High-fidelity Large-eddy Simulation Research Articles

Wind farm wake modeling based on deep convolutional conditional generative adversarial network

Modeling of wind farm wakes is of great importance for the optimal design and operation of wind farms. In this work a surrogate modeling method for parametrized fluid flows is proposed for wind farm wake modeling, based on the state-of-the-art deep learning framework i.e. deep convolutional conditional generative adversarial network. Based on the proposed method and the data generated by high-fidelity large eddy simulations, a novel wind farm wake model is developed. The developed model is first validated against high-fidelity data and the results show that it achieves accurate, efficient, and robust prediction of wind turbine wake flow, at all the streamwise locations including both near wake and far wake, for both streamwise and spanwise velocity components, and at the cases with different inflow wind profiles. Then an extensive parametric study is carried out and the results show that the model generalizes well to unknown flow scenarios. Furthermore, a case study for a wind farm is investigated by the developed model. The prediction results are then compared with high-fidelity simulations, showing that the model can predict the wind farm wake flow (including both the streamwise and spanwise velocity fields) very well.

Robust Nonlinear Tracking Control for Unmanned Aircraft in the Presence of Wake Vortex

The flight trajectory of unmanned aerial vehicles (UAVs) can be significantly affected by external disturbances such as turbulence, upstream wake vortices, or wind gusts. These effects present challenges for UAV flight safety. Hence, addressing these challenges is of critical importance for the integration of unmanned aerial systems (UAS) into the National Airspace System (NAS), especially in terminal zones. This work presents a robust nonlinear control method that has been designed to achieve roll/yaw regulation in the presence of unmodeled external disturbances and system nonlinearities. The data from NASA-conducted airport experimental measurements as well as high-fidelity Large Eddy Simulations of the wake vortex are used in the study. Side-by-side simulation comparisons between the robust nonlinear control law and both linear H∞ and PID control laws are provided for completeness. These simulations are focused on applications involving small UAV affected by the wake vortex disturbance in the vicinity of the ground (which models the take-off or landing phase) as well as in the out-of-ground zone. The results demonstrate the capability of the proposed nonlinear controller to asymptotically reject wake vortex disturbance in the presence of the nonlinearities in the system (i.e., parametric variations, unmodeled, time-varying disturbances). Further, the nonlinear controller is designed with a computationally efficient structure without the need for the complex calculations or function approximators in the control loop. Such a structure is motivated by UAV applications where onboard computational resources are limited.

Dynamic Stall over a Pitching Natural-Laminar-Flow Airfoil

High-fidelity large-eddy simulations are performed to study the unsteady boundary layer sensitivity of a natural-laminar-flow airfoil (NLF-0414) undergoing dynamic stall in comparison to two traditional airfoil sections (NACA 0012 and SD 7003). In this work, the natural-laminar-flow airfoil at and pitches with a constant nondimensional rate of from an initial incidence angle of 4 deg to angles of attack beyond the onset of dynamic stall. It is found that the transition of the NLF-0414’s unsteady boundary layer during the pitchup is much more abrupt than the standard airfoil sections before the formation of the laminar separation bubble. Once the laminar separation bubble is established near the leading edge, the overall dynamic stall process of the NLF-0414 evolves very similarly to that of the SD 7003 and the NACA 0012. Minor differences are observed in the formation of the turbulent separation vortex near the trailing edge and convection of the dynamic stall and shear-layer vortices. The dynamic stall of the NLF-0414 airfoil more closely resembles that of the SD 7003 and the NACA 0012 at reduced Reynolds numbers, suggesting that the extended laminar flow acts to delay Reynolds number effects of traditional airfoil profiles.

Numerical analysis of particle dispersion and deposition in coal combustion using large-eddy simulation

High-fidelity modeling provides a useful approach to investigate the particle dispersion and deposition mechanism in pulverized coal combustion. To be able to analyze these detailed mechanisms, this work couples detailed multiphysics models for poly-dispersed turbulent, particle-laden flow, particle and gas phase reaction chemistry, convective, conductive and radiative heat transfer, ash formation and deposition with a Large-Eddy Simulation (LES) approach to numerically simulate coal combustion in a downfired self-sustained oxy-fuel combustor (OFC). It is necessary to explicitly capture all but the highest frequency dynamics of the turbulence and its coupling with each of the other physical phenomena using LES. Effects of subgrid-scale unresolved turbulence on the particle motions are also analyzed by a novel Stokes number analysis. Due to the inherent relation ship between coal combustion and ash deposition, this study integrates the improved ash deposition model into the numerical simulation. Overall simulation results are compared with experimentally measured data from the OFC. The simulation and measured data for the averaged gas temperature and deposition rates agree with 5% and 28%. This study shows that high-fidelity LES coupled with other detailed multiphysics models running on a exascale computing facility can provide a good representation of complex coal combustion and deposition in a laboratory-scale furnace.

Three-dimensional spatiotemporal wind field reconstruction based on physics-informed deep learning

In this work, a physics-informed deep learning model is developed to achieve the reconstruction of the three-dimensional (3-D) spatiotemporal wind field in front of a wind turbine, by combining the 3-D Navier–Stokes equations and the scanning LIDAR measurements. To the best of the authors’ knowledge, this is for the first time that the full 3-D spatiotemporal wind field reconstruction is achieved based on real-time measurements and flow physics. The proposed method is evaluated using high-fidelity large eddy simulations. The results show that the wind vector field in the whole 3-D domain is predicted very accurately based on only scalar line-of-sight LIDAR measurements at sparse locations. Specifically, at the baseline case, the prediction errors for the streamwise, spanwise and vertical velocity fields are 0.263 m/s, 0.397 m/s and 0.361 m/s, respectively. The prediction errors for the horizontal and vertical direction fields are 2.84° and 2.58° which are important in tackling yaw misalignment and turbine tilt control, respectively. Further analysis shows that the 3-D wind features are captured clearly, including the evolutions of flow structures, the wind shear in vertical direction, the blade-level speed variations due to turbine rotation, and the speed variations modulated by the turbulent wind. Also, the developed model achieves short-term wind forecasting without the commonly-used Taylor’s frozen turbulence hypothesis. Furthermore it is very useful in advancing other wind energy research fields e.g. wind turbine control & monitoring, power forecasting, and resource assessments because the 3-D spatiotemporal information is important for them but not available with current sensor and prediction technologies.

Aeroelastic loads on a 10 MW turbine exposed to extreme events selected from a year-long large-eddy simulation over the North Sea

Abstract. In this article the aeroelastic loads on a 10 MW turbine in response to extreme events (low-level jet, shear, veer and turbulence intensity) selected from a year-long large-eddy simulation (LES) on a site at the North Sea are evaluated. These events are generated with a high-fidelity LES wind model and fed into an aeroelastic tool using two different aerodynamic models: a model based on blade element momentum (BEM) and a free vortex wake model. Then the aeroelastic loads are calculated and compared with the loads from the IEC standards. It was found that the loads from all these events remain within those of the IEC design loads. Moreover, the accuracy of BEM-based methods for modelling such wind conditions showed a considerable overprediction compared to the free vortex wake model for the events with extreme shear and/or veer.

A data-driven approach using machine learning for early detection of the lean blowout

A data-driven approach using machine learning is presented for the identification of the critical flame location for the early detection of an incipient lean blowout (LBO) in a realistic gas turbine engine combustor under engine-relevant conditions. This method is demonstrated by utilizing the temperature (T) and the hydroxyl radical mass fraction (YOH) data from high fidelity large eddy simulations (LES) of Jet-A combustion. The fuel flow rate is progressively reduced in numerical simulations with a fixed airflow rate to mimic experimental studies of LBO in the gas turbine combustor. These simulations are the first of their kind for a fully resolved realistic combustor geometry with adaptive mesh refinement and have accurately captured the dynamics of the LBO process and global lean blowout equivalence ratio. Time-series of T and YOH are extracted in the primary zone of the combustor, from stable flame condition to LBO condition, to train the machine learning model. A Support Vector Machine (SVM) model with radial basis function is successfully developed to identify the critical flame location for early detection of incipient LBO condition in a practical combustor for the first time. The performance of the SVM model is quantified using the F-score, and the critical flame location corresponds to the maximum value of the F-score. The critical flame location is found to be in the flame root region and is effective in the early detection of incipient LBO. The conventional statistical measures are compared with the results of the trained machine learning model to assess the feasibility of the latter for online flame health monitoring. The machine learning model successfully prognosticated the LBO approximately 20 ms before the event, and this study has shown significant promise for the use of the SVM model in engine prognostics and health management.

Numerical Simulation of a Three-Dimensional Axisymmetric Hill: Performance Evaluation of Hybrid RANS–LES Models

Abstract Computational fluid dynamics (CFD) simulation of flow over a three-dimensional (3D) axisymmetric hill presents a unique set of challenges for turbulence modeling. The flow past the crest of the hill is characterized by boundary layer separation, complex vortical structures, and unsteady wake flow. As a result, traditional eddy-viscosity Reynolds-averaged Navier–Stokes (RANS) models have been found to perform poorly for this benchmark test case. Recent studies have focused on the use of large eddy simulation (LES) and hybrid RANS–LES (HRL) methods to improve accuracy. In this study, several different HRL models are investigated, and results from the different models are evaluated relative to each other, to an eddy-viscosity RANS model, and to previously documented high-fidelity LESs and experimental data. Results obtained from the simulations in terms of mean flow statistics, surface pressure distribution, and turbulence characteristics are presented and discussed in detail. Results indicate that eddy-viscosity based HRL models can improve predictions over comparable eddy-viscosity based RANS models such as the shear stress transport (SST) k–ω model used in this study, but only when the development of turbulent velocity fluctuations in the separated shear layer and recirculation region are well resolved.

A COMBINED NEURAL NETWORK AND MULTIPLE MODIFICATION STRATEGY FOR REYNOLDS-AVERAGED NAVIER-STOKES TURBULENCE MODELING

Solving the Reynolds-averaged Navier-Stokes (RANS) equation remains an effective and practical approach in engineering applications, but the uncertainty of Reynolds stress modeling will lead to discrepancies in the prediction accuracy of this approach. With the development of artificial intelligence, the data-driven method of turbulence model combined with machine learning algorithm is more effective than the original RANS model, however, the stability and prediction accuracy of the data-driven method could still be further improved. In the present paper, a fully connected neural network is constructed to predict the eddy viscosity, and this neural network is called as Eddy Viscosity Neural Network (EVNN). Additionally, a tensor-based neural network (TBNN) is also applied to predict the higher-order eddy viscosity relationship between the unclosed quantity and the analytical quantity, and the basis tensors are used to ensure the Galilean invariance. Finally, the closed-loop accuracy of the predicted flow field is realized through multiple modifications. For the method above, the neural network which is combined by EVNN and TBNN, is trained by using the high-fidelity data generated by the large eddy simulation (LES) and the baseline data obtained by the RANS simulation. Compared with the high-fidelity LES results, the results of the modified model exhibit significantly higher accuracy in the posterior velocity field, the mean pressure coefficient, and the mean friction coefficient than the original RANS model. It can be found that the implicit treatment of the linear part of the Reynolds stress can enhance the numerical stability, and the modification of the nonlinear part of the Reynolds stress can better predict the anisotropic characteristics of the flow field. Furthermore, the prediction accuracy is further improved through the multiple modification strategy. Therefore, the combined neural network and multiple modification strategy developed in this paper, have strong potentials in data-driven turbulence modeling and engineering applications in the future.

A numerical study on suspended sediment transport in a partially vegetated channel flow

Turbulent structures generated by vegetation patches play a dominant role in the dispersion of suspended sediment, which in turn is of great significance for ecosystem cycling and river geomorphology development. High fidelity Large Eddy Simulations (LES) coupled with the Discrete Phase Method (DPM) were used to explore the particle distribution and its variance (the non-uniformity in temporal and spatial space) in a partially vegetated straight channel. The novel findings and conclusions are outlined here. Firstly, the contour of the vertical vorticity component coincides well with particle preferential gatherings in the outer edge of the mixing layer in the near-bed region. Large-scale turbulent structures grow in mixing layer along the side of a vegetation patch (VP), which deplete particles away from the mixing layer into the neighbouring region. Also, higher vegetation densities (Dn) promote this depletion trend. Secondly, the Probability Density Function (PDF) and its variance were defined to quantify these phenomena, illustrating that the VP continuously interrupts the flow condition and promotes higher non-uniformity of particle distribution among the vegetated and non-vegetated regions. The variance of the PDF in the non-vegetated region is significantly higher than that in the neighbouring vegetated region located in the same streamwise location. The particle parcels are highly unevenly located along the periphery of the large eddies and are exchanged by the mixing flow between the non-vegetated and vegetated regions. Finally, the vertical entrainment of particles occurs in the vegetated region of the present cases. This is because the horseshoe structures provide an upwards velocity for the current Dn conditions (Dn < 0.1) and an increase of Dn (Dn < 0.1) accelerates the upward suspension. These findings complete our understanding of particles’ transportation in both spanwise and vertical directions.

Towards multiscale modeling of ocean surface turbulent mixing using coupled MPAS-Ocean v6.3 and PALM v5.0

Abstract. A multiscale modeling approach for studying the ocean surface turbulent mixing is explored by coupling an ocean general circulation model (GCM) MPAS-Ocean with the Parallelized Large Eddy Simulation Model (PALM). The coupling approach is similar to the superparameterization approach that has been used to represent the effects of deep convection in atmospheric GCMs. However, the focus of this multiscale modeling approach is on the small-scale turbulent mixing and their interactions with the larger-scale processes in the ocean, so that a more flexible coupling strategy is used. To reduce the computational cost, a customized version of PALM is ported on the general-purpose graphics processing unit (GPU) with OpenACC, achieving 10–16 times overall speedup as compared to running on a single CPU. Even with the GPU-acceleration technique, a superparameterization-like approach to represent the ocean surface turbulent mixing in GCMs using embedded high fidelity and three-dimensional large eddy simulations (LESs) over the global ocean is still computationally intensive and infeasible for long simulations. However, running PALM regionally on selected MPAS-Ocean grid cells is shown to be a promising approach moving forward. The flexible coupling between MPAS-Ocean and PALM allows further exploration of the interactions between the ocean surface turbulent mixing and larger-scale processes, as well as future development and improvement of ocean surface turbulent mixing parameterizations for GCMs.

Analysis and Interpretation of Data-Driven Closure Models for Large Eddy Simulation of Internal Combustion Engine

&lt;div class="section abstract"&gt;&lt;div class="htmlview paragraph"&gt;We present an automatic data-driven machine learning (ML) approach for the development, evaluation and interpretation of deep neural networks (DNNs) for turbulence closures and demonstrate their usage in the context of cold-flow large-eddy simulation (LES) of the four-stroke Darmstadt engine using an open-source compressible multi-dimensional CFD solver OFICE, in a hybrid PDE-ML framework. Rather than explicitly using canonical formulations of closure terms, these DNNs robustly discover the functional relationships between the large-scale features of the resolved flow (cell Re, strain and rotation rate tensors etc.) obtained by solving the Navier Stokes to the small-scale unresolved terms. Experimentally validated high-fidelity LES solutions of the engine at different crank angles are utilized as the ground truth to train the DNN based closure models. Since optimizing these DNNs can be a laborious process for scientific datasets, and often require specialized expertise, we propose a Bayesian optimization framework that automatically determines the best set of network parameters, including the architecture and training hyperparameters - batch size, regularization etc. for optimum performance. We compare and contrast various networks for their effectiveness in an &lt;i&gt;a-priori&lt;/i&gt; testing setting. Finally, the best ‘learnt’ network is integrated with the open-source CFD solver (OFICE), and solutions are obtained over several injection cycles. These experiments reveal that the DNN models temporally track resolved scalar variance with a good accuracy. Additionally, we interpret the artificial neural networks with sensitivity analysis to determine the relevant large-scale features for the learning process.&lt;/div&gt;&lt;/div&gt;

Universal Wind Profile for Conventionally Neutral Atmospheric Boundary Layers.

Conventionally neutral atmospheric boundary layers (CNBLs), which are characterized with zero surface potential temperature flux and capped by an inversion of potential temperature, are frequently encountered in nature. Therefore, predicting the wind speed profiles of CNBLs is relevant for weather forecasting, climate modeling, and wind energy applications. However, previous attempts to predict the velocity profiles in CNBLs have had limited success due to the complicated interplay between buoyancy, shear, and Coriolis effects. Here, we utilize ideas from the classical Monin-Obukhov similarity theory in combination with a local scaling hypothesis to derive an analytic expression for the stability correction function ψ=-c_{ψ}(z/L)^{1/2}, where c_{ψ}=4.2 is an empirical constant, z is the height above ground, and L is the local Obukhov length based on potential temperature flux at that height, for CNBLs. An analytic expression for this flux is also derived using dimensional analysis and a perturbation method approach. We find that the derived profile agrees excellently with the velocity profile in the entire boundary layer obtained from high-fidelity large eddy simulations of typical CNBLs.

Can Water Dilution Avoid Flashback on a Hydrogen-Enriched Micro-Gas Turbine Combustion?—A Large Eddy Simulations Study

Abstract Considering the growing interest in Power-to-Fuel, i.e., production of H2 using electrolysis to store excess renewable electricity, combustion-based technologies still have a role to play in the future of power generation. Especially in a decentralized production with small-scale cogeneration, micro-Gas Turbines (mGTs) offer great advantages related to their high adaptability and flexibility, in terms of operation and fuel. Hydrogen (or hydrogen enriched methane) combustion is well-known to lead to flame and combustion instabilities. The high temperatures and reaction rates reached in the combustor can potentially lead to flashback. In the past, combustion air humidification (i.e., water addition) has proven effective to reduce temperatures and reaction rates, leading to significant NOx emission reductions. Therefore, combustion air humidification can open a path to stabilize hydrogen combustion in a classical mGT combustor. However accurate data assessing the impact of humidification on the combustion is still missing for real mGT combustor geometries and operating conditions. In this framework, this paper presents a comparison between pure methane and hydrogen enriched methane/air combustions, with and without combustion air humidification, in a typical mGT combustion chamber (Turbec T100) using Large Eddy Simulations (LES) analysis. In a first step, the necessary minimal water dilution, to reach stable and low emissions combustion with hydrogen, was assessed using a one-dimensional (1D) approach. The one-dimensional unstretched laminar flame is computed for both pure methane (reference case) and hydrogen enriched methane/air combustion cases. The results of this comparison show that, for the hydrogen enriched combustion, the same level of flame speed as in the reference case can be reached by adding 10% (in mass fraction) of water. In a second step, the feasibility and flexibility of humidified hydrogen enriched methane/air combustion in an industrial mGT combustor have been demonstrated by performing high fidelity LES on a 3D geometry. Results show that steam dilution helped to lower the reactivity of hydrogen, and thus prevents flashback, enabling the use of hydrogen blends in the mGT at similar CO levels, compared to the reference case. These results will help to design future combustor toward more stability.

Supersonic Turbulent Boundary-Layer Separation Control Using a Morphing Surface

Separated flows arising due to shock-wave/turbulent boundary-layer interactions can cause problematic low-frequency unsteadiness with potentially severe structural response. High-fidelity large-eddy simulations are employed to examine surface morphing as a way to reduce the size of the separation region, and thus favorably alter the unsteadiness characteristics. The configuration considers a turbulent Mach 2.7 flow at a Reynolds number of subjected to an impinging shock system with a pressure ratio of , which results in separation and the presence of structurally relevant low-frequency unsteadiness. The control surface, centered about the shock impingement location and extending over the separation region, is allowed to deform under material property-based realizability constraints until an asymptotic state is achieved. The criterion for deformation uses a measure proportional to the directional surface shear stress. At an asymptotic state, the deformed surface reveals a shape consistent with aerostructural optimization and a maximum height of . Control mitigates the sharp initial pressure gradient of the uncontrolled flow to delay and reduce separation extent (by 50%) with diminution of low-frequency content and turbulent kinetic energy. Modal decomposition highlights these effects in the energy content of the prominent modes. Morphing may thus provide a means to adjust the local surface deflection in a manner that reduces some of the problems associated with turbulent separation.

On the impact of fuel injection angle in Euler–Lagrange large eddy simulations of swirling spray flames exhibiting thermoacoustic instabilities

This study deals with the fundamental problem of combustion dynamics in gas turbine combustors operating with liquid fuel. In this framework the present work proposes the study of an academic liquid fueled combustor sensitive to thermoacoustic instabilities, simulated via high-fidelity Large Eddy Simulations. The experimental setup addressed is SICCA-spray from EM2C laboratory featuring both stable and unstable flames depending on the combustion chamber length. The proposed analysis, based on the Euler-Lagrange modeling approach, studies the impact of the spray injection angle θ on both the stable flame and the triggering of the longitudinal combustor acoustic mode when using a longer quartz tube. For the liquid injection modeling, the FIM-UR semi-empirical model is adopted with three different θ values: θ=35∘,45° and 60°. In stable conditions, the spray angle is proven to have a negligible impact on the flame anchoring point, however, the mean flame length and fuel distribution are found to be slightly modified by the velocity at which droplets enter the combustion chamber. For the thermoacoustically unstable conditions, two well-established stable limit cycles with the same frequency and similar amplitudes are obtained when fuel is injected at θ=45∘ and 60°. Contrarily, the system stabilizes when θ=35∘ pointing to the importance of the dynamics of the liquid film layer formed inside the injector for this setup. Likewise, this liquid film layer dynamics and its modeling appear critical as already suggested by previous studies on the same configuration. The detailed analysis of the thermoacoustically unstable different predictions is then performed through the investigation of the spatial fields of the local Rayleigh index obtained following the novel extension in the frequency domain of the Rayleigh criterion complemented by the application of Dynamic Mode Decomposition. It confirms that the injection angle of the liquid spray has a significant effect on the thermoacoustic response of the system. Indeed the influence of θ on the dynamics of the liquid fuel when entering the combustion chamber is proven to have an impact on the synchronization mechanism governing the liquid phase with respect to acoustics sustaining the observed limit cycles. More specifically, couplings at the liquid phase level are evidenced by introducing two novel indices correlating the fluctuations of liquid fuel volume fraction and evaporation rate with pressure.

Industry-Relevant Implicit Large-Eddy Simulation of a High-Performance Road Car via Spectral/hp Element Methods

We present a successful deployment of high-fidelity large-eddy simulation (LES) technologies based on spectral/$hp$ element methods to industrial flow problems, which are characterized by high Reyn...

Large eddy simulation of a linear turbine cascade with a trailing edge cutback

In this work, the effects of trailing edge (TE) cutback on the turbulent flow field and the aerodynamic loss of a turbine blade are investigated by Large-Eddy-Simulation (LES). A high-fidelity LES code was first developed based on a high-order finite difference method and applied to the simulation. The effects of the TE cutback on the blade surface boundary layer, pressure distribution, and loss generation are then discussed on the basis of time-mean flow fields of LES. The results represent that the cutback changed the boundary layer thickness just upstream of the TE and the base pressure around the TE, resulting in larger aerodynamic loss. Next, the time-dependent flow dynamics in the cascade is investigated. The results indicate that the dominant pressure fluctuation after the blade was influenced by the cutback. Furthermore, the phase-locked flow field is analyzed to reveal the flow dynamics responsible for the pressure fluctuation. Finally, the LES approach could clarify the underlying mechanisms of performance degradation of a turbine blade that caused by TE cutback.

Large eddy simulation of natural convection heat transfer and fluid flow around a horizontal cylinder

Natural convection around a single horizontal cylinder has been extensively studied for heat transfer characteristics, but when coupled with fluid flow, the laminar-to-turbulent transition in buoyant plume poses severe challenge on modelling fidelity and physical interpretability. To address this challenge, this paper conducts detailed verification, validation and analysis of high-fidelity large eddy simulation (LES) for an unconfined horizontal cylinder in water with a Rayleigh number of 8 × 107, complemented by Reynolds-Averaged Navier-Stokes (RANS) computations. To the best of the authors' knowledge, this is the first LES study with acceptable accuracy as validate by experimental data for buoyant plume above a single horizontal cylinder. It is found that LES is far more sensitive to mesh resolution and boundary condition setting than RANS. An alarm is set for the use of periodic condition in LES on lateral boundaries of computational domain due to its inferior accuracy than RANS with transition SST model. The finely-tuned LES with pressure condition on lateral boundaries shows satisfactory agreement with experimental data in terms of heat transfer on cylinder surface and buoyant plume velocity, based on which new physical insight on transitional behavior of thermal plume is obtained. It is found that after leaving the cylinder, buoyant plume is laminar and accelerates, subject to work input from buoyancy force, while its temperature keep decreasing due to heat loss to atmosphere. Flow instability appears first in upward velocity at a streamwise Grashof number of 1.5 × 108, where transition to turbulence onsets. Thermal plume continues to accelerate until it begins to sway horizontally, where energy dissipation into turbulence becomes the major loss mechanism of mean flow energy. Thus, cross-stream diffusivity is augmented notably due to turbulent stresses, leading to smoothing of transverse velocity distribution and reduction of transversely-averaged mean kinetic energy. Transversely-averaged turbulent kinetic energy keeps increasing in transitional regime until the streamwise Grashof number reaches 7 × 109 and then declines to approach an asymptotic value, signifying the end of transition. Overall speaking, buoyancy work dominates the change in mean flow energy, while mean shear outweighs buoyancy in producing turbulent kinetic energy.

High-fidelity simulations and field measurements for characterizing wind fields in a utility-scale wind farm

Characterizing wind farm flow fields at high temporal and spatial resolutions is critical prerequisite for the optimal design and operation of utility-scale wind farms and for reducing the levelized cost of energy. However, due to the large disparity of underlying scales, measurements or simulations alone cannot provide high resolution wind fields, which are informed by and account for the effect of both large scale (i.e. hour, day, month and year) and small scale (i.e. second and minute) site-specific variations in the atmosphere. We explore the feasibility of integrating field measurements and high-fidelity large-eddy simulation (LES) to characterize the wind field in a utility-scale wind farm while accounting for flow phenomena across multiple temporal scales. Specifically, we employ field measurements to characterize the monthly wind speed and wind direction distributions and investigate the wind characteristics in turbine wakes. It was found that the probability density function (PDF) of the wind speed in turbine wakes can be reasonably represented using the Weibull distribution but with shape factors smaller than those not in the wake. LES of the wind farm under statistically steady inflow is subsequently carried out for one wind direction. The LES predictions are compared with the measured data conditionally averaged based on the wind speed, wind direction and the root-mean-square of wind speed fluctuations over time intervals of 30 min. Good agreement is obtained for both mean wind speed and turbulence intensity. The present work shows the possibility of integrating field measurements and high-fidelity simulations for improved characterization of the site-specific wind fields in utility-scale wind farms.