Experimental and analytical investigation of alternate block shear failure of beams in steel moment connections.

Abstract

Alternate block shear (ABS) failure in beams needs to be addressed when designing bolted flange plate (BFP) and double Tee moment connections associated with deep beams in seismic areas. This study investigates experimentally and analytically the ABS failure in such moment connections when subjected to monotonic tensile loading. ABS, not yet included in the AISC 358 specifications, is a failure mode that combines full tensile fracture in the beam flange with shear failure in the beam web. The design of BFP and double Tee moment connections requires deep beams and thick plate connections to sustain high moment demands. ABS failure might be considered as one of the potential failure modes in such conditions. Therefore, investigating such failure is necessary for the inclusion in the provisions. To address this issue, FE models are conducted to validate the experimental results of structural Tee connections available in the literature. The same FE technique is conducted on a series of prequalified BFP and double Tee moment connections to examine the ABS failure. The dimensionless ratio of connection length to beam depth ratio is considered the major parameter for identifying the governing failure mode. Then, an experimental program is conducted to examine the ABS failure in the beams associated with thick plate connections. The results of a series of four specimens showed that the ABS failure path is a combination of both yielding and rupture mechanisms leading to a ductile failure in the beam. Moreover, the experimental and FE results conducted in this study are compared with existing strength models to investigate their prediction capability and accuracy while designing moment connections. Results indicate that the ABS failure have a lower capacity than the block shear failure which is included in the ANSI-AISC 358-16 design provisions of the BFP and double Tee connections. FE simulations are developed to predict the experimental results. The FE models can predict with acceptable accuracy the experimental results