Keywords: genetic algorithm, optimization, pre-stress, design, concrete beam, evolution, topology.

A genetic algorithm-based methodology for the structured design is developed in this study. This methodology approaches the design task without the use of defined theoretical designs and obtains sizing, configuration, and topology of the design simultaneously. The final design is evolved under a selection demand that favors survival of those designs that possess specified features. This methodology optimizes design while it is being designed. Because this methodology is not constrained by conceptual designs and engineering experience, it has the potential of evolving truly optimum and innovative designs, especially when applied to complex design problems and is applicable even to those design problems that are lacking in past experience.

In this study the genetic algorithm-based methodology used to design pre-stressed concrete beams (PCB). This method is used to obtain set of optimal geometrical dimensions of symmetrical I-beam cross section defined by four measurements; including the overall cross section thickness, flange width, flange thickness, and width of the web. Additionally, the amount of pre-stressing steel is optimized. Post-tensioned pre-stressed beam with a single duct of parabolic shape is considered in the application. Several parameters are studied including the effect of the span length considering different loading cases. The performance constraints are adopted according to the ACI 318/95 Building Code provisions; including the flexural stresses, the ultimate moment capacity of the section with respect to cracking moment, the maximum crack width, the immediate deflection and the long term deflection in addition to the side constraints.

A genetic algorithm-based methodology for the structured design of the pre-stressed concrete structures is developed in this study. Designs are forced to satisfy specified requirements and to optimize the selected cost function using a genetically-based approach. It simultaneously carries out topological, configurational, and sizing designs. The results are obtained using an evolving genetic algorithm program and the objective function selected to minimize the overall cost of the beam in terms of concrete and pre-stressing steel. As a result, several design charts and their interactive curves are developed and presented. These charts are set for future comparison as well as designers use. This study presented the promising capabilities of the genetic algorithm in evolving to efficient and innovative designs, and showed the practicability of the genetic algorithm for different structural optimization problems.

The development of the genetic algorithm, which is an adaptive computation methodology of the genetic on the biological process of natural selection, has presented an opportunity for the development of fundamentally different approach for structural design. The adaptive computation paradigm of the genetic algorithm has made possible the development of a new methodology, which operates, directly on the physical model as seen in the traditional approach. The freedom from the mathematical formulation helps this methodology to avoid limitations of the traditional approach and makes it applicable to complex design situations. All these features, together with the robustness and efficiency of the genetic, have presented a potentially rewarding approach to structural design.

purchase the full-text of this paper (price £20)

go to the previous paper
go to the next paper
return to the table of contents
return to the book description
purchase this book (price £95 +P&P)