ABSTRACT
During decades the initial dimensioning of ships during the preliminary design stage was based on diagrams and statistics obtained from samples of similar ships (Murphy et al., 1965). Since the second half of the XX’s century more numeric methods become popular due to the computing resources available. The initial approaches were based on the systematic variation of a set of design variables. Each combination of variable values defined a possible ship that was next submitted to a set of constraints to verify if it was technically feasible. The optimum ship was selected from the feasible ones based on an objective function such as the ship’s weight or initial cost. This method is not considered as an optimization because the generation of new sets of variables is not directly obtained from the results of the previous iteration. Later, methods based on direct search algorithms started to be applied. More recently evolutionary methods based on concepts inspired on natural phenomena, such as genetic algorithms, particle swarm optimization, ant colony optimization or artificial immune system have been tested in ship design applications.
