Turbulence-resolving Simulations Research Articles

Scale-Resolving Simulations of Turbulent Flow Retaining the Exact Blade Count With the Time-Inclined Method

Abstract Simulations of the T106 low-pressure turbine linear cascade with incoming wakes have been conducted to demonstrate the suitability of the time-inclined method for retaining the exact blade count in scale-resolving simulations. The time-inclined method has been implemented into a high-order unstructured time-marching code based on the flux reconstruction method. The pitch ratio between the upstream incoming wakes and the cascade is not a small integer, and the time-inclined method is used to reduce simulations to a single-passage computational domain. Three-way comparisons have been generated: the solution of the direct periodic full domain, spanning several blade passages, is compared to the single-passage time-inclined solution and to a single-passage solution that approximates the pitch ratio to the nearest integer. Two sets of virtual experiments, which differ in modeling of the incoming perturbations, are reported. First, the immersed boundary approach is used to introduce a cascade of moving bars that act as the trailing edges of the preceding row and generate incoming wakes. The second set of simulations introduces wake-like perturbations at the inlet section of the computational domain. The present work shows that the time-inclined method, retaining the exact blade count, can produce very accurate results for turbulence-resolving simulations. Moreover, the time-inclined method outperforms the standard single-passage methodology in most of the variables of interest and is equivalent in the rest.

The Role of Turbulence in an Intense Tropical Cyclone: Momentum Diffusion, Eddy Viscosities, and Mixing Lengths

Abstract Improved representation of turbulent processes in numerical models of tropical cyclones (TCs) is expected to improve intensity forecasts. To this end, the authors use a large-eddy simulation (with 31-m horizontal grid spacing) of an idealized category 5 TC to understand the role of turbulent processes in the inner core of TCs and their role on the mean intensity. Azimuthally and temporally averaged budgets of the momentum fields show that TC turbulence acts to weaken the maximum tangential velocity, diminish the strength of radial inflow into the eye, and suppress the magnitude of the mean eyewall updraft. Turbulent flux divergences in both the vertical and radial directions are shown to influence the TC mean wind field, with the vertical being dominant in most of the inflowing boundary layer and the eyewall (analogous to traditional atmospheric boundary layer flows), while the radial becomes important only in the eyewall. The validity of the downgradient eddy viscosity hypothesis is largely confirmed for mean velocity fields, except in narrow regions which generally correspond to weak gradients of the mean fields, as well as a narrow region in the eye. This study also provides guidance for values of effective eddy viscosities and vertical mixing length in the most turbulent regions of intense TCs, which have rarely been measured observationally. A generalized formulation of effective eddy viscosity (including the Reynolds normal stresses) is presented. Significance Statement This study uses a turbulence-resolving simulation of a category 5 tropical cyclone to understand the role of turbulence in intense storms. Results show that turbulence clearly modulates storm structure and intensity. This study provides guidance for the values of turbulent quantities (which are usually parameterized in comparatively coarse operational TC forecast models) in scarcely observed regions of intense storms. Furthermore, a complete formulation of the effective eddy viscosities is proposed, incorporating contributions from typically ignored Reynolds normal stress terms.

BoundaryLayerDynamics.jl v1.0: a modern codebase for atmospheric boundary-layer simulations

Abstract. We present BoundaryLayerDynamics.jl, a new code for turbulence-resolving simulations of atmospheric boundary-layer flows as well as canonical turbulent flows in channel geometries. The code performs direct numerical simulation as well as large-eddy simulation using a hybrid (pseudo)spectral and finite-difference approach with explicit time advancement. Written in Julia, the code strives to be flexible and adaptable without sacrificing performance, and extensive automated tests aim to ensure that the implementation is and remains correct. We show that the simulation results are in agreement with published results and that the performance is on par with an existing Fortran implementation of the same methods.

Rotating Horizontal Convection

Global differences of temperature and buoyancy flux at the ocean surface are responsible for small-scale convection at high latitudes, global overturning, and the top-to-bottom density difference in the oceans. With planetary rotation the convection also contributes to the large-scale horizontal, geostrophic circulation, and it crucially involves a 3D linkage between the geostrophic circulation and vertical overturning. The governing dynamics of such a surface-forced convective flow are fundamentally different from Rayleigh–Bénard convection, and the role of buoyancy forcing in the oceans is poorly understood. Geostrophic balance adds to the constraints on transport in horizontal convection, as illustrated by experiments, theoretical scaling, and turbulence-resolving simulations for closed (mid-latitude) basins and an annulus or reentrant zonal (circumpolar) channel. In these geometries, buoyancy drives either horizontal mid-latitude gyre recirculations or a strong Antarctic Circumpolar Current, respectively, in addition to overturning. At large Rayleigh numbers the release of available potential energy by convection leads to turbulent mixing with a mixing efficiency approaching unity. Turbulence-resolving models are also revealing the relative roles of wind stress and buoyancy when there is mixed forcing, and in future work they need to include the effects of turbulent mixing due to energy input from tides.

The Influence of WENO Schemes on Large-Eddy Simulations of a Neutral Atmospheric Boundary Layer

AbstractThis work explores the influence of Weighted Essentially Non-Oscillatory (WENO) schemes on Cloud Model 1 (CM1) large-eddy simulations (LES) of a quasi-steady, horizontally homogeneous, fully developed, neutral atmospheric boundary layer (ABL). An advantage of applying WENO schemes to scalar advection in compressible models is the elimination of acoustic waves and associated oscillations of domain-total vertical velocity. Applying WENO schemes to momentum advection in addition to scalar advection yields no further advantage, but has an adverse effect on resolved turbulence within LES. As a tool designed to reduce numerically generated spurious oscillations, WENO schemes also suppress physically realistic instability development in turbulence-resolving simulations. Thus, applying WENO schemes to momentum advection reduces vortex stretching, suppresses the energy cascade, reduces shear-production of resolved Reynolds stress, and eventually amplifies the differences between the surface-layer mean wind profiles in the LES and the mean wind profiles expected in accordance with the filtered law of the wall (LOTW). The role of WENO schemes in adversely influencing surface-layer turbulence has inspired a concept of anti-WENO (AWENO) schemes to enhance instability development in regions where energy-containing turbulent motions are inadequately resolved by LES grids. The success in reproducing the filtered LOTW via AWENO schemes suggests that improving advection schemes is a critical component toward faithfully simulating near-surface turbulence and dealing with other "Terra Incognita" problems.

Plunging criterion for particle-laden flows over sloping bottoms: Three-dimensional turbulence-resolving simulations

Hyperpycnal flows are observed when the density of a fluid entering into a quiescent environment is greater than that of the ambient fluid. This difference can be due to salinity, temperature, concentration, turbidity, or a combination of them. Over a sloping bottom, the inflowing momentum decreases progressively until a critical stage is reached where the inflow plunges underneath the ambient and flows adjacent to the bed as an underflow density current. In the present work, a new equation is proposed in order to predict the critical depth for plunging, i.e., the plunging criterion. It differs from previous studies since it includes the role of the settling velocity and the bed slope. The high spatiotemporal resolution from twelve original numerical simulations allows us to validate the initial hypotheses established, in addition to numerical and experimental data available in the literature, and good agreement is found between them. A negative value for the mixing coefficient was observed for the first time for the hyperpycnal flow in a tilted channel. This indicates that if the settling velocity of the suspended material is high enough, the submerged flow may lose fluid to the environment (detrainment), instead of incorporating it. The proposed plunging criterion may assist in the design of future experimental or numerical works.

Potential low bias in high-wind drag coefficient inferred from dropsonde data in hurricanes

AbstractUnderstanding momentum exchange at the air-sea interface is important for accurate hurricane predictions and understanding fundamental storm dynamics. One method for estimating air-sea momentum transfer in high winds is the flux-profile method, which infers surface momentum fluxes and the corresponding drag coefficient from mean velocity profiles obtained from either dropsondes or meteorological towers, under the assumption that the boundary-layer wind profile at low altitudes exhibits a logarithmic profile with height. In this study, we use dropsonde data from reconnaissance aircraft, as well as “virtual sondes” from a turbulence-resolving simulation of an intense tropical cyclone, to critically analyze the diagnosis of drag coefficient CD at hurricane-force wind speeds. In particular, the “roll-off” of the drag coefficient, where CD decreases at 10-m wind speeds ¿ 35 m s−1, is called into question based on uncertainty due to relatively low sample size and a lack of robustness of the flux-profile at high winds. In addition, multiple factors appear to favor an underestimate of CD at hurricane-force winds relative to their true values, including uncertainty in the height of recorded dropsonde data, violation of Monin-Obukhov similarity theory near the eyewall, and the short vertical extent of the logarithmic layer. Due to these and other related sources of uncertainty, it is likely that a quantitative limit has been reached in inferring the specific values of u* and CD using the flux-profile method, while at the same time the potential for underestimation may cast doubt on the CD–U10 relationship inferred from this method at high winds.

Near-Wall Flow in Turbomachinery Cascades—Results of a German Collaborative Project

This article provides a summarizing account of the results obtained in the current collaborative work of four research institutes concerning near-wall flow in turbomachinery. Specific questions regarding the influences of boundary layer development on blades and endwalls as well as loss mechanisms due to secondary flow are investigated. These address skewness, periodical distortion, wake interaction and heat transfer, among others. Several test rigs with modifiable configurations are used for the experimental investigations including an axial low speed compressor, an axial high-speed wind tunnel, and an axial low-speed turbine. Approved stationary and time resolving measurements techniques are applied in combination with custom hot-film sensor-arrays. The experiments are complemented by URANS simulations, and one group focusses on turbulence-resolving simulations to elucidate the specific impact of rotation. Juxtaposing and interlacing their results the four groups provide a broad picture of the underlying phenomena, ranging from compressors to turbines, from isothermal to non-adiabatic, and from incompressible to compressible flows.

Efficacy of the Cell Perturbation Method in Large-Eddy Simulations of Boundary Layer Flow over Complex Terrain

A challenge to simulating turbulent flow in multiscale atmospheric applications is the efficient generation of resolved turbulence motions over an area of interest. One approach is to apply small perturbations to flow variables near the inflow planes of turbulence-resolving simulation domains nested within larger mesoscale domains. While this approach has been examined in numerous idealized and simple terrain cases, its efficacy in complex terrain environments has not yet been fully explored. Here, we examine the benefits of the stochastic cell perturbation method (CPM) over real complex terrain using data from the 2017 Perdigão field campaign, conducted in an approximately 2-km wide valley situated between two nearly parallel ridges. Following a typical configuration for multiscale simulation using nested domains within the Weather Research and Forecasting (WRF) model to downscale from the mesoscale to a large-eddy simulation (LES), we apply the CPM on a domain with horizontal grid spacing of 150 m. At this resolution, spurious coherent structures are often observed under unstable atmospheric conditions with moderate mean wind speeds. Results from such an intermediate resolution grid are often nested down for finer, more detailed LES, where these spurious structures adversely affect the development of turbulence on the subsequent finer grid nest. We therefore examine the impacts of the CPM on the representation of turbulence within the nested LES domain under moderate mean flow conditions in three different stability regimes: weakly convective, strongly convective, and weakly stable. In addition, two different resolutions of the underlying terrain are used to explore the role of the complex topography itself in generating turbulent structures. We demonstrate that the CPM improves the representation of turbulence within the LES domain, relative to the use of high-resolution complex terrain alone. During the convective conditions, the CPM improves the rate at which smaller-scales of turbulence form, while also accelerating the attenuation of the spurious numerically generated roll structures near the inflow boundary. During stable conditions, the coarse mesh spacing of the intermediate LES domain used herein was insufficient to maintain resolved turbulence using CPM as the flow develops downstream, highlighting the need for yet higher resolution under even weakly stable conditions, and the importance of accurate representation of flow on intermediate LES grids.

Toward Exascale: Overview of Large Eddy Simulations and Direct Numerical Simulations of Nuclear Reactor Flows with the Spectral Element Method in Nek5000

At the beginning of the last decade, Petascale supercomputers (i.e., computers capable of more than 1 petaFLOP) emerged. Now, at the dawn of exascale supercomputing, we provide a review of recent landmark simulations of portions of reactor components with turbulence-resolving techniques that this computational power has made possible. In fact, these simulations have provided invaluable insight into flow dynamics, which is difficult or often impossible to obtain with experiments alone. We focus on simulations performed with the spectral element method, as this method has emerged as a powerful tool to deliver massively parallel calculations at high fidelity by using large eddy simulation or direct numerical simulation. We also limit this paper to constant-property incompressible flow of a Newtonian fluid in the absence of other body or external forces, although the method is by no means limited to this class of flows. We briefly review the fundamentals of the method and the reasons it is compelling for the simulation of nuclear engineering flows. We review in detail a series of Petascale simulations, including the simulations of helical coil steam generators, fuel assemblies, and pebble beds. Even with Petascale computing, however, limitations for nuclear modeling and simulation tools remain. In particular, the size and scope of turbulence-resolving simulations are still limited by computing power and resolution requirements, which scale with the Reynolds number. In the final part of this paper, we discuss the future of the field, including recent advancements in emerging architectures such as GPU-based supercomputers, which are expected to power the next generation of high-performance computers.

Using 3D turbulence-resolving simulations to understand the impact of surface properties on the energy balance of a debris-covered glacier

Abstract. Debris-covered glaciers account for almost one-fifth of the total glacier ice volume in High Mountain Asia; however, their contribution to the total glacier melt remains uncertain, and the drivers controlling this melt are still largely unknown. Debris influences the properties (e.g. albedo, thermal conductivity, roughness) of the glacier surface and thus the surface energy balance and glacier melt. In this study we have used sensitivity tests to assess the effect of surface properties of debris on the spatial distribution of micrometeorological variables such as wind fields, moisture and temperature. Subsequently we investigated how those surface properties drive the turbulent fluxes and eventually the conductive heat flux of a debris-covered glacier. We simulated a debris-covered glacier (Lirung Glacier, Nepal) at a 1 m resolution with the MicroHH model, with boundary conditions retrieved from an automatic weather station (temperature, wind and specific humidity) and unmanned aerial vehicle flights (digital elevation map and surface temperature). The model was validated using eddy covariance data. A sensitivity analysis was then performed to provide insight into how heterogeneous surface variables control the glacier microclimate. Additionally, we show that ice cliffs are local melt hot spots and that turbulent fluxes and local heat advection amplify spatial heterogeneity on the surface. The high spatial variability of small-scale meteorological variables suggests that point-based station observations cannot be simply extrapolated to an entire glacier. These outcomes should be considered in future studies for a better estimation of glacier melt in High Mountain Asia.

The wake of a three-dimensional underwater obstacle: Effect of bottom boundary conditions

Three-dimensional (3D) obstacles on the bottom are common sites for the generation of vortices, internal waves and turbulence by ocean currents. Turbulence-resolving simulations are conducted for stratified flow past a conical hill, a canonical example of 3D obstacles. Motivated by the use of slip boundary condition (BC) and drag-law (effectively partial slip) BC in the literature on geophysical wakes, we examine the sensitivity of the flow to BCs on the obstacle surface and the flat bottom. Four BC types are examined for a non-rotating wake created by a steady current impinging on a conical obstacle, with a detailed comparison being performed between two cases, namely NOSL (no-slip BC used at all solid boundaries) and SL (slip BC used at all solid boundaries). The other two cases are as follows: Hybrid, undertaken with slip at the flat bottom and no-slip at the obstacle boundaries, and case DL wherein a quadratic drag-law BC is adopted on all solid boundaries. The no-slip BC allows the formation of a boundary layer which separates and sheds vorticity into the wake. Significant changes occur in the structure of the lee vortices and wake when the BC is changed. For instance, bottom wall friction in the no-slip case suppresses unsteadiness of flow separation leading to a steady attached lee vortex. In contrast, when the bottom wall has a slip BC (the SL and Hybrid cases) or has partial slip (DL case), unsteady separation leads to a vortex street in the near wake and the enhancement of turbulence. The recirculation region is shorter and the wake recovery is substantially faster in the case of slip or partial-slip BC. In the lee of the obstacle, turbulent kinetic energy (TKE) for case NOSL is concentrated in a shear layer between the recirculating wake and the free stream, while TKE is bottom-intensified in the other three cases. DL is the appropriate BC for high-Re wakes where the boundary layer cannot be resolved. The sources of lee vorticity are also examined in this study for each choice of BC. The sloping sides lead to horizontal gradients of density at the obstacle, which create vorticity through baroclinic torque. Independent of the type of BC, the baroclinic torque dominates. Vortex stretch and tilt are also substantial. An additional unstratified free-slip case (SL-UN) is simulated and the wake is found to be thin without large wake vortices. Thus, stratification is necessary for the formation of coherent lee vortices of the type seen in geophysical wakes.

Turbulence and Thermal Structure in the Upper Ocean: Turbulence-Resolving Simulations

The upper layer of the ocean participates directly in the exchange of momentum, heat and moisture with the atmosphere. We consider three examples of upper-ocean flow and heat transfer in the present contribution. These examples range from the canonical problem of a stratified shear layer to the surface boundary layer driven by wind and a diurnally varying heat flux to deep cycle turbulence in the Equatorial UnderCurrents (EUC). These problems illustrate stratified shear flow turbulence, wind-driven entrainment in a stratified, rotating fluid, and the communication of surface forcing to subsurface currents in the upper ocean. We discuss the three cases by including new simulations as well as some of our previous work. Direct numerical simulation (DNS) is our tool for the canonical shear layer and, for the other problems, our tool is large eddy simulation (LES) which is increasingly being used to examine turbulent transport and mixing in the ocean. We discuss how buoyancy and rotation affects the spatial structure and temporal evolution of turbulent fluxes, and thereby the distribution of surface inputs of momentum and heat in the upper ocean.

The submesoscale, the finescale and their interaction at a mixed layer front

The spindown of a geostrophically balanced density front in an upper-ocean mixed layer is simulated with a large eddy simulation (LES) model that resolves O(1000) m down to O(1) m scale. Our goal is to examine the interaction between the submesoscale and the turbulent finescale, and another related goal is to use the turbulence-resolving simulation to better characterize vertical transport, frontogenesis and dissipative processes. The flow passes through symmetric and baroclinic instabilities, spawns vortex filaments of O (100) m thickness as well as larger eddies with cross-front velocity as large as the along-front velocity, and develops turbulence that is spatially localized and organized. A O(100) m physical-space filter is applied to the simulated flow so that the coherent submesoscale is separated from the finescale in a decomposition that preserves the spatial organization of the flow unlike the typical practice of a split into a frontal average and a fluctuation that obscures the coherent submesoscale. The energy spectrum exhibits a change of slope at O(100) m with an approximately −5/3 slope over a subrange of the finescale. Analysis of the submesoscale vertical velocity (as large as 5mm/s) in the upper layer of the front reveals that downwelling is limited to the thin vortex filaments and the junction of the submesoscale eddies with these filaments while upwelling occurs over spatially extensive regions in the eddies. Conditional averaging shows that heavier (lighter) fluid is preferentially downwelled (upwelled) by these coherent submesoscale structures leading to an overall buoyancy flux that is restratifying. The submesoscale is unbalanced with local Rossby number as large as 5. The kinetic energy (KE) transport equations are evaluated separately for the submesoscale and the finescale to understand energy pathways in this problem. The buoyancy flux (associated with coherent motions) transfers the potential energy of the front and acts as the primary source of submesoscale KE which is then transported across the front with a fraction transferred to the finescale. The transfer, limited to thin regions of O(100) m horizontal width, is accomplished by primarily horizontal strain in the upper 10 m and by vertical shear in the rest of the 50-m deep mixed layer. Frontogenetic mechanisms are diagnosed through analysis of the transport equation for squared buoyancy gradient. Horizontal strain is the primary frontogenetic term that is especially strong in the near-surface layer. The frontogenesis is counteracted primarily by horizontal diffusion in the top 10 m while, further below, the balance is with the horizontal gradient of vertical velocity.

Transport by deep convection in basin-scale geostrophic circulation: turbulence-resolving simulations

Direct numerical simulations are used to investigate the nature of fully resolved small-scale convection and its role in large-scale circulation in a rotating $f$-plane rectangular basin with imposed surface temperature difference. The large-scale circulation has a horizontal geostrophic component and a deep vertical overturning. This paper focuses on convective circulation with no wind stress, and buoyancy forcing sufficiently strong to ensure turbulent convection within the thermal boundary layer (horizontal Rayleigh numbers $Ra\approx 10^{12}{-}10^{13}$). The dynamics are found to depend on the value of a convective Rossby number, $Ro_{\unicode[STIX]{x0394}T}$, which represents the strength of buoyancy forcing relative to Coriolis forces. Vertical convection shifts from a mean endwall plume under weak rotation ($Ro_{\unicode[STIX]{x0394}T}>10^{-1}$) to ‘open ocean’ chimney convection plus mean vertical plumes at the side boundaries under strong rotation ($Ro_{\unicode[STIX]{x0394}T}<10^{-1}$). The overall heat throughput, horizontal gyre transport and zonally integrated overturning transport are then consistent with scaling predictions for flow constrained by thermal wind balance in the thermal boundary layer coupled to vertical advection–diffusion balance in the boundary layer. For small Rossby numbers relevant to circulation in an ocean basin, vertical heat transport from the surface layer into the deep interior occurs mostly in ‘open ocean’ chimney convection while most vertical mass transport is against the side boundaries. Both heat throughput and the mean circulation (in geostrophic gyres, boundary currents and overturning) are reduced by geostrophic constraints.

Ablation of sloping ice faces into polar seawater

The effects of the slope of an ice–seawater interface on the mechanisms and rate of ablation of the ice by natural convection are examined using turbulence-resolving simulations. Solutions are obtained for ice slopes $\unicode[STIX]{x1D703}=2^{\circ }{-}90^{\circ }$, at a fixed ambient salinity and temperature, chosen to represent common Antarctic ocean conditions. For laminar boundary layers the ablation rate decreases with height, whereas in the turbulent regime the ablation rate is found to be height independent. The simulated laminar ablation rates scale with $(\sin \unicode[STIX]{x1D703})^{1/4}$, whereas in the turbulent regime it follows a $(\sin \unicode[STIX]{x1D703})^{2/3}$ scaling, both consistent with the theoretical predictions developed here. The reduction in the ablation rate with shallower slopes arises as a result of the development of stable density stratification beneath the ice face, which reduces turbulent buoyancy fluxes to the ice. The turbulent kinetic energy budget of the flow shows that, for very steep slopes, both buoyancy and shear production are drivers of turbulence, whereas for shallower slopes shear production becomes the dominant mechanism for sustaining turbulence in the convective boundary layer.

Turbulence-resolving simulations of wind turbine wakes

Turbulence-resolving simulations of wind turbine wakes are presented using a high-order flow solver combined with both a standard and a novel dynamic implicit spectral vanishing viscosity (iSVV and dynamic iSVV) model to account for subgrid-scale (SGS) stresses. The numerical solutions are compared against wind tunnel measurements, which include mean velocity and turbulent intensity profiles, as well as integral rotor quantities such as power and thrust coefficients. For the standard (also termed static) case the magnitude of the spectral vanishing viscosity is selected via a heuristic analysis of the wake statistics, while in the case of the dynamic model the magnitude is adjusted both in space and time at each time step. The study focuses on examining the ability of the two approaches, standard (static) and dynamic, to accurately capture the wake features, both qualitatively and quantitatively. The results suggest that the static method can become over-dissipative when the magnitude of the spectral viscosity is increased, while the dynamic approach which adjusts the magnitude of dissipation locally is shown to be more appropriate for a non-homogeneous flow such that of a wind turbine wake.

A diagnostic tool for jet noise using a line-source approach and implicit large-eddy simulation data

In this work, we propose a cost-effective approach allowing one to evaluate the acoustic field generated by a turbulent jet. A turbulence-resolving simulation of an incompressible turbulent round jet is performed for a Reynolds number equal to 460,000 thanks to the massively parallel high-order flow solver Incompact3d. Then a formulation of Lighthill's solution is derived, using an azimuthal Fourier series expansion and a compactness assumption in the radial direction. The formulation then reduces to a line source theory, which is cost-effective to implement and evaluate. The accuracy of the radial compactness assumption, however, depends on the Strouhal number, the Mach number, the observation elevation angle, and the radial extent of the source. Preliminary results are showing that the proposed method approaches the experimental overall sound pressure level by less than 4 dB for aft emission angles below 50°.

Anisotropic grid-adaptation in large eddy simulations

Grid-adaptation requires an error indicator that identifies where and how the grid should be refined. In the case of turbulence-resolving simulations (LES, DNS, etc), a simple and intuitive error indicator is the small-scale energy which can be linked to both the modeling errors and the numerical errors. The purpose of this work is to introduce a new error indicator for large eddy simulations that is based on the small-scale energy in a directional manner, i.e., in a way that enables anisotropic grid-adaptation. The formulation is geometrically general and applicable to arbitrary geometries and grid topologies. The error indicator is first assessed on turbulent channel flow to test its ability to predict the optimal anisotropy of the grid, and then assessed on the flow over a backward-facing step to test its performance in a more complex case with an unstructured grid.

LES validation of urban flow, part II: eddy statistics and flow structures

Time-dependent three-dimensional numerical simulations such as large-eddy simulation (LES) play an important role in fundamental research and practical applications in meteorology and wind engineering. Whether these simulations provide a sufficiently accurate picture of the time-dependent structure of the flow, however, is often not determined in enough detail. We propose an application-specific validation procedure for LES that focuses on the time dependent nature of mechanically induced shear-layer turbulence to derive information about strengths and limitations of the model. The validation procedure is tested for LES of turbulent flow in a complex city, for which reference data from wind-tunnel experiments are available. An initial comparison of mean flow statistics and frequency distributions was presented in part I. Part II focuses on comparing eddy statistics and flow structures. Analyses of integral time scales and auto-spectral energy densities show that the tested LES reproduces the temporal characteristics of energy-dominant and flux-carrying eddies accurately. Quadrant analysis of the vertical turbulent momentum flux reveals strong similarities between instantaneous ejection-sweep patterns in the LES and the laboratory flow, also showing comparable occurrence statistics of rare but strong flux events. A further comparison of wavelet-coefficient frequency distributions and associated high-order statistics reveals a strong agreement of location-dependent intermittency patterns induced by resolved eddies in the energy-production range. The validation concept enables wide-ranging conclusions to be drawn about the skill of turbulence-resolving simulations than the traditional approach of comparing only mean flow and turbulence statistics. Based on the accuracy levels determined, it can be stated that the tested LES is sufficiently accurate for its purpose of generating realistic urban wind fields that can be used to drive simpler dispersion models.