ε-constraint guided stochastic search with successive seeding for multi-objective optimization of large-scale steel double-layer grids

This paper proposes a design-driven structural optimization algorithm named ε-constraint guided stochastic search (ε-GSS) for multi-objective design optimization of large-scale steel double-layer grids having numerous discrete design variables. Based on the well-known ε-constraint method, first, the multi-objective optimization problem is transformed into a set of single-objective optimization problems. Next, each single-objective optimization problem is tackled using an enhanced reformulation of the standard guided stochastic search algorithm proposed based on a stochastic maximum incremental/decremental step size approach. Moreover, a successive seeding strategy is employed in conjunction with the proposed ε-GSS algorithm to improve its performance in multi-objective optimization of large-scale steel double-layer grids. The numerical results obtained through multi-objective optimization of three challenging test examples, namely a 1728-member double-layer compound barrel vault, a 2304-member double-layer scallop dome, and a 2400-member double-layer multi-radial dome, demonstrate the usefulness of the proposed ε-GSS algorithm in generating Pareto fronts of the foregoing multi-objective structural optimization problems with up to 2400 distinct sizing variables. •ε-constraint guided stochastic search (ε-GSS) algorithm is developed for multi-objective design optimization.•A successive seeding strategy is proposed to enhance the performance of the ε-GSS in multi-objective structural optimization.•An enhanced reformulation of the guided stochastic search algorithm is proposed for discrete sizing optimization.•The proposed ε-GSS is applied to multi-objective optimization problems of large-scale steel double-layer grids.