460 D. Lam e t al. construction depth. These economies have largely accounted for the dominance of composite steel frame construction in the commercial building sector in the UK in recent years. Composite construction of steel frames with profiled steel decking to support floor slabs is now common in multi- storey construction, but the use of precast prestressed concrete hollow core units (hcu' s) in conjunction with the steel frame to provide composite action is relatively new. Figure 1 shows the composite beam with precast hollow core slabs. Figure 1: Composite beam with precast hollow core slabs Hcu's are now the most widely used type of precast floor in Europe; annual production is about 20 million m 2, representing 40 to 60 per cent of the precast flooring market. This success is largely due to the highly efficient design and production methods, choice of unit depth and structural efficiency. The design of dry cast hcu originated in the United States in the late 1940s following the development of the high strength strand that could be reliably pre-tensioned over distances of 100m to 150m. This coincided with advancements in zero slump (hence the term 'dry') concrete production, which inevitably led to factory made hcu. Units have longitudinal voids and are produced on a long prestressing bed either by slip form or extrusion and are then saw cut to length. The degree of prestress and the depth of unit are the two main design parameters. The depth ranges from 150 to 400mm, with the performance limited to a maximum span / depth ratio of around 50, although 35 is more usual for normal office loading conditions. Although hcu's are widely used in all types of multi-storey buildings and account for approximately 50% of all floors used in steel framed buildings, the steel beam is normally designed in bending in isolation from the hcu' s slab and no composite beam action is considered in design with the hcu' s. The main reason for this is the uncertainty over the ability of the hcu's to satisfactorily transfer the shear and compressive forces. Whilst the potential to generate worthwhile composite action is not in doubt, little research into this problem has previously been undertaken. Recent work by the authors (1) is aimed at rectifying this by carrying out a systematic study into the behaviour of the composite beam in bending. To study the flexural behaviour of the hcu slabs and steel beam composite construction, the major issues that were addressed were: (a) the compression behaviour of the hcu slabs, and (b) the transfer of the horizontal shear forces between the steel beam and the concrete slab. To achieve this, full scale bending tests were supplemented by: (a) horizontal eccentric compression tests and (b) horizontal push-off tests, as shown in Figure 2. In this study, 3 no. of full scale 6.0m long composite beam tests were supplemented by 12 shear stud push-off tests and 5 no. composite slab compression tests. In addition to the experimental work described, analytical studies using the finite element technique were Steel- Concrete Composite Construction with Concrete Hollow Core Floor 461 employed to carry out parametric studies. This paper concentrates on presenting the full scale bending tests and the numerical simulation. Suitable references are also made to the supporting scientific work. Figure 2: Simplification of testing regime for [a] full scale bending test [b] isolation eccentric compression slab tests [c] isolated push-off tests TEST SET UP Three full scale bending tests comprised a 356 x 171 x 51 serial size $275 UB loaded in 4-point bending over a 5.7 m simply supported span as shown in Figure 3, with 150 mm deep x 1200 mm wide hcu's connected through 125 mm high x 19 mm diameter 'TRW Nelson' headed studs at 150 mm spacing(s) along the full length of the beam, giving 11 studs between the support and load positions. A 6.0m nominal length universal beam, with a 150ram thick hcu will be capable to carry a general office floor of 6.0m x 16.0m space free from columns. The characteristic cube strength for the hcu's is taken as 50 N/mm 2. The specimens were simply supported over a span of 5.7 metres and loaded by two point loads spaced symmetrically at 1.5 metres from each end support. All three specimens were similarly constructed, with the differences being the transverse reinforcement and insitu joint. Web stiffeners were used to eliminate local failure due to web buckling or flange yielding at the loading position. Figure 3: Plan view of the beam test 462 D. Lam et al. The slabs were placed directly on to the UB with a minimum bearing of 50 mm. The gap between the ends of the hcu' s was carefully monitored during placing to ensure a 65 mm gap width was maintained throughout. The tops of four cores per hcu, i.e. 2 nd, 4 th, 8 th and 10 th core, were left open for a length of 500mm to allow the placing of transverse reinforcement, giving an average bar spacing of 300 mm. Figure 4 shows the specimen before the insitu infill was cast. Figure 4: Test specimen before casting of the insitu infill Following the horizontal compression tests ~2) and push-off tests ~3), it was decided that transverse reinforcement ofT8 and T16 bars should be used for the full scale bending tests. T16 bars were used in test CB 1 to prevent tensile splitting and to confine the concrete slab from splitting failure, while T8 bars were used in test CB2 to allow tensile splitting to take place in a controlled manner. Insitu concrete with the design cube strength of 25 N/mm 2 was placed into the longitudinal and transverse joints and opened cores and compacted using a 25mm diameter vibrating poker to form the composite slab. In addition, specimen CB3 with debonded joints between the insitu and precast concrete was tested to observe the effect of a debonded insitu joint due to shrinkage. Two sheets of polythene were cast between the insitu concrete infill and the hcu to ensure a proper separation between the insitu infill and hcu, so that bonding and aggregate interlocking between the insitu infill and hcu could not be achieved, see Figure 5. Transverse reinforcement of T8 bars was chosen for this test; identical to the arrangement of Test CB2. Figure 5: Composite beam with debonded insitu joint Steel- Concrete Composite Construction with Concrete Hollow Core Floor 463 TEST RESULTS The results of the bending tests are given in Table 1 and Figure 6, where the increases in moment capacity and flexural stiffness of the composite beam compared to the bare steel UB are apparent. The elastic neutral axis of the composite section normally lies close to the interface between the steel and the concrete. As the bending moment increased, the bottom flange of the steel beam yielded and the neutral axis moved towards the compression zone, causing tensile cracking at the underside of the slab. When the stress at the outer surface of the concrete slab reached approximately 0.67fcu, spoiling of the concrete began and the ultimate strength of the section was then fully mobilized. As the curvature of the section is further increased, the load carried remains approximately constant and crushing of the slab occurs. Failure of the shear connectors may also occur, which would reduce the load carrying capacity of the composite section. No slippage between the slab and the end of the UB occurred for loads in the working load range. However, slip does have a considerable influence on the development of the ultimate moment capacity. TABLE 1 BEAM TEST RESULTS Test Reference Tie-steel area Insitu cube Max. test moment, ratio (%) strength (N/mm 2) MR (kNm) Mid-span Ratio Initial deflection of flexural at MR MR / stiffness, (mm) MR(steel) Ki (kNm/mm) CB1-T16 0.45 32.0 496 32 CB2-T8 0.11 25.5 470 35 CB3-T8* 0.11 26.5 345 27 356•215 UB - - 245 51 *included polythene at insitu-precast interface. 2.02 1.92 1.42 1.00 End slip at max. test moment (mm) 25.5 0.4 25.4 2.6 18.9 5.9 7.7 Note: Span/360 = 15.8ram(Deflection limit, BS5950.) Figure 6" Applied moment vs. vertical mid-span deflection curves of bending tests 464 D. Lam et al. The sudden reduction in strength in tests CB 1 & 2 was due to the fracture of the shear studs at one end of the beam. The span / deflection ratio when this occurred was about 165: 1, i.e. much larger than the allowable limit of 360:1 used in limit state design. Failure of test CB3 was due to concrete failure around the shear studs. In test CB 1, the first crack was observed at an applied moment of 342 kNm. This moment of 342 kNm is about 0.69 times the ultimate strength of test CB 1 and may conveniently be taken as the working load. This caused the neutral axis to move towards the compression zone, which in turn resulted in tension and cracking in the soffit of the precast slab. As the load was further increased, yielding of the steel section and cracking in the underside of the hcu extended over the full length of the slab, with a gradual reduction in stiffness. At the applied moment reached 490 kNm, i.e. twice the ultimate capacity of the bare steel beam, the sudden fracture of several shear studs precipitated a rapid reduction of load. No yielding or bond failure was not detected in the T16 transverse reinforcement - stresses were less than 20% of the yield stress at failure. In test CB2, the deformation was linear up to 245 kNm. Reduction in stiffness continued with yielding in the steel section and extended cracking in the hcu. A maximum load plateau was reached at 470 kNm with continuous deflection. Yielding of the transverse reinforcement occurred as the load reached the maximum. Tensile splitting occurred in the top surface of the concrete slab due to yielding of the transverse bars, causing concrete failure around the shear studs and a gradual reduction in load carrying capacity. There was fracture of some of the shear studs and failure at the interface by the crushing of concrete around the headed studs. The main difference caused by the introduction of the pre-cracked joint in test CB3 was the position of the neutral axis from the start of the test, which was located 20 mm below the steel concrete interface, indicating a reduced effective breadth of concrete slab caused by the pre-cracked joint. Although deformation remained linear up to 150 kNm, the position of the neutral axis moved from 20 mm to 58 mm below the steel concrete interface which suggested further reduction of the effective concrete section. At the ultimate moment of 345 kNm, i.e. some 35 % less than in the previous tests, the transverse reinforcement was fully yielded, leading to further tensile splitting of the slab. The stiffness of the pre-cracked specimen CB3 was approximately 0.71 of that in the former tests, indicating a reduced effective breadth when the interface bond is destroyed. A similar result was found in the eccentric compression tests (z), where the resistance of the pre-cracked specimen was 0.72 times that in the normally bonded specimens, also indicating a reduced concrete section. The ultimate moment in Test CB3 was, by chance, equal to the final post-fracture resistance in tests CB 1 & 2, indicating that in spite of the different modes of failure equilibrium is reached at the same level. NUMERICAL SIMULATION A numerical model based on the finite element method has been developed using ABAQUS (4). A two- dimensional model of the composite steel-concrete beam is shown in Figure 7. The model is set up to the same dimensions as the full scale bending test specimens. Although a 2-D model has its limitations when dealing with a 3-D structure, (the 2-D model used precluded the 3 rd dimensional effect where certain failure mechanisms might be critical), it is extremely useful when the modelling is admissible on account of economy (computational time, input/output), ready visualization and the relative ease with which parametric studies may be conducted. Three types of elements were used: 4-node plane stress elements to model the steel beam, 8-node concrete elements for the concrete slab and spring element for the shear studs. The shear studs were modelled as non-linear springs using the actual load - slip behaviour of the shear studs obtained from the experimental load vs. slip curve of the push-off tests (3). Each node of the steel element is connected to the node of the concrete element at the interface Steel- Concrete Composite Construction with Concrete Hollow Core Floor 465 by the spring element, i.e. at 150c/c. The test parameters, including the material properties used are identical to the ones used for the full scale bending tests. Figure 7: Finite element mesh of composite beam model Figure 8: Moment-deflection curves for beam tests and FE analyses The results of the FE analyses are shown in Figure 8. These show close agreement in terms of failure moment and moment-deflection characteristics for the three composite beams analysed herein. The percentage differences between the test and FE results for CB 1, CB2 and CB3 were 1%, 2% and 7% respectively. The FE analysis accounted for all major material features and was able to model concrete cracking and crushing as well as shear stud failure, although the post failure conditions could not be followed. The results showed that the 2-D model is suitable for the analysis and can be used to carry out parametric studies (5) on composite beams. 466 CONCLUSIONS D. Lam e t al. Based on the authors' own extensive experimental and numerical study, it has been shown that precast slabs may be used compositely with steel beams in order to increase both flexural strength and stiffness at virtually no extra cost, except for the headed shear studs. For typical geometry and serial sizes, the composite beams were found to be twice as strong and three times as stiff as the equivalent isolated steel section. The failure mode was ductile, and may be controlled by the correct use of small quantities of tie steel and insitu infill concrete placed between the precast units. A simple 2-D numerical model has been developed and can be effectively predicted moment and deflection of the composite beam with precast hollow core slabs. REFERENCES 1. Lam, D., 'Composite Steel Beams Using Precast Concrete Hollow Core Floor Slabs', Ph.D. Thesis, School of Civil Engineering, University of Nottingham, March, 1998. 2. Lam, D., Elliott, K. S. and Nethercot, D. A., 'Experiments on Composite Steel Beams with Precast Concrete Hollow Core Floor Slabs', submitted for publication. 3. Lam, D., Elliott, K. S. and Nethercot, D. A., 'Push-off tests on shear studs with hollow-cored floor slabs', The Structural Engineer, Vol. 76, No. 9, 1998, pp 167-174. 4. ABAQUS user manual, Version 5.3.1 (1994), Hibbitt, Karlsson & Sorensen, inc., 1080 Main Street, Pawtucket, RI 02860-4847, USA. 5. Lam, D., Elliott, K. S. and Nethercot, D. A., 'Parametric Study on Composite Steel Beams with Precast Concrete Hollow Core Floor Slabs', submitted for publication. ACKNOWLEDGEMENTS The authors acknowledge the support provided by the EPSRC, Bison Floors Ltd., UK and the skilled assistance provided by the technical staff in the Civil Engineering Laboratory at Nottingham University. Testing and Numerical Modelling of Bi-Steel Plate Subject to Push-Out Loading Simon K. Clubley Robert Y. Xiao Department of Civil & Environmental Engineering University of Southampton, UK ABSTRACT Bi-Steel panels are a newly patented composite construction system developed by British Steel Plc. They comprise of steel plates permanently coupled by a matrix of transverse friction welded rods. The shear strength and deformation capacity of the Bi-Steel unit subject to push out load is discussed in this paper. Numerical modelling by the use of finite element analysis has been conducted on Bi-Steel plates with and without in-filled concrete. The results of non-linear analysis are compared with experimental data. Both material and geometrical non-linearity were considered in the computing analysis. A design model has been suggested for deformation calculation due to shear action. 1. INTRODUCTION Performance of a composite steel and concrete structure is dependent upon the efficient interaction and effective transfer of shear between these two materials. Traditional shear connector design includes provision of adequate resistance to section uplift in addition to longitudinal slip. There are a number of types of shear connectors being used in design such as welded shear studs and mechanical fixed shear connectors. It is important to note that ultimate strength design of shear connectors assumes individual studs has sufficient ductility to redistribute load over the array so that all fail as a group under shear actions. In the absence of heavy concentrated loads, connectors are spaced uniformly between supports and sections of maximum bending moment. Bi-Steel adopts the high-speed friction weld process to attach shear rods to both steel panels as shown in Figure 1. The design philosophy of Bi-Steel is systems modular based. It is envisaged that minimal work for shear connections should be carried out on site, with prefabrication and preparation essentially carried out in workshops. Concrete will be poured once panels are delivered and positioned on site. In situ panels may be connected together by either bolting or welding. Large hydrostatic pressures during pumping can be sustained due to the density of shear connectors. This system has extremely strong potential to be used for many different types of structure. In collaboration with British Steel an extensive laboratory test programme was undertaken at Southampton to establish the ultimate shear strength of Bi-Steel specimens subject to push out load. This will be compared with traditional shear connector push-out testing. Research is available by Moy, Xiao & Lillestone (1), Oehlers & Sved (2), Kalfas, Pavlidis & Galoussis (3), Uy & Bradford (4) & (5) and Schuurman & Stark (6). The performance of shear studs in high and normal strength concrete was examined in the work of An & Cederwall (7). Prediction of expected shear stud strength with associated design formulae will be proposed as for the standard composite construction suggested by Oehlers & Johnson (8). Research conducted into double skin composite panels using profiled sheeting in Hossain & Wright (9) by numerical modelling and laboratory test data will also be examined. 467 468 S.K. Clubley and R.Y. Xiao Matrix voids and material imperfections account for the small difference in repeatable trials of push-out tests. Generally, scatter is accounted for by assuming the lowest recorded or derived failure load. Mechanical shear connectors only exhibit very limited shear/slip capacity when used in composite construction. Analysis in Oehlers & Sved (2) and Kemp & Trinchero (10) suggests development of yield at failure loads disappears as flexural strength reduces. Figure 1: Pre-fabricated Bi-Steel section. This paper presents the application of the finite element analysis program Ansys and associated theory to model experimental behaviour of Bi-Steel panels including a wide range of variable geometric specifications. Subsequent evaluation will suggest theory not readily available from data collected in a laboratory test programme. 2. TEST PROGRAMME The initial laboratory test programme commissioned by British Steel consisted of fifteen specimens. Details of all tests can be found in the confidential report to British Steel. Due to the nature of the contract, not all results of testing will be released here. There are a further twenty large specimens being planned for testing. They will be fully published separately in the future. Primary objectives of the test programme were to examine the strength and stiffness of shear connector studs when the concrete is subject to a shear action relative to the steel plates. The design of test specimens will be briefly introduced here. Plate spacing within the Bi-Steel unit was kept constant at 200mm during which five sizes of plate thickness were investigated. Shear connector size remained constant at 25mm diameter and was arranged in a regular grid matrix of 200mm in both directions. Before commencing the test programme representative steel coupons and concrete cubes were tested from each unit to establish key material properties. Prior to concrete casting, two weldable strain gauges were spot welded to each stud. Spot welding of strain gauges is a new technique for attachment to steel rods. To simulate constraint provided by a larger panel, six 16mm diameter threaded steel studs were located in the mould and the concrete was cast. The moulds were stripped after one day and together with the cubes both left in water to cure. The 100mm cubes were removed and tested at intervals to evaluate if the concrete had achieved the required strength. Following Bi-Steel unit removal from the water a drying period of 24 hours elapsed before 50mm Testing and Modellling of Bi-Steel Plate Subject to Push-Out Loading 469 square nuts and washers were placed on the studs and tightened. Protruding wires from the strain gauges were connected to a data logger. In addition, nine displacement transducers were placed around the unit to measure steel/concrete slip and displacement of the steel panels at the end of the studs. Before commencing specimen test each potentiometer was zeroed and calibrated in conjunction with the 1000KN capacity jack and spot-welded strain gauges. These gauges are pre- welded on shear connector rods and post welded on steel plates. A single push out load was applied by a hydraulic jack as shown in Figure 2 of the test set up. Figure 2: Test set-up for the specimen. Table 1 gives some typical test results for different thickness Bi-Steel plates. The load-strain relationships of steel plates are shown in Figure 3. Figure 3: Load-strain relationship for varying plate thickness. . staff in the Civil Engineering Laboratory at Nottingham University. Testing and Numerical Modelling of Bi -Steel Plate Subject to Push-Out Loading Simon K. Clubley Robert Y. Xiao Department. widely used in all types of multi-storey buildings and account for approximately 50% of all floors used in steel framed buildings, the steel beam is normally designed in bending in isolation. cast between the insitu concrete infill and the hcu to ensure a proper separation between the insitu infill and hcu, so that bonding and aggregate interlocking between the insitu infill and hcu