Service Life Evaluation of Marine Concrete Structures Considering Spatial and Temporal Characteristics: A Framework Based on Multi Training-Mcs-nls
Shiqi Wang,Renjie Wu,Fuyuan Gong,Jin Xia,Yuxi Zhao,Bin Zeng
DOI: https://doi.org/10.1016/j.engstruct.2024.119193
IF: 5.582
2025-01-01
Engineering Structures
Abstract:Corrosion of steel caused by chloride corrosion is the main factor leading to the service life of marine structures shorter than the design life. However, this failure mechanism is affected by the coupling of multiple factors such as material composition, environmental parameters and service time, which cannot be adequately revealed by traditional models. Based on this, this paper proposed a framework to assess the service life of marine concrete structure. Firstly, a database containing 711 sets of concrete surface chloride ion concentrations and diffusion coefficients in different exposure zones (splash, tidal and submerged zone) was constructed. Tree models were used to establish multi-training models to predict the above indicators. The improved service life evaluation model was established by considering multi-parameters. The influence of uncertain parameters on structural failure probability was quantified by MCS. Furthermore, the NLS-LM model was used to optimize the cover thickness under multi-parameter combinations. The results show that random forest exhibits the highest prediction accuracy, with mean prediction values of 1.07,1.11 and 1.13 in splash, tidal and submerged zone, respectively. Evaluation model based on multi training and MCS can effectively consider the influence of spatial and temporal characteristics, material composition on the service life of marine structure. The probability distribution and combinations of uncertain parameters will significantly affect the failure probability. In addition, the optimization model based on NLS-LM can effectively guide the design of cover thickness under any marine environment.