A Parallel Particle-In-Cell (PIC) Solver on Curvilinear Grids for Turbulent Flow Simulation

We present a new particle-in-cell (PIC) framework integrated with a validated, MPI-parallel curvilinear immersed boundary (CurvIB) method.The Eulerian flow field is resolved using a fractional-step projection method with a Smagorinsky Subgrid-scale model for large eddy simulation (LES). At the same time, the Lagrangian particle infrastructure enables efficient and flexible representation of additional transported quantities.

The solver is implemented using PETSc's distributed mesh and swarm modules, demonstrating stable and scalable performance for millions of particles on Cartesian grids and curvilinear geometries. Grid–particle interpolation is performed using a trilinear kernel, and reverse projection currently employs a conservative cell-averaging approach. Validation includes canonical curved-duct flow cases, which demonstrate accurate resolution and showcase the solver’s robustness on stretched and body-fitted meshes.

While the current implementation does not include active particle–field coupling or reactive physics, the infrastructure is designed to serve as a flexible foundation for future extensions such as scalar-filtered density function (FDF) methods, stochastic Subgrid modeling, or passive marker-based diagnostics. The combination of high particle capacity, parallel scalability, and geometric flexibility positions this solver as a promising platform for hybrid Eulerian–Lagrangian approaches in turbulent flow modeling.