Axial Compression Behavior of Thin-Walled Steel Tube Concrete Long Columns

Literature Overview

This 2005 paper by Zhang Yaochun, Xu Hui, and Cao Baozhu from Harbin Institute of Technology, published in Building Structures, presents axial compression tests on eight thin-walled square and octagonal CFT long columns. The study extends previous research on short columns to long columns, addressing the stability behavior of thin-walled CFT members under combined flexural and local buckling effects. The column slenderness ratios (L/b) ranged from 14 to 40, and the plate width-to-thickness ratios (b/t) ranged from 6.7 to 12.5.

Experimental Configuration

The octagonal cross-section provides an interesting comparison to the square section because it has a higher polar moment of inertia relative to its perimeter, which affects the flexural buckling resistance. The thin-walled condition (b/t up to 12.5) means that local buckling of the steel tube plates is expected to occur before the column reaches its material strength limit.

Key Findings and Design Implications

The most important finding is the distinction in failure modes between square and octagonal columns:

• Square thin-walled CFT long columns fail by flexural (global) buckling, where the column bows laterally under axial load. The local buckling of individual plates occurs before global buckling but does not govern the ultimate failure mode.
• Octagonal thin-walled CFT long columns fail by strength (material) failure, meaning the column reaches its material capacity before global buckling occurs. Despite local buckling of the steel tube plates, the octagonal geometry provides sufficient flexural stability to allow the column to develop its full strength.

In both cases, local buckling of the steel tube plates was observed prior to ultimate failure. This is expected for thin-walled sections where the plate b/t ratio exceeds the elastic local buckling limit. However, the composite action with the concrete core provides confinement that prevents the local buckling from being immediately unstable.

The bearing capacity decreases significantly with increasing slenderness ratio, consistent with classical column theory. Based on the test results for thin-walled CFT short columns, a linear regression was performed to determine the ultimate bearing capacity reduction coefficient for square CFT long columns within the tested slenderness range.

Engineering Practice Recommendations

For the design of thin-walled CFT long columns:

1. Section selection: Octagonal sections are preferable to square sections for long columns because they provide higher flexural stability and can develop their full material strength. The additional material cost of octagonal tubes is often justified by the improved structural efficiency.
2. Slenderness limits: The significant capacity reduction with slenderness ratio suggests that design codes should impose practical slenderness limits for thin-walled CFT columns. Beyond certain L/b ratios, the stability penalty becomes economically unjustifiable.
3. Local buckling considerations: Although local buckling occurs in all specimens, it does not immediately lead to failure due to the concrete confinement. Design should account for the reduced stiffness after local buckling but can rely on the composite action for ultimate strength.
4. Capacity reduction factors: The linear regression of the bearing capacity reduction coefficient provides a practical design tool for engineers. However, this should be validated against a broader database before being adopted in formal design codes.

Study Reflections

This paper makes an important contribution to the understanding of CFT column behavior in the long-column regime, which is less well-documented than short-column behavior. The finding that octagonal sections outperform square sections in long-column applications is particularly valuable for practical design, as it suggests a clear section optimization strategy. The observation that local buckling precedes but does not govern failure in CFT columns is consistent with the general understanding that composite action provides post-buckling reserve strength. However, engineers should be cautious about extrapolating these results to even thinner walls or higher slenderness ratios, as the confinement effectiveness may diminish. The study's relatively small specimen count (eight total) limits the statistical confidence in the regression results, but the mechanistic understanding is sound and provides a solid basis for further research and code development.