Use this URL to cite or link to this record in EThOS:
Title: Evolutionary topology optimization via direct and generative encodings : applications to aerospace and heat transfer engineering
Author: Ikonen, Teemu Johannes
ISNI:       0000 0004 7967 0591
Awarding Body: University of Southampton
Current Institution: University of Southampton
Date of Award: 2018
Availability of Full Text:
Access from EThOS:
Full text unavailable from EThOS. Please try the link below.
Access from Institution:
Evolutionary algorithms are global search methods that are well-suited for 'black-box' type objective functions and multi-objective optimization. However, as search methods in topology optimization, they have gained only limited acceptance, mainly due to their poor efficiency; they tend to require more objective function evaluations than gradient-based methods. Motivated by their benefits, the first aim of this work is to improve the performance, i.e. effectiveness and efficiency, of evolutionary topology optimization. We parameterize the design domains using both the ground structure approach (direct encoding) and L-systems-based methods (generative encoding). We investigate the use of two interpretation formalisms of L-systems, i.e. map L-systems and the turtle interpretation. In terms of improving the performance, the main contribution of this work is a statistical analysis of the effects of over 400 genetic control parameter combinations on the performance of the map L-systems-based method, which results we report as a Pareto front in the space of effectiveness and efficiency. The second aim of this work is to identify engineering applications to which L-systems-based methods are particularly suitable. We studied three applications, which are related to aerospace and heat transfer engineering. We found that the method with the turtle interpretation is well-suited to topology optimization of a heat conductor due to its natural tendency to produce bifurcating tree-structures. We show that the method is more effective in 10 out of 12 tested optimization problems and is two orders of magnitude more efficient on all 12 problems than a representative direct encoding method. In addition, our results indicate that the method is more effective than the well-established SIMP method (Solid Isotropic Material with Penalization) in optimization problems where the product of volume fraction and the ratio of high and low conductive material is less or equal to 1.
Supervisor: Sobester, Andras Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available