Professional Documents
Culture Documents
A B C A B C N N
B C A B C A
N
N
Patent Pending Trefoil Cable Configuration.
Come visit us at
the IEEE PES T&D show in Dallas!
For further information please email us at: sales@unitedwc.com
Tel: 1-800-265-8697 • Fax: 1-800-461-4689
Going to great lengths to serve you unitedwc.com
ISO 9001
Certified
FEATURES
Coordinating protective 8 ❮❮
contents
cover story
devices in mission
critical facilities
2 Understanding transfer A coordination study ensures that
switch operation the most reliable electrical system
has been installed. Applicable
Consulting engineers should understand codes and standards help
transfer switch construction, performance engineers get it right.
requirements, selection criteria, and
desired operation.
Reviewing, analyzing 14 ❮❮
NEC 2014 changes
PUBLICATION SERVICES Changes in and additions to the
1111 W. 22nd Street, Suite 250, Oak Brook, IL 60523 NFPA 70: National Electrical Code
phone: 630-571-4070 fax: 630-214-4504
(NEC) have a significant impact on
Jim Langhenry Patrick Lynch
commercial and industrial facilities.
Co-Founder and Publisher, CFE Media Director of Content
jlanghenry@cfemedia.com Marketing Solutions
plynch@cfemedia.com
22 ❮❮
Steve Rourke
Co-Founder, CFE Media Paul Brouch Using warm water
srourke@cfemedia.com Director of Operations
pbrouch@cfemedia.com for data center cooling
Trudy Kelly
Assistant to the Publisher Rick Ellis
There are many ways to cool a
tkelly@cfemedia.com Audience Management Director data center. Engineers should
rellis@cfemedia.com
Kristen Nimmo explore the various cooling options
Marketing Manager Michael Rotz
knimmo@cfemedia.com
and apply the solution that’s
Print Production Manager
mike.rotz@frycomm.com appropriate for the application.
Elena Moeller-Younger
Marketing Manager
emyounger@cfemedia.com
CONTENT SPECIALISTS/EDITORIAL
Amara Rozgus Amanda Pelliccione
Editor in Chief/Content Manager Director of Research
arozgus@cfemedia.com apelliccione@cfemedia.com
PUBLICATION SALES
Midwest International
Matt Waddell Stuart Smith
312-961-6840 SSM Global Media Ltd.
Fax 630-214-4504 +44 208 464 5577
mwaddell@cfemedia.com Fax +44 208 464 5588
stuart.smith@ssm.co.uk
West, TX, OK
Tom Corcoran Poland/Russia/Czech Republic
215-275-6420 Michael Majchrzak
Fax 484-631-0598 +48 22 852 44 15
tcorcoran@cfemedia.com Fax +48 22 899 29 48
mike_majchrzak@trademedia.us ON THE COVER:
U.S., Canada
Brian Gross The photo shows paralleling
630-571-4070 x2217 switchgear equipment that serves
Fax 630-214-4504
bgross@cfemedia.com a large hospitality venue in
Northeast
Las Vegas. The property used
Richard A. Groth Jr. paralleling switchgear to combine
774-277-7266
Fax 508-590-0432 electrical capacity from multiple
rgroth@cfemedia.com generators while providing load
management and transfer.
Pure Power is published quarterly by CFE Media and is mailed as a supplement with Courtesy: JBA Consulting
Consulting-Specifying Engineer and Plant Engineering magazines. Copyright 2015 by CFE Media
Engineers
LLC. All rights reserved. Editorial offices are located at 1111 W. 22nd Street, Suite 250,
Oak Brook, IL 60523. Phone 630-571-4070.
www.csemag.com/purepower
❯❯
PURE POWER // WINTER 2015
2 Cover Story
By Ryan Ishino, PE, RCDD, LEED AP, JBA Consulting Engineers, Irvine, Calif.
Understanding
transfer switch operation
Consulting engineers should understand transfer switch construction, performance requirements,
selection criteria, and desired operation to ensure that critical systems and equipment are supplied
with reliable backup power when needed.
W
hen utility power is BASIC TRANSFER SWITCH OPERATION
LEARNING OBJECTIVES interrupted, power Transfer switches are responsible for transitioning electri-
Explain basic transfer switch system failure is cal power from the primary source to a secondary source
operation. not an option for in the event of primary source interruption, maintenance,
Describe types of standby many facilities. Standby power or failure. The primary source most commonly consists of
systems and transfer switch systems have many components, the utility service. The secondary source typically consists
requirements. including transfer switches that of the backup or emergency power source. The sequence of
Compare types of transfer must be designed correctly. Dur- operation typically occurs as follows:
switches and their operations. ing power transitions, transfer
Evaluate transfer switch switch timing and sequence is 1. The primary source is interrupted or fails.
construction and performance critical to ensure proper system
requirements.
operation. Consulting engineers 2. When the secondary source is stable and within
must understand transfer switch voltage and frequency tolerances, the transfer switch
types, timing requirements, ratings, and the types of transitions to the secondary power source. This tran-
standby systems where transfer switches are used to sition can occur automatically or manually.
transfer to backup power. The basis of this article is
NFPA 70-2014: National Electrical Code (NEC) unless 3. When the primary source is restored and stabilized, the
otherwise noted. transfer switch transitions back to the primary source
and resumes under
normal operation.
This transition
back to the primary
source can occur
automatically or
manually.
www.csemag.com/purepower
Cover Story 3 ❮❮
STANDBY SYSTEM TYPES approved by the AHJ. Transfer equipment shall be designed
Standby system types include emergency systems, legally and installed to prevent inadvertent, simultaneous connec-
required standby systems, optional standby systems, criti- tion of primary and secondary supplies of power. Power
cal operations power systems (COPS), and systems that must be transferred to the secondary source in 60 sec or less.
support health care facilities (see Figure 1). Optional standby systems (NEC Article 702): Optional
Emergency systems (NEC Article 700): Emergency sys- standby systems are defined by the NFPA as “intended
tems are defined by the NFPA as “intended to automatically to supply power to public or private facilities or property
supply illumination, power, or both, to designated areas where life safety does not depend on the performance of
and equipment in the event of failure of the normal supply the system.” These systems may include data processing
or in the event of accident to elements of a system intended and communication systems, and mission critical systems
to supply, distribute, and control power and illumination that are not legally required by the AHJ.
essential for safety to human life.” These systems may Transfer equipment, including transfer switches, for
include fire detection and alarm systems, elevators, fire optional standby systems are not restricted to the same
pumps, and egress lighting. requirements as emergency and legally required system
Transfer equipment, including transfer switches, are transfer equipment. However, transfer equipment shall
required to be automatic, identified for emergency use, and be designed and installed to prevent inadvertent, simul-
approved by the authority having jurisdiction (AHJ). Trans- taneous connection of primary and secondary supplies
fer equipment shall be designed and installed to prevent of power. There are no code requirements for power to
inadvertent, simultaneous connection of primary and sec- be transferred to the secondary source within a certain
ondary supplies of power. Transfer equipment shall supply time frame.
only emergency system loads. Power must be transferred to Critical operations power systems (COPS) (NEC Ar-
the secondary source in 10 sec or less. ticle 708): Interruptions or outages to designated critical
Legally required standby systems (NEC Article 701): operations areas may negatively impact national security,
Legally required standby systems are defined by the NFPA economy, public health, or safety. The requirement to
as “intended to automatically supply power to selected comply with NEC Article 708 is provided by any govern-
loads (other than those classed as emergency systems) in mental agency having jurisdiction or by a facility provid-
the event of failure of the normal source.” These systems ing documentation establishing the necessity for such a
may include heating and refrigeration systems, communi- system. These systems may include power systems, HVAC,
cations systems, ventilation and smoke removal systems, fire alarm, security, and communications in these areas.
and other processes that, when stopped in the event of pri- NFPA 1600-2013: Standard on Disaster/Emergency Manage-
mary source interruption, could create hazards or hamper ment and Business Continuity Programs contains further
rescue or firefighting operations. information on this topic.
Transfer equipment, including transfer switches, are Transfer equipment, including transfer switches, are
required to be automatic, identified for standby use, and required to be automatic and identified for standby use.
www.csemag.com/purepower
❯❯
PURE POWER // WINTER 2015
4 Cover Story
Figure 3: The diagram shows a closed-transition transfer switch operation. The make-before-break operation ensures the switch will connect to the
secondary source while connected to the primary source.
Transfer equipment shall be designed and installed to pre- This means that the transfer switch creates a connection
vent inadvertent, simultaneous connection of primary and to the secondary source while connected to the pri-
secondary supplies of power. mary source (see Figure 3). When the connection to the
Health care facilities (NEC Article 517): Essential elec- secondary source is established, the primary source will
trical systems for hospitals consist of the emergency system disconnect. This enables a continuous source of supply
and equipment system to supply a limited amount of light- to the electrical system as the two sources are paralleled
ing and power essential for life safety and effective hospital together. Paralleled (or interconnected) sources shall com-
operation when the normal service is disconnected. The ply with NEC Article 705: Interconnected Electric Power
number of transfer switches used “shall be based on reli- Production Sources, which addresses basic safety require-
ability, design, and load considerations” in accordance with ments related to parallel operation of generators and
NEC Article 517.30(B)(4). Each branch on the emergency normal/primary sources (commonly the utility service).
system and equipment system, respectively, shall have one Closed-transition switches transfer when both sources
or more transfer switches to serve the system loads. How- are synchronized in phase, voltage, and frequency. The
ever, one transfer switch shall be permitted in a facility length of the synchronization period where both sources
with a maximum demand on the essential electrical system are paralleled is usually limited by the utility company’s
of 150 kVA or less. NFPA 99-2015: Health Care Facilities interconnect agreement and requirements.
Code has additional requirements for transfer switch opera- Fast closed-transition transfer switches: Fast closed-
tion and features. transition transfer switches use a momentary paralleling
of sources (typically less than 100 msec) using a control
TRANSFER SWITCH TYPES system similar to the open-transition transfer switch
Transfer switch types include open-, closed-, fast closed-, system. Some fast closed-transition transfer switches use
soft closed-transition, and bypass/isolation. passive synchronization to sense the phase relationship
Open-transition transfer switches: Open-transition between the two live sources (in the event of parallel-
transfer is commonly described as “break-before-make.” ing) and allow interconnection of sources when they are
This means that the transfer switch disconnects from the synchronized. This is considered passive because there
primary source before establishing the connection to the is no direct control over the generator frequency, and the
secondary source (see Figure 2). There is a short-duration sources are paralleled for such a short duration of time.
electrical system outage during this transition. In addition, Although the intent is to not parallel the sources for an
open transition, by design, does not allow paralleling of extended amount of time, utility service providers com-
the two sources at the same time. Open-transition transfer monly require reverse-power relay protection to protect
switches are the most commonly-used type. They are less their systems from sustained paralleled operation. Fast
expensive than other options. closed-transition switches are more expensive than open-
Closed-transition transfer switches: Closed-transition transition switches but less expensive than soft closed-
transfer is commonly described as “make-before-break.” transition switches.
www.csemag.com/purepower
Cover Story 5 ❮❮
www.csemag.com/purepower
❯❯
PURE POWER // WINTER 2015
6 Cover Story
Transfer Switches, and NFPA 110-2016: Standard for Emer- NFPA 110: NFPA 110 is commonly applied and adopted
gency and Standby Power Systems provide transfer switch within the U.S. Chapter 6 of the standard includes require-
construction and performance requirements. ments for transfer switch equipment. Requirements in the
UL 1008: UL 1008 is the most commonly applied and standard related to automatic transfer switch (ATS) features
adopted standard to address transfer switch construction specify that their capabilities must include:
and testing in the U.S. This standard applies to automatic,
manual or nonautomatic, closed transition, hybrid, fire Electrical operation and mechanical holding
www.csemag.com/purepower
All the capabilities you want...
and some you may not have considered!
1-800-225-5250 russelectric.com
An Employee-Owned Company / An Equal Opportunity Employer
Coordinating
protective devices
in mission critical facilities
A coordination study ensures that the most reliable electrical system has been installed.
Applicable codes and standards help engineers get it right.
A
sudden power failure will have a dramatic effect device. A thermal magnetic breaker may have no adjust-
on business, especially in a critical environment. ment at all, or only minimal adjustment to the instanta-
Isolating a fault condition to the smallest area neous region, whereas a fully adjustable electronic trip
possible is essential in providing the most reliable breaker may have many.
electrical system with maximum uptime for your facility. Adjustment of these parameters allows for what is re-
Expensive electronic distribution protection equipment is ferred to as “curve shaping.” Curve shaping enables better
not worth the extra cost unless a proper protective-device coordination between upstream and downstream overcur-
coordination study is provided by an experienced engineer. rent-protection devices. Typical parameters include:
NFPA 70-2014: National Electrical Code (NEC) defi nes
selective coordination as: “Localization of an overcurrent Overload region (long time trip unit): This is the long
condition to restrict outages to the circuit or equipment time trip setting of an overcurrent protective device.
affected, accomplished by the selection and installation This parameter, also known as continuous amperes, is
of overcurrent protective devices a percentage of the breaker’s nominal rating.
and their ratings or settings
LEARNING OBJECTIVES for the full range of available Long time delay: This setting allows for inrush from
Illustrate the basics of overcurrents, from overload to motors to pass without tripping the breaker. This set-
protective-device coordination
the maximum available fault ting affects the position of the I2t slope just below the
studies.
current, and for the full range continuous-current setting.
Outline elevator protection
of overcurrent protective device
coordination as required per
the NEC. opening times associated with Short-time pickup: This setting is typically provided
those overcurrents.” with an adjustment of 5 to 10 times the inrush cur-
Apply NEC Article 517 and
ground-fault coordination In other words, a properly rent. This setting allows downstream overcurrent-
studies required for health care coordinated system will limit protection devices to clear faults without tripping
facilities. disconnection to the nearest upstream devices. This setting can also be adjusted to
upstream protective device. allow for transformer inrush current.
www.csemag.com/purepower
Protective Device Coordination 9 ❮❮
Time in seconds
PD-0008
the breaker. HLD, RMS
HLD
Trip 600 A
Plug 600 A 10
Ground fault delay: This setting allows for Settings phase
Plug adjust (300-600
a time delay before ground fault pickup. A) 500 A (500 A)
LTD (fixed) fixed
This allows for better selective coordina- STPU (2-8 x LTPU)
7 (3,500 A)
tion between multiple levels of ground fault STD (instantaneous-
300 msec) 100
protection. In addition, the time delay cannot msec (I2t on) 1
Instantaneous (5,500
exceed 1 sec (60 cycles) for ground-fault cur- A) fixed (5,500 A)
rents of 3,000 A or more.
XF2-0003
Reduced arc flash mode: This setting allows Inrush factor 10
TX inrush Inrush time 0.1 sec
the breaker to be manually taken out of PD-0009 0.10
J250 E/S/H
coordination for short periods of time during Trip 50 A
Plug 50 A
maintenance to reduce the arc flash hazards Settings phase
Ir, 100 A sensor C (50) (50 A)
on the system. LTD, (2-24 sec) 24
STPU, (2-12 x Ir) 12 (600 A)
DTD, (flat) fixed (700 A)
0.01
When performing electrical engineering stud-
ies for mission critical environments, the required
documentation includes:
Figure 1: The TCC graph (above) and
Description, rating, make, and Type FDE/FDCE/HFDE one-line diagram (left) indicate a 150-kVA
Sensor/trip 225 A
catalog numbers of protective Setting LTPU H (225) (225 A) transformer protected by a 225-A circuit
devices PD-0007 breaker. The “Tx” refers to the transformer
inrush in red. The 225-A breaker curve
Full-load current at the protec- CBL-0006 is represented by the blue curve. This
tive device (3-phase and line-to- breaker curve is to the right of the “Tx,”
ground) ensuring that the transformer inrush will
XF2-0003
not trip the breaker during system startup.
Transformer kVA, impedance, CBL-0007 The red curves on the top of the graph
and inrush current data and represent the transformer damage curves.
Type HLD, RMS 310
connection type (delta-wye, etc.) Sensor/trip 600 A The breaker should be as far to the left
PD-0008 Setting LTPU 500 A (500 A)
of the damage curves as possible. The
BUS-0012
Available fault current at the green breaker is in series with the blue
protective device PD-0009 PD-0010 breaker for this reason. The breakers do
Type J250 E/S/H, DT 310+ Type J250 E/S/H, DT 310+
not have to fully coordinate because loss
Cable and conductor sizes Sensor/trip 50 A
Setting LTPU C (50) (50 A)
Sensor/trip 50 A
Setting LTPU C (50) (50 A)
of either breaker will have the same effect
on downstream loads. The breaker shown
Protective-device design require- in purple has to fully coordinate with the
ments from the serving utility green breaker because there is a breaker in parallel with it (PD-0010). If
not coordinated, a short downstream of the purple breaker could trip the
Voltage at each bus. green breaker. If this was to happen, power would be lost on the PD-0010
breaker. All graphics courtesy: Lane Coburn & Associates
www.csemag.com/purepower
❯❯
PURE POWER // WINTER 2015
10 Protective Device Coordination
It is common today to perform complicated electrical Article 620-62. Figure 2 shows an example of a coordi-
protection coordination studies with computer software. nation study illustrating the feeder-breaker overcurrent
These software platforms typically contain libraries that protection, the elevator fuse overcurrent protection, and the
include most of the common overcurrent protective devices elevator motor startup curves.
and their available adjustments. The study must ensure that the two fuses will trip in a
After the aforementioned critical information is typed fault condition in any one of the separate elevator feeders
into the software database, the function of protective de- and will not trip the 200-A main breaker. A fault in one of
vices can be graphically presented. The resulting graphic the elevator feeders that takes out the main breaker would
representation is called a time-current curve (TCC). When essentially take out both elevators.
more than one electrical device is overlaid on one graph,
the relationship of the characteristics among the devices is GROUND FAULT STUDIES FOR
presented. Any potential issue, such as overlapping of curves HEALTH CARE FACILITIES
or long time intervals between devices, are illustrated. Fault- NEC Article 517.17 requires that if ground fault protection
current conditions can be illustrated by indicating on the is provided for service disconnecting means, an additional
current scale the maximum and minimum value of short- step of ground-fault protection shall be provided in the next
circuit currents (3-phase and line-to-ground) that can occur level of feeder disconnecting means downstream toward
at various points in the circuit (see Figure 1). the load (see Figure 3). There is clear separation between
the ground-fault curves.
ELEVATOR PROTECTION NEC Article 230.95 indicates that all 480 V, 3-phase
COORDINATION PER THE NEC services rated 1,000 A and higher must be installed with
Selective coordination is required when more than one a ground fault relay. The setting of the ground fault relay
elevator is supplied by a common feeder, according to NEC cannot exceed 1,200 A regardless of the size of the over-
current-protection device. In addition, the time
delay cannot exceed 1 sec (60 cycles) for ground
Current in Amperes
Utility 0001
1,000
EV breaker
CHKD
CHKD
Trip 250 A
Settings phase EV breaker
LTPU (A-H) F (200) (200 A) Type CHKD
LTD (2-24s) 15 Sensor/trip 250 A
100 STPU (2-12 x Ir) 12 (2,400 A) Setting LTPU F (200) (200 A)
STD flat (instantaneous-300 msec)
300 msec (P, Q, R)
Instantaneous of fixed (4,400 A)
1 70 AF
FRS, 600 V, Class K5
FRS
Trip 70 A Figure 2: The TCC graph (left) and one-line diagram
(above) show an example of a coordination study
Rated horsepower 30 hp illustrating feeder breaker and elevator fuse over-
0.10 current protection as well as elevator motor startup
Rated horsepower 20 hp curves. A 200-A breaker in the main distribution
gear feeds an elevator control panel with 100-A
and 70-A fuses. The study must ensure that the two
fuses will trip in a fault condition in any one of the
0.01 separate elevator feeders and will not trip the 200-A
0.5 1 10 100 1K 10K main breaker. A fault in one of the elevator feeders
Reference voltage: 208 V Current in Amperes x 10 that takes out the main breaker would essentially
take out both elevators.
www.csemag.com/purepower
Protective Device Coordination 11 ❮❮
Time in seconds
Plug 400 A
will not permit nuisance tripping, but will protect Settings phase
GFPU (1x Ig) 0.61 (244 A)
the electrical equipment from excessive damage GFD (instantaneous-500
msec) 300 msec (I t off)
2
during an event.
Sometimes, perfect coordination between a
set of devices cannot be obtained. Certain set- 1
tings may be required on a breaker that could
affect the settings of many breakers. In some
cases, there may be many levels of breakers that
may cause overlap of the breaker curves within
0.10
the tolerance of the curves. It is at these times
that experience will allow engineers to make
judgment calls regarding certain compromises in
coordination between devices. The engineering
behind providing protective coordination studies 0.01
is not a perfect science. It is important to ensure 0.5 1 10 100 1K 10K
that protection is not compromised, even if per-
Reference voltage: 208 V Current in Amperes x 10
fect coordination is not achieved.
On many occasions, projects are completed
without protective-device studies. In such cases, Figure 3: This TCC graph illustrates the ground-fault setting for a 2,500-A main breaker
the breaker manufacturer will ship the break- and the ground-fault setting for a 400-A sub-breaker. There is clear separation between
ers with all settings set to the most sensitive. the ground fault curves. A ground fault on the 400-A feeder would trip the 400-A
This will ensure the most protection, but will breaker before tripping the 2,500-A main breaker. If the 2,500-A main breaker trips, the
increase false trips and is typically not good for entire system will go down—not just the 400-A feeder.
the reliability and uptime of the systems. As soon
as the owner complains of a false trip, the facility person- NEC Article 100, Definitions: Coordination (selective):
nel will probably set all of the dials to least sensitive. This Localization of an overcurrent condition to restrict outages
will reduce false trips, but may not adequately protect the to the circuit or equipment affected, accomplished by the
electrical system and could also reduce selective coordina- choice of overcurrent protective devices and their ratings
tion of the system. and settings.
A coordination study is typically required to ensure that
the most reliable electrical system has been installed. In NEC Article 700.27, Coordination: Emergency system
addition, there are instances where the NEC requires that a overcurrent devices shall be selectively coordinated with
study be performed. In either case, the cost of a coordination all supply-side overcurrent protective devices.
study is pretty cheap insurance for most installations that
would be adversely affected by an extensive power outage. NEC Article 701.18, Coordination: Legally required
standby system overcurrent devices shall be selectively coor-
CODE-RELATED ISSUES dinated with all supply-side overcurrent protective devices.
To understand code-related issues involved in selective
coordination studies, it is important to quote the new defi- NEC Article 517.26, Application of Other Articles: The
nition of selective coordination and the new codes in NEC essential electrical system shall meet the requirements of
Articles 100, 517, 700, and 701. Article 700, except as amended by Article 517.
www.csemag.com/purepower
❯❯
PURE POWER // WINTER 2015
12 Protective Device Coordination
For clarity, it is important to include the NEC definition ARC FLASH MITIGATION VIA NONCOORDINATION
(Article 100) of overcurrent and the fine print notes (FPNs) The NEC also includes information to help minimize the
defining emergency systems and legally required standby risk of an arc flash incident.
loads in Articles 700 and 701. NEC-2014 Article 240.87, Arc Energy Reduction: Where
Overcurrent: Any current in excess of the rated cur- the highest continuous-current trip setting for which the
rent of the equipment or ampacity of the conductor. It may actual overcurrent device installed in a circuit breaker
result from overload, short circuit, or ground fault. is rated, or can be adjusted, is 1,200 A or higher, Article
It is significant to note that a “short circuit” is listed as 240.87(A) and (B) shall apply:
one of the items that can cause an overcurrent condition.
The typical molded-case circuit breaker combination with (A) Documentation shall be available to those autho-
the upstream breaker that is somewhat larger than the rized to design, install, operate, or inspect the installa-
downstream breaker does not have a problem coordinating tion as to the location of the circuit breaker.
in the overload area of the TCC, but a high level of current
in the short-circuit area of the TCC can present significant (B) As a method to reduce clearing time, one of the
problems to selective coordination. following or approved equivalent means shall be
NEC Article 700.1, FPN No. 3: Emergency systems are provided:
generally installed in places of assembly where artificial (1) Zone-selective interlocking
illumination is required for safe exit and panic control. (2) Differential relaying
Emergency systems may also provide power for such (3) Energy-reducing maintenance switching with
functions as ventilation, fi re detection and alarm systems, local status indicator
elevators, fi re pumps, public safety communication, and (4) Energy-reducing active arc flash mitigation
industrial processes. system
NEC Article 701.2, FPN: Legally required standby (5) An approved equivalent means.
systems are typically installed to serve loads—such as
heating and refrigeration systems, communication sys- A method for complying with this code section is to
tems, sewage disposal, lighting systems, and industrial provide breakers that have a reduced arc flash mode (RAM)
processes—that, when stopped during any interruption of setting. Most major switchgear manufacturers are providing
normal electrical supply, could create a hazard or hamper RAM in their breakers to enable compliance with this new
rescue or fi refighter operations. code section. While each manufacturer may have a different
Elevators are noted in Article 700.1 as an emergency twist on the solution, the basic principal is the same: provide
system load. Some jurisdictions also consider elevators to a temporary setting where the breaker will trip faster in a
be a legally required standby load. In either case, the NEC fault condition. While this RAM system is effective in reduc-
has required elevators in certain situations to be selective- ing the arc flash hazard, it may take all or some of the break-
ly coordinated for some time. This selective coordination ers downstream from the RAM breaker out of coordination
has been required by NEC Article 620.62. and increase the risk of losing critical loads.
NEC Article 620.62, Selective Coordination: Where It is always recommended that all systems be powered
more than one driving-machine disconnecting means is down prior to personnel working on electrical equipment.
supplied by a single feeder, the overcurrent protective However, in mission critical facilities qualified profession-
device in each disconnecting means shall be selectively als may need to access and work near energized equip-
coordinated with any other supply-side overcurrent pro- ment. Testing, troubleshooting, and diagnostics may
tective devices. require that power remain on to complete the task when
NEC Article 240.6 (C) Restricted-Access Adjustable removing power is infeasible.
Trip Circuit Breakers: A circuit breaker that has restricted
access to the adjusting means shall be permitted to have
an ampere rating that is equal to the adjusted current set- ABOUT THE AUTHORS
ting (long-time pickup setting). Restricted access shall be Keith Lane is president/CEO of Lane Coburn & Associates.
defi ned as located behind one of the following: He is a member of the Consulting-Specifying Engineer
editorial advisory board and is a 2008 40 Under 40 award
Removable and sealable covers over the winner.
adjusting means Scott Coburn is principal of Lane Coburn & Associates. He
oversees design constructability to ensure the highest level of
Bolted equipment-enclosure doors construction documents. He is the co-founder of the
Neher-McGrath Institute and Neher-McGrath.com.
Locked doors accessible only to qualified personnel.
www.csemag.com/purepower
ONE NAME
YOU CAN
COUNT ON
PLANNED An ice storm can cripple an
EVENT
economy and take weeks for
services to return to normal. In
your business, that’s simply not
an option. Your Cat® dealer has
everything you need including:
• Generators
• Heaters
• Chillers
UNPLANNED
• Compressors
• Load Banks
EVENT
As part of the Caterpillar
dealer network, we also have
everything you need for your
planned events, capable of
supporting even the most critical
operating systems.
SINCE 1927 SINCE 1948 SINCE 1923 SINCE 1960 SINCE 1916
Alban CAT Cleveland Brothers H.O. Penn Machinery Milton CAT Ransome CAT
www.albancat.com www.clevelandbrothers.com www.hopenn.com www.miltoncat.com www.ransome.com
Baltimore, MD Murrysville, PA Poughkeepsie, NY Milford, MA Bensalem, PA
800-492-6994 800-538-1020 844-CAT-1923 866-385-8538 877-RANSOME
input #402 at www.csemag.com/information
© 2015 Caterpillar All rights reserved. CAT, CATERPILLAR, their respective logos, “Caterpillar Yellow,” the “Power Edge” trade dress as well as corporate and product identity used herein, are trademarks of Caterpillar and may not be used
without permission. www.cat.com www.caterpillar.com
❯❯
PURE POWER // WINTER 2015
14 NEC 2014 Changes
Reviewing, analyzing
NEC 2014 changes
Changes in and additions to the NFPA 70: National Electrical Code (NEC) have a significant
impact on commercial and industrial facilities.
E
very 3 yr, NFPA 70-2014: National Electrical Code in wind turbine and photovoltaic (PV) systems. The voltage
(NEC) is updated and released. However, not all increase affects numerous articles including 240, 250, 300,
states immediately adopt the new code changes. The 430, 490, 690, 692, and 694.
adoption by many states doesn’t typically occur until While many of these changes are purely editorial in na-
the year after the latest version is released. ture, the change in the transformer protection table, Table
According to NFPA, there were 3,745 proposals submitted 450.3(A), compels system designers to use smaller overcur-
recommending changes to the 2014 rent-protective devices (OCPDs) for protecting transformer
edition of the NEC. In addition, primaries when within the 600- to 1,000-V range. Although
LEARNING OBJECTIVES
there were 1,625 comments con- there are no standard operating voltages within this range,
Outline NFPA 70: National
Electrical Code (NEC). cerning the NEC Code-Making Pan- this predominantly affects renewable energy sources,
els’ responses to these proposals. which can connect at voltages not commonly seen in stan-
Analyze the updates and
changes to the most recent dard distribution. It should also be noted that this reduc-
edition of the NEC. OVERALL IMPACT tion in OCPD sizing facilitates in the mitigation of arc flash
Compare and contrast each of OF CODE CHANGES hazards, which have been in the spotlight for safety aware-
the new items in NFPA 70-2014. There are several new articles as ness and recent code changes. Transformers are typically a
well as some noteworthy changes location of elevated incident-energy levels, and faster OCPD
included in the 2014 tripping can reduce incident-energy levels.
edition of the NEC.
Some changes have IMPACT ON COMMERCIAL
had a significant AND INDUSTRIAL FACILITIES
impact on the elec- Some of the code additions and modifications have substantial
trical trade. An ex- impact on commercial and industrial facilities. The following
ample is a codewide list describes, in sequence, seven changes in NEC 2014 that
change to raise the have a high impact on commercial and industrial systems:
maximum voltage
level from 600 V to 1. Dedicated equipment space
more than 1,000 V,
primarily driven by 2. Ground-fault protection (GFP)
the higher voltages
3. Conductor sizing
Figure 1: This diagram
illustrates outdoor dedi- 4. Arc energy reduction
cated-equipment space
extending 6 ft above the 5. Surge protection
width and depth of the
equipment. All graphics 6. Selective coordination requirements and health care
courtesy: Triad Consulting
Engineers Inc. 7. Solar system rapid-shutdown systems.
www.csemag.com/purepower
NEC 2014 Changes 15 ❮❮
Worst case
Arc fault
(Maintenance mode “off”)
Time in seconds
Main circuit breaker with 10
maintenance mode off
Trip 4,000 A
Plug 4,000 A
Settings phase
LTPU 1 (4,000 A)
LTD 30
STPU 6 (24,000 A)
STD I2t 0.25 (12t out) 1
0.01
1. DEDICATED EQUIPMENT SPACE
Current scale x 100 Reference voltage: 480
NEC 2014 change—Article 110: Requirements for Electri-
cal Installations; Section 110.26(E) Dedicated Equipment
Space, (2). Outdoor: Figure 3: The time-current curve shows a circuit breaker with mainte-
Outdoor installations shall comply with 110.26(E)(2)(a) nance-mode settings.
and (b).
Subsection (b), Dedicated Equipment Space: The space process where an unorderly shutdown will introduce ad-
equal to the width and depth of the equipment and extend- ditional or increased hazards.
ing from grade to a height of 1.8 m (6 ft) above the equip- Exception No. 2—The provisions of this section shall
ment shall be dedicated to the electrical installation. No not apply if ground-fault protection of equipment is pro-
piping or other equipment foreign to the electrical installa- vided on the supply side of the branch circuit and on
tion shall be located in this zone. the load side of any transformer supplying the branch
Analysis of change—As with indoor equipment, dedi- circuit.
cated space is now required for installation of outdoor Prior to this addition to the NEC, GFP was only re-
equipment (see Figure 1). This is clarified in the code quired on each service disconnect rated 1,000 A or more.
to pertain to all switchboards, switchgear, panelboards, Now, GFP is required on each branch circuit with dis-
and motor control centers. This removes the gray area of connects rated 1,000 A or more (see Figure 2). As before,
interpretation and clarifies the issue for design engineers, this applies to solidly grounded wye electrical systems of
contractors, and the authorities having jurisdiction. more than 150 V to ground. Therefore, 208 Y/120 V sys-
tems are still exempt from GFP. Because the maximum
2. GROUND-FAULT PROTECTION setting of the ground-fault protective device on the main
NEC 2014 change—Article 210: Branch Circuits; Section is still limited to 1,200-A pickup with a 1-sec delay (for
210.13 Ground-Fault Protection of Equipment: ground faults exceeding 3,000 A only) per Article 230.95,
Each branch-circuit disconnect rated at 1,000 A or more this will require coordination between the settings of the
and installed on solidly grounded wye electrical systems of GFP on the feeders or branch circuits with the ground-
more than 150 V to ground, but not exceeding 600 V phase- fault protective device on the main disconnect.
to-phase, shall be provided with ground-fault protection of Analysis of change—The intent of this code addition is
equipment in accordance with the provisions of 230.95. to mitigate fires commonly associated with ground faults.
Informational note—For buildings that contain health This is because they have a tendency to burn until the
care occupancies, see the requirements of Article 517.17. event propagates into a more severe incident, eventually
Exception No. 1—The provisions of this section shall not causing an open circuit—whether by tripping of upstream
apply to a disconnecting means for a continuous industrial protective devices or equipment burn-down.
www.csemag.com/purepower
❯❯
PURE POWER // WINTER 2015
16 NEC 2014 Changes
Time in seconds
ATS feeder circuit breaker combination of continuous
1,600 AF/1,600 AS
10 Plug 1,600 A and noncontinuous loads, the
Generator Phase
LTPU 1 (1,600 A) minimum feeder conductor
Generator circuit LTD 45
breaker, phase
Devices are selective STPU 4.61 (7,376 A) size shall have an allowable
above 0.1 sec per STD-I2t 0.11 (I t out)
2
NEC Article 517 Instantaneous 15 (24,000 A) ampacity not less than the
ATS feeder circuit (health care facilities) 1 Ground
breaker, phase E N PU 0.36 (576 A) noncontinuous load plus
TD-I2t 0.2 (I t out)
2
Instantaneous 15 (18,000 A)
be served after the applica-
Ground
GFPU 0.26 (312 A)
tion of any adjustment or
Current scale x 100 Reference voltage: 480
TD-I2t 0.1 (I t out)
2
correction factors.
Analysis of change—The
Figure 4: The time-current curve and corresponding single-line diagram show the coordination and selectivity intent of this code change is to
requirements in NEC Articles 517, 700, 701, and 708. simplify the sizing of conduc-
tors. The changes clarify that the calculation procedure
3. CONDUCTOR SIZING for continuous versus noncontinuous loads are sepa-
NEC 2014 change—Article 210: Branch Circuits; Part II, rate from calculations in which ampacity or correction
Branch Circuit Ratings; Section 210.19 Conductors—Mini- factors (e.g., number of current-carrying conductors
mum Ampacity and Size: in a raceway and/or ambient temperature above 86 F)
Subsection (A), Branch Circuits Not More Than 600 are applied. After each method is calculated, the larger
V, (1) General: Branch-circuit conductors must have an conductor size shall be used. A similar change can be
ampacity of not less than the maximum load to be served. found in Section 230.42 for the minimum size and rating
Conductors shall be sized to carry not less than the larger of service-entrance conductors.
of Article 210.19(A)(1)(a) or (b). Example—The following is an example for a 200-A
(a) Where the branch circuit supplies continuous loads continuous load with three current-carrying conduc-
or any combination of continuous and noncontinuous tors in a single conduit, located in a boiler room with a
loads, the minimum branch-circuit conductor size shall temperature of 100 F:
have allowable ampacity not less than the noncontinuous
load plus 125% of the continuous load. Based on the Article 215.2(A)(1)(a) requirement:
(b) Minimum branch-circuit conductor size shall have
an allowable ampacity not less than the maximum load to 200 A x 1.25 = 250 A
be served after the application of any adjustment or correc-
tion factors. Conductor size = 250-kcmil THWN copper, rated 255
Exception—If the assembly, including the overcurrent A at 75 C, based on Table 310.15(B)(16).
devices protecting the branch circuits, is listed for operation
at 100% of its rating, the allowable ampacity of the branch- Based on the Article 215.2(A)(1)(b) requirement:
circuit conductors shall be permitted to be not less than the
sum of the continuous load plus the noncontinuous load. 250-kcmil THWN copper, rated 255 A at 75 C
NEC 2014 change—ARTICLE 215: Feeders; Section 215.2
Minimum Rating and Size, (A) Feeders Not More than 600 100 F correction factor = 0.88 for 75 C insulation,
V, (1) General: based on Table 310.15(B)(2)(a)
www.csemag.com/purepower
NEC 2014 Changes 17 ❮❮
www.csemag.com/purepower
❯❯
PURE POWER // WINTER 2015
18 NEC 2014 Changes
www.csemag.com/purepower
NEC 2014 Changes 19 ❮❮
www.csemag.com/purepower
❯❯
PURE POWER // WINTER 2015
20 NEC 2014 Changes
5. Equipment that perform the rapid shutdown shall be Alternate configurations involve implementing smart com-
listed and identified. biner boxes or dc-to-dc converters with switching capabil-
ity. Schematics of different configurations are illustrated
Analysis of change—Prior to the NEC 2014 update, in Figure 5.
there were no requirements for a rapid power shutdown of With the plethora of changes witnessed in the 2014
rooftop PV systems. The intent of this change is to increase revision to the NEC, the focus here was on seven key
the safety of first responders by creating a protected zone changes affecting the electrical industry for commercial
in which shock hazards have been mitigated. and industrial systems. In the next revisions expect to see
There has been significant controversy and debate about continued code updates in the next revisions for renew-
the specific requirements in this new section, particularly able energy systems as well as selective coordination and
because the methodology is addressed in the product arc flash hazard protection.
standards, not in the NFPA documents. A key concern is
often defining the location of the device to de-energize the ABOUT THE AUTHORS
controlled conductors of the PV system, and if more than Robert C. Corson is a principal engineer at Triad Consulting
one is required. Engineers. He has experience in the design and analysis of
With the inverter anti-islanding requirements of UL power systems for commercial, industrial, renewable energy,
1741-2010: Standard for Inverters, Converters, Control- and mission critical facilities including 120/208 V through 230
lers, and Interconnection System Equipment for Use with kV, PV plants up to 20 MW, emergency microgrid generation,
Distributed Energy Resources, opening the main service and UPS systems for critical applications.
disconnect will result in the de-energizing of the PV sys- Philip M. Grenci is the chief electrical engineer at Triad
tem. As long as any potential storage devices (e.g., capaci- Consulting Engineers. He has 30 yr of experience in the in-
tors) within the inverters discharge within this time frame, dustry and supervises the electrical engineers and designers
and the inverters are located within 10 ft of the PV array, at the company’s New Jersey office.
the main service disconnect could satisfy this requirement.
Sign up today! Go to
www.csemag.com/newsletters
Engineering is personal.
So is the way you use information. CFE Media
delivers a world of knowledge to you. Personally.
provides the
latest knowledge
on commercial
and institutional facility construction
and management.
Visit www.csemag.com
delivers a wide
array of strategies
and solutions to
help control system designers create a
more efficient process.
Visit www.controleng.com
delivers plant-floor
knowledge and
expertise to help
manufacturers operate smarter, safer
and more efficiently.
Visit www.plantengineering.com
❯❯
PURE POWER // WINTER 2015
22 Data Center Cooling
By Bill Kosik, PE, CEM, LEED AP, BEMP, Data Center Facilities Consulting,
Hewlett-Packard Co., Chicago
T
he Slinky and IT systems. Can there really be computer. Some resorted to removing windows, leaving ex-
common attributes between a spiral metal toy terior doors open, and other rudimentary approaches that
invented in the 1940s and a revolutionary technol- used ambient air for keeping things cool. But others had
ogy that impacts most people on this planet? If already realized that water could be a more practical and
you have ever played with a Slinky or seen one in action operationally viable solution to keep the computers at an
(especially on stairs), you’ll see that as it travels on the acceptable temperature. Different generations of computers
stairs, the mass of the spring moves from the trailing end demonstrate this point:
of the Slinky to the leading end. This shifting of the mass
causes the trailing end to eventually overtake the lead- UNIVAC, released in 1951, demanded 120 kW and
ing end, jumping over it and landing on the next stair. If required 52 tons of chilled water cooling. It had an
the staircase went on forever, so electrical density of 100 W/sq ft.
would the Slinky.
LEARNING OBJECTIVES
The meaning behind this Control Data Corp.’s CDC 7600 debuted in the early
Apply the various ways to cool a
data center. visual metaphor is that as soon 1970s and consumed 150 kW. (The primary inventor
as things seem to settle out in of the CDC 7600 went on to found Cray Computers).
Decide which data center cool-
ing options work best in each the information technology (IT) It was benchmarked at 10 megaflop/sec and used an
part of the world. sector, some other disruptive internal refrigerant cooling system that rejected heat
Model different ways to design force changes the rules and leap- to an external water loop. (FLOPS, is an acronym for
a data center by making use of frogs right over the status quo floating-point operations per second) It had an electri-
various cooling technologies. into completely new territory. A cal density of 200 W/sq ft.
similar effect can be seen when
thinking about cooling systems for data centers, which In 1990 IBM introduced the ES/9000 mainframe
have very tight relationships to the IT systems. By the computer, which consumed 166 kW. It is interesting
time the HVAC equipment manufacturers have it figured to note that 80% of the machine’s heat is dissipated
out, the IT industry throws them a curveball and creates a using chilled-water cooling. The remaining heat is re-
new platform that challenges the status quo in maintain- jected to the air. This system had an electrical density
ing the appropriate temperature and moisture levels in- of 210 W/sq ft.
side the data center. But this is not new—the evolution of
powerful computers has continually pushed the limits of The first part of the 21st century ushered in some
power-delivery and cooling systems required to maintain pretty amazing advances in computing. In 2014,
their reliability goals. Hewlett-Packard announced a new high-performance
computer—also water-cooled—that is capable of
A BRIEF HISTORY OF (COMPUTER) TIME 1.2 quadrillion calculations/sec peak performance
Early in the development of large, mainframe computers (“petascale”-computing capability, which is defined as
used in the defense and business sectors, one of the major 1,015 FLOPS). Computing power of this scale designed
problems that the scientists and engineers faced (and still for energy and space efficiency yield a power density
do today) was the heat generated by the innards of the of more than 1,000 W/sq ft.
www.csemag.com/purepower
Data Center Cooling 23 ❮❮
www.csemag.com/purepower
❯❯
PURE POWER // WINTER 2015
24 Data Center Cooling
Water-based cooling in data centers falls somewhere be- been in use for several years and tends to work best for
tween the basic (although tried-and-true) air-cooled direct high-density applications, especially when the equipment
expansion (DX) systems and complex methods with high is located in an existing low-density data center.
degrees of sophistication. And because water-based data Direct-coupled: One of the primary challenges when
center cooling systems have been in use in some form or cooling a data center is the ability to control how effectively
another for more than 60 yr, there is a lot of analytical and the cooling load can be neutralized. A data center that uses
historical data on how these systems perform and where cold air supplied to the room via raised floor or ducting
their strengths and weaknesses lie. The most common comes with inherent difficulties, such as uncontrolled by-
water-based approaches today can be aggregated anecdot- pass air, imbalanced air delivery, and re-entrainment of hot
ally into three primary classifications: near-coupled, close- exhaust air into the intakes of the computer equipment—all
coupled, and direct-cooled. of which will usually present difficul-
Near-coupled: Near-coupled systems ties in keeping the IT equipment at
include solutions including rear-door
heat exchangers (RDHX), where the
One of the primary allowable temperatures.
Most of these complications stem
cooling water is pumped to a large
coil built into the rear door of the IT
challenges when cooling a from proximity and physical contain-
ment; if the hot air escapes into the
cabinet (see Figure 1). Depending on
the design, airflow-assist fans can data center is the ability room before the cold air can mix with
it and reduce the temperature, the
also come integrated into the RDHX. hot air now becomes a fugitive and
This heat-removal design reduces the to control how effectively the cold air becomes an inefficiency
temperature of the exhaust air coming in the system. In all air-cooled data
from the IT equipment (typically cabi- the cooling load can be centers, a highly effective method for
net-mounted servers). The temperature reducing these difficulties is to use a
reduction varies based on parameters,
such as water temperature, water flow,
neutralized. partition system as part of an overall
containment system that physically
airflow, etc. However, the goal is to re- separates the hot air from the cold
duce the temperature to as close to ambient as possible. For air, allowing for a fairly precise cooling solution. On a
example, with an inlet air temperature of 75 F and a water macro scale, it is possible to carefully control how much air
flow of 1.3 gpm/ton, using 66 F chilled water will cool 85% comes into and out of the containment system and predict
of the heat in the server cabinet to room temperature. With general supply and return temperatures. What cannot be
59 F chilled water, 100% of the heat will be cooled to room done in this type of system is to ensure that the required
temperature. It is assumed that the cooling water for the airflow and temperature across the internal components in
RDHX is a secondary or tertiary loop with a 2 F increase in the computer system are being met. As individual server
water temperature from the chilled water temperature. This fans will vary their speed to control internal temperatures
solution has been in use for several years and tends to work (based on workload), it is possible to starve air, especially
best for high-density applications with uniform IT-cabinet- if the workload is small, causing the internal fans to go to
row distribution. Because the RDHX units are typically minimum speed.
mounted on the back of the IT cabinet, it does not impact
the floor space. NEXT GENERATION
Close-coupled: Close-coupled water-based cooling solu- In computing applications where the electrical density is
tions include water-cooled IT cabinets that have coils and at a point where industry-standard cooling solutions can
circulation fans built into the cabinet (see Figure 2). This no longer properly cool the equipment, using alternative
system allows for totally enclosed and cooled IT equip- methods, such as direct water cooling, becomes necessary.
ment where the heat load is completely contained, with There is no hard-and-fast rule on the electrical-density
only a small amount (~5%) of the heat being released cutoff when air cooling starts to become ineffective. A
into the data center. This system has similar limitations to data center with a homogeneous IT equipment layout and
the RDHX, but generally is more feasible to neutralize the optimized air distribution will be able to use air cooling to
entire cooling load. a much higher threshold than a heterogeneous data center
The cabinet is larger than a standard IT cabinet. How- with legacy IT equipment and less-than-optimal air distri-
ever, because the equipment is enclosed, it is possible to bution. Obviously, this is an IT-equipment-driven outcome,
mount higher densities of IT equipment in the cabinet falling outside the purview of the facility’s HVAC engineer-
due to the ability of the cabinet to maintain a much more ing team. When the opportunity arises to develop a design
uniform temperature, essentially eliminating any chance for a direct-water-cooled computing system, it is essential
of a high-temperature cutout event. This solution also has to understand some important concepts and strategies.
www.csemag.com/purepower
Data Center Cooling 25 ❮❮
Annual hours
825
55 884 800 733 706
60 925
65 733 600
525
70 825 459
400 383
75 706
80 383 250
218 196
200 160
85 196 111
90 111 13 0 0 0
0
95 13 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100 105 110
100 0
Water temperature bin (deg F)
105 0
110 0
Figure 3: When analyzing the effectiveness of using warm-water cooling, it is essential to study the climate where the data center is to be located. In these
two examples, bin data of the number of hours at certain condenser water temperatures has been developed. The condenser water temperature takes into
consideration the hourly wet-bulb temperatures as well as the cooling tower approach, typically algorithmically derived. The weather data for Helsinki
(above) indicates condenser water temperatures that are lower than that of Abu Dhabi (below). When using warm-water heat rejection, generating hour-by-
hour data allows for a detailed energy-consumption analysis.
45 0 898
50 0 835
800
55 0 639
60 9 600
562
65 70
70 330 400 330
75 639
80 898 200
85 835 70
0 0 0 0 0 0 0 0 9
90 1,164 0
20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100 105 110
95 1,176
100 1,200 Water temperature bin (deg F)
105 981
110 562
Direct-cooling strategy: There are several approaches cessor, memory, and other internal components. The other
used by different manufacturers, such as a direct supply side of the thermal plate is connected to the cooling water.
of water to a heat sink attached to the computer’s inter- Water temperature: The term “cooling water” is a
nal components, specifically the processor and memory misnomer in that the supply water can be as warm as 85 F
modules. Another approach is to use a dry-type connection to 105 F. The terms “warm-water cooling” or even “hot-
where the water stays outside of the server. The heat-trans- water cooling” have emerged, even though each appears to
fer process is achieved by using thermal plates connected be a contradiction in terms. This is really the centerpiece
to heat pipes that are also connected to plates on the pro- of water-cooled-computer energy efficiency. Because the
www.csemag.com/purepower
❯❯
PURE POWER // WINTER 2015
26 Data Center Cooling
E a al
Exascale
1E+15
5 1,200
1,
Gigascale
g Terascale
a a Petascale
P s e
FLOP/sec
1E+14
4 1,000
1,
1E+13
3 800
80
1E+12
2 600
60
1E+11
1 400
40
1E+10
0 200
20
1E+09
9 0
90
1990
92
1992
94
1994
96
1996
98
1998
00
2000
02
2002
04
2004
06
2006
08
2008
0
2010
2
2012
4
2014
6
2016
8
2018
0
2020
Figure 4: The capability of supercomputers and other HPC installations has increased significantly since 1990. The FLOP/s and MFLOP/sec/W have increased
steadily until the realm of the petascale computer. The computers are not only becoming more powerful, but the use of power has become more efficient.
internal components of a computer generally can happily use of the CDU must be included in any energy-use metric,
operate far above 150 F, using higher-temperature water such as power usage effectiveness.
works great. Here’s the clincher: Using these water temper- Air-cooled equipment requirements: While the water-
atures to cool computers can eliminate the use of compres- cooled computer may be the largest cooling load in the data
sorized cooling equipment. In most climates—even very center, there will most likely be IT and electrical equip-
warm ones—the energy use associated with traditional ment that still rely on air cooling. In addition, many of the
data center cooling solutions is significant because the water-cooled computers still have a small percentage of the
energy use associated with compressorized cooling (chill- cooling load (radiated heat) that must be air cooled. This
ers, condensing units, etc.) will necessitates the use of computer-
be 50% or more of the total HVAC room air-handling units (AHUs)
energy use. While the water-cooled computer or central AHUs. These units
Geography: Although most can be stand-alone, split-system
climates allow for production of may be the largest cooling load DX units, or—if the facility is
warm water used for cooling com- large enough—a small, separate
puters without compressors, the
type and size of the heat-rejection
in the data center, there will most chilled-water system can be
installed. Regardless of the final
system will depend on the distribu-
tion of the dry-bulb and wet-bulb
likely be IT and electrical equipment design, the energy-use estimates
must account for these systems.
temperatures. Analyzing climatic
data early on in the project will
that still rely on air cooling. Cooling-system redundancy:
Driven by the IT equipment
provide guidance on the location, redundancy requirements, the
physical size, and capacity of the cooling towers, evaporative cooling system must have built-in mechanisms for continu-
coolers, dry coolers, or whatever the heat-rejection schema is ity of operation if there is a loss of power or a major equip-
(see Figure 3). ment failure. Due to the diameter and length of the piping
Water distribution: Typically, cooling loops in the com- in a large water-cooled computer facility, the cooling-water
puters will be isolated from the base building water loop via piping can act as a storage tank that continues to circulate
a heat exchanger. Knowing that there would typically be a water (as one part of the overall reliability strategy) until
primary/secondary arrangement on the house-cooling water the outage is corrected.
distribution system, the computer water loops become tertia- Fan energy: There is another component of data center
ry. Contingent on the computer manufacturer, a cooling dis- energy use that is influenced by the use of water-cooled
tribution unit (CDU) will provide the connection to the house computers: fan energy for the facility air-handling systems.
secondary loop and the water pumping through the piping Depending on the type of water-cooled computer, a certain
within the computer. In addition to the physical requirements percentage of the heat will be dissipated into the data cen-
of the CDU, like piping and electrical connections, the energy ter ambient environment, and not picked up by the cooling
www.csemag.com/purepower
Learn about the latest engineering
trends and technologies while earning
CEU credits.
www.csemag.com/ondemandwebcasts
❯❯
PURE POWER // WINTER 2015
28 Data Center Cooling
water. This amount of heat will range 52%. These numbers could be a signal
from approximately 5% to 30% of the More computing power that private industry has the ability to
total IT cabinet load. This cooling load fund HPC systems to a higher level.
What is all of this massive amount of
will be taken care of by the house Another important organization
computing power used for? Organizations
AHUs. This means that 70% to 95% of will be able to: in the HPC/supercomputing world is
the fan energy of a fully air-cooled data The Green500. The purpose of The
Design new products faster
center will be eliminated. And because Green500, similar to the Top500,
and more efficiently
the fan motor energy will be the first or is to provide a ranking of the most
second highest energy use, it is a very Simulate and anticipate energy-efficient supercomputers in
significant energy-reduction strategy. multidimensional phenomena such the world. Over the years, the result
as climate change
of placing an emphasis on computer
NEXT STEPS: MORE GREEN Analyze massive data flows in speed and throughput without much
It is interesting that many of the first real time consideration for electrical consump-
widely distributed water-cooled com- Identify threats to national security tion, cooling-system energy, or overall
puters, such as the IBM ES/9000, were Provide crisis management. environmental impact has been an
used in the business world. But over extraordinary increase in the long-term
These computing systems will enable
time, the focus has seemed to change. operating cost of the supercomputing
applications to capture, store, compute,
There seems to be a rekindled inter- installation. As The Green500 puts it,
analyze, and visualize very massive data
est in using water-cooled computers models as fast as possible. “ … [We] encourage supercomputing
for new high-performance computing stakeholders to ensure that supercom-
systems. To keep things in perspec- puters are only simulating climate
tive, the percentage of high performance computing (HPC) change and not creating climate change.”
systems classified as “industry” on the Top500 List is 22%; On July 29, 2015, President Obama signed an executive
the percentage of systems listed under the categories “gov- order, “Creating a National Strategic Computing Initiative.”
ernment/academic/research” is 76%. Clearly, the number This order establishes the National Strategic Computing Ini-
of HPC systems for nonindustry use dominates the list. tiative (NSCI). “The NSCI is a whole-of-government effort
However, looking at it from a different angle, if one adds designed to create a cohesive, multi-agency strategic vision
up all of the peak-power demands of the systems on the list and federal investment strategy, executed in collaboration
and analyzes the percentage breakdown, the gap shrinks with industry and academia, to maximize the benefits of
significantly. The percentage of total power of “industry” HPC for the United States.” One of the tenets of the order
HPC systems on the Top500 List is 46%; the systems listed is to continue to reinforce requirements that HPC systems
under the categories “government/academic/research” is must improve in performance and efficiency (see Figure 4).
Less than a decade ago, hitting petaFLOP/s performance
was the absolute high-water mark. Now computer engi-
neers, scientists, and researchers are coming close to reach-
AD INDEX
ing “exascale” computing. Scientists believe that exascale
Caterpillar - Northeast - 13 402 is on the order of processing power of the human brain at
www.NECatDealers.com/power the neural level. And, unbelievably, these same people are
CFE Media, Engineering Is Personal - 21 already looking at “zettascale” computers. However, the
630-571-4070 www.csemag.com onset of big data has given business/enterprise users the
CSE E-newsletters - 20 ability to capture, store, compute, analyze, and visualize
630-571-4070 www.csemag.com/newsletters very massive data models, similar to the HPC industry.
Perhaps it is just a matter of time before water-cooled com-
CSE On-demand Webcasts - 27
630-571-4070 www.csemag.com/ondemandwebcasts puters fully penetrate the business ecosystem and gain a
foothold based on computing- and data-intensive business
Kohler - C-4 404
models. So, when this happens, water-cooled computers
800-544-2444 www.KOHLERPOWER.COM
will have finally come full circle.
Russelectric Inc. - 7 401
800-225-5250 www.russelectric.com
ABOUT THE AUTHOR
Thomson Power Systems - C-3 403 Bill Kosik, is a distinguished technologist, Data Center
604-888-0110 WWW.THOMSONPS.COM
Facilities Consulting at Hewlett-Packard Co. He is a
United Wire & Cable - C-2 400 member of the Consulting-Specifying Engineer editorial
800-265-8697 www.unitedwc.com advisory board.
www.csemag.com/purepower
input #403 at www.csemag.com/information
IF YOU CAN DRAG A MOUSE,
YOU CAN SIZE A GENERATOR.