The popular field welded–field bolted beam-to-column moment frame connection, shown in both of the current AISC LRFD and ASD steel design manuals, has become known as the “pre-Northridge” connection. As a result of surveys of damaged buildings made subsequent to the 1994 Northridge, CA earthquake, it was concluded by the Structural Engineers Association of California (1996 SEAOC Blue Book Commentary) that the pre-Northridge connection is fundamentally flawed and should not be used in the construction of new seismic moment frames. This conclusion was based upon (1) this connection's modes of fracture, (2) an historic review of the 1960 - 1990 tests made at the University of California and Lehigh University that led to this connection's design rationale, and (3) ATC-24 protocol tests in the SAC/FEMA research program.

Extensive finite element analyses and strain gage data from ATC-24 tests by Seismic Structural Design Associates, Inc. (SSDA) of this pre-Northridge connection made subsequent to the Northridge earthquake in 1994 and 1995 showed large stress and strain gradients horizontally across and vertically through the beam flanges and welds at the face of the column. These large flange stress and strain gradients are caused by a very large component of the seismic shear, typically 50% or more, that is carried by the beam flanges. It is these flange shear forces that produce a prying moment in the welds and flanges at the face of the column and at the weld access holes that led to the weld and flange fractures and to the column flange divot modes of connection fracture. Moreover, the large plastic strains associated with these prying moments significantly reduce the fatigue life of the connection (Richard, R.M., et al., "Accumulated Seismic Connection Damage Based upon Full Scale Low Cycle Fatigue Connection Tests", Proceedings of the Structural Engineers Association of California 70th Annual Convention, p. 43-48, September 27-29, 2001). Additionally, failure analyses of the SAC ATC-24 test specimens at the University of Michigan and at Lehigh University demonstrated that fatigue cracks initiated and propagated in all the tested specimens (John M. Barsom, "SAC Steel Project 7.1.3", May, 2000).

A six year $12,000,000 seismic research program by SAC (a consortium of three organizations, (1) SEAOC, (2) the Applied Technology Council, and (3) the Consortium of Universities for Research in Earthquake Engineering), which was funded by the Federal Emergency Management Agency (FEMA), the American Institute of Steel Construction (AISC), and the National Science Foundation (NSF), was concluded in August 2000. A series of seismic connection design seminars were then given that were funded by FEMA. Attendees at these seminars were provided with a set of notes entitled "New Recommended Seismic Design Criteria for Steel Moment Frame Buildings: Speakers Slide's". In these notes it is stated that the large component of seismic shear, typically 50%, and its prying effects still exist in the beam flanges and welds of the SAC pre-qualified Reduced Beam Section (RBS), the WUF-W, and the WUF-B connections, which use the "standard" unreinforced flange connection (James Malley, p. 5-19; Stephen Mahin, p. 3-14,15). However, the published FEMA-350, July 2000, design rationale for the RBS, WUF-W, and WUF-B connections does not account for these large flange shears wherein all of the shear is assumed to be resisted by the beam web connection. Specifically, the beam web connection design rationale in FEMA-350 for the RBS is given on p. 3-41, design steps 4 and 5; for the WUF-W on p. 3-31; and for the WUF-B on p. 3-28, design steps 5 and 6.


Historically, the departure of the connection force distribution from elementary beam theory, wherein the "flanges carry the moment and the web carries the shear", was first analytically determined by Professor Y.Y. Yu in 1959 (Jr. of Applied Mechanics, Vol. 26, No. 3, pp. 415-429, 1959). Moreover, in 1968 John F. Abel and Egor P. Popov of the University of California, using the finite element method, confirmed Yu's analytical solution that showed 100% of the beam shear is carried by the flanges in a cantilever beam at a rigid support (AFFDL T.R. No. 68-150, pp. 213-245, 1968). Additionally, in 1970 Popov and Stevens reported that in their connection test of a W18x50 beam "…with welded flanges only and no web attachment whatsoever, exceeded in capacity the yield moment while transmitting a large shear." They further stated that "Since the welded flanges appear to transfer to the column most of the beam shear, the precise number of bolts required for hybrid connections needs further investigation" (University of California, Berkeley, EERC Report No. 70-3, July 1970).

An alternative connection design using vertical beam flange fins to account for the flange shear has been tested successfully by SAC (AISC Engineering Jr., p. 43, 2nd Qtr., 1997 and AISC Engineering Jr., p. 31, 1st Qtr., 2000) and by Zekioglu, A. et al., "Designing After Northridge", Modern Steel Construction, March 1997, p. 36-42). SAC apparently has not recommended the use of vertical flange fins because of their increased cost of fabrication.