Numerical Study of Mechanical Behavior of Concrete-Filled Steel Tube Members

Literature Overview

The paper by Zhang Bo, Li Shucai, Yang Xueying, Li Mingtian, and Sun Guofu, published in Journal of Shandong University (Engineering Science) (2010, Vol. 40, No. 6, pp. 76-81), presents a numerical study on the mechanical performance of concrete-filled steel tube (CFST) members. Supported by the National Natural Science Foundation of China (Grants 40872203 and 40902083) and Shandong University Innovation Fund, the research derives stress-strain relationships suitable for beam elements using two approaches: three-dimensional finite element method and fiber element method. These derived relationships are then applied in beam element numerical analysis of CFST members under axial compression and eccentric compression.

Core Technical Approach

The study addresses a fundamental challenge in CFST structural analysis: the complexity of the concrete-steel interaction, which makes conventional beam element analysis insufficient for capturing the true mechanical behavior. The authors propose two pathways to derive effective stress-strain relationships for beam elements:

1. Three-dimensional finite element approach: A full 3D finite element model of the CFST member is constructed, and the stress-strain response at a representative section is extracted to form a beam-level constitutive relationship.
2. Fiber element approach: The cross-section is discretized into fibers, each representing a small area of concrete or steel, and the stress-strain relationship for each fiber is defined based on the confined concrete and steel material models. The section response is then obtained by integrating the fiber responses.

The comparison with experimental results reveals a significant finding: the fiber element method produces accurate results for both circular and square cross-sections, while the 3D finite element-derived beam relationship performs well only for square sections and shows substantial deviation for circular sections. This discrepancy is attributed to the anisotropic confinement effect in circular sections, where the steel tube provides uniform radial confinement to the concrete core, whereas square sections have non-uniform confinement with weaker confinement at the corners.

Material Modeling and Confinement Effects

The accuracy of CFST numerical analysis depends critically on the material models used for both the steel tube and the confined concrete. The steel tube is typically modeled with an elastic-perfectly plastic or bilinear model, while the confined concrete requires a model that captures the increased strength and ductility under triaxial confinement. The confinement pressure provided by the steel tube is a function of the hoop stress, which in turn depends on the lateral expansion of the concrete under axial load. For circular sections, the confinement pressure is uniform around the perimeter, leading to a well-defined confinement model. For square and rectangular sections, the confinement pressure varies along the perimeter, with lower values at the corners where the steel tube is less effective at restraining lateral expansion.

The fiber element method naturally captures this variation because each fiber can be assigned a different confinement pressure based on its position in the cross-section. The 3D finite element approach, while more physically detailed, introduces discretization errors and mesh sensitivity that can degrade the accuracy of the extracted beam-level response, particularly for circular sections where the smooth curvature requires fine meshing.

Engineering Practice Implications

For structural engineers designing CFST columns and beams, this study provides clear guidance on the selection of numerical analysis methods. The fiber element method should be the preferred approach for routine analysis of CFST members, as it offers the best balance of accuracy and computational efficiency for all common section shapes. The 3D finite element method is more appropriate for detailed local analysis, such as studying the behavior at connections, local buckling of the steel tube, or the effect of concrete cover thickness. When using the 3D FEM-derived beam relationship for circular sections, engineers should be aware of the potential for significant underestimation of capacity and should validate results against experimental data or use the fiber element method instead.

In practical design, the stress-strain relationship derived from the fiber element method can be directly implemented in structural analysis software through user-defined material models. This enables nonlinear analysis of CFST frames and trusses, capturing the progressive yielding and concrete crushing that govern the ultimate behavior. The method also facilitates parametric studies to optimize the section dimensions and material grades for a given loading condition.

Study Reflections

The research highlights an important principle in computational structural engineering: the level of modeling detail must be matched to the problem being solved. A highly detailed 3D model is not always superior to a well-calibrated beam element model, especially when the extraction of section-level behavior from the 3D model introduces additional sources of error. The fiber element method, though conceptually simpler, proves more robust because it directly represents the physical mechanisms of concrete confinement and steel yielding without the intermediate step of 3D-to-beam reduction. This insight has broader applicability in structural analysis, where the temptation to use increasingly complex models does not always translate into improved engineering accuracy.