Abstract
This paper adds to the work of Venugopal et al. (1986 a, b) and Nakagawa (1979, 1984) on optimization techniques for reliability systems based on varied criteria. The new concept of the ‘cost-optimal-(N ∗, n ∗) system’, based on comprehensive cost considerations is introduced and results obtained for optimal parallel reliability systems (that is, with optimal units, N ∗, subjected to optimal repairs, n ∗). The operational use and applicational value of the results are demonstrated through numerical work concentrating on some typical random laws governing the stochastic failure patterns.