Cloud Computing and Generative Ship Designs for New Age of Shipbuilding

Abstract

Empirical estimations are basis for preliminary design of ships. The errors in these estimations may be marginal or significant depending upon vessel configuration and operation. The present paper talks of a method where the ship design begins with requirements' input. This in turn simulates generative designs, identifying the design space by fitting the design against required criteria and assumptions, then generates CAD model using NURBS surface, and performs iterative optimization and decision making based on multi-variable parameters. Since these are actual CAD models, it helps in better estimation of actual parameters in ship design which improves empirical estimations. This process reduces significant errors and becomes critical in preliminary ship design. With rising cloud computing focus, these calculations are able to be performed faster and more efficiently, helping naval architects to identify design space faster; this paper serves as a baseline for exploring more parallel algorithms using cloud computing on the given philosophy.

Given case is a container ship utilizing global optimization algorithms with freight rate as objective. Focus is on generating full ship design, including compartmentalization, utilizing generative designs and no designer involvement.