Comprehensive comparison of NS-SPH and SWE-SPH models for hydrodynamic simulations

Smoothed particle hydrodynamics (SPH) has been widely adopted to solve both Navier–Stokes (NS) and shallow water equations (SWEs). This study provides a systematic comparison of NS-SPH and SWE-SPH across dam-break flows, fluid–structure interactions, and field-scale floods, using multiple hydrodynamic indicators including pressure, water level, velocity, structural forces, and inundation extent. The results yield several novel findings. We first identify a previously unreported phenomenon in dam-break simulations, in which the wavefront predicted by SWE-SPH propagates faster than that by NS-SPH in the initial phase but it is subsequently overtaken. This behaviour can be explained by the influence of a non-hydrostatic pressure neglected in the SWE formulation. NS-SPH shows better capability in resolving three-dimensional processes such as splashing and peak impact loads on structures, while SWE-SPH, despite its hydrostatic assumption, remains robust for large-scale floods. It requires nearly 80% less computation and shows low sensitivity to particle spacing. Importantly, we reveal that model accuracy depends not only on the underlying assumptions but also on the interplay between particle resolution and problem scale; in large-scale cases, SWE-SPH may outperform NS-SPH when vertical resolution is insufficient. These findings provide new insights for selection of SPH-based models according to problem scale, computational resources, and the required level of hydrodynamic details.