Design and optimization of 3D printed air-cooled heat sinks based on genetic algorithms

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

34 Scopus citations

Abstract

Enhancing power density and reliability of power electronics is extremely important in power electronics applications. One of the key challenges in the design process is to design the optimum heat sink. In this paper, an algorithm is proposed to design air-cooled heat sinks using genetic algorithm (GA) and finite element analysis (FEA) simulations. While the GA generates a population of candidate heat sinks in each iteration, FEA simulations are used to evaluate the fitness function of each. The fitness function considered in this paper is the maximum junction temperature of the semiconductor devices. With an approach that prefers 'survival of the fittest', a heat sink providing better performance than the conventional heat sinks is obtained. The simulation and experimental evaluations of the optimized air-cooled heat sink are also included in the paper.

Original languageEnglish
Title of host publication2017 IEEE Transportation and Electrification Conference and Expo, ITEC 2017
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages650-655
Number of pages6
ISBN (Electronic)9781509039043
DOIs
StatePublished - Jul 26 2017
Event2017 IEEE Transportation and Electrification Conference and Expo, ITEC 2017 - Chicago, United States
Duration: Jun 22 2017Jun 24 2017

Publication series

Name2017 IEEE Transportation and Electrification Conference and Expo, ITEC 2017

Conference

Conference2017 IEEE Transportation and Electrification Conference and Expo, ITEC 2017
Country/TerritoryUnited States
CityChicago
Period06/22/1706/24/17

Keywords

  • 3-phase inverter design
  • 3D printing
  • Forced-air cooled heat sink
  • Genetic algorithm
  • Power density
  • Thermal management

Fingerprint

Dive into the research topics of 'Design and optimization of 3D printed air-cooled heat sinks based on genetic algorithms'. Together they form a unique fingerprint.

Cite this