Automated ship hull form design optimisation through morphing and evolutionary computation

Ang, Joo Hock (2019) Automated ship hull form design optimisation through morphing and evolutionary computation. PhD thesis, University of Glasgow.

Abstract

Eco-friendliness and energy efficiency are now key criteria in the design and construction of all new ships due to stringent environmental regulations aimed at reducing their carbon footprint. An efficient hull form helps in reducing the overall drag acting on a ship, as well as reducing fuel consumption and harmful emissions to the environment. More recently, Simulation-Based Design (SBD) methods have been proposed to improve the efficiency of the hull form design process. However, the entire process still requires considerable human inputs, and success depends heavily on the designer’s experience and knowledge.

To address these limitations, an industry-first design automation tool based on Morphing and Evolutionary Computation (MEC) was developed in this thesis. This was achieved by combining a novel two-dimensional (2D) curve morphing method and a Nondominated Sorting Genetic Algorithm II (NSGA-II) using proven hull form designs as initial starting points. The 2D curve morphing method provides an efficient way to modify the hull form whilst providing more flexible geometric variations. NSGA-II is a multi-objective optimisation algorithm that uses nature-inspired operations to optimise a population of candidate solutions in a generative manner. This tool supplements experienced designers in modifying the shape of the hull form, in a more automated manner, helping them to achieve more diverse and optimal design solutions.

Central to the working of MEC are novel crossover and mutation mechanisms, developed using cross-morph and boundary-morph. Together, these two morphing techniques enable geometric manipulation to be performed directly on the hull, instead of modifying the design parameters as is the case in traditional approaches. By applying the two morphing techniques within NSGA-II, more diverse shapes can be discovered, and the entire design process can further be automated.

MEC is particularly well-suited for the initial design stage, where shape diversity and efficient performance evaluation are key to finding more optimal designs. The final output of this method is a set of Pareto optimal designs, which can further be refined by the designer. Through digitisation and computational intelligence, it is envisioned that automated design will help to improve the overall design efficiency and capability to produce more efficient and smarter ships in the future.

• Item Type: Thesis (PhD)
• Qualification Level: Doctoral
• Additional Information: This research is funded under Economic Development Board (EDB) of Singapore and Sembcorp Marine Ltd. (SCM) under Industrial Postgraduate Programme (IPP) grant no: COY-15-IPP/140002.
• Keywords: Hull form design and optimisation, simulation-based design, curve morphing, evolutionary computation, design automation.
• Subjects: V Naval Science > VM Naval architecture. Shipbuilding. Marine engineering
• Colleges/Schools: College of Science and Engineering > School of Engineering
• Supervisor's Name: Goh, Professor Cindy
• Date of Award: 2019
• Depositing User: Mr Joo Hock Ang
• Unique ID: glathesis:2019-75054
• Copyright: Copyright of this thesis is held by the author.
• Date Deposited: 28 Oct 2019 11:30
• Last Modified: 04 Aug 2023 08:48
• Thesis DOI: 10.5525/gla.thesis.75054
• URI: https://theses.gla.ac.uk/id/eprint/75054

Related URLs

• Enlighten Publications Record
• Article DOI