Documents disponibles écrits par cet auteur

Optimal thin-film topology design for specified temperature profiles in resistive heaters / François Mathieu-Potvin in Journal of heat transfer, Vol. 132 N° 10 (Octobre 2010) 

Public ISBD [article]  in Journal of heat transfer > Vol. 132 N° 10 (Octobre 2010) . - pp. [101302-1/9] Titre : Optimal thin-film topology design for specified temperature profiles in resistive heaters Type de document : texte imprimé Auteurs : François Mathieu-Potvin, Auteur ; Louis Gosselin, Auteur Année de publication : 2010 Article en page(s) : pp. [101302-1/9] Note générale : Physique Langues : Anglais (eng) Mots-clés : Inverse  condition  Thin-film  Topology  optimization  Uniform  temperature  resistive  heater  Conjugate  gradient  Genetic  algorithm Index. décimale : 536 Chaleur. Thermodynamique Résumé : In this paper, we optimized the topology of a thin-film resistive heater as well as the electrical potential of the electrodes on the boundaries. The objective was to minimize the difference between the actual and prescribed temperature profiles. The thin-film thickness was represented by 100 design variables, and the electrical potential at each electrode were also design variables. The topology optimization problem (inverse problem) has been solved with two methods, i.e., with a genetic algorithm (GA) and with a conjugate gradient method using adjoint and sensitivity problems (CGA). The genetic algorithm used here was modified in order to prevent nonconvergence due to the nonuniqueness of topology representation. The conjugate gradient method used in inverse conduction was extended to cope with our electrothermal problem. The GA and CGA methods started with random topologies and random electrical potential values at electrodes. Both the CGA and GA succeeded in finding optimal thin-film thickness distributions and electrode potential values, even with 100 topology design variables. For most cases, the maximum discrepancy between the optimized and prescribed temperature profiles was under 0.5°C, relative to temperature profiles of the order of 70°C. The CGA method was faster to converge, but was more complex to implement and sometimes led to local minima. The GA was easier to implement and was more unlikely to lead to a local minimum, but was much slower to converge. DEWEY : 536 ISSN : 0022-1481 En ligne : http://asmedl.aip.org/vsearch/servlet/VerityServlet?KEY=JHTRAO&ONLINE=YES&smode= [...] [article] Optimal thin-film topology design for specified temperature profiles in resistive heaters [texte imprimé] / François Mathieu-Potvin, Auteur ; Louis Gosselin, Auteur . - 2010 . - pp. [101302-1/9]. Physique Langues : Anglais (eng) in Journal of heat transfer > Vol. 132 N° 10 (Octobre 2010) . - pp. [101302-1/9] Mots-clés : Inverse  condition  Thin-film  Topology  optimization  Uniform  temperature  resistive  heater  Conjugate  gradient  Genetic  algorithm Index. décimale : 536 Chaleur. Thermodynamique Résumé : In this paper, we optimized the topology of a thin-film resistive heater as well as the electrical potential of the electrodes on the boundaries. The objective was to minimize the difference between the actual and prescribed temperature profiles. The thin-film thickness was represented by 100 design variables, and the electrical potential at each electrode were also design variables. The topology optimization problem (inverse problem) has been solved with two methods, i.e., with a genetic algorithm (GA) and with a conjugate gradient method using adjoint and sensitivity problems (CGA). The genetic algorithm used here was modified in order to prevent nonconvergence due to the nonuniqueness of topology representation. The conjugate gradient method used in inverse conduction was extended to cope with our electrothermal problem. The GA and CGA methods started with random topologies and random electrical potential values at electrodes. Both the CGA and GA succeeded in finding optimal thin-film thickness distributions and electrode potential values, even with 100 topology design variables. For most cases, the maximum discrepancy between the optimized and prescribed temperature profiles was under 0.5°C, relative to temperature profiles of the order of 70°C. The CGA method was faster to converge, but was more complex to implement and sometimes led to local minima. The GA was easier to implement and was more unlikely to lead to a local minimum, but was much slower to converge. DEWEY : 536 ISSN : 0022-1481 En ligne : http://asmedl.aip.org/vsearch/servlet/VerityServlet?KEY=JHTRAO&ONLINE=YES&smode= [...]