An Optimality Criteria-Based Method for Discrete Design Optimization under Buildability Constraints

This paper presents a heuristic algorithm for discrete design optimization. Practical applicability is the first concern; special attention is therefore paid to the implementation of buildability constraints. The method is used in this paper for the optimal design of a Warren type truss composed of steel CHS members with welded joints. For each member, the optimal section is chosen from a given steel catalog. Following Eurocode 3 and the CIDECT design guidelines, restrictions are imposed on the deflections, the member forces, the layout of the joints, and the joint forces. The buildability of the design is ensured by requiring all top chord members and all bottom chord members to have the same outer diameter. Practically all methods that have recently been proposed for this type of optimization problems are based on evolutionary computing. However, these methods are slow, and they involve algorithmic parameters that require careful tuning. A new algorithm is therefore proposed in the present paper, based on the optimality criteria method. The algorithm is faster than an evolutionary method and free of tuning parameters. It is successfully applied to the example truss design optimization problem.