Do Explicit Subgrid-Scale Models Improve Implicit LES? A GPU-Oriented High-Order DG Perspective
Please login to view abstract download link
High-order Discontinuous Galerkin (DG) methods are well suited for turbulent flow simulations, particularly on GPU architectures where very high polynomial orders can be employed at moderate computational cost. Despite their favorable accuracy properties, high-Reynolds-number flows remain under-resolved in practical settings, making simulations sensitive to numerical dissipation and aliasing errors. Here, we study the interaction between intrinsic DG dissipation mechanisms---specifically split-form discretizations and Riemann solvers---and explicit subgrid-scale (SGS) models within a Large Eddy Simulation (LES) framework. The analysis is conducted using the three-dimensional Taylor--Green vortex at Re = 1600 and in the inviscid limit. We assess kinetic energy dissipation, spectral behavior, and numerical robustness across different resolution regimes. Results indicate that energy- or entropy-stable split-form DG schemes can provide sufficient dissipation to ensure stable under-resolved simulations without the need for explicit SGS modeling. In well-resolved LES, SGS models do not yield accuracy improvements, as their dissipative action overlaps with the inherent numerical dissipation of the DG scheme. However, in marginally resolved regimes typical of very high Reynolds numbers, explicit SGS models effectively complement numerical dissipation and improve spectral accuracy by removing excess unresolved energy. These findings clarify the respective roles of implicit and explicit dissipation mechanisms in high-order DG-based turbulence simulations and provide practical guidelines for selecting modeling strategies in GPU-oriented LES. The numerical experiments and solver implementation are carried out using the HORSES3D high-order DG solver [1], following the methodology detailed in [2]. REFERENCES: [1] E. Ferrer, G. Rubio, G. Ntoukas, W. Laskowski, O. A. Mariño, S. Colombo, A. Mateo-Gabín, H. Marbona, F. M. de Lara, D. Huergo, et al., “HORSES3D: A high-order discontinuous Galerkin solver for flow simulations and multi-physics applications,” Computer Physics Communications, vol. 287, p. 108700, 2023 [2] G. Rubio, G. Ntoukas, M. Chávez-Módena, O. Mariño, B. Font, O. Lehmkuhl, E. Valero, and E. Ferrer, “Can explicit subgrid models enhance implicit LES simulations? A GPU-oriented high-order-solver perspective,” arXiv preprint arXiv:2512.04574, 2025.
