This week's seminars
Statistics Seminars: Stochastic renewal process models for exact unavailability quantification of highly reliable systems
21 May 2018 14:00 in CM221
In previous research an original methodology for high-performance reliability computing was developed which enables exact unavailability quantification of a real maintained highly reliable system containing highly reliable components with both preventive and corrective maintenance. Whereas the methodology was developed for systems containing components with exponential lifetime distribution, main objective of this research is generalization of the methodology by applying stochastic alternating renewal process models, so as to be used for unavailability quantification of systems containing arbitrary components without any restrictions on the form of the probability distribution assigned to time to failure and repair duration, i.e. ageing components are allowed. For this purpose a recurrent linear integral equation for point unavailability is derived and proved. This innovative equation is particularly eligible for numerical implementation, because it does not contain any renewal density, i.e. it is more effective for unavailability calculation than the corresponding equation resulting from the traditional alternating renewal process theory, which contains renewal density. The new equation undergoes the process of discretization which results in numeric formula to quantify desired unavailability function. Found component unavailability functions are used to quantify unavailability of a complex maintained system. System is represented by the use of directed acyclic graph, which proved to be very effective system representation to quantify reliability of highly reliable systems.
Contact firstname.lastname@example.org for more information