JISE


  [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16]


Journal of Information Science and Engineering, Vol. 37 No. 1, pp. 79-92


A Dynamic Parallel Meshless Method for the Problems with Large-Scale Movable and Deformable Boundary


LIANG WANG1, RUI XUE1, NING CAI1,*, PAN CHEN1, XIAOBO CUI1, WEI WU2,
MIAOMIAO NIU1, DONGLIANG ZHANG1, ZHAO ZHANG1 AND XIAOSONG ZHANG3
1School of Energy and Power Engineering
Nanjing Institute of Technology
Nanjing, 211167 P.R. China

2Beijing Aerospace Technology Research Institute
Beijing, 100074 P.R. China

3School of Energy and Environment
Southeast University
Nanjing, 210096 P.R. China
E-mail: cnnjit@126.com


This paper puts forward a dynamic parallel meshless computing algorithm that efficiently solves flow fields with largescale motions of movable and deformable boundaries. The partition boundary is updated, as the moving boundary moves across the material interface. Meanwhile, the point clouds near the moving boundary are reconstructed. Our algorithm also solves the workload balance between nodes and information exchange in each region of the computational field, using the governing equations in the arbitrary Lagrangian-Eulerian (ALE) form. The AUFS scheme is extended to calculate the numerical convective flux. Take the interaction between a helium bubble and a shockwave as an example. Our algorithm is applied to compute the flow field with different numbers of discrete points (33,044 and 66,089) and partitions (2 and 4). The results show that our algorithm achieves an efficiency of over 80%, and captures the interaction between shockwaves and the bubble accurately. Hence, our parallel algorithm is suitable for solving problems with largescale motions of deformation boundaries. The research results shed new light on the calculation speed for similar problems.


Keywords: dynamic parallel algorithm, meshless method, large-Scale movable boundary, parallel efficiency, arbitrary Lagrangian-Eulerian (ALE) form

  Retrieve PDF document (JISE_202101_06.pdf)