Interactive Optimisation in Marine Propeller Design
The significant amount of design variables related to blade design problems requires a systematic search in a large design space. Automated optimisation has been utilised for a number of blade design applications, as it has the advantage of creating a large set of design alternatives in a short period of time. However, automated optimisation has failed to be used in industrial applications, due to its complex set-up and the fact that in more complex scenarios the majority of the non-dominated design alternatives are infeasible. This necessitates a way of enabling the blade designers to interact with the algorithm during the optimisation process.
The purpose of this thesis is to develop a methodology that supports the blade designers during the design process and to enable them to interact with the design tools and assess design characteristics during the optimisation. The overall aim is to improve the design performance and speed. According to the proposed methodology, blade designers are called during intermediate stages of the optimisation to provide information about the designs, and then this information is input in the algorithm. The goal is to steer the optimisation to an area of the design space with feasible Pareto designs, based on the designer's preference. Since there are objectives and constraints that cannot be quantified with the available computational tools, keeping the "human in the loop" is essential, as a means to obtain feasible designs and quickly eliminate designs that are impractical or unrealistic.
The results of this research suggest that through the proposed methodology the designers have more control over the whole optimisation procedure and they obtain detailed Pareto frontiers that involve designs that are characterised by high performance and follow the user preference.
marine propeller design
interactive genetic algorithms
progressively interactive evolutionary computation
Chalmers, Mekanik och maritima vetenskaper, Marin teknik
Interactive evolutionary computation for propeller design optimization of wind-assisted vessels
AIAA AVIATION 2020 FORUM,; Vol. 1 PartF(2020)p. 1-10
Paper i proceeding
"I. Gypa, M. Jansson, K. Wolff, and R. Bensow. Propeller optimisation by interactive genetic algorithms and machine learning"
Strömningsmekanik och akustik
Chalmers tekniska högskola
Opponent: Ola Isaksson, Professor, Department of Industrial and Materials Science, Chalmers University of Technology, Sweden