Constraint Programming Enabled Automated Ship Hull Geometric Design

Abstract

Techniques for the generation of constrained and optimized geometry play a key role in the automated design of ships, particularly in the early stages. So called “form parameter design” systems are state-of-the-art for such ship hull geometry generation tasks. These systems solve a sequence of constrained nonlinear optimization problems to generate ship hull form geometry which conforms to constraints (form parameters). With such a system, a human expert, or search algorithm capable of discarding infeasible designs, is needed to pick constraint combinations that are at once feasible and also result in a high quality design. Simply finding a feasible set of form parameters is non-trivial enough that form parameter design of ship hulls has remained mostly an academic practice. Commercial tools exist, but their uptake by industry has been slow because this issue has given such tools a reputation for being hard to use. In this paper a modified approach for hull shape generation is proposed, combining form parameter design with interval constraint logic programming. We use a constraint solver to pre-process the design space, ensuring all constraints are feasible. This ensures that the form parameter design tool always operates within the feasible sub-domain of the total design space. The new approach is briefly described. Its effectiveness is demonstrated by using a random number generator together with the constraint solver to generate feasible design candidates, and then valid ship hull geometry is generated by our form parameter design program.