We detail our developments in the high-fidelity spectral-element code Neko that are essential for unprecedented large-scale direct numerical simulations of fully developed turbulence. Major innovations are modular multi-backend design enabling performance portability across a wide range of GPUs and CPUs, a GPU-optimized preconditioner with task overlapping for the pressure-Poisson equation and in-situ data compression. We carry out initial runs of Rayleigh-Bénard Convection (RBC) at extreme scale on the LUMI and Leonardo supercomputers. We show how Neko is able to strongly scale to 16,384 GPUs and obtain results that are not possible without careful consideration and optimization of the entire simulation workflow. These developments in Neko will help resolving the long-standing question regarding the ultimate regime in RBC.
Results from large-eddy simulations of a classical hydraulic jump at inlet Froude number two are reported. The computations were performed using the general-purpose finite-volume-based code OpenFOAM®, and the primary goal was to evaluate the influence of the modelling parameters on the predictive accuracy, as well as establish the associated best-practice guidelines. A benchmark simulation was conducted on a grid with a 1 mm-cell-edge length to validate the solver and provide a reference solution for the parameter influence study. The remaining simulations covered different selections of the modelling parameters: geometric vs. algebraic interface capturing, three mesh resolution levels, and four choices of the convective flux interpolation scheme. Geometric interface capturing led to better accuracy, but deteriorated the numerical stability and increased the simulation times. Interestingly, numerical dissipation was shown to systematically improve the results, both in terms of accuracy and stability. Strong sensitivity to the grid resolution was observed directly downstream of the toe of the jump.