High-Lift Common Research Model: RANS, HRLES, and WMLES perspectives for CLmax prediction using LAVA

Abstract

A unified assessment of three turbulence treatments: Reynolds Averaged Navier-Stokes (RANS), Hybrid RANS/LES (HRLES) and Equilibrium Wall-Modelled Large Eddy Simulation (WMLES) is presented for the High-Lift Common Research Model (CRM-HL). For the free-air configuration, steady-state RANS simulations show very accurate drag polar predictions in the low-AoA linear regime. However, strong grid sensitivity is reported near the maximum lift-state (CLmax ), with finer-grids showing larger errors and predicting erroneous flow topologies on the wing. Our RANS simulations show that several corrections for the Spalart-Allmaras (SA) turbulence model widely used in the community lead to more erroneous results compared to the baseline closure, without exception. Both scale-resolving methods (HRLES and WMLES) address these drawbacks and predict an outboard separation pattern on the main element that is in good agreement with the oil flow photographs taken from the QinetiQ wind tunnel experiments, when LES-appropriate grids and numerical discretizations are used. While RANS simulations with the baseline SA closure do not show any wing-root separation post CLmax , both HRLES and WMLES show onset of corner flow separation with varying degrees of progression, along with a weak pitch break in the wing-contribution of the overall pitching moment. This post-CLmax pitch break seen in the free-air simulations is weaker than the break observed in experiments, with a weaker break reported in WMLES for each iteration of grid-refinement. In-tunnel simulations using both SA-baseline RANS and WMLES show a much stronger post-CLmax break with the WMLES predictions showing excellent agreement with the experiment in terms of both the flow-topology observed and the pressure-coefficients at various spanwise stations. Sensitivity to the tunnel wall boundary layer is characterized via comparisons between viscous and inviscid treatments for the tunnel walls. WMLES predictions show moderate sensitivity at the predicted inboard flow-state at CLmax along with the progression towards a post-CLmax stall; however, this stalled state at AoA ≈ 20◦ (inside the tunnel) obtained with both tunnel wall treatments appears to be largely identical.