Identifying critical locations for connection ductility in steel moment resisting frames

The investigations following the unacceptable performance of moment resisting frames (MRF) in the 1994 Northridge Earthquake led to the development of a variety of alternative connections. Tests have shown that these connections have a reliable component-level performance, leading to their prequalification in US standards. Post-Northridge investigations additionally demonstrated the variation of local connection performance within individual frames. The majority of frames investigated had less than 50% of their connections fractured, and the locations of fractured connections were often grouped at specific floor levels, indicating an unequal connection deformation demand over the height of each frame. However, the current design practice consists of applying a single type of prequalified connection, often the reduced beam section (RBS), uniformly throughout an entire MRF. These connections are installed regardless of local deformation demands and at a considerable overall cost to ensure that each connection has an adequate rotational capacity. This traditional design procedure applies a uniform ductility capacity to a non-uniform displacement demand and limits the opportunity to concentrate resources at locations where the local connection performance has the greatest consequence on the global MRF seismic performance. This paper examines the use of non-linear static methods to attempt to identify floor levels where connections undergo large inelastic rotational demands under dynamic loading. A collapse analysis of a 6-storey MRF was conducted to identify the locations of highest displacement demand. In an effort to decrease the time required for analyses while achieving the same relative floor rotation demands, various pushover analyses, which are less computationally intensive than full collapse analyses, were completed on the original RBS frame. Finally, the accuracy of these less computationally intensive analysis methods is evaluated by comparing the connection rotations obtained using both analysis types.