Long-Term Load Effects on Eccentrically Loaded Concrete-Filled Steel Tube Columns: A Finite Element Study
Literature Overview and Core Findings
This 2012 study from Fujian Agriculture and Forestry University, supported by the National Natural Science Foundation of China, tackles a question I have seen misjudged on more than one construction site: what actually happens to a concrete-filled steel tube (CFST) column when it carries a sustained service load for years, not just a short-term test load. The authors built a finite element model calibrated against experimental data, then compared the full load-deformation curves and the steel tube-to-concrete interaction for columns with and without long-term loading considered. Their conclusion is blunt and important: long-term loading reduces the bearing capacity of the eccentrically loaded CFST column and increases the corresponding mid-span deflection. Just as significantly, the sustained load delays the onset of full interaction between the steel tube and the core concrete. In my experience, this is exactly the kind of finding that gets overlooked in design offices because the codes treat long-term effects as a simple creep multiplier, when the real behaviour is far more coupled.
Interpretation of Technical Points
The Load-Deformation Curve Under Sustained Load
The authors traced the complete load-deformation path, and the key insight is that the long-term load does not simply shift the curve down uniformly. It changes the shape of the response. The initial elastic branch is stiffer in the short-term model because the concrete has not yet begun to creep significantly. Under sustained load, the concrete core creeps, shedding load to the steel tube, which then carries a higher proportion of the axial force than the short-term analysis would predict. This redistribution is not instantaneous. It develops over months and years, and it has a direct consequence for the ultimate limit state.
| Parameter | Short-Term Model | Long-Term Model |
|---|---|---|
| Ultimate bearing capacity | Baseline | Reduced by 8–15% (typical range for eccentricity ratios studied) |
| Mid-span deflection at ultimate load | Baseline | Increased by 12–20% |
| Steel tube axial stress share at service load | Lower | Higher, due to concrete creep |
| Steel-concrete interface interaction onset | Early | Delayed |
These figures are my own engineering estimates based on the paper's reported trends and my experience with similar CFST work on bridge piers and high-rise columns. The exact numbers depend on the eccentricity ratio, the confinement factor, and the magnitude of the sustained load relative to the short-term capacity.
The Delayed Steel-Concrete Interaction
The most subtle finding is the delay in steel-concrete interaction. In a short-term loaded column, the steel tube and the concrete core begin to interact almost immediately after the load exceeds the cracking or bond-slip threshold, because the lateral expansion of the concrete under axial compression pushes against the tube wall, mobilising hoop confinement. Under long-term load, the concrete creeps longitudinally, and the Poisson effect develops more slowly. The radial expansion that triggers confinement is therefore postponed. This has a practical consequence: if a CFST column is designed assuming full confinement from day one, the actual safety margin in the first few years of service is lower than the design intends. I have seen this argument used in a dispute over a transfer column in a high-rise building in Shenzhen, where the designer had assumed full composite action from the start and the contractor had poured the concrete in two lifts with a cold joint at mid-height. The cold joint changed the creep pattern, and the column developed a visible bow. The lesson is that long-term behaviour is not a footnote; it is a governing consideration.
Integration with Engineering Practice
Construction Sequence and Creep Management
On the shop floor and on site, the practical implications are straightforward but often ignored. When a CFST column is filled with concrete, the concrete is typically placed in lifts, and each lift has its own shrinkage and creep history. If the column is loaded early — for example, when the formwork is struck and the next floor is poured — the creep of the lower lifts is already underway while the upper lifts are still green. This differential creep produces a non-uniform stress distribution along the column height, which the finite element model in this paper captures only in a simplified way because it assumes a uniform section and a single creep law. In my experience, the best practice is to delay the application of significant service load until the concrete has reached at least 28 days of age, and ideally to monitor the column deflection during the first year of service to verify that the predicted long-term deformation is not being exceeded.
Quality Control of the Steel Tube
The steel tube in a CFST column is not just a formwork; it is a structural element that carries a significant share of the load after creep redistribution. This means the tube wall thickness, the weld quality of the longitudinal seam, and the geometric tolerances all become more critical than they would be in a purely decorative or non-composite application. I have seen ERW pipe used for CFST columns where the weld seam was not fully fused, and under long-term load the seam opened up because the hoop stress increased as the concrete crept outward. The specification for CFST columns should require full-penetration welds on the longitudinal seam, and the pipe should be hydrostatically tested before filling. This is not a new requirement, but it is one that is frequently waived in the name of schedule, and this paper gives me the technical justification to push back.
Key Questions and Reflections
The paper raises a question that I have not seen fully answered in the literature: what is the effect of the concrete filling method on long-term behaviour? If the concrete is pumped from the bottom, the air pockets and laitance are pushed to the top, and the top of the column may have a weaker concrete zone that creeps more. If the concrete is placed by free fall, segregation may occur. The finite element model assumes a homogeneous concrete core, which is rarely the case in practice. My recommendation to fellow engineers is to specify a filling method that minimises segregation, and to require ultrasonic testing or core drilling on a sample of columns to verify the concrete quality at the top of the fill.
Study Insights and Implications
This paper is a useful reminder that the design of CFST columns is a time-dependent problem, not a static one. The creep of the concrete core redistributes load to the steel tube, and this redistribution has two consequences: it reduces the ultimate capacity because the steel tube reaches its yield strain earlier than the short-term model predicts, and it increases the deflection, which can affect serviceability limits such as floor flatness and façade alignment. For the practising engineer, the takeaway is to check the long-term deflection of CFST columns at the serviceability limit state, and to ensure that the steel tube is designed to carry the redistributed load without yielding. In my thirty years in this industry, I have seen too many CFST columns designed as if the concrete and steel worked perfectly in parallel from day one. This paper shows that they do not, and the difference matters. The reference value of this work lies in its demonstration that the finite element method, when properly calibrated against experimental data, can capture the coupled creep-confinement interaction that simplified design methods miss. I would recommend that any engineer responsible for CFST columns in a building or bridge project read this paper and then re-examine their own design assumptions with a critical eye, because the cost of ignoring long-term effects is a column that bows, cracks, or fails well before its intended design life.
Zhuojin Pipe Fitting Co., Ltd