Failure analysis informing intelligent asset management

Abstract

With increasing demands on the UK’s power grid it has become increasingly important to reform the methods of asset management used to maintain it. The science of Prognostics and Health Management (PHM) presents interesting possibilities by allowing the online diagnosis of faults in a component and the dynamic trending of its remaining useful life (RUL). Before a PHM system can be developed an extensive failure analysis must be conducted on the asset in question to determine the mechanisms of failure and their associated data precursors that precede them. In order to gain experience in the development of prognostic systems we have conducted a study of commercial power relays, using a data capture regime that revealed precursors to relay failure. We were able to determine important failure precursors for both stuck open failures caused by contact erosion and stuck closed failures caused by material transfer and are in a position to develop a more detailed prognostic system from this base. This research when expanded and applied to a system such as the power grid, presents an opportunity for more efficient asset management when compared to maintenance based upon time to replacement or purely on condition.