2026 年 21 巻 1 号 p. JFST0007
This study proposes a novel Mass-Preserving Particle Level Set method with Kernel function correction (mpPLS/KC) method that achieves both mass conservation and interface shape reconstruction in multiphase flow simulations. While the conventional PLS/KC method exhibits high shape reconstruction accuracy, it suffers from mass conservation issues. Therefore, a global mass conservation process was added to address this limitation. The proposed method converts minute mass errors occurring throughout the computational domain into offset values for the level set function, performing mass correction while maintaining the distance function properties. Through this efficient algorithm, mass conservation errors were successfully reduced to below 1.0×10-14 (machine epsilon error level) in all benchmark tests. This represents a significant improvement compared to the mass errors of the conventional PLS/KC method (10-3 to 10-6 order). Regarding interface shape reconstruction, the method demonstrated high-accuracy tracking of complex interface shapes involving large deformations in Zalesak's three-dimensional rigid body rotation test and LeVeque's three-dimensional deformation test. The computational time increase due to the addition of the mass conservation process was approximately 4-6%, which is within a practically acceptable range. For practical validation, the proposed method was applied to three engineering problems: dam-break flow, dam-break flow with obstacles, and tsunami flow through coastal forests. In all cases, the results showed good agreement with experimental data. This study demonstrates that simultaneous achievement of extremely high mass conservation and excellent shape reconstruction, which was previously difficult, is now possible. The method is expected to advance as a practical multiphase flow analysis tool for various engineering fields including ocean engineering and fluid machinery.
