期刊
JOURNAL OF COMPUTATIONAL PHYSICS
卷 338, 期 -, 页码 137-164出版社
ACADEMIC PRESS INC ELSEVIER SCIENCE
DOI: 10.1016/j.jcp.2017.02.050
关键词
PIC; FLIP; MPM; APIC; Hybrid Lagrangian/Eulerian; Particle-grid
资金
- NSF [CCF-1422795]
- ONR [N000141110719, N000141210834]
- DOD [W81XWH-15-1-0147]
- Intel STC-Visual Computing Grant [20112360]
- Division of Computing and Communication Foundations
- Direct For Computer & Info Scie & Enginr [1422795] Funding Source: National Science Foundation
We present a new technique for transferring momentum and velocity between particles and grid with Particle-In-Cell (PIC) [1] calculations which we call Affine-Particle-In-Cell (APIC). APIC represents particle velocities as locally affine, rather than locally constant as in traditional PIC. We show that this representation allows APIC to conserve linear and angular momentum across transfers while also dramatically reducing numerical diffusion usually associated with PIC. Notably, conservation is achieved with lumped mass, as opposed to the more commonly used Fluid Implicit Particle (FLIP) [2,3] transfers which require a full mass matrix for exact conservation. Furthermore, unlike FLIP, APIC retains a filtering property of the original PIC and thus does not accumulate velocity modes on particles as FLIP does. In particular, we demonstrate that APIC does not experience velocity noise that is characteristic of FLIP in a number of Material Point Method (MPM) hyperelasticity calculations. Lastly, we demonstrate that when combined with the midpoint rule for implicit update of grid momentum that linear and angular momentum are exactly conserved. (C) 2017 Elsevier Inc. All rights reserved.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据