Files in this item

FilesDescriptionFormat

application/pdf

application/pdf9712241.pdf (16MB)Restricted to U of Illinois
(no description provided)PDF

Description

Title:Investigation of mean flow and vertical vorticity in turbulent convection using spectral simulations on massively parallel processors
Author(s):Cortese, Thomas Anthony
Doctoral Committee Chair(s):Balachandar, S.
Department / Program:Mechanical Science and Engineering
Discipline:Theoretical and Applied Mechanics
Degree Granting Institution:University of Illinois at Urbana-Champaign
Degree:Ph.D.
Genre:Dissertation
Subject(s):Engineering, Mechanical
Computer Science
Abstract:Direct numerical simulations of Rayleigh-Benard convection are performed at several values of Rayleigh number (Ra). The computational domain consists of a square-planform rectangular box with horizontal-to-vertical aspect ratio equal to four. This domain is periodic in the horizontal directions, and the top and bottom boundaries are isothermal, no-slip, impermeable walls. The flow-fields considered here range from steady horizontally-oriented two-dimensional rolls to unsteady three-dimensional turbulence.
At lower values of Ra the convective flow exhibits regular spatial patterns that possess horizontal symmetry and therefore have zero instantaneous mean horizontal velocity. Also, when the flow is marginally supercritical the energy is primarily in the poloidal component and the flow does not exhibit any vertical vorticity. However, instantaneous mean horizontal motion and the spontaneous generation of concentrated vertical vorticity are observed at high Ra, indicating that turbulent Reynolds stresses have broken the horizontal symmetries.
Thermal plumes, which play an important role in the heat transfer at high Ra, can have significant vertical vorticity associated with them and thus are reminiscent of tornados in the atmosphere. The primary thrust of the present work is to investigate the three closely inter-related phenomena of mean horizontal motion, vertical vorticity, and thermal plumes.
The velocity field is decomposed into mean, toroidal, and poloidal components in order to investigate the physical mechanisms which drive these three phenomena. The poloidal motion is driven directly by the buoyancy that is induced by the temperature gradient. The mean flow is driven by the vertical gradients of the horizontally averaged Reynolds stresses, and the vertical vorticity is generated and maintained by vortex-tilting and vortex-stretching mechanisms, with the poloidal motion acting as catalyst.
Several important computational issues are also addressed. A library of kernel routines central to many different spectral method algorithms was successfully implemented on the Thinking Machines Corp. Connection Machine 5 (CM5), a massively parallel machine. The spectral-method algorithm performed beyond all expectations, at 26 Gflops on a 512-node CM5, in spite of the global data communication involved, thus demonstrating that spectral algorithms can be effectively implemented on massively parallel machines.
Issue Date:1996
Type:Text
Language:English
URI:http://hdl.handle.net/2142/21009
ISBN:9780591197822
Rights Information:Copyright 1996 Cortese, Thomas Anthony
Date Available in IDEALS:2011-05-07
Identifier in Online Catalog:AAI9712241
OCLC Identifier:(UMI)AAI9712241


This item appears in the following Collection(s)

Item Statistics