Skip to main content
NASA Advanced Supercomputing (NAS) Division World-class modeling, simulation, and supercomputing in support of the NASA mission

Scale Resolving Simulations of Corner-Flow Separation in a Wing-Body Juncture

AMS Seminar Series

Speaker: Aditya Ghate, Science and Technology Corporation / NASA Ames Research Center

Presentation

Seminar Slide Deck (PDF-15.7MB)

Abstract

Two distinct simulation methodologies, Delayed Detached Eddy Simulation (DDES) and stress-based Wall-modelled Large Eddy Simulation (WMLES), are evaluated using structured overset curvilinear grids for the NASA Juncture flow model. While both methodologies are shown to mitigate the primary shortcomings of steady state Reynolds-Averaged Navier-Stokes (RANS) simulations, several unresolved aspects are identified. Strong sensitivity to RANS-type grid refinement is observed in the DDES with a substantial deterioration of the solution quality with increasing spatial resolution associated with deficiencies in the shielding function. Mean profiles for attached boundary layers on the fuselage show spurious inflections suggesting modelled stress depletion on finer grids. Lower numerical dissipation in terms of spatial discretization and time step size is seen to improve the solution quality, and the advantage of using RC and QCR2000 corrections in the underlying RANS closure is demonstrated for DDES on RANS-type meshes. Equilibrium Wall-Modelled LES using grids capable of representing kinematic effects the tripping dots over the fuselage nose and the wing leading edge, consistent with the experiment setup is performed using meshes with up to 550 million grid points. These high-resolution simulations are shown to be cost-competitive compared to DDES on RANS-type grids. Although the agreement between WMLES predictions of first and second order single point statistics with experimental measurements is promising, a fundamental shortcoming is noted in terms of an overshoot in streamwise momentum in corner regions of the wing-fuselage juncture. This excess momentum subsequently delays the onset of separation, thereby resulting in an underprediction in length of the separation bubble. While improvement to grid aspect ratios upstream of the separation location shows an improvement, the late onset of separation is not entirely mitigated. Some quantitative sensitivity to numerical discretization and subgrid scale modeling is observed; simulations with absolutely no numerical dissipation on the curvilinear meshes employed, show the best agreement with the experiment.

Biography

Photo of Aditya Ghate
Aditya Ghate joined the Computational Aerosciences Branch of the NASA Advanced Supercomputing (NAS) Division at NASA Ames Research Center in 2018. His research within the LAVA group involves wall-modelled Large Eddy Simulations (WMLES) of non-canonical geometries using the structured overset curvilinear Navier-Stokes formulations and development of Lattice-Boltzmann models for coarse-grid simulations of high-Reynolds number wall-bounded turbulence. His work is focused on development of wall-modelled LES capability in order to address deficiencies of legacy RANS modeling. This research spans a wide array of topic in aeronautics including smooth-body and corner-flow separation seen in high-lift (CLmax) configurations, and shock-induced flow separation seen in transonic flow configurations approaching buffet. Aditya obtained his PhD in Aeronautics & Astronautics from Stanford University in 2018 where his thesis work focused on theory and modeling on subfilter-scale turbulence in LES.