SPH implementation using LAMMPS-Trilinos-1Sandia has implemented a scalable parallel capability to model polarized-fluid dynamics in micro devices driven by an electric potential. The Sandia team discretized the mathematical model (Navier-Stokes and Poisson-Bolzman equations) with an implicit smoothed-particle hydrodynamics (SPH) method. Due to the method’s particle nature, the team used the LAMMPS (Large-scale Atomic/Molecular Massively Parallel Simulator) particle library, which was coupled with Trilinos for linear and nonlinear systems solutions.

SPH implementation using LAMMPS-Trilinos-2This was the first time that LAMMPS and Trilinos were coupled, and their implementation effectiveness has been demonstrated in a paper submitted to Computer Methods in Applied Mechanics and Engineering. The team was able to achieve very good scalability, up to 130 million particles on 32 thousand cores—the largest implicit SPH simulation ever performed, and they verified implementation accuracy on several fluid-dynamic and electrokinetic benchmarks.

The code was used by Wenxiao Pan and Bruce Palmer (of PNNL) for realistic mesoscale porous-media applications such as the aqueous lithium-air batteries. This effort was done in collaboration with Nathan Trask (Brown University).

The work was done as part of the ASCR-funded Collaboratory on Mathematics for Mesoscopic Modeling of Materials (CM4) Mathematical Multi-faceted Integrated Capability Centers (MMICC), a collaborative effort between PNNL; Sandia; Brown, Stanford, Princeton, and Penn State universities; and the universities of Minnesota and California–Santa Barbara.