Abstract

An efficient implementation of Cundall's model on the Cray T3D massively parallel computer is presented. This model is used to simulate granular media where every grain is identified separately, generating very time-consuming simulations. First, we show a data structure based on a triangulation which defines a neighbourhood of each grain. This allows to save computing time. Then we explain how to distribute data and work on the processors and give some preliminary results.

Details

Actions