For more information.......................................................................................................................... 14
Call to action .................................................................................................................................... 14
Abstract
The use of three-phase power distribution allows for simplifying hardware installation, achieving
higher efficiency, and reducing energy costs. A comprehensive three-phase power infrastructure
involves both rack-level (intra-rack) and row-level (inter-rack) solutions. This paper discusses the latest
trends in zoned (end-of-row) power distribution and describes HP solutions that provide customers with
a complete and efficient solution for three-phase power distribution.
Introduction
An enterprise data facility contains a large amount of IT equipment with substantial power
requirements. Energy usage has increased to the point that it is second only to labor in operating
costs for some large facilities. Increased server densities and subsequent rise in power demands on a
per-rack basis make it crucial to interface and balance the electrical load properly with facility power
to ensure safe and cost-efficient operation.
Power distribution in data centers
The typical data center uses power that originates from the utility power grid and is stepped down to
the appropriate range by the building transformer. Power within the data center can be distributed
using either a centralized or distributed infrastructure. Centralized systems provide utility or backup
power to IT equipment from a single distribution point that includes the alternating current (AC)
switchgear, uninterruptible power supplies (UPSs), and power distribution panels. This paper
describes a distributed (zoned) system (Figure 1) where utility or generator power from the AC
switchgear is distributed among equipment groups or rows. In a zoned system, each IT equipment
row includes its own UPS/power distribution infrastructure.
Figure 1. Data center using zoned power distribution
Building
generator
Utility
power grid
AC
switchgear
Building
transformer
UPS
rack
Cooling
Equipment
Power
Distribution
Rack
IT Equipment racks
2
Figure 2 shows basic power distribution at the row level of a data center. The primary distribution
components are the UPSs, the power distribution rack, and the power distribution cables. While this
paper describes the UPSs as being in front of (feeding) the power distribution panels, alternate
configurations are possible depending on data center needs.
Figure 2. AC power distribution at the row level
ServerRack
ServerRack
ServerRack
IT
Equipment
Rack
Power
Distribution
Units
415Y
or
208 VAC
UPS
Rack
UPS
UPS
UPS
UPS
UPS
UPS
415Y
or
208 VAC
Power
Distribution
Rack
Distribution
Panel
Distribution
Panel
415/230 VAC
or
208/110 VAC
Power
Distribution
Cables
Under normal conditions, the UPSs receive input power from the AC mains in the range of 415Y/230
VAC (international) or 208Y VAC (North America). If the input power rises above or falls below the
expected range or if it is interrupted completely, the UPSs go online and apply battery power to the
distribution panels. The UPSs will continue to provide battery power until utility power is restored to
within the accepted operating range or until the building generator can produce the required power,
whichever comes first. Once UPS input power resumes (either because the building generator comes
online or because utility power is restored) the UPSs will go back offline and resume their normal
mode of operation, which is to charge the batteries while allowing power from the generator to pass
through to the power distribution rack. The power distribution rack evenly distributes AC power
through the distribution cables to the IT equipment racks.
Concerns and trends in designing power infrastructures
The power distribution system of a data center should provide the following:
• Flexibility—easily adaptable to reconfigurations of the data center
• Scalability—expandable with IT infrastructure growth
• Reliability—providing constant service with no unscheduled downtime
• Efficiency—minimizing the cost of utility power and reducing greenhouse gas emissions
The actual power infrastructure of a data center will depend on several factors:
• Number of racks:
– Small data center: fewer than 20 racks
– Medium data center: 20 to 100 racks
– Large data center: more than100 racks
3
• Power density per rack:
– Low density: less than 6 kilowatts per rack
– Medium density: from 6 to 12 kilowatts per rack
– High density: more than12 kilowatts per rack
• Data center availability (Uptime Institute Tier Classifications)
– Tier I: single power path with non-redundant UPS (highest probable downtime)
– Tier II: single power path with redundant (N+1) UPS
– Tier III: dual power paths, each with redundant (N+1) UPS, one active and one passive
– Tier IV: dual power paths each with redundant, active UPSs that allow concurrent maintenance
(lowest probable downtime)
Each data center is unique in its requirements based on business needs. The only constant in data
centers is change, and the best solution will involve components that allow data centers to change
and grow easily. For example, adding one or two racks in today’s high-density blade systems can
add significantly to the power infrastructure load. Therefore, it is essential that data centers be
adaptable to change.
Modular systems allow customers to use a flexible “pay as you grow” strategy and address the issues
of adaptability. Three-phase modular systems offer flexibility and cost-effective power distribution.
Why three-phase power?
As power requirements for high density compute systems increase, the benefits of distributing threephase power to individual racks become more significant. Strictly speaking, IT equipment (ITE) does
not use three-phase power
balancing for improved efficiency, cable reduction for simplicity, and larger power rating capability
and expandability.
With single-phase power, the voltage across the hot and neutral conductors can be anywhere
between its peak (maximum) and zero at any given time, and electrical conductors must be large to
meet high amperage requirements. Three-phase power uses four discrete conductors (three hot and
one neutral) handling three cycles that are 120 degrees out of phase (Figure 3). The more constant
voltage across the three hot conductors results in smoother current flow and allows the use of smallergauge conductors to distribute the same amount of AC power.
Figure 3. Comparison of single- and three-phase AC voltage waveforms
Single-phase AC
0 V
1 hot, 1 neutral
1
—the benefits are in its distribution. These benefits include easier load
Three-phase AC
3 hot, 1 neutral
The load balancing and increased power handling capabilities of three-phase AC distribution can
result in more efficient and economical power distribution. The key components of three-phase power
distribution include the UPS, the PDR, and distribution cables.
1
Some HP BladeSystem blade and power enclosures connect directly to three-phase power and internally
distribute split-phase or single-phase power to the power supply units.
4
Three-phase UPS technology
The main purpose of the UPS is to maximize the availability of IT equipment by providing a constant
power source for IT equipment, regardless of the state of the AC power the UPS receives. UPS
performance is determined by its design parameters and topology.
UPS load considerations
A UPS should be designed based on two main factors: the AC input feed, and the needs and
characteristics of the active output load. In this case, the active output load is comprised of the power
supply units (PSUs) of the IT equipment. Table 1 identifies the primary PSU parameters to which a UPS
must be designed.
Table 1. Primary parameters of power supply units that affect UPS design
Typically low-line (100 to 127 VAC)
or high-line (200 to 240 VAC) input
power. High-density server systems
often use power supplies that can
operate with either low-line or highline power, but require high-line
power to operate at maximum rated
performance and peak efficiency.
Frequency tolerated by most power
supply units is typically within the 47
to 64 Hz range to accommodate
worldwide applications.
Power supply circuitry that
compensates for the power loss
inherent with reactive components
using AC voltage
Ability of a power supply unit to
provide rated output during an AC
power interruption, typically 12 to
20 ms.
Ability of a power supply unit to
absorb AC line sags (brownouts) and
surges (spikes)
UPS must maintain a three-phase output with
single-phase and split-phase components
that meet the acceptable range of the power
supply units.
UPS must maintain output frequency within
the acceptable range
The cumulative power requirement of all
power supply units in a circuit branch should
not exceed UPS volt-ampere (VA) and watt
(W) ratings
UPS mode transitions (offline-to-online,
online-to-offline) need to occur much faster (5
ms or less) than the power supply unit
holdup time to ensure constant operation of
power supply unit.
UPS must prevent passing on AC line
transients that would adversely affect power
supply units.
In addition, the UPS must meet the cumulative power consumption of all power supply units included
in the output circuit. While UPS specifications generally include both volt-amperes (VA) and watt (W)
ratings, the rating in watts provides a truer indication of the loading ability of the UPS. The watts
rating will be stated with a time parameter indicating how long the UPS can provide the rated power
in the online (battery) mode.
5
Loading...
+ 9 hidden pages
You need points to download manuals.
1 point = 1 manual.
You can buy points or you can get point for every manual you upload.