Shear capacity evaluation and buckling mechanisms of composite I-girders with corrugated steel webs

Corrugated steel webs provide high shear stability, transverse rigidity, and construction efficiency, motivating the development of steel-concrete composite I-girders in which the web is connected only to a concrete slab at the top. Without a bottom concrete slab, the web experiences asymmetric restraints, rendering the double-sided fixed boundaries assumed in existing shear buckling models and design codes incompatible with this system. Consequently, accurately evaluating the shear buckling strength of corrugated-web I-girders is essential for safe and economical design. In addition, current shear capacity formulations for this structure rely on converting bending–shear interaction relationships, which complicates practical engineering use. This study conducts comparative experiments on composite beams with corrugated and flat steel webs, develops validated nonlinear finite element models, and performs parametric studies on shear span ratio and shear connection degree. Based on the asymmetric restraint configuration, a new top-fixed and bottom-simply supported boundary model is proposed, from which modified formulas for local, global, and interactive shear buckling are derived. A direct shear capacity model accounting for deck contribution and bending shear interaction is further established. Results show that the corrugated web provides a shear buckling resistance approximately 4.5 times that of a flat web, while the concrete deck contributes 20-30% of the total shear force. The proposed shear capacity equation achieves an average prediction ratio of 0.99 across 25 beams, providing a more accurate, user-friendly design tool for composite I-girders with corrugated steel webs. • A new boundary model (top-fixed and bottom-simply supported) is proposed for corrugated-web I-girders. • Modified formulas for local, global, and interactive shear buckling are developed and validated. • A direct shear capacity model accounting for deck contribution and bending-shear interaction is established. • The proposed models show high accuracy against 25 test beams, with an average prediction ratio of 0.99.