Structural optimization of hollow-section steel trusses by differential evolution algorithm

This paper deals with the weight minimization of planar steel trusses by adopting a differential evolution-based algorithm. Square hollow sections are considered. The design optimization refers to size, shape and topology. The design variables are represented by the geometrical dimensions of the cross sections of the different components of the truss, directly involving the size of the structure, and by some geometrical parameters affecting the outer shape of the truss. The topology is included in the optimization search in a particular way, since the designer at different runs of the algorithm can change the number of bays keeping constant the total length of the truss, to successively choose the best optimal solution. The minimum weight optimum design is posed as a single-objective optimization problem subject to constraints formulated in accordance with the current Eurocode 3. The optimal solution is obtained by a Differential Evolutionary (DE) algorithm. In the DE algorithm, a particular combination of mutation and crossover operators is adopted in order to achieve the best solutions and a specific way for dealing with constraints is introduced. The effectiveness of the proposed approach is shown with reference to two case-studies. The analysis results prove the versatility of the optimizer algorithm with regard to the three optimization categories of sizing, shape, topology as well as its high computational performances and its efficacy for practical applications. In particular useful practical indications concerning the geometrical dimensions of the various involved structural elements can be deduced by the optimal solutions: in a truss girder the cross section of the top chord should be bigger than the one of the bottom chord as well as diagonals should be characterized by smaller cross sections with respect to the top and bottom chords in order to simultaneously optimize the weight and ensure an optimal structural behaviour.