Global optimization of grillage-type foundations using a distributed genetic algorithm

Abstract

In this paper, the optimal design of grillage-type foundations is considered. A difficult black-box global optimization problem should be solved to find the optimal positions of piles in this type of foundations. The goal of the research was to enable the solution of real grillage problems by distributed (volunteer) computing using the open infrastructure for network computing BOINC. A distributed genetic algorithm has been developed as well as its implementation on a computational BOINC platform. The algorithm is adapted to suit the BOINC platform and cope with possible occasional faults of client computers. The results of the proposed algorithm are compared with those attained by using the simulated annealing algorithm running on a computational grid infrastructure which previously showed the best performance. The results on 10 real grillage foundations and quantitative comparison reveal that the performance of the proposed algorithm is better.