Large-Eddy Simulation Of Neutral Turbulent Flow Over Rough Sinusoidal Ridges

The requirements for a credible large-eddy simulation of neutral, turbulent flow over hills with an aerodynamically rough surface are discussed, in order to select a suitable case for simulation. As well as providing adequate resolution within the dynamically important inner region, obtaining a realistic upstream or undisturbed mean velocity profile is also of critical importance. A distributed drag canopy formulation has been introduced to the model to allow it to obtain such a profile while resolving very close to the rough surface. Simulations have then been performed of flow over ridges of varying heights. The results from the steepest case, which is just on the verge of separation, are compared with wind-tunnel observations. It is shown that the large-eddy simulation results are in much better agreement with the experimental data than are the results from a simple first-order mixing-length closure model. An encouraging lack of sensitivity of the simulation results to numerical resolution is also demonstrated.     

Edge Flow and Canopy Structure: A Large-Eddy Simulation Study

Sharp heterogeneities in forest structure, such as edges, are often responsible for wind damage. In order to better understand the behaviour of turbulent flow through canopy edges, large-eddy simulations (LES) have been performed at very fine scale (2 m) within and above heterogeneous vegetation canopies. A modified version of the Advanced Regional Prediction System (ARPS), previously validated in homogeneous conditions against field and wind-tunnel measurements, has been used for this purpose. Here it is validated in a simple forest-clearing-forest configuration. The model is shown to be able to reproduce accurately the main features observed in turbulent edge flow, especially the “enhanced gust zone” (EGZ) present around the canopy top at a few canopy heights downwind from the edge, and the turbulent region that develops further downstream. The EGZ is characterized by a peak in streamwise velocity skewness, which reflects the presence of intense intermittent wind gusts. A sensitivity study of the edge flow to the forest morphology shows that with increasing canopy density the flow adjusts faster and turbulent features such as the EGZ become more marked. When the canopy is characterized by a sparse trunk space the length of the adjustment region increases significantly due to the formation of a sub-canopy wind jet from the leading edge. It is shown that the position and magnitude of the EGZ are related to the mean upward motion formed around canopy top behind the leading edge, caused by the deceleration in the sub-canopy. Indeed, this mean upward motion advects low turbulence levels from the bottom of the canopy; this emphasises the passage of sudden strong wind gusts from the clearing, thereby increasing the skewness in streamwise velocity as compared with locations further downstream where ambient turbulence is stronger.     

Large-Eddy Simulation Of Turbulent Separated Flow Over Rough Hills

Large-eddy simulations (LES) have been performed ofneutral turbulent flow over two-dimensional ridges steepenough to cause separation. Both periodic and isolated ridges havebeen considered. The results are compared with wind-tunnel observations and with the predictions of various turbulence closure models.For the periodic case the LES results are qualitatively reasonable,although the depth of the separated region appears to besensitive to the use of a distributed drag near the lower boundary.The isolated ridge results compare very favourably with the experimentaldata, with the LES performance appearing to be at least as good as that ofthe closure models.     

Large-Eddy Simulation of Turbulent Flow Over a Rough Surface

A family of wall models is proposed that exhibits moresatisfactory performance than previousmodels for the large-eddy simulation (LES) of the turbulentboundary layer over a rough surface.The time and horizontally averaged statistics such asmean vertical profiles of windvelocity, Reynolds stress, turbulent intensities, turbulentkinetic energy and alsospectra are compared with wind-tunnel experimental data.The purpose of the present study is to obtain simulatedturbulent flows that are comparable with wind-tunnelmeasurements for use as the wind environment for thenumerical prediction by LES of source dispersion in theneutral atmospheric boundary layer.     

Large-Eddy Simulation of Katabatic Flows

A large-eddy simulation model with rotated coordinates and an open boundary is used to simulate the characteristics of katabatic flows over simple terrain. Experiments examine the effects of cross winds on the development of the slope-flow boundary layer for a steep (20°) slope and the role of drainage winds in preventing turbulence collapse on a gentle slope (1°). For the steep flow cases, comparisons between model average boundary-layer velocity, temperature deficit, and turbulence kinetic energy budget terms and tower observations show reasonable agreement. Results for different cross slope winds show that as the cross slope winds increase, the slope flow deepens faster and behaves more like a weakly stratified, sheared boundary layer. Analysis of the momentum budget shows that near the surface the flow is maintained by a balance between downslope buoyancy forcing and vertical turbulence flux from surface drag. Above the downslope jet, the turbulence vertical momentum flux reverses sign and acceleration of the flow by buoyancy is controlled by horizontal advection of slower moving ambient air. The turbulence budget is dominated by a balance between shear production and eddy dissipation, however, buoyancy and pressure transport both are significant in reducing the strength of turbulence above the jet. Results from the gentle slope case show that even a slight terrain variation can lead to significant drainage winds. Comparison of the gentle slope case with a flat terrain simulation indicates that drainage winds can effectively prevent the formation of very stable boundary layers, at least near the top of sloping terrain.     

Three-dimensional simulations of air flow and momentum transfer in partially harvested forests

In order to predict wind loading on trees (canopy height h) in partially harvested forests, it is necessary to characterize air flow and momentum transfer in progressively more complex patterns where groups of trees (or aggregates) are retained. In this study, we used large-eddy simulation to explore the effects of aggregate size, inter-aggregate spacing, and the ratio between the aggregate size and inter-aggregate spacing on air flow and momentum transfer. Forty-five grid points across an aggregate were needed to achieve an adequate level of turbulence. Using grid sizes of h/15 throughout was too viscous for the smaller aggregates. Vertical and horizontal flow deflection by the leading aggregates sheltered some of the downstream aggregates to varying degrees where turbulence increased for subsequent rows. The number of rows of protected aggregates decreased as aggregate dimensions and the space between aggregates increased. A theoretical treatment of time-dependent wind is presented for the lead aggregate and a simulation case is presented for the case of a gust of reduced wind passing through the aggregate pattern. The leading aggregate responded with decreasing moment for decreasing ambient wind speed as predicted by theory. However, downwind aggregates experienced substantial increases in bending moment. The overall results of the disruptive aspects of time dependence agrees with arguments regarding the role of irrotational (potential) flow to this problem. Our treatment of retention pattern design is only a first step and further research suggestions are presented.     

Three-Dimensional Scalar Microfront Systems in a Large-Eddy Simulation of Vegetation Canopy Flow

Large-eddy simulation is used to reproduce neutrallystratified airflow inside and immediately above a vegetation canopy. A passive scalaris released from the canopy and the evolution of scalar concentration above the canopyis studied. The most significant characteristic of the scalar concentration is the repeatedformation and dissipation of scalar microfronts, a phenomenon that has been observedin nature. These scalar microfronts consist of downstream-tilted regions of highscalar concentration gradients. Computer visualization tools and a conditional samplingand compositing technique are utilized to analyze these microfronts. Peaks in positivepressure perturbation exceeding an experimental threshold are found to be effectiveindicators of scalar microfronts. Convergence of the streamwise velocity componentand divergence of the cross-stream velocity component are observed in the immediatevicinity of scalar microfronts, which helps explain their relatively longlifetimes. Many of these three-dimensional features have been observedin previous field studies of canopy flow.     

Turbulent Pressure Statistics in the Atmospheric Boundary Layer from Large-Eddy Simulation

We use large-eddy simulation (LES) to study the turbulent pressure field in atmospheric boundary layers with free convection, forced convection, and stable stratification. We use the Poisson equation for pressure to represent the pressure field as the sum of mean-shear, turbulence–turbulence, subfilter-scale, Coriolis, and buoyancy contributions. We isolate these contributions and study them separately. We find that in the energy-containing range in the free-convection case the turbulence–turbulence pressure dominates over the entire boundary layer. That part dominates also up to midlayer in the forced-convection case; above that the mean-shear pressure dominates. In the stable case the mean-shear pressure dominates over the entire boundary layer.We find evidence of an inertial subrange in the pressure spectrum in the free and forced-convection cases; it is dominated by the turbulence–turbulence pressure and has a three-dimensional spectral constant of about 4.0. This agrees well with quasi-Gaussian predictions but is a factor of 2 less than recent results from direct numerical simulations at moderate Reynolds numbers. Measurements of the inertial subrange pressure spectral constant at high Reynolds numbers, which might now be possible, would be most useful.     

Large-Eddy Simulation of Flows over Random Urban-like Obstacles

Further to our previous large-eddy simulation (LES) of flow over a staggered array of uniform cubes, a simulation of flow over random urban-like obstacles is presented. To gain a deeper insight into the effects of randomness in the obstacle topology, the current results, e.g. spatially-averaged mean velocity, Reynolds stresses, turbulence kinetic energy and dispersive stresses, are compared with our previous LES data and direct numerical simulation data of flow over uniform cubes. Significantly different features in the turbulence statistics are observed within and immediately above the canopy, although there are some similarities in the spatially-averaged statistics. It is also found that the relatively high pressures on the tallest buildings generate contributions to the total surface drag that are far in excess of their proportionate frontal area within the array. Details of the turbulence characteristics (like the stress anisotropy) are compared with those in regular roughness arrays and attempts to find some generality in the turbulence statistics within the canopy region are discussed.     

Two-Point Correlation Analysis Of Neutrally Stratified Flow Within And Above A Forest From Large-Eddy Simulation

Two-point space-time correlations ofvelocities, a passive scalar and static pressure arecalculated using the resolvable flow fields computedby large-eddy simulation (LES) of neutrally stratifiedflow within and above a sparse forest. Zero-time-lagspatial auto-correlation contours in thestreamwise-vertical cross-section for longitudinal andlateral velocities and for a scalar are tilted fromthe vertical in the downstream direction, as istypical in near-wall sheared flow. On the other hand,auto-correlations of vertical velocity and of staticpressure are vertically coherent. Zero-time-lagspatial auto-correlations in the spanwise-verticalcross-section show no distinct tilt, and those forboth longitudinal and vertical velocities demonstratedistinct negative side lobes in the middle forest andabove, while longitudinal velocity in the subcrowntrunk space is laterally in-phase. Static pressureperturbations appear to be spatially coherent in thespanwise direction at all heights, especially insidethe forest. Near the forest floor, longitudinalvelocity is found to be in-phase with static pressureperturbation and to be closely linked to theinstantaneous streamwise pressure gradient, supportinga previous proposal that longitudinal velocity in thisregion is dominantly modulated by the pressurepatterns associated with the coherent sweep/ejectionevents. Near treetop height, a lack of linkage betweenthe pressure gradient and the local time derivative ofthe longitudinal velocity supports the hypothesis ofadvection dominating turbulent flow.The major phase characteristics of the two-pointcorrelations essentially remained the same from fourLES runs with different domain size and/or gridresolution. A larger LES domain yielded betteragreement with field observations in a real forest onboth the magnitudes of the correlations and thesingle-point integral time scales. A finer gridresolution in the LES led to a faster rate of decreaseof correlation with increasing separation in space ortime, as did the higher frequency fluctuations in theturbulent records from field measurements. Convectivevelocities estimated from the lagged two-pointauto-correlations of the calculated flow fields werecompared with similar calculations from wind-tunnelstudies. At the canopy top, estimates from thecorrelation analyses agree with the translationvelocity estimated from instantaneous snapshots of ascalar microfront using both LES and field data. Thistranslation velocity is somewhat higher than the localmean wind speed. Convective velocities estimated fromlagged correlations increase with height above thecanopy. It is suggested that an appropriate filteringprocedure may be necessary to reduce the effects ofsmall-scale random turbulence, as was reported in astudy over an orchard canopy. The mean longitudinalvelocity near the treetops is found to be moreappropriate than the local mean longitudinal velocityat each height to link single-point integral timescales with directly calculated spatial integralstreamwise length scales.     

Wind Flow Over Complex Terrain: A Historical Perspective and the Prospect for Large-Eddy Modelling

The development of our understanding of turbulent flow overcomplex terrain is described. Despite considerable advances, there are still gaps in our knowledge. Large-eddy simulation of such flows is now becoming possible. This may stimulate further advances but it is computationally very demanding. The application of this technique to flow over hills is discussed.     

Large-Eddy Simulation and Parametrization of the Effects of Shear on Shallow Cumulus Convection

Large-eddy simulations are performed of ensembles of shallow convective clouds under varying shear. The scalar transports are shown to be largely unaffected by shear, while the momentum transport is strongly affected. The parametrization of the momentum fluxes is discussed. The cross-cloud pressure gradient, commonly neglected in shallow cumulus parametrizations, is shown to be a dominant term in the in-cloud momentum budget. The performance of two possible representations of this pressure gradient, each previously proposed for use in deep convection parametrizations, is assessed.     

Large-Eddy Simulation Of Radiation Fog

In order to study the three-dimensional structure of radiation fogand to obtain a basic understanding of its generation mechanism,a numerical experiment is performed with a large-eddysimulation model and compared with the observation at Cabauw in the Netherlands. After confirming that the results are insatisfactory agreement with the observations, the structure of thefog and its generation mechanism are examined in more detail.Before the fog forms, the atmosphere is stable and an inversionlayer exists almost adjacent to the ground surface. As the fog grows, however, the stratification is destabilized and a mixed layerdevelops gradually. The longwave radiative cooling near thefog top contributes to the destabilization more than thecondensational heating does.The evolution of the fog can be classified into three stagesaccording to the behaviour of turbulent kinetic energy (TKE):formation, development, and dissipation stages.The fog layer has different flow structures at each stage.During the formation stage, longitudinal rolls similar tostreaks in channel flows appear near the ground surface.The development stage is characterized by an initiation oftransverse bands due to Kelvin–Helmholtz instability anda sudden increase of TKE. During the dissipation stage, longitudinalrolls and polygonal cells due to convective instability are organized.     

Large-Eddy Simulation Of The Stably Stratified Planetary Boundary Layer

In this work, we study the characteristics of a stably stratifiedatmospheric boundary layer using large-eddy simulation (LES).In order to simulate the stable planetary boundary layer, wedeveloped a modified version of the two-part subgrid-scalemodel of Sullivan et al. This improved version of themodel is used to simulate a highly cooled yet fairly windy stableboundary layer with a surface heat flux of(W)_o = -0.05 m K s^-1and a geostrophic wind speed of U_g = 15 m s^-1.Flow visualization and evaluation of the turbulencestatistics from this case reveal the development ofa continuously turbulent boundary layer with small-scalestructures. The stability of the boundary layercoupled with the presence of a strong capping inversionresults in the development of a dominant gravity wave atthe top of the stable boundary layer that appears to be relatedto the most unstable wave predicted by the Taylor–Goldsteinequation. As a result of the decay of turbulence aloft,a strong-low level jet forms above the boundary layer.The time dependent behaviour of the jet is compared with Blackadar'sinertial oscillation analysis.     

Large-Eddy Simulation of Flow and Pollutant Dispersion in High-Aspect-Ratio Urban Street Canyons with Wall Model

A large-eddy simulation (LES) with a one-equation subgrid-scale (SGS) model was developed to investigate the flow field and pollutant dispersion inside street canyons of high aspect ratio (AR). A 1/7th power-law wall model was implemented near rigid walls to mitigate the demanding near-wall resolution requirements in LES. This LES model had been extensively validated against experimental results for street canyons of AR = 1 and 2 before it was applied to the cases of AR = 3 and 5. A ground-level passive pollutant line source, located in the middle of the street, was used to simulate vehicular emissions. Three and five vertically aligned primary recirculations were developed in the street canyons of AR 3 and 5, respectively. The ground-level mean wind speed was less than 0.5% of the free stream value, which makes it difficult for the pollutant to be transported upward for removal. High pollutant concentration and variance were found near the buildings where the air flow is upwards. It was found that the velocity fluctuation, pollutant concentration and variance were all closely related to the interactions between the primary recirculations and/or the free surface layer. Several quantities, which are non-linear functions of AR, were introduced to quantify the air quality in street canyons of different configurations.     

Effect of Roughness on Surface Boundary Conditions for Large-Eddy Simulation

An important parameterization in large-eddy simulations (LESs) of high- Reynolds-number boundary layers, such as the atmospheric boundary layer, is the specification of the surface boundary condition. Typical boundary conditions compute the fluctuating surface shear stress as a function of the resolved (filtered) velocity at the lowest grid points based on similarity theory. However, these approaches are questionable because they use instantaneous (filtered) variables, while similarity theory is only valid for mean quantities. Three of these formulations are implemented in simulations of a neutral atmospheric boundary layer with different aerodynamic surface roughness. Our results show unrealistic influence of surface roughness on the mean profile, variance and spectra of the resolved velocity near the ground, in contradiction of similarity theory. In addition to similarity-based surface boundary conditions, a recent model developed from an a priori experimental study is tested and it is shown to yield more realistic independence of the results to changes in surface roughness. The optimum value of the model parameter found in our simulations matches well the value reported in the a priori wind-tunnel study.     

A Variable Mesh Spacing for Large-Eddy Simulation Models in the Convective Boundary Layer

A variable vertical mesh spacing for large-eddy simulation (LES) models in a convective boundary layer (CBL) is proposed. The argument is based on the fact that in the vertical direction the turbulence near the surface in a CBL is inhomogeneous and therefore the subfilter-scale effects depend on the relative location between the spectral peak of the vertical velocity and the filter cut-off wavelength. From the physical point of view, this lack of homogeneity makes the vertical mesh spacing the principal length scale and, as a consequence, the LES filter cut-off wavenumber is expressed in terms of this characteristic length scale. Assuming that the inertial subrange initial frequency is equal to the LES filter cut-off frequency and employing fitting expressions that describe the observed convective turbulent energy one-dimensional spectra, it is feasible to derive a relation to calculate the variable vertical mesh spacing. The incorporation of this variable vertical grid within a LES model shows that both the mean quantities (and their gradients) and the turbulent statistics quantities are well described near to the ground level, where the LES predictions are known to be a challenging task.     

A Wind-Tunnel Study of Airflow in Waving Wheat: An EOF Analysis of the Structure of the Large-Eddy Motion

We have conducted an Empirical Orthogonal Functionanalysis (EOF) of a three dimensional, 2-pointvelocity covariance field, measured in a wind tunnel. The rate of convergence of the EOF sequence was usedas an objective test for the presence of distinctlarge turbulent structures. We found that in theroughness sublayer (2h > z > 0) the sequence convergedmuch more rapidly than in the lower surface layer(6h > z > 0), 75% of the total velocity variancebeing captured by the first three of 42 eigenmodes;h is the canopy height. The analysis was extended to three dimensions, whereover 50% of the variance and most of the spatialstructure of the covariance fields were captured by aneven smaller fraction of the total number ofeigenmodes. With some relatively weak additionalassumptions we were able to construct the velocityfield of a characteristic eddy or large coherentstructure. This consisted of a pair ofcounter-rotating streamwise vortices centred above thecanopy. The sense of rotation of the vortex pair wasopposite to that found in the wall region of boundarylayers but matched that found in plane mixing layers. A strong gust or sweep motion generated between thevortices was responsible for most of the shear stresscarried by the large eddies. The region of significanttransport of streamwise momentum by the characteristiceddy is much smaller than the region of coherence of theeddy's velocity field.     

Large-Eddy Simulation Of A Nocturnal Stratocumulus-Topped Marine Atmospheric Boundary Layer: An Uncertainty Analysis

A large-eddy simulation (LES) model has been used to study a nocturnalstratocumulus-topped marine atmospheric boundary layer. The main objectivesof our study have been first to investigate the statistical significance of LES-derived data products. Second, to test the sensitivity of our LES results with respect to the representation of subgrid-scale mixing and microphysical processes, and third to evaluate and to quantify the parametric uncertainty arising from the incomplete knowledge of the environmental parameters that are required to specify the initial and boundary conditions of a particular case study. Model simulations were compared with observations obtained in solid stratocumulus during the third flight of the first 'Lagrangian' experiment of the Atlantic Stratocumulus Transition Experiment (ASTEX). Based on these simulations the following conclusions could be drawn. Resolution(50 × 50 × 25 m³) and domain size (3.2 × 3.2 × 1.5 km³) of the LES calculations were adequate from a numerical point of view to represent the essential features of the stratocumulus-topped boundary layer. However, the ensemble runs performed in our study to investigate the statistical significance of LES-derived data products demonstrate that the area-time averaging procedure for the second-order moments produces only a low degree of statistical reliability in the model results. This illustratesthe necessity of having LES model results that are not only of adequate resolution but also of sufficiently large domain. The impact of different subgrid schemes was small, but the primary effects of drizzle were found to influence the boundary-layer structure in a climatologically significant way. The parametric uncertainty analysis revealed that the largest contribution to the variance of the LES-derived data products is due to theuncertainties in the cloud-top jump of total water mixing ratio and the net radiative forcing. The differences between the model and measurements for most of the simulated quantities were within the modelling uncertainties, but the calculated precipitation rate was found to differ significantly from that derived in the observations.