An approximate model for optimizing Bernoulli columns against buckling

• Published in: Engineering structures Vol. 141; pp. 316 - 327
• Main Authors: Ruocco, E., Wang, C.M., Zhang, H., Challamel, N.
• Format: Journal Article
• Language: English
• Published: Kidlington Elsevier Ltd 15.06.2017; Elsevier BV; Elsevier
• Subjects: Algorithms; Axial loads; Buckling; Columns (structural); Comparative studies; Cross-sections; Differential equations; Differential geometry; Engineering Sciences; Evolutionary algorithms; Genetic algorithms; Geometry; Load distribution (forces); Mathematical models; Optimization; Parameter optimization; Stress concentration; Structural engineering
• ISSN: 0141-0296; 1873-7323
• DOI: 10.1016/j.engstruct.2017.01.077

•The paper presents a new model able to define the optimal design of columns against buckling.•Columns are inhomogeneous and subjected to concentrated and distributed loads.•An optimization scheme based on a parallel genetic algorithm is here proposed.•The model is free of any form of load or geometrical constrains.•The model deals with a large range of sample types. Proposed herein is a simple but powerful method for optimization of inhomogeneous, elastically restrained columns against buckling when subjected to both compressive concentrated and distributed axial loads that include self-weight. Unlike previously published studies on the subject, we do not have to specify any prescribed geometrical variation and analysis may be readily performed on columns with any complex geometrical shape. In the proposed method, the differential equation governing the buckling of Euler columns is discretized by adopting the Hencky bar-chain model, and critical buckling loads are evaluated by seeking the lowest eigenvalue of the resulting system of algebraic equations. The discrete nature of the formulation, as well as the reduced number of parameters to be optimized, is well suited for the adopted optimization process that is based on evolutionary algorithms. We propose an optimization scheme based on a parallel genetic algorithm. A comparison study between the obtained optimal column shape and buckling loads on homogeneous and isotropic columns with circular cross section, and the numerical and analytical solutions found in the open literature shows fast convergence, high accuracy and flexibility of the proposed method.