Distributed power architecture in the context of the cost effective data center

Abstract

As new technology and serviw continue to feed the migration of networks from voice-based to data-based, multitudes of frames that required relatively small capacity DC feeds will he replaced with a relatively smaller quantity of digital devices hut that require much larger capaaty feeds. This dramatic shift in the power requirements of the network equipment is an opportunity for operators to explore new approaches in the distribntiou of DC power, as well as new methods in its deployment. It is also an opportunity to adopt methods that minimize the cosl in infrastructure.The traditional method of establishing centralized DC power plants, in mom located sometim= over a hundred f e t away from the loads that they feed, is loo rigid and eostly for the needs of tomorrow's operators. This paper presents a more flexible and cost-eftective approach, based on distributing multiple power frames, in clme proximity to the loads, eliminating the need for secondary distribution bays. The Traditional ArchitectureThe traditional architecture of power plants mirrors the architecture of the telecommunication systems to which it feeds power. The classic central ofilce, as its name suggests, consists of hardware and software deployed around a central element, the telephone switch. In the legacy central office, the DC power plant can as also be qualified as highly centralized, and can be described as a 3-layer model, consisting of: batteries;the secondary DC distribution.the rectifiers and the primary DC distribution;The first two layers of the model, the primary power infrastructure, the batteries and rectifiers are often located together, in a central location, the power room.