Voting for enhanced security - how far is far enough

Microprocessor relays have reduced the cost of protection by an inflation adjusted factor of ten or more in the past 25 years. At the same time, the availability of fault reports and sequence of events from these relays have improved our understanding of how relays respond to fault, and non-fault events. One problem with these relays is that by using one relay for an entire protection scheme it has become possible for that one relay to incorrectly operate without supervision from a separate relay. Advances in microprocessor relays make it possible to apply voting schemes without significant cost increases although just adding additional relays is also possible. This paper discusses the fault tree analysis mathematics that prove the advantage of voting schemes. New relay measurement methods that provide for multiple systems in the same device are introduced. Application techniques to incorporate voting are considered such as double line protection, using relay-to-relay logic for voting implementation, and reversion to nonvoting on relay alarm. Of course much more than relay capabilities need to be involved in the selection of an overall relaying system. Installation, operating, and testing considerations are brought up and evaluated. Conclusions regarding appropriate installations and applications are presented.