Cross-scale numerical simulations using discrete particle models

W. Dzwinel, W. Alda, D. A. Yuen

Research output: Contribution to journalArticlepeer-review

25 Scopus citations

Abstract

We propose a concept for a homogenous computational model in carrying out cross-scale numerical experiments on liquids. The model employs the particle paradigm and comprises three types of simulation techniques: molecular dynamics (MD), dissipative particle dynamics (DPD) and smoothed particle hydrodynamics (SPH). With respect to the definition of the collision operator, this model may work in different hierarchical spatial and time scales as: MD in the atomistic scale, DPD in the mesoscale and SPH in the macroscale. The optimal computational efficiency of the three types of cross-scale experiments are estimated in dependence on: the system size N-where N is the number of particles - and the number of processors P employed for computer simulation. For the three-hierarchical-stage, as embodied in the MD-DPD-SPH model, the efficiency is proportional to N 8/7 but its dependence on P is different for each of the three types of cross-scale experiments. The problem of matching the different scales is discussed.

Original languageEnglish (US)
Pages (from-to)397-418
Number of pages22
JournalMolecular Simulation
Volume22
Issue number6
DOIs
StatePublished - 1999

Bibliographical note

Funding Information:
We thank Professor Dr. J. Moicinski, Dr. J. Kitowski, Dr. M. Bubak, M. Pogoda and Mr. M. Slowik from the AGH Institute of Computer Science (Poland) and Dr. George Fann from PNNL for their contributions to this work. The work is supported by the Polish Committee for Scientific Research (KBN) Grant No. 8TllC00615 and by DOE “Office of Science’s Laboratory Technology Research Program”.

Keywords

  • Cross-scale simulations
  • Dissipative particle dynamics
  • Molecular dynamics
  • Parallel implementation
  • Smoothed particle dynamics

Fingerprint

Dive into the research topics of 'Cross-scale numerical simulations using discrete particle models'. Together they form a unique fingerprint.

Cite this