A Fast Algorithm for Non-Newtonian Flow. An Enhanced Particle-Tracking Finite Element Code for Solving Boundary-Valve Problems in Viscoelastic Flow

Abstract

This project concerned the development of a new fast finite element algorithm to solve flow problems of non-Newtonian fluids such as solutions or melts of polymers. Many constitutive theories for such materials involve single integrals over the deformation history of the particle at the stress evaluation point; examples are the Doi-Edwards and Curtiss-Bird molecular theories and the BKZ family derived from continuum arguments. These theories are believed to be among the most accurate in describing non-Newtonian effects important to polymer process design, effects such as stress relaxation, shear thinning, and normal stress effects. This research developed an optimized version of the algorithm which would run a factor of two faster than the pilot algorithm on scalar machines and would be able to take full advantage of vectorization on machines. Significant progress was made in code vectorization; code enhancement and streamlining; adaptive memory quadrature; model problems for the High Weissenberg Number Problem; exactly incompressible projection; development of multimesh extrapolation procedures; and solution of problems of physical interest. A portable version of the code is in the final stages of benchmarking and testing. It interfaces with the widely used FIDAP fluid dynamics package.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Jan 05, 1989
Accession Number
ADA206997

Entities

People

  • David S. Malkus

Organizations

  • University of Wisconsin–Madison

Tags

Communities of Interest

  • Air Platforms

DTIC Thesaurus Topics

  • Computational Fluid Dynamics
  • Computational Science
  • Finite Element Analysis
  • Flow Visualization
  • Fluid Dynamics
  • Fluid Flow
  • Fluid Mechanics
  • Geometry
  • Materials
  • Mechanics
  • Numerical Analysis
  • Steady Flow

Readers

  • Computational Fluid Dynamics (CFD)
  • Fluid Dynamics.