H. Ted Davis, Fellow

Fluid Physics for Technologies

These researchers used Supercomputing Institute resources to investigate fluid physics and associated phenomena underlying current and future process technologies. A concomitant to this research was the development of efficient computational mathematics and numerical methods for the problems and machines at hand. The problems are typically nonlinear and multidimensional, often involving coupled partial differential and integral equations, and also frequently have time dependence. The mathematical methods principally employed include:

• Subdomains, adaptive sub-domaining, and finite element basis functions
• Isoparametric mapping into a computational domain
• Newton iteration with continuation
• Vectorized calculation of the coefficients of basis functions (adaptive time-stepping and iterative diagonalization of large Hermitian matrices)

Usually the structure of the solution spacefolds, bifurcations, etc.—that is, the number of solutions and their stability—is much more important than individual solutions. Most of the problems are not excessively large but have to be solved many times to map solution space. Transient behavior is often important and typically multiplies by two orders of magnitude the number of times the basic problem must be solved. The problems are usually highly vectorizable.

This team is also developing parallel computer methods for simulating the interior region of a liquid chomatographic column. A detailed flow field is developed from a pore-scale calculation of low Reynolds number flow in a packed bed of surface-retentive particles. The fluid-mechanical calculations are performed with a lattice-Boltzmann technique. Convection, diffusion, and retention of a solute are modeled by stochastic algorithms based on the Langevin equation. The computational scheme offers the ability to reproduce essential dynamics of the chromatographic process from fundamental considerations of packed bed geometry, flow velocity, solute diffusivity, and adsorption parameters.

Research Group and Collaborators

Ganapathy Ayappa, Department of Chemical Engineering, Indian Institute of Science, Karnataka, India
Daniel M. Kroll, Adjunct Faculty Collaborator
Robert S. Maier, Network Computing Services Inc., Minneapolis, Minnesota
Alon McCormick, Faculty Collaborator
Kishore Mohanty, Faculty Collaborator
Mark R. Schure, Rohm and Haas, Philadelphia, Pennsylvania
L. E. Scriven, Faculty Collaborator
Pieter Van Remoortere, Research Associate