Computationally enhanced techniques for practical optimum design of steel structures


Thesis Type: Doctorate

Institution Of The Thesis: Orta Doğu Teknik Üniversitesi, Faculty of Engineering, Department of Civil Engineering, Turkey

Approval Date: 2014

Student: SAEID KAZEMZADEH AZAD

Supervisor: OĞUZHAN HASANÇEBİ

Abstract:

Practical optimum design of structural systems via modern metaheuristic algorithms suffers from enormously time-consuming structural analyses to locate a reasonable design. This study is an attempt to reduce the computational effort of optimization process involved in real-life applications through development of alternative techniques to the existing computationally expensive methods. Basically two main approaches are considered as (i) investigating the algorithmic structure of the existing metaheuristics and enhancing their performances in sizing optimization problems (ii) developing new design-driven optimization techniques based on the principles of structural mechanics. In the first approach enhanced reformulations of modern metaheuristics are developed and tested using real-life instances. Furthermore, an upper bound strategy is proposed wherein non-improving candidate designs are identified and excluded from the structural analysis stage, diminishing the total computational effort. vi In the second approach a guided stochastic search (GSS) technique is developed wherein the search direction is determined by the principle of virtual work and response computations of the generated designs. In the GSS, the information provided in the structural analysis and design check stages are utilized for handling strength constraints. Moreover, the principle of virtual work is used to detect the most effective structural members for satisfying displacement constraints. The optimum sizing of a structure is then performed where both strength and displacement criteria are taken into account for reduction of the member sizes along the way the aforementioned constraints are handled. The numerical results indicate the computational efficiency of the proposed techniques in sizing optimization of steel skeletal structures.