Definition of conditions of transition from corrective recovery to planned recovery strategy of uninterruptible power supply

B. Yu. Volochiу, D. S. Kuznetsov

Lviv Polytechnic National University

The problem of the conditions definition of the transition from the corrective recovery to the planned recovery strategy of an uninterruptible power supply, while a reliability is kept on the same level, is considered in this paper.
We consider the uninterruptible power supplies (UPS), which have a modular structure of the power supply unit (PSU). They are designed as a fault-tolerant systems (FTS) with N+M and 2×(N+M) redundant configurations, where N - is number of the PSU modules required to power the workload, and M - the number of the redundant modules.
When there are the specific time intervals where an object, where the UPS is installed, is inaccessible, or there are problems with an access to it, it is reasonable to perform the transition from the corrective recovery (CR) to the planned recovery (PR) strategy of the UPS. In order to perform this, the parameters of the FTS and the recovery strategy must be defined. Those parameters must provide the equivalent UPS's reliability.
For the UPS with N+M redundant configuration the CR strategy is applied (UPS №1). The CR strategy is performing recovery immediately after the finding the fact that UPS reached threshold condition (but can be operable as well). From this moment begins the countdown of recovery procedure.
For an UPS with 2×(N+M) redundant configuration the PR strategy is applied (UPS №2). It differs from the previous one in that the recovery procedure starts at the pre-determined scheduled time moments. In these time moments the condition of the UPS is compared with its threshold operable condition and a decision to send a repair crew to the object is made. If the condition does not meet the threshold condition, the repair crew is not sent.
To solve this problem the models of fault-tolerant systems as the systems of Kolmogorov - Chapman differential equations had been developed. With its help, the dependence of the mean time of a failure-free operation (TFFO) of the UPS №1 on the duration of the recovery of its components have been determined. For the input data, the periodicity values of the PR (TPER) for the UPS №2, which provides the equivalent reliability with the UPS №1 by the TFFO index, have been determined. Though, those recovery strategies, while providing the same mean times of a failure-free operation, provide the different values of  the probability of a failure for a given operation time.