What is XDEM-HPC?

XDEM-HPC is a Julia framework for high-performance Discrete Element Method (DEM) simulations with full two-way coupling to Computational Fluid Dynamics (CFD) solvers.

It is the successor to the C++ XDEM platform, redesigned from the ground up for modern HPC hardware: multi-core CPU clusters and GPU accelerators (NVIDIA CUDA, AMD ROCm).

XDEM-HPC is currently under active development and has not yet been released as open source. We plan to make the code publicly available once a stable baseline is established. Watch this page or the GitHub organisation for announcements.

Planned features

GPU-native kernels via KernelAbstractions.jl — single codebase runs on CPU and GPU
Two-way CFD-DEM coupling to OpenFOAM (ESI v2512 and Foundation 9) via MPMD MPI
Float32 / Float64 selectable at runtime — maximise GPU throughput or precision as needed
Thermochemical particle physics — heat conduction, drying, devolatilisation, combustion
Co-located MPI partitioning — W-matrix projection is rank-local, no all-to-all communication

Background

XDEM (eXtended Discrete Element Method) was originally developed at the University of Luxembourg. That codebase is no longer maintained or accessible. XDEM-HPC is a full redesign led by Vilnius Gediminas Technical University, continuing the scientific vision with a modern Julia implementation focused on scalable HPC and GPU computing.