Random field modelling of mechanical behaviour of corroded thin steel plate specimens

Highlights

• Corrosion degradation is modelled with the use of random field approach.

• Nonlinear FE Analysis with the use of an explicit dynamic solver is performed.

• A fully nonlinear material model is adopted in order to obtain the proper response.

• Sensitivity analysis considering main descriptors of the random field is performed.

• The results of the analysis are validated with the available experimental data.

• Irregularities in the corroded surface are main reason for the strength decrease.

Abstract

The objective of this work is to explore the possibility of corrosion degradation modelling of thin steel plate specimens with the use of random field approach. The mechanical properties are obtained via the nonlinear Finite Element Analysis with the use of an explicit dynamic solver. The fully nonlinear material model is adopted to obtain the proper stress-strain response. Sensitivity analysis considering the main statistical descriptors of the random field is performed. The results of the analysis are validated with the available experimental data showing a good agreement for lower levels of Degree of Degradation and significant deviations for severely corroded specimens. The analysis shows that the irregularities in the corroded plate surface are one of the main reason for the mechanical properties reduction. Random field modelling revealed to be a swift and practical tool for representing the corroded surfaces in steel structures.

Introduction

Many steel structures operate in different marine environmental conditions subjected to corrosion degradation [1]. In the last decades, various models of thickness reduction due to corrosion were developed [1], [2], [3]. Recently attention was paid to the fact that not only thickness reduction concerning corrosion is observed, but also mechanical properties changes are observed as well [4] leading to severe strength reduction [5], [6], [7], [8], [9].

The discussion about the reduction of mechanical properties of steel specimens with the subsequent corrosion degradation is vital. During the last couple of years, several studies have been done, mainly experimentally, but also in the numerical domain. Garbatov et al. [4] tested corroded plate specimens with different corrosion severity. They concluded that for the specimens with degradation level higher than 20%, the mechanical properties reduction is significant. Not only yield strength, and the Young modulus is reduced, but also the elongation is significantly reduced. Kashani et al. [10] analysed the stress-strain behaviour of non-uniformly corroded steel circular bars. In this case, the degree of degradation was up to 20%. The reduction of yield strength was up to 30% and the reduction of the mean strain at failure was even up to 80%. Zhan et al. [11] tested 267 corroded steel bars that were corroded in different conditions (in a laboratory and natural environment). They concluded that the mechanical properties are significantly deteriorated, and the yield plateau of the steel bar became shorter or even disappeared with the development of the corrosion. In the study performed by Fernandez and Berrocal [9], 30-years old, naturally corroded steel reinforcing bars where examined. Beside the tensile tests, they measured the corrosion distribution along the bars. They concluded that the reduction of the mechanical properties is governed by the minimal cross-section area of the specimen. Additionally, Li et al. [12], [13] performed the tensile test of pre-stressed and corroded specimens, and the effect of the mechanical properties reduction was magnified with comparison to only corroded specimens.

In case of the ship and offshore structures, thin-walled structures such as plates and slender beam are more often used. The mechanical properties of corroded flat specimens were investigated in [14], [15], [16], [17], [18], and the conclusions were similar to those obtained for bars. The reduction was observed not only in severe corrosion conditions but also in lower corrosion degradation. In the case of very thin plates, such as one millimetre of thickness, the reduction of the mechanical properties was up to 70% from their initial values [15].

Most researchers agreed that if the corrosion is uniform, it would not have a significant effect on the mean stress-strain response. Thus induce, that the reduction of mechanical properties is caused by the non-uniform distribution of corrosion pits [19], [20], [21], [22]. This hypothesis is supported by experimental results of Garbatov et al. [23], where specimens after cleaning showed higher values of mechanical properties compared to corroded non-cleaned ones. In the case of the marine environment and especially for ships structures, the non-uniform corrosion is the most common one, and this phenomenon cannot be neglected.

Based on the experimental results, several mathematical models for the prediction of mechanical properties of corroded bars and flat specimens were developed [24], [25]. Li et al. [26] developed a simplified constitutive model for corroded bars based on experimental and numerical studies. Garbatov et al. [4] developed a model for corroded coupon specimens. Other mathematical models for estimation of different mechanical properties of flat bars that are based on experimental investigations can be found in [27], [28], [29], [30]. Although couple models were developed, it seems that more work is still needed towards in developing a more unified approach with regards to mechanical properties change in terms of corrosion degradation.

With the development of more advanced measuring techniques, the detailed measurements of 3D corrosion morphology are possible. Examples of such measurements can be found in [31], [32], [33]. Based on that, the mechanical properties of corroded specimens or fatigue properties can be evaluated [33], [34], [35] and even the structural response of beam elements may be analysed [36]. Wang et al. [14] compared some experimental results with the nonlinear FE analysis, where the corrosion surface was scanned and implemented in the finite element model. The agreement between experimental and numerical results was good.

Based on the data collected about mechanical properties changes [37], a numerical model of corroded structures can be developed, such as presented in [38]. In this case, the ultimate strength of corroded stiffened plates was evaluated and compared to the experimental results reported in [39]. However, in this case, the degree of degradation levels was very high (above 40%). Usually, the plates with that levels of corrosion are replaced in real structures. The study presented in the sections explores the mechanical properties change with lower levels of degradation.

Due to the complexity of methods for measurements of different structural imperfections, such as corrosion degradation, initial distortions etc., the random field modelling [40] seems to be a powerful tool for the modelling of such imperfections. The examples of random field modelling to reflect the geometric and material imperfections can be found in [41], [42], [43]. The possibility of stochastic modelling of corrosion field was also presented in [44]. Additionally, the random field approach can provide many samples which are usually hard to obtain, especially for relatively large structures such as ships or offshore platforms. Thus, it can be exploited in the reliability analysis [45].

In the presented study, the random field approach is applied in modelling corroded surfaces to evaluate the mechanical properties of corroded specimens subjected to general corrosion. The nonlinear FE method with the use of the explicit dynamics solver is used. The sensitivity analysis concerning the governing factors of the random field is performed. The results of experimental investigations presented in [15] are used to validate the present study.