Name: BRENO DIAS BRÊDA
Type: MSc dissertation
Publication date: 19/03/2021
Advisor:
Name |
Role |
|---|---|
| ELCIO CASSIMIRO ALVES | Advisor * |
Examining board:
| Name |
Role |
|---|---|
| ELCIO CASSIMIRO ALVES | Advisor * |
| MARCOS ANTONIO CAMPOS RODRIGUES | Co advisor * |
| WALNÓRIO GRAÇA FERREIRA | Internal Examiner * |
Summary: A complete process of optimization of a steel structure depends on the calculated design stress, the code and steel shapes tables used in the design and the optimization algorithm considered. With that in mind, this work integrates different simplified geometric nonlinear analysis methods, such as the B1-B2 Method, the Two Cycles Iterative Method (TCIM) and the Purely Incremental Analysis (PIA), with the EN-1993-1-1 (2005), brazilian and european steel shapes tables and the structural optimization is associated to the process using the Genetic Algorithm (GA). To achieve this objective, the present work presents: (1) the validation of the nonlinear analysis module using ANSI/AISC 360-16 (2016) benchmark examples, (2) the comparison of the design module that uses EN-1993-1-1 (2005) with the results of a commercial software and, finally, (3) results of the GA optimization of benchmark structures in the literature and using the developed modules, observing possible gains. Results show that TCIM presents accurate results and low computational cost when compared to B1-B2 Method and PIA. The GA is efficient in the optimization of steel structures because it uses discrete variables and this facilitates the use of the steel shapes table provided by the manufacturers. Observed gains are associated: to the use of more accurate analysis methods, to the use of EN-1993-1-1 (2005) and to the use of different profile catalogs, according to the cases analyzed.
Keywords: Genetic Algorithm, Steel Structures, Geometric Nonlinear Analysis.
