This paper describes experimental research on three full-scale tests of beam-to-column connections under cyclic loading. The specimens were configured as a T-shaped assembly, with two specimens having a wide flange section column and partial joint penetration (PJP) groove welds at the ends of the beam flanges (BH-1, BH-2). BH-1 has the area of incomplete penetration of 50% of the beam flange thickness (16mm), and BH-2 has the area of incomplete penetration of 75% of the beam flange thickness (16mm). BH-1 had insufficient cross-sectional area of welded joints, while BH-2 had sufficient cross-sectional area due to larger fillet welds. Tests on BH-1 and BH-2 concern the applicability of PJP groove welds for beam-to-column connections. The third specimen had a wide flange beam field-welded to a cold-formed square hollow section column (SHS) (BS-3). BS-3 had an internal diaphragm at the position of both the beam top and bottom flanges, and used a cover-plate to reinforce the connection. The aim of the test was to investigate the plastic rotation capacity and ultimate strength of this type of beam-to-column connection.
Both the 1994 Northridge and the 1995 Kobe earthquakes took structural engineering professionals by surprise in that many of the welded connections in modern steel building frames sustained brittle fractures. Before these earthquakes, a common practice for engineers was to assume that the joints between the beam flanges and columns using complete joint penetration (CJP) groove welds satisfied overstrength criteria to allow formation of plastic hinges in beams. These fractures most frequently occurred in regions around the beam bottom flange groove welds. In Northridge, brittle fractures initiated at a very low level of plastic demand. In Kobe, however, the majority of beam-to-RHS column connections that fractured during the earthquake were accompanied by extensive yielding or local buckling at the beam ends.
