The paper presents an adaptive finite volume-particle method (AFVP) for simulating free surface flows, combining the efficiency of finite volume methods with the flexibility of smoothed particle hydrodynamics. The method dynamically partitions the computational domain into Eulerian mesh and Lagrangian particle regions based on proximity to free surfaces, ensuring mass and momentum conservation. Validation against analytical solutions shows second-order convergence rates, with performance comparisons indicating a 2.5x speedup over pure SPH methods.
Key findings
AFVP achieves second-order convergence for smooth solutions and first-order for discontinuous flows.
Mass is conserved to machine precision with physically consistent momentum and energy behavior.
Performance metrics show a 2.5x speedup over pure SPH methods while maintaining accuracy.
The adaptive partitioning strategy effectively manages computational resources.
Limitations & open questions
Further testing needed for more complex free surface scenarios.
Potential limitations in handling very large interface deformations.