Development of fault tree for reliability studies of a data-processing system

Abstract

In this paper a systematic approach is developed for generating the fault-tree logic, which is the basic requirement in any reliability studies of large complex systems using Monte Carlo simulation or deterministic approaches. The entire network is decomposed into sub-networks, each sub-network involving several loops, and it becomes quite easy to determine failure logic for these loops and thus for the subnetworks. Failure of such subnetworks between the source and the sink node leads to a major system failure and therefore this successive development results in a fault tree for the whole network. These loops can be automatically generated. The failure event for each node in such an approach can also be accounted for. The advantage of the present methodology is that the fault-tree logic is automatically developed, which can be directly used as a sub-programme on a computer for various reliability studies of the system.