STATIC TESTS ON VARIOUS TYPES OF SHEAR CONNECTORS FOR COMPOSITE STRUCTURES

Transcription

1 STATIC TESTS ON VARIOUS TYPES OF SHEAR CONNECTORS FOR COMPOSITE STRUCTURES Hans (J.) C. Galjaard* and Joost C. Walraven** *Van Hattum en Blankevoort/Volker Stevin Construction Europe, The Netherlands **Delft University of Technology, Division of Concrete Structures, The Netherlands Abstract The results and interpretations of push-out tests on shear connector devices for steelconcrete composite structures/bridges carried out in the Stevin laboratory are presented. The devices under investigation are headed studs, perfobondstrips, oscillating perfobondstrips, waveform strips and s. The oscillating-perfobondstrip, waveform strip and have specially been developed for this research project. Different concrete grades and types are used, like ordinary concrete, lightweight concrete, high strength concrete and lightweight high strength concrete, with and without steel fibres. The tests are done as specified in EuroCode 4 for standard push-out test. Large differences in ductility and strength between the various connector devices and concrete types have been observed. Sometimes unexpected behaviour occurred during testing, like brittle splitting failure in case of the (oscillating) perfobondstrip in normal concrete. A direct comparison for a specific shear connector device in combination with a certain concrete grade/type is possible. Although the number of samples in a test series is limited, it is still realistic to draw conclusions with respect to their strength/ductility based on EuroCode Introduction Application of high strength (lightweight) concrete could be useful for bridge decks, not only because of its enhanced strength, but moreover because of its improved endurance in a harsh environment. The design code for steel-concrete composite bridges EuroCode 4-2 [1] is however limited to concrete grades up to a strength of C5/6 (3.1.1(2)). This discourages the use of these materials for the decks of steel-concrete bridges. In steelconcrete composite bridges headed studs are often applied as an economic shear connection device. A major drawback is that the strength for concrete grades higher than C3/37 is governed by the strength of the steel cross section of the stud. Hence higher 1313

2 concrete grades will not be utilised by this connector device. Another drawback is the impossibility to automate the welding of headed studs. A prime objective of this research project is therefore the development of new shear connection devices with less disadvantages. For the purpose of comparison, and to investigate their behaviour in high strength (lightweight) concrete, existing shear connection devices like the headed stud and the perfobondstrip have been included in the research project. 2. Connector devices New connector devices were proposed, from which the three most promising devices have been selected based on economics, feasibility and diversity. The devices tested, including the old headed-stud and perfobondstrip, are: - φ 19 mm and a length of 125 mm, see Fig Continuous perfobondstrip with a height of 1 mm, a thickness of 12 mm and 5 holes φ 5 mm, Fig. 1: φ19 mm see Fig. 2. A continuous strip was selected because it is better weldable, using standard shop welding equipment, than a discontinuous one. Two fillet welds of 7 mm have been used to weld in on the HE-section. - Oscillating perfobondstrip with a height of 1 mm, a thickness of 8 mm, 5 holes φ 5 mm, and bend in 1.5 wave with an amplitude of 11 mm, see Fig. 3. It is believed that the curved form will Fig. 2: Continuous perfobondstrip give a better force transfer between steel and concrete compared with a straight connector. It is however recognised that welding might be difficult using present automated weld equipment. The strip is welded to the HE-shape with two fillet welds of 5 mm. - Waveform strip with a width of 5 mm, a thickness of 6 mm and bend in 2 waves with amplitude 11 mm, see Fig. 4. The idea was to use Fig. 3: Oscillating perfobondstrip point weld equipment for the production. Equipment with sufficient capacity is however very scarce, and it is doubtful whether the connector could be successfully welded in this way. For the tests it was welded to the HE-shape with propwelds φ25 mm. Fig. 4: Waveform strip 1314

3 - : a section with a length of 3 mm of a standard T-shape 12 welded to the HE shape with two fillet welds of 6 mm, see Fig. 5. This connector evolved from the observation by Oguejiofor [2] that a large part of the bearing capacity of a perfobondstrip was the result of the direct bearing of the concrete at the front end of the (discontinuous) perfobondstrip. Therefore a T- Fig. 5: shape, which has a larger contact area than a single strip, and because of its shape will prevent vertical separation between HE-shape and concrete, seemed a good alternative. The main objective of this project is to get information about the behaviour of the connection between steel and concrete. Other kinds of failure also had to be excluded. This resulted in a, sometimes, vast amount of splitting/shear reinforcement, see Fig. 7, and welds of the perfobondstrip, oscillating perfobondstrip and purposely made stronger than the connector itself. 3. Experimental set-up Three modifications have been made to the standard push-out test as described in chapter 1 of EuroCode 4 [1]: Design assisted by testing : - The test-specimens have been placed on a sliding support made of a greased Teflon plate between two stainless steel plates. These support conditions are similar to those used by Mainstone and Menzies [3], and are considered to give conservative results compared to the fixed supports of EuroCode 4. - For some connector devices concrete panels of 15 mm thick are too thin, and plates of 2 mm thick have been used. In particular for all tests with the (oscillating) perfobondstrip and for the later series with the T- connector because of spalling problems with the first series. - For some tests the standard HE 26B section would not have sufficient capacity, and was replaced by a HE 24M section. Preliminary calculations for the capacity of the connector devices learned that the combination of (oscillating) perfobondstrip and high strength concrete could have a failure load of nearly 4 kn. A special closed testing frame with an ultimate capacity of 5 kn was developed for Fig. 6: Test set-up these tests, see Fig. 6. The longitudinal displacement between the concrete panels and the steel section was measured by LVDT s with a range of 2 mm. The transverse separation between steel section and concrete panels was measured by LVDT s with a range 1315

4 of 2, 3 and 5 mm respectively. The elongation at the outside of the concrete panels was measured by LVDT s with a range of 2 mm. All data has been collected at an interval of.1 mm longitudinal displacement. The specimens have been tested in a deformation controlled way until failure or the maximum possible deformation of approximately 45 mm. For the test specimens certain combinations of connector device and concrete grade are used. The concrete grades vary from an ordinary C3/37 to high strength concrete C7/85. Lightweight concrete is included in the series as normal strength LC3/37 and high strength LC62/75. Concrete was used with and without added steel fibres giving a total of 8 possible concrete grades/compositions. The mix designs used are presented in Table 1. This table also shows the number of tests executed for a certain combination. Table 1: Concrete mixes used [kg/m 3 ], averaged strengths [MPa], and number of tests Mix C3/37 C7/85 LC3/37 LC62/75 C3/37 C7/85 LC3/37 + fibres + fibres + fibres Portland cement/ Blast-furnace cement LC62/75 + fibres 32/ 238/237 32/ 238/237 32/ 238/237 32/ 238/237 Microsilica slurry Water Sand -4 mm Gravel 4-16 mm * * - - Lytag 4-8 mm Lytag mm Liapor F Dramix RC 8/ BN 6 BP 8 BN 6 BP Strength Average cube strength [MPa] Average splitting tensile strength [MPa] Number of tests Headed-studs Oscillating perfobondstrip Waveform strip * Broken gravel + 1% saturated BN Type BN BP Type BP Splitting or shear failure of the concrete panels had to be avoided, sometimes resulting in heavy reinforcement, see Fig. 7. All reinforcement is standard grade S5. The type of reinforcement provided is listed in Table

5 Table 2: Type of reinforcement per test-specimen Concrete grade C3/37 C7/85 LC3/37 LC62/75 C3/37 + fibres C7/85 LC3/37 + fibres + fibres Headed-studs Oscillating LC62/75 + fibres perfobondstrip Waveform strip Type: 1 2 x 5 φ1 BF and 2 x 4 stirrups φ1 BF per panel 2 2 x 5 φ16 BF and 2 x 4 stirrups φ1 BF per plate (for panel of 15 mm) 3 2 x 5 φ16 BF and 2 x 4 stirrups φ1 BF per plate (for panel of 2 mm) 4 2 x 5 φ2 BF and 2 x 4 stirrups φ1 BF per panel 5 2 x 5 φ25 BF and 4 x 1 mm anchor-plate and 1 hairpins and 2 x 4 stirrups φ1 BF per panel 6 2 x 5 φ25 BF and 2 x 1 mm anchor-plate and 2 x 4 stirrups φ1 BF per panel 7 2 x 5 φ12 BF and 2 x 4 stirrups φ1 BF per panel 8 2 x 5 φ25 BF and 4 x 1 mm anchor-plate and 1 hairpins and 2 x 4 stirrups φ1 BF per panel BF : Both Faces 4. Test results Fig. 7: Reinforcement perfobondstrip in high strength concrete The differences between the various connector-devices with respect to ultimate load, ultimate displacement and general behaviour are considerable. An overview of the maximum loads and the accompanying displacements for these loads is given in Table 3. Only 2 tests per test specimen have been carried out. The results of these tests can nevertheless be regarded as a good indication for the behaviour of the connector device since the differences found between these 2 tests, and especially for the pre-peak behaviour, are generally very small. The graphs in this section always show the minimum force at a certain displacement for the 2 test-specimens investigated. The discussion of the results will be subdivided into two parts: 1 A comparison of the behaviour of the connector devices for a particular concrete type. 2 A comparison of the characteristic strength and ductility according to EuroCode 4. Due to its rather disappointing behaviour the waveform-strip is only included in the graphs for C3/37, and is not discussed any further. 1317

6 Table 3: Maximum load [kn]/displacement at maximum load [mm] per test Concrete grade C3/37 + C7/85 + C3/37 C7/85 LC3/37 LC62/75 fibres fibres Headedstuds Perfobondfobondstrip Oscillating perfobondstrip Waveform strip T- connector 961/5.6 91/6.7 94/ / / / /2.4 51/ / / / / / / / / / / / /3. 177/ / /1.7 11/ / / / / / /5.4 9/ /3. 15/ / / / /5.4 LC3/37 LC62/ fibres fibres 891/ / / / / /8.9 31/ / / / / / / / / / / / Comparison of connector devices for a particular concrete type The comparisons of connector devices for a particular concrete type are shown in the Figs. 8 and 9 for all concrete types without fibres and for C3/37 with fibres. The dotted lines in these figures show the characteristic capacity obtained according to the testing procedure of EuroCode 4. The following conclusions can be drawn per concrete type: - The behaviour of the (oscillating) perfobondstrip is a bit disappointing for concrete C3/37 without fibres compared to, for instance, headed studs. This mainly has to do with the fast drop of the load capacity after the peak caused by splitting of the concrete panels in the plane of the bottom reinforcement (running through the holes in the perfobondstrip). It is not directly possible to conclude that the (oscillating) perfobondstrip should not be applied for C3/37; the splitting failure mode may have been induced by the (sliding) support conditions, what could be too conservative. The capacity of the perfobondstrip, and especially the oscillating-perfobondstrip, is larger than that of the headed studs. Both strips hardly deformed during testing, and could have been reused. An expected vertical splitting crack parallel to the longitudinal direction of the strip was not observed. Although this crack was observed for the it there seemed quite innocent compared to the spalling of the concrete cover of the bottom part of the during testing. This spalling however hardly effected the behaviour of the, which performed very well compared to headed studs. The equals the capacity of the perfobondstrip, but has a much larger ductility. The graph also reveals the rather disappointing behaviour of the waveform strip. - Concrete type C3/37 is the only mix where all connectors have been tested in combination with fibres (except the waveform connector. Fig. 8b. clearly reveals the effects which fibres have on the behaviour of and perfobondstrip, and especially the oscillating perfobondstrip; a significant greater ductility is obtained. 1318

7 - All connectors show a better ductility for LC3/37 compared to C3/37 without fibres, but it is especially remarkable to notice the difference in behaviour of the oscillating-perfobondstrip. This despite that LC3/37 is regarded to be a more brittle material than C3/37. The T- connector still behaves very well with respect to strength and ductility. - The graphs for LC62/75 reveal that there is a significant increase in strength and ductility of the (oscillating) perfobondstrip, and some increase for the, compared to LC3/37. They all behave much better than headed studs, and could very well be applied in combination with LC62/75. For the the concrete is no longer decisive, but the strength of the connector itself. - The graphs for C7/85 in Fig. 9 immediately reveal the possibilities of the (oscillating) perfobondstrip. They both have a much higher strength combined with a much better ductility. High strength concrete really could use the potential of these connectors. The strength of the is comparable to that of LC62/75. For all connectors the strength of the concrete is no longer decisive, but the force [kn] force [kn] force [kn] force [kn] Waveform strip C3/37 without fibres displacement [mm] C3/37 with fibres displacement [mm] LC3/37 without fibres displacement [mm] LC62/75 without fibres displacement [mm] Fig. 8: Comparison of connector devices for a particular concrete type (Dotted line = cap. acc. EC4) 1319

8 strength of the connector itself. 4.2 Strength and ductility according to EuroCode 4 The procedure for the determination of the characteristic resistance and slip capacity of connector devices based on pushout tests is given in section of ENV (Euro- Code 4-1) [1]. The characteristic resistance P Rk should be taken as the minimum failure load of 3 tests reduced by 1%. Results for P Rk for the connector devices investigated based on Table 3 can be found in Table 4. The slip capacity of a connector should be taken as the maximum slip measured at the characteristic load level, as shown in Fig. 1. The characteristic slip capacity δ uk should be taken as the minimum test value of δ u reduced by 1%. Conservatively not the maximum slip is measured for individual connector devices, but for the bottom limit of the curves of a particular connector device as shown in Figs 8 and 9. The calculated values for δ uk are shown in Table 4 and Fig. 11. Section 6.1.2(3)(b) of ENV states that connector devices with a characteristic slip of not less than 6 mm have force [kn] C7/85 without fibres displacement [mm] Fig.9: Comparison of connector devices for a particular concrete type Fig. 1: Determination of characteristic resistance and slip capacity connector according EC 4-1 the same deformation capacity as headed studs. of sufficient length in turn may be regarded to be ductile connectors. Examination of Table 4 learns that a lot of connectors, even headed studs, do not fulfil this requirement. It should however be noted that this ductility is only required when ideal plastic behaviour of the shear connection in the structure is considered. This is generally not the case for bridge structures where a number of shear connector devices discussed in this article could be well applied. Fig. 11 shows that there is an increase in capacity for all connector devices for higher cube strengths. This increase in cube strength is most beneficial for the (oscillating) perfobondstrip. It should however be remarked that the failure modes for the T- connector and the (oscillating) perfobondstrip are different for lower and higher concrete strengths; the connectors device itself becomes governing for the higher concrete strengths. The capacity of the could easily be increased by providing a 132

9 longer connector, what is less easy for the (oscillating) perfobondstrip. The increase in strength for headed studs looks like a contradiction to the general accepted idea that the steel strength of the studs is governing from concrete grade C3/37 and up, and that the concrete strength therefore should have no effect. Due to the increase in Youngs-modulus of the concrete, and hence the increase in the bedding stiffness of the stud, the failure mode of the studs changes from a combined tension and shear failure to a pure shear failure mode. As a result the tensile force in the stud will reduce, and leave more capacity for the shear mode of failure. F rk [kn] δ u [mm] Cube strength - Characteristic strength (EC 4) Cube strength [Mpa] Cube strength - Ductility (EC 4) Cube strength [Mpa] Fig. 11: Strength and ductility according to EC 4 Table 4: Strength and ductility according to EuroCode 4 Concrete grade C3/37 C7/85 LC3/37 LC62/75 C3/37 C7/85 LC3/37 LC62/75 +fibres +fibres +fibres +fibres Headed-studs P Rk [kn] (8) δ uk [mm] P Rk [kn] (2) δ uk [mm] perfobondstrip δ uk [mm] 1.9 Oscillating P Rk [kn] (2) δ uk [mm] 1.8 Waveform strip P Rk [kn] P Rk [kn] (2) δ uk [mm] (number of connectors per test-specimen) For headed-studs and the there is some decrease in ductility for higher concrete strengths, whilst there is some increase for the (oscillating) perfobondstrip. Despite 1321

10 the decrease in ductility of the this device still is the most ductile connector for higher concrete strengths. 5. Conclusions The following general conclusions can be drawn with respect to the connector devices and concrete mixes investigated: - A failure mechanism not known from literature was observed for the (oscillating) perfobondstrip in concrete C3/37, LC3/37 and LC75/8 without fibres. - The addition of steel fibres had a very beneficial effect on the behaviour of the connection in general. The only exception are the headed studs where the behaviour of the steel limited the strength. - The connection in high strength concrete, which was expected to be very brittle, behaved in a very ductile way. The observation that the concrete panels did not split, and that the steel connector device had to deliver most of the deformations, seems an explanation for this ductility. - The behaviour of the is very promising, but there may be questions about its behaviour under dynamic loads. It actually has a very small bearing area on the concrete. Subsequently the stresses in the concrete must be very high, and local crushing of the concrete may occur. This in particular could cause a degradation of the connection under a dynamic load. This has to be further investigated. - Lightweight aggregate concrete often seemed to behave in a little bit more ductile way when compared to normal weight concrete, especially in the post-peak stage. 6. References 1 EuroCode 4: ENV :1997, Design of composite steel and concrete structures Part 2: Composite bridges (1997), European Committee for Standardisation (CEN) 2 Oguejiofor E.C. and Hosain M.U. (1994). A parametric study of perfobond rib shear connectors. Can. J. Civ. Eng. 21, Mainstone R.J. and Menzies J.B. (1967). Shear connectors in steel-concrete composite beams for bridges 1: Static and fatigue tests on push-out specimens. Concrete 1:9,