Stochastic optimization of system performance using cold-standby redundancy

摘要

The target life of a system cannot be shortened, in many practical applications so that the system has to be designed to perform at least up to that time. Such situations are common in space applications, defense applications with high performance such as communications and missile carrying flights. The systems meant for such purposes should have to function for the required period of time without failure. Adding redundancy is one way to ensure this. However without proper allocation redundancy may not ensure this. The two types of redundancies normally used are standby or passive redundancy and active redundancy. In a redundancy state if a unit carries a weak load then it is called partial redundancy and the unit is said to be in warm state. When the unit is passive redundancy it is said to be in cold standby state. In cold redundancy the unit works when a parallel system fails. It is assumed that a unit cold state will not fail and take over from a failed unit will not cause system failure in case of perfect switching. In case of imperfect switching, the system can work successfully with probability 1 - p when the unit fails and the redundant unit starts working. The decision to have active or passive redundancy depends on the expected system performance and cost considerations. No general rule can be applied to the optimal allocation of cold redundancy. The proposed method is not design or component specific and hence can be applied to all types of system designs and component-life distributions. (6 refs.)