Development of structural optimization and neural network design tool for buckling behaviour of skin-stringer structures under combined compression and shear loading

---

Tezin Türü: Yüksek Lisans

Tezin Yürütüldüğü Kurum: Orta Doğu Teknik Üniversitesi, Fen Bilimleri Enstitüsü, Türkiye

Tezin Onay Tarihi: 2019

Tezin Dili: İngilizce

Öğrenci: AYDIN OKUL

Danışman: Ercan Gürses

Özet:

Stiffened panels are commonly used in aircraft structures in order to resist high compression and shear forces with minimum total weight. Minimization of the weight is obtained by combining the optimum design parameters. The skin panel dimensions, the stringer spacing and the stringer dimensions are some of the critical parameters which affect the global buckling behaviour of the stiffened panel. The aim of this thesis is to develop a neural network design tool and to carry out a geometric optimization for panels having a large number of stringers under combined loadings. Before the design tool creation, a simplified panel with minimized number of stringers and the boundary conditions to be substituted for the side stringers are found. Then the effect of some critical design parameters on the buckling behavior is investigated and it is determined which parameter is used in order to increase or decrease the strength effectively. In the Artificial Neural Network (ANN) phase, approximately seven thousand finite element (FE) models are created and analyzed in ABAQUS FE program with the help of a script written in Phyton. The script changes the parametric design variables for the analyses and collect the results. These design variables and analysis results are grouped together in order to create an ANN in MATLAB NNTOOL toolbox. This process allows faster determination of buckling analysis results than the traditional FE analyses. This tool can predict the critical buckling load and the margin of safety for a given geometry and given loads. In the last phase, a structural optimization study is carried out by using MATLAB OPTIMTOOL toolbox for a specific region with given external dimensions and the applied compression-shear loads by using genetic algorithm method. The only constraint is not to buckle under given loadings. In this way, optimum weight is obtained for optimum design variables, which are the stringer and skin thicknesses, stringer dimensions and stringer placement.