Eventos Anais de eventos
COBEM 2019
25th International Congress of Mechanical Engineering
Reconstruction of Spacewise Varying Boundary Heat Flux Employing a Combination of Explicit Estimation with Markov Chain Monte Carlo Method
Submission Author:
Emerson de Lima Sanches , RJ
Co-Authors:
Emerson de Lima Sanches, Diego Knupp, Luiz A. S. Abreu, Fabricio Mascouto, Antônio Silva Neto
Presenter: Antônio Silva Neto
doi://10.26678/ABCM.COBEM2019.COB2019-1785
Abstract
The main objective of the present work is the validation of a new hybrid inverse methodology based on the Markov Chain Monte Carlo method (MCMC) and an explicit method, applying it to estimate a boundary heat flux. The idea is to first solve an inverse heat conduction problem using the explicit method, with the truncated eigenfunction expansions to regularize the measured temperatures. Posteriorly, the obtained estimates are used as the initial candidate for the first Markov Chain’s state, which are sampled adopting the Total Variation technique as a priori information. Besides that, this second approach use the regularized temperature obtained with the explicit methodology as the input data for the inverse problem. This proposed procedure can possibly reduce the computational time of the MCMC method, since the initial state of the Markov Chain will be more similar to the final estimate, accelerating the convergence of the method and improving the final results. As a case study, it is chosen a steady state heat conduction problem with spatial variation along two directions of a thermally thin sample, with a two-dimensional heat flux applied on the boundary. For this purpose, the measured temperature is simulated with the addition of experimental noise on the direct problem solution, representing synthetic data acquired through infrared thermography.
Keywords
heat conduction, Inverse problem, Regularization Techniques, explicit formulation, Bayesian inference
