Strategies for parallelizing the three-dimensional Navier-Stokes equations on the Cray T3E: analysis of thermocapillary flow bifurcations

Research output: Chapter in Book/Report/Conference proceedingChapter

7 Downloads (Pure)

Abstract

The implementation and performance of a parallel spatial direct numerical simulation of the three-dimensional, incompressible, non linear and time-dependent Navier-Stokes equations on the Cray T3E massively parallel computer system are documented.
The feasibility of using this approach to perform transition studies of Marangoni flow in half zone liquid bridges on this computer is examined.
A parallel method for solving the elliptic Navier-Stokes equations for incompressible fluid flows is presented. The parallel implementation strategy is based on grid-partition.
Message-passing protocol MPI has been coded into the code so that it is portable to systems that support this interface for interprocessor communications.
The code is based on a 3D Simplified Marker and Cell primitive variable method applied to a staggered finite-difference grid.
A parallel multisplitting kernel is introduced for the solution of the pseudo-pressure elliptic equation which represents the most time-consuming part of the algorithm.
Numerical experiments and parallel performance measurements are made on the implemented code to check the numerical properties and parallel efficiency.
Original languageEnglish
Title of host publicationScience and Supercomputing at CINECA
EditorsMarco Voli
Place of PublicationBologna, Italy
Pages326-340
Number of pages5
Volume11
Publication statusPublished - 1997

Keywords

  • three-dimensional Navier-Stokes equations
  • Cray T3E
  • numerical experiments
  • parallel implementation strategy

Fingerprint Dive into the research topics of 'Strategies for parallelizing the three-dimensional Navier-Stokes equations on the Cray T3E: analysis of thermocapillary flow bifurcations'. Together they form a unique fingerprint.

  • Cite this