Eventos Anais de eventos
ENCIT 2016
16th Brazilian Congress of Thermal Sciences and Engineering
BAYESIAN ESTIMATION OF THERMOPHYSICAL PROPERTIES IN A TWO-DIMENSION SEMI-TRANSPARENT MEDIUM
Submission Author:
Paulo Moreira , RJ
Co-Authors:
Helcio Orlande, Olivier Wellele
Presenter: Paulo Moreira
doi://10.26678/ABCM.ENCIT2016.CIT2016-0584
Abstract
Semi-transparent materials are largely used on industrial applications, as the manufacturing of optic fibers, electronic components, glasses, thermal protection devices, etc. In order to describe how the heat transfer occurs in such cases, it is necessary to determine the values of the related thermo-physical properties. In a static, semi-transparent material, heat is transferred at the same time by conduction and radiation, and such phenomenon is described by a coupled conduction-radiation mathematical model. The present work brings a Bayesian approach for the parameters estimation of thermophysical parameters of a ceramic plate subjected to a radiative heat flux, where the parameters related to the coupled heat transfer model were dealt as probability densities. The Metropolis-Hasting algorithm for the Monte Carlo Markov chain (MCMC) method, was applied for the simultaneous calculation of the posterior distributions of the physical parameters related to the mathematical model, e.g. the thermal conductivity, the thermal capacity and the convective heat transfer coefficient, using non-informative prior distributions. The estimation of such parameters was carried out using simulated temperature measurements taken in observation points of a semi-transparent ceramic plate. These measurements were simulated by calculating the temperatures at each point and adding normally distributed errors to each calculated value. The mean values of the posterior distributions of the thermophysical parameters were used to calculate the estimated temperatures at the observation points and these temperatures were compared to the measured ones. The Markov chain converged for all parameters and normal posterior distributions were obtained for all parameters. The observed results indicate the accuracy of the Bayesian approach for the inverse problem of parameters estimation applied in this work.
Keywords
Bayesian inference, Radiation heat transfer, Inverse problem, thermophysical properties
