System reliability optimization with k-out-of-n subsystems and changing k

A system reliability optimization approach with multiple k-out-of-n subsystems connected in series is presented. The design objective is to select multiple components to maximize system reliability while satisfying system requirement constraints such as cost and weight. Mixing of non-identical component types is allowed in each subsystem as well as at system level. In practice, different design configurations can be associated with different values of k. Thus k is considered a decision variable that can be selected as part of the design process. Previously, k was generally considered to be one to maximize the reliability of a series-parallel system. In this paper, k >; 1 is considered. The approach is demonstrated on a well-known test problem with interesting results. A Genetic algorithm is effectively applied to solve the optimization problem.