Strength of shear connection in composite bridges with precast decks using high performance concrete and shear-keys

Available online at www.rilem.net Materials and Structures 38 (March 2005) 173-181 Strength of shear connection in composite bridges with precast decks using high performance concrete and shear-keys D. L. Araújo 1 and M. K. El Debs 2 (1) Civil Engineering School, Federal University of Goiás, Brazil (2) Engineering School of São Carlos, São Paulo University, Brazil Received: 26 April 2004; accepted: 13 July 2004 ABSTRACT In this paper, a connection between precast beam and precast slab to be used in decks of precast composite bridges is studied. The difference between this connection and the common connection used in bridge construction is that in this paper the steel connector is associated with a shear-key, which is formed on the top face of the precast beam, and a high performance concrete is used to fill out the shear pockets. The connector is formed by steel bars bent in a hoop form, which are inserted in pockets of the precast slab. The connection is made filling out the shear pockets with steel fiber reinforced concrete. Push-out tests were carried out to evaluate the strength of the shear connection. From these tests expressions based on a shearfriction model are proposed to evaluate the strength of the shear connection taking into account the compressive strength of the concrete cast in the pocket, the diameter of the connector and the addition of steel fibers to cast-in-place concrete. These expressions were shown to be appropriate to evaluate the strength of a shear connection, since the failure of precast concrete is avoided near the connection. 1359-5997 2004 RILEM. All rights reserved. RÉSUMÉ Dans cet article, on étudie le joint entre poutres et dalles pré-fabriquées en béton, employées dans les plateaux des ponts composés, préfabriqués. La différence entre ce joint et ceux habituellement employés dans la construction de ponts demeure entre l association des connecteurs métalliques avec des joints crantés confectionnés sur le côté supérieur de la poutre pré-fabriquée et l emploi de béton de haute performance pour remplir la niche de la dalle. Les connecteurs métalliques sont formés d armatures en acier courbé en boucle qui sont introduites dans les niches de la dalle pré-fabriquée. Le joint est à posteriori réalisé en remplissant les niches avec du béton renforcé de fibres d acier. On a réalisé des essais de cisallement direct à partir desquels sont proposées des formules empiriques basé sur théorie du frottement-cisaillement pour l évaluation de la résistance du joint. Celles-ci prennent en considération la résistance elle-même à la compression du béton moulé dans les niches, le diamètre du connecteur et l addition de fibres d acier au joint. Ces formules se sont montrées convenables à l évaluation de la résistance du joint, pourvu que la rupture des piéces pré-fabriquées soit évitée à sa proximité. 1. INTRODUCTION The connection of precast elements with cast-in place concrete is one of the most common applications in precast construction, generally referred to as composite beam. Such a connection is successfully used in the construction of bridges where the longitudinal beams are precast and the slab is cast-in-place. Some of the main advantages of employing composite beams are the time oonstruction, which is shorter than a solution with cast-in-place concrete only, and the reduction of formwork. These advantages can be maximised if the slab is precast (Fig. 1). In precast concrete structures, there are several ways and materials to make connections between two elements. However, the use oast-in-place concrete is still one of the simplest ways. A connection type that can be used between precast beams and precast slabs consists of the combination of steel connectors with cast-in-place concrete. The connectors are formed by steel bars bent in a hoop and placed in the precast beam during its moulding. They are inserted in the shear pocket of the slab during the assembly of the structure, and the connection is later carried out by placing concrete in these pockets. The shear transfer at the interface is ensured by the connectors and by the contact area between the precast beam and the concrete cast in the slab pockets. Transfer by friction in the contact surface between the precast beam and the precast slab is generally ignored in design since it is not reliable. Although the combination of precast beams and precast slabs has been frequently used in the construction of bridges, there are no reliable methods for the design of the connection between both elements. There are some results obtained from 1359-5997 2004 RILEM. All rights reserved. doi:10.1617/14034

174 D.L. Araújo, M.K. El Debs / Materials and Structures 38 (2005) 173-181 Pockets filled with castin-place concrete Connector Pockets filled with high performance concrete Slab Connector Beam Longitudinal intersect Transversal intersect Precast slab Precast slab Precast beam Pockets filled with castin-place concrete Top view Fig. 2 - Proposed connection between precast beam and precast slab using connector and shear-key. 2. EVALUATION OF THE CONNECTION USING PUSH-OUT TESTS Connector Transversal section Top view Fig. 1 - Usual connection between precast beam and precast slab in concrete bridge decks. tests with the combination of steel beams with precast concrete slabs [1-7]. However, the conclusions obtained from these tests cannot be directly applied to the combination of precast concrete beams and precast concrete slabs because for composite steel beams stud bolts were used as shear connectors. This paper proposes an alteration of the connection between precast beams and precast slab through the manufacture of a shear-key on the top face of the precast beam (Fig. 2). In addition, it also proposes the application of high strength concrete with steel fibers. From push-out tests performed, expressions to evaluate the strength of such a connection type based on a shear-friction model are proposed considering the influence of the strength of the cast-in-place concrete, the connector diameter and the addition of steel fibers. Previous studies using push-out tests demonstrated that the use of shear-keys, associated with steel fiber reinforced concrete, increases significantly the strength of the connection compared to connections with plane surfaces (smooth or rough). Such an increase is limited by the strength of the precast concrete and the strength oast-in-place concrete used in the pockets [8, 9]. 2.1 Description of tests Nineteen push-out tests with shear-keys were carried out. The variables analysed in the tests were the compressive strength ooncrete cast in the pockets, the diameter of the connector and the volume of fibers added to the connection. The main characteristics of the specimens are given in Table 1. The specimen used to apply shear stresses to the connection is shown in Fig. 3. It was composed of a central piece (simulating the precast beam) and two lateral pieces (simulating the precast slab). The connector consisted of a steel bar bent in a hoop shape. The connection between the central piece and the lateral pieces was made oast-in-place concrete in the existent pockets of the lateral pieces (Fig. 4). The dimensions of the specimens used in these tests are also shown in Fig. 3. These dimensions were chosen taking into account both the recommendations of BS 5400 [10] and the internal dimensions of the equipment available for the tests. The M1-0-0 specimen was made without a connector in the connection. The aim of this test was to quantify the contribution of the concrete to the connection strength. An adhesive was applied to the surface of the precast piece prior to casting the concrete in the pocket to guarantee the adherence between the shear-key and the precast central piece. This should ascertain that the failure would occur in the shear-key and not by the loss of adherence between the concrete cast in the pocket and the precast concrete. The tests were carried out with displacement control seven days after the second connection was cast. A closedloop equipment with a capacity of ±3000 kn in static load and ±2500 kn in dynamic load was used.

D.L. Araújo, M.K. El Debs / Materials and Structures 38 (2005) 173-181 175 Table 1 Details of push-out tests Connection 1 s (mm) V f (%) 2 Specimen 3 0 - M1-0-0 8 - M1-8-0 Mixture 1 10 - M1-10-0 12.5 - M1-12.5-0 12.5 1.50 M1-12.5-1.50 8 - M2-8-0 8 0.75 M2-8-0.75 8 1.50 M2-8-1.50 10 - M2-10-0 10 0.75 M2-10-0.75 Mixture 2 10 1.50 M2-10-1.50 12.5 - M2-12.5-0 12.5 0.75 M2-12.5-0.75 12.5 0.75 M2-12.5-0.75-b 12.5 0.75 M2-12.5-0.75-c 12.5 1.50 M2-12.5-1.50 8 - M3-8-0 Mixture 3 10 - M3-10-0 12.5 1.50 M3-12.5-1.50 (1) The compressive strength ooncrete used in connection ranged from 50 MPa to 100 MPa; (2) The steel fibers used were DRAMIX RL-45/30 BN with hooked ends; (3) In the nomenclature of the specimens, the first two letters are the mixture used in connection, the following number is the diameter of the connector and the last number is the volume of fibers added to the connection. For example, M1-0-0 is a specimen in which mixture 1 was used in the connection, without connector and without addition of fibers. Direction of loading 120 120 300 135 140 185 75 150 75 300 Side View 30 120 Pocket (140 x 150) 50 Front View 150 460 540 Top View 610 Fig. 3 - Dimensions of the specimen used in push-out tests. 300 Fig. 4 Details of specimen manufacture for push-out test. 2.2 Characteristics of the materials employed in the tests Three mixtures were used to make the connection between the precast elements, with a compressive strength ranging from 50 MPa to 100 MPa as shown in Table 3. The mix proportions per cubic meter ooncrete are given in Table 2. In this table, mixtures 1, 2 and 3 were used to cast the connections and mixture 4 was used to cast the precast elements. By using high-early-strength Portland cement, it was possible to perform a test only seven days after casting the connection. Silica fume was added to the concrete employed in the connection using a proportion of 10% of the cement weight. The sand and coarse aggregates were obtained in the region, and the coarse aggregate was 19 mm in maximum size. A superplasticizer was also added to improve the workability of the mixture. Table 2 Mixture proportion per cubic meter ooncrete Mixture Cement (kg) Silica fume (kg) Sand (kg) Coarse aggregate (kg) W/C Superplasticizer (%) 1 380 38 631 1132 0.56 1.50 2 460 46 616 1104 0.42 1.80 3 640 64 563 1011 0.32 2.20 4 345-759 1173 0.55 0.37 (1) Percentage by weight oement

176 D.L. Araújo, M.K. El Debs / Materials and Structures 38 (2005) 173-181 Table 3 Concrete strength on push-out tests Specimen Precast concrete Cast-in-place concrete m tm,sp m tm,sp M1-0-0 53.58 4.39 55.10 4.17 M1-8-0 47.40 3.90 54.00 3.90 M1-10-0 47.40 3.90 54.00 3.90 M1-12.5-0 51.06 3.83 48.01 3.49 M1-12.5-1.50 51.60 4.00 55.75 7.50 M2-8-0 73.49 4.58 83.80 4.92 M2-8-0.75 84.61 5.11 88.60 7.61 M2-8-1.50 73.08 4.23 80.00 9.42 M2-10-0 67.63 4.04 72.81 4.65 M2-10-0.75 66.79 3.52 71.87 5.89 M2-10-1.50 66.79 3.52 72.07 8.41 M2-12.5-0 73.49 4.58 83.80 4.92 M2-12.5-0.75 53.58 4.39 80.92 7.68 M2-12.5-0.75-b 53.58 4.39 80.92 7.68 M2-12.5-0.75-c 84.61 5.11 88.60 7.61 M2-12.5-1.50 67.63 4.04 71.35 8.79 M3-8-0 51.06 3.83 96.76 5.45 M3-10-0 51.06 3.83 96.76 5.45 M3-12.5-1.50 51.60 4.00 96.35 10.05 The steel fibers employed were DRAMIX RL-45/30 BN with hooked ends. They were 30 mm long with a diameter of 0.62 mm, resulting in an aspect ratio of 48 and a minimum tensile strength of 1250 MPa. Two volumes of fibers were used, i.e., 0.75% and 1.50%, which corresponded to 60 kg and 120 kg of fibers per cubic meter ooncrete, respectively. The fibers were added at the end of the mixture process of all other components; subsequently mixing was continued for one more minute. The addition of the fibers to the concrete did not alter its compressive strength, but it improved its tensile strength obtained from third point load tests up to 34%. An increase in the ductility of high strength concrete up to 145% was observed. Other information about the influence of fibers on the behaviour ooncrete can be found in [9]. The steel employed for the manufacture of the precast elements and the connectors had a well-defined yield strength. In direct tension tests an average yield strength of 553 MPa and an average modulus of elasticity of 210 GPa had been determined. 2.3 Results of push-out tests Table 4 shows the ultimate load of the push-out tests. This table also shows the normal stress in the shear-off plane at the moment oonnection failure, n, and the average slide between precast elements, m, also at the moment oonnection failure. The normal stress in the shear-off plane is transmitted by the connector crossing this plane, and can be calculated by a shear-friction model dealt in more detail in [11]. In this model, a rough interface is idealised by several small teeth without friction. When a horizontal load is applied, one portion tends to slide from the other portion. However, because of the small teeth, the portions tend to separate and the transversal reinforcement will be tensioned and will produce a normal load at the interface. In the push-out tests, electric strain gauges were glued to the connector in the shear-off plane and the steel strains ( s ) were measured. From the strain measurements ( s ), the normal stress ( n ) is calculated by dividing the normal load applied by the connector through the area of the shear-off plane which corresponds to the pocket area: 2 s 2 4 A lig Table 4 Results of push-out tests Specimen n u m (1) (mm) M1-0-0 0 0 196.05 9.34 0.57 M1-8-0 1.89 2.65 230.70 10.99 0.62 M1-10-0 3.09 4.13 250.10 11.91 0.75 M1-12.5-0 5.03 6.46 259.95 12.38 0.90 M1-12.5-1.50 4.66 6.46 358.30 17.06 1.81 M2-8-0 2.65 2.65 286.15 13.63 0.76 M2-8-0.75 2.65 2.65 320.10 15.24 0.96 M2-8-1.50 2.65 2.65 345.30 16.44 1.15 M2-10-0 3.65 4.13 316.55 15.07 0.89 M2-10-0.75 4.01 4.13 363.10 17.29 0.93 M2-10-1.50 4.13 4.13 383.85 18.28 1.12 M2-12.5-0 6.34 6.46 349.40 16.64 0.99 M2-12.5-0.75-6.46 461.90 22.00 1.43 M2-12.5-0.75-b - 6.46 453.40 21.59 2.40 M2-12.5-0.75-c 5.74 6.46 519.50 24.74 1.19 M2-12.5-1.50 6.24 6.46 459.55 21.88 1.32 M3-8-0 2.65 2.65 310.00 14.76 0.96 M3-10-0 4.13 4.13 345.75 16.46 1.08 M3-12.5-1.50 6.46 6.46 421.75 20.08 2.21 (1) n is calculated by expressions (1a) and (1b), but not larger than f y. (1a) (1b) n ses f y f y F exp (kn) where is the geometric ratio of transversal steel in the connection, is the connector diameter, A lig is the cross section of key base, which is equal to the pocket area (150 mm x 140 mm), s is the measured strain in the connector at the failure, E s is the modulus of elasticity of steel and f y is the yield strength of the steel. Table 4 includes also the product of the geometric reinforcement ratio and the reinforcement yield strengths (.f y ) which for the shear-friction model is the maximum value the normal stress ( n ) could reach. 2.3.1 Influence oompressive strength ooncrete cast into the pockets In most push-out tests, the connection failure occurred by shear of the concrete in the shear-key, allowing the conclusion that the connection strength is influenced significantly by the strength of the concrete cast into the pocket. In fact, from the analysis of specimens without fibers it was possible to observe that increasing the compressive strength of the concrete cast into the pocket from about 50 MPa to about 100 MPa increased the connection strength by 35%. This relative increase, however, was smaller than the relative increase in

D.L. Araújo, M.K. El Debs / Materials and Structures 38 (2005) 173-181 177 compressive strength of the concrete cast into the pocket. This is so because in the case of high strength concrete the shear plane intercepts the coarse aggregates, while in the case of normal strength concrete this plane follows the interface between paste and aggregates, being, therefore, rougher than in the cast of high strength concrete. Thus, to obtain for high strength concrete a relationship between the shear strength of the connection and compressive strength of the pocket concrete ( u /m ) which is similar to the relationship for normal strength concrete, the employment of more resistant aggregates becomes necessary. 2.3.2 Influence of the connector diameter Increasing the connector diameter from 8 mm to 12.5 mm resulted in a moderate increase in the ultimate load of the connection. This increase was more significant for the specimen where the connection was made with mixture 3. In that case, there was an increase by 22%, while in the specimen with the connection made with mixture 1 the increase was only 13%. The addition of fibers to the connection seems to increase the influence of the connector. When the diameter of the connector increased from 8 mm to 12.5 mm in specimens with mixture 2, the connection strength increased by 62% and by 33% when 0.75% and 1.5% of fibers were used, respectively. The addition of 1.5% of fibers provided a smaller contribution of the connector. This fact can be related to the casting defects, which occurred when the largest volume of fibers was used. Although obtaining a workable mixture, some difficulty occurred to mould the concrete using a needle vibrator when the concrete was cast into the pocket. There exist many empirical expressions in the literature based on the shear-friction model to evaluate the ultimate shear strength ooncrete interfaces. The comparison of the test results obtained from specimens without fibers with some empirical expressions based on the shear-friction model extracted from the literature is shown in Fig. 5. It is observed that these expressions do not represent the authors test results satisfactorily. In the expression proposed by [12], the ultimate shear strength is just a function of the normal stress acting on the interface. On the other hand, in the expressions proposed by [13] and [14], a portion of the connection strength is attributed to the shear strength of the concrete. This portion, however, is much smaller than the one observed experimentally in the test of the M1-0-0 specimen. Bakhoum [15] carried out tests on shear-keys employed in connection with precast segmental construction of bridges. The comparison between the expressions proposed in [15] and the results obtained in the present study are shown in Fig. 6. Two expressions are given: one obtained from tests with dry joints and another from tests with epoxy joints similar to the test carried out on the M1-0-0 specimen (in the case of a dry joint, one of the precast elements was simply placed on top of the other). It is possible to observe that the strengths deduced from the expressions of Bakhoum [15] are smaller than those obtained in the present study, however, the expressions for epoxy joints are closer to the results found in this research. Shear strength, u Shear Strength, u 20 18 16 14 12 10 8 6 4 2 0 20 18 16 14 12 10 8 6 4 2 0 u =( n 0.5 0.73 )/3.82 < 0.3 (MATTOCK, 1994 [12]) = 100 MPa =75MPa =50MPa =75MPa Tests u =K 1 + 0.8 n < 0.3 e 16.56 MPa with K 1 = 0.1 < 5.52 MPa e n >K 1 /1.45 (MATTOCK, 2001 [14]) 0 1 2 3 4 5 6 7 Normal stress, n Tests u = 0.467 0.545+ 0.8 n < 0.3 (MATTOCK, 1988 [13]) = 100 MPa =50MPa 0 1 2 3 4 5 6 7 Normal stress, n Fig. 5 - Comparison between push-out tests and the ultimate shear strength evaluated by the shear-friction model (V f = 0%). Shear Strength, u 20 18 16 14 12 10 8 6 4 2 0 Tests - V f =0% = 100 MPa =50MPa Dry joint - u = 0.648 0.5 +1.36 n Epoxy joint - u =0.922 f 0.5 c +1.20 n 0 1 2 3 4 5 6 7 Normal stress, n Fig. 6 - Comparison between push-out tests and the ultimate shear strength evaluated by empirical expression proposed by [15]. 2.3.3 Influence of fiber volume The addition of steel fibers to the concrete improved some of its properties, such as its tensile strength [9]. Thus, an increase in the strength of the connection was expected due to the addition of fibers to the concrete cast in the pocket. In fact, an increase by up to 32% was observed when 1.5% of fibers were added.

178 D.L. Araújo, M.K. El Debs / Materials and Structures 38 (2005) 173-181 In the connection made with mixture 2, the addition of 0.75% of fibers increased the ultimate load of the connection by, approximately, 13%, and the addition of 1.50% of fibers increased the ultimate load of the connection by, approximately, 21%. These values were observed on specimens with a connector of 8 mm and 10 mm, respectively. In the case of the connector with a diameter of 12.5 mm, there was a more significant increase. The addition of 0.75% of fibers provided an average increase by 37%, and the addition of 1.50% resulted in an average increase by 32%. The most significant increase in the strength in the case of the connector with the largest diameter can be attributed to the dowel action of such a connector. In this case, close to failure, the concrete area around the connector is subjected to compression and tension splitting. The connection failure occurs due to crushing of the concrete that serves as a support to the connector. Such phenomena are even more pronounced for connectors with diameters larger than 12.5 mm. The addition of fibers improves the behaviour of the concrete when it is subjected to tension splitting, increasing the dowel action resistance of the connector. Such behaviour was observed by other researchers, who noticed an increase in the dowel action strength due to the addition of fibers [16]. Comparing specimens M1-12.5-1.50 and M3-12.5-1.50, it is possible to verify the influence of the compressive strength of the concrete cast in the pockets when maximum values of the diameter of the connector and of the volume of fibers are chosen. In this case, the increase in the connection strength was smaller than the increase observed in the tests with mixture 2. The reason is that failure did no longer occur in the shear-key, but by the crushing of the central piece of the specimen which simulated the precast beam (Fig. 7). This result shows that a pronounced increase in the compressive strength of the concrete at the connection transfers the failure to the precast concrete. At this stage the assumption appears to be reasonable that the failure no longer occurs in the shear-key, but rather in the precast beam if the ratio between the tensile strength of the cast-in-place concrete and of the precast concrete exceeds 2.5. Fig. 8 shows the influence of the fibers on the energy absorbed by the connection until failure. In this paper, the energy, in kn.mm, is defined by the area under the load-slide curve obtained in the tests, being limited by the maximum load of the connection. In this figure, the energy is divided by both the compressive strength of the concrete and the diameter of the connector to eliminate the influence of these variables. Despite the scatter of the results, it is possible to clearly notice the energy increase with an increase in the fiber volume. Such increase is caused by the increase in the connection strength, but, mainly, by the increase of the average slide between precast elements at failure, which result in the increase in the ductility of the connection. 3. STRENGTH OF THE CONNECTION BETWEEN PRECAST BEAM AND PRECAST SLAB WITH SHEAR-KEYS The objective of this research was to obtain an expression based on the shear-friction model to evaluate the connection strength with a shear-key. For this, the normal stress to the shear plane, n, was divided by the square root of the compressive strength of the concrete cast in the pocket, as shown in Fig. 9. In this figure, the results are also shown for the assumption that the connector always reaches the yield strength of the steel before the connection reaches the failure, i.e., n = f y. From a linear regression, the following expressions were obtained: u 1.302 m 0.767n 1.8 m if V f = 0% (R = 0.94) (2a) Fig. 7 - Failure of the specimens with shear-key and 1.50% of fibers. Energy (mm 2 ) m s 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 Tests 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 Volume of fibers, V f (%) Fig. 8 Influence of fibers on the energy absorbed by connection.

D.L. Araújo, M.K. El Debs / Materials and Structures 38 (2005) 173-181 179 2.2 2.0 1.8 τ = 1.302f cm + 0.767 n (R = 0.94) u 2.8 2.6 2.4 2.2 /f u cm 1.6 1.4 1.2 1.0 Tests - without fiber Linear regression 0.0 0.2 0.4 0.6 0.8 1.0 u /m 2.0 u = 1.238f cm + 1.791 n 1.8 (R = 0.922) 1.6 Tests - with fiber 1.4 Linear regression 1.2 0.2 0.3 0.4 0.5 0.6 0.7 0.8 n /m n /m 2.2 2.0 1.8 τ u = 1.270 (m ) + 0.798 f y < 1.8 (m ) (R = 0.929) 2.8 2.6 2.4 2.2 u /m 1.6 1.4 1.2 1.0 Tests - without fiber Linear regression 0.0 0.2 0.4 0.6 0.8 1.0 f y /m /f u c m 2.0 1.8 1.6 1.4 1.2 = 1.730 f 0.7075 ( f ) 0.4150 u cm y (R = 0.948) Tests - with fiber Non-linear regression 0.8 1.2 1.6 2.0 2.4 2.8 (f y m ) Fig. 9 - Empirical expression to evaluate the connection strength with shear-key and without fibers. u 1.270 m 0.798f y 1.8 fcm if V f = 0% (R = 0.93) (2b) where m is the average compressive strength of the concrete cast in the pockets. Comparing these expressions with Bakhoum s [15], relations shown in Fig. 6, one can observe a larger contribution of the concrete to the strength of the connection. On the other hand, the portion of strength due to the normal stress of the shear plane is smaller. This fact can be justified as in the tests by [15] the normal stress to the shear plane was guaranteed by external stresses that simulated the effect of a prestress in the connection. On the other hand, in the tests carried out in this study, the normal stress was just generated by the deformation of the connector. From the authors test results, an upper limit was proposed for the strength of the connection based on the concrete strength of the concrete cast in the pocket. To estimate the connection strength with shear-keys, expression (2b) is suggested for engineering applications as it is easier to use. From a linear regression of the results of tests on specimens with fibers, the following expressions were obtained to evaluate the connection strength (Fig. 10): Fig. 10 - Empirical expression to evaluate the connection strength with shear-key and fibers. u 1.238 m 1.791 n 2.6 m, if 0.75% V f 1.50% (R = 0.92) 0.415 f y 2.6 fcm 0.708 u 1.730f cm, if 0.75% V f 1.50% (R = 0.95) (3a) (3b) where m is the average compressive strength of the concrete cast in the pocket. Expression (3b) is valid only for > 0.005, as tests were not carried out on specimens with fibers and without a connector. In cases where this limit is exceeded, expression (2b) may be used. According to expressions (3a) and (3b) the contribution of the normal stress is larger than it is for connections without fibers. An upper limit for the connection strength based on the strength of the concrete cast in the pocket is also proposed. However, a clear upper limit was not observed, as the strength of the connection also depends on the compressive strength of the precast concrete. This fact means that since with an increase in the shear-key strength, eventually failure occurs no longer in the shear-key, but rather in the precast concrete elements. Therefore, a limit of the ratio between tensile strength of the cast-in-place

180 D.L. Araújo, M.K. El Debs / Materials and Structures 38 (2005) 173-181 concrete and of the tensile strength of the precast concrete equal to 2.5 is recommended. It was not possible to correlate the strength of the connection with the volume of fibers, as only insufficient experimental results from specimens with 1.5% of added fibers were available. To evaluate the connection strength with shear-key and fibers, the expression (3b) is suggested, as it is easier to be used. Expressions (2) and (3) were applied to the specimens tested, and the results are shown in Table 5. The ratios of the strengths obtained from the empirical equations (2a) and (3a) and the experimental values are as an average 0.994 with a standard deviation of 0.045. The ratios of the strengths obtained from the empirical equations (2b) and (3b), for which the assumption n =.f y is valid, and the experimental values are as an average 1.002 with a standard deviation of 0.044. This value is practically identical to the value obtained with the expressions (2a) and (3a), which justifies its use in cases where the normal stress n is not known for the design of a connection. Taking into account only the tests on specimens without fibers, the ratio of the values obtained from expression (2b) and the experimental values are as an average 1.000 with a standard deviation of 0.038. For tests on specimens with fibers, an average ratio of the strengths estimated by expression (3b) and the experimental values of 1.003 with standard deviation of 0.052 was to be drawn. Table 5 - Comparison between the ultimate shear strength obtained from tests and from empirical expressions proposed Specimen u u,1 (1) u1 u u,2 (2) M1-0-0 9.34 9.66 1.035 9.43 1.009 M1-8-0 10.99 11.02 1.002 11.45 1.042 M1-10-0 11.91 11.94 1.002 12.63 1.060 M1-12.5-0 12.38 12.47 1.007 12.47 1.007 M1-12.5-1.50 17.06 17.59 1.031 17.44 1.022 M2-8-0 13.63 13.95 1.024 13.74 1.008 M2-8-0.75 15.24 16.40 1.076 16.72 1.097 M2-8-1.50 16.44 15.82 0.962 15.56 0.946 M2-10-0 15.07 13.91 0.923 14.13 0.938 M2-10-0.75 17.29 17.68 1.022 17.34 1.003 M2-10-1.50 18.28 17.91 0.980 17.37 0.950 M2-12.5-0 16.64 16.48 0.990 16.48 0.990 M2-12.5-0.75 22.00 - - 22.70 1.032 M2-12.5-0.75-b 21.59 - - 22.70 1.052 M2-12.5-0.75-c 24.74 21.93 0.887 24.21 0.978 M2-12.5-1.50 21.88 21.63 0.989 20.77 0.949 M3-8-0 14.76 14.84 1.005 14.61 0.990 M3-10-0 16.46 15.98 0.971 15.79 0.959 M3-12.5-1.50 20.08 23.72 1.181 (3) 25.52 1.271 (3) Average 0.994 1.002 Standard deviation 0.045 0.044 Variation coefficient 4.5% 4.4% (1) Values obtained with expressions (2a) e (3a); (2) Values obtained with expressions (2b) e (3b), i.e., with n = f y ; (3) Value not considered in the calculation of the average due to the failure in the precast concrete u2 u 4. CONCLUSIONS The main conclusions obtained from push-out tests with shear-keys are: An increase in the compressive strength of the concrete cast in the pockets from approx. 50 MPa to 100 MPa increased the connection strength by 35%, on average. An increase in the connector diameter also increased the connection strength, particularly when fibers were added. Increases in the connection strength up to 62% were observed if the connector diameter was increased from 8 mm to 12.5 mm. The addition of 0.75% by volume of steel fibers to the concrete cast in the pockets resulted in an increase of the connection strength up to 37%. More significant, however, was the increase in energy absorbed until the connection had reached its ultimate load due to the increase in the slide between the precast elements. The expressions proposed to evaluate the strength of the shear connection between precast beam and precast slab were a function of the compressive strength of the concrete cast in the pocket, the amount of steel perpendicular to the connection and the addition of steel fibers. The ratio between the strength obtained from the proposed expressions and the strength obtained from tests was as an average 1.002 with a standard deviation of 0.044. Thus, the proposed expressions are appropriate to evaluate the strength of this type oonnection, if failure of the precast concrete in the vicinity of the connection is prevented, and if the ratio between the tensile strength of the concrete employed in the connection and the tensile strength of the precast concrete is smaller or equal to 2.5. These values were obtained assuming that the normal stress actions on the shear-off plane and generated by the connector at maximum load is f y. ACKNOWLEDGEMENTS The authors would like to acknowledge CAPES and FAPESP for the financial support of this research, and the Brazilian companies Camargo Corrêa Cimentos S.A. and Belgo-Mineira Bekaert Arames S.A. for the donation of materials used in the tests. NOTATION E s : modulus of elasticity of steel F exp : maximum load oonnection obtained from tests R: correlation factor V f : volume of fibers in percent, m : average compressive strength ooncrete obtained from tests on 100 mm x 200 mm cylinders tm,sp : average tensile splitting strength ooncrete obtained from tests on 100 mm x 200 mm cylinders

D.L. Araújo, M.K. El Debs / Materials and Structures 38 (2005) 173-181 181 f y : yield strength of steel m : average slide at maximum load oonnection in push-out tests s : strain oonnector at maximum load of connection in push-out tests s : diameter oonnector : geometric ratio of transversal steel in connection. n : normal stress by connector on shear-off plane at maximum load u, u,1, u,2 : ultimate shear strength oonnection obtained from tests and using empirical expressions REFERENCES [1] Shim, C.S., Lee, P.G. and Chang, S.P., Design of shear connection in composite steel and concrete bridges with precast decks, Journal of Constructional Steel Research 57 (3) (2001) 203-219. [2] Lam, D., Elliott, K.S. and Nethercot, D.A., Designing composite steel beams with precast concrete hollow-core slabs, Proceedings of the Institutional Engineering Structures & Buildings 140 (2000) 139-149. [3] Lam, D., Elliott, K.S. and Nethercot, D.A., Experiments on composite steel beams with precast concrete hollow core floor slabs, Proceedings of the Institutional Engineering Structures & Buildings 140 (2000) 127-138. [4] Issa, M.A., Yousif, A.A. and Issa, M.A., Experimental behavior of full-depth precast concrete panels for bridge rehabilitation, ACI Structural Journal 97 (3) (2000) 397-407. [5] Yamane, T., Tadros, M.K., Badie, S.S. and Baishya, M.C., Full depth precast, prestressed concrete bridge deck system, PCI Journal 43 (3) (1998) 50-66. [6] Issa, M.A., Yousif, A.A., Issa, M.A., Kaspar, I.I. and Khayyat, S.Y., Analysis of full depth precast concrete bridge deck panels, PCI Journal 43 (1) (1998) 74-85. [7] Lam, D., Elliot, K.S. and Nethercot, D.A., Push-off tests on shear studs with hollow-cored floor slabs, The Structural Engineer 76 (9) (1998) 167-174. [8] Araújo, D.L., Shear between precast beam and precast slab joined by pockets filled with high performance concrete, PhD Thesis. (Engineering School of São Carlos, University of São Paulo, 2002) [only available in Portuguese]. [9] Araújo, D.L. and Debs, M.K., Application of high performance concrete to the connection between precast beam and precast slab, in High-Performance Concrete: Performance and Quality of Concrete Structures, Proceedings of the 3 rd International Conference, Recife, 2002. (ACI, Farmington Hills, 2002, SP-207) 339-359. [10] British Standards Institution - BSI. Steel, concrete and composite bridges. Part 5: code of practice for design of composite bridges (BS 5400: Part 5: 1979). [11] Birkeland, P.W. and Birkeland, H.W., Connections in precast concrete construction, Journal of American Concrete Institute, Proceedings 63 (3) (1966) 345-367. [12] Mattock, A.H., Comments on Horizontal shear strength of composite concrete beams with a rough interface, PCI Journal 39 (5) (1994) 106-108. [13] Mattock, A.H., Comments on Influence ooncrete strength and load history on the shear friction capacity ooncrete members, PCI Journal 33 (1) (1988) 166-168. [14] Mattock, A.H., Shear friction and high-strength concrete, ACI Structural Journal 98 (1) (2001) 50-59. [15] Bakhoum, M.M., Shear behavior and design of joints in precast concrete segmental bridges, PhD Thesis. (Massachusetts Institute of Technology, Cambridge Mass, 1991) [16] Soroushian, P. and Mirza, F., Effects of fiber reinforcement on cyclic behavior of dowel bars, Journal of Structural Engineering 117 (3) (1991) 822-828.