Beam-column joint models, extracted from representative connections, were tested by applying bending loads. Experimental variables adopted are:
type of weld tab;
welding procedure;
test temperature;
weld defects; and
loading pattern.
Test results showed that the location and shape of weld defects influenced the rotation capacity of joint models significantly. This paper presents the effects of these variables on the plastic deformation capacity and the relationship between stress triaxiality and J-integral to brittle fracture. Steel weld tabs, as compared with flux tabs, gave detrimental effects on the plastic deformation capacity. When the stress triaxiality was low at the notch root, brittle fractures initiated at a large J-integral value.
The 1994 Northridge and 1995 Kobe Earthquakes revealed that the plastic deformation capacity of steel building frames was significantly reduced by cracks starting from various discontinuities in beam-column joints. In Kobe, cracks frequently started firstly from the toes of weld access holes and secondly from the ends of CJP (Complete Joint Penetration) groove welded joints between the beam flanges and diaphragms. Post-earthquake investigations demonstrated that improvements in profiles of beam copes could retard the former cracks (AIJ Kinki, 1997), while it is possible that weld defects (lack of fusion, slag inclusions, undercut, overlap) occur easily around the ends of CJP groove welds. Weld defects can easily lead to brittle fracture because these are discontinuities with a sharp notch. It was observed in the Kobe Earthquake and in post-earthquake investigations that ductile cracks originating from weld defects led to brittle fracture. Brittle fracture from the ends of CJP groove welded joints in beam-to-column connections is the subject of this study. If weld defects appear in steel structures, defect regions may need to be repaired if the defects are rejectable.