論文ID: 17-00451
In mesh-based fluid-structure interaction (FSI) analysis, mesh moving techniques are essential for preventing mesh distortion and improving the robustness of FSI analysis. In our previous study, we have proposed a new mesh moving technique with minimum-height-based stiffening, which is more robust against the mesh distortion compared to existing techniques. Although this technique provides reasonable distribution of stiffness in space depending on minimum height of each mesh, there is a parameter that determines relationship between the minimum height and the stiffness. Optimal value of this parameter varies depending on problem. However, it is empirically determined in usual. In this study, we develop optimization approaches for the parameter both in space and time directions adopting three optimization methods: static optimization, global optimization and local optimization. Benchmark problems were analyzed for quantitative evaluation of these three approaches. The static optimization provides the best parameter in the minimum-height-based stiffening technique without optimization in time direction. The results with the global optimization contain the best in all results in this study. However, this approach requires unacceptable computational cost from a practical point of view. The local optimization has a good balance between computational cost and robustness since it can achieve comparable to or more robustness against the mesh distortion with much chaper cost compared with the static optimization. This approaches are expected to be applicable to FSI analyses, and improve their accuracy and stability.
