Thesis

Molecular dynamics simulation in arbitrary geometries for nanoscale fluid mechanics

Creator
Rights statement
Awarding institution
  • University of Strathclyde
Date of award
  • 2008
Thesis identifier
  • T12335
Qualification Level
Qualification Name
Department, School or Faculty
Abstract
  • Simulations of nanoscale systems where fluid mechanics plays an important role are required to help design and understand nano-devices and biological systems. A simulation method which hybridises molecular dynamics (MD) and continuum computational fluid dynamics (CFD) is demonstrated to be able to accurately represent the relevant physical phenomena and be computationally tractable. An MD code has been written to perform MD simulations in systems where the geometry is described by a mesh of unstructured arbitrary polyhedral cells that have been spatially decomposed into irregular portions for parallel processing. The MD code that has been developed may be used for simulations on its own, or may serve as the MD component of a hybrid method. The code has been implemented using OpenFOAM, an open source C++ CFD toolbox (www.openfoam.org) . Two key enabling components are described in detail. 1) Parallel generation of initial configurations of molecules in arbitrary geometries. 2) Calculation of intermolecular pair forces, including between molecules that lie on mesh portions assigned to different, and possibly non-neighbouring processors. To calculate intermolecular forces, the spatial relationship of mesh cells is calculated once at the start of the simulation and only the molecules contained in cells that have part of their surface closer than a cut-off distance are required to interact. Interprocessor force calculations are carried out by creating local copies of molecules from other processors in a layer around the processor in question. The process of creating these copied molecules is described in detail. A case study of flow in a realistic nanoscale mixing channel, where the geometry is drawn and meshed using engineering CAD tools, is simulated to demonstrate the capabilities of the code for complex simulations.
Advisor / supervisor
  • Reese, Jason
Resource Type
DOI
EThOS ID
  • uk.bl.ethos.501789
Date Created
  • 2008
Former identifier
  • 997969953402996
Funder

Relations

Items