Optimum Design of Welded Steel Plate Girder using Genetic Algorithms

Abstract

The design of structures whether reinforced concrete or steel are governed by code provisions. In structural engineering the primary goal of optimization lies in either minimizing the weight of the structure or reducing its cost, keeping in view the overall safety and serviceability aspects. Engineering design is influenced by several parameters and associated constraints hence while designing a structure numerous solutions are obtained, thereby making it difficult to choose an optimal solution. The essence of Darwin’s principle of “Survival of the Fittest”, has led to the development of genetic algorithms which are nowadays being used as global optimization tools for identifying optimal or near optimal solutions for a given problem. Genetic Algorithms based on the mechanism of natural evolution, consider multiple design solutions at a time and cleverly explore the solution space to achieve a global minima or maxima. Genetic Algorithms work well than Classical optimization algorithms as these do not require the gradient information for the function to be optimized. Genetic algorithm based optimization approach works well on discrete, continuous or mixed search spaces, thereby making it versatile for engineering optimisation. The paper presents the capability of genetic algorithms as a directed search technique for optimum design of welded plate girder governed by the mixed nature of design variables.