Load Transfer in CFT Systems
Three primary mechanisms are responsible for load transfer in CFT systems: natural bond, shear connection, and direct bearing. However, because of deformation compatibility, the mechanisms typically do not act in conjunction and their strengths cannot be superimposed.
Natural bond strength is generally attributed to three primary mechanisms: adhesion, friction, and wedging (Parsley et al., 2000 and Johansson 2003). Adhesion, provided by the chemical bond between the concrete and steel, is a brittle mechanism and only active at most during the early stages of load. It may not be active at all depending on the relative amplitudes of radial enlargement of the steel tube caused by the wet concrete, shrinkage of the concrete, and the roughness of the steel tube (Roeder et al., 1999). Friction is the product of the roughness of the steel-concrete interface and the contact pressure existing at the interface. Wedging occurs as the motion of the concrete core is resisted by geometric irregularities in the steel tube.
Roeder et al. (1999) stated that the bond between steel and concrete depends on three factors, including radial enlargement of wet concrete due its pressure on the steel tube, roughness of the tube wall, and shrinkage of the concrete. They identified three states of bond depending on the relative magnitude of the aforementioned factors. In state A, the amount of radial enlargement due to concrete pressure is larger than the shrinkage of the concrete. State B has the condition that shrinkage of the concrete is greater than the summation of radial displacement and the amplitude of surface roughness, which indicates the loss of contact between the steel and concrete. State C is common in practical applications, and in this case the shrinkage of concrete is greater than radial displacement, while the difference between the two is smaller than the amplitude of the surface roughness. Roeder et al. (1999) also indicated that chemical adhesion enhanced the initial bond between the steel and concrete. Moreover, they found that microlocking between steel and concrete results from mechanical interlocking of concrete with surface irregularities as well as the friction between steel and concrete due to lateral pressure of concrete. In their push-out tests, Roeder et al. (1999) obtained an exponential bond stress distribution along the column length under low axial loads, and as slip starts to occur the bond stress distribution became more uniform. Roeder et al. (1999) also tested several push-out specimens under eccentric loading, determining that eccentric loading improved the bond strength up to approximately 2.5 times the bond strength from the concentric loading tests. Roeder et al. (1999) provided an equation for the bond strength of circular CFTs that is a function of the D/t ratio.
Experimental studies on bond behavior of CFT members have most frequently been conducted through the use of push-out tests. The boundary conditions of push-out tests induce a near constant bond stress at the ultimate limit state
Virdi and Dowling (1980) established a characteristic bond strength of 150-160 psi and concluded that surface preparation and the amount of compaction are the only significant parameters that will increase the amount of bond. Parameters such as concrete strength, length of the concrete/steel interface, the tube thickness, and the tube diameter had only negligible effects on the amount of bond. Shakir-Khalil and Zeghiche (1989) performed push-out bond tests on rectangular steel tubes filled with concrete and found that there was less bond than for reinforcement bars or even circular steel tubes. The effect of shrinkage and the relative flexibility of the rectangular tube walls reduced the bond strength. Bond has also been shown to be less in rectangular sections than circular sections due to the shrinkage of the concrete, which will have a greater effect on the less uniform rectangular section.
Morishita et al. (1979) and Tomii et al. (1980) conducted push-out tests on circular and square CFTs by loading the steel tube alone. They found that the concrete strength did not have any influence on the bond strength for square shapes, while in circular tubes, bond strength decreased when high strength concrete was utilized. They also determined that mean bond stress remained constant even if large values of slip occurred between steel and concrete. Morishita et al. (1979) specified bond strength values of 28.46 to 56.92 psi and 21.34 to 42.69 psi for the circular and square CFTs, respectively. Morishita and Tomii (1982) performed push-out tests on square CFTs under constant axial load and cyclic shearing force. It was found that when the shear forces acting on the columns increased, the magnitude of mean bond stress improved, and as the amount of slip increased, the bond stresses became constant. In addition, the concrete strength was again found to have little effect on the mean bond stress.
Despite the large number of push-out tests available in the literature, they provide limited information for bond strength and bond stiffness. The scatter of these two quantities obtained from the push-out tests is large (Hajjar et al. 1998). However, these tests provided insight to the shape of the load-slip behavior.
Effect of shear tab rotation
Push-out tests do not always represent well the conditions in typical composite CFT frames. In typical shear connections, load is applied to a CFT column through a shear tab. Moreover, connection regions are where the highest bond demand exists in a frame. Push-out tests where force is applied to the concrete core and resisted by shear tabs attached to the steel tube include these beneficial effects and thus provide the closest analogs to typical shear connections used in practice.
Therefore, connection tests are often more suitable to investigate bond transfer between steel and concrete. Dunberry et al. (1987) tested CFT columns framed by steel girders at the mid-height. They found that the rotation of girders in the lower part of the connection caused a pinching effect, increasing the bond strength and the rate of load transfer from the steel to the concrete. Dunberry et al. (1987) also determined that capping and grouting the end of CFT columns caused some portion of load transfer to be achieved in the upper part of the column rather than restricting the load transfer mechanism to the connection region. This caused more favorable load transfer from the steel to the concrete. Shakir-Khalil and Mahmoud (1995) performed simple beam-to-column connection tests similar to Dunberry et al. (1987). The load transfer between the steel and concrete was completed within a distance of D above the connection.