Four full-scale exterior beam-to-column connections comprising H-section beams and square hollow-section (SHS) columns were tested under low-cycle reversed loading to investigate two engineering-oriented T-stub section configurations and the effects of the presence or absence of triangular stiffeners. Failure modes, moment–rotation response, stiffness degradation, energy dissipation, and cumulative damage were evaluated, together with nonlinear finite element simulations and a modified Park–Ang damage assessment. All specimens progressed from bolt-hole slip through plastic deformation to localized fracture. In the unstiffened connections, damage concentrated near the T-stub flange-to-web junction; stiffeners redistributed critical demand toward the stiffener welds, adjacent T-stub webs, and SHS column walls. The maximum differences in initial rotational stiffness relative to J1A were 15.69% and 15.07% in the positive and negative loading directions, indicating that the elastic-stage response reflected the combined deformability of the T-stub, blind-bolt assembly, and column wall. The maximum increases in yield moment, peak-resistance moment, and ductility coefficient were 20.59%, 43.71%, and 45.91%, respectively. Complete-history energy dissipation varied non-monotonically across the tested configurations. The finite element model reproduced the global and local responses, while the damage-index results showed overall correspondence with the observed failure progression. The findings emphasize stiffness compatibility and rational distribution of plastic demand rather than maximum local stiffness.

To investigate the effects of end-plate thickness on the seismic performance and damage evolution of K-shaped eccentrically braced frames with end-plate connections (EPEBFs), two scaled specimens with different end-plate thicknesses were tested under low-cycle reversed loading. The influence of end-plate thickness on mechanical behavior and seismic response was evaluated from failure modes, hysteretic response, skeleton curves, ductility, energy dissipation, and stiffness degradation, together with damage assessment using different models. Refined finite element models were then established for parametric analysis. The results show that EPEBFs exhibit a well-defined plastic development path. Damage in the reference specimen was mainly concentrated in the link and its end-plate connection region, thereby protecting the frame columns, beams, and braces. For the two thicknesses tested, the specimen with thicker end plates showed higher lateral resistance, greater ultimate deformation capacity, and better late-stage resistance retention, but lower displacement ductility, indicating enhanced absolute deformation capacity but reduced post-yield deformation reserve. Thicker end plates also altered stress transfer and plastic development in the connection regions, causing damage to extend toward the frame-beam ends. Among the damage models considered, the elastic–plastic energy dissipation ratio model better captured the accumulation of energy dissipation and plastic deformation. The finite element results agreed well with the tests. Within the investigated range, increasing end-plate thickness mainly improved ultimate deformation capacity and late-stage resistance retention, without a clear monotonic effect on lateral resistance. The findings provide a basis for prefabricated design, damage control, and post-earthquake replacement and repair of EPEBFs.