Roeder, Cameron, and Brown 1999
The load transfer mechanism between steel and concrete in circular CFT columns was examined. The authors presented bond strength and slip characteristics observed in experimental studies from the available literature. Moreover, they processed the experimental results and proposed equations for bond strength.
Experimental Study, Results, and Discussions
Static and dynamic analysis of CFT systems in past studies from the literature showed that bond stress demand was highest at the connections and foundation supports. In addition, braced frames were found to have more bond stress demand compared to unbraced frames. This was attributed to high vertical load coming from the braces at the connections.
Three states of bond (A, B, C) were presented depending on radial displacement of concrete, amount of shrinkage and amplitude of irregularity on steel tube surface. The radial displacement of concrete could be in the form of expansion due to lateral concrete pressure or in the form of contraction due to shrinkage. In state A, both cohesive chemical bond and mechanical resistance was effective for bond strength. However, concrete pressure should be about 174 psi to reach state A in typical applications. The steel and concrete was separated from each other and mechanical resistance was poor in state B. The most common case for CFT applications in the U.S. was state C. The chemical bond was not significant and the mechanical resistance was decreasing rapidly. The behavior of state C was changing with D/t ratio and shrinkage. The mechanical resistance was low for large D/t ratios due to reduced radial stiffness. However, the effect of concrete shrinkage was not clear particularly for long columns.
In the available literature, bond tests were mostly conducted on specimens with D/t ratios between 15 and 35. The maximum diameter was on the order of 12 in.. According to the results, the average bond strength of rectangular CFTs was about 70% smaller than that of circular CFTs. The concrete strength did not have a significant effect on bond strength. Due to large scatter of the data, it was not possible to determine a trend of bond stress depending on diameter and D/t ratio.
Two series of experiments were conducted to examine bond strength of circular CFT columns. The specimens were tested under axial load, which was applied to concrete alone. The specimens had an air gap of 2.31 in. at the bottom and rested on the steel tube. The axial load was either concentrically or eccentrically applied. The tests were monotonic, except that one specimen was loaded cyclically. The main variables of the tests were the diameter of the concrete core, the steel tube thickness, and the shrinkage of concrete. The diameter of steel tubes varied from 9.84 to 25.6 in.. The range for the D/t ratio was 20 to 110. Specimens in test series I had concrete with moderate shrinkage, while the specimens in test series II had concrete with low shrinkage. The compressive strength of the concrete was 7.40 ksi. The mechanical properties of the steel tubes was not provided as the applied loads were not high enough to yield the steel tube.
After achieving the ultimate load, the slip resistance was found to decrease with the increase in slip. This trend was typical among the specimens of test series II. In the axial load versus slip curves of the specimens in test series II, a point at which slip started to increase sharply was identified. At that point, the initial bond between steel and concrete was broken and friction started to dominate the bond resistance. This occurred at load levels of 40 to 80% of the ultimate capacity of the column. Initial breakage of the concrete-steel bond was not evident among the specimens of test series I. The bond stress potentials of the specimens were estimated using their acoustic responses. For this purpose, the type of sound generated by tapping outside of the steel tubes was examined. If the sound was solid, bond response was estimated to be close to state A or in state C. If the sound was hollow, the bond response was close to state B or in state C. The majority of the specimens in test series II had a solid sound, while hollow sounds were common in test series I. The acoustic responses agreed well with the experimental bond stresses. The bond stress had an exponential distribution at low load levels, and the distribution became more uniform after slip took place. This type of bond stress-slip response was observed for all of the specimens. For the specimens with low bond strength, the circumferential strains were negligible. Those strains were significant if the bond capacity of the specimens was high. This showed that the specimens with high bond strength developed large contact stresses between the steel and concrete due to the Poisson effect. The eccentrically loaded columns had average bond stress values 1.23 and 2.52 times the average bond stress of the concentrically loaded specimens. The cyclic loading caused deterioration in ultimate strength and bond strength of the specimen. The reduction in bond strength was about 50%.
The bond strength of the specimens in test series II was approximately two to three times larger than the specimens of test series I. This trend and the acoustic response of the columns proved that shrinkage took place in long CFT members.
CFT columns were analyzed using finite element analysis for typical section sizes to study the interaction between the steel and concrete. Three-dimensional solid elements were utilized and columns were subjected to axial forces and bending moments. At the beginning, loading was applied to either steel or concrete alone. Elastic load transfer then took place until composite action was developed. The results showed that the bond stress had an exponential distribution when slip was prevented. The maximum bond stress occurred at the end of the column where loading was applied. However, it decreased to zero over a distance of approximately D/2. This distance was smaller for high D/t ratios and larger for low D/t ratios. When slip was allowed, the bond stress was found to have a uniform distribution and then decreased to zero exponentially after a distance of 0.2D from the loaded end.
The bond strength data obtained in this experimental study was compared with other experimental data from the literature. It was found that bond strength decreased rapidly with an increase in the D/t ratio. However, the individual effects of D and t did not exhibit a clear trend. A linear regression analysis on the available data was performed and the following formula was proposed for maximum average bond strength (in MPa) for circular CFTs:
Roeder, C. W., Cameron, B., and Brown, C. B. (1999). “Composite Action in Concrete Filled Tubes.” Journal of Structural Engineering, 125(5), 477-484. doi:10.1061/(ASCE)0733-9445(1999)125:5(477)