Reliability-Equivalent-Separation of a Node

A criterion is obtained for separating a node into two or more nodes without changing the network reliability. If a network with exactly one input node, a network with exactly one output node or a network composed of only nonoriented lines has such a reliability-equivalent-separable node, then the network can be separated into two or more s-independent subnetworks by separating the node into two or more nodes. The network reliability can be obtained as a product of the subnetwork work reliabilities. If a network has such a separable node, then computation of the network reliability can be appreciably reduced by separating the nodes. Examples are provided.