Wall-modeled large-eddy simulation of a separated flow over the NASA wall-mounted hump

Wall-modeled large-eddy simulation (WMLES) attempts to provide a predictive but affordable framework for LES by modeling the effect of the computationally demanding inner-layer structures on the outer-layer flow. The grids employed in LES are designed to resolve the outer layer only, thereby reducing the computational cost of LES significantly to affordable levels. The saving in the computational cost can easily be a factor of hundreds to thousands for external aerodynamics configurations at flight conditions. Over the past five years, significant improvements to the WMLES technique have been made by the researchers in the Center for Turbulence Research at Stanford University. These include extending the technique to unstructured grid flow solvers (Bodart & Larsson 2011; Park & Moin 2016), identifying and removing some of the error sources in WMLES (Kawai & Larsson 2012), removing ad-hoc tunings in the wall model (Park & Moin 2014), and developing a new class of wall model that is free from the RANS legacy (Bose & Moin 2014). The Center now plans to validate various wall-modeling methods against separating flows over non-trivial geometries using a common computational framework. This is important, because each wall model has been validated independently of other wall models, often against relatively simple flows using different codes. The final goal is applying the state-of-art wall models to realistic three-dimensional configurations relevant to external aerodynamics. This work is a part of the newly embarked NASAsupported project, aimed at increasing the technology-readiness level of WMLES for aerospace applications that is deemed low at this stage. In this article, we report WMLES of a separated flow over the NASA wall-mounted hump using two standard wall models, as a preliminary work to the aforementioned project. The flow configuration, numerical setup, and wall-modeling approaches are reviewed in Section II. The mean flow predictions from WMLES and the no-slip LES calculation are presented in Section III with comparison to the experimental measurements. The importance of resolving the thin, attached shear layer in the pre-separation region will be highlighted, with an emphasis on local mesh adaptations.