Final report: shear friction and interface shear of ultra high performance concrete [June 2010]

School of Civil and Environmental Engineering
Structural Engineering, Mechanics and Materials Research Report No. 2010-2
Shear Friction & Interface Shear of Ultra High Performance Concrete
Final Report
Prepared for Office of Materials and Research Georgia Department of Transportation GDOT Research Project No. 07-05
Task Order No. 02-37
by C. Kennan Crane and Lawrence F. Kahn
June 2010

Contract Research Task Order No. 02-37 GDOT Research Project No. 07-05
Final Report:
Shear Friction & Interface Shear of Ultra High Performance Concrete
Prepared for Office of Materials and Research Georgia Department of Transportation
by
C. Kennan Crane and Lawrence F. Kahn
June 2010
The contents of this report reflect the views of the authors who are responsible for the facts and accuracy of the data presented herein. The contents do not necessarily reflect the official views
or policies of the Georgia Department of Transportation. This report does not constitute a standard, specification or regulation.

Executive Summary
A multi-scale research approach was used to experimentally investigate interface shear and diagonal tension shear in composite Ultra-high Performance Concrete (UHPC) bridge girders. Thirty-eight shear friction push-off specimens were made using monolithic UHPC and using UHPC with cold joints between it and high-strength concrete. Five 10-ft. long precast UHPC sections were made with varying surface conditions, and high strength decks were cast on top. Three full-scale bulb-tee (BT) shaped girders 32.4-in. deep and between 34-ft. and 54-ft. long were constructed using UHPC with as-cast (smooth) and fluted top surface conditions along with varying amounts of shear reinforcement. High strength 8-in. thick decks were cast atop these three girders.
The UHPC specimens had compressive strengths between 26,000 psi and 30,000 psi. The high strength concrete strengths ranged between 8,000 psi and 12,000 psi.
The shear-friction and girder tests demonstrated that current ACI and AASHTO provisions are conservative for estimating interface shear capacity and shear friction of composite UHPC/HPC structures and of monolithic UHPC. Formliners can be successfully used to create a fluted surface finish in UHPC that is comparable to the 6 mm (1/4-in.) surface roughness recommended by current codes for composite construction. This surface finish provides significant increase in interface shear capacity, particularly when used in conjunction with reinforcing steel crossing the interface. In monolithic UHPC, steel fibers allow for significant shear transfer across pre-existing cracks even when no additional shear reinforcement is used; however, smooth cold joints in UHPC have very low interface shear capacities when transverse shear reinforcement is not used.
Both the push-off tests and small beam tests showed that for smooth, cold joint UHPC surfaces, a friction coefficient = 0.6 be used and that for a fluted cold joint between HPC and UHPC a friction coefficient = 1.0 be used. It was found for the latter that friction coefficient = 1.4 was safe, but more testing is recommended before a value larger than that given in AASHTO (2007) be used.
The seven tests of the three full-size girders showed that most of the sections, even those with no stirrups, failed in compression of the girder which demonstrated that the UHPC possesses outstanding diagonal tension, shear capacity. The wider 18-in. flange of Girders 1 and
ii

2 was necessary to develop the full flexural strength of the section. The narrow flange of Girder 3 diminished the ultimate strength of the section and lead to the compression, flexural failure of both 3-1 and 3-2.
It is recommended that the current AASHTO code (2007) be used for the design of interface shear for UHPC-HPC composite bridge girders. Further, it is recommended that the top surface of UHPC girders be fluted (roughened) to assure adequate service load performance. Wide top surface girders are needed to develop the fluted surface, so BT type girders are recommended.
While shear reinforcement is not needed to provide diagonal tension shear strength of UHPC girders, it is recommended that more than minimum shear reinforcement be used to connect the HPC deck to the UHPC girder.
iii

Acknowledgments
The Georgia Department of Transportation provided the financial support for this project through Research Project No. 07-05, Task Order No. 02-37. The opinions and conclusions expressed herein are those of the authors and do not represent the opinions, conclusions, policies, standards or specifications of the Georgia Department of Transportation or of other cooperating organizations.
Many GDOT employees were very helpful and provided information and reviewing the report. We specifically acknowledge Mr. Paul Liles, State Bridge Engineer, and Ms. Supriya Kamatkar, Research Engineer.
Tindall Corporation personnel, especially Mr. Kevin Kirkley, assisted in all aspects of the casting of the UHPC specimens. Mr. Vic Perry and Mr. Peter Calcetas of Lafarge NorthAmerica gave many valuable suggestions and supported the placement of UHPC at Tindall. The support of Tindall Corporation and Lafarge North-America and of their personnel is gratefully acknowledged.
Dr. Ben Graybeal of the Federal Highway Administration was very helpful in providing guidance and advice concerning mixing, placing, curing and testing of UHPC. The following Georgia Tech research assistants aided in the study: John Bennett, Jennifer Dunbeck, Murat Engideniz, Luis Fajardo, Roshanak Gharaat, Kate Howard, Danielle Simpson, and Katherine Snedecker.
iv

Table of Contents
Executive Summary.................................................................................ii Acknowledgments...................................................................................iv Chapter 1: Introduction...........................................................................1-1 Chapter 2: Background...........................................................................2-1 Chapter 3: Experiment Design...................................................................3-1 Chapter 4: Specimen Construction.................................................................4-1 Chapter 5: Shear-friction Push-off Tests.......................................................5-1 Chapter 6: Interface Shear Composite Beams.................................................6-1 Chapter 7: Full-scale Bridge Girders............................................................7-1 Chapter 8: Conclusions and Recommendations...............................................8-1 References........................................................................................R-1
v

Notation
= friction factor; 1.4 for a monolithic concrete connection; 1.0 for a surface intentionally roughened to amplitude of 6mm (1/4 in.); 0.6 for a smooth cold joint
Acv = area of concrete engaged in shear transfer Avf = area of interface shear reinforcement c = cohesion factor: 2.758 MPa (0.4 ksi) for a monolithic concrete connection; 1.655 MPa (0.24
ksi) for a surface intentionally roughened to amplitude of 6mm (1/4 in.); 0.517 MPa (0.075 ksi) for a smooth cold joint fc' = compressive strength of concrete determined using 4x8-in. cylinders at date of test fy = design yield strength of steel reinforcing bars not be taken as greater than 413.7 MPa (60 ksi) K1= fraction of concrete available to resist interface shear: 0.25 for a monolithic concrete connection; 0.25 for a surface intentionally roughened to amplitude of 6mm (1/4 in.); 0.2 for a smooth cold joint K2=limiting interface shear resistance: 10.342 MPa (1.5 ksi) for a monolithic concrete connection; 10.342 MPa (1.5 ksi) for a surface intentionally roughened to amplitude of 6mm (1/4 in.); 5.516 MPa (0.8 ksi) for a smooth cold joint Pc = net compressive force normal to shear plane Vni = nominal interface shear resistance ni = nominal shear stress Vni/Acv ( 11.0 MPa [1.6 ksi]) v = shear friction reinforcement ratio Avf/Acv
vi

1. Introduction
1.1 Purposes and Objectives
The two purposes of this research study were the following: (1) determine the interface shear and shear friction capacity between ultra high performance concrete (UHPC) used for bridge girders and high strength concrete used for deck slabs so that the connection between a precast prestressed concrete girder and its cast-in-place deck may be designed accurately; and (2) determine the shear friction capacity of UHPC by itself so that the post-cracking strength of UHPC in the shear zone of precast UHPC girders may be calculated and the shear design of UHPC girders may be accurately and conservatively developed.
Specific objectives were the following: (1) use traditional shear friction push-off specimens to gain shear friction parameter for use in AASHTO shear friction relations for monolithic UHPC, (2) use traditional shear friction push-off specimens to gain shear friction parameter for use in AASHTO shear friction relations for UHPC-to-high performance concrete (HPC) cold joints between UHPC bridge girders and HPC cast-in-place deck, (3) use small beams to determine interface shear between cast-in-place deck and precast UHPC beams, (4) use full-size bridge girders with varying top surface and shear reinforcement to calibrate shear friction tests for determining the applicability of AASHTO design relations for use with UHPC bridge girders.
1.2 Scope
The research was conducted in three phases as detailed in Chapter 3. In the first phase 38 push-off specimens with different surface properties and different amounts of reinforcement crossing the interface were tested. The shear planes were smooth or roughened cold joints, or monolithic sections with or without a pre-existing crack. The transverse
1-1

reinforcement ratio varied between 0% and 0.75%. These specimens matched traditional shearfriction push-off specimens so that their results could be readily compared to past research.
In the second phase, 5 small scale 10-ft. long composite beams were built to compare interface shear with the results from the push-off tests. The 10-in. deep, 4-in. wide rectangular UHPC section was precast and thermally treated. Different top surface roughness and reinforcement conditions were used. A 5 -in. thick deck made of normal strength concrete was cast atop the precast UHPC sections. The five beams were tested in four-point bending.
In the third phase, three full size BT-32 UHPC precast prestressed girders were cast. Each girder was tested twice to determine interface shear and diagonal tension shear capacity. The surface condition and amount of shear reinforcement varied at each end of each beam; so, the six tests were different.
In all cases, the UHPC was made with Lafarge Ductal premix utilizing 2% by volume steel fiber reinforcement.
1.3 Significance of Research
New ultra high performance concretes (UHPCs) have begun being used in precast prestressed concrete bridge girders. The high strength and durability of these concretes make them attractive alternatives to steel when low maintenance and a 100-year lifespan are desired. Specifically, when looking to replace steel bridge girders with girders of approximately equivalent capacity, depth, and weight, UHPC is one of the only options currently available. Highly optimized beam cross-sections have been created for use of this material, but their construction requires sophisticated formwork that is both expensive to manufacture and more difficult to use than formwork for traditional I-shaped girders. Therefore, it is necessary to determine if significant benefit can be gained from using this new material with traditional precast prestressed I-girder construction.
Currently, bridges constructed from precast prestressed I-girders usually rely on composite action with a cast-in-place concrete deck for carrying live loads. This connection has been created by two means: 1) roughening of the girder surface to 1/4 inch by raking to expose coarse aggregate, and 2) extending shear reinforcement from the girder into the deck. The selfconsolidating nature of UHPC prevents the surface from being roughened by raking. Also, according to the manufacturer, UHPC does not require shear stirrups to resist shear forces
1-2

because of the steel fibers in the concrete. In light of these differences, the effects of varying surface preparations and transverse reinforcement on interface shear capacity of UHPC must be evaluated.
The interaction between a UHPC girder and cast-in-place concrete deck must be understood if this structural system is to be used safely. Specifically, the flexural and shear performance of the composite system must be studied. The shear performance may be further divided into the interface shear performance of the connection between the girder and the deck and the diagonal tension shear performance of the girder. The interface shear performance will determine how to design for composite action in systems containing UHPC. The diagonal tension shear performance will validate the manufacturer's claim that no shear reinforcement is necessary in UHPC bridge girders. Few tests have been performed on the shear performance of UHPC bridge girders and none on interface shear across the bonded connection between a precast UHPC section and poured-in-place high performance or ultra high performance concrete.
Most importantly, we emphasize that UHPC has the potential to provide an economic alternative to steel girders by allowing a prestressed concrete beam made with UHPC to have as small a depth as a steel girder, be fully reinforced with prestressing strands, be extremely durable, and require no maintenance. It is predicted that a UHPC girder would not require shear reinforcement except for an amount required to connect the cast-in-place deck slab to the girder. Overall, UHPC has the potential to provide maintenance free, long-life bridge structures which would result in millions of dollars of savings.
1-3

2. Background
2.1 Material Characterization of Ultra High Performance Concrete
Ultra High Performance Concretes (UHPCs), sometimes referred to as Reactive Powder Concretes (RPCs), comprise a class of concretes with compressive strengths above 22 ksi (150 MPa) made by elimination of large aggregate and refinement of the concrete microstructure (Shah and Weiss, 1998). These materials are designed using particle-packing models to densify the material microstructure leading to low porosities and high strengths (Delarrard and Sedran, 1994).
Currently, the only UHPC widely available in North America is Ductal by Lafarge, Inc. (Graybeal, 2004). As part of the Federal Highway Administration's research in UHPC, extensive material property tests of this material have been performed in the general areas of strength, durability, and long-term stability (Graybeal, 2006a). The research showed very high compressive strengths of 28.9 ksi (199 MPa) for UHPC thermally treated at 195F (90C) for 48 hours immediately following demolding as specified by Lafarge. The modulus of elasticity of the thermally treated material was 7600 ksi (52.5 GPa). Tensile strengths ranged from 1.2-1.7 ksi (8.3-11.7 MPa) for thermally treated UHPC. The variation in observed tensile strength was due to different tensile testing procedures.
Graybeal showed the compressive and tensile strengths as well as the modulus of the material to be highly dependent on the curing method (2006a). For example, when no thermal curing was used, 28 day compressive strength was only 17.2 ksi (119 MPa). When thermal treatment was delayed until 28 days after casting, the 30 day compressive strength was 24.7 ksi (170 MPa). When thermal treatment was performed at 140F (60C) for 72 hours, the 28 day compressive strength was 21.3 ksi (147 MPa). These compressive strengths are 15-40% lower than that seen when the recommended curing procedure was followed. Modulus decreased by 219% when the recommended procedure was not used. Tensile capacity decreased by up to 30% when no thermal treatment was used, but did not appear to be affected by type of thermal treatment.
2-1

Victor Garas (2009) recently completed a thesis which details short and long-term compressive and tensile properties of UHPC.
2.2 Interface Shear
Very limited shear-friction or interface shear testing has been performed on UHPC, and no tests of this kind have been performed on UHPC beams or girders. Two main types of interface shear tests were developed by Hanson (1960) for his research on composite action of precast prestressed bridge girders with cast-in-place decks performed for the Portland Cement Association. Since then, different forms of these same tests have continued to be the basis for the majority of research in interface shear. Push-off tests directly measure the shear capacity of an interface between two concretes by creating a shear force between two compositely cast concrete blocks. Tests of composite beams create shear across the interface between a precast beam and cast-in-place deck by bending the composite beam in flexure.
2.2.1 Push-off tests
The original specimens (Figure 2-1) used for interface shear push-off tests were designed by Hanson and had interface areas varying from 48 to 192 in2 (Hanson, 1960). The width of interface in shear was held constant at 8 in. and the length was varied between 6, 12, or 24 in. The amount of steel crossing the interface was varied between reinforcement ratios of .002 and .008. Both smooth and roughened surface preparations were investigated in both bonded and unbounded setups. In the unbounded setups, a silicone compound was applied to the surface of the girder prior to casting of the slab. Various means of removing surface paste and the use of shear keys were also evaluated in some specimens.
2-2

Figure 2-1: The push-off specimens used by Hanson (1960) varied the interface area between 48 and 192 in2. The reinforcement ratios varied between .002 and .008.
Figure 2-2 shows the typical stress-slip curves observed by Hanson (1960) for various types of interfaces. In reporting the interface shear capacity of the push-off tests, the contribution of the steel stirrups was removed by subtracting a reference curve with the same amount of stirrups and a smooth unbounded interface from each experimental data set. This method was utilized in order to be able to compare data sets more directly; however, by subtracting the steel contribution, it was implicitly assumed that there were no linked effects of reinforcement and surface preparation. This assumption has been challenged by most researchers and a linked effect between surface preparation and reinforcement ratio is considered by both current ACI and AASHTO codes.
2-3

Figure 2-2: Typical stress-slip curves for push-off tests by Hanson (1960). Note that the effect of stirrups has been removed from these curves.

Based on his work, Hanson proposed a preliminary equation of the form:

v = A +17,500 psi,

(1)

where A is a constant based on surface roughness (300 psi for smooth, 500 psi for rough) and

is the shear reinforcement ratio across the interface. It should be noted that this equation was not

presented as a suggested design equation, but rather as an observation. Hanson further

2-4

concluded that roughened bonded contact surfaces were the only advisable interface between precast beams and cast-in-place slabs.
Anderson (1960) performed similar tests on a slightly different specimen shape as shown in Figure 2-3. The advantage of this test shape is that it allows for the shear loads to be applied directly opposite one another, eliminating any moment due to load eccentricity. The interface area for all of these tests was100 in2. These tests varied the reinforcement ratio across the interface between 0.002 and 0.0248. The effect of concrete strength was also evaluated. For the precast side of the push-off, the concrete strength was 7,500 psi for all specimens. For the castin-place portion of specimens, the concrete strength was either 3,000 psi or 7,500 psi. The precast surface of each push-off specimen was roughened prior to setting. Immediately before casting of the other half of the specimen, the roughened surface was coated with a neat cement slurry.
Figure 2-3: The push-off specimen used by Anderson (1960) had a 100 in2 interface area and varied the reinforcing ratio between .002 and .0248.
The results of Anderson's tests are shown in Figure 2-4. Two clear trends can be seen from this data. As the concrete strength of the weaker concrete (in this case the cast-in-place concrete) increases, the interface shear capacity increases. Also, as the reinforcement ratio
2-5

increases, the interface shear capacity increases. The trend lines provided by Anderson for 3000

psi concrete and 7500 psi concrete are approximately

v = 650 + 34, 000 (psi)

(2)

and

v = 800 + 41, 000 (psi)

(3)

respectively.

Where

v = interface shear stress capacity

= shear friction reinforcement ratio Avf/Acv, area of stirrup reinforcement divided by area of concrete interface

These equations cannot be directly compared to Equation 1 because Hanson's data attempted to remove the effect of steel.

Figure 2-4: Push-off results from Anderson (1960) compare steel ratio to the ultimate interface shear stress in the concrete. Two different concrete strengths were used for the cast-in-place
portion of specimens.
2-6

Birkeland and Birkeland proposed a shear friction hypothesis for explaining the mechanics of interface shear transfer (Birkeland and Birkeland, 1966). This theory views interface shear resistance as being provided by friction across a roughened surface. The friction force is therefore the product of the normal clamping force across the interface and the tangent of the contact angle across the surface. This idea is depicted visually in Figure 2-5.

T

T

V

V (=T tan )

V T

T N
T tan

Figure 2-5: Shear friction hypothesis (Birkeland and Birkeland, 1966) In the equation

,

the force, T, is the force provided by the steel reinforcement crossing the interface, Asfy. The term, tan , can be thought of as the coefficient of friction, . This gives the shear friction equation as

, or

(4)

. where,

Vu = Interface ultimate shear strength, lbs. vu = interface shear stress, psi As = area of transverse steel reinforcement crossing interface, in2 s = ratio of transverse reinforcement to concrete interface area s = shear friction coefficient fy = yield stress of reinforcement, psi

This is the basic form of the equation currently used by both ACI 318-08 and AASHTO LRFD 2007. The friction factor, , varies depending on the surface preparation of the interface between the two concretes. Birkeland and Birkeland (1966) suggested values of 1.7, 1.4, and 0.8-1.0 for monolithic concrete, intentionally roughened construction joints, and smooth construction joints respectively.

2-7

The shear friction equation was validated by comparison to the work of both Hanson and Anderson (Hanson, 1960; Anderson, 1960; Birkeland and Birkeland, 1966). In order to compare these data sets, Birkeland and Birkeland added the steel effect into Hanson's data. Figure 2-6 shows these data plotted along with the values predicted by the shear friction equation. Based on unpublished tests by Mast, an upper bound of 800 psi was suggested when the shear friction equation was used.
Figure 2-6: Shear friction equation compared to experimental data for push-off tests and girder tests from Birkeland and Birkeland (1966).
2-8

Mast (1968) expanded the use of the shear friction equation from existing interfaces to potential cracks. Using this design method, a notional crack is assumed at a given location and the reinforcement is designed across that interface using shear friction. This process allows for safe design even when unplanned cracks are present. The values of tan suggested by Mast are 1.4, across a crack in monolithic concrete, 1.4 across roughened interfaces, and 0.7 across smooth interfaces.
Mast (1968) also suggested a change to the way interface shear stresses were calculated. Previously, most concrete researchers had calculated the shear stress at an interface via

(5)

where,

v = interface shear stress, psi
V = Interface shear strength force, lbs. I = Section moment of inertia in4 Q = First moment of compression area about neutral axis, in3
b = Width of interface, in

Mast noted that this equation was invalid outside of the elastic range and could not be used for shear stress calculations at ultimate. Instead, Mast suggested a force balance approach whereby the tension in the steel must be resisted by compression in the concrete. Therefore, if the compression block is on one side of the interface and the tension steel is on the other, the shear stress across the interface is simply the tension force (or compression force) divided by the interface area:

(6)

where,

v = interface shear stress, psi
T = Tension force in flexural reinforcement, lbs. Ai = Area of interface, in2

Hofbeck et. al. (1969) tested Mast's theory that shear friction could be used to calculate interface shear capacity across a crack in monolithic concrete. Thirty-eight push-off specimens were cast similar to those used by Anderson (see Figure 2-3). Instead of being cast in two phases with a roughened cold joint between, the specimens were cast monolithically. Most of the specimens then had a crack induced along the desired shear plane by placing opposing line loads on the section until a crack formed. Some of the monolithic specimens were left uncracked in

2-9

order to evaluate whether push-off tests can be used to determine shear performance of monolithic concrete.
Figure 2-7 shows the results of Mast's (1968) tests plotted along with the shear friction equation proposed by Birkeland and Birkeland(1966). When a value of 1.4 was used for tan and the upper bound of 800 psi suggested by Birkeland and Birkeland was used, the shear friction equation was shown to provide a conservative lower bound for shear transfer across preexisting cracks. It was also shown that, when uniaxial compressive capacity of the concrete was

taken to be

, and the uniaxial tensile capacity of the concrete was taken to be

, the

results of the monolithic push-off tests very closely followed the Mohr circle failure envelope

proposed by Zia (1961).

Figure 2-7: Results of precracked push-off shear tests compared to shear friction equation from Hofbeck et. al. (1969).
2-10

Mattock and Hawkins (1972) used a pull-off specimen and a modified push-off specimen

to study the effects of tension parallel to the shear plane and additional compressive forces

perpendicular to the shear plane. The pull-off specimens were similar to the push-offs

previously used except that they had steel end pieces that allowed them to be tested in tension.

The modified push-off tests created additional compression perpendicular to the shear plane by

angling the shear plane from the load direction. The results of the pull-off tests showed that

tensile stresses perpendicular to the shear plane decreased the shear capacity of monolithic

specimens, but had no effect on the capacity of precracked specimens. The results of the

modified push-off tests showed that additional compressive forces perpendicular to the shear

plane can be considered as additive to the clamping force provided by the steel reinforcement

when using the shear friction equation. This research also suggested including a cohesion term

in the shear friction equations when considering cracks in monolithic concrete given that certain

minimum amounts of clamping force were maintained. Based on this suggestion, the shear

friction equation suggested for design becomes:

( ) vu = 200 + 0.8 f y + Nx

(7)

where

Nx = Stress normal to shear surface, compression positive, psi

Hermansen and Cowan (1974) derived a similar formula, but suggested a cohesion term of 580 psi for monolithic concrete based on evaluation of corbel tests by Kriz and Raths (1965). In response to this work, Mattock (1974a) pointed out that a cohesion term for monolithic concrete goes against the original basis for shear friction, i.e. that a pre-existing crack may form anywhere. Somerville (1974) also looked at the addition of a cohesion term for use with the shear friction equations when designing corbels, but noted that more research must still be done to determine what this term is at low reinforcement ratios.
Mattock (1974b) further tested the validity of the shear friction hypothesis when applied to reinforcement crossing the shear plane at an angle. Twenty-three push-off tests were performed with stirrups at various angles to the shear plane. Some of these tests included two sets of stirrups that were at right angles to each other at various angles to the shear plane. It was found that the shear friction equations can adequately be used to predict behavior of shear

2-11

interfaces with steel at an angle to the shear plane if both the parallel and perpendicular components of the forces in these bars are included in the shear friction equation.
Paulay and Loeber (1974) examined the specific effects of aggregate interlock in transferring shear across an interface. Pre-cracked monolithic push-off specimens were created in which crack width opening could be controlled via external restraint. Three types of tests were performed as part of this research. For the first set of tests, push-off specimens with fixed crack width were tested under monotonic loading until failure occurred. The second set of tests used similar fixed width cracks, but tested the specimens under cyclical loading and unloading. The third set of tests increased the crack width as load was being applied. From these tests, it was determined that crack width was the single most important factor in determining interface shear performance of concrete push-off specimens.
Further tests by Paulay et. al. (1974) investigated the interaction of dowel action, surface roughness, reinforcement ratio, and cyclic loading on the push-off specimen shown in Figure 28. This specimen type is symmetric so that loading can be reversed in order to study the effect of cyclical loading of the interface. The portion of the specimen below the dashed line was precast and the remainder was cast-in-place. This research concluded that because of the large deformation required to develop dowel action, dowel action cannot be counted on for design purposes. This research also supported previous findings that roughening of the interface between precast and cast-in-place concrete is the best method for preventing large displacements in composite construction.
P
P
Figure 2-8: Push-off specimen for cyclic loading of interface. The dashed line represents the interface between precast and cast-in-place concretes (Paulay et al., 1974).
2-12

Mattock et. al. (1975) studied the combined effects of moment and shear or tension and shear on interface shear capacity. Push-off tests with intentionally eccentric loads were used for creating moment and shear across the interface. For examining the effect of tension, push-off specimens were created with anchor bolt inserts that could be simultaneously be pulled during the interface shear test. The results of these tests showed that, while moments less than the ultimate capacity of the section do not decrease interface shear capacity, shear steel should be placed in the tension region of the interface if it is to be fully effective.
Mattock, et. al. (1976) expanded the research in interface shear transfer to include sanded-lightweight and all lightweight concretes. They concluded that lightweight concretes had lower interface shear capacities than normal weight concretes of equivalent compressive strength. They recommended reducing the shear friction design capacity by adding a multiplier of 0.75 for all-lightweight concrete or 0.85 for sanded lightweight concrete.
Shaikh (1978) suggested a parabolic equation for the relationship between clamping stress and ultimate shear stress for the PCI design guideline. The proposed equation was

( ) vu = 1000 v fy psi

(8)

where all symbols are the same as previously identified.

The author noted that the units do not work out cleanly, but the curve fits the data more closely than the previously suggested linear equation.
Walraven (1981) suggested that the main mechanism for shear transfer across an interface was aggregate interlock. This aggregate interlock was either due to the protrusion of aggregates on a roughened interface or cracks propagating around aggregates (rather than through them) if a pre-existing crack was assumed. Using a mathematical model, the shear friction across an interface was related to the volume fraction, gradation, and size of the large aggregate. The model then derived a distribution of protruding spheres from the large aggregate distribution. Finally a contact model between protruding spheres of aggregate and indentations in the paste was used to predict shear capacity. This model was compared to data from push-off tests with varying aggregate fractions, gradations, and sizes. The model was observed to match the experimental data when a coefficient of friction of 0.4 was used. Note that this is different to the coefficient of friction in other models because it is the coefficient around each individual aggregate rather than across the interface. This changes the assumed geometry significantly and
2-13

prevents this coefficient of friction from being compared with those suggested by previous researchers.
Walraven, et. al. (1987) evaluated the influence of concrete strength, load cycling, and load duration on interface shear capacity. Load duration and load cycling were found to not have significant effect on the ultimate interface shear capacity when the prolonged load or cyclic load were kept below 82% and 66% of the ultimate static interface shear capacity respectively. Concrete strength, however, was found to have a large influence on interface shear capacity. A design chart was presented that could be used for including the influence of concrete strengths between 2,500 and 9,000 psi when clamping force across the interface was between 100 and 1500 psi. In discussion of this research, Mattock (1988) noted that this table was slightly unconservative for some previously published data and that explicit consideration of concrete strength may make design equations overly cumbersome.
Bass et. al. (1989) examined the effect of various surface preparations of existing concrete in order to determine the applicability of shear friction equations to interfaces between existing and new concretes. Unlike planned composite construction where the surface of the precast element can be roughened prior to set, this research evaluated sandblasting, chipping, and making shear keys in existing concretes prior to casting new concrete against the surface. For some specimens, holes were also drilled to varying depths in the existing concrete and dowels were epoxied into the holes prior to placement of new concrete. The results of the tests showed that 6 in. embedment was preferable for development of full shear capacity, but otherwise, shear friction equations provided a conservative estimate of experimental behavior.
Walraven and Stroband (1994) examined the use of shear friction equations for predicting interface shear performance in high-strength concretes. Previous research had been limited to concrete with cylinder strengths below 9,000psi. It was observed that, in higher strength concretes, the cracks tended to propagate through aggregates rather than around them. This crack mechanism does not allow for aggregate interlock to transfer shear. Walraven and Stroband evaluated the applicability of shear friction equations for concretes with cylinder strengths up to 14,500psi. In push-off specimens with these higher concrete strengths, lower values of interface shear resistance are observed, or rather, the increase in concrete strength did not correlate to an equivalent increase in interface shear capacity. The basic form of the shear
2-14

friction equation used in codes still provided an adequate lower bound for performance. This research did not suggest any modifications to the existing shear friction equations.
Ali and White (1999) proposed a contact model for calculation of interface shear capacity that included the effect of cracking aggregate in high-strength concretes. This mathematical model predicted behavior across concrete strengths very accurately, but was too cumbersome for design use. The simplified form of the equation was still shown to be more accurate than the basic shear friction equation, but it was still unconservative for high values of f'c, with measured/theory values between 0.7 and 1.0 for compressive strengths of 13,600psi.
Valluvan et. al. (1999) evaluated the upper limits placed on the shear friction equation:

. Based on new research and an extensive evaluation of previous

research, it was suggested that the upper limits be changed to

. This

suggested change significantly decreased the overconservatism in the shear friction equations for

the presented data without giving unconservative estimates.

Hwang et. al. (2000) suggested a departure from shear friction theory in calculating

interface shear capacity. Instead, they showed how a softened strut-and-tie model can be used to

predict failure along the interface by a mechanism similar to shear in monolithic concrete. This

method has not been adopted by current codes, presumably because of the significant increase in

complexity of the equations. The authors also note that method presented for calculating the

shear angle is not applicable to high-strength concretes or highly roughened interface surfaces.

Mattock (2001) suggested a change to the form of the shear friction equation that could

account for some contribution of concrete strength, making the predicted values of interface

shear in high-strength concrete better match experimental data. A variable related to concrete

strength and cohesion was defined for existing cracks as K1 = 0.1fc' but not more than 800psi.

When ,

v

f y

<

K1 1.45

psi

insufficient data exists to consider the effects of concrete strength, so

the shear friction capacity across an existing crack is taken as

( ) vn = 2.25 v fy + Nx psi

(9)

When reinforcement ratios are higher, the shear friction capacity across an existing crack is

( ) vn = K1 + 0.8 v f y + Nx psi

(10)

but not greater than K1 fc' nor K2 psi,

2-15

where v is the ratio of shear friction reinforcement to the contact area, Nx is the normal stress, K1 is defined above, K2 = 0.3, and K3=2,400psi. For normal weight concrete cast against intentionally roughened normal weight concrete, equations 8 and 9 can still be used, but K1 = 400psi, K2 = 0.3, and K3=2,400psi. For sand-lightweight concrete, K1 = 250psi, K2 = 0.2, and K3=1,200psi. For all lightweight concrete, K1 = 200psi, K2 = 0.2, and K3=1,200psi. For concrete placed against concrete but not intentionally roughened,

vn = 0.6 Avf f y psi

(11)

but not more than 0.2f'c nor 800 psi. where is defined by ACI 318, and Avf is the area of shear friction reinforcement.
Kahn and Mitchell (2002) tested 50 push-off specimens with concrete strengths up to

17,000 psi. These tests were performed using specimens (Figure 2-8) with interface areas of 60 in2 and reinforcement ratios between .0037 and .0147. Roughened surfaces as well as

precracked and monolithic shear planes were evaluated. In all cases, the shear friction equation

(4) with the upper bounds of 800 psi and 0.2f'c imposed by ACI318-99 was found to be a conservative predictor of behavior. It was suggested that the upper bound of 800 psi be

removed, but the upper bound of 0.2f'c be kept in order to make better use of increased concrete strengths. An alternative equation that more closely matched test data was also proposed:

vn = 0.05 fc '+1.4v f y 0.2 fc ' psi

(12)

The effect of steel fibers in monolithic steel-fiber reinforced concrete has also been evaluated by push-off testing (Barragn et al., 2006). In these tests, concrete blocks were monolithically poured and then cuts were made to form a shear push-off specimen. This preparation method prevented preferential fiber alignment around extraneous formwork as well as wall effects. The specimens were then tested with LVDTs (Linearly Variable Displacement Transducers) measuring both slip and crack width opening. Two different mixes with different concrete strength, fiber type, and fiber volume ratio were used. Increasing fiber volume fraction showed increased shear capacity and shear toughness for both normal and high-strength steel fiber reinforced concretes as shown in Figure 2-9. The increased ability of concretes to continue carrying shear over large crack openings was of particular interest. The authors noted that fiber orientation is a critical parameter in determining this interface shear strength, so push-off specimens should be formulated to mimic fiber orientation in actual members.

2-16

Figure 2-8: The push-off specimens used by Kahn and Mitchell (2002) had interface areas of 60 sq. in and reinforcement ratios v varying from .0037 and .0147.
.
2-17

Figure 2-9: Shear stress versus slip of high performance concrete with varying amounts of steel fiber. C70/00 contains no steel, C70/20 contains 33.7 lb of steel fibers per cubic yard, and
C70/40 contains 67.4 lb of steel fibers per cubic yard. "A" denotes first peak. (Barragn et al., 2006)
Recent research has evaluated the use of shear friction equations for describing the interface shear capacity of lightweight concrete against ultra-high performance concrete (UHPC) (Banta, 2005). Three main variables were considered: amount of reinforcing steel (1, 2, 4, or 6 #3 bars across interface), the interface area (120, 180, or 240 in2), and the preparation of the UHPC surface (smooth, deformed with half rounds, shear keys, or chipped surface). The UHPC used for these tests was Ductal by Lafarge.
The setup of the test specimen is shown in Figure 2-10. The test setup measured the load being created by the actuator and the displacement of the lightweight concrete block. It did not measure the displacement of the UHPC block to normalize the results. Strains in the rebar were also measured close to the interface. In addition to shear across the interface, this setup also creates a moment due to the P forces not acting along the same line of action. The thesis does not fully describe the placement of forces (other than to say that they are non-concurrent) during the test so this moment can't be quantified with the information given. Strain in the steel was affected by the presence of this moment. In the specimens with 6 stirrups, the rear stirrups
2-18

experienced compression causing up to 850 microstrain due to this moment. This moment was present in all tests so it may have been manifest as a reduction (or increase) in strain in the reinforcement of the other specimens.
wn
Lightweight Concrete
P P
Ductal

Figure 2-10: Setup for interface shear tests conducted by Banta (2005). The normal force (wn) mimics the dead load on the structure and the P forces create shear across the interface between
the two materials.

Table 3-1 shows the average results for the different interface types. The failure mechanism was slightly different for each. The smooth and deformed both sheared cleanly across the interface without cracking either surface. In the keyed specimen, the lightweight concrete keys sheared during failure. In the chipped specimen, the fibers ruptured or pulled out of the lightweight concrete. All of these failures were brittle with negligible resistance experienced after failure.

Table 3-1: Effect of Surface Preparation on Max Load and Slip at Failure Surface Preparation Max Load (180 in2 interface) [k] Slip at Failure [in]

Smooth

16

.12

Deformed

29

.11

Keyed

50

.12

Chipped

65

.28

The specimens with varying amounts of shear reinforcement failed in different ways depending on the amount of reinforcement used. Those with 4 or more #3 bars failed in a ductile

2-19

manner. Those with only 1-2 #3 bars failed in a brittle manner (accompanied by large amounts of slip) before regaining some strength. It was not clear from the paper whether this was due to low reinforcement ratios or just lack of redundancy. Larger scale tests would be necessary to determine this. Crushing occurred adjacent to the rebar in the lightweight concrete in all tests. In the tests with more rebar, some crushing occurred in the UHPC but in the less heavily reinforced specimens, only minor spalling was observed.
The results are compared to a strut and tie model similar to that suggested by Hwang et. al. (2000). Banta notes, however, that more research needs to be done in this area to determine a rational for choosing the strut angles. The results were also compared to ACI, AASHTO LRFD, and AASHTO Standard codes. Each of the codes conservatively predicted the results observed even when no reduction factor is used for lightweight concrete. No new equations or changes to existing equations are suggested for improving accuracy.
No push-off shear testing has been done to date examining the interface shear capacity of monolithic UHPC with or without a pre-existing crack. Also no research has been done on interface shear capacity between cast-in-place normal strength, high-strength, or ultra-high strength concrete and precast UHPC. The effects of reinforcement ratio, interface area, and surface preparation must be evaluated for each of these cases to have a complete understanding of interface shear in UHPC structures.
2.2.2 Composite beams
While shear friction specimens allow for a direct testing of interface shear capacity, they do not exactly mimic the shear that exists across an interface in a precast beam with cast-in-place deck. In order to validate shear push-off tests and the interface shear equations, several researchers have tested composite beams.
In addition to the tests mentioned in the previous section, Hanson (1960) performed tests of compositely cast T-beams. Two lengths of girder were used with the same cross section (see Figure 2-11). The interface shear stress reported in these tests was calculated from the elastic equation (5), and is compared to the push-off data in Figure 2-12. This comparison is not the most helpful because of the method used for calculating the shear stress, but the comparison between tests still shows a good correlation between test types. From the test results of the push-
2-20

off and beam tests, the author suggested a maximum slip of .005 in. for determining the loss of composite action between the girder and deck.
Figure 2-11: Design for composite girder casting and testing performed by Hanson (1960). In the elevation view, the X's indicate slip dial locations and the triangles indicate deflection measurement locations.
2-21

Figure 2-12: Comparison of push-off and T-beam results by Hanson (1960)
2-22

Grossfield and Birnstiel (1962) performed similar tests of eight 10-foot-long T-Beams. The results of these test showed that it is difficult to separate the effects of slip and shear cracking from dial gauge readings in beam tests. Push-off tests, however, do not capture the effects of web cracking. They also note that this cracking effect may make a maximum slip value unrealistic for composite concrete design.
Saemann and Washa (1964) performed a comprehensive study of 42 composite T-beams. The influence of surface roughness, neutral axis position, shear span length, reinforcement ratio, shear keys, and concrete strength were evaluated. Beams with lengths of 8 and 11 ft were more likely to fail in interface shear than 20 ft beams. In these two shorter beam lengths, it was found that increasing the roughness of the interface increased the interface shear capacity. Also, when some surface roughness was present, increasing the amount of steel across the interface dramatically increased the interface shear capacity. From these tests, the following equation was suggested:

Y

=

2700 X +5

+

300P



33 - X X 2 + 6X + 5

where Y is the shear stress in psi, P is the interface reinforcement ratio, and X is the shear span to depth ratio. The authors used the elasticity solution for shear stress found in equation(5) for calculating stresses at failure.
Loov and Patnaik (1994) tested 16 composite T-Beams with a length of 126 in. and a span of 120 in. Half of the beams had their flanges discontinued 15.75 in. from each end of the beam (12.80 in. from the support); the other half had flanges that continued to the end of the beam. The surfaces of all beams were left as cast with protruding large aggregate, and the reinforcement ratio was varied from 0.1% to 1.9%. The authors found that slip was insignificant up to the point when the shear stress equaled 220 to 290 psi as calculated by equation (5). After this point, slip and shear stress both increased up to a slip of 0.01 to 0.03 in. at which time slip continued to increase, but shear stress began to decrease. They suggested a departure from the shear friction methodology with the use of the equation:

( ) vn = k 15 + v f y fc ' 0.25 f 'c

In this equation, k is a constant defined as 0.5 for compositely cast concrete and 0.6 for monolithic concrete.

2-23

Patnaik (1999) conducted additional tests on 6 composite beams with roughened interfaces and no interface shear reinforcement. Based on these tests, he suggested a lower bound interface shear capacity of:
vn = 4.215 fc ' psi when no shear reinforcement is present. This expression was only validated for concrete strengths below 8,700 psi. Patnaik also suggests the use of the equilibrium equation (6) for calculating the shear stress, particularly in prestressed beams where the "elastic equation will produce unrealistic stresses."
Patnaik (2001) also investigated the interface shear strength of composite beams with smooth interfaces and varying interface areas and interface shear reinforcement ratios. He showed that existing codes for smooth interfaces are conservative, but suggested the equation:
vn = 87 + v f y 0.2 fc ' and 800 psi or vn = 0.0 when v f y < 50 psi He also noted that composite action was usually maintained up to about 65% of the factored design load for each beam. Below this load, the use of the cracked composite section was valid for calculating deflections, but after this, the deflections increased dramatically. Figure 2-13 compares observed deflection behavior with that predicted by ACI for a monolithic section.
Kahn and Slapkus (2004) studied interface shear capacity of 6 composite beams cast with high performance concretes with strengths up to 11,300 psi. Interface shear reinforcement ratios were varied from 0.186% to 0.371%. The results of these tests showed the equation proposed by Kahn and Mitchell (2002) as well as the equation proposed by Loov and Patnaik (1994) to be good predictors of interface shear capacity in composite high-strength concrete beams. The ACI and AASHTO code equations were also shown to be conservative for these higher concrete strengths.
2-24

Figure 2-13: Typical load deflection curve for a composite beam with smooth interface compared to ACI predicted behavior (Patnaik, 2001).
No composite beam tests have been performed with UHPC, so little is known about the interface shear performance of these beams. Particularly, the higher strength of this material allows for much higher moment and shear capacities at ultimate. It is unclear if interface shear capacity will increase along with these properties. It is important to study these phenomena in order to avoid a situation in which a beam with very high expected moment and shear capacities fails prematurely due to low interface shear strength.
2.3 Diagonal Tension Shear
ACI 426R-74 (1973) and ACI 445R-99 (1998) describe many methods for calculating shear capacity of concrete members including Compression Field Theory (CFT), Modified Compression Field Theory (MCFT), and various truss analogies. Most of these assume a minimum amount of transverse reinforcement or a minimal shear capacity of concrete without transverse reinforcement. None of the methods in these documents, however, deal explicitly with the shear capacity of fiber-reinforced concrete or ultra-high performance concrete.
A preliminary French code for the use of UHPC (Association Franaise de Gnie Civil, 2002) suggests an additive relationship between the shear resistances provided by the concrete,
2-25

the steel fibers, prestressing strands, and any other passive reinforcement present. This is similar to the Vn = Vc +Vs relationship allowed in current ACI code, where Vn is the nominal shear capacity, Vc is the nominal concrete shear capacity, and Vs is the nominal shear reinforcement capacity.
The Federal Highway Administration performed three shear tests on prestressed UHPC AASHTO Type II girders with varying shear spans (Graybeal, 2006b). Based on the results of this research, Graybeal showed that the French code underestimates the shear capacity by more than 50% in prestressed UHPC girders with no shear stirrups. This code also doesn't account for shear carried in a cracked section.
Graybeal suggested using a simplified model for shear failure that bases the shear capacity on the diagonal tension carried by the UHPC. Using this method, he back-calculated a tensile capacity for the UHPC in the section where shear failure occurs. This tensile capacity was 38%-77% greater than Graybeal's best estimate of tensile cracking strength (fct) from smallscale tests. Thus this method would provide a lower bound for shear capacity, but may be overly conservative for this material.
The research only focused on girders without shear stirrups, so the effect of additional mild reinforcement was not evaluated. This research also focused only on shear capacity of the girder, rather than looking at the girder-deck system used in most bridge construction. The interactions between UHPC and mild steel reinforcement and between girder and deck must be understood in order to safely design highway bridge structures.
Shear friction principles can also be applied to diagonal tension shear stresses in concrete members. Mast (1968) originally suggested that any given plane in concrete should be considered a potential interface because a pre-existing crack could exist in this location. Based on this principle, the shear transfer required at any given section could be designed by assuming it to be a cracked interface.
Krauthammer (1992) used the shear friction theory to propose an alternative method for minimum shear reinforcement. It was suggested that the minimum shear reinforcement be the amount that provides the same magnitude of interface shear resistance as full aggregate interlock. This guideline would require a reinforcement ratio of at least 0.001 in all areas subject to shear. This is one of the first uses of shear friction theory to explain shear transfer in areas other than along interfaces or pre-defined shear planes.
2-26

In order to make a comprehensive design code for the use of UHPC in shear, more tests must be performed to better characterize and quantify the mechanics of failure. Specifically, it is important to determine how concrete strength, fiber reinforcement, and conventional stirrups interact in carrying shear stresses in UHPC. If shear friction theories can be applied to these phenomena, it will allow for the use of small push-off specimens for evaluating diagonal tension shear capacity of UHPC members without the expense or time required for full-scale testing in every case.
2-27

3. Research and Experiment Design
3.1 Research Approach
A multi-scale research approach is used to experimentally investigate interface shear and diagonal tension shear in composite UHPC bridge girders. The goal of this approach is to provide a thorough understanding of shear in composite UHPC bridge girders and to correlate the results of full-scale tests to results of smaller tests. Once this is accomplished, smaller tests can be more accurately used to predict structural behavior of full-scale designs. The full-scale composite girders used for these tests had interface lengths over 28 feet. The medium scale Tbeams had interfaces that were approximately 10 feet in length. The push-off specimens had interfaces that were 1 foot in length. Figure 3-1 shows a graphical representation of this approach.

Push-off Specimens

Composite T-Beams

Full-Scale Composite Bridge Girder and Deck

Design Guidelines and Recommendations for Field Use

Figure 3-1: Schematic overview of research approach.

3-1

Figure 3-2 shows an overview of specimens used in the current multi-scale interface shear research. Figure 3-3 shows a similar overview for diagonal tension shear.

Specimen
Full-Scale Girders

Interface
Roughened
Smooth

Reinf. Ratio
0.00000 0.00104 0.00417
0.00000 0.00104

Composite Tee-Beams

Roughened
Partially Roughened
Smooth

0.00000 0.00268 0.00000
0.00179 0.00268

Push-Offs

Roughened
Partially Roughened
Smooth

0.00000 0.00506
0.00000
0.00000 0.00253 0.00506 0.00759

Figure 3-2: Overview of specimens used in multi-scale interface shear tests.

3-2

Specimen
Full-Scale Girders

Reinf. Ratio
0.00000 0.00417 0.00833

Shear Plane
monolithic monolithic monolithic

Push-Offs

0.00000

monolithic precracked cold joint

0.00506

monolithic precracked cold joint

Figure 3-3: Overview of specimens used in multi-scale diagonal tension shear tests.

3.2 Full-Scale Girders and Decks
Three full-scale precast prestressed UHPC bridge girders with cast-in-place HPC were designed to test the flexural, diagonal tension shear, and interface shear capacities of the composite UHPC / HPC bridge girders. One longer girder with a span-to-depth ratio of 19.3 (15.2 with deck) was cast to test the flexural capacity of the system in four-point symmetrical bending. After the flexural test, the span length could be adjusted to test the shear capacity of each end of the beam by subjecting it to asymmetrical 3-point bending. Two shorter girders with span-to-depth ratios of 11.9 (9.3 with deck) were cast to provide additional shear tests. Each end of each of these beams was designed to be subjected to asymmetrical 3-point bending in order to force shear failure.
The six shear spans from the three beams were designed with differing amounts of transverse shear reinforcement in order to study the effect of shear reinforcement on the diagonal tension shear capacity of UHPC beams. The interface between the UHPC girder and HPC deck

3-3

was also varied along each of these shear spans. Differing amounts of interface reinforcement, shear areas, and concrete surface deformation were used in order to test the effect of these parameters on interface shear capacity. Table 3-1 lists the shear tests with the test parameters. The following sections will detail the design and construction of the girders and decks as well as the experimental setup and instrumentation.

Table 3-1: Comparison of interface steel and surface preparation for shear tests.

Shear Test

Surface Preparation

Stirrup Spacing, s

Reinforcement Ratio, v

1-1

smooth

24 in.

0.00104

1-2

fluted

N/A

0

2-1

fluted

24 in.

0.00104

2-2

smooth

N/A

0

3-1

smooth

12 in.

0.00417

3-2

fluted

N/A

0

3.2.1 Design and construction of girders and decks The design of the girder cross section for these tests was based largely in the desire to use
UHPC with existing formwork. Tindall Corporation, Inc. agreed to cast the material at their precast plant in Conley, GA, and suggested a preliminary beam shape based on existing formwork that they already possessed. These forms set the basic shape of the girders which was then modified to provide the girder designs shown in Figure 3-4 and 3-5. These shapes were chosen to more fully utilize the material properties of the UHPC. The bottoms of the girders were enlarged to contain 28 prestressing strands. The webs were able to be narrowed to 4 in. due to the high flowability of UHPC combined with its ability to carry high shear forces. This helped to minimize dead weight of the girders. All of these changes required minimal adjustments to be made to existing formwork. The final girder design for girders 1 and 2 had a cross sectional area of 298 in.2 and a moment of inertia of 37,420 in.4. For Girder 3, the top flange width was reduced and the cross-sectional area and moment of inertia changed to 271 in.2 and 28,865 in.4 respectively. Table 3-2 compares section properties of the Georgia Tech girders with several common highway bridge girders whose cross sections are shown in Figures 3-6 through 3-9.
3-4

2 13/32 in.

18 in.

8 in.

No. 4

stirrups

2 in.

3 in.

7 in.

No. 5 lifting loops
No. 4 stirrups

22 in.

17 13/32 in.

1 1/4 in. C.L.

4 in.

7 in.

3 in. 2 in.
2 in. 7 in.
2 in. 2 in

2 in.

7 spaces at 2 in. 2 in.

18 in.

SECTION AT MIDPOINT

No. 3 doghouse bars
1 1/8 in. C.L.
SECTION AT END

Figure 3-4: Cross section for Girder 1 and Girder 2.

No. 4 Ubars

3-5

2 13/32 in.

9 in. 3 1/2 in.

2 1/2 in.

No. 4 stirrups
3 15/16 in. 1 1/16 in.

No. 5 lifting loops
No. 4 stirrups

24 in.

17 13/32 in.

1 1/4 in. C.L.

4 in.

7 in.

3 in. 2 in.
2 in. 7 in.
2 in. 2 in.

2 in.

7 spaces at 2 in.

2 in.

18 in.

SECTION AT MIDPOINT

No. 3 doghouse bars
1 1/8 in. C.L.
SECTION AT END

Figure 3-5: Cross section for Girder 3 (reduced top flange).

No. 4 Ubars

3-6

Table 3-2: Section Properties of Various Bridge Girder Shapes

Girder
Georgia Tech Girders 1 & 2
Georgia Tech Girder 3
AASHTO Type I
AASHTO Type II
PCBT-29 Virginia Bulb Tee
Modified PCI Bulb Tee
(Silva et al., 2004)

Height, in. 32.4 32.4 28 36 29
28

Area, in.2 298 271 276 369 644
559

Moment of Intertia, in.4
37,420 28,865 22,750 50,980 66,800
54,352

Max number of strands in bulb
28 28 18 28 48
38

12 in.

3 in. 4 in.

28 in. 11 in.

3 in. 6 in.
5 in.

5 in. 5 in.

16 in. Figure 3-6: AASHTO Type I Girder
3-7

3 in. 6 in.

15 in.

36 in.

6 in. 6 in.

12 in. 3 in.
6 in. 6 in.
18 in. Figure 3-7: AASHTO Type II
3-8

2 in. 3.5 in. 2 in.

10 in.

28 in.

Figure 3-8: PCBT-29 Virginia Bulb Tee 42 in.
6 in. 2 in. 16 in. 10 in.
26 in. Figure 3-9: Modified PCI Bulb Tee (Silva et al., 2004)
3-9

6 in. 4.5 in.

Based on the formwork available, the total length of UHPC girder that could be cast was 125 ft. This length allowed for casting three girders. Girder 1 was 54 ft in length, and Girder 2 and Girder 3 were 34 ft in length. These lengths were chosen so that one flexural test could be performed on Girder 1 and two shear tests could be performed on each girder one on each end. The top flange width of Girder 3 was reduced as shown in Figure 3-5 in order to study the effects of interface area on the performance of the girder/deck system.
The Georgia Tech girders were constructed with 30-9/16-in. low-lax prestressing strands stressed to 70 percent of their 270-ksi ultimate strength. This number of strands and level of stress were determined using a Magnel Diagram with the modifications proposed by Krishnamurthy (1983) as shown in Figure 3-10, 3-11 and 3-12. The limit states of initial tension in the top of the girder, initial compression in the bottom of the girder, compression in the top of the girder at deck placement, and tension in bottom of the girder at deck placement along with the geometric constraints lead to the shaded "safe zone" shown for the 54-ft long Girder 1. The initial allowable tension was taken as 500 psi based on Graybeal's work in tensile strength of early age UHPC (2006a). Allowable tension in the top of the girder at strand release in Girder 3 controlled strand design. From these diagrams, the maximum eccentricity of -7.53 in. (-5.83 in Girder 3) was shown to be acceptable when all strands were fully prestressed. After prestress losses were calculated, the total prestressing force in the girder was 850 kips giving a force per area of girder of 2,860 psi. These values are 89% and 134% greater than the total prestressing force and force per area of the girder tested by FHWA (Graybeal, 2006b). Since the same strands were run through all three girders, the same configuration was used for Girder 1 and Girder 2 as was used in Girder 3.
3-10

Figure 3-10: Magnel Diagram for design of prestressing in Girder 1. 3-11

Figure 3-11: Magnel Diagram for design of prestressing in Girder 2. 3-12

Figure 3-12: Magnel Diagram for design of prestressing in Girder 3.
In order to test the interface shear capabilities of the girder/deck system, the surfaces of the girders were designed with two surface conditions. In some areas, the surface was left unfinished and allowed to cure with plastic covering as recommended by the UHPC manufacturer. Other areas were fluted by using a formliner depressed into the surface of the plastic concrete. The latter method of fluting was chosen because the self-leveling properties of the UHPC along with the steel fibers were known to prevent traditional roughening by brooming or raking of the surface. Figure 3-13 shows the formliner dimensions. The formliner used had regular deformations that were -in. wide and -in. deep. This formliner was chosen to replicate the -in. deformations required by ACI 318 Section 17.5.3.3 (2008) and AASHTO
3-13

LRFD Section 5.8.4.3 (2007). Differing amounts of shear reinforcement also protruded from the surface of the beams to vary the interface shear reinforcement. Figures 3-14, 3-15, and 3-16 show the detailed elevations of the girders along with their surface preparations.

1/4 in. 1/4 in.

19/32 in. 1/4 in.

Figure 3-13: Dimensioned drawing of formliner used in creating fluted UHPC surface.

3-14

4 spaces at 3 in. No. 4 stirrups

11 in.

11 in. 2 No. 5 lifting loops
24 in.

2 in.

Girder length = 54 ft
17 spaces at 24 in. No. 4 stirrups, no formliner

2 3/4in.

5 spaces at 12 in. No. 3 doghouse bars

11 in. 2 No. 5 lifting loops
1 space at 120 in. with formliner

11 in. 4 spaces at 3 in. No. 4 stirrups
2 in.

5 spaces at 12 in. No. 3 doghouse bars

2 3/4 in.

Center of bearing 1 ft

Center of bearing

Bearing to bearing length = 52 ft

1 ft

Figure 3-14: Girder 1 elevation.

3-15

11 in.
4 spaces at 3 in. No. 4 stirrups

11 in. 2 No. 5 lifting loops
24 in.

2 in.

Girder length = 34 ft
6 spaces at 24 in. No. 4 stirrups, with formliner

2 3/4in.

5 spaces at 12 in. No. 3 doghouse bars

11 in. 2 No. 5 lifting loops
1 space at 120 in. no formliner

11 in.
4 spaces at 3 in. No. 4 Stirrups
2 in.

5 spaces at 12 in. No. 3 doghouse bars

2 3/4 in.

1 ft

Bearing to bearing length = 32 ft

1 ft

ELEVATION VIEW

Figure 3-15: Girder 2 elevation

3-16

11 in.
4 spaces at 3 in. No. 4 stirrups
2 in.

11 in. 2 No. 5 lifting loops
12 in.

Girder length = 34 ft
17 spaces at 12 in. No. 4 stirrups, no formliner

2 3/4in.

5 spaces at 12 in. No. 3 doghouse bars

11 in. 2 No. 5 lifting loops
1 space at 120 in. with formliner

11 in. 4 spaces at 3 in. No. 4 stirrups
2 in.

5 spaces at 12 in. No. 3 doghouse bars

2 3/4 in.

1 ft

Bearing to bearing length = 32 ft

1 ft

ELEVATION VIEW

Figure 3-16: Girder 3 Elevation

3-17

A 5-ft wide, 8-in. thick high performance concrete (HPC) deck was cast on top of each of the girders as shown in Figures 3-17 and 3-18. The formwork for the deck was completely supported by the girder, which was simply supported during this construction. This unshored deck construction mimicked that which would most likely occur in the field. The decks were discontinued 3 feet from each end of the girders in order to facilitate testing of the girders. Temperature and shrinkage reinforcement were provided in the form of 7 No. 4 bars placed longitudinally in the decks and No. 4 bars at 10 inches on center running transversely. The steel was chaired 5 inches from the bottom of the decks. Figures 3-17 and 3-18 show the deck reinforcement and Figure 3-19 shows Girder 1 with formwork prior to pouring of the deck.
60 in.

8 in.
3/4 in. No. 4 reinforcing bars (typ.)

5 in. clear 1 in.

Figure 3-17: Girder 1 and 2 with composite HPC deck. Notice 3/4-in. haunch. 3-18

60 in.

8 in.
3/4 in. No. 4 reinforcing bars (typ.)

1/2 in.

5 in. clear

Figure 3-18: Girder 3 with composite HPC deck. Notice 3/4-in. haunch.

Figure 3-19: Temperature and shrinkage reinforcement for HPC deck on Girder 1. 3-19

Table 3-1 compares the different shear reinforcement and surface preparation combinations used on the interfaces of the three test beams. The first number of the test represents the beam number; the second number represents the different ends of the beams. Shear tests 1-1, 1-2, 2-1, and 2-2 were designed to analyze the interaction of steel interface reinforcement and surface preparation on the overall interface shear performance. Shear test 3-1 was designed to analyze the effect of increased steel reinforcement on the interface shear performance. Shear test 3-2 was designed to analyze the effect of a decreased interface shear area. Comparison of tests 2-2, 1-1, and 3-1 will show the effect of interface reinforcement ratio on interface shear capacity of girders with smooth interfaces. Comparison of tests 1-2 and 2-1 will show the effect of interface reinforcement ratio on shear capacity of girders with fluted interfaces. Tests 2-2 and 1-1 will evaluate the shear capacity of interfaces without reinforcement. Since shear friction theory requires clamping force, these tests will rely on traditional diagonal tension shear based on concrete cohesion to transfer forces across the interface.
During these tests, shear in the web of the UHPC portion of the girder will also be evaluated. Shear test 2-2, 1-1, and 3-1 have shear reinforcement ratios of 0%, .4% and .8% respectively in their webs. Evaluation of the performance will help determine if steel fibers and other shear reinforcement have an additive effect on shear capacity of monolithic UHPC.
3.2.2 Experimental setup and instrumentation of composite bridge section The flexural test of Girder 1 was conducted in 4-point bending. Vibrating wire strain
gauges (VWSGs) were embedded in the girder during casting (Figure 3-20) to determine prestress loss and were monitored during the flexural testing of the girder to obtain a strain profile. One gauge was placed between the top two prestressing strands. Two gauges were placed near the bottom of the girder between the first and second row of strands along each side of the bottom flange. All internal instrumentation was placed at midspan of the girder.
In addition to the internal instrumentation, strain gauges made using LVDTs were affixed to the exterior of the bridge section as shown in Figure 3-21. One 700-kip load cell was used to determine the load being applied to the system. Half of this load was placed at each of the locations marked P/2. Five string potentiometers were used to measure the deflected shape of the composite girder during the tests. Each potentiometer was attached to the underside of the
3-20

girder. On each side of the girder and deck, 5 linearly varying displacement transducers (LVDTs) were used to determine the strain profile. Each LVDT had a gauge length of 30 inches in order to capture cracking phenomena. Two additional LVDTs were placed on each side of the beam at the interface of the girder and the deck. One end was attached to the girder and the other to the deck in order to measure any differential motion between the two elements. The gauge length on these slip LVDTs was 30 in.
11 in.
15 in.
1 in. 3 in.
Vibrating Wire Strain Gauge Type K Thermocouple Figure 3-20: Section view of internal instrumentation in Girder 1. Girders 2 and 3 are identical except Girder 3 has reduced top flange width and Girder 2 has no thermocouples.
3-21

P/2

P/2

2 ft

2 ft

11 ft

6 1/4 in. 1 in.

12 in.

10 in. 2 in.

2 ft

11 ft

1 ft

Bearing to bearing length = 52 ft

String potentiometer Linear varying displacement transducer (LVDT) [30-in. gauge length] LVDT slip gauge [30-in. gauge length]

13 ft 1 ft

Figure 3-21: Elevation view of external instrumentation for flexure testing of Girder 1. (All instrumentation except string potentiometers is mirrored on other side of beam.)

3-22

The shear tests on Girder 1 were designed to utilize the beam that had already failed in flexure. Initially it was thought that both sides of the beam would stay intact enough to perform shear tests of each half of the beam as its own composite system. During flexural testing, however, the end of the beam with transverse shear reinforcement and a smooth interface surface experienced interface shear failure by debonding of the two concretes and shear failure of all of the transverse steel. Since the interface was instrumented with slip gauges, the load-slip data were recorded.
While the interface underwent complete failure, the UHPC girder remained relatively undamaged. Therefore, in lieu of a test of the composite section, this end of the girder was used for a shear test of the girder without composite deck. Figure 3-22 shows the potential experimental setup for this test. Because of the indeterminacy of the system, a load cell was used under the far right support to calculate the exact reaction forces and forces in the shear span of the girder.
3-23

P

6 ft

10 ft

1 ft
String potentiometer Linear varying displacement transducer (LVDT) [10-in. gauge length] Linear varying displacement transducer (LVDT) [30-in. gauge length]
Figure 3-22: Suggested instrumentation for shear test of girder without deck. A load cell under the far right support will allow for precise measurement of the reaction forces. The six-ft span to the left of the load point is 2.2 beam depths and is where shear
failure is expected.
3-24

The proposed setup for shear test 1-2 was not changed significantly by the interface shear failure during flexural testing and is shown in Figure 3-23. Again, a load cell on the far left support will allow for computation of reactions and internal forces during the beam testing. The interface being evaluated in this test is 16 in. wide and fluted by use of formliners. Slip gauges similar to those used during the flexural test will measure the relative displacement between the girder and the deck in order to quantify the interface shear capacity.
The web of Girder 1 did not have any transverse reinforcing bars in the shear span tested in shear test 1-2, so any resistance to diagonal tension shear was provided by the concrete and the steel fibers. Strain gauge rosettes made from LVDTs were place in the zone of high shear between the load point and the nearest support. These sets of gauges measured the principle strains and strain angles in the web of the girder. When diagonal tension shear failure occurred, these data were used to quantify the capacity of the web with no stirrups.
The setups for shear tests 2-2 and 3-2 were identical based on the similarity between the beam designs. Similar to the instrumentation for shear test 1-2, LVDTs were used for measuring slip along the interface between the deck and girder. LVDTs were also used to calculate a vertical strain profile and shear strains and strain angles. Figure 3-24 shows the instrumentation for these two tests.
Figure 3-25 shows the setup for shear tests 2-1 and 3-1. For these tests, one support was moved in to exclude the failed portions of the beams from the previous tests. Unlike tests 1-1 and 1-2, these tests were statically determinate, so a load cell under a support was not necessary to calculate reactions and internal forces. With the exception of the different total span length, the instrumentation for these tests is identical to that used for shear tests 2-1 and 3-1. The load for all of the full-scale tests was applied with a 1,000-kip ENERPAC hydraulic cylinder. Data from all LVDTs, potentiometers, and load cells were recorded in LabView.
3-25

P

12 ft

7 ft

1 ft
String potentiometer Linear varying displacement transducer (LVDT) [10-in. gauge length] Linear varying displacement transducer (LVDT) [30-in. gauge length] LVDT slip gauge [30-in. gauge length]
Figure 3-23: Experimental setup and instrumentation for shear test 1-2. All instrumentation except potentiometer mirrored on rear of girder.
3-26

P

24 ft

8 ft

2 ft 6 1/4 in.
1 in.
7 in.
12 in.

10 in. 2 in.

1 ft

Bearing to bearing length = 32 ft

1 ft

String potentiometer Linear varying displacement transducer (LVDT) [10-in. gauge length] Linear varying displacement transducer (LVDT) [30-in. gauge length]
LVDT slip gauge [30-in. gauge length]

Figure 3-24: Experimental setup and instrumentation for shear tests 2-2 and 3-2. All instrumentation except potentiometers mirrored on rear of girders.

3-27

P

8 ft

16 ft

2 ft 6 1/4 in. 1 in.
7 in.
12 in.

10 in. 2 in.

8 ft

1 ft

Bearing to bearing length = 24 ft

9 ft

String potentiometer Linear varying displacement transducer (LVDT) [10-in. gauge length] Linear varying displacement transducer (LVDT) [30-in. gauge length]
LVDT slip gauge [30-in. gauge length]

Figure 3-25: Experimental setup and instrumentation for shear tests 2-1 and 3-1. All instrumentation except potentiometers mirrored on rear of girders.

3-28

3.2 Composite T-Beams
Five cast-in-place UHPC beams with cast-in-place HPC composite decks were created in order to replicate the T-beam tests run by Kahn and Slapkus (2004). The interface shear capacity of the beam-to-deck connection is evaluated as the beams are tested in 3-point bending. The results of these tests will be compared to the interface shear capacity of the full-scale composite girders. This comparison will allow for a correlation to be drawn between interface shear failures in structures of different sizes. The interface shear results of these tests will also be compared with shear friction push-off tests discussed in the following section.
Each UHPC beam was 120-in. long with 114-in. span between supports. The cast-inplace HPC deck had a reduced length of 88 in. in order to force an interface shear failure. The "precast" beam had a depth of 10-in. and a width of 6-in. while the deck was 5-in. deep by 16.5-in. wide as shown in Figure 3-26. Figure 3-27 illustrates the reinforcement. Figure 3-28 shows the beam formwork with reinforcement prior to placement of the UHPC.

No. 3 stirrups 5 in.

No. 3 bars No. 3 stirrups

10 in.

2 in. 1 in.

5 in.

6 in.

No. 3 bars No. 9 bars 5 in.

Figure 3-26: Typical tee-beam cross section

3-29

16 in.

4 No. 3 top longitudinal steel (typ.)
88 in.

No. 3 double leg stirrups crossing interface (varied spacing)
2 No. 3 (typ.)

3 in.

114 in.

3 in.

4 No. 9 bottom longitudinal steel (typ.) No. 3 double leg stirrups (typ.)
Figure 3-27: Typical reinforcement for tee-beams.

Figure 3-28: Beam formwork and reinforcement prior to casting of UHPC. 3-30

The web was cast using UHPC at Tindall Corporation precast concrete plant, Conley, GA. Two days after casting, the web was thermally treated at 194F for 48 hours. The UHPC was Lafarge Ductal using 2% by volume steel fiber reinforcement. The 28-day mean compressive strength was 28,930 psi. Three hundred and forty two days after casting the web, the deck was placed using conventional high performance concrete (-in. maximum size aggregate) delivered by ready-mix. The HPC had a compressive strength at the time of testing of 12,170 psi. All reinforcement was A615, Grade 60. The elastic modulus for both the No. 9 and No. 3 bars was taken as 29,000 ksi.
The main variation between each of the composite beams was the interface preparation and reinforcement between the web and the flange. The top surface of the UHPC beams proved impossible to roughen by raking, brooming, or cutting with a trowel. To create a rough interface, a form liner was used to create in. flutes similar to the roughness amplitude required by codes. The variations between all five of the beams tested can be seen in Table 3-3. The reinforcement ratios are based on the interface area between the loading plate and the end of the deck. This means that on the 7-stirrup specimens, only 6 stirrups are actually engaged in interface shear.
At the suggestion of Mr. Peter Calcetas of Lafarge North America, burlap was placed atop the interface of one beam containing no reinforcement. The burlap was supposed to create a textured surface, which was to provide an improved bond to the deck slab. The burlap proved difficult to remove from the surface after the initial 48-hour cure and significant wire brushing was required. Further, during moving of a sixth beam intended to be tested that contained a smooth interface and no reinforcement, the deck fell off; that is, there was little bond between the web and slab. The latter beam was tested to give a strength comparison of a beam with no deck.

Table 3-3: Summary of Experimental Results

Beam
0-B 4-S 7-S 0-FL 7-FL

Number of Stirrups
0 4 7 0 7

v (percent)
0 0.179 0.268
0 0.268

Type of Interface
Smooth Smooth Smooth Fluted Fluted

3-31

The beams were designated using the following convention X-Y where the first term indicates the number of double-legged No. 3 stirrups crossing the interface and the second term indicates the type of interface between the web and the flange. B denotes a burlap-roughened cold joint interface, S denotes a smooth cold joint interface and FL denotes a cold joint interface with a fluted surface of -in.
The beam was loaded in three-point bending as illustrated in Figure 3-29 using a 1,000kip ENERPAC hydraulic actuator. The load was measured with a 700-kip load cell. Four mechanical strain gauges made with linear variable displacement transducers (LVDTs) were placed at the center of the beam. Each had a gauge length of 16-in. and these LVDTs were placed on both sides of the beam at the bottom and top of the web. Along with the strain gauges, two LVDT slip gauges were placed on one side of the beam. The body of the LVDT was attached to the web while the extension portion of the LVDT was attached to the underside of the cast-in-place deck measuring the slip between the beam and deck. Finally, the midspan deflection was measured using a string potentiometer. Figure 3-30 shows beam 7-FL set up for testing.

Slip LVDT, both sides of centerline, this side of beam only

Load application point 10 in. 33 in.

2 in. 2 in.

LVDT strain gauges, front and back

8 in. 8 in.

Vertical deflection at midspan measured by string potentiometer (not shown)

Figure 3-29: Loading and instrumentation of composite T-beams.

3-32

Figure 3-30: Beam 7-FL prior to flexural testing.
3.4 Push-Off Specimens
Thirty-nine push-off specimens were designed to test the interface shear capacity of HPC cast against UHPC as well as the shear friction behavior of monolithic UHPC and UHPC with cold joints and with pre-existing cracks. Twenty-one interface shear specimens will be compared to both the composite T-beams as well as the full-scale girders. Eighteen monolithic, pre-cracked, and cold-joint UHPC specimens will be compared to the diagonal tension shear capacity of the full-scale girders. The goal of correlating these small tests with larger ones is to allow for future design of full-scale girders to be based on shear-friction behavior defined by this type of smaller test.
In order to create interface shear push-off specimens with UHPC, several modifications had to be made to the traditional push-off design based on Anderson (1960). Because of the very high flowability of the material, it was not possible to cast an L-shape with one leg vertical. Therefore, the base leg of the L-shape was formed by embedding a steel wide flange section, and the specimen was cast in the orientation shown in Figure 3-31. This was the method of casting used for all of the interface shear specimens as well as the UHPC cold joint specimens. The push-off specimens were designed to have an interface shear area of 87 in2, which was approximately the same interface area used by Anderson (1960) and in one set of tests by Hanson (1960). The reinforcement of the specimens had to be changed slightly from that
3-33

previously used based on the expected increased capacity of the UHPC specimens. The complete composite specimen is shown in Figure 3-32.
In order to test the shear friction capacity of monolithic UHPC and UHPC with a preexisting crack, 12 monolithic specimens were created. The design of the monolithic push-off specimens was largely the same as the composite push-off specimens except that the breadth of the specimen and the size of the reinforcement had to be increased because of the expected increased capacity of the shear interface (Figure 3-33). Table 3-4 gives the properties of each different type of push-off specimen made. Three identical specimens were created with each set of properties.
Wide Flange
UHPC
Figure 3-31: Casting orientation of composite UHPC push-off specimen.
3-34

#4 Bar 7 " 1"

1'2 "

7 "

7 "

3 #4 Bars
A 1'8" 3 #5 Bars

8 "

3"

3"

3" 12"

2'4 "

3"

A 3"
2"

6"

W 6X15

A A

7/8"

5 "

7 "

7/8"

5/8"

1'1 "

5/8"

1'2 "

Figure 3-32: Composite push-off specimen with cold joint represented by dashed line.

3-35

W 8X35 7 " X 3" A36
Steel Plate 1 "
" #3 Stirrups
3 #4 Bars
A 3 #5 Bars

1'7"

9 "

9 "

10"

2" 4"

12" 2'8" 4"

4"

A

2"

8"

A A

7/8"

5 "

7 "

7/8"

5/8"

1'5 "

5/8"

1'7" Figure 3-33: Monolithic sections cast with no cold joint. Some specimens were intentionally
cracked along interface.

3-36

Table 3-4: Push-off specimen properties. Three specimens of each type were created.

Phase 2

Shear

Phase

(if

Interface Reinforcement Reinforcing

Specimen 1

present)

Type

Type

Ratio

U-H-S-0 UHPC HPC

smooth

none

0.0000

U-H-FL-0 UHPC HPC

fluted

none

0.0000

U-H-S-1 UHPC HPC

smooth 1 No. 3 U stirrup 0.0025

U-H-S-2 UHPC HPC

smooth 2 No. 3 U stirrups 0.0051

U-H-FL-2 UHPC HPC

fluted 2 No. 3 U stirrups 0.0051

U-H-S-3 UHPC HPC

smooth 3 No. 3 U stirrups 0.0076

U-H-B-0 UHPC HPC

burlap

none

0.0051

U-U-S-0 UHPC UHPC smooth

none

0.0000

U-U-S-2 UHPC UHPC smooth 2 No. 3 U stirrups 0.0051

U-U-PC-0 UHPC

precracked

none

0.0000

U-U-PC-2 UHPC

precracked 2 No. 3 U stirrups 0.0051

U-U-UC-0 UHPC

uncracked

none

0.0000

U-U-UC-2 UHPC

uncracked 2 No. 3 U stirrups 0.0051

U=UHPC, H=HPC, S=smooth cold joint interface, FL=fluted cold joint interface, B=burlap-

roughened cold joint interface, PC=precracked monolithic interface, UC=uncracked monolithic

interface, 0-3=number of dual-leg No. 3 stirrups crossing interface

3-37

4. Specimen Construction

4.1 Introduction
All UHPC construction was performed at the Tindall Corporation, Georgia Division Plant in Conley, GA. For these castings, the UHPC used was Ductal manufactured by Lafarge Corporation. Batching was done in Tindall's 4.5 cu.yd. shear mixer by pouring bags of the Ductal premix, water reducing admixture, water and steel fibers into the mixer. Table 1 shows the mix design.

Table 4-1: Mix design for 1 yd3 of UHPC.

Component Ductal Premix
Premia 180
Water
Steel Fibers

Weight lb. (kg) 3700 (1680)
68.2 (30.93)
209 (94.8) 263 (119)

Weight Fraction 87.3%
1.6%
4.9%
6.2%

4.2 Placement of Push-off and Small Beam Specimens
Formwork for the push-off and small beam specimens are shown in Figures 4-1 through Figure 4-3. Figure 4-4 shows the concrete from this batch flowing down a 10 foot long beam. The monolithic push-off specimens were poured directly from the bucket (Figure 4-5) while the UHPC for the cold-joint push-off specimens was scooped from a wheelbarrow into the forms (Figure 4-6). Finishing the UHPC was done in three different ways for the specimens. Most specimens were finished by screeding the surface and immediately covering the specimen with 6 mil. plastic tight against the surface of the concrete. The screeding process was more difficult than with traditional concretes, largely due to the stickiness of the mix and the presence of steel fibers.

4-1

Figure 4-1: Formwork for UHPC portion of composite shear friction specimen.
Figure 4-2: Formwork for monolithic push-off specimen prior to casting of UHPC.
4-2

Figure 4-3. Formwork for 10-ft long beams prior to casting of UHPC.
Figure 4-4. UHPC flowing down 10-ft. long beam.
4-3

Figure 4-5: Placement of UHPC in large push-off form directly from bucket.
Figure 4-6: Placement of UHPC in small push-off forms with wheelbarrow and scoop.
4-4

The plastic that was pressed into the surface of the concrete made a glassy finish. Any extra air in the uncured concrete would rise to the surface over time and leave trapped air bubbles on the surface of the concrete, beneath the plastic. This created blisters that were noticeable upon curing.
The second method of finishing involved depressing a ridged formliner backed with plywood into the surface of the wet concrete. The formliner was secured in place using screws through the plywood into the formwork. This process created a roughness on the surface of the cured concrete. In order for the formliner to work, the form has to be filled to the surface with UHPC. On one specimen, a small leak allowed for the level of UHPC to lower in the form after the formliner was placed. This resulted in a smooth surface -in. below the surface of the form.
Without formliners, a roughened surface finish was difficult to achieve with UHPC. It was very difficult to brush or rake the surface because of the steel fibers and stickiness of the mix. Even if a rake or broom were used to deform the surface of the plastic concrete, the material would self level before setting could occur.
The third type of finished surface was created by pressing burlap into the surface of the plastic concrete. This created a roughened surface with much smaller deformations than those created by the formliner. The problem with this process, however, was that the burlap was removed too late and much of the material was permanently trapped in the surface of the UHPC. Further work could be done to find a time between final set and significant strength gain where the material could be removed while still leaving the desired surface preparation.
Immediately after the UHPC was placed and finished, all specimens were covered with plastic to prevent the loss of moisture from the concrete. The plastic was secured to the forms to increase the quality of the seal. The specimens were allowed to cure at ambient temperature for 48 hours. No water or wet burlap was used due to the plastic that was pressed into the surface of the wet concrete.
After 48 hours, the forms were removed and the specimens were stacked around the steam line and again covered with plastic. The recommended temperature for thermal treatment of 95 degrees C was possible to achieve. Temperatures between 94 and 98 degrees C were maintained for the 48 hours of requisite thermal treatment.
4-5

Three 4x8-in. cylinder and three 6x12-in. cylinder specimens were cast for each UHPC beam and for each three UHPC push-off specimens. All test cylinders were cured under identical conditions as the test specimens. The 4x8 cylinders were used for compressive strength tests while the 6x12's were used for modulus of elasticity tests.
4.3 Girder Construction
One 54-ft.-long precast prestressed bridge girder, two 34-ft.-long precast prestressed bridge girders, and 60 cylinders were cast in a separate batching and placing operation. Since 13 yds3 of UHPC were being cast, 4 batches were required. Ice was again substituted for 90% of the water to keep the mixing temperatures below 77 F (25 C). Each of these batches was mixed as previously described and then transferred via concrete bucket to a Ready-mix truck. This Ready-mix truck was used because the 4.0 yd3 mixer at Tindall could not facilitate enough concrete for casting these specimens from one batch. Each beam needed to be cast in one continuous pour to prevent cold joints from occurring in the beams. After batches 1 and 2 were in the Ready-mix truck, the first 54-ft.-long beam was cast. Batches 3 and 4 were then added to the truck and the two 34-ft.-long beams were cast. UHPC was placed in wheel barrows and was scooped into the cylinder test molds.
In casting the UHPC girders, the most striking characteristics of the concrete were the very high flowability and the homogeneity of the material. These traits were noticeable in the casting of smaller specimens, but were much more so in the casting of the large girders. Figure 4-7 shows the material as it pours down the chute of the Ready-mix truck. The material did not exhibit any visible segregation or bleeding whatsoever. Figure 4-7 and 4-8 show the material as it is being poured into one end of the beam form and down the length of the beam. The large scale placement of the material further showed the atypical consistency of the material; it flows like cake batter.
4-6

Figure 4-7: Highly homogeneous UHPC pouring fluidly from Ready-mix truck.
Figure 4-8: Beam forms are filled by pouring UHPC in one end and allowing the material to flow along the axis of the beam.
As the forms were being filled, the amount of material had to be controlled very carefully. Any overfill was difficult to remove due to the fibers in the concrete. The time required for removing excess material also contributed to drying of the surface of the material. Also, unlike normal
4-7

concrete, the excess cannot be easily screeded off. This top surface formed a leather-like skin over the plastic concrete below. This dry skin also served to trap rising air bubbles from the unset concrete below. These air bubbles then formed blisters immediately below the surface of the skin.
Nineteen hours after casting, the first cylinder was broken to determine if the beams were strong enough to remove the forms. A compressive strength of 440 psi was measured, so the forms were not released at that time. At 42 hours after casting, the cylinder strength of the material was 15,460 psi, which was strong enough for the forms to be removed. The forms were removed between 44 and 47 hours after casting. The beams were covered with plastic whenever possible to prevent moisture loss. The strands were cut approximately 68 hours after casting.
Steam curing was begun after cut-down. After 24 hours of steam curing, Beam 1 averaged 144 F (62 C), and Beam 3 averaged 122 F (50 C). These values were 56 and 78 F (32 and 45 C) less than the temperatures recommended by Lafarge. Due to the low steam temperatures, the beams were left to steam for an additional 9 days making a full 10 days of thermal treatment. Figure 4-9 shows the UHPC strength gain with this curing.

Compressive Strength (psi)

30000 25000 20000 15000 10000
5000 0 0

steam curing

5

10

Days After Casting

Figure 4-9: Strength gain for UHPC steam cured in field. Low temperatures increased steaming time and even so 30,000 psi strengths were not attained.

The 56-day, post thermal treatment, compressive strength of the girder UHPC was 24,890 psi. At test day, the compressive strength and modulus of elasticity were 24,425 psi and 7,366ksi respectively.

4-8

The girders were then transported to the Georgia Tech Structural Engineering Laboratory. Formwork was erected and an 8-in. thick deck, 5-ft. wide deck was cast atop each girder as shown in Figures 4-10 and 4-11. As ordered, the deck concrete strength was designed to be 7,000 psi. Yet, at 90-days, the compressive strength of the deck concrete was 10,260 psi.
Figure 4-10. Construction of the composite deck in the Structures Laboratory
4-9

Figure 4-11. Drawing of the composite deck slab showing instrumentation

Vibrating wire strain gauge LVDT strain gauge Linear string potentiometer Electrical resistance strain gauge
4-10

5. Shear Friction Push-off Tests

5.1 Introduction

The purpose of this experimental study was to determine if the 2008 ACI1 and 2007 AASHTO2 code equations for shear friction are applicable for ultra high performance concrete

(UHPC) and for cold joints between UHPC and cast-in-place high performance concrete (HPC).

The concept of shear friction uses the idea of a coefficient of friction to explain shear transfer

across a given plane, especially at a cold joint or at an existing or potential crack. The current ACI 1 and AASHTO2 shear friction equations are given in Eqs. (1) & (2) respectively

Vni = Avf fy

(1)

not to exceed the smallest of:

0.2 fc ' Ac psi

0.89 fc ' Ac kN (480 + 0.08 fc ') Ac psi
(3.309 + 0.0005516 fc ') Ac kN

and 11.03Ac , kN.

Vni = cAcv+ (Avf fy+Pc)

(2)

not to exceed the smallest of:

K1 fc ' A

And K2 Ac .

Where Ac is the area of concrete resisting the shear. Using equations (1) and (2), shear stress is calculated by dividing nominal shear strength

by the area of the concrete engaged in shear transfer as follows in Eqs. (3) and (4)

ni = vfy

(3)

ni = c + (vfy + Pc/Acv)

(4)

Eq. (1) was derived from what was originally proposed by Birkeland-Birkeland (1966) and has

since been modified based on experimental research [Mast, 1968; Hofbeck, Ibrahim and

Mattock, 1969; and Kahn and Mitchell, 2002].

5-1

The current ACI (2008) and AASHTO (2007) code provisions need to be validated in order to determine if current provisions can be used for UHPC girder shear resistance and for UHPC/HPC interface shear resistance for composite structures. Because of UHPC's high tensile capacity and the presence of steel fibers in the matrix, it has been proposed that UHPC girders require no transverse shear reinforcement. Also, the amount of shear reinforcement required to develop shear transfer across the interface between a precast UHPC bridge girder and a cast-inplace deck slab is unknown. The interface connection is further complicated because the high flowability of the UHPC, and the presence of the fibers which make roughening the UHPC surface by brooming or raking impossible.
5.2 Test Set up
Twelve monolithic specimens were cast from UHPC with a compressive strength, fc', of 200 MPa (28,900 psi) as shown in Figure 3-33. Twenty-six composite specimens were manufactured by first casting one-half from UHPC with a compressive strength, fc', of 200 MPa (28,900 psi). After this half had cured, the other half was cast from either HPC or UHPC, creating a cold joint. Figure 3-32 shows the design of the composite specimens. The HPC used to create the second half of some composite specimens had compressive strength of 84.4 MPa (12,200 psi). The UHPC used to create the second half of some composite specimens had compressive strength of 191 MPa (27,000 psi). For all specimens, the interface shear plane was rectangular with width of 174 mm (7 1/4 in.) and length of 288mm (12 in.).
In the all-UHPC specimens, the two variables evaluated were the interface surface and transverse reinforcement ratio. The monolithic specimens were either left uncracked or were pre-cracked; the composite specimens had a smooth cold joint to represent a UHPC girder in which concrete placement was delayed and an accidental cold joint was created. The monolithic pre-cracked specimens were prepared by placing a specimen on its front side while aligning a knife-edge plate perpendicular to the shear plane as shown in Figure 5-1.
5-2

Figure 5-1. Section cut showing loading and position of monolithic specimen during precracking.
Load was applied at an initial rate of 500 lbs per second and continued until a 3000 lb. drop in load was observed. For each surface preparation, either 0 or 2 two-leg No. 3 stirrups crossing the shear plane were used giving reinforcement ratios of 0% or 0.5%.
Similarly, in the UHPC/HPC cold joint composite specimens, the interface surface preparation and interface reinforcement ratio were varied. Three surface preparations were used. The first interface used a 6 mm (1/4 in.) deep form liner which was pressed into the surface of the UHPC to create a fluted, "roughened" condition. In the second interface, burlap was placed on the cast UHPC to create a mildly rough surface. The third surface was the smooth, as-cast cold joint surface; no troweling, brooming, or raking was used. Either 0, 1, 2, or 3 two-leg No. 3 stirrups crossed the shear plane to give reinforcement ratios of 0%, 0.25%, 0.5%, and 0.75%. Figures 5-2 shows the forms for the two types of specimens.
5-3

Figure 5-2. Specimen formwork for monolithic pour (top) and composite pour (bottom).
5-4

The Grade 60 No. 3 reinforcing bars used for all closed stirrup transverse reinforcement had an average yield stress of 506.8 MPa (73.5 ksi); however, the ACI (2008) recommended fy of 413.7 MPa (60 ksi) was used in the following calculations of predicted shear stress values.
Table 5-1 summarizes the details of all specimens. The first two letters represent the type of concrete on either side of the shear plane, U for UHPC and H for HPC. The third set of letters indicates the interface type: monolithic uncracked (UC), monolithic precracked (PC), fluted cold joing (FL), burlap-roughened cold joint (B), smooth cold joint (S). The last character indicates the number of stirrups crossing the shear plane (0,1, 2, or 3). Three identical push-off specimens termed A, B, and C were constructed for each type except for U-H-FL-0 which had specimens A and B.
Table 5-1. Specimen Specification

Specimen Identification*

Concrete 1

Concrete 2

U-U-S-0

UHPC

UHPC

U-U-S-2

UHPC

UHPC

U-U-PC-0

UHPC

UHPC

U-U-PC-2

UHPC

UHPC

U-U-UC-0

UHPC

UHPC

U-U-UC-2

UHPC

UHPC

U-H-S-0

UHPC

HPC

U-H-S-1

UHPC

HPC

U-H-S-2

UHPC

HPC

U-H-S-3

UHPC

HPC

U-H-B-0

UHPC

HPC

U-H-FL-0

UHPC

HPC

U-H-FL-2

UHPC

HPC

* 3 specimens of each type were tested

Interface Type
Smooth Cold Joint Smooth Cold Joint Pre-Cracked Monolithic Pre-Cracked Monolithic Un-Cracked Monolithic Un-Cracked Monolithic Smooth Cold Joint Smooth Cold Joint Smooth Cold Joint Smooth Cold Joint Burlap-Roughened Cold Joint Fluted Cold Joint Fluted Cold Joint

Transverse Reinforcement
Ratio, v 0%
0.50% 0%
0.50% 0%
0.50% 0%
0.25% 0.50% 0.75%
0% 0% 0.50%

All push-off specimens were tested using the set up illustrated in Figure 5-3. Each specimen was centered in a 1,780 kN (400 kip) capacity hydraulic testing machine. Load was applied at a rate of 445, 890, or 2,224 N/s (100, 200, or 500 lb/s) depending on the expected interface shear capacity. Relative slip movement across the interface was measured by dial gages located at center of the interface on both front and back of the specimen. Prior to testing, the width and height of the interface shear surface were measured and recorded for determining interface shear stress.

5-5

Figure 5-3. Typical loading and push-off test of a composite specimen.
5.3 Experimental Results and Discussion
5.3.1 Monolithic UHPC Push-off Results The ultimate interface shear recorded for each specimen is presented in Table 5-2 and 5-3 along with the predicted capacities from Equation (1) and (2). The monolithic uncracked specimens with and without shear reinforcement behaved similarly. Initial cracks were observed at loads between 20% and 65% of the peak ultimate capacity and were only visible when alcohol was applied. These cracks were between 1 to 4 inches long and oriented diagonally between 0 to 30 degrees to the shear plane from top and bottom as illustrated in Figure 5-4. This behavior was similar to what was observed by Hofbeck et al (1969) and Kahn and Mitchell (2002). The
5-6

unreinforced specimens showed ultimate capacities that were approximately 33% lower than those with the 2-double leg #3 bar stirrup reinforcement.
In the pre-cracked monolithic specimens, measurable slip was observed at much lower loads and increased at a greater rate than in the uncracked specimens. As the load approached ultimate, the pre-existing cracks became wider and caused the failure to be localized. The ultimate load in the pre-cracked specimens was 30-35% less on average than the corresponding uncracked specimens, but it varied widely between specimens. Inconsistent crack width between specimens may have caused this variation. The all-UHPC specimens with a smooth cold joint at the interface showed much ultimate capacities that were only 10-20% of the uncracked or precracked monolithic UHPC push-offs.

Table 5-2. Monolithic push-off specimen ultimate shear experimental and predicted values, 1kN=224.8lb

Specimen Identification
A U-U-S-0 B
C A U-U-S-2 B C A U-U-PC-0 B C A U-U-PC-2 B C A U-U-UC-0 B C A U-U-UC-2 B C

Test value Vexp [kN]
91.2 73.4 77.0 219.7 227.3 203.3 631.6 914.1 658.3 1069.7 573.8 1396.7 1005.2 1000.8 1113.3 1467.8 1561.2 1641.3

ACI1 predicted Eq. (1)
[kN] 0 0 0
70.3 70.3 70.3
0 0 0 164.6 164.6 164.6 0 0 0 164.6 164.6 164.6

AASHTO2 predicted Eq. (2)
[kN] 30.7 31.1 29.4 100.5 100.5 99.6 157.0 156.6 153.9 320.7 320.3 323.4 158.3 159.7 156.6 323.8 319.8 318.5

Vexp / VACI
predicted
0 0 0 3.1 3.2 2.9 0 0 0 6.5 3.5 8.5 0 0 0 8.9 9.5 10.0

Vexp / VAASHTO
predicted
3.0 2.3 2.6 2.2 2.3 2.0 4.0 5.8 4.3 3.3 1.8 4.3 6.4 6.3 7.1 4.5 4.9 5.2

5-7

Table 5-3. Experimental and predicted shear capacities of UHPC/HPC push-off specimens, 1kN=224.8lb

Specimen Identification

Test value Vexp [kN]

A U-H-FL-0
B

A

U-H-FL-2 B

C

A

U-H-S-0

B

C

A

U-H-S-1

B

C

A

U-H-S-2

B

C

A

U-H-S-3

B

C

A

U-H-B-0

B

C

203.7 207.3 391.4 520.4 369.2 55.6 52.5 80.1 118.3 129.4 119.7 218.0 179.3 184.1 249.1 188.6 270.4 146.8 120.1 160.1

ACI1 Predicted Eq. (1)
[kN]
0 0 14.8 19.7 14.0 0 0 0 14.9 16.3 15.1 13.7 11.3 11.6 10.5 7.9 11.4 0 0 0

AASHTO2 predicted Eq. (2)
[kN]
93.4 95.2 209.5 210.8 211.3 30.2 30.2 29.8 63.6 64.9 64.5 100.1 100.5 100.1 135.2 134.8 135.2 29.8 29.4 29.4

Vexp / VACI predicted

Vexp / VAASHTO predicted

0

2.2

0

2.2

3.3

1.9

4.4

2.5

3.1

1.7

0

1.8

0

1.7

0

2.7

3.4

1.9

3.7

2.0

3.4

1.9

3.1

2.2

2.5

1.8

2.6

1.8

2.4

1.8

1.8

1.4

2.6

2.0

0

4.9

0

4.1

0

5.5

5.3.2 UHPC/HPC Cold Joint Results The UHPC/HPC specimens with reinforcement all tended to have more gradual changes
to the slope of the load-slip curve than did the all-UHPC specimens. This could be due to a more gradual transfer of force from cohesion to shear friction. The fluted UHPC specimens were the exception to this trend. They exhibited much more consistent stiffness up to failure loads. The reinforced fluted specimens carried the most load of the UHPC/HPC specimens, and even the unreinforced fluted specimens carried more load than all but the most heavily reinforced smooth specimens. In general, the unreinforced UHPC/HPC specimens exhibited very brittle failures at
5-8

much lower loads than their reinforced analogs. Figure 5-4 shows the typical interface crack after loading.

a)

b)

c)

Figure 5-4. (a) Specimen after testing, (b) slip at joint visible at reinforcement markings, (c) failed fluted interface.
5-9

5.3.3 Comparison with code equations
Figure 5-5 shows the ultimate shear stress, u, as a function of the clamping stress, vfy, where fy is limited to 413.7 MPa (60 ksi). The solid and dashed lines represent the ACI shear friction prediction with varying coefficients of friction. ACI (2008) gives an 11.03 Mpa (1600 psi) limit. All shear stresses are greater than those predicted by equation (4) with coefficient of friction, , of 1.0. Figure 5-5 shows that the ACI (2008) equation is conservative when extended to UHPC.
Figure 5-6 shows the same experimental results compared to the AASHTO (2007) equations. These equations also provide a conservative estimate of the experimental results in all cases.
5.3.4 Effect of interface on performance of all-UHPC specimens Figure 5-7 shows typical load-slip curves for each type of all-UHPC specimen tested. It
was observed that the pre-cracked specimens and cold-joint specimens exhibited similar initial slopes to their load-slip curves, which were both less than those slopes observed when testing the uncracked specimens. This difference in slope suggests that microslipping began across the precracked and cold joint interfaces even at very low loads. The pre-cracked specimens with and without reinforcement, however, had capacities that were 670-800 kN (150-180 kip) greater than similar specimens with cold joints. This increase in capacity shows the large contribution of the steel fibers in transferring shear across the interface.
5.3.5 Effect of reinforcement ratio on performance of all-UHPC specimens As expected, the specimens with reinforcement exhibited much more ductile failures than
the unreinforced all-UHPC specimens. The reinforced specimens also exhibited increased load carrying capacity with the uncracked, pre-cracked, and cold joint specimens with reinforcement carrying 50, 38, and 170%, respectively, more load than the equivalent specimens without reinforcement. The increase in clamping force acts in the way predicted by shear friction theory. For smooth cold joints, a friction coefficient, = 0.6, was a conservative predictor of performance as clamping force increased. For monolithic specimens with and without existing cracks, = 1.4 provided an overly conservative prediction, underestimating the capacity by at least 300%.
5-10

Figure 5-5. Comparison of all specimens to ACI shear friction equation, Eqn. (1).
5-11

Figure 5-6. Comparison of all specimens to AASHTO shear friction equation, Eqn. (2).
5-12

Figure 5-7. Typical load-slip curves for all-UHPC push-off specimens.
5-13

5.3.6 Effect of surface preparation on performance of cold joint UHPC/HPC specimens Figure 5-8 shows the typical load-slip curves for the UHPC/HPC push-off specimens.
The surface roughness of these specimens had a large impact on the interface shear capacity. In unreinforced specimens (data given in Figure 5-6), the burlap-roughened surface increased the interface shear capacity over the smooth cold joint by 127% while the fluted surface increased the shear capacity by 228%. In specimens with 0.5% reinforcement ratios (two stirrups), the fluted surface increased the interface shear capacity of the specimens by 120%. For all fluted surfaces specimens, the ACI and AASHTO recommended friction coefficient, = 1.0, was conservative. Even when the coefficient was taken as 1.4, the ACI and AASHTO codes provided conservative estimates. 5.3.7 Effect of reinforcement ratio on performance of UHPC/HPC specimens
In specimens with a smooth interface, the shear capacity appeared to increase linearly with the increase in reinforcement ratio. This supports the shear friction theory and its applicability to UHPC/HPC interfaces even when a smooth joint is used with a friction coefficient = 0.6. When a fluted joint was used, the use of a 0.5% reinforcement ratio increased the interface shear capacity by 108%. For both types of fluted-joint specimen, the shear friction theory was conservative with the ACI and AASHTO specified friction coefficient = 1.0.
5-14

Figure 5-8. Typical load-slip curves for UHPC/HPC push-off specimens. The curves for the unreinforced specimens had approximately the same slope as that shown for U-H-S-0 with maximum values listed in Table 5-3.
5-15

5.4 Conclusions Regarding Push-off Tests
Based on the current research in shear transfer in monolithic and compositely cast pushoff specimens, the following conclusions were made:
1. Current ACI and AASHTO provisions are conservative for estimating interface shear capacity of composite UHPC/HPC structures and monolithic UHPC.
2. Formliners can be successfully used to create a fluted surface finish in UHPC that is comparable to the 6 mm (1/4-in.) surface roughness recommended by current codes for composite construction. This surface finish provides significant increase in interface shear capacity, particularly when used in conjunction with reinforcing steel crossing the interface.
3. In monolithic UHPC, steel fibers allow for significant shear transfer across pre-existing cracks even when no additional shear reinforcement is used; however, smooth cold joints in UHPC have very low interface shear capacities when transverse shear reinforcement is not used.
5-16

6. Interface Shear in Small Composite Beams

6.1 Introduction
Five T-beams were constructed using UHPC for the web with a nominal compressive strength of 29 ksi (200 MPa) and HPC for the deck with a nominal compressive strength of 12 ksi (83 MPa). Each beam contained a smooth interface (no roughening), burlap roughened, or a form liner (-in. (6 mm) roughened) interface between the web and the flange. The transverse reinforcement ratio ranged between 0 and 0.292 percent for all beams.
AASHTO LRFD (2007) provides an equation for shear friction along an interface between two concretes cast at different times. As long as minimum transverse reinforcement requirements are met, the nominal shear resistance of the interface plane is given in Eq. (1):

Vn = cAcv + [Av fy + Pc ] Vn max

(1)

where

Vn = nominal shear strength, lb Vn max = the smaller of 0.2fyAcv or 800Acv, lb c = cohesion factor, 75 psi (0.52 MPa) for a clean concrete surface, not roughened,

280 psi (1.9 MPa) for a roughened surface with a -in. amplitude,

400 psi (2.76 MPa) for concrete cast monolithically Acv = area of concrete considered to be engaged in interface shear transfer, in2 = friction factor equal to 1.0 for a roughened surface with a -in. (6 mm) amplitude,

0.6 for a not intentionally roughened surface,

1.4 for concrete cast monolithically

Av = area of interface shear reinforcement crossing the shear plane within the area Acv, in2
fy = yield stress of transverse reinforcement, psi Pc = permanent net compressive force normal to the shear plane, lb

6-1

ACI 318 (2008) gives a similar equation for shear friction along an interface with minimum transverse reinforcement. This equation is given in Eq. (2) in customary in.-lb units:

( ) Vn = 260 + 0.6v fy bvd 500bvd

(2)

where

s = spacing of transverse reinforcement, in.

v

=

transverse

reinforcement ratio,

Av bv s

bv = interface width considered to be engaged in shear transfer, in. d = distance from top of slab to centroid of bottom tensile reinforcement, in.

Furthermore, both AASHTO (2007) and ACI (2008) permit the shear friction equation given in Eq. (3) to be used when evaluating interface shear strength:
Vn = Av fy the smaller of 0.2 fcAcv, (480 + 0.08 fc)Acv or 1600Acv (3) where
= coefficient of friction, 1.4 for a monolithic concrete connection, 1.0 for a cold joint with in. (6 mm) roughness amplitude, and 0.6 for a cold joint at a smooth concrete interface
Acv = area of shear reinforcement across shear plane, in2 fy = yield stress of transverse reinforcement ( 60,000 psi)

Neither Eq. (1) or Eq. (2) takes into account the concrete compressive strength (fc'), and Eq. (3) only uses compressive strength as an upper boundary. Yet several authors have investigated the influence of fc' on nominal shear strength of beams. Loov and Patnaik (1994) performed tests on 16 composite beams with varying concrete strengths, web widths, stirrup spacing and two different flange lengths. They also made sure that the surface between the flange and web was left rough and that course aggregate was protruding. Loov and Patnaik (1994) proposed Eq. (4), which calculates nominal shear stress capacity and takes into account the roughed surface as well as fc':

6-2

vn = k

(15

+

v

fy)

f


c



0.25

f


c

(4)

where n = nominal shear stress, psi k = roughness constant equal to 0.6 for rough surfaces and 0.5 for smooth surfaces = correction factor related to concrete density fc' = weaker compressive strength of the flange or web concrete, psi

Loov and Patnaik (1994) further concluded that the stirrups did not attribute to shear

resistance until the horizontal shear stress reached 220 to 290 psi (1.5 to 2 MPa), suggesting that

the roughened surface provided adequate shear resistance before this stress range.

Fifty pushoff specimens were tested by Kahn and Mitchell (2002) to determine if current

design standards may be used for high strength concretes. Concrete compressive strengths

ranged between 6,800 and 17,900 psi (47 and 123 MPa), and transverse reinforcing ratios

between 0.0037 and 0.0147. Testing concluded that both AASHTO (2007) and ACI (2008) were

conservative estimates for shear resistance when using high strength concretes. Kahn and

Mitchell introduced Eq. (5) that is applicable for both monolithic and rough, cold joints, with a

friction coefficient equal to 1.4:

vn

=

0.05

f


c

+ 1.4v

fy



0.2

f


c

(5)

where

fy = yield stress of transverse reinforcement (60,000psi)

Kahn and Slapkus (2004) tested 6 composite beams with precast, high strength concrete webs with compressive strengths of 12,120 psi (83.6 MPa), and cast-in-place decks with compressive strengths either 7,280 or 11,290 psi (50.2 or 77.8 MPa). The tests concluded that both AASHTO (2007) and ACI (2008) provisions are a conservative estimate for interface shear resistance of composite beams with high strength concrete made with an intentionally roughed interface with protruding aggregate.

6-3

6.2 Test Set up
The five composite beams were created in order to replicate the tests run by Kahn and Slapkus (2004) and to compare shear friction pushoff results with the interface shear results of these composite beam tests. Each beam was 120-in. (3.05 m) long with 114-in. (2.9 m) span between supports. The cast-in-place deck slab had a reduced length of 88 in. (2.2 m) in order to force an interface shear failure. The "precast" web had a depth of 10-in. (254 mm) and a width of 6-in. (152 mm) while the slab was 5-in. (140 mm) deep by 16.5-in. (419 mm) wide as shown in Figures 6-1 and 6-2.

5 in.

No. 3 stirrups

No. 3 bars No. 3 stirrups

10 in.

No. 3 bars

2 in. 1 in.

No. 9 bars

5 in.

6 in.

5 in.

Figure 6-1: Typical T-beam cross section. Note: 1 in. = 25.4 mm; No. 3 bar = 10 M; No. 9 bar = 29 M

6-4

16 in.

4 No. 3 top longitudinal steel (typ.)
88 in.

No. 3 double leg stirrups crossing interface (varied spacing)
2 No. 3 (typ.)

3 in.

114 in.

3 in.

4 No. 9 bottom longitudinal steel (typ.) No. 3 double leg stirrups (typ.)

Figure 6-2: Typical reinforcement for all beams (top).
As discussed in Chapter 4, the web was cast using UHPC at Tindall Corporation precast concrete plant, Conley GA. Two days after casting, the web was thermally treated at 194F (90C) for 48 hours. The UHPC was Lafarge Ductal using 2% by volume steel fiber reinforcement. The 28-day mean compressive strength was 28,930 psi (199.5 MPa). Three hundred and forty two days after casting the web, the deck was placed using conventional high performance concrete (-in. (19 mm) maximum size aggregate) delivered by ready-mix. The HPC had a compressive strength at the time of testing of 12,170 psi (83.9 MPa).
All reinforcement was A615, Grade 60 (415 MPa). The measured yield stress of the No. 9 (29 M) bars was 62,000 psi while the yield strength of the No. 3 (10 M) stirrups was 72,000 psi. The elastic modulus for both the No. 9 (29 M) and No. 3 (10 M) bars was taken as 29,000 ksi (200 GPa).
The main variation between each composite beam was the interface between the web and the flange. The top surface of the UHPC beams proved impossible to roughen by raking, brooming, or cutting with a trowel. To create a rough interface, a form liner was used to achieve the in. (6 mm) roughness amplitude required by codes. The variations between all five of the
6-5

beams tested can be seen in Table 6-1. At the suggestion of Mr. Peter Calcetas of Lafarge North America, burlap was placed atop the interface of one beam containing no reinforcement. The burlap was supposed to create a textured surface, which was to provide an improved bond to the deck slab. The burlap proved difficult to remove from the surface after the initial 48-hour cure and significant wire brushing was required. Further, during moving of a sixth beam intended to be tested that contained a smooth interface and no reinforcement, the deck fell off; that is, there was little bond between the web and slab. The latter beam was not included in any tables.

Beam
0-B 4-CJ 7-CJ 0-FL 7-FL

Table 6-1: Summary of Experimental Results

v Maximum Load at (percent) Load (lb) slip (lb)

Number of Stirrups

0

63,067

13,724

0

0.167 64,754

14,719

4

0.292 71,083

17,890

7

0

66,657

49,270

0

0.292 93,321

93,321

7

Type of Interface
Smooth Smooth Smooth Grooved Grooved

The beams were designated using the following convention X-Y where the first term indicates the number of double-legged No. 3 (10 M) stirrups crossing the interface and the second term indicates the type of interface between the web and the flange. B denotes a burlap interface, CJ denotes a smooth cold joint interface and FL denotes a form liner interface with a grooved surface of -in. (6 mm).
Each beam was tested between 697-703 days after web casting and 355-361 days after deck casting. The beam was loaded in three-point bending as illustrated in Figure 6-3 using a 1,000-kip (4.45 MN) ENERPAC hydraulic actuator. The load was measured with a 700-kip (3.11 MN) load cell. Four mechanical strain gauges made with linear variable displacement transducers (LVDTs) were placed at the center of the beam. Each had a gauge length of 16-in. (406 mm) and these LVDTs were placed on both sides of the beam at the bottom and top of the web. Along with the strain gauges, two LVDT slip gauges were placed on one side of the beam. The body of the LVDT was attached to the web while the extension portion of the LVDT was attached to the underside of the cast-in-place deck measuring the slip between the beam and deck. Finally, the midspan deflection was measured using a string potentiometer.

6-6

Slip LVDT, both sides of centerline, this side of beam only

Load application point

10 in.

33 in.

LVDT strain gauges, front and back

2 in. 2 in.

8 in. 8 in.

Vertical deflection at midspan measured by string potentiometer (not shown)

Figure 6-3: Typical loading and instrumentation
6-7

6.3 Experimental Results and Discussion
For all five beams, the primary failure mode was cracking and slipping between the web and the flange interface shear failure. Figure 4a compares the load deflection curves for all five beams. Interface failure was assumed to occur when the load dropped but there was a significant decrease in flexural stiffness. This interface failure is denoted on Figure 4a with a circle on each load deflection curve. The point at which the load dropped was compared to load slip curves to verify that the interface of the beam had failed. Figure 4a shows that there is a drastic difference between the beams that had a smooth or burlap interface and the beams that had formlinerroughened interfaces. It was observed that the greatest interface shear capacity was obtained from the beam that contained a form liner interface along with 7 stirrups crossing the interface. Similarly, the least interface shear capacity was observed to be the beam that contained burlap along the interface and no stirrups crossing the interface. Finally, it was observed that when interface failure occurred and all composite action was lost, the beams behaved as the rectangular plain beam. When composite action was lost, deflection requirements were no longer satisfied. This can be seen in Figure 4b after interface failure has occurred. Figure 4b also shows the maximum load that each beam achieved and is denoted with a circle at this point.
Figure 6-5 compares the moment curvature diagrams for all beams to the theoretical moment curvature calculated using the computer program Response 2000 developed by Bentz6. As mentioned above, when composite action was lost, each beam began to behave as the plain beam or the theoretical non-composite curve. Stiffer behavior in experimental beams than predicted by Response 2000 can be explained by the non-negligible tensile capacity of the UHPC.
Table 6-1 compares the maximum load and the load at which the interface failed for all five beams. The grooves made with the form liner were the key component in providing composite strength. The beams that did not use form liner failed at an average load of 15,444 lbs (68,698 N), while the beams that did use form liner failed at an average load of 71,296 lbs (317,140 N). For the beam containing both stirrups and form liner, the maximum load was the load at interface failure; therefore, this combination provided the greatest composite interface shear capacity.
6-8

Figure 6-4a. Load-deflection plots for beams with slip location denoted.
Figure 6-4b. Load-deflection plots for beams with maximum load denoted.
6-9

Figure 6-5: Moment-curvature plots for beams with theoretical moment-curvature curves plotted.
As mentioned above, when composite action is lost, deflection requirements are lost or serviceability is not maintained. In order to quantify these deflection requirements, comparisons were made to ACI (2008) deflection requirements. ACI states that immediate deflection due to live load should not exceed L/360, where L is the span length of the beam in inches. For all five beams, the live load was obtained using the equation Pmax = 1.2DL + 1.6LL. This deflection that occurred at this live load was then noted and converted L/X value. These values can be seen in Table 6-2. It can be seen that the only beam that satisfies serviceability requirements based on ACI (2008) is the beam that contains 7 stirrups with a form liner interface. However, when comparing the two beams with no stirrups and different interfaces, there is a drastic difference between the calculated serviceability and the form liner interface is much closer to meeting the serviceability requirements. This trend is also seen with the two 7 stirrup beams. Also, when comparing all five beams, the two beams with the form liner are the only two that come close or do in fact meet the serviceability requirements stated by ACI (2008). It is noted that the AASHTO (2007) deflection requirement of L/800 is 2.2 times less deflection than permitted by ACI. In all cases but beam 7-FL, the composite action fails AASHTO deflection criteria.
6-10

Table 6-2: Live loads and corresponding deflections and calculated serviceability requirement.

Beam

Maximum Load (lb)

Live Load (lb)

Deflection at Live Load (in.)

Serviceability Value

0-B

63,067

38,428

0.671

L/170

4-CJ

64,754

39,483

0.609

L/187

7-CJ

71,083

43,438

0.531

L/215

0-FL

66,657

49,270

0.383

L/298

7-FL

93,321

57,337

0.204

L/559

Table 6-3 compares the experimental interface shear stress based on two loads: the load

at interface failure and the maximum load. These two criteria were chosen because while there is

still a substantial amount of strength remaining in the beam once the interface has failed, all

serviceability requirements have been lost. The experimental interface shear stress was obtained using AASHTO LRFD1 Equation (6) below:

exp

=

Vu bv dv

(6)

where

Vu = failure load being considered

bv = interface width considered to be engaged in shear transfer, in.

dv = the distance between the centroid of the tension steel and the mid-thickness of the

slab, in.

Along with the experimental interface shear stress, Table 6-3 contains the interface shear

stress capacities calculated using the equations 1 through 5.

Table 6-3: Experimental and predicted interface shear stress

Beam

Simplified Simplified ACI Shear Kahn and Loov and

Experimental Experimental Friction Mitchell Patnaik

(Max) vexp

(Slip) vexp

Eq. (3)

Eq. (5) Eq. (4)

AASHTO LRFD Eq. (1)

ACI Eq. (2)

0-B*

956

208

0

608

214

75

40.6

4-CJ

981

223

60

748

591

135

50

7-CJ

1077

271

105

853

760

180

57

0-FL

1010

747

0

608

256

280

40.6

7-FL

1414

1,414

175

853

912

455

57

*0-B assumed unroughened surface because in. amplitude was not achieved with the burlap

interface.

6-11

Table 6-4 and Table 6-5 give the ratio of the experimental interface shear stress (both maximum and slip loads) to the theoretical interface shear stress calculated using the equations 1 through 5 above. It was observed that the Kahn and Mitchell (2002) Equation provided the best estimate for stress along the interface when compared to the experimental interface shear stress due to maximum loads. Similarly, the AASHTO LRFD (2007) Equation provided the best estimate for the stress along the interface when compared to the experimental interface shear stress due to slip loads.

Table 6-4: Ratio of exp max, experimental interface shear stress at maximum load to predicted interface shear stress

Beam

exp max /ACI Shear Friction
Eq. (3)

exp max /Kahn and Mitchell
Eq. (5)

exp max /Loov and Patnaik
Eq. (4)

exp max /AASHTO LRFD Eq. (1)

exp max /ACI Eq. (2)

0-B*

-

1.57

4.47

12.75

23.55

4-CJ

16.35

1.31

1.66

7.27

19.62

7-CJ

10.26

1.26

1.42

5.98

18.89

0-FL

-

1.66

3.95

3.61

24.88

7-FL

8.08

1.66

1.55

3.11

24.81

Mean

11.56

1.49

2.61

6.54

22.35

Standard Deviation

4.29

0.17

1.32

3.46

2.58

Coefficient of Variation

37.1%

11.5%

50.5%

52.8%

11.5%

*0-B assumed unroughened surface because in. amplitude was not achieved with the burlap

interface.

Table 6-5: Ratio of exp slip, experimental interface shear stress at slip load to predicted interface shear stress

Beam

exp slip /ACI exp slip /Kahn exp slip /Loov

Shear Friction and Mitchell and Patnaik

Eq. (3)

Eq. (5)

Eq. (4)

exp slip /AASHTO LRFD Eq. (1)

exp slip /ACI Eq. (2)

0-B*

-

0.34

0.97

2.77

5.12

4-CJ

3.72

0.30

0.38

1.65

4.46

7-CJ

2.58

0.32

0.36

1.51

4.75

0-FL

-

1.23

2.92

2.67

18.40

7-FL

8.08

1.66

1.55

3.11

24.81

Mean

4.79

0.77

1.23

2.34

11.51

Standard Deviation

2.90

0.57

0.95

0.64

8.49

Coefficient of Variation

60.6%

73.8%

76.9%

27.4%

73.8%

6-12

Figures 6-6 through 6-10 compare the experimental interface shear stress (both maximum and slip loads) to the interface shear stress calculated from the five equations. When the experimental interface shear stresses due to maximum loads were compared with theoretical predictions, all of the equations were conservative estimates of the behavior. However, when the experimental interface shear stress due to slip loads were compared, Equations 4 and 5 were not conservative. However, it was observed that when the equations were deemed to be unconservative, it was because form liner was not used in the beam. Only the beams without form liner flutes had an experimental interface shear stress around 200 psi, which was extremely low and similar to the values found by Loov and Patnaik (1994).

1600 1400 1200

T-Beam tests vmax AASHTO Eq. (1) c=75psi and mu = 0.6 AASHTO Eq. (1) c=280 psi and mu=1.0 ACI Eq. 2 T-Beam tests vslip

1000

exp , psi

800

600

400

200

0

0

50

100

150

200

250

300

v f y , psi

Figure 6-6: Comparison of interface shear stress results to predictions using AASHTO LRFD Eq. (1) and ACI Eq. (2). Note: 1 psi = 0.0069 MPa.

6-13

Figure 6-7: Comparison of interface shear results with ACI Shear Friction Eq. (3).
Figure 6-8: Comparison of interface shear stress results to predictions using Loov and Patnaik Eq. (4).
6-14

exp , psi

1600

1400

1200

1000

800

600

400

T-Beam tests vmax

T-Beam tests vslip

Eq. 5 200

0

0

50

100

150

200

250

300

v f y , psi

Figure 6-9: Comparison of interface shear results with Kahn and Mitchell Eq. (5).

Figure 6-10: Comparison of interface shear results with ACI Shear Friction Eq. (3), Loov and Patnaik Eq. (4), and Kahn and Mitchell Eq. (5).
6-15

6.4 Conclusion on Interface Shear Composite Beams
There are no current code requirements on ultra high performance concretes. The only requirement for UHPC is that adequate shear reinforcement must be provided across the interface but that a smooth interface is permitted. It is evident that a smooth interface, even when reinforcement is present, does not provide enough composite strength and fails serviceability requirements. Although this may be acceptable when compared to ultimate loads, it is recommended that any "precast" girder constructed using UHPC with an HPC cast-in-place deck use form liners to create flutes (roughness), which satisfy AASHTO LRFD (2007) conditions for a roughened surface.
The trend in results of the small beams matches those found in the push-off tests.
6-16

7. Full-scale Composite Bridge Girders

7.1 Introduction
Three 32.4-inch deep bridge girders were constructed using UHPC as presented in Chapters 3 and 4. Each was reinforced with 28 bottom and 2 top 9/16-inch diameter grade 270 low relaxation prestressing strands. The overall length of Girder 1 was 54 ft. with center-tocenter bearing length of 52 ft. while that of Girders 2 and 3 was 34 ft. with center-to-center bearing length of 32 ft. In the Bulb Tee (BT) shape of Girders 1 and 2, the top flange width was 18 in. while for Girder 3, the top flange width was 9 in.
A 5-ft wide by 8-in. thick HPC slab was cast atop each girder as illustrated in Figures 317 through 3-19.
The girders were designed so that a shear test could be performed on each end; thus, shear tests were designated as 1-1, 1-2, 2-1 and the like representing girder number followed by the test number. Table 7-1 shows the girder test specimens along with the condition of the top surface of girder for each test, the amount of shear reinforcement and shear reinforcement ratio, area of reinforcement divided by the distance between stirrups and the width of the top flange.

Table 7-1. Bridge Girder Shear Test Parameters

Girder Shear span, a/d ratio surface

test

"a" (in.) and (d=27.83") condition

number a/d ratio

1-1

288 in.

10.3

smooth

1-2

96

3.44

fluted

2-1

96

3.44

smooth

2-2

96

3.44

fluted

3-1

96

3.44

smooth

3-2

96

3.44

fluted

Shear

reinforcement ratio

reinforcement v = Av sbv

2 #4 @ 24-in. 0.000926

none

0.0

none

0.0

2 #4 @ 24-in. 0.000926

none

0.0

2 #4 @ 24-in. 0.00185

7-1

7.2 Test Set-up and Instrumentation
As initially discussed in Chapter 3, all girders were instrumented with LVDT strain gauges and linear potentiometers. Strain gauges made using LVDTs were affixed to the exterior of the beam section as shown in Figure 7-1. One 700-kip load cell was used to determine the load being applied to the system. Half of this load was placed at each of the locations marked P/2. Five string potentiometers were used to measure the deflected shape of the composite girder during the tests. Each potentiometer was attached to the underside of the girder. On each side of the girder and deck, 5 linearly varying displacement transducers (LVDTs) were used to determine the strain profile. Each LVDT had a gauge length of 30 inches in order to capture cracking phenomena. Two additional LVDTs were placed on each side of the beam at the interface of the girder and the deck. One end was attached to the girder and the other to the deck in order to measure any differential motion between the two elements. The gauge length on these slip LVDTs was 30 in. Strain rosettes made using LVDT's within a 12-in. square were applied on one side of the girder midway between the support and the nearest load point.

P/2

P/2

2 ft

2 ft

11 ft

6 1/4 in. 1 in.

12 in.

10 in. 2 in.

2 ft

11 ft

1 ft

Bearing to bearing length = 52 ft

String potentiometer Linear varying displacement transducer (LVDT) [30-in. gauge length] LVDT slip gauge [30-in. gauge length]

13 ft 1 ft

Figure 7-1: Elevation view of external instrumentation for flexure testing of Girder 1. (All instrumentation except string potentiometers is mirrored on other side of beam.)

7-2

Figures 7-2 and 7-3 show photos of the set-up for Test 2-1 through 3-2. Figures 7-4 through 7-6 show typical test set-up photos.

P

16 ft

8 ft

6 1/4 in. 1 in. 7 in.
12 in.
10 in. 2 in.

2 ft 12 in.
12 in. 12.8 in.

9 ft
String potentiometer Linear varying displacement transducer (LVDT) Linear varying displacement transducer (LVDT) [30-in. gauge

4 ft

1 ft

Figure 7-2. Experimental setup and instrumentation for shear tests 2-2 and 3-2

P

8 ft

16 ft

2 ft 6 1/4 in. 1 in.
7 in.
12 in.

10 in. 2 in.

8 ft

1 ft

Bearing to bearing length = 24 ft

9 ft

String potentiometer Linear varying displacement transducer (LVDT) [10-in. gauge length] Linear varying displacement transducer (LVDT) [30-in. gauge length]
LVDT slip gauge [30-in. gauge length]

Figure 7-3. Experimental setup and instrumentation for shear tests 2-1 and 3-1

7-3

Figure 7-4. Overall view of test 1-1, load a midspan of 54-ft. girder
Figure 7-5. Instrumentation on test 1-2, typical of all tests.
7-4

Figure 7-6. Set up for test 2-2, typical for all short shear span tests.
7.3 Experimental Results
All girders were loaded with monotonically increasing load up to failure. Typically, there was an interface shear failure between the deck and the girder. Loading was continued until the girder failed by compression crushing of the top flange or by diagonal tension failure. Testing and results are presented below for each test.
7.3.1 Test 1-1, 54-ft. girder loaded at midspan The load-deflection curve is shown in Figure 7-7 along with a theoretical curve. The
behavior was linear up to a load of 72,400 lbs. at which point slip was noted along the interface, particularly along the end with the smooth interface and #4 stirrups spaced 24-in. on-center. The strain diagram just before slip is shown in Figure 7-8 which indicates the beginning of noncomposite action.
7-5

200

180

160

140

120

Load (kips)

100

80

60

40

20

0

0

1

2

3

4

5

6

Displacement (in.)

Figure 7-7a. Experimental load-deflection curve for test 1-1. Interface shear failure occurred at a total load of 70,050 lbs, a shear of 35,000 lbs. Compression failure of the girder occurred at a
total load of 173,400 lbs.

Figure 7-7b Theoretical and experimental load-deflection curves. With the exception of a low ultimate strength, the girder behaved as would be expected for a partially composite system.
7-6

Figure 7-8. Strain profile of girder and deck at applied load of 60,000 lbs., Test 1-1. Notice the discontinuity in the profile between the girder and the deck.
From this point on in the test, the slope of the load deflection curve decreased significantly, and became much less regular. It is the opinion of the authors that this corresponded to partial loss of composite action as the concrete to concrete bond failed along the interface. From this point on, shear friction along with dowel action became the main mechanism for carrying shear across the interface.
The load slip curve is shown in Figure 7-9.
7-7

Figure 7-9 Slip between girder and deck at different loads. Notice the sharp change in behavior of the east end after 70,000 lbs.
The ultimate failure of the girder occurred following a peak load of 174,480 lbs. due to crushing failure at the top of the girder as shown in Figure 7-10.
Figure 7-10. Crushing of top flange of test 1-1. Note the flexural crack in the deck showing non-composite action.
7-8

7.3.2 Test 1-1A, 20-ft. non-composite girder loaded 10 ft. from end After Test 1-1, the girder was sawn in half so that shear tests could be performed at each
end. The top of the half where the deck has been sheared off was removed. The remaining beam called 1-1A was tested with the single point load 120 inches from the end as illustrated in Figure 7-11. Figure 7-12 shows the compression failure of the top flange after extensive shear cracking during the test.
Figure 7-11. Set-up for test 1-1A. Note transverse shear reinforcement epoxy bonded to the top flange.
Figure 7-12. Compression-shear failure of test 1-1A
7-9

7.3.3 Test 1-2, 24-ft. span loaded at 8 ft. from the end, fluted surface, no stirrup reinforcement The cut half of Girder 1 used for test 1-2 is shown in Figure 7-13. Note that the crushed
region is located beyond the right roller support. The left end of the beam had the formliner with fluted surface.
Figure 7-13. Test 7-13, left half of Girder 1 As the girder was loaded, an interface shear failure occurred through the fluted jointed surface. The crack was later shown to go through the deck concrete. Loading was continued until shear failure occurred as shown in Figure 7-14. The horizontal shear/bond failure occurred after the diagonal tension failure. Note that the horizontal failure was arrested by the vertical reinforcement at the end of the girder. Figure 7-15 gives the load-deflection curve.
Figure 7-14. Diagonal shear failure of test 1-2 at a shear of 431,300 lbs
7-10

700

600

500

400

Load (kips)

300

200

100

0

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

Deflection (in.)

Figure 7-15. Load-deflection curve for test 1-2. Interface shear failure through fluted surface at a total load of 524,950 lbs, shear force of 349,950. Diagonal tension failure of the girder occurred at a total load of 647,000 lbs, shear force of 431,300 lbs.

7.3.4 Test 2-1, 24-ft. span with 8-ft shear span. Smooth surface, no stirrups Figure 7-16 shows the test 2-1. As the beam was loaded, an interface failure occurred
and the deck lifted off the girder as illustrated in Figure 7-17. With further loading, the girder failed in compression as shown in Figure 7-18. The load-deflection curve is shown in Figure 719.

7-11

Figure 7-16. Overall view of test 2-1 with a smooth surface and no stirrup reinforcement at the left end of the girder. Gap liftoff
Figure 7-17. Inteface shear failure with deck liftoff
7-12

Load (kips)

800 700 600 500 400 300 200 100
0 0

Figure 7-18. Test 2-1 compression failure

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

Deflection (in.)

Figure 7-19. Test 2-1, Load-deflection curve. Interface shear failure occurred at a shear force of 27,000 lbs. (40,000 lbs load) and compression failure at 673,000 lbs. load, (449,150 lbs. shear)

7-13

7.3.5 Test 2-2, 24-ft span with 8-ft shear span. Fluted surface, #4 stirrups @ 24-in. o.c. The overall view of test 2-2 before loading is shown in Figure 7-20 . The shear failure of
the girder is shown in Figure 7-21. The load-deflection curve is shown in Figure 7-22.
Figure 7-20. Test 2-2 before loading.
Figure 7-21. Shear failure crack between vertical shear reinforcement.
7-14

800

700

600

500

Load (kips)

400

300

200

100

0

0

0.2

0.4

0.6

0.8

1

1.2

1.4

Deflection (in.)

Figure 7-22. Test 2-2 load-deflection curve. Interface shear of the fluted surface at total load of 632,050 lbs and a shear force of 421,400 lbs; and diagonal tension shear failure occurred at 720,150 lbs. total load, 480,100 lbs. shear force.

7.3.5 Test 3-1, 24-ft span with 8-ft shear span, smooth surface, no stirrups, 9-in. top flange
The narrower top flange of Girder 3 lead to lower interface shear and lower compression failure ultimate load results. Figure 7-23 shows the set up. Figure 7-24 shows the compression failure of the top flange while Figure 7-25 shows the load-deflection curve. Early in the loading, the deck separated from the top flange, an interface shear failure. At the ultimate load, the top flange failed in compression.

7-15

Figure 7-23. Test 3-1 set-up. Notice the narrow 9-in. top flange.
Figure 7-24. Compression failure of top flange. Notice deck separation at the right of the compression failure.
7-16

700

600

500

400

Load (kips)

300

200

100

0

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

Deflection (in.)

Figure 7-25. Test 3-1 load-deflection curve. Interface shear failure occurred at a total load less than 10,000 lbs, shear force less than 6,700 lbs. Compression failure occurred at a total load of
632,500 lbs., and shear force of 421,650 lbs.

7.3.7 Test 3-2, 24-ft span with 8-ft shear span. Fluted surface, #4 stirrups @ 24-in. o.c. Figure 7-26 shows the compression failure of test 3-2.

7-17

Figure 7-26. Compression failure of test 3-2. Note the flexure crack in the deck and the interface crack between the girder and deck.
700

600

500

400

Load (kip)

300

200

100

0

0.00

0.20

0.40

0.60

0.80

1.00

1.20

1.40

1.60

Deflection (in.)

Figure 7-27. Test 3-2 load-deflection curve. Interface shear failure occurred at total load of 458,950 lbs. and a shear force of 153,000 lbs. Compression failure occurred at total load of
666,150 lbs., shear force of 444,100 lbs.

7-18

7.4 Discussion of Results
Table 7-1 shows the overall results. In all cases, the deck separated from the girder in interface shear. Yet the fluted surface with minimum reinforcement did develop the expected interface shear based on results from the push-off tests. Therefore, that one interface condition can be expected to provide adequate service load performance. Even without reinforcement (test 1-2) the fluted surface developed substantial interface shear. When the width of the fluted flange was reduced by 50% (test 3-2), the interface shear decreased by 56%.
That most of the sections, even those with no stirrups, failed in compression of the girder demonstrated that the UHPC possesses outstanding diagonal tension, shear capacity. Test 1-2 demonstrated that the compression and shear capacities were about equal. The wider 18-in. flange of Girders 1 and 2 was necessary to develop the full flexural strength of the section. The narrow flange of Girder 3 diminished the ultimate strength of the section and lead to the compression, flexural failure of both 3-1 and 3-2.
The non-composite girder possesses sufficient flexural and shear capacity for long-span bridge girders. But composite action is needed for stiffness and deflection resistance. To achieve composite action, a fluted surface is needed and more than minimum shear reinforcement is recommended.

Girder test number 1-1 1-2

Shear span,
288 in. 96 in.

2-1

96 in.

2-2

96 in.

3-1

96 in.

3-2

96 in.

(d=27.83")

Table 7-1. Bridge Girder Shear Test Parameters

a/d surface Shear

Interface Ultimate Girder

ratio condition reinforcement Failure

Load

Failure

shear (lbs) (lbs)

Mode

10.3 smooth 2 #4 @ 24-in.

35,000 173,400 flexure

3.45 fluted none

349,950 647,000 diagonal

tension

3.45 smooth none

27,000 673,000 flexure

3.45 fluted 2 #4 @ 24-in.

421,400 720,150 diagonal

tension

3.45 smooth none

6,700 632,500 flexure

3.45 fluted 2 #4 @ 24-in.

153,000 666,150 flexure

7-19

8. Conclusions and Recommendations
8.1 Conclusions
The 38-push-off tests demonstrated that current ACI and AASHTO provisions are conservative for estimating interface shear capacity and shear friction of composite UHPC/HPC structures and of monolithic UHPC. Formliners can be successfully used to create a fluted surface finish in UHPC that is comparable to the 6 mm (1/4-in.) surface roughness recommended by current codes for composite construction. This surface finish provides significant increase in interface shear capacity, particularly when used in conjunction with reinforcing steel crossing the interface. In monolithic UHPC, steel fibers allow for significant shear transfer across pre-existing cracks even when no additional shear reinforcement is used; however, smooth cold joints in UHPC have very low interface shear capacities when transverse shear reinforcement is not used.
Both the push-off tests and small beam tests showed that for smooth, cold joint UHPC surfaces, a friction coefficient = 0.6 be used and that for a fluted cold joint between HPC and UHPC that a friction coefficient = 1.0 be used. It was found for the latter that friction coefficient = 1.4 was safe, but more testing is recommended before a value larger than that given in AASHTO (2007) be used.
The seven tests of the three full-size girders showed that most of the sections, even those with no stirrups, failed in compression of the girder which demonstrated that the UHPC possesses outstanding diagonal tension, shear capacity. The wider 18-in. flange of Girders 1 and 2 was necessary to develop the full flexural strength of the section. The narrow flange of Girder 3 diminished the ultimate strength of the section and lead to the compression, flexural failure of both 3-1 and 3-2.
The non-composite girder possesses sufficient flexural and shear capacity for long-span bridge girders. But composite action is needed for stiffness and deflection resistance. To achieve composite action, a fluted surface is needed and more than minimum shear reinforcement is recommended.
8-1

8.2 Recommendations
It is recommended that the current AASHTO code (2007) be used for the design of interface shear for UHPC-HPC composite bridge girders. Further, it is recommended that the top surface of UHPC girders be fluted (roughened) to assure adequate service load performance. Wide top surface girders are needed to develop the fluted surface, so BT type girders are recommended.
While shear reinforcement is not needed to provide diagonal tension shear strength of UHPC girders, it is recommended that more than minimum shear reinforcement be used to connect the HPC deck to the UHPC girder.
8-2

References
ACI Committee 318 (2008). Building code requirements for structural concrete : (ACI 318-08) ; and commentary (ACI 318R-08). Farmington Hills, Mich., American Concrete Institute.
Ali, M. A. and R. N. White (1999). "Enhanced Contact Model for Shear Friction of Normal and High-Strength Concrete." ACI Structural Journal 96(3): 348-360.
American Association of State Highway and Transportation Officials (2007). AASHTO LRFD bridge design specifications. Washington, D.C., American Association of State Highway and Transportation Officials.
Anderson, A. R. (1960). "Composite designs in precast and cast-in-place concrete." Progressive Architecture 41(9): 8.
ASCE-ACI Committee 426 (1973). The Shear Strength of Reinforced Concrete Members: (ACI 426R-74). Farmington Hills, Mich., American Concrete Institute.
ASCE-ACI Committee 445 (1998). Recent Approaches to Shear Design of Structural Concrete: (ACI 426R-99). Farmington Hills, Mich., American Concrete Institute.
Association Franaise de Gnie Civil (2002). Interim recommendations for ultra high performance fibre-reinforced concretes. France.
Banta, T. E. (2005). Horizontal shear transfer between Ductal and lightweight concrete. Civil Engineering. Blacksburg, VA, Virginia Polytechnic Institute and State University. Masters of Science: 132.
Barragn, B., R. Gettu, et al. (2006). "Shear Failure of Steel Fiber-Reinforced Concrete Based on Push-Off Tests." Aci Materials Journal 103(4): 251.
Bass, R. A., R. L. Carrasquillo, et al. (1989). "Shear Transfer Across New and Existing Concrete Interfaces." ACI Structural Journal 86(4): 383-393.
Birkeland, P. W. and H. W. Birkeland (1966). "Connections in precast concrete construction." ACI Journal 63(3): 345-367.
Delarrard, F. and T. Sedran (1994). "Optimization of Ultra-High-Performance Concrete By the Use of a Packing Model." Cement and Concrete Research 24(6): 997-1009.
Garas, V. Y. (2009). Multi-Scale Investigation of Tensile Creep of Ultra-High Performance Concrete for Bridge Applications. Civil and Environmental Engineering. Atlanta, GA, Georgia Institute of Technology. Ph.D.
R-1

Graybeal, B. A. (2004). Fabrication of an optimized UHPC bridge. PCI National Bridge Conference. Atlanta, GA.
Graybeal, B. A. (2006a). Material Property Characterization of Ultra-High Performance Concrete. F. H. Administration. McLean, VA, U.S. Department of Transportation.
Graybeal, B. A. (2006b). Structural Behavior of Ultra-high Performance Concrete Prestressed IGirders. F. H. Administration. Mclean, VA, U.S. Department of Transportation.
Grossfield, B. and C. Birnstiel (1962). "Tests of T-Beams wiht Precast Webs and Cast-in-Place Flanges." ACI Journal 59(6): 843-851.
Hanson, N. W. (1960). "Precast-Prestressed Concrete Bridges: 2. Horizontal Shear Connections." Journal of the PCA Research and Development Laboratories 2(2): 38-58.
Hermansen, B. R. and J. Cowan (1974). "Modified Shear-Friction Theory for Bracket Design." ACI Journal 71(2): 55-60.
Hofbeck, J. A., I. O. Ibrahim, et al. (1969). "Shear transfer in reinforced concrete." ACI Journal 66(2): 119-128.
Hwang, S.-J., H.-W. Yu, et al. (2000). "Theory of Interface Shear Capacity of Reinforced Concrete." Journal of Structural Engineering 126(6): 700-707.
Kahn, L. F. and A. D. Mitchell (2002). "Shear Friction Tests with High-Strength Concrete." ACI Structural Journal 99(1): 98-103.
Kahn, L. F. and A. Slapkus (2004). "Interface Shear in High Strength Composite T-Beams." PCI Journal 49(4): 102-110.
Krauthammer, T. (1992). "Minimum Shear Reinforcement Based on Interface Shear Transfer." ACI Structural Journal 89(1): 99-105.
Krishnamurthy, N. (1983). "Magnel Diagrams for Prestressed Concrete Beams." Journal of Structural Engineering 109(12): 2761-2769.
Kriz, L. B. and C. H. Raths (1965). "Connections in Precast Concrete Structures - Strength of Corbels." PCI Journal 10(1): 18-61.
Loov, R. E. and A. K. Patnaik (1994). "Horizontal Shear Strength of Composite Concrete Beams." PCI Journal 39(1): 48-67.
Mast, R. F. (1968). "Auxiliary reinforcement in concrete connections." Journal of the Structural Division, ASCE 94(6): 1485-1504.
R-2

Mattock, A. H. (1974a). "Discussion of the paper "Modified Shear Friction Theory for Bracket Design," by B. R. Hermansen and J. Cowan." ACI Journal 71(8): 421-423.
Mattock, A. H. (1974b). Shear Transfer in Concrete Having Reinforcement at an Angle to the Shear Plane. ACI Special Publication 42, Shear in Reinforced Concrete 2. Detroit, Michigan, American Concrete Institute: 17-42.
Mattock, A. H. (1988). "Discussion of the paper "Influence of Concrete Strength and Load History on teh Shear Friction Capacity of Concrete Members," by Walraven, et. al." PCI Journal 33(1): 165-166.
Mattock, A. H. (2001). "Shear Friction and High-Strength Concrete." ACI Structural Journal 98(1): 50-59.
Mattock, A. H. and N. M. Hawkins (1972). "Shear Transfer in Reinforced Concrete - Recent Research." PCI Journal 17(2): 55-75.
Mattock, A. H., L. Johal, et al. (1975). "Shear Transfer in Reinforced Concrete with Moment or Tension Across the Shear Plane." PCI Journal 20(4): 76-93.
Mattock, A. H., W. K. Li, et al. (1976). "Shear Transfer in Lightweight Reinforced Concrete." PCI Journal 21(1): 20-39.
Patnaik, A. K. (1999). "Longitudinal Shear Strength of Composite Concrete Beams with a Rough Interface and No Ties." Australian Journal of Structural Engineering 1(3): 157-166.
Patnaik, A. K. (2001). "Behavior of Composite Concrete Beams with Smooth Interface." Journal of Structural Engineering 127(4): 359-366.
Paulay, T. and P. J. Loeber (1974). Shear Transfer by Aggregate Interlock. ACI Special Publication 42, Shear in Reinforced Concrete 2. Detroit, Michigan, American Concrete Institute: 1-15.
Paulay, T., R. Park, et al. (1974). Horizontal Construction Joints in Cast-in-place Reinforced Concrete. ACI Special Publication 42, Shear in Reinforced Concrete 2. Detroit, Michigan, American Concrete Institute: 17-42.
Saemann, J. C. and G. W. Washa (1964). "Horizontal shear connections between precast beams and cast-in-place slabs." American Concrete Institute -- Journal 61(11): 1383-1409.
Shah, S. P. and W. J. Weiss (1998). Ultra High Performance Concrete: A Look to the Future. 1998 ACI Spring Conference. Houston, TX.
Shaikh, A. F. (1978). "Proposed Revisions to Shear-Friction Provisions." PCI Journal 23(2): 1221.
R-3

Silva, J., L. F. Kahn, et al. (2004). Analytical investigation: ultra high performance concrete for precast prestressed bridge girders: Task 1 report. Atlanta, GA, Georgia Department of Transportation. Somerville, G. (1974). The Behavior and Design of Reinforced Concrete Corbels. ACI Special Publication 42, Shear in Reinforced Concrete 2. Detroit, Michigan, American Concrete Institute: 477-502. Valluvan, R., M. E. Kreger, et al. (1999). "Evaluation of ACI 318-95 shear-friction provisions." ACI Structural Journal 96(4): 473-481. Walraven, J. and J. Stroband (1994). Shear Friction in High-Strength Concrete. ACI International Conference, High-Performance Concrete, Singapore, American Concrete Institute. Walraven, J. C. (1981). "Fundamental Analysis of Aggregate Interlock." Journal of the Structural Division, ASCE 107(11): 2245-2270. Walraven, J. C., J. Frenay, et al. (1987). "Influence of Concrete Strength and Load History on the Shear Friction Capacity of Concrete Members." PCI Journal 32(1): 66-84. Zia, P. (1961). "Torsional Strength of Prestressed Concrete Members." ACI Journal 57(4): 13371360.
R-4