This Page Intentionally Left BlankThis Page Intentionally Left BlankThis Page Intentionally Left BlankThis Page Intentionally Left BlankThis Page Intentionally Left BlankThis Page Intentionally Left BlankThis Page Intentionally Left BlankThis Page Intentionally Left BlankThis Page Intentionally Left BlankThis Page Intentionally Left BlankThis Page Intentionally Left BlankThis Page Intentionally Left BlankThis Page Intentionally Left BlankThis Page Intentionally Left Blank CONCRETE-FILLED STEEL TUBES AS COUPLING BEAMS FOR RC SHEAR WALLS J.G. Teng 1, J.F. Chen 2 and Y.C. Lee 1 1 Department of Civil and Structural Engineering Hong Kong Polytechnic University, Hong Kong, China 2 Built Environment Research Unit, School of Engineering and the Built Environment, Wolverhampton University, Wulfruna Street, Wolverhampton WV 1 1SB UK ABSTRACT Coupling beams in a reinforced concrete coupled shear wall structure are generally designed to provide a ductile energy dissipating mechanism during seismic attacks. This paper explores the use of concrete-filled rectangular tubes (CFRTs) as coupling beams and describes an experimental investigation into this form of construction to study their load carrying capacity, ductility and energy absorption characteristics. Results from six tests on simplified CFRT coupling beam models subject to static and cyclic loads are presented. These results demonstrate that CFRT beams have good ductility and a good energy absorption capacity. They are therefore suitable as coupling beams for shear walls particularly if the effect of local buckling is minimised by the use of steel plates of an appropriate thickness. KEYWORDS Coupling beams, concrete-filled steel tubes, shear walls, tall buildings, seismic design, ductility. INTRODUCTION Reinforced concrete (RC) coupled shear walls are commonly found in high-rise buildings. For buildings subject to seismic attacks, properly designed coupled walls offer excellent ductility through inelastic deformations in the coupling beams, which can dissipate a great amount of seismic energy. It is thus essential that the coupling beams be designed to possess sufficient ductility. The traditional way of constructing a ductile RC coupling beam is to use a large amount of steel reinforcement, particularly diagonal reinforcement (e.g. Pauley and Binney, 1974, Park and Paulay, 1975). However, diagonal reinforcement is only effective for coupling beams with span-to-depth ratios less than two. For larger span-to-depth ratios, the inclination angle of diagonal bars becomes too small for them to contribute effectively to shear resistance (Shiu et al, 1978). However, deep coupling beams 391 392 J.G. Teng et al. are often not desirable because their depths may interfere with clear floor height. Furthermore, with the increased use of high strength concrete, it is more difficult to achieve ductility in RC beams as the section size reduces and the brittleness of the concrete increases. Therefore, the exploration of alternative coupling beam forms offering good ductility is worthwhile. As an alternative to RC coupling beams, Harries et al. (1993) and Shahrooz et al. (1993) studied the use of steel I-beams as coupling beams. As a structural material, steel is much stronger and much more ductile than concrete. However, steel beams may suffer from inelastic lateral buckling and local buckling which limit their ductility. Although local buckling may be prevented by the proper use of lateral stiffeners (Harris et al., 1993), such stiffening is labour intensive and may lead to uneconomic designs. More recently, steel coupling beams encased in normally reinforced concrete have been studied (Liang and Han, 1995; Wang and Sang, 1995; Gong et al., 1997). These studies show that the encasement of concrete leads to increases in stiffness and strength which should be properly considered in design and that the concrete is likely to spall during cyclic deformations. This paper explores the use of concrete-filled rectangular tubes (CFRTs) as coupling beams and describes an experimental investigation into this form of construction to study their load carrying capacity, ductility and energy absorption characteristics. Extensive recent research has been carried out on the behaviour of concrete filled steel tubes, particularly as columns (e.g. Ge and Usami, 1992; Shams and Saadeghvaziri, 1997; Uy, 1998). In such tubes, the concrete infill prevents the inward buckling of the tube wall while the steel tube confines the concrete and constrains it from spalling. The combination of steel and concrete in such a manner makes the best use of the properties of both materials and leads to excellent ductility. To the authors' best knowledge, CFRTs have not previously been used as coupling beams, although their use in buildings and other structures, particularly as columns, has been extensive. Apart from ductility considerations, CFRT beams are much simpler to construct than RC beams because both the placement of complicated reinforcement and temporary formwork are eliminated. Compared with steel coupling beams, CFRT beams are more economic due to significant savings in steel. ." ~hear ~lall ' L .A dShe~'.wall , ". 'I L/2 ~1 w I LIlllllllllll Shear orco Bending moment a) Prototype structure IIIIllllllLll Model structure Figure 1: Modelling of coupling beams SPECIMEN DESIGN AND PREPARATION Modelling of Coupling Beams During an earthquake, the coupling beams provide an important energy dissipation mechanism in a coupled wall structure through inelastic deformations. These beams are subject to large shear forces Concrete-Filled Steel Tubes as Coupling Beams for RC Shear Walls 393 and bending moments, with the effect of axial forces being small. The shear force and bending moment distributions in a coupling beam with the point of contraflexure at the mid-span are shown in Figure 1 a. These force distributions can be modelled by a cantilever beam under a point load at its free end (Figure l b). This cantilever beam system was thus used in the present study to simulate the behaviour of a coupling beam under seismic loading. The effect of embedment was not considered and the wall was assumed to provide a rigid support to the beam. In practical applications, a sufficient embedment length should be used to prevent premature failures in the embedment zones. An existing approach for designing the concrete embedment for steel coupling beams (Marcakis and Mitchell, 1980; Harries et al., 1993) can be used for designing the concrete embedment for CFRT coupling beams. Design of Specimens Eight cantilever beams were tested in this study, consisting of two control rectangular hollow section (RHS) tubes and six CFRTs. All tubes had a wall thickness of 2 mm, with a cross-sectional height of 200 mm and width of 150 mm. The variable for the CFRTs was the concrete strength, designed to cube strengths of 40, 60 or 90 MPa (referred to as Grade 40, Grade 60 and Grade 90 concrete respectively in the paper). The eight specimens were divided into two series, each consisting of one RHS tube and three CFRTs filled with concrete of different grades. The two series of specimens were tested under static loads and cyclic loads respectively. Preparation of Specimens The fabrication of the RHS tubes was by cold-bending and welding. Two channels were first made from steel sheets using a bending machine. Subsequently, the two channels, with their edges facing each other, were welded together to form a RHS tube with a welding seam at the mid-height of each web. Two types of steels with slightly different properties were used (Table 1). These properties were determined by tensile tests using samples from the same plates used for fabricating the RHS tubes. TABLE 1 SPECIMEN DETATILS Specimen Steel properties, MPa Concrete properties, MPa Yield stress Ultimate stress Young' s modulus Compressive Strength, 28th day Compressive strength, day of beam test Splitting tensile strength, 28 th days RHSs 290 441 194,000 N/A N/A N/A TG40s 290 441 194,000 45.3 45.2 3.03 TG60s 290 441 194,000 87.6 84.3 4.57 TG90s 290 441 194,000 92.5 94.1 4.68 RHSc 290 365 216,000 N/A N/A N/A TG40c 290 441 194,000 41.3 45.1 3.36 TG60c 290 441 194,000 87.6 90.62 4.57 TG90c 290 365 216,000 112.0 109.5 6.51 Test type Static Static Static Static Cyclic Cyclic Cyclic Cyclic The fabricated RHS tubes were then filled with fresh concrete. For each of the specimens, six 100x100x100 mm 3 concrete cubes and three concrete cylinders with a diameter of 100mm and a height of 200mm were cast to test their compressive and splitting tensile strengths. Measured concrete properties are shown in Table 1. The actual concrete strength for Grade 60 (Specimens TG60s and TG60c) was as high as that for Grade 90 (Specimens TG90s and TG90c) probably due to mixing problems. While this was undesirable, the specimens were still suitable for the present study and are still referred to using their intended concrete grades (ie TG60 and TG90) in this paper. 394 J.G. Teng et al. In order to prevent the concrete core from being pushed out when a CFRT specimen was loaded, two 6 mm thick steel plates were welded to the ends of each CFRT beam when the concrete age was 28 days. This simulated the antisymmetric condition at the point of contraflexure in a full coupling beam. EXPERIMENTAL SET-UP The experimental set-up for static loading tests is shown in Figure 2. Beam specimens were clamped between two large angle plates, which were in turn fixed on the floor by four high strength bolts. The embedment length of the beams was 440mm. Loads were applied at 460 mm from the fixed end. The span to depth ratio of the beams was 460/200=2.3 which was the smallest value possible because of restrictions of the pre-installed anchor plates on the strong floor. A hydraulic jack was fixed onto the floor to load the beam horizontally for convenience. Displacements at the loading position, the mid- span and near the fixed end were measured by electronic displacement transducers. Furthermore, a number of strain gauges were installed near the fixed end (Figure 2). Two displacement transducers were also used to measure the translation and rotation of the fixed end support. The effect of small support movements has been removed in the values of displacements presented in this paper. For cyclic loading tests, two hydraulic jacks were used. Because of this arrangement, the displacement transducer at the loading point was moved to the tip of the beam. The positions of other transducers were the same as in the static tests. The deflection at the loading position was inferred from the measured values at the tip in an approximate manner assuming either the beam deformed elastically or rigid-plastically with a plastic hinge at the fixed support. Details are given in Lee (1998). No strain measurement was undertaken in the cyclic tests. L I -HSS~ q -~C) I -~C) 'Dlsptacenent Tronsducers '='I 45 Degree ~ S• Rosett __~Str~In G~uge a) Plan b) Section A-A Figure 2: Experimental set-up for static loading test TEST PROCEDURE Static Loading Test In static loading tests, the specimens were monotonically loaded until failure. The strains and displacements were recorded at different load levels, from which load-deflection curves were plotted. These curves were used to determine the values of the 'yield load' Py and the corresponding deflection at the loading position dy (Lee, 1998). Based on observations during the experiments, the 'yield load' was defined as the load when local buckling of the compression flange occurred and corresponds to a strong change in slope of the load deflection curve. This yield load Py and the deflection dy were later used to control the load/displacement levels in the cyclic loading tests. Concrete-Filled Steel Tubes as Coupling Beams for RC Shear Walls 395 Cyclic Loading Test The loading sequence used in the cyclic loading tests is shown in Figure 3. Load control was used before the yield load was reached. Two cycles of reversed cyclic loading were carried out at a load level of P=0.8Py. Three additional cycles were then carried out at P = Py. Thereafter, deflection control at multiples of dy was used. Three complete cycles were carried out at each selected value of deflection until the specimen failed. The loads or displacements were carefully controlled during cyclic tests, nevertheless, some small deviations from the intended values still existed. Displacements were monitored and recorded throughout the test. Figure 3: Loading history for cyclic tests Figure 4: Load-deflection curves under static loading STATIC TEST RESULTS Figure 4 shows the load-deflection curves of the loading point for all four static test specimens. The rapidly descending load-deflection curve after buckling of the RHS tube indicates that its load carrying capacity was reduced quickly, exhibiting very limited ductility. The ultimate strengths of the concrete filled tubes are almost triple that of the corresponding RHS tube. The extended plateaux in the load- deflection curves alter yielding show that CFRT beams are very ductile. These effects of the concrete infill are well known. The ductile behaviour of the CFRT beams was terminated by tensile rupture of the tension flange which occurred significantly earlier in Specimen TG90s than in the other two CFRT beams. Specimens TG40s and TG60s showed similar ductility, though they were filled with concrete of rather different strengths. The effect of the concrete strength on ductility is thus believed to be small. Although local buckling of the steel tube was observed in all tests, the final failure modes were different for RHS and CFRT specimens (Figure 5). The local buckling of the compression flange near the fixed end occurred at a load of approximately 40 kN, leading to immediate collapse of Specimen RHSs. Shear buckling occurred on both webs at the same load. No crack was found on the tensile flange of Specimen RHSs. For the three CFRT beams, outward local buckling was observed on the compression flanges at a load of approximately 80kN. Shear buckling occurred later on the webs at about 100kN. Clearly, the concrete infill constrained the plate to buckle only away from it, which led to a higher buckling strength, as has been shown by many authors (eg Wright, 1993; Smith et al., 1999). Strain readings showed that the tensile flange had yielded and the compression flange was close to yielding when local buckling occurred. Fracture cracks were found on the tension flanges of CFRT specimens at final failure, indicating the full use of the steel strength. The final failure of CFRT members was by rupture of steel of the tension flange and is referred to as a flexural failure. 396 J.G. Teng et al. Figure 5: Static loading test: buckling of the compression flange Table 2 shows the experimental ultimate loads for all the static test specimens. The calculated ultimate flexural failure loads according to the approach in BS 8110 (1985) for reinforced concrete beams and using the ultimate stress of steel are also listed for comparison. Clearly, experimental observations are in good agreement with theoretical predictions for CFRT specimens, with discrepancies within 3%. These calculations did not consider local buckling effects, so the calculated ultimate flexural failure load of Specimen RHSs of 93.78kN is more than double the value actually achieved during the test (42.51kN). The chief contribution of the concrete infill is thus to provide constraint to the steel tube. The ultimate strength of CFRT beams increases with the concrete strength. However, this increase is small. Table 2 shows that the concrete strength for TG60s and TG90s is almost twice that for TG40s, but the increase in the experimental ultimate load is only less than 3% while the theoretical increase is less than 6%. TABLE 2 STATIC ULTIMATE LOADS Specimen fcu, MPa Test ultimate load, Predicted ultimate Test / Prediction kN load, kN RHSs N/A 42.51 93.78 0.453 TG40s 45.2 117.14 114.2 1.026 TG60s 84.3 120.24 119.55 1.006 TG90s 94.1 119.57 120.38 0.993 CYCLIC TEST RESULTS Test Observations and Failure Modes Local buckling was observed on both flanges of all the cyclic test specimens. During load reversal, a buckled flange was straightened again under tension. The compression-tension cyclic stresses caused degradation in both steel and concrete, so that the maximum load reached in a cyclic test is considerably lower than that in the corresponding static test. For Specimen RHSc, local buckling was observed in both flanges. No crack developed in the flanges, indicating that the steel tensile strength was not fully utilised. By contrast, cracks developed in both flanges of TG40c and TG60c, and in one of the flanges of TG90c at final failure. Figure 6 shows one Concrete-Filled Steel Tubes as Coupling Beams for RC Shear Walls 397 of the flanges for each of the three cyclic test specimens after final failure. All the CFRT specimens failed after 14 to 15 loading cycles. Figure 6: Failure mode under cyclic loading Hysteretic Responses The hysteretic load-deflection responses of the loading position from all four cyclic tests are shown in Figure 7. The load carrying capacity of Specimen RHSc (Figure 7a) was quickly reduced from about 40kN in the first few cycles, to less than 20kN at the 8 th cycle and to less than 10kN at the 14 th cycle, confirming the lack of ductility as observed in the static loading test. Because the areas surrounded by the hysteresis loops represent the amount of energy absorbed by the test specimen, the energy absorption capacity of RHS tubes is thus very limited and reduces quickly under large cyclic deformations. As observed in the static tests, the ultimate strength of CFRT beams is significantly higher than their hollow counterparts. While the differences in the load carrying capacity in the plastic range are not large between the three CFRT cyclic test specimens, it is worth noting that TG90c, which had the highest concrete strength (Table 1), showed the lowest load carrying capacity. Compared with the results from the static loading tests, the maximum load carrying capacities of CFRT beams under cyclic loading are about 20-30% lower, with the difference between the two TG40 specimens being the smallest and that between the two TG90 specimens the largest. This indicates that a CFRT beam with a lower strength concrete behaves better than one filled with concrete of a higher strength. The CFRT beams exhibited strength and stiffness degradations under reversed cyclic loading and pinching is seen for all of them (Figure 7). The main reason is believed to be the degradation of concrete in strength and stiffness when subject to reversed cyclic loading which leads to shear cracks in both directions. Slipping between the steel tube and the concrete may also have been a significant factor. The slipping behaviour may be improved by using shear connectors such as those used by Shakir-Khalil et al. (1993). Overall, the hysteretic responses of these beams are good and are better than normal reinforced concrete beams, but are not as good as deep RC beams with proper diagonal reinforcement (Park and 398 J.G. Teng et al. Paulay, 1975). Significant improvements to the cyclic behaviour of these beams should be achievable by using thicker steel plates so that the effect of local buckling is minimised. Figure 7: Hysteretic load-deflection responses at loading position CONCLUSIONS This paper has explored the use of concrete filled steel tubes as coupling beams for reinforced concrete coupled shear wall structures. Six concrete filled rectangular steel tubes and two rectangular hollow steel tubes have been tested under static and cyclic loadings. The mutual constraints of the steel tube and the concrete infill lead to higher strength and good ductility. The strength and ductility of these beams are insensitive to concrete strength, but cyclic degradation seems to increase with concrete strength. The use of high strength concrete thus seems to be undesirable. The hysteretic responses of these beams under cyclic loads show that they have a good energy absorption capacity. Therefore, these beams are suitable as coupling beams, particularly if local buckling is minimised by using relatively thick steel plates and slipping between the steel and concrete is reduced using some form of shear connectors. Further research is required to better understand this form of coupling beams. Concrete-Filled Steel Tubes as Coupling Beams for RC Shear Walls ACKNOLWEDGEMENTS The authors are grateful to Dr. Y.L. Wong for helpful discussions on the subject. 399 REFERENCES BS 8110 (1985). Structural Use of Concrete. British Standards Institution, London. Gong B., Shahrooz B.M. and Gillum A.J. (1997). Seismic Behaviour and Design Of Composite Coupling Beams. Proc. of the Engineering Foundation Conference 1997, ASCE, New York, NY, USA, 258-271. Ge, H.B. and Usami, T. (1992). Strength of Concrete-Filled Thin-Walled Steel Box Columns: Experiment. Journal of Structural Engineering, ASCE, 118:11, 3036-3051. Harries K.A., Mitchell D., Cool W.D. and Redwood R.G. (1993). Seismic Response of Steel Beams Coupling Concrete Walls. Journal of Structural Engineering, ASCE, 119:12, 3611-3629. Lee Y.C. (1998). Concrete Filled Steel Tubes as Coupling Beams for Concrete Shear Walls, BEng Dissertation, Dept of Civil & Structural Engineering, The Hong Kong Polytechnic University, Hong Kong, China. Liang, Q. and Han, X. (1995). The Behaviour of Stiffening Beams and Lintel Beams under Cyclic Loading. Journal of South China University of Technology (Natural Science), 23:1, 26-33. Marcakis, K. and Mitchell, D. (1980). Precast Concrete Connections with Embedded Steel Members", PCI Journal, 25:4, 88-116. Park, R. and Paulay, T. (1975). Reinforced Concrete Structures, John Wiley and Sons, New York, N.Y. Paulay T. and Binney J.R. (1974). Diagonally Reinforced Coupling Beams of Shear Walls. Shear in Reinforced Concrete: Publication No. 42, ACI, Detroit, Mich., 579-598. Shahrooz B.M., Remmetter M.E. and Qin F. (1993). Seismic Design and Performance of Composite Coupled Walls. Journal of Structural Engineering, ASCE, 119:11,3291-3309. Shakir-Khalill, H. and Hassan, N.K.A. (1993) Push Out Resistance of Concrete-Filled Tubes. Tubular structures VI, (ed by Grundy, Holgate & Wong), Balkema, Rotterdam. Shams, M. and Saadeghvaziri, M.A. (1997). State of the Art of Concrete-Filled Steel Tubular Columns. A CI Structural Journal, 94:5, 558-571. Shiu K.N., Barney G.B., Fiorato A.E. and Corley W.G. (1978) Reversed Load Tests of Reinforced Concrete Coupling Beams. Proc., Central American Conference on Earthquake Engineering, 239- 249. Smith, S.T., Bradford, M.A. and Oehlers, D.J. (1999). Elastic Buckling of Unilaterally Constrained Rectangular Plates In Pure Shear. Engineering Structures, 21,443-453. Uy, B. (1998). Concrete Filled Fabricated Steel Box Columns for Multistorey Buildings: Behaviour and Design. Progress in Structural Engineering and materials, 1:2, 150-158. Wang, Z. and Sang, W. (1995). Beating Behaviour and Calculation Method of Steel Reinforced Concrete Coupling Beams. Journal of South China University of Technology (Natural Science), 23:1, 35-43. Wright, H. (1993). Buckling of Plates in Contact with a Rigid Medium. The Structural Engineer, 71:2, 209-215. . Engineering, 23 9- 249. Smith, S.T., Bradford, M.A. and Oehlers, D.J. (1999). Elastic Buckling of Unilaterally Constrained Rectangular Plates In Pure Shear. Engineering Structures, 21,44 3-4 53 are suitable as coupling beams, particularly if local buckling is minimised by using relatively thick steel plates and slipping between the steel and concrete is reduced using some form of shear. the RHS tubes was by cold-bending and welding. Two channels were first made from steel sheets using a bending machine. Subsequently, the two channels, with their edges facing each other, were