Algorithm Data Science student designs best ship in less time
Partly thanks to Roy de Winter, ship design and engineering company C-Job Naval Architects now develops optimal ships in a short time. The master's student Computer Science from Leiden developed the CEGO algorithm, which eliminates the classic design cycle and makes people in the maritime optimization process redundant.
Do all desired compartments fit into the hull? What impact does the ship have on the environment? And what are the total costs of ownership? In the design phase, shipbuilders deal with all sorts of variables, limitations and objectives. With every minimal change in the design of a ship, they must reconsider their previous choices at the drawing board. This time-consuming and repetitive work now belongs to the past: an optimisation algorithm makes is possible to develop better ships in less time.
From student to engineer
Roy de Winter developed the algorithm CEGO (Constrained Efficient Global Optimization) together with Thomas Bäck and Bas van Stein, scientists from the Leiden Institute of Advanced Computer Science (LIACS). Data scientist De Winter did his research at C-Job Naval Architects as part of his master's programme in Leiden. He has written about his results in his thesis 'Designing Ships using Constrained Multi-Objective Efficient Global Optimization'.
De Winter is now employed as a research & development engineer at C-Job Naval Architects, the largest independent maritime engineering company in the Netherlands. 'What I really like about working at an engineering firm is that you work on a product that you can actually touch. This means that the code I am currently writing will eventually be converted into a ship, that will actually go into the water! Working on a real product in combination with the latest developments in science is what I have always wanted to do', says De Winter.
New way of working
Shipbuilders have to make assessments continuously. For example: should the ships they design be short or long? A short design is cheaper to build, but will be more expensive once in use. And while the construction of a longer design is more expensive, it's operating costs will be lower. Shipbuilders simply cannot oversee all links between the decisions they make and their eventual consequences. The CEGO algorithm however is able to do this. From now on, CEGO will assist shipbuilders in making the right choices. This new way of working is much faster and is therefore called the 'Accelerated Concept Design' method.
No people required
A ship designer used to have to make his own choices, but now CEGO can take well thought-out decisions, eliminating annoying repetitive work. This makes the design process considerably faster. Where the desinging process used to take up months, now an optimal ship can be designed in 2 to 4 weeks. After the shipbuilders have done their work using CEGO and the Accelerated Concept Design, a set of optimised design variants remains. These are designs of a ship's hull, the layout of the ship, and a steel model. Moreover, all variants comply with the laws of physics, the rules of the authorities and customer demands. The customer can then choose which optimised design will be further developed.
Maritime Designer Award
For his innovative contribution to the maritime industry and the application of insights from mathematical optimisation techniques, De Winter has been nominated for the 2018 Maritime Designer Award. 'This research shows that interdisciplinary work can lead to great new insights. I hope the jury will see this and will award my research. That would be an extra incentive to make the algorithm even better than it already is', says De Winter. The awards ceremony takes place on 12 November 2018 during the Maritime Awards Gala.