Parameter sensitivity analysis and evaluation method of axial compression bearing capacity of corroded circular steel tubes

Abstract

Corrosion-induced durability problems of spatial structures are coming to the fore. However, the understanding of the effect of local corrosion on the bearing capacity of components such as steel tubes, remains a challenge due to the large number of parameters involved. In this paper, a simulation method for determining the axial compression bearing capacity of corroded circular steel tubes (CCSTs) was introduced and calibrated by experimental data. After simplifying the shape of corrosion, 3,000 models with different parameters were produced. Based on the calculated results, Sobol sensitivity analysis and the maximum information coefficient method were used to analyze the sensitivity of different parameters. Subsequently, the relationship between the sensitive parameters and the corrosion degradation factor was analyzed. After comparison with Chinese and American codes, fitting equations and machine learning were used to obtain the degradation factor and discuss how the effects of corrosion should be studied on the basis of the above codes. The results demonstrated that both methods can adequately predict the bearing capacity of CCSTs. Lastly, the two approaches were compared and their application scenarios were discussed.