Article
Meteorology & Atmospheric Sciences
Alfredo Pena, Jeffrey D. Mirocha, M. Paul van der Laan
Summary: This study evaluates the wind-farm parameterization of the Weather Research and Forecasting Model and finds that it shows excellent agreement in predicting velocity within the turbine area. However, there are discrepancies in predicting turbulence kinetic energy (TKE) due to the requirement of higher TKE at the turbine position in mesoscale simulations. The impact of inversion height and strength is small, while resolution has a low impact on large-eddy simulations but a high impact on mesoscale simulations.
MONTHLY WEATHER REVIEW
(2022)
Article
Geosciences, Multidisciplinary
Domingo Munoz-Esparza, Hyeyum Hailey Shin, Jeremy A. Sauer, Matthias Steiner, Patrick Hawbecker, Jennifer Boehnert, James O. Pinto, Branko Kosovic, Robert D. Sharman
Summary: In recent years, there has been significant interest and investment in integrating aerial vehicles into urban environments for various purposes. While these Advanced Air Mobility operations have the potential to alleviate transportation bottlenecks, challenges remain, particularly in accurately predicting weather effects within urban areas.
Article
Green & Sustainable Science & Technology
Sam T. Fredriksson, Goran Brostrom, Bjorn Bergqvist, Johan Lennblad, Hakan Nilsson
Summary: The Deep Green technique for tidal power generation is suitable for moderate flows and operates typically at mid-depth, moving in a figure-eight path. A unique wake is created by Deep Green with increased bottom shear locally, and the flow disturbance can be scaled with its horizontal path width.
Article
Engineering, Marine
Gang Wang, Tobias Martin, Liuyi Huang, Hans Bihs
Summary: This study focuses on investigating fluid flow through a fixed net panel using large eddy simulations. The computational meshing strategies suitable for different twine materials are proposed and validated. It is found that simulating a small portion of the net panel is enough to replicate the full-scale net, while the diameter and length of the twines have an impact on the surrounding turbulence fields.
Article
Automation & Control Systems
Jingying Zhao, Yifan Song, Likun Wang, Hai Guo, Fabrizio Marigentti, Xin Liu
Summary: A hybrid ensemble Gaussian process regression (HEGPR) model is proposed in this paper to address the issue of non-Gaussian distribution in the sample space of wedge winding eddy current losses of large generators. The HEGPR model consists of three layers, with four tree regression models in the first layer and multiple Gaussian regression models in the second layer. The results demonstrate that the HEGPR model has good prediction performance, with a root mean squared error (RMSE) of 0.0282 and a goodness of fit (R2) of 0.9973. Compared to other Gaussian process models and traditional ensemble learning models, the HEGPR model has higher prediction accuracy and is more suitable for forecasting eddy current loss in large generators. It effectively addresses the issue of insufficient regression accuracy in Gaussian process when the sample space does not follow a Gaussian distribution.
ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE
(2023)
Article
Environmental Sciences
Robert S. Arthur, Katherine A. Lundquist, Jeffrey D. Mirocha, Stephanie Neuscamman, Yuliya Kanarska, John S. Nasstrom
Summary: Models used for nuclear detonation cloud rise by emergency planning and response teams are typically simplified and based on historical nuclear tests, lacking full consideration of complex emergency scenarios. This study introduces a new multiscale framework within the Weather Research and Forecasting (WRF) model to simulate nuclear cloud rise with higher fidelity, including time-varying weather fields and complexities like atmospheric moisture and terrain. The simulation results show good agreement with observations, especially for high-air bursts, demonstrating the potential for improved cloud rise predictions through a multiscale simulation framework.
ATMOSPHERIC ENVIRONMENT
(2021)
Article
Engineering, Civil
Quentin Bucquet, Isabelle Calmet, Laurent Perret, Magdalena Mache
Summary: This work assesses the performance of the drag-porosity model implemented in ARPS atmospheric Large-Eddy Simulation (LES) solver for simulating the atmospheric boundary layer over the urban canopy. The flow over an idealized urban canopy consisting of cubes with various packing densities is investigated. The model is able to reproduce the key features of the flow over urban terrain, including turbulent coherent structures and their characteristic scales.
JOURNAL OF WIND ENGINEERING AND INDUSTRIAL AERODYNAMICS
(2023)
Article
Multidisciplinary Sciences
Soukayna Mouatadid, Paulo Orenstein, Genevieve Flaspohler, Judah Cohen, Miruna Oprescu, Ernest Fraenkel, Lester Mackey
Summary: Subseasonal forecasting is crucial for water allocation, wildfire management, and disaster mitigation. However, current models suffer from errors in representing atmospheric dynamics and physics. To address this, we introduce an adaptive bias correction method that combines dynamical forecasts with observations using machine learning, leading to significant improvements in temperature and precipitation prediction skill.
NATURE COMMUNICATIONS
(2023)
Article
Meteorology & Atmospheric Sciences
Sina Khani, Fernando Porte-Agel
Summary: This study introduces a new scale-aware subgrid-scale parameterization method and implements it in the Weather Research and Forecasting Model to capture the dynamics of the transition process from laminar to turbulent flow in stratified shear layers at coarse-resolution simulations. The proposed method skillfully represents the unresolved fluxes and reduces biases in climate models, leading to higher accuracy in weather predictions without a high computational cost.
MONTHLY WEATHER REVIEW
(2022)
Article
Agriculture, Multidisciplinary
Kittakorn Sriwanna
Summary: This paper proposes a system for predicting rice blast disease using weather data and employs an ensemble method to rank weather features. Experimental results demonstrate that the top ten features outperform others in rice blast disease prediction. Average visibility, amount of rainfall, hours of sun, maximum wind speed, and days of rain are identified as the five most effective weather features.
COMPUTERS AND ELECTRONICS IN AGRICULTURE
(2022)
Article
Engineering, Civil
Yong Cao, Tao Tao, Yujiang Shi, Shuyang Cao, Dai Zhou, Wen-Li Chen
Summary: The limitations of large-eddy simulation (LES) mode in predicting microscale flows for engineering purposes in an atmospheric model (WRF) are not well understood. This study compared the performance of WRF-LES with the popular CFD solver (OpenFOAM) using a typical separated turbulent flow past a three-dimensional axisymmetric hill. The results showed that both models were able to produce the primary flow features with high similarity, but WRF-LES underestimated the turbulent kinetic energy in the near wake compared to OpenFOAM-LES. The energy spectra suggested that WRF-LES had a stronger capacity for generating and maintaining small-scale turbulent motions than OpenFOAM-LES. Additionally, the deviation of numerical dissipation behavior between the two solvers was examined.
JOURNAL OF WIND ENGINEERING AND INDUSTRIAL AERODYNAMICS
(2023)
Article
Meteorology & Atmospheric Sciences
Zhaoyi Shen, Akshay Sridhar, Zhihong Tan, Anna Jaruga, Tapio Schneider
Summary: This paper aims to create a public library of large-eddy simulations (LES) of clouds to improve parameterizations in global climate models (GCMs). The LES are driven by large-scale forcings from GCMs and are used to study cloud behavior in different climate states.
JOURNAL OF ADVANCES IN MODELING EARTH SYSTEMS
(2022)
Article
Automation & Control Systems
Chenbei Lu, Jinhao Liang, Wenqian Jiang, Jiaye Teng, Chenye Wu
Summary: High-resolution probabilistic load forecasting is vital for the reliable operation of the future power grid with a high penetration of renewables. Existing linear combination-based model ensemble approaches for load forecasting may not fully utilize the advantages of different models, limiting the performance. We propose a learning ensemble approach that directly learns the optimal nonlinear combination from data, outperforming conventional ensemble approaches. A Shapley value-based method is introduced to evaluate the contributions of each model to the ensemble, and numerical studies confirm its remarkable performance.
JOURNAL OF THE FRANKLIN INSTITUTE-ENGINEERING AND APPLIED MATHEMATICS
(2023)
Article
Engineering, Aerospace
Prashant Shrotriya, Ping Wang, Hai-xiang Zeng, Xi-rui Zhou, Antonio Ferrante, Fei Tian
Summary: This study investigates turbulent partially premixed flames using a dynamic thickened flame (DTF) model in large eddy simulations (LES). The DTF model is adjusted based on laminar flame thickness and speed parameters, which serve as fitting functions for mixture fraction values. Comparison with experimental results shows good agreement in terms of mixture fraction and temperature radial statistics. The sub-grid scale combustion modeling approach is further validated through investigations of important combustion characteristics, such as local extinction and blow-off limits. The simulations are quantitatively assessed using the Wasserstein metric, which demonstrates good performance.
AEROSPACE SCIENCE AND TECHNOLOGY
(2023)
Article
Green & Sustainable Science & Technology
Hannah M. Johlas, David P. P. Schmidt, Matthew A. A. Lackner
Summary: This study investigates the effect of angling wind turbine rotors on wake behavior using large eddy simulations. The results show that tilting the rotor vertically can steer the wake downward, increasing the available power for downwind turbines and generating stronger counter-rotating vortices. Although tilted and non-tilted wakes recover similarly in terms of wake velocity deficit, tilted wakes can provide available power to downwind turbines faster.