Eventos Anais de eventos
ENCIT 2022
19th Brazilian Congress of Thermal Sciences and Engineering
A DATA-DRIVEN APPROACH FOR ATMOSPHERIC BOUNDARY LAYER FLOWS: LARGE-EDDY SIMULATION AND DIMENSIONALITY REDUCTION
Submission Author:
Pedro Roberto Barbosa Rocha , RJ
Co-Authors:
Pedro Roberto Barbosa Rocha, Marcos Sebastião de Paula Gomes
Presenter: Pedro Roberto Barbosa Rocha
doi://10.26678/ABCM.ENCIT2022.CIT22-0637
Abstract
Spatiotemporal analyses of atmospheric systems are extremely important for a variety of environmental studies, especially for those related to the wind energy sector, where there is a strong interest in finding optimal parameters for the operation of wind farms. For this optimization problem, many heavy numerical simulations are needed, which becomes unfeasible to be achieved in most situations since extremely high computer processing power and storage capacity are required. The present work tackled this challenge by applying dimensionality reduction tools to the outputs of a large eddy simulation (LES) of an atmospheric boundary layer (ABL) flow. This high-fidelity simulation was performed using ANSYS Fluent. To mimic the three-dimensional (3D) transient air flow over a mountainous terrain, a Gaussian-shaped bump was placed in the left-hand side of a rectangular domain and a velocity profile was prescribed at the inlet plane. The incremental principal components analysis (IPCA) algorithm built in Python was employed to reduce the dimensionality of the system and an excellent agreement was achieved between original and reconstructed fields. This efficient data reduction allows building reduced-order models (ROMs), which are much faster than their full-order counterparts and still portray the main features of the flow.
Keywords
Large Eddy Simulation, Dimensionality Reduction, atmospheric boundary layer, reduced order model
