Método Lattice Boltzmann em GPU
Carregando...
Data
2025-08-25
Autores
Verneck, Gustavo Arruda
Título da Revista
ISSN da Revista
Título de Volume
Editor
Universidade Federal do Espírito Santo
Resumo
Computational FLuid Dynamics (CFD) is an essential field of physics and engi neering, enabling the simulation and analysis of fluid behavior through numerical methods. This work focuses on the development of a high-performance application for CFD simulations using the Lattice Boltzmann Method (LBM), an efficient and highly parallelizable computational approach based on the Boltzmann Kinetic The ory. LBM is particularly well-suited for parallelization on Graphics Processing Units (GPUs), offering significant advantages in terms of performance and scalability. The application was implemented in the Rust programming language and employs the OpenCL framework to leverage computational parallelism. This dissertation presents the theoretical foundation of the method, the development process of the proposed library, the validation of results, and the performance analysis across different simula tion scenarios. The results demonstrate the feasibility and efficiency of the proposed approach, while also paving the way for future optimizations and applications in more complex and diverse fluid dynamics contexts
Descrição
Palavras-chave
Método Lattice Boltzmann (LBM) , Fluidodinâmica computacional
(CFD) , Simulação numérica , Paralelização em GPU , Computação de Alto Desempenho (HPC) , Lattice Boltzmann Method (LBM) , Computational Fluid Dynamics (CFD) , Numerical simulation , GPU parallelization , High-Performance Computing (HPC)