Kawano and Matsui 1997

From Composite Systems
Jump to: navigation, search

Beam-to-column connections with external diaphragm details might create difficulty when placed near external walls if the diaphragm plates are large. New connection types were designed to overcome this problem. These connections had vertical stiffeners welded to the column face, substituting for the external diaphragm on one side. In this paper, simple tension tests and cyclic cruciform specimen tests were presented, both of which were conducted on various connection details with vertical stiffeners. In addition, the authors described formulations to calculate the strengths of these connections.

Experimental Study, Results, and Discussion

In the tension tests, four types of connection details for square CFT and square tubular columns were tested. Their labels were A, B, C, and D. Type A was a typical external diaphragm detail. Types B and C both had vertical stiffeners. For type B, the centerlines of the column and the girder coincided with each other. However, they did not coincide for type C. In type D, only the vertical stiffeners were used and the external diaphragms were removed. The simple tension specimens had single girder flanges that were attached to the CFT or tubular columns through vertical stiffeners or external diaphragms. Fillet welding was utilized to connect the elements. A total of fifteen specimens were tested with varying column types, connection details and dimensions. The L/D ratio for the steel tubes was 2. The columns were square and had a D/t ratio of 33.3. The girder flanges, diaphragms and stiffeners had a thickness of 0.177 in.. The average measured compressive strength of concrete for the CFT columns was 5.01 ksi. The measured yield strength of the steel was equal to 62.8 ksi for the steel tube and it was equal to 44.4 ksi for the girders, diaphragm plates and stiffeners.

The specimens were subjected to monotonic tension through the flanges until fracture. All of the specimens except for one failed at the predicted locations. Failure in the stiffener was common for the specimens with connection types B, C and D. For two of the specimens with connection type C, early fracture in the flange took place. This fracture initiated at the edge of the flange, which was close to the vertical stiffener weld, and then moved transversely along the whole flange. The fracture was due to the fact that the vertical stiffener length and the connection length were not equal to each other. This caused a discrepancy in flexural stiffness at the sides of the connection.

Two series of cyclic tests were conducted on cruciform CFT and cruciform hollow tubular specimens. Before testing, the strengths of the specimens were calculated based on the girder capacity, column capacity, and connection capacity, separately. The specimens in series I were designed to have weak connections, while the specimens in series II were designed to have either weak columns or weak girders. The specimens had connection types A, B, or C. The same kinds of materials as were used for the simple tension test specimens were utilized to manufacture the cruciform specimens . The columns were simply supported and subjected to constant axial load throughout the tests. Cyclic shear force was applied at the girder ends until fracture.

For the first test series, the experimental capacity of the specimens showed good correlation with the predicted values. However, for the tubular specimen with connection type C and a weak stiffener, the experimental and theoretical results did not match. Panel zone buckling took place in the connection region of this specimen due to eccentricity of the girder, which caused high shear at connection region. For the specimens in series II having weak columns, the experimental yield strength values were equal to or greater than the theoretical yield strength values. However, the experimental ultimate strength values were less than the theoretical values due to the low yield strength of steel used for the vertical stiffeners. In the case of the specimens in series II having weak girders, the theoretical estimates gave conservative strength values. The horizontal load-deflection diagrams showed that the frames with vertical stiffener connections generally experienced stable hysteresis loops similar to the ones obtained for the specimens with outside diaphragm connections. For some of the specimens, the load-deflection curve behavior improved when the vertical stiffener was increased in size.

Analytical Study

The yield strength of the connections with the vertical stiffener (Py) was taken as the minimum of the yield strengths of the connections of type A and type D. For their ultimate strength (Pu), the same method was proposed. The formulations from the AIJ (1990) design code provisions were presented to calculate the yield strength (Pdy) and ultimate strength (Pdu) of connection type A. On the other hand, the following equations were proposed for the connection type D to calculate its yield strength (Psy) and ultimate strength (Psu) (all stress values are in MPa):

The first and second terms in the equations above represent the force transferred to the column through the stiffener and girder flange, respectively. The empirical factor, β, which is related to the effective width of the column web, was equal to 4 in the case of connections to CFT columns and it was equal to 3 in the case of connections to hollow tubular columns. In addition, a different set of equations was proposed for connection type C to calculate its yield and ultimate strengths. These equations accounted for the unbalanced transfer of force apparent in this connection type:

where Psy and Psu are the yield strength and ultimate strength of the vertical stiffener connection, Pdy and Pdu are the yield strength and ultimate strength of the external diaphragm connection, a and b are distances from the beam center line to the edges of the steel tube.

According to the test results, good correlation was achieved between the experimental and theoretical strength values. Moreover, the failure in the stiffener for specimens with connection types B, C and D proved the accuracy of the given equations.

The authors concluded by presenting two conditions for the proposed equations to be used safely: the thickness of the vertical stiffener should be at least as thick as the girder flange; and the D/t ratio for the steel tube should be less than 35.


Kawano, A. and Matsui, C. (1997). “New Connections Using Vertical Stiffeners Between H-Shaped Beams and Hollow or Concrete-Filled Square 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. 172-185.