A computer-based, interactive genetic algorithm optimisation design tool (GENOD) for reinforced concrete structures

This paper describes the application of genetic algorithms to the optimum cost design of realistic reinforced concrete structures, set within an artificial intelligence computer design environment. The interactive optimisation design tool GENOD developed by authors combines structural analysis with design and offers advantages over the current use of computer technology in design offices. Traditional design of reinforced concrete structures requires the adherence to precise guidelines as specified in the relevant Codes of Practice. This process relies on the designer's intuition and experience and often it is not clear which direction will lead towards a more economical structure. GENOD, however, replaces this conventional approach by a systematic, goal-orientated design process that uses artificial intelligence in searching and sorting through similar design concepts to achieve an economical design. Genetic algorithms are implemented using an object-orientated visual programming language offering facilities for continual monitoring, assessing and changing the current state of the search control parameters. These facilities are shown to be essential when determining the most suitable control parameter settings for a given structural problem. Results obtained so far have shown that genetic algorithms can be successfully applied to the minimum cost design of reinforced concrete skeletal structures, overcoming the difficulties associated with the practical assessment of the structural costs, discontinuity of the design equations and their complex interrelationship with the design variables.