GENETIC ALGORTIHMS FOR MIXED DISCRETE/CONTINUOUS OPTIMIZATION IN MULTIDISCIPLI NARY DESIGN

Genetic algorithms have been applied to a simple demonstration problem which had previously been used in the validation of a multidisciplinary design optimization framework referred to as Concurrent Subspace Design. Discrete optimization in the most recent implementation of that framework was performed by simulated annealing. The results obtained by applying both genetic algorithms and simulated annealing to the demonstration problem demonstrated that the performance of both methods was very similar, both in terms of their ability to locate optimal designs and the computational requirements involved in doing so. It was therefore concluded that the capabilities and computational cost of Concurrent Subspace Design would be similar regardless of which discrete optimization method was incorporated into the framework.