Eventos Anais de eventos
COBEM 2023
27th International Congress of Mechanical Engineering
CONTACT THERMAL CONDUCTANCE ESTIMATION BY THE APPLICATION OF OPTIMIZATION ALGORITHMS
Submission Author:
Cairo Ximenes , RJ
Co-Authors:
Cairo Ximenes, Marcelo Colaco, Luiz A. S. Abreu
Presenter: Cairo Ximenes
doi://10.26678/ABCM.COBEM2023.COB2023-1540
Abstract
The following paper aims to solve inverse problems (IP) in the Heat Transfer context in order to estimate contact thermal conductivity (CTC), a property that represents the adherence between two or more bodies subjected to heat exchange phenomenon. This function is then evaluated within the interface of a bicomposite material, by indirect measurements of temperature in the external surface of the object of study. The distribution of temperature is synthetically obtained by the solution of the direct problem (DP) by applying the Finite Difference Method (FDM). As hypothesis, it was considered a steady-state regime, inexistence of internal source of heat, a bidimensional body and homogeneous media. In the DP framework, from the prior knowledge of thermal physical properties (in which CTC is included), as well as geometry and boundary conditions, it becomes possible to calculate temperature distribution in the external surface. The IP approach, then, addresses to indirectly estimate CTC when one is able to measure the temperature distribution in a certain region of the body, such as in a laboratory environment. It is therefore possible when some optimization techniques, such as the gradient-based heuristics Levenberg-Marquadt and Gauss-Newton are applied. The algorithms were developed in a Wolfram Mathematica environment alongside with C executables to solve the PD, in an integrated methodology. Firstly, to verify the feasibility of the proposed techniques, to the estimation process it was not included uncertainties in the synthetic experimental data, obtained by the solution of the DP. Once done, the second part consists of adding different levels of noise to the income distribution of temperature, so as to evaluate the influence into the solution heuristic. In the solution of the problems of this paper, six types of functions to represent the CTC distribution were analysed. To guarantee stabilility, Tikhonov regularization was added to the minimization process.
Keywords
Inverse problems, Optimization, Levenberg-Marquardt, Gauss-Newton, Contact Thermal Conductivity
