Given the situation in the Republic of Korea that all nuclear power plants are located at the seaside, the interim storage facility is also likely to be located at seaside and the maritime transportation of Spent Nuclear Fuel is considered inevitable. The Republic of Korea does not have an independently developed maritime transportation risk assessment code, and no research has been conducted to evaluate the release rate of radionuclides from a submerged transportation cask in the sea. Therefore, there is a need to develop a technology that can assess the impact of immersion accidents and establish a regulatory framework for maritime transportation accidents. The release rate of radionuclides should be calculated from the flow rate through a flow path in the breached containment boundary. According to the cask design criteria, it is anticipated that even under severe accident conditions, the flow path size will be very small. Previous studies have evaluated fluid flow passing through micro-scale channel by integrating internal and external flows within and around a transport cask. As part of the evaluation, a comprehensive “Full-Field Model” incorporating external flow fields and a localized “Local-Field Model” with micro-scale flow paths were constructed. Sub-modeling techniques were employed to couple the flow field calculated by the two models. The aforementioned approach is utilized to conduct the evaluation of fluid flow passing through micro-scale flow paths. This study aims to evaluate fluid flow passing through micro-scale flow paths using the aforementioned CFD (Computational Fluid Dynamics) method and aims to code the findings. The Gaussian Process Regression technique, a machine learning model, is utilized for developing a mathematical metamodel. The selected input parameters for coding are organized and their respective impacts are analyzed. The range of these selected parameters is tailored to suit domestic environments, and computational experiments are planned through Design of Experiments. The flow path size is included as an input parameter in the coded model. In cases where the flow path size becomes extremely small, making it impractical to use CFD techniques for calculations, Poiseuille’s law is employed to calculate the release rate. In this study, a model is developed to evaluate the release rate of radionuclides using CFD and mathematical equations covering the whole possible range of flow path size in a lost cask in the deep sea. The model will be used in the development of a maritime transportation risk assessment code suitable for the situation and environment in Korea.