An approach for determining the best maintenance policy strategy in power infrastructures using stochastic modeling

Abstract

Data centers are constantly growing, driving demands for new technologies such as cloud computing, e-commerce, which has forced the availability of these systems 24 hours a day, 7 days a week under penalty of great losses. In this way, in addition to static or dynamic redundancy techniques applied to its components for a correct and continuous functioning of these systems, it is of great importance to adopt adequate strategies for maintenance policies. Our study proposed analytical models along with a methodology with the aim of determining the most appropriate maintenance policy strategy, based on both technical and business-oriented aspects, which will be adopted in a power infrastructure architecture. The technical aspects refer to the unavailability metrics resulting from the architecture/infrastructure from different maintenance strategies. In turn, the business-oriented aspects refer to the labor cost metrics, associated with these strategies, applied to the considered infrastructure architectures. The case study used a power infrastructure in architectures with increasing levels of redundancy for the application of the proposed methodology and models. The results showed that the most suitable maintenance strategy for an architecture/infrastructure does not necessarily have the lowest unavailability, but it suits together the lowest unavailability and the lowest maintenance costs.