Random field modelling of mechanical behaviour of corroded thin steel plate specimens
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.
Section snippets
Corrosion degradation modelling using random field
Due to the natural origin of the corrosion environment, the surface of the corroded plate can be modelled with the use of a random field. The spatial distribution of corrosion pits may be modelled by the random field approach as the one that is the most suitable for this purpose, due to the set of an infinite number of spatially correlated random variables. For engineering purposes, one needs to find a random field with a finite number of random variables, and with this respect, the Gaussian
Finite element modelling
To analyse the mechanical behaviour of the tensile tested steel specimen, ANSYS LS-DYNA [50] software is used. As an input stress-strain relationship of mechanical properties, the multilinear true stress-strain curve is applied, and the specimen is modelled with the use of SOLID164 elements for explicit dynamics. The element fractures, when the strain reaches the failure strain. Nevertheless, the tensile test is quasi-static, the explicit dynamics solver is used to overcome the convergence
Corrosion degradation modelling
The corrosion degradation may be one or two-side developed depending on the environmental conditions and structural component location and position. Since the results of this study are validated with the experimental data of [15], the corrosion is considered as a one-side. In both cases, some corrosion degradation descriptors are distinguished. The approach which is adopted in the present study is presented in Fig. 9, where a typical cross-section of the corroded specimen with a randomly
Sensitivity analysis
To establish the most influencing parameters in the random field generation, the global sensitivity analysis is performed. The analysed random field parameters are presented in Table 2. The ratio between the maximum and mean corrosion depth varies between 2 and 5. By changing the maximum corrosion depth to the mean corrosion depth, the shape of the random field will change. With a higher corrosion depth, the field will be sharper. In this way, the variance of the random field is calibrated to
Finite element analysis of corroded specimens
The corrosion field, modelling with the use of the random approach, is compared to experimental results of very thin plates as reported in [15] (the initial thickness is 1 mm). Seven specimens were cut from a real structure and exposed to atmospheric corrosion conditions. However, only the maximum residual thickness and minimum cross-sectional area are provided as a result of the corrosion morphology analysis. The corrosion is one-side in this case. The dimensions of the gauge area of the
Conclusions
The methodology developed here can be used in predicting the reduction of mechanical properties in terms of corrosion degradation. The developed methodology overestimates the mechanical properties only in the case of higher DoD. However, in operation, the highly corroded plates are usually replaced. The presented study confirms that the irregularities in the surface due to corrosion degradation are the main reason for the mechanical properties reduction. However, in the case of very high Degree
CRediT authorship contribution statement
Krzysztof Woloszyk: Conceptualization, Funding acquisition, Investigation, Methodology, Writing - original draft, Writing - review & editing. Yordan Garbatov: Conceptualization, Funding acquisition, Methodology, Supervision, Writing - original draft, Writing - review & editing.
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgements
This work has been supported by the National Science Centre, Poland (grant No. 2018/31/N/ST8/02380). The ANSYS software used in presented simulations in this paper was available as a part of the partnership cooperation agreement between ANSYS Inc., MESco sp. z o.o. and the Gdansk University of Technology. Part of the calculations was carried out at the Academic Computer Centre in Gdańsk.
References (61)
- et al.
A time-dependent corrosion wastage model for seawater ballast tank structures of ships
Corros Sci
(2004) - et al.
Tensile strength assessment of corroded small scale specimens
Corros Sci
(2014) - et al.
Fatigue strength experiments of corroded small scale steel specimens
Int J Fatigue
(2014) - et al.
Assessment of characteristic values of the ultimate strength of corroded steel plates with initial imperfections
Eng Struct
(2013) - et al.
Ultimate strength analysis of aged steel-plated structures exposed to marine corrosion damage: a review
Corros Sci
(2014) - et al.
Nonlinear stress-strain behaviour of corrosion-damaged reinforcing bars including inelastic buckling
Eng Struct
(2013) - et al.
Predicting the residual strength and deformability of corroded steel plate based on the corrosion morphology
Constr Build Mater
(2017) - et al.
Experimental investigation of mechanical properties of corroded cold-formed steels
J Constr Steel Res
(2019) - et al.
Study on the degradation of mechanical properties of corroded steel plates based on surface topography
J Constr Steel Res
(2016) - et al.
Effect of cleaned corrosion surface topography on mechanical properties of cold-formed thin-walled steel
Constr Build Mater
(2019)