Investigation of Enhanced Polygon Wall Boundary Model in PNU-MPS Method

2017 
With regard to demonstration of fluid flow, there are two descriptions which are Eulerian description and Lagrangian description. In the field of CFD (Computational Fluid Dynamics), a number of studies relevant to grid method based on Eulerian description have been conducted generally. However, when the grid method is employed to simulate flow field, it is inevitable to give consideration to convection term which generates severe numerical diffusion and fluctuation. To obtain the accuracy of solution, a different type of method based on Lagrangian description is come to the fore. Numerical approaches following Lagrangian description have been called meshfree or particle method. Even though particle method does not accompany convection term and fully satisfies conservation of mass, its studies have not been carried out extensively because it is difficult to implement the boundary conditions correctly due to insufficient number of particles in the vicinity of boundary. It affects directly the stability of flow field and accuracy in computation. In MPS (Moving Particle Semi-implicit) method [1], fixed-type of dummy particles are placed inside wall boundary. By placing extra particles as the wall, it seems to be not easy to satisfy the boundary condition for sharp-edged or extremely thin body configuration. In this study, the enhanced polygon wall boundary model, which was suggested originally by Mitsume et al. [2], is employed to the PNU-MPS (Pusan-National-University-modified MPS) method [3] to improve and stabilize the analysis of fluid flow with arbitrary-shaped body including sharp-edged body configuration without any additional particles. The developed simulation method, called as PNU-MPS-POLY, is adopted to the Couette flow and the lid-driven cavity flow with various corner angles. The present simulation results are validated through comparison with the analytic solutions, the experiments [4], and other simulation results [5,6].
    • Correction
    • Source
    • Cite
    • Save
    • Machine Reading By IdeaReader
    0
    References
    0
    Citations
    NaN
    KQI
    []