AC or DC—or Both?
In the past, telecommunications and data-communication operations were distinctly different entities. Housed in separate departments within a firm, each had its own unique power needs—and its own way of meeting those needs. But with the development of computer networks and the Internet, telecom and datacom centers have converged. As a result, power system designers must consider the many strategies for delivering two kinds of electrical power—alternating current (AC) and direct current (DC)—to the contemporary data center.
A bit of history is in order. From its inception the telecommunications industry—which has a much longer history than the information industry—used 48-volt DC power as its standard. Pioneering telecom engineers, faced with the need for guaranteeing high reliability of telephony operations, knew that DC power was the easiest and most efficient means of delivering power reliability.
More specifically, telecommunications equipment usually requires -48-volt DC power. Typically, the power systems rely on multiple parallel-redundant rectifiers that charge lead-acid storage batteries. Backup generators may be employed to maintain operation during a power event.
For the modern datacommunications industry, however, it was different. Computers require a DC power supply, and computer makers decided that the easiest way to deal with the issue was to convert the standard AC power from the mains into DC by providing each computer with its own power supply that converts AC to the required DC. Most emergency and backup power for data centers were—and still are—AC uninterruptible power-supply (UPS) systems.
AC vs. DC
The rise of mission-critical high-tech centers has fueled a discussion about the possibility of delivering DC power directly to such facilities. Peter Huber and Mark Mills are industry observers who are strong proponents of DC power plants. In an article that appeared last year in their newsletter, The Digital Power Report, the authors argue for the inherent superiority of DC-power plants: “DC power plants, in short, are inherently scaleable. Their strengths reinforce. Their weaknesses compensate.”
Huber and Mills are DC-power true believers. In the engineering community, however, one doesn’t necessarily find a consensus on this issue. In fact, there are some distinct differences of opinion. DC-power engineers—those who have designed for telecommunications facilities—tend to believe in the reliability of DC electrical-power systems. They also may be less convinced of the dependability of standby AC-power systems. Designers of AC systems are of the opposite opinion.
The coalescence of data and telecommunications has meant that designers of high-tech facilities will need to design a system around both AC and DC loads. The case studies presented below describe some of the possible solutions.
BridgePoint International Inc. of Montreal provides telecom/datacom centers for telecommunications, content providers, active service pages and Internet service providers. Company officials understand the company’s fundamental mission to be providing reliable power. Overall power reliability for all equipment at these co-location centers depends on 3-phase AC uninterruptible power-supply (UPS) systems, designed with a parallel bypass solution that accommodates up to four UPS units for redundancy and scalability.
However, DC power plants from the same supplier were also installed along with the AC equipment, primarily as a reliable source of DC power for telecom equipment. The multi-layered approach to power protection is the firm’s preferred choice these days.
The Hybrid Approach
The two case studies that follow involve telecommunications-industry giants that have moved rapidly to stay competitive as information and communications technologies have converged:
AT&T Corporation serves more than 80 million customers. When a power-system failure late one Friday evening threatened to cause a complete service outage at a local services switching office in Denver Colo., the company needed an emergency-power solution.
The supplier that AT&T officials turned to recommended installation of an emergency-recovery DC-power system (ERPS) at the site to power the load and charge the batteries. Within 24 hours, the supplier was able to assemble an ERPS at its plant and deliver the equipment to the Denver site. By Sunday, AT&T had all its customers back online.
Originally developed to protect power systems from Y2K-related interruptions, ERPS provides a plug-and-play solution to help a communications providers recover from almost any power event: AC utility problems such as spikes, surges, brownouts or over-voltage; weather-related problems like lightning strikes, tornadoes, flooding or hurricanes; and equipment-related issues including generator-induced harmonics and end-of-life on rectifiers and controllers.
The system includes rectifiers with 600-ampere total capacity, controllers, bussing, 5-foot cabinet on wheels, all necessary pre-attached AC/DC cables, additional 15-ft AC twist-lock connector and a simple-to-follow installation guide.
Nortel Networks, another global leader in telephony, data, e-business and wireless applications for the Internet, is experiencing phenomenal sales of its optical networking equipment that enable high speed, content-rich broadband services. Nortel’s optical networking business in 1999 grew more than 80% compared to the year earlier.
Nortel relies on surface-mount DC/DC power converters to ensure clean and reliable power for the firm’s fiber-optic networks. During 1999, its supplier was providing Nortel with 2,000 to 4,000 of these units each week. This year’s production has been increased to 6,000 to 9,000 DC/DC power converters per week. In addition, the supplier has implemented a multi-year program to meet the company’s increased needs for a wide range of power converter products used in its optical networks.
A large Internet-hosting company was building a number of sites nationwide. Each site was approximately 10,000 square feet, with a planned load density of up to 75 watts per square foot. The mix of AC- and DC-powered load was unknown, because it would depend on individual tenants’ selected equipment.
Facility designers set the criterion that up to 10 percent of the total load would be allowed for DC-powered equipment. To assure maximum availability, the design required redundant, dual-distribution power systems to support load equipment having dual input power connections.
The selected AC UPS configuration was distributed redundant 750-kVA systems—two independent systems with three 375-kVA modules each for N+1 redundancy—with 15 minutes of battery backup at full load. Permanently sited, redundant N+1 standby engine-generators were included. The ultimate maximum total -48VDC power load to be accommodated was 1,600 amps. The initial design required the ability to support half the ultimate load. All of the loads were expected to be dual-input DC- or AC-powered equipment. Single-input AC equipment was supplied through static transfer switches that allow the equipment to be powered from both UPS sources. Several different DC power system configurations were considered.
Two design alternatives were selected for detailed engineering and cost analysis. One configuration would use centralized 1,600-amp DC power plants. The alternative was distributed battery-less DC-rectifier systems—in capacities of 400 amps each—powered from the AC UPS systems.
The hybrid distributed DC-power system configuration met the high availability requirements of the Web-hosting company by using the dual-redundant AC UPS with redundant-power distribution paths. A high degree of fault tolerance was obtained without any single points of failure in the AC- or DC-power system. Any component within the AC- or DC-power system could fail or be maintained without disrupting the critical load equipment operation.
From Pure Power, Summer 2001.