Session T43: Turbulence: Environmental and Planetary Boundary Layer

CFD's Carbon Footprint in 2022

Contemporary fluid dynamics simulations leverage growing computational resources and consequently enable direct simulation at increasing flow Reynolds number. However, electricity needed to power these computers, when derived from non-renewable sources yields carbon production. A recent analysis suggested that the carbon footprint of a direct simulation grows with nearly the fourth power of the Reynolds number. This work leverages that analysis and considers a larger database of simulations. A set of simulations from 2022 is considered to construct a sample distribution of fielded simulation Reynolds numbers. We develop confidence intervals to determine a typical Reynolds number simulated and, using the previously derived correlation for carbon footprint, we estimate the average carbon footprint for a direct simulation in 2022. The distribution and total carbon footprint of the database is considered and we conclude with an extrapolation for total carbon footprint of CFD in 2022. The goal of this work is that better quantitative assessment of CFD’s contribution to climate change may influence modelling choices such as the Reynolds number considered, modelling paradigm, location of computing resources, and choice of research problem.   

Recreating River Fluid Dynamics in a Water Channel

Rivers are multifaceted systems that play a vital role in helping model the Earth’s surface and maintain the ecological system. Studying river fluid dynamics in controlled environments aids in modeling and testing riverine fields and replicating the complexity of natural rivers poses challenges. This research aim is to provide a riverine environment in a water channel in terms of a mean velocity profile, a turbulence intensity profile, and free surface fluctuations that are typical of large rivers. The velocity and turbulence profiles were achieved by using alternating winglets on a passive turbulence grid with angles set to obtain a velocity shear while free surface deflections and deformations were achieved by matching the Froude number of the Kvichak River. The data for velocity was collected with a hotwire probe at various vertical positions while the free surface was characterized with qualitative and quantitative image processing.  The results indicated a mean velocity profile with a shear exponent of about 0.1, a turbulence profile with an average intensity of about 10%, and a free surface with a surface deflection of about 10% of the water depth.  This fluid dynamic representation of typical riverine conditions can be used to investigate particle transport,etc.   

Large-eddy simulations of scalar dispersion around an idealized building

Pollutant dispersion in urban environments is strongly influenced by the complex roughness geometry of the urban canopy. Detailed simulations of idealized geometries can provide vital insight into fundamental flow physics and can facilitate the development of reduced-order models. Here, Large Eddy Simulations (LES) are performed for the fluid flow and passive scalar transport from point releases near a simplified cubical building geometry using a spectral element method code. A thorough validation is conducted for the velocity field to ensure the sufficient accuracy of the LES approach compared with Direct Numerical Simulations (DNS) from the literature. The LES results provide a detailed characterization of scalar dispersion and its relation to flow structures and a Lagrangian description of the flow.   

Predicting Wind Damage in a City During a Typhoon: A Meteorological Model/LES Approach

The frequency of intense typhoon landfalls is rising in many parts of the world, heightening the likelihood of urban areas facing high wind speeds, previously considered exceedingly rare.

This emphasizes the need to verify the wind safety of each city in such extreme weather conditions, considering their diverse geometries resulting from the unique developmental histories and ways of living.

To achieve this goal, this study utilizes the meteorological model/LES hybrid method to replicate wind characteristics of typhoons, including distinctive vertical profiles, fluctuations in wind speed and direction originating from the coherent structure, and non-stationarity. An inflow wind emulating the strong typhoon Hagibis (2019) approaching Tokyo was generated.

Subsequently, an LES calculation of an actual urban area with buildings and topography over a wide area of 6km*3km was performed using the numerical fluid dynamics solver CUBE specifically designed for large-scale parallel computing by RIKEN R-CCS on the supercomputer Fugaku. Furthermore, a detailed analysis of the urban space with a sub-meter resolution for the central area was carried out.

The study assessed the risk distribution in the city arising from peak wind speeds in the near-ground region, while also estimating the potential building damage caused by the pressure acting upon buildings in the target area. The quantitative risk potential was analyzed by considering the concept of ensemble analysis.   

Comparison of wind profile models across the Ekman layer against DNS data

Knowledge of the wind profile in the planetary boundary layer is key to many applications from wind power engineering via boundary-layer schemes in the atmosphere to PBL closure in large- and mesoscale models. In this presentation, we will present an account on existing, operational wind profile models against a novel representation of the wind vector across the Ekman layer. Our novel representation of the wind profiles is based on a consistent non-dimensionalization of the entire boundary layer down to the surface and thus takes into account rotational effects in vicinity to the bottom boundary. We compare our theory to both existing wind profile models and to data from direct numerical simulation up to Re_τ∽4000.   

Field Measurements of Lagrangian Statistics in the Atmospheric Surface Layer

We present unique field measurements of Lagrangian trajectories in the lower atmospheric boundary layer. A field-scale, three-dimensional, particle tracking system was used to obtain the statistics. A massive measurement volume—greater than one hundred meters cubed—enables the study of a range of turbulent transport phenomena, including single-particle and two-particle dispersion. The scaling of Lagrangian statistics and comparisons to classical homogeneous isotropic turbulence will be presented. These findings contribute to our understanding of mixing and scalar dispersion in the atmosphere.   

Surface drag scaling in truly neutral atmospheric boundary layer using Laboratory ideas and ideal WRF LES

Turbulent drag is one of the most essential parameters in any turbulent flow. Accurately estimating drag is challenging due to a sharp velocity gradient near the wall. An approach for the scaling of drag for the hydrodynamically smooth wall-bounded laboratory flows, such as zero pressure gradient (ZPG) Laboratory Turbulent Boundary Layers (TBLs), are derived in Dixit et al. (2020 & 2022), which is known as the asymptotic drag law or the –1/2 power law. TBLs can be considered a specific case of Atmospheric Boundary Layer (ABL) flows, which are exceedingly complex. Comparing an ABL to a Lab TBL, the Reynolds number of an ABL is very high. Also, the surface of the atmosphere is rough. Therefore, the ABL presents a unique opportunity to examine drag scaling of laboratory TBL relations across orders of magnitude range in Reynolds number. As stability plays a special role in the ABL, the near-neutral conditions of the ABL are closer to the Lab TBLs, i.e., the buoyancy flux at the surface is close to zero. Such conditions are challenging to find in ABLs. In recent years, WRF (Weather Research and Forecasting ) has emerged as a better tool to simulate atmospheric flow conditions. Ideal WRF LES(large eddy simulation) mimics the neutral ABLs. Within the WRF LES, the boundary layer can quickly grow in neutrally stratified fluid background conditions. Results suggest that the TBLs and ABLs both follow the same asymptotic drag law.   

Calibration and uncertainty quantification of subgrid scale parameterizations for atmospheric flows

Physical processes such as turbulence and surface-atmosphere interactions are parameterized in the atmospheric boundary layer (ABL) flow models that drive predictions used for decision-making in a wide range of engineering and sustainability applications. Based on the Reynolds Averaged Navier Stokes (RANS) equations, parameterizations of subgrid scale (SGS) processes in weather and climate models are generally calibrated manually by tuning ABL turbulence parameterizations to available high-fidelity data. Manual tuning only targets a limited subset of observational data and parameters. We develop methods for the computationally-efficient calibration and uncertainty quantification (UQ) of model parameters. Uncertainty quantification is performed using the calibrate-emulate-sample approach, which combines stochastic optimization and machine learning emulation to speed up Bayesian learning. The methods are demonstrated in a perfect-model setting through the calibration and UQ of a convective parameterization in an idealized general circulation model (GCM) with a seasonal cycle. Calibration and UQ based on seasonally averaged climate statistics, compared to annually averaged, reduces the calibration error by up to an order of magnitude and narrows the spread of posterior distributions by factors between two and five, depending on the variables used for UQ. The reduction in the size of the parameter posterior distributions leads to a reduction in the uncertainty of climate model predictions. The UQ methodology is extended to perform Bayesian experimental design, where the locations and time periods of data acquisition to maximally reduce parameter uncertainty are identified. Extensions of the methodology to UQ of turbulence modeling in the ABL will be discussed.