Eventos Anais de eventos
ENCIT 2022
19th Brazilian Congress of Thermal Sciences and Engineering
THERMAL CONTACT RESISTANCE ESTIMATION USING FUTURE TIME REGULARIZATION
Submission Author:
Luiz Eduardo Silva Resende , SP
Co-Authors:
Luiz Eduardo Silva Resende, Elisan dos Santos Magalhães, Humberto Machado
Presenter: Humberto Machado
doi://10.26678/ABCM.ENCIT2022.CIT22-0438
Abstract
In mechanical engineering, some critical parameters are hard to measure directly. For example, estimating the friction heat flux in a machining tool processing involves calculating high temperatures, something almost impossible with conventional equipment. Another good example could be the measurement of the aerodynamic heat flux on the nose of a space vehicle in a reentry condition. This heat flux can generate temperatures bigger than 10000 °C in some cases. However, the estimation of these parameters can be done indirectly using inverse techniques. These techniques are in the scope of optimization problems. This work aims to solve an Inverse Problem by applying a Future Time Regularization combined with two minimization techniques: The Quadrilateral Optimization Method and Bayesian Inference (Metropolis-Hastings algorithm). The reason behind using a regularization method is that Inverse Problems are ill-posed. Thus, the estimation of the parameters can be highly harmed by errors in the data measurements. This work aims to estimate the thermal contact resistance in a unidimensional heat conduction problem. The Finite Volumes method will be applied using Parallel Cyclic Reduction for the direct problem. The problem will be solved by an in-house algorithm developed using Python language, and the performance of the different inverse methods will be compared. Furthermore, different experimental conditions will be applied to the problem for studying the thermal contact resistance behavior.
Keywords
Inverse problems, Time Traveling Regularization, Thermal Contact Resistance, Quadrilateral Optimization Method, Bayesian inference
