Aerospace and Mechanical Engineering
Adjust Font Size: Normal Large X-Large
simulation of liquid-solids

National Science Foundation: KDI: Direct Numerical Simulation and Modeling of Solid-Liquid Flows

Principal Investigator: Professor Daniel Joseph


Professor Roland Glowinski University of Houston
Professor Howard Hu  University of Pennsylvania
Professor Yousef Saad  University of Minnesota
Professor Ahmed Sameh  Purdue University
Professor Vivek Sarin  Texas A & M University
Pushpendra Singh  New Jersey Institute of Technology

A Project Summary

The KDI/NCC proposal builds upon a prior NSF HPCC Grand Challenge award for the high-performance computation of the motion of solid particles in Newtonian and viscoelastic liquids; it aims to use these computations to elucidate fundamental dynamics and solve problems of engineering interest. The Grand Challenge goal was to develop software packages, called particle movers, capable of carrying out direct numerical simulations ("DNS") of the motion of thousands of particles in 2-D and hundreds in 3-D, in parallel implementation. Two state-of-the-art particle movers were developed. One of these (ALE particle mover) uses a moving unstructured, body-fitted grid and has given rise to the first and only package which can presently move particles by a field of Lagrange multipliers which enforce the constraint of rigid body motion. The DLM particle mover is inventive and appears to have great potential with respect to parallel implementation of large numbers of particles in realistic flow conditions. Both methods use a new combined weak formulation in which the fluid and particle equations of motion are combined into a single weak equation of motion from which the hydrodynamic forces and torques on the particles have been eliminated. These and other results achieved in the Grand Challenge project have opened new and promising lines for investigation.

One objective of KDI/NCC is to further develop the state-of-the-art particle movers started in the Grand Challenge project; this is incremental research, building great codes brick by brick, in which the innovation enters in the formulation of more efficient preconditioners and solvers. The handling of collisions in dense slurries is one challenging computational and physical problem suggested by the Grand Challenge work, which is intrinsic to particulate flow; it has not been faced in computational work before. Code enhancement for the particle movers will be achieved by a major overhaul of the solvers and preconditioners. The ALE particle mover will be enhanced by implementing an operator-splitting scheme to avoid the problems created by nonlinear iteration, by a matrix-free formulation to avoid excessive memory requirements, and by multilevel and domain decomposition methods to effect better 3-D parallel performance. A viscoelastic version of the DLM particle mover will be implemented.

A second objective is to pioneer ways to convert simulation results into forms which can be used by the fracturing industry. To do this, the proposal focuses on sand transport in fractured reservoirs, a surrogate for all forms of slurry transport. The research is to be done and partly financed by partners from the oil and oil service industries in the consortium STIMLAB. The oil industry makes extensive use of models to guide field operations. DNS is a new tool for modeling forces on particles, and can even be used to generate correlations to two-fluid models on a PC. The research on direct numerical simulation will impact both industry and academia. DNS can change the way fracturing models are constructed and used in this multi-billion-dollar-a-year industry. New microstructural problems associated with lift-off, resuspension, and slip velocities can be expected to impact the modeling and multi-phase flow communities, and to open a new frontier for mathematicians who study Navier-Stokes equations and the mysterious motions of particles in viscoelastic fluids. Clean mathematical problems suitable for proving theorems are generated by this work. The challenges of DNS for computer science are brand new, and the "solvers" as part of the software generated to meet these challenges will be made publicly available.

More project information can be found at:/Solid-Liquid_Flows/

Last Modified: 2011-11-14 at 16:16:44 -- this is in International Standard Date and Time Notation