37 kN.m, respectively. The energy

absorption capacities of the CFCC and CFRP

beams were approximately 17 and 25 % less

than the energy absorption capacity of the steel

beam, respectively. Table 3 shows the results of

tested beams.

Table 3 Results of experimental investigation

Crack

load

(kN)

Failure

Load

(kN)

Max.

deflection

(mm)

Energy

absorbed

(kN.m)

Failure

mode

Steel

42 191 348 49 Tension

CFCC

44 205 329 41 Comp.

CFRP 44 169 359 37 Tension

3. NUMERICAL SIMULATION

A numerical investigation was conducted to

examine the performance of the tested decked bulb T beams using commercially available

software ABAQUS. The concrete beam was

modeled using a three dimensional solid element

C3D8R. A continuum, plasticity-based, damage

model for concrete was used to model the

material behavior. The concrete damaged plasticity model uses concepts of isotropic

damaged elasticity in combination with isotropic

tensile and compressive plasticity to represent the inelastic behavior of concrete. It assumes

that the main two failure mechanisms are tensile

cracking and compressive crushing of the concrete. Consequently, the concrete was

defined by its uniaxial compressive and tensile

performance in addition to the elastic properties.

For the compressive behavior, the response was

assumed linear until the initial yield, which was

assumed to occur at stress equal to

approximately 60 % of the concrete ultimate

strength. After initial yield, the material started

the plastic response, which was typically

characterized by stress hardening followed by

strain softening beyond the ultimate stress. For

the tension side, the stress-strain response followed a linear elastic relationship until the

cracking stress was reached, which corresponded

to the onset of micro-cracking in the concrete material. Beyond the cracking stress, the

formation of micro-cracks was represented

macroscopically with a softening stress-strain

response, which included strain localization in

the concrete structure (Fig.9).

For the Steel and the CFCC beams, the modulus

of elasticity for the concrete was taken as 33.8

GPa, the direct tensile strength was taken as 4.8

MPa, and Poisson’s ratio was taken as 0.2. For the CFRP beam, the modulus of elasticity for the

concrete was taken as 30.5 GPa and the direct

tensile strength was taken as 4.3 MPa. These

values were calculated based on section 5.4.2 of

AASHTO LRFD [8] for the material properties

of the concrete.

The longitudinal reinforcement, with mechanical

properties as shown in Fig.10, and the shear reinforcement were modeled with a two-node

linear 3D truss element (T3D2). The truss

elements were embedded inside the host concrete brick elements. Each node of the truss

embedded elements had three degrees of

freedom ��� , �� , ��� and these degrees of

freedom were constrained to the interpolated values of the corresponding degrees of freedom

of the host element nodes.

The numerical cracking load was approximately

53 kN for all three beams, which was slightly

higher than the exhibited experimental cracking

load (44 kN). The difference between the

experimental and numerical cracking load was

attributed to various factors on both the

experimental and numerical sides. For instance,

it was difficult to precisely evaluate the tensile

capacity of the concrete in the beam and the exact loss in the prestressing force at the time of

testing. On the other hand, the element size in

the numerical analysis influenced the cracking load. Smaller element size tended to show lower

cracking load. However, the analysis became

intractable with such small element size.

Nevertheless, the response of the numerical

models closely matched that of the tested beams

after cracking until failure. For instance, in the

numerical model of the steel beam, when the

applied vertical load reached 169 kN, the strain

in the bottom reinforcement reached 5,080 µε

and the strain in the concrete at the top surface of the beam at the mid-span reached

approximately 1,100 µε. These values of strain

matched the measured strain values in the experimental investigation just before the yield

stage, where the strain in the concrete averaged

1,200 µε and the strain in the bottom

reinforcement averaged 4,700 µε. The ultimate

load predicted by the numerical model of the

steel beam was 198 kN, with a difference of 3 %

from the experimental ultimate load. At the

ultimate load, the FE analysis indicated

excessive yielding of the steel strands before the

crushing of the concrete in the top flange.

Fig.9 Stress-strain response of concrete

material as defined in the FEA

Fig.10 Stress-strain response of

reinforcement as specified in the FEA

The FE analysis of the CFCC beam showed that

the failure load of the beam was approximately 213 kN with a difference of 4 % from the

experimentally achieved ultimate load (205 kN).

In addition, the stress in the concrete at the top

flange reached 63 MPa in the FE model, which

was in close agreement with the ultimate

compressive strength of the concrete (64 MPa).

Furthermore, the strain in the bottom CFCC

strands reached 8,500 µε. This numerical CFCC

strain at failure matched the measured strain from the experimental investigation (8,900 µε).

It should be noted that when adding this strain to

the strain due to prestressing (4,220 µε), the total strain in the CFCC strands at failure would be

12,720 µε. This strain level is less than the

ultimate strain of the CFCC strands (16,000 µε)

and thus confirms the compression failure of the

CFCC beam. Likewise, the FE analysis of the

CFRP beam predicted a failure load of 165 kN

with a difference of 2 % from the experimentally

achieved ultimate load (169 kN). At failure, the

strain in the concrete at the top flange

approached 2,200 µε and the strain in the bottom CFRP tendons approached 10,130 µε. Adding

the CFRP strain to the strain due to prestressing

after losses (6,740 µε), the total strain would be

16,870 µε. This was the ultimate strain as

defined by the manufacturer and was provided

through the input file of the FE analysis. The

experimentally measured concrete strain at

failure was approximately 2,400 µε, while the

experimentally measured strain in the CFRP tendon before failure (excluding the strain due to

prestressing) was around 10,900 µε.

The experimental vs. numerical load-deflection

curves for all three beams are given in Fig.11.

The load-deflection curves obtained from the FE analysis of all three beams were in close

agreement with those obtained from the

experimental investigation. Therefore, it is

reasonable to extend the FE investigation to

model complete decked bulb T beam bridge

models with different geometries and different

loading configurations.

Fig.11 Load-deflection curves for tested

beams (experimental vs. numerical)

-70

-60

-50

-40

-30

-20

-10

0

10

-0.004 -0.003 -0.002 -0.001 0 0.001 0.002 0.003

Str

ess

(MP

a)

Strain (µε)

0

500

1,000

1,500

2,000

2,500

3,000

0 0.01 0.02 0.03 0.04 0.05

Str

ess

(MP

a)

Strain (µε)

Steel strands

CFCC strands

CFRP tendons

0

40

80

120

160

200

0 50 100 150 200 250 300 350

Lo

ad (

kN

)

Deflection (mm)

Steel-Exp.

CFCC-Exp.

CFRP-Exp.

Steel-FEA

CFCC-FEA

CFRP-FEA

4. CONCLUSIONS

Based on the results obtained from the

experimental investigation and the numerical

simulation, the following conclusions are drawn: 1. Under service limit state, the flexural

performance of the decked bulb T beams

prestressed with CFCC strands or CFRP tendons was comparable with the

performance of beams prestressed with

steel strands. No significant difference

was observed between tested beams.

2. Beyond service limit state, flexural

crack pattern and cracks spacing were

nearly similar in all tested beams. On the

other hand, the crack width in CFRP and

CFCC beams was slightly larger than

that in the steel beam. This suggests that flexural distress signs of the FRP-

prestressed decked bulb T beams are

similar to those of the steel-prestressed

beams but larger deflection is expected

in FRP-prestressed beams.

3. The flexural load carrying capacity and

the corresponding maximum deflection

of the CFCC beam were 107 % and 94

% of those of the steel beam, respectively. On the other hand, the

flexural load carrying capacity and

corresponding maximum deflection of the CFRP beam were 88 % and 103 %

of those of steel beam, respectively. In

addition, the total energy absorption capacity of the CFCC and the CFRP

beams were 84 % and 76 % of the total

energy absorption capacity of the steel

beam, respectively. Therefore, it is

reasonable to conclude that the overall

flexural performance of the CFCC and

the CFRP beams was comparable with

the flexural performance of the steel

beam with respect to the load carrying capacity and the deformation. However,

the FRP reinforced beams have the

advantage of corrosion resistance over the steel reinforced beams.

4. Numerical models for all the tested

beams accurately predicted the cracking

pattern, deflection, ultimate load, and

failure modes. Therefore, the numerical

approach can be adequately employed in

the design of the decked bulb T beams

reinforced and prestressed with FRP

materials.

ACKNOWLEDGMENT

This investigation was sponsored through a

consortium assembled of the National Science

Foundation, (Award No. CMMI-0969676), Michigan-DOT Center of Excellence, US-DOT

(Contract No. DTOS59-06-G-00030), Tokyo

Rope MFG. CO. LTD., Japan, and Diversified Composites, Inc. KY, U.S.A. The authors

gratefully acknowledge their supports.

REFERENCES

[1] Washington State Department of

Transportation, 2008, “Bridge Standard

Drawings,” online version, http://www.

wsdot.wa.gov/eesc/bridge/drawings/inde

x.cfm?fuseaction=drawings&section_nbr=6&type_id=24

[2] Hill, J.J.; McGinnis, L.G.; hughes,

W.R.; and Shirole, A.M., 1988, “Design

and Construction of Transversely

Posttensioned Concrete Bulb Tee Beam

Bridge,” Transportation Research

Record, Issue 1180, Washington, DC,

U.S.A.

[3] Tokyo Rope, 2007, “CFCC Quality Report” Tokyo Rope Mfg. Co. Ltd.,

Japan.

[4] Leadline™ Carbon Fiber Tendon/ Bars, October 1994, product manual,

Mitsubishi Chemical Corporation,

Tokyo, Japan. [5] Grace, N. F.; Navarre, F.; Nacey, R.;

Bonus, W.; and Collavino, L., 2002,

“Design-Construction of Bridge Street

Bridge-First CFRP Bridge in the United

States”, PCI Journal, Sept.-Oct., pp. 20-

35.

[6] Rohleder, Jr.; W.; Tang, B.; Doe, T. A.;

Grace, N. F.; and Burgess, C. J., 2008,

“Carbon Fiber-Reinforced Polymer Strand Application on Cable-Stayed

Bridge, Penobscot Narrows, Maine”,

Journal of the Transportation Research Board, Vol. 2050, pp. 169-176.

[7] Diversified Composites, Inc., 2004, DCI

Tendon Product Specifications,

Erlanger, KY, U.S.A. (www.

Diversified-Composites.com).

[8] American Association of State

Highways and Transportation Officials

(AASHTO), 2007, “AASHTO LRFD

Bridge design specification.”

Washington, D.C., U.S.A.