Stochastic Sensitivity Analysis of Creep Effect in Steel Pipe Concrete Columns

Literature Overview

The paper by Zhao Jingang, Zhao Rendda, and Zhan Yulin, published in the Journal of Architecture and Civil Engineering (Vol. 33, No. 4, 2016, pp. 44-50), addresses the stochastic nature of creep behavior in steel pipe concrete (SRC) columns. The authors employ Support Vector Machine (SVM) regression to fit explicit functions for the creep effect, combined with Monte Carlo simulation for probabilistic analysis. An adaptive hybrid particle swarm optimization algorithm is used to optimize the SVM parameters for computational efficiency. The study analyzes two SRC creep model specimens and validates the proposed method against direct Monte Carlo simulation results.

Core Technical Points

Creep in steel pipe concrete columns is a time-dependent deformation mechanism that affects long-term structural performance. The paper recognizes that creep is inherently stochastic due to the variability in concrete mix design, curing conditions, loading history, and environmental factors. The authors identify the following key random variables influencing the creep effect:

• Concrete compressive strength at loading age
• Concrete modulus of elasticity
• Self-weight of the column
• Applied axial load
• Temperature and humidity conditions
• Steel pipe material properties

The SVM-based approach provides an efficient surrogate model that approximates the complex nonlinear relationship between input random variables and the creep response. The sensitivity coefficients quantify the relative influence of each random variable on the creep effect, enabling engineers to identify the most critical parameters for quality control.

Sensitivity Analysis Results

The study finds that the probability density function of the creep effect is approximately normal, which simplifies the reliability analysis. The relative error between the SVM-Monte Carlo method and the direct Monte Carlo method is small, validating the efficiency and accuracy of the proposed approach.

Relevance to Steel Pipe Manufacturing and Welding

From a steel pipe engineering perspective, the creep behavior of SRC columns is influenced by several factors related to the steel pipe itself. The steel pipe acts as a permanent formwork and a structural reinforcement, and its properties directly affect the long-term performance of the composite column.

Weld Quality and Creep

The quality of the longitudinal weld in the steel pipe has a significant but often overlooked impact on creep behavior. Welding introduces residual stresses in the HAZ and weld metal, which can alter the stress distribution in the steel pipe under sustained loading. If the residual tensile stresses in the HAZ are not adequately relieved, they can accelerate creep deformation by effectively increasing the mean stress in the pipe wall.

The following welding-related considerations are critical for SRC columns:

• Post-weld heat treatment (PWHT) should be performed to relieve residual stresses in the pipe weld, particularly for thick-walled pipes used in large-diameter SRC columns.
• The weld procedure must comply with applicable standards (e.g., GB/T 150, ASME Section VIII) to ensure full penetration and minimize porosity.
• The HAZ microstructure should be examined metallographically to confirm that no brittle phases (such as martensite) are present, which could crack under sustained loading.

Material Selection for Creep Resistance

The steel grade selection for SRC columns should consider the long-term creep behavior. Carbon structural steels (Q235, Q345 per GB/T 1591) are commonly used, but for high-temperature or high-load applications, low-alloy steels with better creep resistance may be required. The following material properties are relevant:

Engineering Practice Integration

The stochastic sensitivity analysis methodology presented in this paper can be directly applied to quality control planning for SRC column steel pipes. By identifying the most sensitive parameters, engineers can prioritize inspection and testing resources:

1. For pipes where concrete strength at loading age is the most sensitive variable, the construction schedule must ensure that the concrete reaches sufficient strength before the column is loaded.
2. For pipes where the steel pipe thickness is a significant variable, the wall thickness tolerance must be tightly controlled during manufacturing (typically ±10% per GB/T 8163).
3. For pipes where temperature is a critical factor, the construction environment must be monitored and, if necessary, temperature-controlled during the concrete curing period.

The Monte Carlo simulation approach also provides a framework for risk assessment. Engineers can calculate the probability of exceeding a specified creep limit, which is essential for serviceability limit state design. The sensitivity coefficients guide the allocation of inspection effort, ensuring that the most influential parameters receive the most rigorous quality control.

Summary

This paper presents a sophisticated stochastic analysis framework for evaluating creep in steel pipe concrete columns. The SVM-based surrogate modeling approach offers significant computational efficiency while maintaining accuracy, making it practical for engineering applications. The sensitivity analysis results highlight that concrete properties and loading conditions are the dominant factors, while steel pipe properties have a secondary but still meaningful influence. For steel pipe engineers, the key implications are: ensuring weld quality to minimize residual stress effects on creep, selecting appropriate steel grades with adequate creep resistance, and controlling wall thickness tolerances. The probabilistic approach also provides a rational basis for allocating quality control resources, focusing inspection efforts on the most critical parameters that influence long-term structural performance.