Design approach and dimensional similarity in lay-out optimization of structural systems

A design approach to lay-out optimization of structural systems is proposed as a less software-intensive alternative to the analysis approaches of ground structure, and of homogenization. The proposed approach is implemented in two design methods based on cognitive modelling. A Lamarckian method is developed which perceives the layout optimization as one of force transmission problem, and a dimensional method perceives it as a selection problem. The proposed design approach can include buckling constraints. Furthermore, the layout of a structure may provide multiple failure paths so that its system reliability is an important criterion in the search for the optimal configuration. For a given space-load situation a corresponding generic structure can be hypothesized which attempts to satisfy the postulated function of force transmission. The generation of the hypothesized configuration starts from a primitive configuration with a minimal number of members and nodes, and the optimum is sought by growing, i.e. by addition of members and nodes to the primitive structure. Furthermore, for a given space-load situation, a measure of loading transfer intensity, known as the structural index can be formulated. Using this dimensional similarity concept, prediction of the optimal weight and proportion of a structure with a different structural index, without performing any optimization routine, can be facilitated once the optimal weight and proportion of a different, but configurationally equivalent, structure is known. Optimal design charts can be developed, so that lay-out selection can be facilitated at early design stages. The dimensional similarity approach can be a cost-effective alternative to the artificial intelligence and artificial neural nets approaches. The Lamarckian method is applied to several examples which have been solved using the ground structure approach, the homogenization method, and an interactive structural optimization system. The dimensional method is demonstrated with an application to a truss model of a jacket-type structure.