Estimativa simultânea de propriedades térmicas usando uma superfície de aquecimento ativa e inferência Bayesiana: aplicação em superfícies revestidas e texturizadas

• Ano de defesa: 2022
• Autor(a) principal: Santos Junior, José Aguiar dos
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Uberlândia; Brasil; Programa de Pós-graduação em Engenharia Mecânica
• Palavras-chave em Português: Problemas inversos de condução de calor; Medições de propriedades térmicas; Soluções analíticas de condução de calor; Texturização; Revestimento; CNPQ::ENGENHARIAS::ENGENHARIA MECANICA; CNPQ::ENGENHARIAS::ENGENHARIA MECANICA::ENGENHARIA TERMICA; Engenharia Mecânica; Mecânica dos fluidos; Meios de transferência de calor
• Link de acesso: https://repositorio.ufu.br/handle/123456789/35799; http://doi.org/10.14393/ufu.te.2022.324

This work proposes an experimental method to simultaneously estimate the thermal diffusivity and thermal conductivity of conductive materials, ABNT 1045 carbon steel and class K carbide with 5% Co, using a single access surface. Subsequently, these samples underwent surface modifications, the ABNT 1045 carbon steel was textured and the class K carbide with 5% Co was coated and its effect on the determination of effective thermophysical properties was analyzed. Modifications of surface materials are carried out in different sectors and have different applications to obtain functional surfaces. They offer the possibility to design the surface according to different requirements. An experimental apparatus was set up and the samples were exposed to a vacuum medium to obtain optimal results. The samples were partially heated on an active surface, the temperature was measured at two different points on the surface to estimate the thermal properties. The method was applied to two different thermal models using the same experimental data set. The first model used the ratio of two acquired surface temperatures to determine thermal diffusivity. The inverse problem was then solved using Bayesian inference. The second model applied the Bayesian inference on the theoretical and experimental values of temperatures to obtain the maximum probability of the squared error function of the temperature to estimate the thermal conductivity. The estimation of the thermophysical properties was performed with and without the Monte Carlo technique via Markov Chains (MCMC), with the Metropolis-Hastings sampling algorithm. The approach without the MCMC technique was called off-line Bayesian inference. Both techniques proved to be adequate for estimating properties. The results found are in good agreement with the literature, with a difference of less than 9,0% and an expanded uncertainty of less than 8,0% was found at a confidence level of 95,45%, from the uncertainty analysis. The proposed technique using only one surface indicates excellent potential for application to finished surfaces. Texturing and coating did not significantly affect the estimation of effective thermal diffusivity but promoted a change in the estimation of effective thermal conductivity. Texturing promoted an increase of up to approximately 13.0% in effective thermal conductivity and coating a reduction of approximately 11.0%.