Matsui, Tsuda, and El Din 1993
This paper, along with El Din et al. (1993), reports the results of a series of 30 tests on long CFT and HT beam-columns under eccentric and concentric axial load. While the first paper gives a summary of the results and important conclusions, the second paper presents a more detailed description of the test parameters, materials, and experimental results. The design formulas from the Japanese code (AIJ) are compared to the experimental results, and the effect of column slenderness is discussed.
Experimental Study, Discussion, and Results
Thirty column and beam-column tests were performed on CFT and HT sections. The column tests were performed with simple support conditions, and the beam column tests were on pinned end columns with applied eccentric loading. The main test parameter was the slenderness ratio Lk/D (= 4, 8, 12, 18, 24, and 30). The secondary parameter was the eccentricity of the axial load, with four cases tested for each column group (e = 0, k, 3k, and 5k, where k is the eccentricity required to induce tension at the extreme fiber of the cross-section). One column test from each group was performed on a HT under concentric loading. The tubes were all 150 x 150 x 4.5 mm. square sections filled with 4.6-5.0 ksi concrete.
The specimens were instrumented to record strains and deflections at the midhieght of the tube. The tests were terminated when a stable post-ultimate load was achieved or when the deflection limit of the test rig was reached. The results are presented as plots of load verses midhieght deformation, with the results of a material nonlinear analysis plotted for comparison. The HT test and analytical results varied the most due to local buckling and other geometric nonlinear effects. The CFT tests, however, showed very good correlation between analysis and experiment. The largest difference was in the post-ultimate region when the analysis tended to give results which were below the experimentally measured values. This was due in part to the material models chosen for the analysis, as the concrete model did not allow for a plateau of the stress-strain diagram for large strains [such as in Tomii and Sakino (1979b)].
The authors note that the limiting slenderness ratio for negligible geometric nonlinear effects is approximately 12. The HT limiting slenderness ratio is also observed to be around 12. The authors describe three modes of failure which depend on the eccentricity and the slenderness of the section. The first is a squashing failure, which is observed for short columns (Lk/D<12) with low to moderate eccentricities. This mode of failure is dominated by local buckling of the steel section at the ultimate load of the CFT. For short sections with large eccentricities, the critical load is also reached as local buckling is initiated. The second mode of failure is observed in longer columns which have a moderate to high eccentricity. The moment in the section at ultimate load is near the capacity of the section, and the author attributes failure to a three-hinge beam mechanism. For long columns (Lk/D>18) under concentric loading, the dominant failure mode is flexural buckling. Finally, the authors compared the Japanese code recommendations for ultimate capacity of CFT and HT sections with the experimental results. They found that the code was very conservative when applied to CFTs, but less conservative for HTs.
Reference
Matsui, C., Tsuda, K., and El Din, H. Z. (1993) “Stability Design of Slender Concrete Filled Steel Square Tubular Columns” Proceedings of the 4th East Asia-Pacific Conference on Structural Engineering and Construction, Vol. 1, pp. 317-322.
El Din, H. Z., Matsui, C., and Tsuda, K. (1993). “Stability and Post-Buckling Behavior of Concrete Filled Square Tubular Columns,” Journal of Structural Engineering, Japan Society of Civil Engineering, Vol. 39B, pp. 323-334.