ITEMS

modern scholarly publishers in the finest tradition

Journals

	Search

International Journal for Multiscale Computational Engineering

ISSN for PRINT: 1543-1649

Institutional price:	$747.00
Issues per year:	6

Best Paper Award Selection - Editorial Board Site

2006, Volume4

271 pages

DOI: 10.1615/IntJMultCompEng.v4.i5-6

Issue price - $256.00

Wavelet-based Spatiotemporal Multiscaling in Diffusion Problems with Chemically Reactive Boundary

George Frantziskonis
University of Arizona

Sudib K. Mishra
Civil Engineering & Engineering Mechanics, University of Arizona, P.O. Box 210072, University of Arizona, Tucson, AZ 85721-0072

Sreekanth Pannala
Computer Science & Mathematics Division, Oak Ridge National Laboratory

Srdjan Simunovic
Computer Science & Mathematics Division, Oak Ridge National Laboratory

C. Stuart Daw
Computer Science & Mathematics Division, Oak Ridge National Laboratory

Phani Nukala
Computer Science & Mathematics Division, Oak Ridge National Laboratory

Rodney O. Fox
Iowa State University

Pierre A. Deymier
Materials Science amp;& Engineering, University of Arizona

ABSTRACT

Chemically reacting flows over catalytic and noncatalytic surfaces are one of the elementary operations in chemical processing plants. The underlying physical phenomena span time and length scales over several orders of magnitude, which a robust and flexible modeling framework must efficiently account for. With this purpose as the eventual goal, we propose a wavelet-based multiscale numerical framework and demonstrate it on the coupling of two prototype methods for the problem of species generated on a chemically reactive boundary and diffusing through the bulk. The two methods consider different time and length scales. The first method in this coupling, termed "fine," models the chemical reactions on the reactive boundary stochastically by the kinetic Monte Carlo method and the diffusion in the medium deterministically using relatively small time increments and small spatial discretization mesh size. The second method, termed "coarse," models both the reaction and the diffusion deterministically and uses drastically larger time increments and spatial discretization size than the fine model. The two methods are coupled by forming a spatiotemporal compound wavelet matrix that combines information about the time and spatial scales contained in them.