The simulation of particle suspensions in fluid flow is a key aspect of numerical modeling for complex phenomena, including blood flow and industrial processes. This work aims to extend the LEDDS framework by implementing ellipsoidal particles, enabling a more accurate representation of real particles and improving the simulation of particle-laden fluids.

Most existing Discrete Element Method (DEM) frameworks represent particles as spheres, simplifying particle-particle and particle-fluid interactions while allowing high particle volume fractions. The LEDDS framework provides a portable, GPU-optimized C++ implementation of sphere-based DEM, coupled with fluid dynamics using the Lattice Boltzmann Method (LBM).

This work introduces ellipsoidal particles by extending the existing DEM formulation and coupling it with LBM while maintaining the framework’s performance efficiency. We followed the data-oriented strategy to fully leverage GPU computational power. The implementation uses parallel algorithms from the C++17 standard, ensuring seamless parallelization and GPU deployment via compilers such as NVIDIA’s nvc++.

This addition to the open-source LEDDS framework enhances the accuracy of granular suspension simulations. The use of GPU computational power and standard C++ enables efficient large-scale simulations, opening new possibilities for research and improving our understanding of particle-fluid dynamics.