We present a fully parallel version of the contact dynamics simulation method. The algorithm works for an arbitrary number of processors while dynamically balancing the load using a hierarchical domain decomposition method. We notably achieve 100% efficiency for large enough systems. Each processor calculates independently the contact forces belonging to its domain via an iterative scheme, while data exchange between domains happens only after each iteration step. The partially parallel update at the domain boundaries leads to additional iterations. However, this effect vanishes with increasing number of particles allowing for an effective parallelization.

This content is only available via PDF.