Design of steel MRF connections before and after 1994 Northridge earthquake

Abstract This paper presents the studies of pre-Northridge and post-Northridge earthquake welded beam-to-column connections used in typical steel moment-resisting frames (MRFs). The purpose of these studies is to give explanations on both fracture locations and failure modes of the aforementioned connections in rational ways. The stress concentration at the juncture of welded beam flange and column flange is analysed by three-dimensional elastic–plastic finite elements. The result clearly explains why the weak beam flange breaks off right at the weld due to triaxial actions in that region with no apparent yielding. The effect of backing bar in connection failure is analysed by fracture mechanics methods. The unfused backing bar surface next to the column flange is interpreted as an artificial crack. During the bending of a beam, the tension in the flange opens the artificial crack and initiates the rupture. The stress intensity factors at the artificial crack tips of both top and bottom backing bars are calculated by the J-integral method. The result explains why the rupture primarily started at the bottom flange but not at the top flange. The analytical cyclic load–deflection curves are compared with the SAC 2 full-size specimens tested at UC Berkeley. Good agreements between the analytical results and the experimental tests are found. Two remedy designs using a dog bone and reinforcing plates are presented at the end. Experimental results demonstrated their merits over the conventional design: large ductility without any brittle fracture.