On optimal grouping and stochastic comparisons for heterogeneous items

Abstract

In this paper, we consider series and parallel systems composed of n independent items drawn from a population consisting of m different substocks/subpopulations. We show that for a series system, the optimal (maximal) reliability is achieved by drawing all items from one substock, whereas, for a parallel system, the optimal solution results in an independent drawing of all items from the whole mixed population. We use the theory of stochastic orders and majorization orders to prove these and more general results. We also discuss possible applications and extensions. © 2017 Elsevier Inc.