A multicriteria optimisation model for reliability design of series‐parallel systems: part 1

Abstract

Purpose

Reliability is a major quality characteristic which has grown in importance as products/systems have become ever more sophisticated. Neglecting it could spell great losses in terms of patronage, revenue, and even lives. The purpose of this paper is to present a multi‐criteria optimisation model and methodology for the Pareto optimal assignment of reliability to the components of a series‐parallel system in order to maximise its reliability.

Design/methodology/approach

The subsystems' reliabilities are maximised independently but simultaneously in order to maximise the overall system reliability, while a penalty function modelling cost of reliability improvement is minimised. The resultant continuous and nonlinear optimisation problem is scalarised by a convex combination of the criteria and the MATLAB Optimisation Toolbox is used to generate the solutions.

Findings

The results for an illustrative example problem extracted from the literature show that: higher reliabilities could be assigned to the components, in order to achieve or exceed target system reliability; cost increased sharply with slight improvements in the component reliabilities, and the model was stable under the weighting scheme used.

Originality/value

The novelty of this work is in: the multi‐criteria optimisation view taken of the design problem; the focus on the subsystems' reliabilities and cost as the criteria to be optimised; the use of the two aforementioned qualities for the purpose of Pareto assignment of component reliabilities in a system's design; and the use of the model and methodology in the context of series‐parallel systems.

Citation

Twum, S., Aspinwall, E. and Fliege, J. (2012), "A multicriteria optimisation model for reliability design of series‐parallel systems: part 1", International Journal of Quality & Reliability Management, Vol. 29 No. 9, pp. 1038-1055. https://doi.org/10.1108/02656711211272944