Authors:
Ramkumar, P.; Suribabu, C.R.;
Abstract:
Economical attention in the field of steel structures is gaining importance over the years and has motivated researchers for the development of many algorithms, such as : Genetic, Particle Swarm, Artificial Bee Colony,… etc. These algorithms aim at achieving optimal design of truss structures. In this paper, Differential
Evolution Algorithm with discrete design variables is applied to optimize the design of plane truss structures. Optimization of trusses aims at reducing their weight; i.e., in search of a most economic section. The optimum values for cross-sectional areas of truss members are to be such that they minimize the structural
weight of truss, without violating the constraints. The constraints are a specified range of area, as well as deflection and stress values with upper and lower limits which have to be satisfied by each and every individual member of the truss. Fitness values are calculated for the population from the objective function
derived and penalized if the constraints are violated. The process of differential evolution starts with an initial population and continues until the termination condition is satisfied. C++ programming language is employed for efficient operation. To study the efficiency of Differential Evolution Algorithm applied to truss
optimization, five plane trusses are designed. When compared with manually designed truss structures, the result of DEA is found to give robust and superior truss structures. Final results show that the optimized weight of trusses obtained is 30% less than that of the same truss designed manually. Ultimately, a common DEA applied program to optimize any plane truss is developed and optimal F and CR values are found.
Keywords:
Optimization, Plane truss, Differential evolution, Deflection, Stress