home · mobile · calendar · defenses · 2005-2006 · 

Thesis Defense - Schreüder

Parallel Numerical Solution of Groundwater Flow Problems
Computer Science PhD Candidate
12/15/2005
12:00pm-2:00pm

Basin scale groundwater flow models are non-trivial because the domain typically covers several decades and thousands of square miles. The flow is governed by

equation

which is highly non-linear as a result of the h-b and S(h) terms.

This research investigates the use of parallel processing techniques to solve the groundwater flow equations. Reduced execution times allow more complex models to be used, generally resulting in the ability to better simulate features such as stream-aquifer interaction.

The BeoFLOW program was created by parallelizing the popular USGS MODFLOW program. File compatibility was retained to encourage adoption by the groundwater community.

BeoFLOW uses the Krylov Subspace (KSP) linear parallel solvers and Scalable Nonlinear Equation Solver (SNES) parallel solvers from the Portable, Extensible Toolkit for Scientific Computation (PETSc). One processor is dedicated to solving the surface water flow equations which cause complex, long-distance interactions. BeoFLOW also implements a new relaxation technique to prevent inadvertent occurrence of h-b < 0.

The parallel techniques are evaluated using synthetic and practical problems, including the Rio Grande Decision Support System (RGDSS) Groundwater Model.

Committee: Xiao-Chuan Cai, Professor (Chair)
Oliver McBryan, Professor
Elizabeth Jessup, Associate Professor
Kenneth Knox, University of Denver
Stephen McCormick, Department of Applied Mathematics
Department of Computer Science
University of Colorado Boulder
Boulder, CO 80309-0430 USA
webmaster@cs.colorado.edu
www.cs.colorado.edu
May 5, 2012 (14:20)
XHTML 1.0/CSS2
©2012