Skip to:

Scalable, Shared and Distributed Memory Algorithms for Computational Solids, Fluids and Geometry

Ron Fedkiw, Stanford University
David Grove, Army Research Laboratory
Brent Kraczek, Army Research Laboratory

Numerical simulation of physical phenomena is of great importance to the military as it can be used for designing better weaponry, training new recruits or for testing equipments to be used in the field.  The core research is to develop an open-source platform for physics simulations which can be used by researchers in academia, industry and the military for tackling real-world problems.  The PI is working to develop:

  • New algorithms for taking large time steps for rigid and deformable solids, plasticity, fracture, fluids, solid-fluid coupling, etc
  • Algorithms for reducing the computation per time step including faster projection methods, multigrid, preconditioners, Krylov solvers such as conjugate gradient, etc.
  • Algorithms for domain decomposition techniques including Chimera grids
  • Hybridized MPI/threaded framework
  • Publish the new system as part of the PhysBAM open-source release, portability, Android and Windows

 

There are many problems of interest to the Army that would benefit from computational resources.  Consider unexploded ordinance that is run over by an army vehicle, and the subsequently explodes producing high velocity projectiles that can penetrate the vehicle.  This can injure the driver, passengers and damage the vehicle as well as its cargo.  As another example, suppose a projectile hits a tank, then there can be similar issues where people inside the tank can get injured.  It is cheaper and easier to these and other wide range of tests computationally than is feasible experimentally.  The PI will release a newer version of PhysBAM, an open source library that can be used by the army to create such tests.