Tampere University of Technology

TUTCRIS Research Portal

Optimization of Fin Arrays Cooled by Forced or Natural Convection

Research output: Book/ReportDoctoral thesisCollection of Articles

Details

Original languageEnglish
PublisherTampere University of Technology
Number of pages71
ISBN (Electronic)978-952-15-4175-9
ISBN (Print)978-952-15-4171-1
Publication statusPublished - 24 Aug 2018
Publication typeG5 Doctoral dissertation (article)

Publication series

NameTampere University of Technology. Publication
Volume1558
ISSN (Print)1459-2045

Abstract

Electronic components must be cooled to maintain their operating temperatures below the specified limits. If the maximum permissible limit of a component is exceeded, its service life decreases considerably. With increasing power densities in recent decades, the use of heat sinks to improve component cooling has become virtually mandatory in many applications. However, designing a heat sink, which optimally compromises its material weight and its heat transfer performance, is a difficult task because the result depends heavily on its geometrical structure and its operating conditions.

In this thesis, a fast way to optimize industrial heat sinks with a fixed set of heat dissipating components is presented. In a typical optimization case, several hundred temperature field evaluations are needed to find the optimal geometry. These evaluations consume a lot of CPU time if done with conventional CFD. The main objective of this thesis is, therefore, to present a new calculation model, which can handle these temperature field evaluations in a much shorter time. In the model, the speedup is obtained by replacing the slow 3D CFD -solution of air velocity and temperature distributions with 1D solutions for the mean values of these distributions, where convective heat transfer and shear stress are calculated from analytical correlations. A complete 3D numerical solution is only performed for the solid temperature field. These modifications make the new model at least a thousand times faster than CFD.

The calculation model is then tested for accuracy in many test cases, where its results are compared to those calculated with CFD and analytical solutions. These comparisons ensure that the model operates with the precision needed for optimization to predict the maximum temperature of the components. This is important because, in optimization, the maximum temperature of the components is the most crucial quantity.

After accuracy testing, the use of the model as part of an efficient multi-objective optimization algorithm is demonstrated in many distinct cases. Instead of just one optimal solution, multiobjective optimization results in a set of best compromise solutions, called the Pareto optimal set, according to the chosen criteria. Usually, the optimization criteria are the maximum temperature of the components and the weight of the material, or the external volume, of the heat sink. A wellperformed optimization can allow a significant reduction of the solid material used. In the heat sink manufacturing industry, the potential for total savings in material, energy, and CO2 emissions is significant as the global market size of thermal management technology is vast.

Field of science, Statistics Finland

Downloads statistics

No data available