Shakir-Khalil and Al-Rawdan 1995, 1996
An experimental study on simple beam-to-column connections of CFT members was presented in several companion papers. A total six series of connections (A, B, C, D, E, and F) were tested, including both circular and square sections for the steel tubes (Shakir-Khalil, 1993a, 1993b, 1994a, 1994b; Shakir-Khalil and Mahmoud, 1995, Shakir-Khalil and Al-Rawdan, 1995, 1996). An analytical study was also performed for the specimens in test series F to predict their failure load and failure pattern.
Experimental Study, Results and Discussions
A total of eight specimens were tested in test series A, B, E, and F, while test series C and D consisted of two specimens. In test series A and C, circular sections were used and for the remaining test series (B, D, E, and F), square sections were utilized. Either finplates (i.e., shear tabs) or tee cleats, which were cut from wide flange sections, were used as the connection elements, with series A to D using finplates, and series E and F using tee cleats. The finplates and tee cleats were fillet welded to the steel tubes and bolted to the beam webs. Hilti nails were provided as shear connectors in the CFTs for some of the specimens in test series A and B to investigate the effect of shear connectors on the connection performance. The test setup consisted of a vertical CFT column having two beams framing in at the mid-height. However, the first group of specimens in test series F were exterior connections that had single beam attached to the columns. The beams were subjected to symmetrical upward shear loading, while an axial load proportional to and in the same direction as the beam loading was also applied at the bottom of the CFT columns. Only the beams of the specimens in the second group of test series F were loaded unsymmetrically. The specimens were laterally braced at mid-height, and the column ends were free to rotate. The nominal length of the columns was 435 in. and their D/t ratio varied from 30.00 to 34.78. The ranges for the measured yield strength of the steel tube and the measured compressive strength of the concrete were 43.9-60.2 ksi and 4.25-5.90 ksi, respectively.
In test series A, overall column failure took place at the upper part of the columns. The tube walls at the finplate locations underwent insignificant deformation. No connection failure occurred despite the local yielding of the steel tubes at the lower part of the connections. In this region, the local yielding of the steel tubes was caused by the combined effects of the compressive stresses from the axial load applied to the columns, transverse tensile stresses due to the moment from the beams and residual compressive stresses as a result of the welding process. The longitudinal strain of the steel tubes was found to increase at the same rate for the gauged points located away from the finplates at a distance more than the diameter of the steel tube. This showed that the load transfer between steel and concrete was completed within a distance less than the diameter of the steel tube above and below the finplates. The failure load of the specimens was found to increase with the use of shear connectors at the connection region. Providing higher finplate depth and smaller moment arms for the beam loads also increased the failure load of the specimens. At the end of the tests, no concrete crushing at the connection region was observed and the major damage was the elongation of the bolt holes. In test series B, the specimens having low eccentricity in the beam loading exhibited column failure in the upper part of the connection. Steel yielding at the top of the column was observed and local buckling took place for some of the specimens. On the other hand, the specimens with large eccentricity in the beam loading underwent large rotations and failed by outward bending of the steel tube at the toe of the finplate. The concrete core prevented a complete yield line failure, as it did not allow the steel tube corners to deform inward and caused a membrane mechanism to develop making the section stifffer. However, the failure load of these specimens was lower than the ones that experienced column failure. The strain distribution above the finplate position showed that beam force was transferred to the composite section within a distance of approximately half of the depth of the square tube. At the end of the tests, the concrete was examined and no crushing was observed.
The performance of the finplate connections for square CFTs of test series B and D was not as good as for the circular CFTs of test series A and C. The connection stiffness of the specimens in test series A and C were found to be approximately 2.7 times those of the specimens in test series B and D. This was due to the flexibility of the rectangular tube walls. In contrast to test series B and D, tee cleats were utilized for the square CFTs of test series E. The tee cleat connections were found to be about 2.4 times stiffer than the finplate connections, comparing the result of test series E to those of test series B. For some of the tee cleat connections, transverse welds were applied along their flanges, which caused an increase in stiffness about 30%. In test series A to E, most of the specimens had a sudden increase in rotation between 88.51 kips-in. and 221.28 kips-in. moments. This was due to the slip occurring at the bolt holes and the outward bending of the steel tube walls.
The first group of specimens in test series F, which were designed as exterior connections, suffered in-plane deformations at the mid-height, since the braces were subjected to high forces close to the failure load of the columns. The strains along the column heights indicated the combined effect of axial compressive force and bending. On the connection side of the columns, the bending effect caused high compressive and tensile strains at the top and bottom of the connection regions, respectively. The maximum strains in the steel tube occurred at 2.95 or 5.91 in. above tee cleat and decreased toward the end of the column. The location of the minimum strains in the steel tube was 2.95 or 5.91 in. below the tee cleat and the strains increased gradually toward the lower end of the steel tube. On the opposite side of the column at 2.95 to 5.91 in. above the connection, the strains were minimum, although they increased toward the upper end of the column. On the opposite side of the column, below the connection region, the strains were the highest at the end of the tee cleat and decreased toward the lower end of the column. Yielding on the steel tube was observed above the tee cleats at the connection side between 204.59 and 260.79 kips axial loads. The local buckling of the steel tube generally occurred 4.72 to 7.09 in. above the tee cleats at the connection side. During the initial stages of loading, the stiffnesses of some of the connections were observed to decrease, which was attributed to slip at the boltholes. It was speculated that the specimens having no reduction in initial stiffness might have bolts that were overtightened. Moreover, it was probable that the connection with lowest initial stiffness had lightly tightened bolts. The experimental failure loads were compared with BS5950 (1985) and BS5400 (1979) design code provisions and the experimental values were conservatively estimated .
The second group of the specimens in test series F were configured like the other test series, with two beams per specimen at the mid-height. However, these specimens were subjected to unsymmetrical beam loading. Close to the failure load, large forces developed on to the lateral braces and this caused in-plane deformations at the midheight and at the ends of the columns. The specimens exhibited similar responses to the first group of specimens in stiffness and slip at the bolt holes caused sudden drop of the initial connection stiffness. For all the specimens in the second group of test series F, overall column failure was observed. The BS5400 (1979) and BS5950 (1985) procedures predicted the experimental load conservatively.
An analytical study was also performed in order to estimate the behavior of the specimens in the second group of test series F. The failure loads were calculated and the bending moment-axial load interaction diagrams were generated for the column sections. For this purpose, a rectangular stress block approach and analyses using ABAQUS were used. In addition, the connections were also modeled in ABAQUS with 3D finite elements. The beams were not modeled separately and the stem of the tee cleat was extended to represent the beams. The main objective of the numerical model was to investigate the yield line failure of the tee cleats. For all the specimens, the rotation in the yield line failure took place at the level of second bolt from the top of the tee cleat. The experimental results also proved this response when the concrete core was investigated after testing. The analysis results indicted a gap along the yield line between the tee cleat flange and the column. It was observed that after sufficient amount of deformation along the yield line, the yield line failure pattern changed into a stable membrane behavior. The failure loads found by 3D finite element analysis of the connection gave the best correlation with the experimental results by giving 2% to 6% higher failure loads than experimental values.
Shakir-Khalil, H. (1993a). “Full-Scale Tests on Composite Connections,” Composite Construction in Steel and Concrete II, Easterling, W. S. and Roddis, W. M. K. (eds.), Proceedings of the Engineering Foundation Conference, Potosi, Missouri, June 14-19, 1992, American Society of Civil Engineers, New York, New York, pp. 539-554.
Shakir-Khalil, H. (1993b). “Connection of Steel Beams to Concrete-Filled Tubes,” Proceedings of the Fifth International Symposium on Tubular Structures, Nottingham, England, U.K., August, pp. 195-203.
Shakir-Khalil, H. (1994a). “Finplate Connections to Concrete-Filled Tubes,” Steel-Concrete Composite Structures, Proceedings of the Fourth International Conference on Steel-Concrete Composite Structures, Javor, T. (ed.), Kosice, Slovakia, 20-23 June 1994, held by the Association for International Cooperation and Research in Steel-Concrete Composite Structures, Expertcentrum, Bratislava, Slovakia, 1994, pp. 181-185.
Shakir-Khalil, H. (1994b). “Beam Connections to Concrete-Filled Tubes, “ Tubular Structures VI, Proceedings of the Sixth International Symposium on Tubular Structures, Grundy, P., Holgate, A., and Wong, W. (eds.), Melbourne, Australia, 14-16 December 1994, A. A. Balkema, Rotterdam, The Netherlands, 1994, pp. 357-364.
Shakir-Khalil, H. and Al-Rawdan, A. (1995). “Behaviour of Concrete-Filled Tubular Edge Columns,” Proceedings, Istanbul, Turkey, June.
Shakir-Khalil, H. and Mahmoud, M. A. (1995). “Steel Beam Connections to Concrete-Filled Tubular Columns,” Nordic Steel Construction Conference '95, Sweden, June.
Shakir-Khalil, H. and Al-Rawdan, A. (1996). “Behaviour of Asymmetrically Loaded Concrete-Filled Tubular Columns,” Tubular Structures VII, Proceedings of the Seventh International Symposium on Tubular Structures, Farkas J. and Jarmai, K. (eds.), University of Miskolc, Hungary, August 28-30, 1996, pp. 363-370.