Subgrid Terms Research Articles

A filtered Rankine–Hugoniot method for consistent stochastic particle LES-FDF modelling of combustion in high-speed flows

A new model consistency scheme based on the filtered form of the Rankine–Hugoniot (R–H) relation is developed for hybrid Eulerian/Lagrangian Large Eddy Simulation-Filtered Density Function (LES-FDF) methods to model combustion in compressible high-speed flows. The approach particularly pertains to the evolution of the FDF stochastic Lagrangian particles in enthalpy space, accounting for the subgrid effects in the filtered R–H relation to ensure stable solutions and model consistency with the total energy field solved by an Eulerian compressible finite volume scheme. An analytical form of the subgrid term appearing in the filtered R–H relation is derived and it is shown to have a significant effect on the ability to accurately produce the post-shock thermochemical state. Subsequently, two novel numerical schemes are proposed based on either a modelled subgrid scale kinetic energy or a relaxation over a physical timescale. In testing for 1D normal shock cases, the former approach is found to be strongly dependent on the filter width and Mach number, while the latter approach produces consistent results. The relaxation method is further tested for a turbulent 3D shock tube case and a 1D detonation case and found to produce numerically consistent and accurate temperature and species fields. It is shown that the subgrid term is critical for accurately and consistently predicting the temperature fields in both cases and for achieving the correct detonation structure involving a leading shock and autoignition in the second case.Novelty and significance statementLES-FDF methods have not been widely used for high-speed combusting flows with strong shocks. Such flows are common in aerospace combustors and accurate modelling of these is essential for their optimisation, especially with rapid transition to alternative fuels. The first novelty is the development of a filtered Rankine–Hugoniot model to ensure energy consistency between the Eulerian LES and the FDF modelled through Lagrangian particles. The inclusion of a subgrid kinetic energy term in the Lagrangian particle energy equation is shown to be essential for producing consistent solutions and accurate prediction of thermodynamics states through discontinuities. This subgrid term is not closed and, therefore, the second novelty lies in proposing a practical numerical scheme which accounts for it. The consistent model and numerical scheme is validated in canonical shock tube and detonation cases.

Statistical Dynamics and Subgrid Modelling of Turbulence: From Isotropic to Inhomogeneous

Turbulence is the most important, ubiquitous, and difficult problem of classical physics. Feynman viewed it as essentially unsolved, without a rigorous mathematical basis to describe the statistical dynamics of this most complex of fluid motion. However, the paradigm shift came in 1959, with the formulation of the Eulerian direct interaction approximation (DIA) closure by Kraichnan. It was based on renormalized perturbation theory, like quantum electrodynamics, and is a bare vertex theory that is manifestly realizable. Here, we review some of the subsequent exciting achievements in closure theory and subgrid modelling. We also document in some detail the progress that has been made in extending statistical dynamical turbulence theory to the real world of interactions with mean flows, waves and inhomogeneities such as topography. This includes numerically efficient inhomogeneous closures, like the realizable quasi-diagonal direct interaction approximation (QDIA), and even more efficient Markovian Inhomogeneous Closures (MICs). Recent developments include the formulation and testing of an eddy-damped Markovian anisotropic closure (EDMAC) that is realizable in interactions with transient waves but is as efficient as the eddy-damped quasi-normal Markovian (EDQNM). As a similarly efficient closure, the realizable eddy-damped Markovian inhomogeneous closure (EDMIC) has been developed. Moreover, we present subgrid models that cater for the complex interactions that occur in geophysical flows. Recent progress includes the determination of complete sets of subgrid terms for skilful large-eddy simulations of baroclinic inhomogeneous turbulent atmospheric and oceanic flows interacting with Rossby waves and topography. The success of these inhomogeneous closures has also led to further applications in data assimilation and ensemble prediction and generalization to quantum fields.

Volume average of two-fluid RANS equations and a priori estimation of subgrid terms on subcooled boiling experiments

A statistic-volume averaged two-fluid model, convenient for subchannel and system codes in nuclear industry, is established from RANS equations of Neptune_CFD code and leads to the appearance of subgrid terms. A database of simulations is made on experiments with flow representative of nuclear reactor cores with cylindrical (KIT) and subchannel (PSBT) configurations. The subgrid terms are then compared to the convection terms on the database of 82 test cases. Concerning momentum and energy equations, subgrid terms reach respectively 7.5% and 16% of convection terms. Overall, the comparison shows that dispersion terms are always greater than turbulence terms, but both remain in the same order of magnitude, and are not negligible compared to convection. Moreover, the geometry has an impact on the dispersion terms. This observation suggests that caution should be exercised when using correlation established on a different geometry.

Toward discretization-consistent closure schemes for large eddy simulation using reinforcement learning

This study proposes a novel method for developing discretization-consistent closure schemes for implicitly filtered large eddy simulation (LES). Here, the induced filter kernel and, thus, the closure terms are determined by the properties of the grid and the discretization operator, leading to additional computational subgrid terms that are generally unknown in a priori analysis. In this work, the task of adapting the coefficients of LES closure models is thus framed as a Markov decision process and solved in an a posteriori manner with reinforcement learning (RL). This optimization framework is applied to both explicit and implicit closure models. The explicit model is based on an element-local eddy viscosity model. The optimized model is found to adapt its induced viscosity within discontinuous Galerkin (DG) methods to homogenize the dissipation within an element by adding more viscosity near its center. For the implicit modeling, RL is applied to identify an optimal blending strategy for a hybrid DG and finite volume (FV) scheme. The resulting optimized discretization yields more accurate results in LES than either the pure DG or FV method and renders itself as a viable modeling ansatz that could initiate a novel class of high-order schemes for compressible turbulence by combining turbulence modeling with shock capturing in a single framework. All newly derived models achieve accurate results that either match or outperform traditional models for different discretizations and resolutions. Overall, the results demonstrate that the proposed RL optimization can provide discretization-consistent closures that could reduce the uncertainty in implicitly filtered LES.

Physically evocative meso-informed sub-grid source term for energy localization in shocked heterogeneous energetic materials

Reactive burn models for heterogeneous energetic materials (EMs) must account for chemistry as well as microstructure to predict shock-to-detonation transition (SDT). Upon shock loading, the collapse of individual voids leads to ignition of hotspots, which then grow and interact to consume the surrounding material. The sub-grid dynamics of shock-void interactions and hotspot development are transmitted to macro-scale SDT calculations in the form of a global reactive “burn model.” This paper presents a physically evocative model, called meso-informed sub-grid source terms for energy localization (MISSEL), to close the macro-scale governing equations for calculating SDT. The model parameters are explicitly related to four measurable physical quantities: two depending on the microstructure (the porosity ϕ and average pore size D¯void), one depending on shock–microstructure interaction (the fraction of critical voids ξcr), and the other depending on the chemistry (the burn front velocity Vhs). These quantities are individually quantifiable using a small number of rather inexpensive meso-scale simulations. As constructed, the model overcomes the following problems that hinder the development of meso-informed burn models: (1) the opacity of more sophisticated surrogate/machine-learning approaches for bridging meso- and macro-scales, (2) the rather large number of high-resolution mesoscale simulations necessary to train machine-learning algorithms, and (3) the need for calibration of many free parameters that appear in phenomenological burn models. The model is tested against experimental data on James curves for a specific class of pressed 1,3,5,7-tetranitro-1,3,5,7-tetrazoctane materials. The simple, evocative, and fast-to-construct MISSEL model suggests a route to develop frameworks for physics-informed, simulation-derived meso-informed burn models.

Variational multiscale super‐resolution: A data‐driven approach for reconstruction and predictive modeling of unresolved physics

AbstractThe variational multiscale (VMS) formulation formally segregates the evolution of the coarse‐scales from the fine‐scales. VMS modeling requires the approximation of the impact of the fine‐scales in terms of the coarse‐scales. In linear problems, our formulation reduces the problem of learning the subscales to learning the projected element Green's function basis coefficients. For this approximation, a special neural‐network structure—the variational super‐resolution N‐N (VSRNN)—is proposed. The VSRNN constructs a super‐resolution model of the unresolved scales as a sum‐of‐the‐products of individual functions of coarse‐scales and physics‐informed parameters. Combined with a set of locally nondimensional features obtained by normalizing the input coarse‐scale and output subscale basis coefficients, the VSRNN provides a general framework for the discovery of closures for different Galerkin discretizations. By training it on a sequence of projected data and using the subscale to compute the continuous Galerkin subgrid terms, and the super‐resolved state to compute the discontinuous Galerkin fluxes, we improve the optimality and the accuracy of these methods for the convection‐diffusion and linear advection problems. Finally, we demonstrate that the VSRNN allows generalization to out‐of‐sample initial conditions and nondimensional numbers.

Subgrid Parameterization of Eddy, Meanfield and Topographic Interactions in Simulations of an Idealized Antarctic Circumpolar Current

AbstractThe ocean circulation dynamics can be represented as a high‐dimensional multi‐scale nonlinear system with inhomogenous meanfields and topography. Numerical simulations of the ocean dynamics resolve the large scales of motion on a computational grid, with the unresolved subgrid interactions parameterized. These simulations are highly dependent upon the grid resolution, unless the subgrid terms are appropriately parameterized. There are five fundamental classes of subgrid interactions: eddy‐eddy; eddy‐topographic; eddy‐meanfield; meanfield‐meanfield; and meanfield‐topographic. Scale dependent parameterizations representing each of these interaction classes are presented here in oceanic flows for the first time. Subgrid parameterizations are calculated and validated in baroclinic quasi‐geostrophic simulations of idealized Antarctic Circumpolar Current flows with representative mean currents and ocean floor topography. The parameterization coefficients are derived from the coarse grained statistics of high resolution reference simulations in spectral space. Stochastic and deterministic parameterizations are developed for the eddy‐eddy interactions, and deterministic forms for the remaining classes. The kinetic energy spectra and meanfield resulting from large eddy simulations (LES) adopting these coefficients accurately replicate those of the reference simulation. The eddy‐eddy interactions are dominant, but all classes need to be parameterized for best results. For LES where baroclinic instability is explicitly resolved the stochastic variants out‐perform the deterministic ones across all scales. When baroclinic instability is not explicitly resolved, the stochastic variants out‐perform the deterministic cases at the large scales, but introduce some distortions at the smallest resolved scales. This study provides an assessment of the relevant strengths of the subgrid interaction classes in an idealized yet representative ocean.

Large Eddy Simulation of Combustion for High-Speed Airbreathing Engines

Large Eddy Simulation (LES) has rapidly developed into a powerful computational methodology for fluid dynamic studies, between Reynolds-Averaged Navier–Stokes (RANS) and Direct Numerical Simulation (DNS) in both accuracy and cost. High-speed combustion applications, such as ramjets, scramjets, dual-mode ramjets, and rotating detonation engines, are promising propulsion systems, but also challenging to analyze and develop. In this paper, the building blocks needed to perform LES of high-speed combustion are reviewed. Modelling of the unresolved, subgrid terms in the filtered LES equations is highlighted. The main families of combustion models are presented, focusing on finite-rate chemistry models. The density-based finite volume method and the reaction mechanisms commonly employed in LES of high-speed H2-air combustion are briefly reviewed. Three high-speed combustor applications are presented: an experiment of supersonic flame stabilization behind a bluff body, a direct connect facility experiment as a transition case from ramjet to scramjet operation mode, and the STRATOFLY MR3 Small-Scale Flight Experiment. Several combinations of turbulence and combustion models are compared. Comparisons with experiments are also provided when available. Overall, the results show good agreement with experimental data (e.g., shock train, mixing, wall heat flux, transition from ramjet to scramjet operation mode).

Frame invariant neural network closures for Kraichnan turbulence

Numerical simulations of geophysical and atmospheric flows have to rely on parameterizations of subgrid scale processes due to their limited spatial resolution. Despite substantial progress in developing parameterization (or closure) models for subgrid scale (SGS) processes using physical insights and mathematical approximations, they remain imperfect and can lead to inaccurate predictions. In recent years, machine learning has been successful in extracting complex patterns from high-resolution spatio-temporal data, leading to improved parameterization models, and ultimately better coarse grid prediction. However, the inability to satisfy known physics and poor generalization hinders the application of these models for real-world problems. In this work, we put forth a frame invariant closure approach to improve the accuracy and generalizability of deep learning-based subgrid scale closure models by embedding physical symmetries directly into the structure of the neural network. Specifically, we utilized specialized layers within the convolutional neural network in such a way that desired constraints are theoretically guaranteed without the need for any regularization terms. We demonstrate our framework for a two-dimensional decaying turbulence test case mostly characterized by the forward enstrophy cascade. We show that our frame invariant SGS model (i) accurately predicts the subgrid scale source term, (ii) respects the physical symmetries such as translation, Galilean, and rotation invariance, and (iii) is numerically stable when implemented in coarse-grid simulation with generalization to different initial conditions and Reynolds number. This work opens up a possibility of connecting physics-based theories and data-driven modeling paradigms, and thus represents a promising step towards the development of physically consistent data-driven turbulence closure models.

Large eddy simulation of two separated hypersonic shock/turbulent boundary layer interactions

We present a detailed statistical description of turbulence in the high-fidelity Large Eddy Simulation (LES) of hypersonic shock-and-turbulent boundary layer interactions with strong separation. Our high-fidelity LES method uses a 4th-order, bandwidth-optimized Weighted Essentially Non-Oscillatory (WENO) scheme combined with a robust shock-capturing method and dynamic mixed model formulations for the closure of subgrid scale terms. Our mixed models, which contain a scale-similar term in addition to the purely dissipative one, are shown to accurately and significantly improve separation prediction compared to purely dissipative closures, such as the dynamic eddy viscosity formulation.

Understanding loss generation mechanisms in a centrifugal pump using large eddy simulation

Water pumps are amongst the most frequently used turbomachines, which consume about 9% of the global energy production. This paper provides insight on the loss generation mechanisms in inline centrifugal pumps operating under realistic conditions through a detailed analysis based on large eddy simulation (LES). The equations for the resolved, sub-grid and wall entropy generation terms are derived using the LES filtering and Smagorinsky-Lilly formalism. Specific exergy is calculated at the grid points. Entropy generation and exergy transfer within the flow domain are visualised to highlight the loss generation mechanisms within the pump. Results show that there is a strong component interaction. The double-bended inlet pipe initiates Dean-vortices at the eye of the impeller. This asymmetric inflow coupled with the disturbances at the cut-off cause a non-uniform flow through the blade passages. Other drivers of the asymmetrical loading of the blade passages are high leakage and unsteady and asymmetrical mass and exergy transfer between the dead zone. The paper also investigates the effect of the grid resolution on the first and second order statistics of the flow field and its turbulent characteristics. A thorough evaluation of the discretisation error is presented to provide a measure for the required grid density and demonstrate the extent and impact of the temporal and spatial discretisation error in centrifugal turbomachinery.

Finding Closure Models to Match the Time Evolution of Coarse Grained 2D Turbulence Flows Using Machine Learning

Machine learning is used to develop closure terms for coarse grained model of two-dimensional turbulent flow directly from the coarse grained data by ensuring that the coarse-grained flow evolves in the correct way, with no need for the exact form of the filters or an explicit expression of the subgrid terms. The closure terms are calculated to match the time evolution of the coarse field and related to the average flow using a Neural Network with a relatively simple structure. The time dependent coarse grained flow field is generated by filtering fully resolved results and the predicted coarse field evolution agrees well with the filtered results in terms of instantaneous vorticity field in the short term and statistical quantities (energy spectrum, structure function and enstropy) in the long term, both for the flow used to learn the closure terms and for flows not used for the learning. This work shows the potential of using data-driven method to predict the time evolution of the large scales, in a complex situation where the closure terms may not have an explicit expression and the original fully resolved field is not available.

Effect of orographic gravity wave drag on Northern Hemisphere climate in transient simulations of the last deglaciation

Long transient simulations of the last deglaciation are increasingly being performed to identify the drivers of multiple rapid Earth system changes that occurred in this period. Such simulations frequently prescribe temporal variations in ice sheet properties, which can play an important role in controlling atmospheric and surface climate. To preserve a model’s standard performance in simulating climate, it is common to apply time dependent orographic variations, including parameterised sub-grid scale orographic variances, as anomalies from the pre-industrial state. This study investigates the causes of two abrupt climate change events in the Northern Hemisphere extratropics occurring between 16 and 14 thousand years ago in transient simulations of the last deglaciation from the Hadley Centre coupled general circulation model (HadCM3). One event is characterized by regional Northern Hemisphere changes comprising a centennial scale cooling of ~ 10 °C across Fennoscandia followed by rapid warming in less than 50 years as well as synchronous shifts in the Northern Annular Mode. The second event has comparable but temporally reversed characteristics. Sensitivity experiments reveal the climate anomalies are exclusively caused by artificially large values of orographic gravity wave drag, resulting from the combined use of the orographic anomaly method along with a unique inclusion of transient orography that linearly interpolates between timesteps in the ice sheet reconstruction. Palaeoclimate modelling groups should therefore carefully check the effects of their sub-grid scale orographic terms in transient palaeoclimate simulations with prescribed topographic evolution.

Numerical simulations of gas explosion using Porosity Distributed Resistance approach Part −1: Validation against small-scale experiments

A new model, PDRFOAM, has been developed for the prediction of gas explosions in congested plant. It uses the PDR approach to model the effect of small-scale obstacles, e.g., pipes and vessels on flame propagation and on explosion overpressure. While the effects of large obstacles i.e., that is the large scales are explicitly resolved. The equations for mass, momentum, enthalpy and Favre-averaged regress variable are solved. In addition, porosity modified standard k-ε turbulence model and the transport equations for the flame wrinkling parameter are solved. The model PDRFOAM, is built as a new application in OpenFOAM, suite of models. OpenFOAM is an open-source CFD package of routines for solution of systems of partial differential equations.In addition, to the PDRFOAM model, the CADPDR program was developed which generates various fields (volume blockage, area blockages, surface area, sub-grid drag and turbulence generation parameters) needed by PDRFOAM. CADPDR needs as input obstacle files that list coordinates and dimension of the obstacles e.g. pipes and vessels. The PDRFOAM code solves porosity-modified momentum and continuity equations with sub-grid source terms. The combustion model in PDRFOAM is based on flame area transport. The turbulent burning velocity correlation used is based on Markstein and Karlowitz number. Flame area generation due to the folding of the flame around obstacles is explicitly modelled.This paper presents the formulation of PDRFOAM and validation of the PDRFOAM code against three series of small- and medium-scale experiments i.e., ERGOS, MERGE and Buxton S-Series experiments. A total of more than 150 experiments were used for validation which includes variation in blockage ratio, grid pitch, size of the congested region, equivalence ratio, confinement, partial fill, obstacle diameter, different obstacle shapes and three different fuels. The model is compared against flame position, flame speed as a function of time and maximum overpressure obtained from experiments. Simulations predict the experimental trends of increase in overpressure with increase in blockage ratio, laminar burning velocity, partial fill of the gas cloud, size of the congested region, confinement, grid pitch and obstacle diameter. It also predicts the trends of increase in overpressure with increase in equivalence ratio, variation of maximum overpressure with ignition location, change in obstacle shape and obstacle configuration. The maximum overpressure predicted by simulations (see Fig. 1) is in general within the uncertainty of a factor of two.

Computations and measurements of the global drag force on a Tension-Leg Platform

We present experimental and computational results for the global drag force on a model Tension-Leg Platform (TLP) in uniform current for various values of the Reynolds number in the subcritical regime (Re = 104 - 1.06×105). The objective of the experiments, which were performed in a towing tank on a representative TLP configuration, was to provide data suitable for validating numerical model predictions. The purpose of the simulations was to determine the extent to which Large-Eddy Simulations (LES), when used as a truly predictive tool, can be relied upon to provide reliable estimates of the total drag force. Thus the simulations were performed only once with no further computations performed in order to bring about closer agreement with the measurements. The dependence of the computed results on the numerical grid was checked using the Grid-Convergence Index (GCI) method applied to results from benchmark flows. These results also served to assess the dependence on the value of the Smagorinsky constant in the model for the sub-grid scale terms. It was found that, at the highest value of Reynolds number considered, the variation between the measured and predicted values of the global drag coefficient was under 10% - a result which is in line with the limitations inherent in both.

Sub-grid Scale Modelling and a-Posteriori Tests with a Morphology Adaptive Multifield Two-Fluid Model Considering Rising Gas Bubbles

The predictive simulation of gas–liquid multiphase flows at industrial scales reveals the challenging task to consider turbulence and interfacial structures, which span a large range of length scales. For simulation of relevant applications, a hybrid model can be utilised, which combines the Euler–Euler model for the description of small interfacial structures with a volume-of-fluid model as a scale-resolving multiphase approach. Such a hybrid model needs to be able to simulate interfaces, which are hardly resolved on a coarse numerical grid. The goal of this work is to improve the prediction of interfacial gas–liquid flows on a numerical grid with comparably large grid spacing. From the low-pass filtering of the two-fluid model five unclosed sub-grid scale terms arise. The convective and the surface tension part of the aforementioned contributions are individually modelled with multiple closure formulations. Those models are a-posteriori assessed in cases of two- and three-dimensional gas bubbles rising in stagnant liquid. It is shown, that the chosen closure modelling approach is suitable to improve the predictive power of the numerical model utilised in this work. Hence, simulation results on comparably coarse grids are changed towards results obtained with higher spatial resolution.

Direct simulations and subgrid modeling of turbulent channel flows asymmetrically heated from both walls

Thermal large-eddy simulations (T-LES) and a direct numerical simulation are carried out in a bi-periodical channel with hot and cold wall temperatures of, respectively, 900 and 1300 K. The mean fluid temperature is lowered below the cold wall temperature thanks to a heat source, resulting in a both walls heating of the fluid. The hot and cold wall friction Reynolds numbers are, respectively, 640 and 1000. These conditions are representative of the working conditions of gas-pressurized solar receiver of solar power tower. The low Mach number Navier–Stokes equations are solved. The coupling between the dynamic and the temperature effects is considered. In the T-LES, both the momentum convection and the density–velocity correlation subgrid terms are modeled. Functional models, structural models, and mixed models are considered. A tensorial version of the anisotropic minimum-dissipation (AMD) model is also investigated. The Quick and the second-order-centered schemes are tested for the discretization of the mass convection term. First, an overview of the results of 17 T-LES on first- and second-order statistics is proposed. It permits selecting 6 of these simulations for a detailed analysis consisting in the investigation of profiles of mean quantities and turbulent correlations. Particular attention is given to the wall heat fluxes because they are a critical point for the design and the optimization of solar receivers. Overall, the first-order statistics are better predicted than the second-order's. The tensorial AMD model takes advantage of the classical AMD model properties and better reproduces the anisotropy of the flow thanks to its formulation. The tensorial AMD model produces the most reliable and efficient results among the considered models.

A finite-volume method for simulating contact lines on unstructured meshes in a conservative level-set framework

Accurate numerical simulation of moving contact lines on complex boundaries with surface wettability effect remains a challenging problem. In this paper, we introduce a robust and accurate method to perform 3D contact line simulations on unstructured meshes with an imposed contact angle. The contact angle is imposed through a sub-grid scale curvature term and the contact line motion is enabled thanks to partial slip on the wall. Moreover, an original strategy has been designed to improve normal and curvature computation at contact line from the level-set field. The whole method is validated on 2D and 3D test cases and shows good mass conservation properties. The drop detachment from a horizontal fiber due to gravity or surface tension is then investigated. For this purpose, dynamic mesh adaptation is used to keep high resolution around the interface with moderate number of cells. These realistic cases demonstrate the ability of the numerical method to handle surface wettability effects on resolved complex geometries.

Subgrid scale modeling considerations for large eddy simulation of supercritical turbulent mixing and combustion

This paper presents a systematic investigation of large eddy simulation (LES) and subgrid scale (SGS) modeling with application to transcritical and supercritical turbulent mixing and combustion. There remains uncertainty about the validity of extending the LES formalism developed for low-pressure, ideal-gas flows to simulations of high-pressure real-fluid flows. To address this concern, we reexamine the LES theoretical framework and the underlying assumptions in the context of real-fluid mixing and combustion. Two-dimensional direct numerical simulations of nonreacting and reacting mixing layers of gaseous methane and liquid oxygen in the thermodynamically transcritical and supercritical fluid regimes are performed. The computed results are used to evaluate the exact terms in the LES governing equations and associated SGS models. Order of magnitude analysis of the exact filtered and subgrid terms in the LES equations and a priori analysis of the simplifications are performed at different filter widths. It is shown that several of these approximations do not hold for supercritical turbulent mixing. Subgrid scale terms, which are neglected in the LES framework for ideal-gas flows, become significant in magnitude compared to the other leading terms in the governing equations. In particular, the subgrid term arising from the filtering of the real-fluid equation of state is shown to be important.

Investigation of thermal large-eddy simulation approaches in a highly turbulent channel flow submitted to strong asymmetric heating

This study deals with thermal large-eddy simulation (T-LES) of anisothermal turbulent channel flow in the working conditions of solar receivers used in concentrated solar power towers. The flow is characterized by high-temperature levels and strong heat fluxes. The hot and cold friction Reynolds numbers of the simulations are, respectively, 630 and 970. The Navier–Stokes equations are solved under the low-Mach number approximation and the thermal dilatation is taken into account. The momentum convection and the density–velocity correlation subgrid terms are modeled. Functional, structural, and mixed subgrid-scale models are investigated. A tensorial version of the classical anisotropic minimum-dissipation (AMD) model is studied and produces good results. A Quick scheme and a second-order-centered scheme are tested for the discretization of the mass convection term. First, a global assessment of 22 large-eddy simulations is proposed, then six are selected for a careful analysis including profiles of mean quantities and fluctuation values as well as a comparison of instantaneous fields. Probability density functions of wall heat fluxes are plotted. The results point out that T-LESs performed with the Quick scheme tend to underestimate the wall heat flux whereas the second-order-centered scheme significantly improves its estimation. T-LESs tend to overestimate the peaks of velocity correlations. When regarding the dimensionless profiles of fluctuations, the tensorial AMD model provides better results than the other assessed models. For the heat flux estimation, the best agreement is found with the AMD model combined with the second-order-centered scheme.