Search NASA⌕ Search

SEARCH · Search NASA

Results for “Large Eddy 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 1,207 records · Page 67

Flow adjustment inside homogeneous canopies after a leading edge – An analytical approach backed by LES

A two-dimensional analytical model for describing the mean flow behavior inside a vegetation canopy after a leading edge in neutral conditions was developed and tested by means of large eddy simulations (LES) employing the LES code PALM. The analytical model is developed for the region directly after the canopy edge, the adjustment region, where one-dimensional canopy models fail due to the sharp change in roughness. The derivation of this adjustment region model is based on an analytic solution of the two-dimensional Reynolds averaged Navier–Stokes equation in neutral conditions for a canopy with constant plant area density (PAD). The main assumptions for solving the governing equations are separability of the velocity components concerning the spatial variables and the neglection of the Reynolds stress gradients. These two assumptions are verified by means of LES. To determine the emerging model parameters, a simultaneous fitting scheme was applied to the velocity and pressure data of a reference LES simulation. Furthermore a sensitivity analysis of the adjustment region model, equipped with the previously calculated parameters, was performed varying the three relevant length, the canopy height ( h ), the canopy length and the adjustment length ( L c ), in additional LES. Even if the model parameters are, in general, functions of h / L c , it was found out that the model is capable of predicting the flow quantities in various cases, when using constant parameters. Subsequently the adjustment region model is combined with the one-dimensional model of Massman, which is applicable for the interior of the canopy, to attain an analytical model capable of describing the mean flow for the full canopy domain. As a result, the model is tested against an analytical model based on a linearization approach.

54 ENVIRONMENTAL SCIENCES↗

Turbulence kinetic energy budget and conditional sampling of momentum, scalar, and intermittency fluxes in thermally stratified wind farms

A description of the turbulence kinetic energy budget is investigated to understand the dynamics and turbulence energy transfer between the atmospheric boundary layer and large wind farms. The turbulence kinetic energy budget terms are a dominant factor in explaining the energy supply process in wind farms. Large eddy simulations are used to generate the thermally stratified wind turbine array boundary layer, where stable and unstable conditions are considered. The wakes represent a sink of mean kinetic energy, and turbulence transport redistributes energy in the wake region. The energy balance between the budget terms is dependent on the physical location in the wind farm. During the unstable scenario, turbulence is driven by the production, transport, and buoyancy terms. Further, the contribution of the turbulence kinetic energy budget terms is reduced with increasing atmospheric stability, especially above the shear layer, where the mechanical production is the dominant term. Pressure diffusion redistributes the turbulent energy from the streamwise direction into the other components of the normal Reynolds stresses. This distinctive feature of the changing thermal stratification is presented through the quadrant analysis. The momentum ejection and sweep frequencies are both comparable and dominant. The scalar flux shows an increase in the contribution of the negative correlation in the quadrants. The quadrant analysis is extended to account for intermittency based on the pointwise Holder exponents. As a result, positive correlations are dominant in the wake and in higher layers of the domain for both considered cases.

17 WIND ENERGY↗

a priori uncertainty quantification of reacting turbulence closure models using Bayesian neural networks

While many physics-based closure model forms have been posited for the sub-filter scale (SFS) in large eddy simulation (LES), vast amounts of data available from direct numerical simulations (DNS) create opportunities to leverage data-driven modeling techniques. Albeit flexible, data-driven models still depend on the dataset and the functional form of the model chosen. Increased adoption of such models requires reliable uncertainty estimates both in the data-informed and out-of-distribution regimes. Here, in this work, we employ Bayesian neural networks (BNNs) to capture both epistemic and aleatoric uncertainties in a reacting flow model. In particular, we model the filtered progress variable scalar dissipation rate which plays a key role in the dynamics of turbulent premixed flames. We demonstrate that BNN models can provide unique insights about the structure of uncertainty of the data-driven closure models. We also propose a method for the incorporation of out-of-distribution information in a BNN, which can be used for out-of-distribution query detection. The efficacy of the model is demonstrated by a priori evaluation on a dataset consisting of a variety of flame conditions and fuels.

97 MATHEMATICS AND COMPUTING↗

Machine learning assisted modeling of mixing timescale for LES/PDF of high-Karlovitz turbulent premixed combustion

Accurate modeling of mixing in the transported probability density function (PDF) method remains a great challenge, especially for turbulent premixed combustion under extreme conditions such as high Karlovitz number Ka. Recently, a power-law based mixing timescale model was developed for the large-eddy simulations (LES)/PDF modeling of high-Ka number turbulent premixed flames. It is found in this work that the power-law mixing timescale model is highly sensitive to the model parameters. It is thus critically needed to develop accurate calibration of these model parameters. The empirical specification of the model parameters developed in Zhang et. al. is found to be inadequate for accurate modeling of the mixing timescale. Here, machine learning is introduced as an attractive alternative in this work for the specification of the model parameters. A high-Ka number DNS jet flame is used as the training and validation of the machine learning models. The choices of the input parameters are discussed and compared for the machine learning models. The effect of differential molecular diffusion on mixing is examined by including the effect of the Lewis number in the training of the machine learning models. The performance of different machine learning algorithms is compared for the specification of the mixing model parameters. Overall, excellent performance of the machine learning models is observed for assisting the mixing modeling. The feasibility, interpretability, applicability, generality, and portability of using machine learning are discussed in general to provide a perspective on applying data-driven machine learning for turbulent combustion modeling studies.

42 ENGINEERING↗

Optimal smoothing length scale for actuator line models of wind turbine blades based on Gaussian body force distribution: Wind energy, actuator line model

The actuator line model (ALM) is a commonly used method to represent lifting surfaces such as wind turbine blades within large-eddy simulations (LES). In the ALM, the lift and drag forces are replaced by an imposed body force that is typically smoothed over several grid points using a Gaussian kernel with some prescribed smoothing width e. To date, the choice of e has most often been based on numerical considerations related to the grid spacing used in LES. However, especially for finely resolved LES with grid spacings on the order of or smaller than the chord length of the blade, the best choice of e is not known. In this work, a theoretical approach is followed to determine the most suitable value of e, based on an analytical solution to the linearized inviscid flow response to a Gaussian force. We find that the optimal smoothing width eopt is on the order of 14%-25% of the chord length of the blade, and the center of force is located at about 13%-26% downstream of the leading edge of the blade for the cases considered. These optimal values do not depend on angle of attack and depend only weakly on the type of lifting surface. It is then shown that an even more realistic velocity field can be induced by a 2-D elliptical Gaussian lift-force kernel. Some results are also provided regarding drag force representation.

17 WIND ENERGY↗

New Capabilities and Improvements to the High-Order Glenn Flux Reconstruction Code

The Glenn Flux Reconstruction (GFR) code is a computational fluid dynamics (CFD) code under development at NASA Glenn Research Center. GFR is based on the high-order flux reconstruction (FR) method and provides a large-eddy simulation (LES) capability that is both accurate and efficient for complex aeropropulsion flows. Three significant new capabilities have been added to the code that improve its performance and functionality. First, a variety of explicit Runge-Kutta methods, including some with adaptive time stepping, were added to GFR with two methods offering a 33% improvement in time-to-solution. Second, GFR can now utilize fully unstructured, mixed-element meshes to more easily facilitate the grid generation process for complex geometries. Finally, a rotating reference frame capability has been added to GFR for solving rotating turbomachinery problems. A selection of results demonstrating these new capabilities are presented in this work. The Taylor-Green vortex problem is used to verify the new unstructured capability by showing similar accuracy and resolution for all element types. LES of the Turbulent Heat Flux Phase III (THX3) experiment with comparison to another high-order LES code and a popular Reynolds-averaged Navier-Stokes (RANS) code demonstrates the accuracy of the code for complex aeropropulsion flows. Finally, LES of a spacecraft cabin ventilation fan shows the ability of GFR to efficiently establish a fan performance map and identify operating points for further analysis at high orders of accuracy.

High-Order Methods↗

TURBULENT CONVECTION IN STELLAR INTERIORS. III. MEAN-FIELD ANALYSIS AND STRATIFICATION EFFECTS

We present three-dimensional implicit large eddy simulations of the turbulent convection in the envelope of a 5 M{sub Sun} red giant star and in the oxygen-burning shell of a 23 M{sub Sun} supernova progenitor. The numerical models are analyzed in the framework of one-dimensional Reynolds-Averaged Navier-Stokes equations. The effects of pressure fluctuations are more important in the red giant model, owing to larger stratification of the convective zone. We show how this impacts different terms in the mean-field equations. We clarify the driving sources of kinetic energy, and show that the rate of turbulent dissipation is comparable to the convective luminosity. Although our flows have low Mach numbers and are nearly adiabatic, our analysis is general and can be applied to photospheric convection as well. The robustness of our analysis of turbulent convection is supported by the insensitivity of the mean-field balances to linear mesh resolution. We find robust results for the turbulent convection zone and the stable layers in the oxygen-burning shell model, and robust results everywhere in the red giant model, but the mean fields are not well converged in the narrow boundary regions (which contain steep gradients) in the oxygen-burning shell model. This last result illustrates the importance of unresolved physics at the convective boundary, which governs the mixing there.

79 ASTRONOMY AND ASTROPHYSICS↗

Evaluation of Wall-Modeled Les for Flow Over A Multi-Element Airfoil

We perform wall-modeled large-eddy simulations (WMLES) of turbulent flow over a 30P30N multielement airfoil at various angles of attack to assess its ability to accurately predict lift using an unstructured node-based finite-volume methodology. The flow conditions are based on the experiments reported in Klausmeyer & Lin [NASA /TM-112858]. The oncoming flow Mach number is 0.2, and the stowed chord-based Reynolds number (𝑅𝑒𝑐) is 9 million. The angle of attack (𝛼) is varied between 8◦ and 23◦ with the expected stall angle of around 21◦. The simulated geometry is periodic in the span, thus any three-dimensional effects present in the experiment are ignored. We first analyze the predictions with baseline isotropic and anisotropic grids containing over 16 grid points per boundary-layer thickness. The near-wall streamwise/nominal spacing ratio is 0.4 at the leading/trailing edges of the slat, main element and flap, and 1 elsewhere, and the first wall-normal/nominal spacing ratio is 0.375 over most of the airfoil for the anisotropic grid, while these are unity for the isotropic grid. It is found that the predicted lift and wall pressures, near-wall velocity profiles, and the location of onset of resolved turbulence are sensitive to the near-wall grid anisotropy. Further sensitivities were explored at lower angles of attack by independently varying the streamwise and wall-normal grid anisotropies. The effects of both off-wall and near-wall grid refinement were also quantified. For the grid resolutions used here, our results suggest that the first wall-normal spacing, which coincides with the WMLES exchange location, has a dominant effect on the predictions for this flow configuration.

Computational Fluid Dynamics↗

A Finite Element Method for Compressible and Turbulent Multiphase Flow Instabilities with Heat Transfer

We present a new finite element framework for modeling compressible, turbulent multiphase flows with heat transfer. For two-fluid systems with a free surface, the Volume of Fluid (VOF) method is implemented without the need for interface reconstruction, while turbulence is resolved using a dynamic Vreman large eddy simulation (LES) model. Unlike most two-phase VOF studies, which neglect heat transfer, the present approach incorporates energy transport equations within the VOF formulation to account for heat exchange, an effect particularly important in turbulent flows. Conjugate heat transfer is often challenging in finite volume methods, which require explicit specification of heat fluxes at the solid–fluid interface, limiting accuracy and predictive capability. By contrast, the finite element formulation does not require heat flux inputs, allowing more accurate and robust simulation of heat transfer between solids and fluids. The method is demonstrated through three representative cases. First, a two-fluid instability with a single-mode perturbation is simulated and validated against analytical growth rates. Second, conjugate heat transfer is examined in a high-temperature flow over a cold metal cylinder, with validation performed both quantitatively—via pressure coefficient comparisons with experimental data—and qualitatively using vector field topology. Finally, compressible spray injection and breakup are modeled, demonstrating the ability of the framework to capture interfacial dynamics and atomization under turbulent, high-speed conditions. In the compressible spray injection and breakup case, the results indicate that the finite element formulation achieved higher predictive accuracy and robustness than the finite-volume method. With the same mesh resolution, the FEM reduced the root mean square error (RMSE) and mean absolute percentage error (MAPE) from 6.96 mm and 26.0% (for the FVM) to 4.85 mm and 12.7%, respectively, demonstrating improved accuracy and robustness in capturing interfacial dynamics and heat transfer. The study also introduced vector field topology to visualize and interpret coherent flow structures and instabilities, offering insights beyond conventional scalar-field analyses.

97 MATHEMATICS AND COMPUTING↗

A two-step chemical scheme for kerosene-air premixed flames

A reduced two-step scheme (called 2S-KERO-BFER) for kerosene-air premixed flames is presented in the context of Large Eddy Simulation of reacting turbulent flows in industrial applications. The chemical mechanism is composed of two reactions corresponding to the fuel oxidation into CO and H{sub 2}O, and the CO - CO{sub 2} equilibrium. To ensure the validity of the scheme for rich combustion, the pre-exponential constants of the two reactions are tabulated versus the local equivalence ratio. The fuel and oxidizer exponents are chosen to guarantee the correct dependence of laminar flame speed with pressure. Due to a lack of experimental results, the detailed mechanism of Dagaut composed of 209 species and 1673 reactions, and the skeletal mechanism of Luche composed of 91 species and 991 reactions have been used to validate the reduced scheme. Computations of one-dimensional laminar flames have been performed with the 2S{sub K}ERO{sub B}FER scheme using the CANTERA and COSILAB softwares for a wide range of pressure ([1; 12] atm), fresh gas temperature ([300; 700] K), and equivalence ratio ([0.6; 2.0]). Results show that the flame speed is correctly predicted for the whole range of parameters, showing a maximum for stoichiometric flames, a decrease for rich combustion and a satisfactory pressure dependence. The burnt gas temperature and the dilution by Exhaust Gas Recirculation are also well reproduced. Moreover, the results for ignition delay time are in good agreement with the experiments. (author)

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Effect of soot concentration on radiative transport in a 2 m JP-8 pool fire

Hubbard et al. (2024) investigated conjugate heat transfer in a 7.9 m JP-8 pool fire, utilizing both heavily sooting and non-sooting fuel models. Their findings indicated that all models produced pool radiative heat fluxes within the measurement uncertainty for large-scale outdoor experiments. This study aims to further investigate the influence of soot concentration on radiative transport. To this end, Large Eddy Simulations (LES) of a 2 m diameter JP-8 pool fire were conducted, employing four turbulent combustion chemistry models: the Eddy Dissipation Concept (EDC) and three variants of flamelet model. Additionally, three variations of the one-equation subgrid scale turbulence kinetic energy model, and multiple mesh resolutions were utilized. The primary validation metrics were pool radiative heat flux and plume radiative fraction, with simulation data compared to experimental results published by Blanchat et al. (2010). The EDC model produced high soot concentrations near the pool surface, resulting in a significant shielding effect. In contrast, the flamelet combustion model has different kinetic limiters for soot production, leading to soot formation higher in the flame, where it can be more readily oxidized. Flamelet models incorporating both acetylene and benzene contributions to soot growth demonstrated better agreement with experimental data. To assess the sensitivity of pool radiative heat flux to soot mass concentration, we used a one-dimensional approximation over a wide range of soot concentrations and compared it to three-dimensional solutions obtained using Discrete Ordinates. Furthermore, this analysis illustrates the impacts of soot shielding and Turbulence Radiation Interactions (TRI), with TRI emission enhancement factors predicted to be in the range of 3–5 near the edges of the plume.

Computational fluid dynamics↗

Preconditioning of overcast-to-broken cloud transitions by riming in marine cold air outbreaks

Marine cold air outbreaks (CAOs) commonly form overcast cloud decks that transition into broken cloud fields downwind, dramatically altering the local radiation budget. In this study, we investigate the impact of frozen hydrometeors on these transitions. We focus on a CAO case in the NW Atlantic, the location of the multi-year flight campaign ACTIVATE (Aerosol Cloud meTeorology Interactions oVer the western ATlantic Experiment). We use MERRA-2 (Modern-Era Retrospective analysis for Research and Applications, version 2) reanalysis fields to drive large eddy simulations with mixed-phase two-moment microphysics in a Lagrangian framework. We find that transitions are triggered by substantial rain (rainwater paths >25 g m−2), and only simulations that allow for aerosol depletion result in sustained breakups, as observed. Using a range of diagnostic ice nucleating particle concentrations, Ninp, we find that increasing ice progressively accelerates transitions, thus abbreviating the overcast state. Ice particles affect the cloud-topped boundary layer evolution, primarily through riming-related processes prior to substantial rain, leading to (1) a reduction in cloud liquid water, (2) early consumption of cloud condensation nuclei, and (3) early and light precipitation cooling and moistening below cloud. We refer to these three effects collectively as “preconditioning by riming”. Greater boundary layer aerosol concentrations available as cloud condensation nuclei (CCN) delay the onset of substantial rain. However, cloud breakup and low CCN concentration final stages are found to be inevitable in this case, due, primarily, to liquid water path buildup. An ice-modulated cloud transition speed suggests the possibility of a negative cloud–climate feedback. To address prevailing uncertainties in the model representation of mixed-phase processes, the magnitude of ice formation and riming impacts and, thereby, the strength of an associated negative cloud–climate feedback process, requires further observational evaluation by targeting riming hot spots with in situ imaging probes that allow for both the characterization of ice particles and abundance of supercooled droplets.

Marine cold air outbreaks↗

Analysis of direct and indirect noise in a next-generation aviation gas turbine combustor

A large-eddy simulation (LES) of a next-generation combustor is performed to examine effects of this combustor concepts on direct and indirect combustion noise characteristics. Direct noise is computed considering the unsteady heat release predictions while entropy fluctuations in the downstream part of the combustor are used to estimate indirect noise through the transfer function of the outlet nozzle. A low-order acoustic reconstruction technique, which utilizes the Green’s function of the configuration, is developed to compute the acoustics inside the combustor. Comparisons to experimental results are reported, showing the reasonable accuracy of the LES and combustion noise computations. The direct noise spectrum peaks at frequencies around 3 kHz because of the fast timescales associated with the chemical reactions inside the combustor. In addition, the indirect noise is dominant at low frequencies, i.e., less than 400 Hz, because of the characteristics of the entropy spectrum at the outlet nozzle. Further, compared to a conventional rich-quench-lean (RQL) combustor configuration, significant differences in the acoustics are observed, and are explained by two specific technological novelties. Firstly, the direct noise spectrum peaks at significantly higher frequencies due to the combustor’s compactness. Secondly, the entropy inhomogeneties downstream of the combustion region are an order of magnitude smaller in amplitude due to the lack of dilution with cold air. This leads to a significant reduction in the indirect noise amplitude at low frequencies, where it is the dominant source of noise. These results suggest opportunities for advanced combustion technologies to achieve substantial reductions in combustion-related noise.

42 ENGINEERING↗

On the interaction of a wind turbine wake with a conventionally neutral atmospheric boundary layer

In this work, we investigate the dynamics of wind turbine tip-vortex breakdown in a conventionally neutral atmospheric boundary layer (ABL). To this end, high-resolution data are collected from large-eddy simulations of a wind turbine operating within a neutral ABL and studied by means of proper orthogonal decomposition (POD) and Fourier analysis. The high resolution of the generated data in both space and time allows us to gain insight into the tip-vortex breakdown mechanisms by (i) capturing the energy modes of the coherent structures, (ii) studying their contribution to the tip-vortex breakdown through their power spectra functions and mean kinetic energy (MKE) flux, and (iii) analysing the growth rate of each contributing perturbation frequency along tip vortices. Our analysis shows that under a fully turbulent scenario, the growth rate of perturbations along the tip vortices is largest for low wave numbers, i.e. long-wave perturbations. Additionally, the MKE flux reaches its highest value at two diameters downstream of the rotor plane, a behaviour that can be attributed to the coexistence of multiple interacting POD modes, with the streamwise vortex roller mode being the primary contributor to the total MKE flux budget, contributing approximately . Finally, comparisons with a laminar, uniform flow scenario subject to a single-frequency perturbation highlight the differences between the two ambient flow conditions. In the non-turbulent, uniform flow scenario, the growth rate attains its maximum value at a wave number corresponding to the out-of-phase mutual-inductance mechanism, whereas the MKE flux exhibits local minima and maxima along the wake and at different downstream locations depending on the perturbation frequency. Our analyses suggest that the breakdown of the wind turbine tip vortices under a fully turbulent neutral ABL inflow is due to complex interactions across a range of excitation frequencies, in which the mutual-inductance instability may not be the dominant one.

17 WIND ENERGY↗

Transition to turbulence and noise radiation in heated coaxial jet flows

Laminar-turbulent transition and noise radiation of a parametrized set of subsonic coaxial jet flows with a hot primary (core) stream are investigated numerically by Large-Eddy Simulation (LES) and direct noise computation. This study extends our previous research on local linear stability of heated coaxial jet flows by analyzing the nonlinear evolution of initially laminar flows disturbed by a superposition of small-amplitude unstable eigenmodes. First, a baseline configuration is studied to shed light on the flow dynamics of coaxial jet flows. Subsequently, LESs are performed for a range of Mach and Reynolds numbers to systematically analyze the influences of the temperature and the velocity ratios between the primary and the secondary (bypass) stream. The results provide a basis for a detailed analysis of fundamental flow-acoustic phenomena in the considered heated coaxial jet flows. Increasing the primary-jet temperature leads to an increase of fluctuation levels and to an amplification of far-field noise, especially at low frequencies. Strong mixing between the cold bypass stream and the hot primary stream as well as the intermittent character of the flow field at the end of the potential core lead to a pronounced noise radiation at an aft angle of approximately 35{sup ∘}. The velocity ratio strongly affects the shear-layer development and therefore also the noise generation mechanisms. Increasing the secondary-stream velocity amplifies the dominance of outer shear-layer perturbations while the disturbance growth rates in the inner shear layer decrease. Already for r{sub mic} > 40R{sub 1}, where r{sub mic} is the distance from the end of the potential core and R{sub 1} is the core-jet radius, a perfect 1/r{sub mic} decay of the sound pressure amplitudes is observed. The potential-core length increases for higher secondary-stream velocities which leads to a shift of the center of the dominant acoustic radiation in the downstream direction.

42 ENGINEERING↗

Prediction of extinction and reignition in nonpremixed turbulent flames using a flamelet/progress variable model. 2. Application in LES of Sandia flames D and E

An extension of the flamelet/progress variable (FPV) model for the prediction of extinction and reignition is applied in large-eddy simulation (LES) of flames D and E of the Sandia piloted turbulent jet flame series. This model employs a presumed probability density function (PDF), in which the marginal PDF of a reactive scalar is modeled by a statistically most likely distribution. This provides two advantages. First of all, the shape of the distribution depends on chemical and mixing time-scale information, and second, an arbitrary number of moments can be enforced. This model was analyzed in an a priori study in the first part of this work. In the present LES application, the first two moments of mixture fraction and reaction progress variable are used to constrain the shape of the presumed PDF. Transport equations for these quantities are solved, and models for the residual scalar dissipation rates, which appear as unclosed terms in the equations for the scalar variances, are provided. Statistical flow field quantities for axial velocity, mixture fraction, and temperature, obtained from the extended FPV model, are in good agreement with experimental data. Mixture-fraction-conditioned data, conditional PDFs, and burning indices are computed and compared with the delta-function flamelet closure model, which employs a Dirac distribution as a model for the marginal PDF of the reaction progress parameter. The latter model considerably underpredicts the amount of local extinction, which shows that the consideration of second-moment information in the presumed PDF of the reaction progress parameter is important for the accurate prediction of extinction and reignition. Mixture-fraction-conditioned results obtained from the extended FPV model are in good agreement with experimental data; however, the overprediction of the consumption of fuel and oxidizer on the fuel-rich side results in an overprediction of minor species. The predictions for the conditional PDFs and burning indices are in good agreement with measurements. (author)

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

A priori analysis on deep learning of subgrid-scale parameterizations for Kraichnan turbulence

In the present study, we investigate different data-driven parameterizations for large eddy simulation of two-dimensional turbulence in the a priori settings. These models utilize resolved flow field variables on the coarser grid to estimate the subgrid-scale stresses. We use data-driven closure models based on localized learning that employs a multilayer feedforward artificial neural network with point-to-point mapping and neighboring stencil data mapping, and convolutional neural network fed by data snapshots of the whole domain. The performance of these data-driven closure models is measured through a probability density function and is compared with the dynamic Smagorinksy model (DSM). The quantitative performance is evaluated using the cross-correlation coefficient between the true and predicted stresses. We analyze different frameworks in terms of the amount of training data, selection of input and output features, their characteristics in modeling with accuracy, and training and deployment computational time. We also demonstrate computational gain that can be achieved using the intelligent eddy viscosity model that learns eddy viscosity computed by the DSM instead of subgrid-scale stresses. Here, we detail the hyperparameters optimization of these models using the grid search algorithm.

42 ENGINEERING↗

Scalar Flux Profiles in the Unstable Atmospheric Surface Layer Under the Influence of Large Eddies: Implications for Eddy Covariance Flux Measurements and the Non‐Closure Problem

How convective boundary-layer (CBL) processes modify fluxes of sensible (SH) and latent (LH) heat and CO 2 (F c ) in the atmospheric surface layer (ASL) remains a recalcitrant problem. Here, large eddy simulations for the CBL show that while SH in the ASL decreases linearly with height regardless of soil moisture conditions, LH and F c decrease linearly with height over wet soils but increase with height over dry soils. This varying flux divergence/convergence is regulated by changes in asymmetric flux transport between top-down and bottom-up processes. Such flux divergence and convergence indicate that turbulent fluxes measured in the ASL underestimate and overestimate the “true” surface interfacial fluxes, respectively. While the non-closure of the surface energy balance persists across all soil moisture states, it improves over drier soils due to overestimated LH. The non-closure does not imply that F c is always underestimated; F c can be overestimated over dry soils despite the non-closure issue.

Geology↗