Applied and Computational Mathematics Seminar (ACMS)

The ACMS is held, unless otherwise indicated, on Wednesdays, from 14:30 to 15:30 (2:30 to 3:30 PM), room 418 Machray Hall. If you want to give a talk, or have a visitor willing to give a presentation, contact Julien Arino or Stéphanie Portet.

A mailing list has been set up, to distribute the announcements of the seminars. Click here to subscribe.
In the following table, click on a title for details. See here for past seminars.

Date Speaker Title
10/11/2010 Bing-Chen Wang Modelling of Turbulent Flow and Dispersion in Complex Environments
24/11/2010 Shaun Lui Domain Decomposition Methods for Partial Differential Equations

Wednesday, November 10, 2010, 2:30-3:30 PM, 418 Machray Hall
Bing-Chen Wang
Department of Mechanical & Manufacturing Engineering
University of Manitoba
Modelling of Turbulent Flow and Dispersion in Complex Environments

The prediction of the turbulent dispersion of a passive scalar, resulting from an accidental or deliberate release of a hazardous material into an urban environment, has become an increasingly important problem in recent years. The major challenge associated with this problem involves obtaining a deeper understanding of the interaction of the dynamically evolving flow structures with the complex building geometry in an urban environment, as well as the coupling of the momentum and scalar transport processes. Over the past decade, significant advances have been made in our current understanding of urban flow and dispersion using both experimental and numerical approaches. In this research, several new sets of high-quality water-channel data for turbulent dispersion of passive scalars released from a localized source in idealized urban environments (consisting of regular and staggered obstacles) will be analyzed. Furthermore, the Reynolds-averaged Navier-Stokes (RANS) method (based on several newly proposed scalar dissipation rate models) will be used to numerically simulate the physical processes of turbulent dispersion in the obstacle arrays. Finally, the current programs on turbulent flow and dispersion based on some advanced numerical approaches (such as large-eddy simulation (LES) and hybrid RANS/LES) conducted at the University of Manitoba will be briefly introduced.
Wednesday, November 24, 2010, 14:30-15:30, 418 Machray Hall
Shaun Lui
Department of Mathematics
University of Manitoba
Domain Decomposition Methods for Partial Differential Equations

In domain decomposition methods, the domain of a partial differential equation (PDE) is split up into smaller subdomains. We solve the PDE in each subdomain, with appropriate boundary conditions along artificial boundaries, and stitch together the local solutions to form a global solution. This is an iterative method which is optimal in the sense that the convergence rate is independent of the discretization parameter and the number of subdomains. These methods have been used successfully by engineers and scientists to solve large-scale problems involving many millions of unknowns.

The first part of this talk will be an introduction to these methods and will be followed by a brief discussion of some recent work on a class of methods known as optimized Schwarz methods.