【正文】 l stiffener assisted in transferring the axial forces and delayed the formation of web local buckling. 1. Introduction Aimed at evaluating the structural performance of reducedbeam section (RBS) connections under alternated axial loading and lateral displacement, four fullscale specimens were tested. These tests were intended to assess the performance of the moment connection design for the Moscone Center Exp ansion under the Design Basis Earthquake (DBE) and the Maximum Considered Earthquake (MCE). Previous research conducted on RBS moment connections [1,2] showed that connections with RBS profiles can achieve rotations in excess of rad. However, doubts have been cast on the quality of the seismic performance of these connections under bined axial and lateral loading. The Moscone Center Expansion is a threestory, 71,814 m2 (773,000 ft2) structure with steel moment frames as its primary lateral forceresisting system. A three dimensional perspective illustration is shown in Fig. 1. The overall height of the building, at the highest point of the exhibition roof, is approxima tely m (116ft) above ground level. The ceiling height at the exhibition hall is m (27 ft) , and the typical floortofloor height in the building is m ( ft). The building was designed as type I according to the requi rements of the 1997 Uniform Building Code. The framing system consists of four moment frames in the East–West direct ion, one on either side of the stair towers, and four frames in the North–South direction, one on either side of the stair and elevator cores in the east end and two at the west end of the structure [4]. Because of the story height, the con cept of the CoupledGirder MomentResisting Framing System (CGMRFS) was utilized. By coupling the girders, the lateral loadresisting behavior of the moment framing system changes to one where structural overturning moments are resisted partially by an axial pression–tension couple across the girder system, rather than only by the individual flexural action of the girders. As a result, a stiffer lateral load resisting system is achieved. The vertical element that connects the girders is referred to as a coupling link. Coupling links are analogous to and serve the same structural role as link beams in eccentrically braced frames. Coupling links are generally quite short, having a large shear tomoment ratio. Under earthquaketype loading, the CGMRFS subjects its girders to wariab ble axial forces in addition to their end moments. The axial forces in the Fig. 1. Moscone Center Expansion Project in San Francisco, CA girders result from the accumulated shear in the link. Fig 2. Analytical model of CGMRF Nonlinear static pushover analysis was conducted on a typical onebay model of the CGMRF. Fig. 2 shows the dimensions and the various sections of the model. The link flange plates were mm ??254 mm (1 1/8 in ??10 in) and the web plate was mm ??476 mm (3 /8 in ??18 3/4 in). The SAP 2020 puter program was utilized in the pushover analysis [5]. The frame was characterized as fully restrained(FR). FR moment frames are those frames for 1170which no more than 5% of the lateral deflections arise from connection deformation [6]. The 5% value refers only to deflection due to beam–column deformation and not to frame deflections that result from column panel zone deformation [6, 9]. The analysis was performed using an expected value of the yield stress and ultimate strength. These values were equal to 372 MPa (54 ksi) and 518 MPa (75 ksi), respectively. The plastic hinges’ load–deformation behavior was approximated by the generalized curve suggested by NEHRP Guidelines for the Seismic Rehabilitation of Buildings [6] as shown in. Fig. 3. △y was calcu lated based on Eqs. () and () from [6], as follows: P–M hinge load–deformation model points C, D and E are based on Table from [6] for △y was taken as rad per Note 3 in [6], Table . Shear hinge load load–deformation model points C, D and E are based on Table [6], Link Beam, Item a. A strain hardening slope between points B and C of 3% of the elastic slope was assumed for both models. The following relationship was used to account for moment–axial load interaction [6]: where MCE is the expected moment strength, ZRBS is the RBS plastic section modulus (in3), is the expected yield strength of the material (ksi), P is the axial force in the girder (kips) and is the expected axial yield force of the RBS, equal to (kips). The ultimate flexural capacities of the beam and the link of the model are shown in Table 1. Fig. 4 shows qualitatively the distribution of the bending moment, shear force, and axial force in the CGMRF under lateral load. The shear and axial force in the beams are less significant to the response of the beams as pared with the bending moment, although they must be considered in design. The qualita tive distribution of internal forces illustrated in Fig. 5 is fundamentally the same for both elastic and inelastic ranges of behavior. The specific values of the internal forces will change as elements of the frame yield and internal for ces are redistributed. The basic patterns illustrated in Fig. 5, however, remain the same. Inelastic static pushover analysis was carried out by applying monotonically increasing lateral displacements, at the top of both columns, as shown in Fig. 6. After the four RBS have yielded simultaneously, a uniform yielding in the web and at the ends of the flanges of the vertical link will form. This is the yield mechanism for the frame , with plastic hinges also forming at the base of the columns if they are fixed. The base shear versus drift angle of the model is shown in Fig. 7 . The sequence of inelastic activity in the frame is shown on th