A Global Optimization Model for Ship Design


This paper presents a global optimization model for ship design. A multicriteria optimization model for ship design has been identified as a problem with multiple local optima representing widely varying sets of designs. The model as developed handles the ship design optimization problem as a multicriteria constrained multivariable nonlinear optimization process and aims at a global optimum solution of the problem. The use of any local nonlinear constrained multivariable tool for such a problem can only assure the attainment of local optima which may be quite far from the global optimum. The model incorporates a global optimization tool, a decision system handler and several accepted naval architectural estimation methods. Simulated annealing is a method based on the analogy between problems of combinatorial optimization and statistical mechanics and has been successfully applied to the ship design optimization problem as a global optimization tool. The decision system handler is based on the analytic hierarchy process. Though they do not ensure the attainment of the global optimum, solutions from the process of simulated annealing successfully identify the main cluster which is very close to the reported global optimum predicted by pure random methods and genetic algorithms. The main cluster of points identified by simulated annealing provides the designer the region of safe and economic designs.