Dunberry, LeBlanc, and Redwood 1987

From Composite Systems
Jump to: navigation, search

Short square CFT columns were loaded axially by means of shear tab connections coupled with direct axial compression. The axial load capacity of the specimens and the load transfer mechanism between steel and concrete were investigated. Using the test results, the authors proposed an equation for the cross section strength of CFT columns.

Experimental Study, Results, and Discussions

A total of four series of tests (A, B, C, and D) were performed on CFT columns. Axial load was applied to the specimens both at the top of the CFT and at the connections. The ranges for the D/t and L/D ratios were 20.8-37.1 and 3.7-29.4, respectively. The nominal yield strength of the steel was 50.77 ksi and the measured compressive strength of concrete varied from 2.52 to 4.29 ksi. The ratio of the axial load applied through the connections to the total axial load (β) was changing from 0 to 1. All of the specimens were grouted at the bottom, while grouting was applied to the top if the specimens were subjected to an axial load at their top ends. In the case of no axial load acting at the top of a specimen, either grouting and steel capping were applied together or the top end was left without grouting and capping. Bracing against overall buckling was provided for all of the specimens. Test series A consisted of columns having standard-tee connections and the smallest D/t ratios. Test series B was identical to test series A, but higher D/t ratios were used. Test series C included the specimens having no grouting or capping at the top, and the specimens in this test series that did not have any load applied at the connections. In test series D, the columns had different connection details including standard- tee, extended-tee, single plate, and shortened-tee details. For all of the specimens, the steel tube and concrete core were instrumented separately to measure their strain values.

In most of the experiments, the total axial load capacity achieved in the specimens was close to their squash loads. The specimens loaded through the connections tended to have less strength compared to the specimens loaded only at the top. The difference in strength was at most 8%. The steel and concrete strains along the column length exhibited incompatibility at the connection regions and the relative slip between concrete and steel was found to be between 0.00315 and 0.00630 inches. The development length over diameter ratio was close to 3 at the top of the connections, and it was varied from 1 to 2 at the bottom of the connections. The failure pattern was generally local buckling, which took place within the connection region or below the connection region. Some specimens experienced overall buckling due to low stiffness and strength of the bracing. The axial load carried by concrete was found to increase along the connection, with the largest gain in force taking place in the bottom half of the connection. The concrete load also continued to increase below the connection region.

Local buckling was observed in the test series D. For single plate connections, the location of local buckling was a distance 4D below the connection, and a rapid increase in concrete load was observed in the bottom half of the connection due to concentrated pinching action of the plate. On the other hand, local buckling took place close to the connection region in extended-tee type connections, and a gradual transfer of load to the concrete was observed. The rotation of joints took place without slip at the bolt holes. This showed that load transfer to the concrete might also occur by joint rotation. Local buckling generally took place after yielding of the steel. Thus, the presence of concrete did not improve the capacity of these CFT columns significantly. However, it increased the failure strain of the specimens by between 2 to 29%. The occurrence of local buckling was mainly affected by the transfer of shear force to the concrete within the connection length. If the shear load transferred to the concrete within the connection was large, a lesser amount of load could be transferred to the concrete outside of the connection. This might cause excessive steel stress and subsequent local buckling in that region.

Using the experimental results, the following formulation was proposed to calculate the cross section strength of CFT columns (all stress values are in MPa):


The γ factor accounted for the reduction in concrete strength due to local buckling of steel tube. The reduction in concrete strength from size and curing effects was accounted for with the factor α (which ranged between 0.85 or 1).


Dunberry, E., LeBlanc, D., and Redwood, R. G. (1987). “Cross-Section Strength of Concrete-Filled HSS Columns at Simple Beam Connections,” Canadian Journal of Civil Engineering, Vol. 14, pp. 408-417.