Abstract
Abstract This paper addresses the optimum design of axially loaded non-prismatic steel columns, in which the objective is to minimize their volume under a given load by changing their shape, and assuming that are subject to buckling and strength constraints. We use a genetic algorithm (GA) to move through the search space of possible column designs, and choose the best one. Several issues arise when using the GA, such as how to decide which is the most appropriate representation scheme and how to fine tune its parameters. Both floating point and binary representation (with and without Gray coding) were used and compared to a more traditional optimization technique based on the generalized reduced gradient method. Our results show that the floating point representation provides the best solutions overall, reducing the volume of the column up to 30% in some cases.
Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
