Please use this identifier to cite or link to this item:
Scopus Web of ScienceĀ® Altmetric
Type: Theses
Title: An intelligent approach to inverse heat transfer analysis of irradiative enclosures
Author: Mirsepahi, Ali
Issue Date: 2017
School/Discipline: School of Chemical Engineering
Abstract: This is a thesis by publication for a PhD degree of engineering in the university of Adelaide. The current dissertation comprises five published/submitted journal articles. Three of these journal papers have already been published in the journal of "International Communications in Heat and Mass Transfer" and one has been accepted by the editorial board of "Chemical Engineering Communications". This study, based on research undertaken in the area of Inverse Heat Transfer Problems (IHTP), aims at analyzing the applicability of Intelligent Techniques (ITs) to solve sequential (real-time) heat flux estimation class of IHTPs, especially those involving in the most complicated form of heat transfer, radiation. Currently, several optimization based methods have been developed and applied to solve heat flux estimation problems. These methods normally require detailed and accurate information regarding physical properties. Often, the measurement of such physical properties is extremely difficult, if not impossible. Moreover, all optimization-based methods require that the direct problem must be solved first. This constraint of the need for iterated direct problem solutions can produce significant computing errors and calculations may be excessively time-consuming. This thesis offers new inverse models to estimate heat flux based on a sequence of measured temperatures. The offered models developed by ITs, in accordance with the achievement of this research, only requires a series of temperature-input heat data for a few minutes of operation; the dimensions and thermophysical properties are not needed. As another significant advantage, the estimation stage by the trained ITs only includes a small number of simple calculations excluding any recursive computation; this means the method is very fast-paced in comparison with classical avenues of numerical heat transfer for similar problems. At the outset, the most general form of ITs in engineering applications, Artificial Neural Networks (ANNs), employed to formulate an inverse model in the studied furnace/dryer (see chapter 4). The promising results confirmed that ITs are sound candidates to create inverse models. In that study, some deficiencies in ANNs such as finding the relevant parameters by trial and errors motivated the authors to check GA-ANNs and ANFIS as the possible alternatives for ANNs. The comparison study between aforementioned methods (see chapter 5) provided good outlines to find the best method in different situation. As the ANNs optimized by Genetic Algorithms (GA) discovered as the best method in the chapter 5, different types of ANNs were compared to find the best one (see chapter 6) in terms of accuracy and computation time. The results demonstrated that Multilayer Perceptron (MLP) optimized by GA can perform the best among all studied ANNs. Since the literatures lack of a practical comparison between the proposed and optimization based methods, as the next phase of study, these two method were compared (see chapter 7) to reconfirm the superiority of inverse models developed by ITs. In the last stage (chapter 8), a two-input/ two-output problem defined to check the capability of the proposed method in the problems more closer to the real-world industrial applications. In short, a series of very accurate methods for inverse heat transfer problems is proposed and successfully tested using experimental data.
Advisor: Chen, Lei
O'Neil, Brian K.
Dissertation Note: Thesis (Ph.D.) (Research by Publication) -- University of Adelaide, School of Chemical Engineering, 2017.
Keywords: inverse heat transfer problems
fuzzy logic
Research by Publication
Provenance: This electronic version is made publicly available by the University of Adelaide in accordance with its open access policy for student theses. Copyright in this thesis remains with the author. This thesis may incorporate third party material which has been used by the author pursuant to Fair Dealing exceptions. If you are the owner of any included third party copyright material you wish to be removed from this electronic version, please complete the take down form located at:
DOI: 10.4225/55/592630b550cda
Appears in Collections:Research Theses

Files in This Item:
File Description SizeFormat 
01front.pdf172.68 kBAdobe PDFView/Open
02whole.pdf16.57 MBAdobe PDFView/Open
PermissionsLibrary staff access only704.65 kBAdobe PDFView/Open
RestrictedLibrary staff access only5.77 MBAdobe PDFView/Open

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.