Session J11: CFD: DNS and LES

Wall-Modeled Large-Eddy Simulation of the Lockheed Martin X-59 QueSST

A wall-modeled large-eddy simulation of the experimental aircraft X-59 Quiet SuperSonic Technology (QueSST) developed by Lockheed Martin at Skunk Works for NASA's Low-Boom Flight Demonstrator project is conducted using the solver charLES. The capabilities of large-eddy simulation to predict the noise level in cruise conditions are evaluated and compared with other numerical studies. The main quantities of interest are the fairfield pressure field and the intensities and locations of the shock waves. This approach will enable the detection of design deficiencies prior to aircraft construction, resulting in financial benefits and accelerating certification by analysis efforts.   

Wall-Modelled Large-Eddy Simulations of flows with non-uniform roughness

Wall-Modelled Large-Eddy Simulations (WMLES) and hybrid RANS/LES

simulations of wall-bouned flows with non-homogeneous roughness

were carried out. The flow under consideration is an open channel,

with alternating rough- and smooth-wall strips oriented transverse

to the flow. Results were compared to turbulent boundary-layer

experiments in a similar configuration. Despite the fact that all

models considered rely on an equilibrium assumption

(production=dissipation), and are designed to yield the logarithmic

law-of-the-wall at the inner/outer-layer interface, they predict

the readjustment of the flow to the new surface conditions fairly

accurately. They compare well to the experiments, both in terms of

mean velocity and wall-shear stress. The hybrid method predicts the

sharp variation of the skin friction coefficient after the

roughness transition more accurately, at the expense of a higher

computational cost. The implications and limitations of the

equilibrium assumption will be discussed.

Research supported by the Natural Sciences and Engineering Council

of Canada (NSERC).   

Improved slip wall models using optimal finite element projections.

Slip-based models constitute a class of wall models that allow easy implementation of the wall boundary condition and enable the possibility to model flow separation. To improve its performance, an optimal projection framework is developed to offer a unified perspective of WMLES models and a wide range of other eddy-resolving models, viz. LES and Hybrid RANS-LES. Estimates for the slip-wall model coefficient C_w computed using optimal projection of the DNS solution and Reichardt profile show a strong dependence on the numerical method and the Reynolds number based on the slip velocity and the cube root of cell volume. This slip Reynolds number-based formulation allows easy computation of C_w without requiring any iterative procedure. In an attempt to formulate a universal model, a resolution normalizer λ is introduced, which effectively parametrizes the effect of the numerical method through just a single parameter. Further corrections to the model form due to subgrid modeling errors are made by estimating C_w from the solutions of the traditional WMLES approach. The resulting slip wall model with minimal variation of the new model constant λ generalizes to different Reynolds numbers and mesh resolutions.   

Tripping effects on the flow around a 6:1 prolate spheroid using large-eddy simulation

The flow around an inclined 6:1 prolate spheroid is a commonly studied canonical problem that exhibits a variety of complex phenomena found in immersed bodies at angle of attack.  These features include three-dimensional boundary layers, smooth separation and several mechanisms of natural transition including Tollmien-Schlichting, centrifugal and crossflow instabilities. The boundary layer of the prolate spheroid is often tripped in experiments and numerical studies to ensure reproducibility and high Reynolds number behavior. The present study aims at understanding the effects of tripping on the prolate spheroid flow using large-eddy simulation. A wall-resolved and trip-resolved approach is taken to investigate the performance of the trip at 20^o angle of attack and at a Reynolds number of 4.2 million based on length and freestream velocity. An overset methodology is used to model the trip setup used in the reference experiment, which consisted of a set of cylindrical posts at 20% of the length. Comparison to experiment and analysis of the results will be presented.

    

Extension of residual-based variational multiscale LES to the finite volume method

The finite volume method has been previously combined with the Streamline-upwind Petrov Galerkin (SUPG) method, originally developed for finite elements. The purpose of SUPG is to stabilize the numerical method for advection-dominated flows. The control volume formulation augmented with SUPG has been coined the streamline upwind control volume (SUCV) method by its original developers and has been shown to lead to convergent results for heat transfer problems. Here we review the SUCV and cast it within the context of residual-based variational multiscale (RBVMS) modeling for large-eddy simulation (LES). The subgrid-scale (SGS) model that results is expressed in terms of a SGS anisotropic viscosity and stress. The new RBVMS LES formulation on a finite volume discretization shows improved results relative to traditional LES with the dynamic Smagorinsky model on the same discretization in terms of mean velocity and root mean square of velocity for turbulent channel flows at various friction Reynolds numbers.   

A statistical framework using LES to assess the effect of internal heating and natural convection on airborne transmission

Under standard ventilation within confined spaces, the flow corresponding to the natural convection from a heat source such as a human adult, a heated wall or any electronic equipment may be substantial. The present work builds on a recently proposed statistical framework using high-fidelity large-eddy simulations and particle overloading (Salinas, J. S., Krishnaprasad, K. A., Zgheib, N., Balachandar, S. (2022). Improved guidelines of indoor airborne transmission taking into account departure from the well-mixed assumption. Physical Review Fluids, 7(6), 064309). More specifically, we investigate the effect of heating on the mixing of pathogen concentration. Several high-fidelity simulations were considered within a canonical room of size 10m × 10m × 3.2m with over 20 million droplet nuclei that were individually tracked. The heat sources consisted of a heated wall and volumetric heat sources, representing one or more adults/children in the room. In each simulation, the strength and positioning of the source were varied. Our statistically relevant results indicate that provided the flow is turbulent, the presence of heat sources or heated boundaries have an effect on the room averaged statistics, which can be accounted for in the well-mixed theory. Furthermore, we investigate the added effect of filtration, and optimal ACH (air changes per hour) as it pertains to reducing the risk of airborne transmission.   

A Generalized Method for External Forcing of DNS of Complex, Unsteady, and Anisotropic Turbulent Flows

Direct numerical simulations (DNS) of statistically stationary, homogeneous, isotropic turbulence have historically served as one of the primary tools for theoretical studies of turbulence. To sustain such flows, a variety of external forcing approaches have been suggested, which intend to represent the action of larger energetic flow scales not captured in the DNS. The key limitation of the prior strategies is that the effect of larger scales is condensed into a single scalar parameter - rate of kinetic energy injection, which decouples DNS from any realistic external flow features. Here we propose a generalized method for forcing fully resolved simulations of complex turbulent flows, which can be highly unsteady (non-equilibrium), anisotropic, and inhomogeneous. Large-scale flow structure is extracted from an appropriately filtered field from a large eddy simulation (LES) and is then used to force a fully resolved simulation of a particular flow subregion. This approach is implemented in physical space in contrast to prior spectral nudging strategies. The approach allows one to capture the details of a specific region of interest, which cannot be otherwise resolved in lower-fidelity LES. A detailed a priori and a posteriori analysis of the method performance is presented.   

Turbulent flow around a 3D stepped cylinder: statistical and modal analysis of different flow regimes

The present study focuses on the numerical investigation of the turbulent external flow around a 3D stepped cylinder. This flow consists of two cylinders with different diameters joint at one extremity and it represents a good model for real-world applications, e.g. the foundations of offshore wind turbines. We perform direct numerical simulations with the adaptive mesh refinement technique in the spectral elements code Nek5000. The Reynolds number varies from Re_D=150 to Re_D=5000, which is the highest considered for a DNS for that flow. The case is ideal for performing a methodological study of the spectral error indicator (SEI), which is used to measure interpolation and quadrature errors. Furthermore, this configuration shows interesting physical behaviours, as the junction vortex dynamic strongly depends on the local Re_D and three different cell structures are developing in the wake regions. We present new physical insights into the irregular wake regime, looking at the time-averaged first and second orders statistical moments. Eventually, we perform a proper orthogonal decomposition to quantify the most energetic coherent structures and establish a reduced-order model. A new mode responsible for the updraft has been identified for the higher Re_D.   

Quantifying the effect of Rotation on 2D Ellipsoids using Nonconforming Schwarz-SEM

Vortex shedding and the related added-mass effect’s impact on drag and lift felt by rotating spherical and ellipsoidal bodies moving in flows at moderate Reynolds numbers has been recently quantified using a novel moving nonconforming Schwarz-spectral element method, which consists of two overlapping computational domains. The study showed that rotation has a significant effect on drag and lift on the ellipsoids, which isn't captured by static simulations that only account for angle of attack of the body. In the current study we used 2D direct numerical simulations (DNS) to quantify the effect of Reynolds numbers, nondimensional angular velocities, and the ellipsoid’s aspect ratio. We conduct 2D DNS because they are computationally cheaper than 3D DNS, and appropriate for a wide parametric sweep. The results are compared with 3D DNS reported in Mittal et al., potential flow solutions, and empirical relationships from the literature. DMD analysis is also conducted on the velocity data to identify the dynamically important modes.

    

Predictive LES of aircraft icing aerodynamics

Predicting the aerodynamic performance of an aircraft in icing conditions is critical as failures in an aircraft’s ice protection system can compromise flight safety. Aerodynamic effects of icing have typically relied on RANS modeling, which usually struggles to predict stall behavior, including those induced by surface roughness. Encouraged by recent studies using LES that demonstrate the ability to predict stall characteristics on full aircraft with smooth wings at an affordable cost (Goc et al. FLOW, 2021), this study seeks to apply this methodology to icing conditions. Measurements of lift, drag, and pitching moments of a NACA23012 airfoil under clean and iced conditions are collected at Re = 1.8M. Preliminary results for the clean airfoil are completed and predict the correct stall angle and max lift coefficient. Using laser scanned, detailed representations of the icing geometries, LES calculations are conducted to compare integrated loads against experimental measurements in both clean and iced conditions at various angles of attack through the onset of stall (Broeren et al., Journal of Aircraft 2018).   

Physics Guided Neural Networks for Spatio-temporal Super-resolution of Turbulent Flows

Direct numerical simulation (DNS) of turbulent flows is computationally expensive and is not practical for simulating flows at high Reynolds numbers. Low-resolution large eddy simulation (LES) is a pragmatic alternative, but its success depends on modeling of the small scale flow dynamics. Reconstructing DNS from low-resolution LES is critical for many scientific and engineering disciplines, but it poses many challenges to existing super-resolution methods due to the complexity of turbulent flows and computational cost of generating frequent LES data. In this work, we propose a physics-guided neural network for reconstructing frequent DNS from sparse LES data by enhancing its spatial resolution and temporal frequency. Our proposed method consists of a partial differential equation (PDE)-based recurrent unit for capturing underlying temporal processes and a physics-guided super-resolution model that incorporates additional physical constraints. We demonstrate the effectiveness of both components in reconstructing the data generated by simulating the Taylor-Green Vortex sparse LES data. Moreover, we show that the proposed recurrent unit can preserve the physical characteristics of turbulent flows by leveraging the physical relationships in the Navier-Stokes equation.   

LES modeling of gas turbine combustor using Nek5000

Understanding the behavior of fuel and combustion within gas turbine combustors is key to the stable operation of engines. The fuel properties may also have significant impacts on the flame stability, structure and shape during engine operation. These combustors operate at very lean conditions and lean blow out (LBO) can occur in the event of flame instability. In this study, wall-resolved large-eddy simulations (WRLES) of the turbulent flow in the Army Research Laboratory's midsize combustor (ARC-M1) were performed using a high-order spectral element method Computational Fluid Dynamics (CFD) code, Nek5000. The statistical behaviors of the turbulent structures within the ARC-M1 are first validated against the available experimental data and further analyzed using proper orthogonal decomposition (POD) to investigate the flow structures and advance turbulence models for engineering simulations.