The 1994 Northridge earthquake caused widespread and unexpected damage to s
teel moment-resisting joints and connections. Shortly following the earthqu
ake, the Federal Emergency Management Agency funded a series of full-scale
tests of steel moment-resisting joints and connections to characterize the
behavior of pre-earthquake connections and to evaluate the efficacy of st s
elected number of repair schemes. Twelve pre-earthquake connections were te
sted. Three of the twelve connections were tested by the authors to failure
, and then repaired and re-tested. The response of the pre-earthquake conne
ctions was highly variable and uniformly poor. Premature fractures were obs
erved in all twelve connections, and the types of fractures were similar to
those observed in the field following the earthquake. The mean beam plasti
c rotation was 0.005 rad: one-sixth of the target value of 0.03 rad. The re
sponse of those moment-resisting connections that were repaired by replacin
g fractured weld and parent metal with toughness-rated weld filler metal wa
s also poor. On the basis of the studies described in the paper, the rotati
on capacity of large-size moment-resisting connections built prior to the N
orthridge earthquake is smaller than the target values established followin
g the earthquake; rehabilitation of earthquake-damaged moment-resisting con
nections by re-welding only will likely be ineffective; beam-column panel z
ones should be designed to remain elastic for the forces associated with pl
astic hinging in the beams; design equations for continuity plates should b
e revised; and design checks for flange compactness should be based on expe
cted rather than nominal material properties. (C) 1998 John Wiley & Sons, L
td.