Global optimization of grillages using genetic algorithms

Abstract

The mathematical models and different solution algorithms for global optimization of pile placement schemes in grillage-type foundations are presented: the local search from a quasioptimal solution and the global search by Branch-and-Bound algorithm and genetic algo-rithms. The maximum in absolute value vertical reactive force arising at either support is to be minimized. The op-timization problem is nonlinear and nonconvex. The value of objective function is supplied by a “black box” finite element program. Comparison of different optimization techniques reveals the potential of genetic algorithms for this type of problems: the rational solution may be ob-tained in an appropriate for engineering practice solution time. Moreover, a several different topologies of solution with a close objective function values may be obtained in a several runs of genetic algorithm, and the most relevant for the engineering practice topology may be chosen.