Showing papers by "Parviz Moin published in 2022"

Large Eddy Simulation of the NASA High-Lift Common Research Model

Abstract: This brief describes follow-on work to earlier research at the Center for Turbulence Research (Lehmkuhl et al. 2018; Goc et al. 2019, 2020a,b, 2021), where wall-modeled large-eddy simulation (WMLES) was used to simulate a realistic aircraft in landing configuration across the lift curve. This work differs from the previous investigations in that the focus is no longer on the JAXA Standard Model configuration (Ito et al. 2006), but rather on the NASA High-Lift Common Research Model (CRM-HL), which is set to become the new benchmark validation case in the field of computational aerodynamics of high-lift flows (Lacy & Sclafani 2016). In this work, we seek to apply lessons related to modeling choices and gridding approach from three years of experience in the simulation of flows in this regime to accurately predict aircraft maximum lift to within the tolerances (e.g., ∆CL ≤ 0.03 at maximum lift) required by the aerospace industry (Clark et al. 2020).

Non-Boussinesq subgrid-scale model with dynamic tensorial coefficients

Abstract: A major drawback of Boussinesq-type subgrid-scale stress models used in large-eddy simulations is the inherent assumption of alignment between large-scale strain rates and filtered subgrid-stresses. A priori analyses using direct numerical simulation (DNS) data has shown that this assumption is invalid locally as subgrid-scale stresses are poorly correlated with the large-scale strain rates [Bardina et al., AIAA 1980; Meneveau and Liu, Ann. Rev. Fluid Mech. 2002]. In the present work, a new, non-Boussinesq subgrid-scale model is presented where the model coefficients are computed dynamically. Some previous non-Boussinesq models have observed issues in providing adequate dissipation of turbulent kinetic energy [e.g.: Bardina et al., AIAA 1980; Clark et al. J. Fluid Mech., 1979; Stolz and Adams, Phys. of Fluids, 1999]; however, the present model is shown to provide sufficient dissipation using dynamic coefficients. Modeled subgrid-scale Reynolds stresses satisfy the consistency requirements of the governing equations for LES, vanish in laminar flow and at solid boundaries, and have the correct asymptotic behavior in the near-wall region of a turbulent boundary layer. The new model, referred to as the dynamic tensor-coefficient Smagorinsky model (DTCSM), has been tested in simulations of canonical flows: decaying and forced homogeneous isotropic turbulence (HIT), and wall-modeled turbulent channel flow at high Reynolds numbers; the results show favorable agreement with DNS data. In order to assess the performance of DTCSM in more complex flows, wall-modeled simulations of high Reynolds number flow over a Gaussian bump exhibiting smooth-body flow separation are performed. Predictions of surface pressure and skin friction, compared against DNS and experimental data, show improved accuracy from DTCSM in comparison to the existing static coefficient (Vreman) and dynamic Smagorinsky model.

Large-eddy simulation of a Gaussian bump with slip-wall boundary conditions

Abstract: Computational fluid dynamics is a powerful tool that has a growing presence in the aerospace industry for analyzing external aerodynamic applications. In particular, recent advances in computational methods have improved our abilities to predict and understand complex turbulent flows. These advances show promise that it soon may be possible to predict aerodynamic quantities of interest via affordable computations; accurate computer simulations can reduce the time and cost of aircraft certification by reducing the number of expensive wind tunnel tests. The field of applied computational fluid dynamics has focused heavily on the method of Reynolds-averaged Navier-Stokes (RANS) simulations, in which the mean flow quantities are computed directly while the effects of turbulence are modeled. However, RANS faces the turbulence closure problem that results from averaging the nonlinear Navier-Stokes equations. Existing closure models are challenged in complex flows and often require tuning of coefficients. Recent advances in computational hardware have allowed for the use of higher-fidelity methods, in particular large-eddy simulation (LES), where the large, energy-containing scales of turbulence are resolved by the numerical grid, and the effect of the unresolvable small scales on the large scales is modeled. LES provides improved predictive capability over RANS in complex flows because it resolves the large scales of turbulence, which are flow and geometry dependent, and models the small scales, which are more universal. However, LES still faces a challenge in modeling wall-bounded turbulent flows. In wall-bounded turbulence, the energy-containing eddies scale with distance from the wall, meaning significant isotropic grid refinement is needed close to the wall. In this method, referred to as wall-resolved LES (WRLES), the grid point requirements become intractable for high-Reynolds-number flows (Choi & Moin 2012), even nearing the cost of direct numerical simulation (DNS) (Yang & Griffin 2021). To avoid the need for intractable grid refinement, the near-wall flow is represented by a wall model, and the effect on the outer flow is imposed through boundary conditions. This method is commonly referred to as wall-modeled LES (WMLES). Traditional wall models for WMLES are wall-stress models. The most common wallstress models rely on the assumption of local equilibrium to find the wall stress, which is then imposed through the boundary conditions on the outer LES. For a review of this simulation paradigm, see Cabot & Moin (2000) and Bose & Park (2018). Commonly, the boundary conditions impose the desired wall stress via a Neumann condition in the wallparallel direction and a no-penetration condition in the wall-normal direction. Recently, the slip wall (Robin) boundary condition was derived for the LES velocity field (Bose & Moin 2014). This boundary condition has been shown to provide benefits over the traditional boundary conditions, including proper convergence behavior and improved

Stable, entropy-consistent, and localized artificial-viscosity method for capturing shocks and contact discontinuities

Abstract: Simulations of high-Mach-number compressible flows, and for high Reynolds numbers, require an accurate and stable discontinuity-capturing method. In this work, we propose a novel, entropy-consistent, and stable localized artificial-viscosity/diffusivity (LAD)-based method for capturing shock and contact discontinuities in compressible flows. Using an analogy between the Lax-Friedrichs (LF) flux and the artificial-viscosity methods, a discrete LF-type flux formulation is proposed for the LAD method. The proposed method satisfies the discrete kinetic energy– and entropy-consistency conditions presented by Jain & Moin (2020), hence the name entropy-consistent localized artificial diffusivity (EC-LAD). We also propose new sensors that localize where the artificial viscosity is acting, and show that the proposed method is suitable for large-eddy simulation (LES) of compressible turbulent flows with shocks. The sensors are designed in a way that the resulting method does not require tuning coefficients, depending on the problem being solved, that are typical of LAD methods. At the end, the extension of the proposed method to compressible two-phase flows is also presented.

Atomization of the optimally disturbed liquid jets

Abstract: This paper is associated with a video winner of a 2021 American Physical Society's Division of Fluid Dynamics (DFD) Gallery of Fluid Motion Award for work presented at the DFD Gallery of Fluid Motion. The original video is available online at the Gallery of Fluid Motion, https://doi.org/10.1103/APS.DFD.2021.GFM.V0060.

Predicting the nonlinear amplification of disturbances in a boundary layer using the spatial perturbation equations

Abstract: Predicting the laminar-turbulent transition process requires capturing the linear and nonlinear amplification of disturbances in laminar flows. Direct numerical simulations (DNS) faithfully predict disturbance growth, but at significant and often impractical expense (Slotnick et al. 2014). Utilizing the properties of slowly evolving shear flows, we can implement spatial marching techniques to reduce computational expense while accurately predicting the amplification of disturbances. The parabolized stability equations (PSE), which are based on a decomposition of the perturbations into slowly and fast-evolving components, are perhaps the best-known example (Bertolotti et al. 1992; Herbert 1997). The PSE can accurately capture the evolution of disturbances for many flows of interest at a fraction of the cost of DNS and thus allow predictions of transition to turbulence (Lozano-Durán et al. 2018). The particular formulation of the PSE, however also introduces fundamental limitations. Neglecting terms in the governing equations based on scaling analysis results in a set of equations that are not fully parabolic (Haj-Hariri 1994). Specifically, the streamwise pressure gradient term introduces residual ellipticity that requires numerical treatment to stabilize the marching procedure as discussed by Towne et al. (2019). Existing remedies include the use of an implicit Euler integration scheme in combination with a minimum step size (Li & Malik 1996, 1997), the disregard of the streamwise pressure gradient term in the governing equations (Chang et al. 1991; Li & Malik 1996), and the modification of the marching procedure via an additional damping term (Andersson et al. 1998). These modifications have undesirable disadvantages that have motivated recent development of alternative spatial marching techniques (e.g. Towne & Colonius 2015; Ran et al. 2019). A recently developed method of one-way spatial marching introduced by Towne & Colonius (2015) stabilizes the spatial marching of linear perturbations by splitting the governing equations using characteristic variables and removing the upstream-traveling components. Further development has led to a projection operator, which allows the inclusion of forcing terms in the governing equations (Towne 2016). The inclusion of any forcing, however, has led to observations of unstable behavior if certain grid conditions are not satisfied (Kamal et al. 2020). Ran et al. (2019) introduced the parabolized Floquet equations to study modal interaction in boundary-layer flows.The method captures the growth of a range of harmonics during the nonlinear amplification of disturbances through the successive linearization of the governing equations. In this work, we present a new approach using the spatial perturbation equations (SPE). The method has been developed as a fully predictive technique that captures the nonlinear amplification of disturbances in slowly varying shear flows. As shown in (Harris

A near-wall patch wall model for large-eddy simulation

Abstract: The direct numerical simulation (DNS) of high-Re wall-bounded turbulent flows is prohibitively expensive, with its cost scaling as Re 5/2 τ in channel flows, where Reτ is the Reynolds number defined based on the friction velocity (Mizuno & Jiménez 2013). The direct application of the large-eddy simulation (LES) approach, where only the energycontaining eddies are resolved, to wall-bounded turbulent flows, termed wall-resolved LES (WRLES), yields a marginally better cost scaling than DNS, Reτ (Mizuno & Jiménez 2013). The still prohibitive cost of WRLES is due to the reduction in size of the energycontaining eddies as the wall is approached (Tennekes & Lumley 1972; Townsend 1976). Wall-modeled LES (WMLES) attempts to further reduce the cost by only resolving the energy-containing eddies in the outer region of the flow, extending the near-isotropic grid which scales in outer units to the wall. The effect of the severely under-resolved near-wall region on the outer flow is replaced by wall models (Bose & Park 2018). The cost scaling of WMLES is potentially reduced to Reτ assuming the same arguments made by Choi & Moin (2012) and Yang & Griffin (2021) in the context of spatially-evolving boundary layers. The wall models employed in WMLES rely primarily on some form of the Reynoldsaveraged Navier-Stokes (RANS) equations (Bose & Park 2018). This could range from the entirety of the RANS equations solved on a separate embedded grid such as the nonequilibrium wall model (NEQWM) of Park & Moin (2014) to the equilibrium wall model (EQWM), where all terms in the RANS equations are ignored except for the wall-normal diffusion, equivalent to the assumptions of local equilibrium and a constant stress layer. The use of the RANS equations limits the predictive capabilities of these wall models to the mean wall-shear stress, with higher-order statistics and fluctuating quantities such as the wall-shear stress fluctuations being severely underpredicted, regardless of grid refinement (Park & Moin 2016). This is expected, because the near-wall eddies responsible for the wall-shear stress fluctuations are under-resolved and not modeled explicitly. Furthermore, the use of the gradient-diffusion hypothesis to model the Reynolds shear stress in both the EQWM and NEQWM prevents them from behaving correctly under viscously induced three-dimensional non-equilibrium effects (Lozano-Durán et al. 2020). This is due to the misalignment between the mean shear and the Reynolds shear stress. The objective of this study is to devise a wall model that incorporates the near-wall turbulent structures, hence addressing the misprediction of both the wall-shear stress fluctuations and the non-equilibrium behavior of the near-wall layer. To do so, we propose to use a near-wall patch with DNS resolution. The patch has a size that is constant in inner units in all three Cartesian directions and is independent of the LES grid size. To couple the patch to the outer LES flow, we utilize the wall-normal self-similar structure

Continuum numerical modeling of rarefied flows with slip-wall boundary conditions

Abstract: Rarefied gas flows are encountered in low-pressure environments such as in high-flying aircraft, re-entry of space vehicles, and dynamics of Earth satellites, and also at standard pressure conditions in microfluidics and microelectromechanical systems. One such application is the manufacturing process of layers of organic light-emitting diodes (OLEDs). In the production process, organic molecules pass through a series of pipes between the chambers as well as nozzles, which causes a severe pressure drop due to an abrupt change in the cross-section of the flow, before they are deposited on to a substrate. Maintaining the uniformity of the thickness of the OLED layers is crucial for the production process, and this requires an accurate control of the system. The pressure of the OLED production process ranges from 0.1 Pa down to 10−5 Pa. At such low pressures, the flow becomes rarefied and can no longer be considered a continuum. The classical continuum methods break down for rarefied gas flows and nonequilibrium effects need to be included, which pose severe challenges in both experiments and numerical simulations of these flows. The traditional Navier–Stokes equations with a no-slip boundary condition becomes inaccurate and underpredicts the mass flow rate (Maurer et al. 2003; Ewart et al. 2007). The behavior of a rarefied gas flow is accurately described by the Boltzmann equation (Sone 2002). However, solving the Boltzmann equation is computationally challenging due to the cost and the complicated structure of the molecular collision term (Dongari & Agrawal 2012). Therefore, the objectives of this study are to perform numerical simulations of rarefied flow using the traditional NavierStokes equations and to assess the accuracy of slip boundary conditions in accounting for the non-equilibrium effects in the rarefied regime.