Publications

Computational physicists at Sandia National Laboratories have moved the Eulerian CTH code, and the arbitrary-Lagrangian-Eulerian ALEGRA code to distributed memory parallel computers. CTH is a three-dimensional solid mechanics code used for large-deformation, shock wave analysis. ALEGRA is a three-dimensional arbitrary Lagrangian-Eulerian solid-mechanics code used for coupled large-deformation, shock and structural mechanics problems. This paper discusses our experiences moving the codes to parallel computers, the algorithms we used and our experiences running the codes.

Computational physicists at Sandia National Laboratories have moved their production codes to distributed memory parallel computers. Such an effort required the development of parallel algorithms, parallel data bases and parallel support tools. The Eulerian CTH code was rewritten. Moving both ALEGRA and PRONTO to parallel computers required only a modest number of modifications. It involved restructuring the restart and graphics data bases to make them parallel and minimize the I/O to the parallel computer. It also involved developing mesh decomposition tools to divide a rectangular or arbitrary connectivity into sub-meshes. It also involved developing new visualization tools to process the very large, parallel data bases. This paper also discusses Sandia's experiences running these codes on its 1840 compute node Intel Paragon, 1024 processor nCUBE and networked stations.

This paper is an introductory discussion of stress pulse phenomena in simple solids and fluids. Stress pulse phenomena is a very rich and complex field that has been studied by many scientists and engineers. This paper describes the behavior of stress pulses in idealized materials. Inviscid fluids and simple solids are realistic enough to illustrate the basic behavior of stress pulses. Sections 2 through 8 deal with the behavior of pressure pulses. Pressure is best thought of as the average stress at a point. Section 9 deals with shear stresses which are most important in studying solids.

Continuum dynamics codes are categorized as Lagrangian or Eulerian according to the motion of the mesh. A Lagrangian code`s mesh moves with the material, so no mass flows between cells. An Eulerian code`s mesh is stationary, so mass flows between the cells. Eulerian codes have improved to the point where they are routinely used to solve a broad variety of large deformation solid and fluid dynamics problems ranging from air flow over an airplane wing to meteor impact on space structures. This presentation will concentrate on multi-fluid Eulerian codes capable of modeling transient were propagation in solids. These codes use a two-step process to integrate the physics across a time step. The first step, referred to as the Lagrangian step, integrates the physics on a Lagrangian mesh across the time step. The field values are then at the new time, but they are on the distorted Lagrangian mesh. The second step, referred to as the remap step, remaps the data on the distorted Lagrangian mesh back to the original Eulerian mesh thus completing one time step. The algorithms used in the first step are similar to those used in modern Lagrangian codes but they must be extended to handle multi-material cells. The algorithms used in the second step are complex and must be very carefully chosen to minimize errors. These algorithms include second-order, monotone advection equations to calculate the quantities flowing between cells. They also require algorithms that construct material interfaces inside multi-material cells. The strength and limitations of currently used numerical techniques will be discussed. New code development activities that combine the best features on both Lagrangian and Elueian codes will also be discussed. These new codes will employ the strengths of both technologies to address problems that cannot be adequately solved at this time.