Shakir Khalil and Al-Rawdan 1997
This paper concluded a series of four papers by the first author (see also Shakir-Khalil and Zeghiche, 1989; Shakir-Khalil and Mouli, 1990; Shakir-Khalil, 1991; Shakir-Khalil, 1994) documenting tests of full-scale rectangular CFT columns. This segment contained the test and analysis results of 11 members subjected to uniaxial bending about the major and minor axes. The experimental failure loads were compared to the predictions of the British Standard, BS5400 (1979), as well as, to the predictions of three-dimensional finite element studies.
Experimental Study, Results, and Discussions
Stub columns of rectangular CFTs were first tested to determine the squash load of CFT members. Column CFT specimens were then tested monotically in a horizontal position. Pin-ended support conditions were simulated by the test setup. The L/D ratios were ranging from 21 to 49. For major and minor axis bending, the D/t ratio of the specimens was 30 and 20, respectively. The applied eccentricities did not exceed one half the diameter of the column. The yield strength of steel was varying from 47.0 ksi to 53.3 ksi. The compressive strength of concrete ranged between 5.3 ksi and 6.0 ksi. The stub columns exhibited 16 to 30% higher strength than their nominal axial load capacity calculated according to BS5400 (1979). Their strength was observed to decrease with an increase in length due to local buckling. The local buckling generally took place at the longer side of the tubes. The concrete was investigated after testing. It was crushed but kept its integrity, thus facilitating the achievement of the large strengths in the stub columns. In addition, the CFT columns were found to have an increase in strength of 25-37% over similar hollow tubes. Except for one case, the failure load decreased with an increase in end eccentricity. This was because that specimen experienced pure bending response about the major axis. The authors noted that the behavior of columns subjected to small eccentricities about the major axis was especially sensitive to any imperfections, most notably, out-of-straightness.
The results showed that the BS5400 (1979) design code conservatively estimated failure due to major axis bending by 20-66% (although the author indicated that this may not be the case for longer columns). The BS5400 (1979) specification, however, overestimated the strength in minor axis bending, and the author deemed it imperative that a change in the code be implemented.
Analytical Study
Using a standard stress block approach, the interaction diagram for the stub column sections was generated. The same graph was also derived in the finite element program ABAQUS. The CFT specimen was modeled by using brick elements for the concrete and shell elements for the steel. The brick elements close to the supports were rigid to simulate the effect of loading plates. Perfect bond between the steel and concrete were assumed. The interaction diagrams obtained from the two methods correlated well. In the three-dimensional finite element analysis, the tensile strength of concrete was accounted for, which resulted in somewhat higher strength values than the stress block approach. However, the difference for the maximum moment values from the two methods was found to be approximately 5%.
The column specimens were modeled in ABAQUS in the same way as the stub column specimens. The columns under major axis bending were analyzed with zero and 0.118 in. minor axis eccentricity, separately, where 0.118 in. was the minimum eccentricity about the minor axis required by BS5400 (1979) for the tested column sizes. The finite element results showed that the effect of the 0.118 in. eccentricity was important in the case of long columns subjected to a small amount eccentricity about their major axis. The load capacities for these specimens were found to be lower and their mode of failure was governed by out-of-plane displacements when minor axis eccentricity was introduced. The computational results obtained with minor axis eccentricity were conservative and showed better agreement with the experimental results than the BS5400 (1979) predictions. Another observation from the finite element analysis was that the in-plane mid-height deflections were found to get larger with an increase in column length and major axis eccentricity.
When the columns that were tested under minor axis bending were analyzed, the load-displacement relationships matched well with the computational results from ABAQUS. Although. the computational results were found to overestimate the experimental failure loads by 6 to 12%, they were still better than the BS5400 (1979) predictions.
Reference
Shakir-Khalil, H. and Zeghiche, Z. (1989). “Experimental Behavior of Concrete-Filled Rolled Rectangular Hollow-Section Columns,” The Structural Engineer, Vol. 67, No. 19, pp. 345-353.
Shakir-Khalil, H. and Mouli, M. (1990). “Further Tests on Concrete-Filled Rectangular Hollow-Section Columns,” The Structural Engineer, Vol. 68, No. 20, pp. 405-413.
Shakir-Khalil, H. (1991). “Tests on Concrete-Filled Hollow Section Columns,” Proceedings of the Third International Conference on Steel-Concrete Composite Structures, Wakabayashi, M. (ed.), Fukuoka, Japan, September 26-29, 1991, Association for International Cooperation and Research in Steel-Concrete Composite Structures, pp. 89-94.
Shakir-Khalil, H. (1994). “Experimental Study of Concrete-Filled Rectangular Hollow Section Columns,” Structural Engineering Review, Vol. 6, No. 2, pp. 85-96.
Shakir-Khalil, H. and Al-Rawdan, A. (1997). “Experimental Behavior and Numerical Modelling of Concrete-filled Rectangular Hollow Section Tubular Columns,” Composite Construction in Steel and Concrete III, Buckner, C. D. and Shahrooz, B. M. (eds.), Proceedings of the Engineering Foundation Conference, Irsee, Germany, June 9-14, 1996, American Society of Civil Engineers, New York, New York, pp. 222-235.