Authors: Colin Bell; Cristinel Mares; Romeo Glovnea
Addresses: Faculty of Technology, Design and Environment, Oxford Brookes University, Wheatley Campus, Wheatley, Oxfordshire, OX33 1HX, UK ' Department of Mechanical Engineering, School of Engineering and Design, Brunel University, Uxbridge, Middlesex, UB8 3PH, UK ' School of Engineering and Design, University of Sussex, Brighton, Sussex, BN1 9RH, UK
Abstract: Manufacturing companies face tough competition across global markets with continuous demands to decrease product-development-cycle time whilst lowering costs without compromising quality. This paper presents a method of improving the efficiency of one aspect of the development cycle: functional optimisation. This optimisation technique is based around a modified genetic algorithm. This is used to develop a set of dimensions that fulfil stated, functional targets relating to the performance of a continuously variable transmission. These targets are prioritised based on an adapted Pugh's decision matrix for a number of different applications to automatically obtain weighted targets. Additionally, it is shown that for this particular problem, stochastic restarting of the genetic algorithm can lead to superior results without affecting computational time. This paper demonstrates that the methodology discussed can be successfully applied to a number of multi-objective problems in order to quickly yield the most favourable set of dimensions.
Keywords: design optimisation; continuously variable transmissions; CVTs; genetic algorithms; multicriteria optimisation; functional optimisation.
International Journal of Design Engineering, 2014 Vol.5 No.3, pp.232 - 255
Received: 19 Jan 2013
Accepted: 17 Nov 2013
Published online: 04 Jun 2014 *