Dissertação de Mestrado de Alexandre de Quadro Tacques Filho

Defesa de Dissertação de Mestrado Acadêmico

Aluno(a): Alexandre de Quadro Tacques Filho
Orientador(a): Marcio Dorn

Título: Generating reliable computational data through the Lattice Boltzmann Method: a simulation of the Lauric Acid melting process.
Linha de Pesquisa: Algoritmos e Otimização

Data: 01/09/2025
Hora: 10:00
Local: Esta banca ocorrerá de forma remota. Acesso público disponibilizado pelo link https://meet.google.com/gpx-iaaa-pto.

Banca Examinadora:
-Flávia Schwarz Franceschini Zinani (UFRGS)
-Guilherme Henrique Fiorot (UFRGS)
-José Palazzo Moreira de Oliveira (UFRGS)

Presidente da Banca: Marcio Dorn

Resumo: Latent Heat Energy Storage Systems (LHESS) utilizing Phase Change Materials (PCMs) offer high thermal energy storage density and effective temperature regulation due to their ability to absorb and release heat at nearly constant temperatures. However, simulating the nonlinear heat transfer and phase transition processes during PCM melting remains challenging. This study employs the Lattice Boltzmann Method (LBM), implemented in the open-source OpenLB framework, to simulate the melting process of lauric acid in a rectangular cavity with a single horizontal fin. The primary objective is to validate a new numerical model implemented in OpenLB by comparing its results with experimental data, focusing on liquid fraction evolution and temperature profiles. By doing that, we can understand how to reliably and efficiently generate physics simulation data. The secondary objective is to develop a prototype structured dataset from OpenLB simula tions, serving as a foundation for future data-driven modeling, such as training physics-informed neural networks or other machine learning models. The numerical approach leverages the enthalpy-based LBM, where the total enthalpy evolves via a distribution function that accounts for both sensible and latent heat. Temperature and liquid fraction are derived from the enthalpy field, enabling accurate modeling of the melting process. Validation is performed by comparing simulation results with experimental data from a setup involving lauric acid in a rectangular cavity. The model accurately reproduces the liquid fraction over time, particularly at higher spatial and temporal resolutions, and temperature solutions at multiple points align with trends observed in experimental thermocouple data, confirming the model’s physical consistency. To explore the impact of fin geometry, the study evaluates twelve geometric configurations, assessing their performance as heat exchangers. Results show that fin positioning significantly influences melting rates, with lower fin positions yielding 11% to 26% faster melting compared to higher positions. Longer and thinner fins further enhance phase change rates, with melting time improvements ranging from 1% to 25%. The combined effect of fin geometry and position results in a 63.7% difference between the minimum and maximum total melting times. OpenLB’s structured output in VTK format facilitates the generation of simulation data in a structured format, enabling automated data processing/extraction. This dataset is designed to support future research into data-driven modeling, enabling the development of surrogate models and accelerated predictive computations using machine learning techniques. OpenLB’s open-source nature, multiphysics capabilities, and GPU acceleration support make it ideal for generating high-fidelity simulation data and scaling to larger domains or longer time periods. This study prioritizes a rigorous validation of the OpenLB model against experimental data, ensuring reliability for LHESS applications, while establishing a foundation for future data-driven research through a structured dataset prototype.

Palavras-Chave: Latent Heat Energy Storage, Phase Change Material, Lattice Boltzmann Method, OpenLB, Lauric Acid, Melting, Numerical Simulation, Open Source.