Comparison of MPS and SPH methods for solving forced motion ship flooding problems

In order to assess the safety of damaged ships quantitatively, advanced numerical simulation methods are required for the prediction of ship dynamic behavior coupled with floodwater. For the prediction of fast and complicated free-surface flows in immediate flooding, particle methods are promising because largely deformed free-surface can be naturally and accurately followed without numerical diffusion. In this paper, systematic comparisons of the MPS (Moving Particle Semi-implicit) and SPH (Smoothed Particle Hydrodynamics) methods are made on ship flooding problems. Firstly, numerical simulations based on the MPS and SPH methods are performed on forced roll tests of a two-dimensional damaged compartment and are compared with a dedicated model experiment. Secondly, similar comparisons are made for a three-dimensional flooded compartment. Through these comparisons between MPS and SPH methods and to model-scale experiments, the capability of particle methods in simulating complicated flooding flows is investigated and the difference of prediction accuracy of the MPS and SPH methods is discussed. For the case of two-dimensional damaged compartment, it is also studied what effects has a single-phase approximation in comparison to simulations where air is modelled, and how the trapped air in damaged compartments influences the water flow evolution. Results from this study show that both particle methods are capable to simulate complex flooding flows in good quantitative agreement with respect to experiments and with similar execution times while there is no possibility to draw superiority of one method. However when dealing with air modeling, using a simplified approach limits computational times but induces quite large discrepancies compared to fully modeling the air phase.

Automated summary

This study evaluates the application of particle methods in simulating complex ship flooding problems, finding that both MPS and SPH methods have similar capabilities in simulating flooding flows and are in good agreement with experimental results. Additionally, the handling of air modeling also affects the results.