Tampere University of Technology

TUTCRIS Research Portal

Tubular truss optimization using heuristic algorithms

Research output: Book/ReportDoctoral thesisCollection of Articles

Details

Original languageEnglish
Place of PublicationTampere
PublisherTampere University of Technology
Number of pages106
ISBN (Print)978-952-15-1891-1
Publication statusPublished - 2007
Publication typeG5 Doctoral dissertation (article)

Publication series

NameTampereen teknillinen yliopisto, Julkaisu
PublisherTampere University of Technology
Volume706
ISSN (Print)1459-2045

Abstract

The topic of this thesis is discrete tubular truss optimization using heuristic multipurpose algorithms. The multicriteria topology, shape and sizing optimization problem has been formulated based on practical real life needs. This means that in addition the material cost also the manufacturing cost has been taken into account, design constraints satisfy the demands of steel design rules (Eurocode 3) and the selection of available tubular profiles does not have to be in a special way limited. Mass, cost or displacement have been used as the objective function criteria in the example problems. The design constraints ensure that truss members and joints are strong enough, single members or the whole truss do not buckle, certain displacements are small and natural frequencies do not locate in forbidden intervals. The tubular truss optimization problem has been solved using four multipurpose heuristic algorithms. Two of them are local search algorithms simulated annealing (SA) and tabu search (TS) and two of them population based methods genetic algorithm (GA) and particle swarm optimization (PSO). The efficiency of heuristic algorithms has been studied empirically in several example problems. In the first academic ten-bar truss example problem heuristic algorithms have also been compared to a different type of branch&bound algorithm. The rest of the example problems deal with tubular plane and space trusses. Also the conflict of mass and cost and the effect of the number of different design variables to the final solution have been studied.

Publication forum classification

Downloads statistics

No data available