iSALE-3D: A three-dimensional, multi-material, multi-rheology hydrocode and its applications to large-scale geodynamic processes

摘要

The numerical simulation of rapid geodynamic processes calls for codes that can handle compressible flows, often referred to as hydrocodes, which are in many respects similar to impact and shock physics codes used in Engineering Sciences. Codes that are specifically designed to simulate large-scale, rapid geodynamic processes such as meteorite impacts, collisions of planetesimals, explosions and volcanic eruptions, landslides and the formation of tsunami waves require sophisticated models of the mechanical and thermodynamic response of geomaterials (rocks, ice, etc.) to shock compression and large rapid deformations. In the study of geodynamie processes many of the constraining parameters, such as material properties, are often only vaguely known and different scenarios varying the unknown parameters have to be tested. To enable large parameter studies the simulation of a single scenario needs to be computable in a reasonable timeframe. Providing much more processors for a single calculation is one possibility which is often used. However, this requires a large computer cluster or even supercomputers. To enable such calculations within a much smaller infrastructure, specifically optimized algorithms have to be developed. Nevertheless, such simulations make high demands on computing power and are often more expensive in terms of memory and computation time than other hydro- or fluiddynamical problems generally solved in an incompressible manner. Here we present iSALE-3D, a three-dimensional multi-material, multi-rheology hydrocode. Originally developed to study meteorite impacts, it is applicable for a broad range of rapid dynamical problems. The paper aims at describing the algorithms and strategies used in the code. In particular we focus on the implementation of an adaptive multi-material interface reconstruction technique. This increases the stability, accuracy, and in particular efficiency of the code substantially.