유체-입자 연동 개별요소해석 기법의 데이터 구조와 알고리즘

Abstract

To investigate the efficiency of data structures and algorithms in solving particle-laden flows, two particle-solving methodologies, array-based data structures and linked-list-based data structures are examined. In order to mimic various particle-involved fluid flow physics, three cases, particle-laden flow in a channel, random-deletion of particles, and collision pair searching algorithms, are selected and tested. The results demonstrate that array-based data structure is more advantageous in terms of computational speed and memory usage in the case of particle-laden flow in a channel. On the other hand, linked-list-based data structure is more advantageous in case of random-deletion of particles which indicates linked-list-based data structure has better performance in data insertion or deletion. Linked lists show outstanding performance in particle pair searching algorithm used with cell-based loop method which find a collision pair particle 657 times faster than particle-based loop method.