Search NASA⌕ Search

SEARCH · Search NASA

Results for “Direct numerical simulation”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 649 records · Page 36

Turbulent Transport by Diffusive Stratified Shear Flows: From Local to Global Models. I. Numerical Simulations of a Stratified Plane Couette Flow

Shear-induced turbulence could play a significant role in mixing momentum and chemical species in stellar radiation zones, as discussed by Zahn. In this paper we analyze the results of direct numerical simulations of stratified plane Couette flows, in the limit of rapid thermal diffusion, to measure the turbulent viscosity and the turbulent diffusivity of a passive tracer as a function of the local shear and the local stratification. We find that the stability criterion proposed by Zahn, namely that the product of the gradient Richardson number and the Prandtl number must be smaller than a critical values (JPr){sub c} for instability, adequately accounts for the transition to turbulence in the flow, with (JPr){sub c}≃0.007. This result recovers and confirms the prior findings of Prat et al. Zahn’s model for the turbulent diffusivity and viscosity, namely that the mixing coefficient should be proportional to the ratio of the thermal diffusivity to the gradient Richardson number, does not satisfactorily match our numerical data. It fails (as expected) in the limit of large stratification where the Richardson number exceeds the aforementioned threshold for instability, but it also fails in the limit of low stratification where the turbulent eddy scale becomes limited by the computational domain size. We propose a revised model for turbulent mixing by diffusive stratified shear instabilities that properly accounts for both limits, fits our data satisfactorily, and recovers Zahn’s model in the limit of large Reynolds numbers.

79 ASTRONOMY AND ASTROPHYSICS↗

Simulation and Modeling of Hypersonic Turbulent Boundary Layers Subject to Favorable Pressure Gradients due to Streamline Curvature

Direct numerical simulations (DNS) of favorable-pressure-gradient turbulent boundary layers are presented for a nominal freestream Mach number of 5, with the objective of assessing the limitations of the currently available Reynolds-averaged Navier-Stokes (RANS) models. The favorable pressure gradient is induced by the streamwise curvature of the two-dimensional, planar, convex measurement surface used during experiments at the Texas A&M University. The DNS data shows good comparison with the measured velocity profiles, strain rates, and some, but not all, of the Reynolds-stress components. The discrepancies between the predicted and the measured wall-normal as well as shear stress components are primarily attributed to the lower than actual values inferred from typical PIV measurements of turbulent boundary layers. The DNS data shows a zero or slightly negative Reynolds shear stress in the outer part of the boundary-layer, which is indicative of the decaying turbulent motion under a strong favorable pressure gradient. The DNS data is also compared with the results of RANS computations based on commonly used zero, one, and two equation eddy-viscosity models. The RANS models yield reasonable comparisons with the DNS-based skin friction under zero and weak pressure gradients, but significant discrepancies under a strong pressure gradient. The k-w SST model provided the best overall predictions of skin friction, except in the region where the flow transitions from a favorable to an adverse pressure gradient. While the RANS models examined herein also give good predictions of the Reynolds shear stress under a sufficiently weak pressure gradient, none of those models are able to appropriately capture the reduction in the Reynolds stresses when the flow was subjected to a strong pressure gradient. An a priori assessment of the turbulent heat-flux prediction based on the assumption of a constant turbulent Prandtl number with the DNS data shows that while the constant turbulent Prandtl number model is effective in predicting the wall-normal component of turbulent heat flux, it does not capture the turbulent heat transfer in the streamwise direction for all the pressure gradient cases. The failure of the constant turbulent Prandtl number model highlights a requirement for more advanced models of the turbulent heat flux.

turbulent boundary layers↗

Numerical simulation of transition in wall-bounded shear flows

The current status of numerical simulation techniques for the transition to turbulence in incompressible channel and boundary-layer flows is surveyed, and typical results are presented graphically. The focus is on direct numerical simulations based on the full nonlinear time-dependent Navier-Stokes equations without empirical closure assumptions for prescribed initial and boundary conditions. Topics addressed include the vibrating ribbon problem, space and time discretization, initial and boundary conditions, alternative methods based on the triple-deck approximation, two-dimensional channel and boundary-layer flows, three-dimensional boundary layers, wave packets and turbulent spots, compressible flows, transition control, and transition modeling.

Kleiser, Leonhard↗

A localized artificial diffusivity method to simulate compressible multiphase flows using the stiffened gas equation of state: A localized artificial diffusivity method to simulate compressible multiphase flows using the stiffened gas equation of state

The development of numerical approaches to perform direct numerical simulations of compressible multiphase flows has been an active field of research for several years. Proper treatment of fluid interfaces is crucial as important physics occur in this infinitesimally small region. Furthermore, the compressibility of the fluid requires proper treatment of discontinuities. Artificial diffusivity is among a number of methods widely used for compressible flows. This study develops a general form of consistent artificial diffusion fluxes and extends the localized artificial diffusivity method for high-order central schemes to solve multiphase flows with an interface-capturing method. These fluxes ensure an oscillation-free interface for pressure, velocity, and temperature without employing a sharpening technique. Moreover, the high-order representation of all scales in the flow helps capture the wide range of instabilities inherent in these flows. The goal is to develop an approach capable of performing high-fidelity simulations supported by physics-driven validation. lastly, this is achieved by solving the five-equation model with the stiffened-gas equation of state using the proposed method for multicomponent and multiphase flows on a variety of 1D and 2D problems.

42 ENGINEERING↗

Simulation and Modeling of Hypersonic Turbulent Boundary Layers Subject to Favorable Pressure Gradients due to Streamline Curvature

Direct numerical simulations (DNS) of favorable-pressure-gradient turbulent boundary layers are presented for a nominal freestream Mach number of 5, with the objective of assessing the limitations of the currently available Reynolds-averaged Navier-Stokes (RANS) models.The wall geometry and flow conditions of the DNS are representative of the experimental data for a Mach 4.9 turbulent boundary layer that was tested on a two-dimensional planar convexwall model in the high-speed blow-down wind tunnel located at the National Aerothermochemistry Laboratory at Texas A&M University. The favorable pressure gradient is induced bythe streamwise surface curvature. The DNS results will be compared with experimental measurements and RANS computations for the same flow conditions and wall geometry. The DNS datasets will also be analyzed to provide an improved physical understanding of hypersonic boundary-layer turbulence subject to mechanical nonequilibrium, including the validity of Morkovin’s hypothesis and the various available compressibility transformations, as well as to obtain turbulence statistics that are relevant to RANS modeling, including the distribution ofeddy viscosity, the budgets of turbulent kinetic energy and Reynolds stresses, and the velocity-temperature correlations. The RANS predictions will be obtained by using both the commonly used one- and two-equation models that assume a constant turbulent Prandtl number and an algebraic, variable Prandtl number model for the turbulent energy flux.

turbulent boundary layers↗

Coarse grained simulation of convectively driven turbulent mixing, transition, and turbulence decay

Accurate predictions with quantifiable uncertainties are essential to many practical turbulent flow applications exhibiting extreme geometrical complexity and broad ranges of length and time scales. Under-resolved computer simulations are typically unavoidable in such applications, and implicit large-eddy simulation (ILES) often becomes the effective strategy. In this work, we focus on assessing ILES initialized with well-characterized 25 6 3 homogeneous isotropic turbulence datasets generated with direct numerical simulation (DNS). ILES solutions based on the LANL xRAGE code are studied as function of resolution for 6 4 3 , 12 8 3 , 25 6 3 , and 51 2 3 grids. ILES performance of new directionally-unsplit high-order numerical hydrodynamics algorithms in xRAGE is examined in this context. Compared to the initial 25 6 3 DNS, we find longer inertial subranges and higher turbulence Reynolds number (Re) for 25 6 3 and 51 2 3 xRAGE — attributed to having linked DNS (Navier–Stokes based) solutions to nominally inviscid (higher Re) Euler based ILES solutions. For fixed 25 6 3 resolution, we find that significantly larger simulated turbulence Re can be achieved with the higher-order unsplit (vs. split) discretizations.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Uncertainty quantification in LES of channel flow

Here, in this paper, we present a Bayesian framework for estimating joint densities for large eddy simulation (LES) sub-grid scale model parameters based on canonical forced isotropic turbulence direct numerical simulation (DNS) data. The framework accounts for noise in the independent variables, and we present alternative formulations for accounting for discrepancies between model and data. To generate probability densities for flow characteristics, posterior densities for sub-grid scale model parameters are propagated forward through LES of channel flow and compared with DNS data. Synthesis of the calibration and prediction results demonstrates that model parameters have an explicit filter width dependence and are highly correlated. Discrepancies between DNS and calibrated LES results point to additional model form inadequacies that need to be accounted for.

97 MATHEMATICS AND COMPUTING↗

Supercriticality to subcriticality in dynamo transitions

Evidence from numerical simulations suggests that the nature of dynamo transition changes from supercritical to subcritical as the magnetic Prandtl number is decreased. To explore this interesting crossover, we first use direct numerical simulations to investigate the hysteresis zone of a subcritical Taylor-Green dynamo. We establish that a well defined boundary exists in this hysteresis region which separates dynamo states from the purely hydrodynamic solution. We then propose simple dynamo models which show similar crossover from supercritical to subcritical dynamo transition as a function of the magnetic Prandtl number. Our models show that the change in the nature of dynamo transition is connected to the stabilizing or de-stabilizing influence of governing non-linearities.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Consistent Large-Eddy Simulation of a Temporal Mixing Layer Laden with Evaporating Drops: A Posteriori Modelling - Part 2

Large-eddy simulation (LES) is conducted of a three-dimensional temporal mixing layer whose lower stream is initially laden with liquid drops which may evaporate during the simulation. The gas-phase equations are written in an Eulerian frame for two perfect gas species (carrier gas and vapour emanating from the drops), while the liquid-phase equations are written in a Lagrangian frame. The effect of drop evaporation on the gas phase is considered through mass, species, momentum and energy source terms. The drop evolution is modelled using physical drops, or using computational drops to represent the physical drops. Simulations are performed using various LES models previously assessed on a database obtained from direct numerical simulations (DNS). These LES models are for: (i) the subgrid-scale (SGS) fluxes and (ii) the filtered source terms (FSTs) based on computational drops. The LES, which are compared to filtered-and-coarsened (FC) DNS results at the coarser LES grid, are conducted with 64 times fewer grid points than the DNS, and up to 64 times fewer computational than physical drops. It is found that both constant-coefficient and dynamic Smagorinsky SGS-flux models, though numerically stable, are overly dissipative and damp generated small-resolved-scale (SRS) turbulent structures. Although the global growth and mixing predictions of LES using Smagorinsky models are in good agreement with the FC-DNS, the spatial distributions of the drops differ significantly. In contrast, the constant-coefficient scale-similarity model and the dynamic gradient model perform well in predicting most flow features, with the latter model having the advantage of not requiring a priori calibration of the model coefficient. The ability of the dynamic models to determine the model coefficient during LES is found to be essential since the constant-coefficient gradient model, although more accurate than the Smagorinsky model, is not consistently numerically stable despite using DNS-calibrated coefficients. With accurate SGS-flux models, namely scale-similarity and dynamic gradient, the FST model allows up to a 32-fold reduction in computational drops compared to the number of physical drops, without degradation of accuracy; a 64-fold reduction leads to a slight decrease in accuracy.

two-phase flow↗

The Discrete Equation Method (DEM) for Fully Compressible, Two-Phase Flows in Ducts of Spatially Varying Cross-Section

For the simulation of light water nuclear reactor coolant flows, general two-phase models (valid for all volume fractions) have been generally used which, while allowing for velocity disequilibrium, normally force pressure equilibrium between the phases (see, for example, the numerous models of this type described in H. Städtke, Gasdynamic Aspects of Two-Phase Flow, Wiley-VCH, 2006). These equations are not hyperbolic, their physical wave dynamics are incorrect, and their solution algorithms rely on dubious truncation error induced artificial viscosity to render them numerically well posed over a portion of the computational spectrum. The inherent problems of the traditional approach to multiphase modeling, which begins with an averaged system of (ill-posed) partial differential equations (PDEs) which are then discretized to form a numerical scheme, are avoided by employing a new homogenization method known as the Discrete Equation Method (DEM) (R. Abgrall and R. Saurel, Discrete Equations for Physical and Numerical Compressible Multiphase Mixtures, J. Comp. Phys. 186, 361-396, 2003). This method results in well-posed hyperbolic systems, this property being important for transient flows. This also allows a clear treatment of non-conservative terms (terms involving interfacial variables and volume fraction gradients) permitting the solution of interface problems without conservation errors, this feature being important for the direct numerical simulation of two-phase flows. Unlike conventional methods, the averaged system of PDEs for the mixture are not used, and the DEM method directly obtains a well-posed discrete equation system from the single-phase conservation laws, producing a numerical scheme which accurately computes fluxes for arbitrary number of phases and solves non-conservative products. The method effectively uses a sequence of single phase Riemann problem solutions. Phase interactions are accounted for by Riemann solvers at each interface. Non-conservative terms are correctly approximated. Some of the closure relations missing from the traditional approach are automatically obtained. Lastly, the continuous equation system resulting from the discrete equations can be identified by taking the continuous limit with weak-wave assumptions. In this work, this approach is tested by constructing a DEM model for the flow of two compressible phases in 1-D ducts of spatially varying cross-section with explicit time integration. An analytical equation of state is included for both water vapor and liquid phases, and a realistic interphase mass transfer model is developed based on interphase heat transfer. A robust compliment of boundary conditions are developed and discussed. Though originally conceived as a first step toward implict time integration of the DEM method (to relieve time step size restrictions due to stiffness and to achieve tighter coupling of equations) in multidimensions, this model offers some unique capabilities for incorporation into next generation light water reactor safety analysis codes. We demonstrate, on a converging-diverging two-phase nozzle, that this well-posed, 2-pressure, 2-velocity DEM model can be integrated to a realistic and meaningful steady-state with both phases treated as compressible.

42 ENGINEERING↗

As a Matter of Force—Systematic Biases in Idealized Turbulence Simulations

Many astrophysical systems encompass very large dynamical ranges in space and time, which are not accessible by direct numerical simulations. Thus, idealized subvolumes are often used to study small-scale effects including the dynamics of turbulence. These turbulent boxes require an artificial driving in order to mimic energy injection from large-scale processes. In this paper, we show and quantify how the autocorrelation time of the driving and its normalization systematically change the properties of an isothermal compressible magnetohydrodynamic flow in the sub- and supersonic regime and affect astrophysical observations such as Faraday rotation. For example, we find that δ-in-time forcing with a constant energy injection leads to a steeper slope in kinetic energy spectrum and less-efficient small-scale dynamo action. In general, we show that shorter autocorrelation times require more power in the acceleration field, which results in more power in compressive modes that weaken the anticorrelation between density and magnetic field strength. Thus, derived observables, such as the line-of-sight (LOS) magnetic field from rotation measures, are systematically biased by the driving mechanism. Finally, we argue that δ-in-time forcing is unrealistic and numerically unresolved, and conclude that special care needs to be taken in interpreting observational results based on the use of idealized simulations.

79 ASTRONOMY AND ASTROPHYSICS↗

Simulation and Modeling of Hypersonic Turbulent Boundary Layers Subject to Adverse Pressure Gradients due to Concave Streamline Curvature

Direct numerical simulations (DNS) of adverse-pressure-gradient turbulent boundary layers over a planar concave wall are presented for a nominal freestream Mach number of 5, with the objective of assessing the limitations of the currently available Reynolds-averaged Navier-Stokes (RANS) models. The wall geometry and flow conditions of the DNS are representative of the experimental data for a Mach 4.9 turbulent boundary layer that was tested on a two-dimensional planar concave wall model in the high-speed blow-down wind tunnel located at the National Aerothermochemistry Laboratory at Texas A&M University (TAMU). The DNS was validated against the experimental results of TAMU for the same flow conditions and wall geometry. An analysis of the DNS datasets was also conducted to provide an assessment of the validity of Morkovin’s hypothesis and the strong Reynolds analog for turbulence subject to mechanical nonequilibrium. In addition to the DNS results, RANS predictions are obtained by using the Baldwin-Lomax (BL), Spalart-Allmaras (SA), and the k - w SST turbulence models. The comparisons between RANS and DNS showed little impact of an adverse pressure gradient on the accuracy of these models, at least up to an incompressible Clauser pressure gradient parameter of beta(sub inc) 1.22. While the Boussinesq assumption provided reasonable predictions for the Reynolds shear stress, it failed to adequately predict the normal components of the Reynolds stress.

turbulent boundary layers↗

Dynamics of premixed flames in a narrow channel with a step-wise wall temperature

The effect of channel height, inflow velocity and wall temperature on the dynamics and stability of unity Lewis number premixed flames in channels with specified wall temperature is investigated with steady and transient numerical simulations using a two-dimensional thermo-diffusive model. The simplified model is capable of capturing many of the transitions and the combustion modes observed experimentally and in direct numerical simulations in micro- and meso-scale channels, and indicates that the thermal flame/wall interaction is the mechanism leading to the observed flame instabilities. Finally, an ad-hoc one-dimensional model based on the flame-sheet approximation is tested in its capacity to reproduce some of the flame dynamics of the two-dimensional thermo-diffusive model. (author)

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

A New Method for Accurate Treatment of Flow Equations in Cylindrical Coordinates Using Series Expansions

The motivation of this work is the ongoing effort at the Center for Turbulence Research (CTR) to use large eddy simulation (LES) techniques to calculate the noise radiated by jet engines. The focus on engine exhaust noise reduction is motivated by the fact that a significant reduction has been achieved over the last decade on the other main sources of acoustic emissions of jet engines, such as the fan and turbomachinery noise, which gives increased priority to jet noise. To be able to propose methods to reduce the jet noise based on results of numerical simulations, one first has to be able to accurately predict the spatio-temporal distribution of the noise sources in the jet. Though a great deal of understanding of the fundamental turbulence mechanisms in high-speed jets was obtained from direct numerical simulations (DNS) at low Reynolds numbers, LES seems to be the only realistic available tool to obtain the necessary near-field information that is required to estimate the acoustic radiation of the turbulent compressible engine exhaust jets. The quality of jet-noise predictions is determined by the accuracy of the numerical method that has to capture the wide range of pressure fluctuations associated with the turbulence in the jet and with the resulting radiated noise, and by the boundary condition treatment and the quality of the mesh. Higher Reynolds numbers and coarser grids put in turn a higher burden on the robustness and accuracy of the numerical method used in this kind of jet LES simulations. As these calculations are often done in cylindrical coordinates, one of the most important requirements for the numerical method is to provide a flow solution that is not contaminated by numerical artifacts. The coordinate singularity is known to be a source of such artifacts. In the present work we use 6th order Pade schemes in the non-periodic directions to discretize the full compressible flow equations. It turns out that the quality of jet-noise predictions using these schemes is especially sensitive to the type of equation treatment at the singularity axis. The objective of this work is to develop a generally applicable numerical method for treating the singularities present at the polar axis, which is particularly suitable for highly accurate finite-differences schemes (e.g., Pade schemes) on non-staggered grids. The main idea is to reinterpret the regularity conditions developed in the context of pseudo-spectral methods. A set of exact equations at the singularity axis is derived using the appropriate series expansions for the variables in the original set of equations. The present treatment of the equations preserves the same level of accuracy as for the interior scheme. We also want to point out the wider utility of the method, proposed here in the context of compressible flow equations, as its extension for incompressible flows or for any other set of equations that are solved on a non-staggered mesh in cylindrical coordinates with finite-differences schemes of various level of accuracy is straightforward. The robustness and accuracy of the proposed technique is assessed by comparing results from simulations of laminar forced-jets and turbulent compressible jets using LES with similar calculations in which the equations are solved in Cartesian coordinates at the polar axis, or in which the singularity is removed by employing a staggered mesh in the radial direction without a mesh point at r = 0.

Constantinescu, G.S.↗

Part 2 of a Computational Study of a Drop-Laden Mixing Layer

This second of three reports on a computational study of a mixing layer laden with evaporating liquid drops presents the evaluation of Large Eddy Simulation (LES) models. The LES models were evaluated on an existing database that had been generated using Direct Numerical Simulation (DNS). The DNS method and the database are described in the first report of this series, Part 1 of a Computational Study of a Drop-Laden Mixing Layer (NPO-30719), NASA Tech Briefs, Vol. 28, No.7 (July 2004), page 59. The LES equations, which are derived by applying a spatial filter to the DNS set, govern the evolution of the larger scales of the flow and can therefore be solved on a coarser grid. Consistent with the reduction in grid points, the DNS drops would be represented by fewer drops, called computational drops in the LES context. The LES equations contain terms that cannot be directly computed on the coarser grid and that must instead be modeled. Two types of models are necessary: (1) those for the filtered source terms representing the effects of drops on the filtered flow field and (2) those for the sub-grid scale (SGS) fluxes arising from filtering the convective terms in the DNS equations. All of the filtered-sourceterm models that were developed were found to overestimate the filtered source terms. For modeling the SGS fluxes, constant-coefficient Smagorinsky, gradient, and scale-similarity models were assessed and calibrated on the DNS database. The Smagorinsky model correlated poorly with the SGS fluxes, whereas the gradient and scale-similarity models were well correlated with the SGS quantities that they represented.

Okongo, Nora↗

Classification of machine learning frameworks for data-driven thermal fluid models

Thermal fluid processes are inherently multi-physics and multi-scale, involving mass-momentum-energy transport phenomena at multiple scales. Thermal fluid simulation (TFS) is based on solving conservative equations, for which – except for “first-principles” direct numerical simulation – closure relations (CRs) are required to provide microscopic interactions or so-called sub-grid-scale physics. In practice, TFS is realized through reduced-order modeling, and its CRs as low-fidelity models can be informed by observations and data from relevant and adequately evaluated experiments and high-fidelity simulations. This paper is focused on data-driven TFS models, specifically on their development using machine learning (ML). Five ML frameworks are introduced including physics-separated ML (PSML or Type I ML), physics-evaluated ML (PEML or Type II ML), physics-integrated ML (PIML or Type III ML), physics-recovered (PRML or Type IV ML), and physics-discovered ML (PDML or Type V ML). The frameworks vary in their performance for different applications depending on the level of knowledge of governing physics, source, type, amount and quality of available data for training. Notably, outlined for the first time in this paper, Type III models present stringent requirements on modeling, substantial computing resources for training, and high potential in extracting value from “big data” in thermal fluid research. Here the current paper demonstrates and investigates ML frameworks in three examples. First, we utilize the heat diffusion equation with a nonlinear conductivity model formulated by convolutional neural networks (CNNs) and feedforward neural networks (FNNs) to illustrate the applications of Type I, Type II, Type III, and Type V ML. The results indicate a preference for Type II ML under deficient data support. Type III ML can effectively utilize field data, potentially generating more robust predictions than Type I and Type II ML. CNN-based closures exhibit more predictability than FNN-based closures, but CNN-based closures require more training data to obtain accurate predictions. Second, we illustrate how to employ Type I ML and Type II ML frameworks for data-driven turbulence modeling using reference works. Third, we demonstrate Type I ML by building a deep FNN-based slip closure for two-phase flow modeling. The results show that deep FNN-based closures exhibit a bounded error in the prediction domain.

42 ENGINEERING↗