Experimental performance analysis of concrete-filled steel column to concrete-filled steel beam connections

Abstract

Existing literature and practical engineering practice have comprehensively examined the behaviour of concrete-filled steel tubular (CFST) columns, CFST beams, and their associated connection systems involving either steel or reinforced concrete (RC) beams. Despite these advancements, limited research has focused on the direct beam–column interaction in fully CFST-to-CFST connection configurations. The absence of established design specifications and systematic experimental evidence has hindered the reliable adoption of such connections in structural applications. This feasibility study addresses this knowledge gap by conducting an integrated theoretical and experimental investigation into the structural performance of moment-resisting connections between CFST columns—locally strengthened with internal stiffening plates and configured with external bolted flange connections—and CFST beams of matching geometry. To provide a meaningful benchmark, a comparable hollow steel column–steel beam connection with identical cross-sectional dimensions and bolt arrangements was also evaluated. The experimental setup involved cyclic loading tests designed to capture load–rotation behaviour, quantify flexural stiffness, and identify critical limit states governing connection performance. Detailed measurements of moment–displacement response, local deformation patterns, and strain distribution were collected to assess connection rigidity, load-transfer mechanisms, and potential vulnerability to local buckling. The resulting data allowed for direct comparison between the proposed CFST-to-CFST connection configuration and the hollow steel reference specimen, enabling a clearer understanding of the composite action and confinement effects provided by the infilled concrete. The findings contribute foundational evidence for the feasibility of those moment connections and offer preliminary insights to support future analytical modelling, design recommendations, and full-scale implementation.