Optimum design of reinforced concrete skeletal systems using non-linear programming techniques.

The last 30 years has seen a rapid development in a wide variety of different techniques and approaches applied to solve optimisation problems in structural design. However,
recent surveys have shown insufficient penetration of optimisation methods in structural design practice, especially in the field of optimum design of realistic reinforced concrete structural systems. This research investigates new approaches in the use of cost-efficient optimisation and applies these to the multi-level design of skeletal systems. It has produced implementation theory and general computer programs for the automatic optimum
(improved) design of realistic, rigidly jointed reinforced concrete structures based on the ultimate limit state theory, as embodied in BS 8110. Guided by the adopted problem-seeks-optimum design approach and conscious of its practical application, this research provided novel approaches to the development of realistic cost objective function formulations, to design constraint handling, to the assessment of multiple and worst-scenario loading arrangements and to the development of a model that groups
structural elements in a manner that both improves efficiency of the optimisation algorithm and mirrors design office practice. The validity of structural optimisation has been established, dependent directly on the balance between the mathematical model of the objective function and the design
constraints, the algorithm that is applied, and the physical reality of the structural problem and its practical application. Hence, taking into consideration the main
components of skeletal system superstructures and substructures, suitable optimisation algorithms were implemented, and their performance reported to the structural problem formulation and the limitations encountered. The research has shown that optimising structural problems with single load case does not give a realistic minimum cost of a
structure, and that frames consisting of multiple beam and column groups in general produce a more cost efficient design. Further work is suggested both for improving
structural problem formulations and for implementing appropriate optimisation techniques.