Container ships are widely used and their strength has always been the concern of designers and engineers because of the large openings in deck. Additionally, container ships are of similar arrangements and structures because the containers are of same standards worldwide. So investigation into the strength of typical structures is meaningful. This paper focuses on the ultimate strength of typical bottom structures in container ships under both longitudinal and transverse loads in corrosive environments. On one hand, the ultimate strength of container ship's bottom structures is calculated by Nonlinear Finite Element Analysis (NFEA), and the limit state of failure (i.e. interaction relation of bi-axial loads) is derived through Minimum Square Error (MSE) technique based on a series of NFEAs; On the other hand, uniform and not simultaneous corrosion across different structures is studied, and surrogate models by Gaussian Process (GP) are built for both longitudinal thrust and transverse thrust ultimate strengths individually, and the corresponding longitudinal and transverse ultimate strength probabilistic characteristics of the typical bottom structures are investigated. Finally, corrosion effects on interaction between longitudinal and transverse stresses at ultimate state are studied.
Corrosion takes place and can be a significant source of ship structural strength degradation. Many researchers have devoted in incorporating corrosion growth models regarding to available statistics, e.g. several researchers supposed corrosion follows a linear decrease of plate thickness with time (Sun and Bai, 2003; Wirsching, Ferensic and Thayamballi, 1997; Guedes Soares and Garbatov, 1996); Melchers (1999) extended original linear and bilinear models (Southwell, Bultman and Hummer, 1979) and introduced second statistical moment; Yamamoto and Ikegami (1998) studied the degradation of coating and corrosion of ship's hull, and a consistent corrosion model is proposed and verified; Soares and Garbatov (1999) proposed a nonlinear corrosion model which assumes corrosion doesn't happen until coating life is over, and the probability density function (PDF) can be determined by adopting Weibull or Log-normal distribution assumptions (Garbatov, Guedes Soares and Wang, 2007; Guo, Wang, Ivanov and Perakis, 2008).