A qualitative discussion was presented including the topics of bond in CFTs, the behavior of CFTs as columns in frames, and the behavior of CFT columns under various combinations of axial compression, bending, and shear. Both monotonic and cyclic loading was studied and emphasis was placed on the deformation capacity of the tubes. The discussions centered around previous tests conducted by the author and others.
Experimental and Theoretical Discussion
Axially Loaded Columns. The author included a scatter-plot of the axial load ratio to the slenderness ratio of a number of previous tests. The plot showed that the ultimate load of circular CFTs is considerably larger than the nominal load, which is the sum of the two component strengths. This is due to strain hardening of the steel and confinement of the concrete. Although the confinement effect diminishes with increasing column length and is generally neglected for columns of practical length, it ensures that the column behaves in a ductile manner, a distinct advantage in seismic applications. Previous tests by the author and others resulted in a classification based on test parameters including cross-section shape, D/t ratio, and concrete strength. Three categories emerged: strain hardening type, elastic-perfectly plastic type, and degrading type, which was characterized by a loss of load carrying capacity after the yield point was reached. All of the square columns that were tested exhibited a degrading type of axial load versus strain curve. The circular tubes all produced either strain hardening or elastic-perfectly plastic type curves. The author proposed using the yield load rather than the ultimate load as the capacity of the section, because the ultimate load was much more sensitive to the effects of varying the slenderness ratio and the D/t ratio.
Beam Columns. Tomii and Sakino (1979a) conducted tests on short beam-columns to obtain moment-thrust-curvature relationships for square CFTs with an increasing moment under constant axial load. Despite the occurrence of local buckling in the steel tube, the CFT was more ductile than a comparable reinforced concrete column. The paper discussed an earlier method proposed by the author and Sakino of determining the axial load-moment interaction diagram. This “simple method” used for square columns is based on simple rectangular stress blocks for the tube and the concrete core without a limiting concrete strain since the concrete stress-strain curve used was very ductile (Furlong proposed a more complicated method involving a limiting crushing strain of 0.003). Limited details were shown regarding this method, but it was stated that it could also be used conservatively for circular cross-sections.
Columns under Compression, Bending, and Shear. Members under this combination of loading are very important in seismic applications. Although designs follow the weak beam-strong column philosophy, many instances arise where the base of columns are subjected to high cyclic shear. Eighty tests of short (columns in which shear cannot be neglected) and medium length square columns bent in double curvature were studied, the details of which were omitted in the paper's discussion. For monotonic tests, two failure mechanisms occurred. The short columns (with a shear span to depth ratio of 1.0 or smaller, where the shear span is L/2) showed a shear failure, while medium columns (shear span to depth ratio of 1.5 or larger) produced a flexural failure. Therefore a critical value existed somewhere between 1.0 and 1.5. For the cyclic load tests of square tubes, which were conducted under a high axial load ratio (0.5), the specimens exhibited a small amount of strength degradation, but the hysteretic loops were very stable and a large amount of energy was dissipated. The strength degradation was due to the local buckling of the tube walls at the points of maximum shear stress (the top and bottom of the column) and a subsequent crushing of the concrete at those locations. The tube becomes circular at the buckling/crushing points, which stabilizes the behavior of the column. The short columns, which failed by shear, showed less strength deterioration that the medium columns, which failed by flexure. An additional problem with the local buckling of the steel and the crushing of the concrete was a considerable axial shortening (up to 30% of the column depth). Therefore, the magnitude of the axial load should be limited to mitigate this effect.
Columns in Frames. The author discussed frame tests conducted by Matsui (1985), which compared square concrete-filled steel tubes and square hollow steel tubes. The frames were devised such that the plastic hinges would form in the columns, thus making the CFT or HT columns the critical members. A considerable improvement in both strength and ductility was shown by filling the hollow tubes with concrete. Local buckling of the steel tube, which produced severe strength degradation in the hollow tube tests, did not affect the behavior of the CFTs. The post-local buckling behavior was strongly influenced not only by the difference in the local buckling mechanism between the CFT and HT, but more so by the transfer of axial load from the steel to the concrete in the CFT. This greatly enhanced the energy absorption capacity of the CFT.
Bond. In practice, only limited bond strength is realized. Therefore the full composite action cannot be counted on. The author has proposed design criteria to transfer axial force from the steel tube to the concrete. For long-term loading, a portion of the column should have a zero bond stress resultant such that the concrete core can carry as much long-term axial compression force as possible. For seismic loading, the concrete core should carry the maximum compressive force at the critical sections at the top and bottom of the column such that the CFT can be designed for a moment corresponding to this axial force. Bond tests performed by Morishita et al. (1979a, b) showed that the concrete strength had an insignificant effect on the bond strength and the bond strength was considerably less than that observed between reinforcement bars and concrete. Their tests resulted in the following recommendations for the bond strength between the steel and concrete: for long term monotonic loading -- 21.8 psi for circular sections and 14.5 psi for rectangular; for cyclic loading -- 32.0 psi for circular sections and 21.8 psi for rectangular. They proposed using either a checkered internal tube surface or an expansive concrete to improve bond, the latter of which was deemed more effective.
Tomii, M. (1991). “Ductile and Strong Columns Composed of Steel Tube, Infilled Concrete, and Longitudinal Steel Bars,” Proceedings of the Third International Conference on Steel-Concrete Composite Structures, Special Volumne, Wakabayashi, M. (ed.), Fukuoka, Japan, September 26-29, 1991, Association for International Cooperation and Research in Steel-Concrete Composite Structures, pp. 39-66.