ENERGY STAR
UNDERSTANDING AND DESIGNING
ENERGY-EFFICIENCY PROGRAMS FOR
DATA CENTERS
November, 2012
The U.S. Environmental Protection Agency (EPA) is providing this guide to help inform energy efliciency program administrators about
opportunities to save energy in data centers, and to share emerging practices forprograrn design and imple mentation based on the
experiences of recent data center programs.
WHY DATA CENTERS?
Data centers consume up to 50 tirnesthe electricity ofstandarl office space.' In alio, between 1.7% and 22% ofttie total electricity
use in Tfie United States was consumed by data centers. United States data center electricity use nearty doubled between 2100 and
2(05, and increased by approximated 36% between 2005 and 2:10. Despte some recent efliciency gains, data centers remain a
significant and growing energy end use.' Industry analysts expect data center energy consumption to continue to grow at a rate of
more than 9% peryearthrough 2iK0(froma base of 3!0trillion end-use BTUs in 2003 to eootrillion end-use BTUs in 202!).3
Utilities and other energy-eficiency pro gram administrators can play a significant role in helping customers reduce data center
operating costs,while also reducing energy demand.
DELIVERING SOLUTIONS FOR DATA CENTERS
Information technology (IT is intrinsic to our econony, society, and culture. For most enterprises, IT provides crucial support for
financial operations, data storage and analysis, and all levels of management.
0 ata centers consume a significant amount of energy persquare foot, even when the physical space they occupy is small. In addition
to operating atver>> high energy intensities, data centers operate 24 hours perday, 385 days peryear.This high loadfactorpresentsan
important target forenergy-efliciency program;.
The U.S. Department of Energy(000 estimates ttiatfor ever>> t watt of energy used to direct^1 operate a computer,as much as two
additional watts are used by the data center to support tfiat computer. However, the OOE hasfound that energy-efficient data centers
are capable of reducing that consumption by up to £0%.'
While data center energy-efficiency programs are relative^1 new, early pioneers are realising signiicant savings. Dofte Energy and
several utilities in Wfoconsiti manage data center programs that achieved an average energy savings of 3S6 MWh per project during
their 20G3 to 2!12 pro gram cycles. During that time, they processed, paid, and claimed energy savings for 174 applications, ranging in
savingsfrorn under25! MWh to greater than 3,COD fuTWh.5
This guide will:
¦	Characterise the data centerrnarfcet;
¦	Highlight energy-efficiency prograrnopportunties in data centers;
¦	Provide an overview of data center program; tfiroughouttfie country;
¦	Oiscusstfie rrari;etstructure and resulting challenges, and suggestsolutions to overorne those challenges;
¦	Summarize appropriate programmodels and measures;
¦	Explain prograrnplanning strategies and evaluation, measurement, and verification (EM&V) best practices ttiat can mitigate program
implementation bamers; and
¦	Sugg est implementation strategies for data center program;.
1 E^r? on oortor	irfoimrtfen a&rrtftedtott* EfCFCV STJP	;rn ri E fX'z	d tr*  MO. h>y& 1. #11.
xSww: CM Gt vivdo. H J. Gv/ts. X fcriadi. R Faroso, S. Nyf J&t QstrowsW. McttiwoyGbtal	Maomb. UJo&rg EntfgyEflieitrt yintf* US. Err>n^
-So-jrcv: U$. &>p*rtr««tof Er*r$v. Er.«# led: ErtttyyEflkttrt C**i C*rt*r;. May21. #11. ArciaUo ntk»n (V.KC) rrf	oflk* revproyvrc Djko Entigyi ufkich 50iuc*5N«rth Cvolri^ Soldi Cvdii^ Ctio. srrd
ii Vtawart S^jrcv: V.tCC. p*r*>ral	ril #12

-------
UNDERSTANDING DATA CENTER CUSTOMERS
In most traditional data centers, IT equipment, which includes
servers, data storage equipment, and communications gear, is
housed in racks that are often placed on a raised floor system.
Computer room air conditioners (CRACs) deliver cool air through
the under-floor plenum via grate tiles in the floor.
Data centers differ from other IT infrastructure in that they
commonly have dedicated cooling systems and power delivery
and conditioning systems that require a high degree of reliability.
Power delivery equipment can include:
¦	Uninterruptible power supplies (UPSs)that provide backup
power using batteries or flywheels;
¦	Power delivery units and distribution transformers; and
¦	Backup generators.
Virtually all commercial and industrial (C&l) customers have
some level of IT infrastructure, and larger C&l customers
typically operate data centers. Table 1 provides information
about different data center types.
Data Center
Type
Description
Table 1. Data Center Classifications
Facilities	Total Servers
Square Footage (2009 U.S. estimation) (2009 U.S. estimation)
Utility Scale
Generally measured by the size of the facility's
total load (in MW), or the amount of power
available to the IT equipment. Usually larger than
10 MW, and commonly built with 40 MW load. This
category includes most retail and wholesale co-
location data centers.*
>100,000
7,006
3,604,678
515
Enterprise
Typically operated by large corporations and
institutions. Generally occupy spaces in the
lowtens of thousands of square feet (10,000
square feet can support approximately 1 MW of
IT equipment load, with almost another 1 MW
needed for cooling and power delivery systems).
>5,000
Localized
These facilities may serve only the local, specific
needs of a call center or office operation (for
example), with general, large-scale IT services
provided by a data center in another location.
500-5,000
73,987
3,977,187
54
Server Room**
These data centers often do not have dedicated
cooling or power delivery systems or climate
conditioning equipment.
200-499
1,170,399
3,057,834
3
Server Closet**
Smallest-scale data center.
<200
1,345,741
2,135,538
2
Source: Bailey, Michelle et al. IDC Special Study. Data Center of the Future. Filing Information: April 2006, IDC #06C4799.
* Retail and co-location data centers provide rental space to multiple tenants and are only responsible for managing their tenants' IT equipment.
Wholesale co-location providers develop new data centers, which then lease these facilities on a long-term basis to single tenants who take over the
responsibility of operating and maintaining the entire facility, including the cooling and power delivery systems. Seattle City Light \s starting to see
more interest from co-location providers.6
** Server rooms and server closets are not considered data centers in the ENERGY STAR Buildings Program.
Different data center types require different efficiency program
approaches:
¦	Operators of utility-scale data centers are typically very
sophisticated and may implement many energy-efficiency
measures on their own—particularly when building
new facilities or upgrading IT equipment. Consequently,
freeridership concerns are important to recognize and
manage, and incentives may need to be limited to highly
innovative measures or those measures with higher payback
periods. For example, customers that operate utility-scale data
centers, such as the technology, communications, financial
services, and government sectors, most likely do not need
incentives rewarding them for server virtualization upgrades.
¦	Enterprise data centers represent an excellent opportunity for
energy-efficiency retrofits, and energy-efficiency programs
have had an impact on this market. At this time, there are
few freeridership concerns. As a result, this segment is an
important target market for efficiency program managers.
Enterprise data centers are used by both commercial and
noncommercial entities across widely diverse business
sectors.
¦ Localized data centers, server rooms, and server closets
pose difficulties for utility programs because the market
is disaggregated and the savings per project are small.
However, they are considered an untapped market. The
Natural Resources Defense Council believes that program
implementers can increase energy efficiency opportunities for
small server rooms and closets with education and evaluation
tools, and extend education programs with on-site efficiency
evaluations. These tools and services could, subsequently,
support a select package of incentive programs to encourage
energy efficiency upgrades in this market. Program strategies
that minimize administrative costs need to be considered.7
6 Some market actors believe that this response from co-location providers was caused by pressure from their customers to become more efficient in order to avoid spending a lot of money and
resources to power and cool their data center.
'Source: Bramfitt, Mark and P. Delforge. Natural Resources Defense Council. Utility Energy Efficiency Program Design: Server Room Assessments and Retrofits. April 2012. Available online:
http://d ocs.nrdc.org/energy/files/ ene_12041101 a.pdf.

-------
OPPORTUNITIES FOR IMPROVING DATA
CENTER EFFICIENCY
Data center IT managers have varied motivations when
considering facility improvements. These include reducing
the ownership cost of IT operations, which have become so
high that they threaten to take capital from other investments
and projects;8 and addressing existing or upcoming capacity
shortfalls, such as a lack of cooling capacity for additional
equipment or a lack of power delivery and conditioning capacity.
IT managers know that adding more power or cooling capacity
to an existing building is challenging, time-consuming, and
costly. As a result, energy-efficiency measures that help reclaim
capacity and IT measures that improve utilization rates and
decrease power load can be very attractive. Fortunately, data
centers have abundant opportunities for improving efficiency,
depending on existing conditions. Potential improvements
include the following:
¦	Cooling system efficiency can be improved with premium-
efficiency equipment, such as chillers, pumps, fans, cooling
towers, and ultrasonic humidifiers, as well as through the use
of variable frequency drives, installation of air- or water-side
economizers, and improved air flow management.
¦	Power delivery and conditioning capacity can be improved
with the use of ENERGY STAR-certified UPSs and premium-
efficiency distribution transformers and direct current power
systems.
¦	IT equipment efficiency can be improved by upgrading to an
ENERGY STAR-qualified or other premium efficiency product,
virtualizing and consolidating servers and data storage
equipment,9 and by employing other data storage measures.
Table 2 summarizes measures that utility-funded energy-
efficiency programs commonly target. Not all efficiency
measures and technologies are applicable to every data center
type and, as is discussed in subsequent sections, program
design and delivery strategies vary by market and data center
type.
Table 2. Measures Commonly Targeted by Energy-Efficiency Programs1
Utility Programs
cc 0? '55 -2

_ ts ® CD ®
2 
-------
UNDERSTANDING AND ADDRESSING KEY
MARKET BARRIERS
Effective program design requires understanding market
barriers that prevent greater investment in energy efficiency
and developing strategies to overcome those barriers. Common
barriers in the data center market include:
¦	Lack of knowledge and risk aversion: IT managers demand
very high reliability for power and cooling systems. As a result,
they may be wary of projects that could affect that reliability,
and/or they may have misperceptions about the tradeoff
between energy efficiency and performance.
¦	Disincentive for trade allies: Tradealliessometimeshavea
disincentive to encourage efficient measures. For example,
server virtualization and consolidation may reduce future sales
of servers.
¦	Higher first cost and split incentives: In m any ca ses, efficient
equipment and related services cost more to purchase. This
barrier is often compounded when the IT manager purchasing
the equipment is not responsible for paying data center
operating costs (i.e., the electricity bill is paid by another
party).
In addition to these market barriers, program administrators
face challenges in ensuring that their investment in data center
efficiency generates energy savings beyond what would occur
naturally in this marketplace. The following sections provide
further detail about the key barriers listed above and strategies
for addressing them.
Educating the Marketplace
A smoothly running data center is essential to business
operations, and many IT managers are averse to taking risks
and cautious about implementing changes. A recent study found
that data center downtime (when a system is unavailable) costs
companies over $5,000 per minute.10 Risk aversion can manifest
in several ways, including:
¦	Building in more power redundancy than necessary (which
drives the utilization of power delivery and conditioning
equipment well below 50%);
¦	Overcooling the data center (e.g., setting the temperature too
low, setting the air volume too high); and
¦	Resisting technological solutions that can improve utilization
(e.g., virtualization, thin provisioning, data storage measures).
For all of these examples, energy-efficiency measures can be
employed without reducing reliability, and in some cases, can
improve reliability or help recapture capacity. For example:
¦	Removing non-critical equipment from power conditioning
systems minimizes potential points of failure and reduces the
energy overhead of these systems.
¦	Overcooling is a burden on cooling equipment and could lead
to premature failure.
¦	Some new software solutions, such as virtualization, feature
the ability to move IT workloads to other servers, providing
more operational flexibility to balance equipment load and
respond to system failures.
Implementing a comprehensive data center training program
can increase the information that IT managers have at their
disposal and help address hesitation to making energy-efficiency
improvements caused by risk aversion. These trainings typically
focus on the many energy-efficiency measures and technologies
available for these facilities, are often organized by end use (e.g.,
cooling, power delivery and conditioning, IT measures), and are
popular with both vendors and customers.
Internal Training
Educating the following internal parties gives them the tools to
properly identify energy-saving opportunities for data center
managers:
¦	Account representatives and call center specialists manage
all aspects of the commercial customer relationships, and
represent a critical channel for marketing energy-efficiency
programs.
¦	Project managers and engineers support account
representatives and should have a comprehensive technical
understanding of efficiency measures, rebate processing,
and administration.
In 2010, Consolidated Edison of New York held a two-hour
training program for account managers, during which they
provided an overview of leading energy-efficiency measures
for data centers and desktop IT equipment.
External Training
Utilities in California, the Pacific Northwest, Ohio, North
Carolina, South Carolina, and Wisconsin have demonstrated
that education programs focused on data center and IT
equipment energy efficiency are popular with customers and
vendors.
¦	SMUD has held an annual Business Computing showcase
for the past five years, featuring a seven-hour training
course and a show area for up to 30 vendors.
¦	Pacific Gas and Electric Company (PG&E) has been holding
two data center training courses each year since 2005,
and notes there has been increasing attendance by IT
professionals, along with the usual participation by facility
managers and members of the design and engineering
community. PG&E later began offering one general program
course and a targeted training course (i.e., data center
design, evaluation tools, and retrofit opportunities) annually.
¦	Duke Energy and utilities in Wisconsin have held five all-
day Data Center Efficiency Summits since 2010. Over 350
customer representatives have attended.
'"Source: Emerson Network Power. Understanding the Cost of Data Center Downtime: An Analysis of the Financial Impact on Infrastructure Vulnerability. 2011. Available online: http://
emersonnetworkpower.c om/en-U S/Brand ^Li ebert/D oc uments/White%20 Pa per^data - center-u ptime_24661 -R05-11. pdf
4
-------
Successful trainings include:
¦	An industry overview of market trends and challenges;
¦	Descriptions of available efficiency programs and services.
For example, an airflow management strategy class would
describe eight to ten measures that improve airflow, and
would cite customer case studies and describe a utility retro-
commissioning program; and
¦	A call to action message asking participants to consider
implementing measures and participating in efficiency
programs.11
In addition, most data center operators have little context of how
their facility's energy use compares to their peers' data center
energy use. EPA's measurement and tracking tool, ENERGY STAR
Portfolio Manager, can help data center operators assess their
facility's energy performance by providing an easy to understand
score between 1 and 100. Understanding energy performance
can capture the attention of not only the operator, but also other
business executives that make investment decisions. Data
centers that perform in the top quartile are eligible to apply for
ENERGY STAR certification.12 Also, the ENERGY STAR Low Carbon
IT Champion campaign recognizes organizations that reduce the
energy costs of their data centers. Education about assessing
energy use and recognizing energy-saving opportunities can
help motivate data center operators.
PG&E, Sacramento Municipal Utility District (SMUD), Seattle City
Light, and Southern California Edison (SCE) have promoted the
use of ENERGY STAR Portfolio Manager and participation in the
Low Carbon IT Champion campaign through customer education
programs.
Engaging Trade Allies
Within the data center industry, there are numerous trade allies
that could influence their customers' decisions about more
efficient options, thereby helping program managers deliver
efficient data center programs. The first three groups of trade
allies listed below focus on IT engagements (including equipment
and software solutions); the second three are important partners
for facility improvements:
1.	Value-Added Resellers (VARs): Sell IT equipment and services,
often providing integrated solutions to customers (including
project planning, equipment specification, and software
integration). VARs are crucial in the sale of premium-efficiency
equipment and for solutions such as server virtualization and
consolidation.
2.	System Integrators: Offer the same services as VARs, but can
also offer customized equipment and, in some cases, their own
branded IT equipment.
3.	IT Equipment Manufacturers: The largest firms offer
premium-efficiency equipment (including ENERGY STAR-rated
equipment) and often sell directlyto customers.
4.	Design and Engineering Firms: Design new data centers and/
or make significant modifications to existing facilities.
5.	Electrical and Cooling System Contractors and Maintenance
Firms: Can recommend retrofits and/or upgrades to existing
systems.
6.	Data Center Cooling Equipment Manufacturers: Market
leaders provide technical support for both new installations
and retrofits.
Trade allies often have an established relationship with
their customers and are relied upon for technical advice.
Unfortunately, not all IT vendors promote efficiency with their
customers due to a lack of knowledge or, in some cases, a
disincentive to promote efficient options that would reduce
sales. This disincentive can be mitigated to some extent by the
following strategies:
¦	Structuring incentives to reward vendor participation. For
example, upstream incentives to manufacturers or vendors
can align their interests directly with energy-efficiency
programs. This concept is covered in more detail in the
Designing Program Incentives section.
¦	Using incentives that are geared towards certain efficiency
behavior. Some utility programs incentthe removal of
inefficient servers rather than the purchase of new servers to
encourage server virtualization and consolidation.
¦	Offering energy-efficiency assessments to customers using
internal staff or contracted consultants. Utilities that choose
this program approach
(e.g., PG&E and SCE]
typically offer free
assessments if customers
commit to implementing
recommended measures
that meet a specified
financial return criteria.
Assessments performed
by consultants have the
advantage of providing
independent analyses that
customers can then use
to seek bids from multiple
vendors, thus engaging
the trade ally community.
"Trade allies are essential
to the success of energy
efficiency programs due to
their customer relationships
and their marketing and
sales efforts. It is crucial for
the trade ally community to
incorporate efficiency and
program resources into their
messaging. Investing time
and resources to help trade
allies succeed with energy
efficiency marketing and
selling will improve the cost
effectiveness ..."
-Geoff Overland, WECC
¦ Offering subsidies for
energy assessments
provided by trade allies.
Utilities that choose this approach typically offer to pay half
of the costs for energy assessments that meet established
criteria. Vendors often use the assessment process as a sales
tool.
Engaging trade allies in data center efficiency programs is
important to successful program implementation. A variety of
strategies can be used to identify potential trade allies, such as:
» Working with utility customers to generate leads by asking for
contractor and supplier referrals, thereby fostering utility-
customer relationships.
" A call to action message might focus on key drivers to implementing energy-efficiency measures, such as realizing cost savings, maximizing the use of existing capacity {and avoiding future
capital outlay for new capacity), and maintaining IT competitiveness.
12 As of April 2012,22 data centers had qualified for ENERGY STAR certification.
5
-------
¦	Pursuing productive outreach opportunities including regional
data center conferences, such as Data Center Dynamics.
Once trade allies are identified, the following outreach and
educational activities can lead to their active support of utility
energy-efficiency programs and services:
¦	Inviting vendors to customer education and training sessions.
¦	Creating efficiency program marketing materials that can be
co-branded with the trade allies' information.
¦	Encouraging vendors to invite efficiency program managers to
deliver presentations attheir customer outreach events.
Overcoming Split Incentives
Financial barriers related to the cost of efficient products and
services are often exacerbated by split incentives created
when the decision maker responsible for authorizing an energy-
efficiency project does not receive direct benefits from the
project. In data centers, the most common split incentive is
between the IT manager and the facility manager. The split
incentive can occur because, in some instances:
¦	The IT manager, responsible for selecting and deploying IT
equipment in a data center, typically makes financial decisions
based on available capital budget, and is not responsible for
power use or its associated costs.
¦	The facility manager, responsible for the power delivery,
cooling systems, and related utility bills, primarily manages the
operating budget and often cannot influence how the capital
budget is spent.
Generally, in organizations where this dynamic is present, the IT
manager seeks to stretch the capital budget as far as possible,
often by buying less efficient, lower cost IT equipment. This
eventually leads to a cooling system or power delivery capacity
shortfall, which must be rectified by the facility manager. The
facility manager must overcome this capacity shortfall with
limited capital funding, as well as absorb the ongoing operational
cost increases.
Outreach, education, and incentive programs should be designed
with the understanding that data center efficiency measures
cannot be implemented without the full commitment of both the
IT and the facility manager. As such, marketing and outreach
efforts should be targeted to both local IT and facility trade
associations, and program opportunities should be discussed
when the facility and IT managers are both present.13 After
learning about the large incentives available for data centers,
facility and IT managers can collaborate on efficiency program
activities and share incentive dollars. Incentive design strategies
are discussed in the following section.
In addition, for data centers located in leased buildings,the
property owner, leasing agent, and management firm may need
to be included in the decision-making process, along with the IT
and facility managers. Tenants are often hesitantto make capital
improvements to a facility that they do not own, and in some cases,
energy costs accrue to the building owner rather than the tenant.
Designing Program Incentives
Well-designed incentives can address key barriers in a given
target market by improving project cost-effectiveness and
addressing split incentive barriers. Incentives designed using
input from customers and trade allies are far more likely to be
effective than those designed in the abstract. This should be
complemented by continuing regular feedback and market
observation to allow program designers and operators to update
the program as market conditions change. Financial incentives
can be combined with nonfinancial incentives, such as
information and/or technical services, to maximize results.14
The following subsections describe incentive strategies and their
applicability to measures commonly targeted by programs.
Prescriptive Incentives
A prescriptive incentive (also referred to as a deemed incentive)
is commonly applied to measures that offer reliable savings on a
per-unit basis, regardless of installation practice. Only a handful
of measures in the data center are commonly promoted using
prescriptive incentives. Yet, prescriptive incentives can be an
important market entry strategy for developing relationships with
trade allies and for engaging customers seeking straightforward
savings opportunities.
Efficiency programs have successfully developed prescriptive
rebate programs for desktop computing and data center
IT equipment. These rebates are based on energy savings
generated by premium-efficiency power supplies and, in
some cases, by power management capabilities. In addition,
prescriptive rebates in other sectors for specific end-use
technologies (e.g., variable speed drives (VSDs), HVAC, pumps
and motors) could be marketed to data center operators.
Network PC power management, and server virtualization and
consolidation measures have been successful in the past,15 but
there are high freeridership concerns in customer segments
where IT is a core business function. For data centers with a
distributed infrastructure (server rooms, server closets, and
localized data centers), server virtualization has about a 40%
market penetration.16 As a result, a combination of technical
assistance and an easy-to-access incentive program may be the
best program delivery mechanism.
Depending on the incremental costs and incremental savings
associated with an efficient product, the incentive can be
targeted downstream to the purchaser, midstream to the
supplier, or upstream to the manufacturer. If an incentive only
represents 1% or 2% of the total measure price, it may be more
effective to engage the marketplace with midstream, or upstream
incentives: that is, to introduce an efficiency incentive into the
13One program manager illustrates the long-standing disconnect between IT and facilities departments with an anecdote: a facility managers swipe card was not programmed to let him enter
the data center.
"National Action Plan for Energy Efficiency. Customer Incentives for Energy Efficiency Through Program Offerings. Prepared by William Prindle, ICF International, Inc. February 2010. pp.12.
Available online: http://www.epa.gov/cleanenergy/document^suca/program_incentives.pdf.
15There are concerns about freeridership for server virtualization and consolidation projects because of high market saturation. However, freeridership claims associated with large businesses
should be carefully examined because industry sources indicate that market penetration hovers around 40%. As a result, opportunities remain within that market, and incentives for virtualization
can convince an organization to apply virtualization to high-value business functions.
16Source: Virtualization Industry Quarterly Survey: http://www.v-index.com/full-report.html. August 2012.
6
-------
production, stocking, and promotion of efficient IT equipment.
Successful midstream and upstream programs typically
involve the collaboration of multiple efficiency programs for a
larger customer base in order to attract vendor participation,
and possibly manufacturer interest (e.g., leading server
manufacturers sell directly to customers). Manufacturers have
historically been interested in changing production lines only if
there is a sizable market for energy-efficient models.
"Utilities can benefit from collaborating and partnering
with other utilities. Combining forces allows sets of utilities
to enter the space quickly representing a larger market.
This could lead to additional opportunities with customers,
vendors, and manufacturers."
- Lee Cooper, PG&E
Midstream and upstream incentives can be used to help
overcome the disincentive that vendors may have to promoting
efficiency. Both PG&E and Seattle City Ught have indicated that
upstream incentives can influence distributors to sell a more
efficient product by increasing those products' profit margins.
Upstream rebates can be offered for thin client, ENERGY STAR-
qualified servers, storage, and networking technologies.
Customized Incentives
The lack of measures appropriate for prescriptive incentives
leads to a reliance on more costly customized or performance-
based program models, which can be more complicated for
vendors and customers due to additional program rules and
requirements. This incentive structure may require:
¦	Qualified technical support for project verification,
¦	Application qualification processes that require engineering
estimates of energy savings and expected project costs prior
to project initiation (pre-installation), and
¦	Application approval that requires review of the calculations.
Consequently, data center program portfolios tend to have
higher administration costs than other market sector-targeted
programs. Despite this, these programs are cost-effective (at a
level similar to industrial sector programs), and they deliver some
of the best results of any sector-directed program. In addition,
data center projects deliver high peak energy demand and usage
savings. Customized incentives are particularly important for
data center cooling measures.
The following measures are generally cost-effective and suitable
for customized incentives:
¦	Water-side economizers
¦	Server removal
¦	Control systems for CRACs and air handlers
¦	Conversion of cooling systems from fixed volume to variable
volume by installing VSDs for fans and associated control
systems (using a customized approach, substantiated by field
measurements, may result in much higher verified energy
savings compared to a prescriptive incentive approach)
¦	Airflow management measures (measures that prevent supply
and return air from mixing, including containment systems)
While many data center energy-efficiency measures can be
evaluated using calculation methodologies, some (notably those
related to airflow isolation measures) will benefit from accurate
energy-savings values via field measurements and monitoring.
One advantage of data centers' very high load factor
and low seasonal variation in energy use is that pre- and
post-measurements to determine energy savings can be
accomplished in short timeframes (a week or two of monitoring
before and after project implementation is usually sufficient).
Efficiency programs requiring post-measurement of on-site
energy use to more accurately quantify savings may wish to
consider using simple standard spreadsheet-based calculation
models for evaluating some measures.
Retro-Commissioning Based Incentives
Retro-commissioning (RCx), or tune-up services, represents a
new program area for utilities. These programs typically feature
a facility audit that recommends no- and low-cost efficiency
measures and provides financial incentives for measures
that are implemented by the customer or a contractor. Some
programs feature pre- and post-measurement of actual savings.
Retro-commissioning has become more important as equipment-
efficiency opportunities have become more difficult to capture
and less cost-effective. RCx services can directly generate
cost-effective energy savings, and can also serve as a gateway
service by establishing a utility/customer relationship that can
lead to more substantial retrofit projects.
Retro-commissioning, or RCx, opportunities depend on data
center size. Permanent monitoring and control systems can
be installed to measure actual efficiency gains, but this is
generally only cost-effective for data centers larger than
10,000 square feet. For such facilities, customers should be
encouraged to consider installing a monitoring and control
system, not only to measure energy savings from airflow
management upgrades, but also to ensure the persistence
of energy savings. An RCx service addressing smaller
data centers can use temporary metering and monitoring
equipment to measure pre- and post-conditions, entering
data into a calculation model that predicts and calculates
energy savings. P£?<$£pioneered an RCx program in 2008
targeting smaller data centers.
¦ Replacement of infrared humidifiers with adiabatic units17
17 Adiabatic humidification raises humidity without adding heat, and includes ultrasonic, fogger, and mister technologies.
7
-------
For data centers with dedicated cooling infrastructures
(excluding most server rooms or closets), ample opportunity
exists for addressing poor airflow conditions, which can
dramatically improve cooling system efficiencies. As a side
benefit, often of paramount concern to the data center operator,
improved airflow can recapture cooling system capacity,
allowing the addition of more IT equipment over time.
Some low- or no-cost airflow management improvement and
cooling system retrofits—separately or in combination—reduce
energy use and improve cooling system effectiveness. These
improve airflow delivery by:
¦	Removing obstructions from under-floor plenums
¦	Blocking airflow to ancillary equipment (such as power
delivery units and distribution transformers)
¦	Blocking airflow through cable outlets
¦	Installing blanking panels in all IT equipment racks
¦	Rebalancing airflow by moving grate tiles
¦	Reorienting IT racks in a hot aisle/cold aisle configuration
¦	Ducting return air
Please Note: Fire Suppression Safely
In order to comply with fire safety recommendations and
codes, customers may have to take additional installation
or protection steps when pursuing airflow management
strategies.
For example, hot aisle/cold aisle containment structures
may be required to have separate "in-aisle" fire protection
systems because their components (e.g., enclosed ceilings,
plastic curtains) can:
¦	Shield the hot aisle from data center's regular fire
suppression; and
¦	Prevent smoke from reaching detection systems.
Current and local codes may require certain construction
materials or patterns.
These measures allow data center operators to:
¦	Raise supply air temperatures to an optimum level so that
cooling is effectively provided to all IT equipment (typically
when equipment does not receive sufficient cooling, the
response is to lower the supply temperature rather than
improve airflow delivery).
¦	Avoid supply and return air mixing, which results in lower
return air temperatures. This is important because cooling
units have a rated capacity based on the difference in supply
and return air temperatures (generally, capacity is listed at a
20-degrees Fahrenheit delta). If the difference in temperatures
falls below the specification, the unit provides a de-rated
cooling capacity.
¦	De-commission a portion of their cooling units, which are
operating attheir full capacity. This is one reason that
airflow management measures can be so valuable: they
not only improve energy efficiency, but they often reclaim
cooling capacity that was unavailable due to poor operating
conditions.
Table 3 summarizes classes of energy-efficiency measures and
indicates which retrofit program approach typically applies. By
balancing program elements (offering prescriptive incentives
and retro-commissioning services, for example), energy savings
can be generated in a shorter timeframe and improve the
performance ofthe entire program portfolio.
PROGRAM PLANNING AND ROLL OUT
Understanding market potential and the market penetration of
data center efficiency measures provides valuable insights into
how a program should be delivered and what incentive levels
would be cost-effective and successfully affect the market.
Establishing Market Potential and Projecting Program Targets
The data center program planning process should include a
baseline market assessment ofthe energy savings potential for
data centers. This baseline will allow program managers to set
realistic savings goals and design programs that are well-suited
for the target market in their service area. Baseline assessments
for data center programs typically capture the information
outlined in the following subsections.
Proportional Estimation of Energy Use for the Data Center
Market
Data centers of all types account for approximately 1.6% of
national energy sales.18 This proportion can be applied to a utility
load to establish a market assessment baseline. Utilities that
primarily serve rural areas might decrease this estimate; those
that serve highly urbanized areas might increase the estimate.
Cataloguing Data Centers within Service Territory
Utilities should consider adjusting the load estimate based on
direct knowledge of enterprise and utility-scale data centers.
The number of utility-scale and enterprise data centers (defined
previously) can be used as a proxy to validate the energy use
estimate. Industry analysts report that approximately 40% of
data center load is attributable to these two classes. A utility
can validate a program's assessment with a reasonable estimate
of this load, approximating 40% ofthe estimated load with the
proportional approach. Planners should catalogue:
¦	All customer loads in excess of 10 MW, and which of those are
utility-scale data centers; and
¦	Customer loads between 1 MW and 10 MW, and which of
those are enterprise data centers.
It is important to recognize that any assessment of market
potential is an estimation; British Columbia Hydros (BC Hydros)
data center program, which had low adoption rates during initial
implementation, found that data center customers were smaller
and fewer than they had expected.
18Source: Masanet, Eric R.(2012). Estimating the Energy Use and Efficiency Potential of U.S. Data Centers. Lawrence Berkeley National Laboratory, Paper LBNL-5168E. Retrieved from: http://
esc hola rshi p. org/u c/item/1475d 9h0
8
-------
Table 3. Illustrative Program Treatment for Data Center Measures
Measure	Program Model
Measure or Technology	Description	Prescriptive Incentive	Customized Incentive	Performance/RCx
Data Center Cooling Systems
Premium Efficiency Equipment
Premium efficiency chillers,
pumps, fans, cooling towers,
and other components; ultra-
sonic humidifiers.
Can use programs established
for other markets (e.g., com-
mercial offices), but will likely
underestimate savings.
Preferred for retrofits to cap-
ture savings from high load
factor use of equipment.

Variable Speed Fans
Variable speed fans on com-
puter room air conditioners/
handlers.
Can use programs established
for other markets (e.g., com-
mercial offices), but will likely
underestimate savings.
Preferred for retrofits to cap-
ture savings from high load
factor use of equipment.

Cooling System Controls
Upgrading controls, integrating
controls of multiple cooling
units, or installing metering
and monitoring to improve
control of system.

Generally must be undertaken
with suitable airflow manage-
ment upgrades and/or variable
speed fans.

Air-side and Water-side
Economizers
Use of outside air or water-
side cooling when environ-
mental conditions a re favor-
able.

Water-side retrofits are very
attractive; air-side retrofits are
more difficult.

Airflow Management
Upgrades
A set of no- and low-cost mea-
sures that ensure the proper
delivery and return of cooling
air, limiting mixing.


Ideal approach, with pre- and
post-measurement of supply
and return temperatures and
calculation of achieved sav-
ings.
Power Delivery and Conditioning
ENERGY STAR certified UPS
and Premium Efficiency
Distribution Transformers
Specification of premium-
efficiency equipment.

Easily handled using standard
calculation model; note that
loading of equipment is crucial.

Direct Current Power Systems
Use of high-voltage, direct
current power delivery and
conditioning schemes.



Self Generation
Use of fuel cells, solar PV, or
other generation.
Generally treated as a
prescriptive incentive.


Premium Efficiency IT Equipment
Premium-Efficiency IT
Equipment
Specification of premium-
efficiency power supplies in IT
equipment, including ENERGY
STAR-certified equipment.
Applicable as a downstream,
midstream, or upstream pro-
gram.


Virealization and
Consolidation
Softwa re that allows for
higher utilization of comput-
ing (server) and data storage
equipment.
Can be delivered as a prescrip-
tive rebate.
Most often handled using a
standard calculation model.

Data Storage Measures
Premium-efficiencyequipment
(power supplies), solid-state
storage, thin provisioning,
compression, de-duplication.

Some measures can be accu-
rately assessed using calcula-
tion models (other technolo-
gies do not have predictable
savings).

Estimating Potential Program Performance
Program implementers can develop a well-documented market
baseline for the data center segment. The following should be
considered when establishing baselines:
¦	Impending changes in technology that improve energy
efficiency:
¦	Estimates of freeridership and spillover based on results from
data center program evaluations; and
¦	Changes in existing and upcoming standards. For example,
in 2013, the State of California will adopt new construction
standards for data centers that include requirements for free
cooling systems, which will obviate those measures from utility
new construction incentive programs in California. Similarly,
the EPA's ENERGY STAR program will adopt new criteria for
classes of IT equipment, which may lead to changes in market
conditions overtime.
Retrofit opportunities in existing data centers can often capture
savings in the range of 25% to 50%, depending on the age of the
data center and the operator's willingness to make significant
upgrades. For example, eBay's 139,000 square-foot data center in
9
-------
Phoenix (of which 65,000 square feet is white space) houses all
of eBay's business units.19 This facility, built in 2004, has been the
target of numerous energy-efficiency upgrades.
A new construction project might yield energy savings of 10% of
the typical baseline design by implementing power delivery and
conditioning measures, specifying premium efficiency cooling
equipment, and fully isolating airflow.
Once savings claims are established, it is importantto prepare
work papers for prescriptive rebate measures (e.g., server
virtualization and consolidation, premium-efficiency servers) and
prepare standard calculation models for selected customized
Table 4. Ultrasonic Humid ilication Payback

Power
Hours of
Number of
Energy Use

Draw(kW)
Use*
Units
(kWh)
Steam
Humidifier
30.3
4,380
6
796,284
Ultrasonic
Humidifier
0.67
4,380
8
23,477
Energy savings (kWh)
772,807
Energy savings ($) at 6.5 cents per kWh
$50,232
Costs (labor and ultrasonic humidifier purchases)
$94,270
Payback (years)
1.9
Potential incentive at 11 cents per kWh
$85,009
Actual incentive limited to 75% of costs
$70,703
Payback (years) with incentive
0.5
* This calculation is based on humidifiers operating half of the time.
Other data centers may have much lower average hours-of-use and
need to adjusttheirsavings accordingly.
Program Tracking
A program tracking system is important for measuring program
progress and tracking energy savings. As with other commercial
programs, program administrators have found that, in addition to
customer information, it is helpful to trackthe data listed in Table
6 for each installed measure.
EM&V Planning
Data center operations feature several unique characteristics
that complicate verification of claimed energy savings and
program attribution. Prior to launching a program, program
managers can recognize, understand, and prepare for these
complexities by meeting with evaluators to plan the program and
performance expectations, thereby reducing risks associated
with freeridership, load growth, and short equipment production
cycles. Other strategies for mitigating risk include discouraging
program participation by utility-scale data center operators
(for most measures), establishing defensible energy-efficiency
baseline standards, and collecting information from customers to
document program influence.
measures. These work papers and models should be submitted
to regulators for approval.
Table 4 and Table 5 outline the costs and benefits associated
with two retrofit measures that eBay instituted in its Phoenix
facility: ultrasonic humidification and VSDs. The tables show the
two measures' simple paybacks (costs divided by annual energy
savings), with and without incentives offered by Arizona Public
Service (APS) Company. The incentives allowed for reasonable
payback periods, which enabled eBay to pursue these retrofits.
In general, data centers, even utility-scale data centers, need a
one-year payback or less on retrofit measures to move forward
with projects.
Table 5. Variable Speed Drive (VSD) Payback

Power
(HP)
Efficiency
Hours of
Use
Number
of Units
Energy
Use (kWh)
Constant
Speed
10
90%
8,760
83
6,026,685
Variable
Speed
10
95%
8,760
83
1,663,675
Energy savings (kWh)
4,363,010
Energy savings ($) at 6.5 cents per kWh
$283,596
VSD equipment costs
$341,960
Harmonic transformer costs
$338,984
VSD installation costs
$69,056
Payback (years)
2.6
Potential incentive at 11 cents per kWh
$479,931
Actual incentive limited to 50% of costs or $300,000
$300,000
Payback (years) with incentive
1.6
In addition, utility program managers should consider reviewing
the latest evaluations that offer analyses of data center
programs. Few of these evaluations have been conducted,
because few mature data center programs exist.20 Three
evaluations of early programs had the following conclusions:
¦	Given market potential, evaluators believe that utility managers
may have difficulty growing data center programs quickly
enough to meet demand.
¦	The split incentive challenge is particularly acute for the data
center market and associated decision makers, especially in
co-location facilities.21
Addressing Program Implementation Challenges
Understanding market conditions and program implementation
challenges is important for effective program planning and for
developing a reasonable forecast of energy savings. The major
challenges for data center program administrators are the
technical complexity, long lead times, and product production
cycles associated with data centers, as well as the risk of
freeridership.
"This includes server applications for the eBay services, analytical databases and storage devices for internal analysis, and router^switches to direct traffic to the applicable system arrays. In
addition, telecom gear will be required to interconnect to a fiber optic backbone.
20This data is as of May 2012. Mature programs are defined as those that have been deployed for multiple program cycles.
21 Seattle City Light reported seeing more interest from co-location providers.
10
-------
Table 6. Typical Program Tracking Data
Measure-Level Information
Power Consumption Information
Energy Savings Information
¦	Measure Type
¦	Measure Brand Name
¦	Measure Model Number
¦	Measure Description
¦	Measure Capacity
¦	Percent of Load on Measure
¦	Quantity of Measure
¦	Level of Incentive
¦	Installation Date
¦	Power Draw of Installed Equipment
¦	Power Draw of Typical Equipment Installed at Time of
Purchase
¦	Power Draw of Old Equipment
¦	Energy Savings
¦	Summer Demand Savings
¦	Winter Demand Savings
¦	Years of Useful Life Remaining on Old
Equipment
¦	Years of Useful Life for Installed Equipment
¦ Technical Complexity: All energy-using elements of data
center operations are technically complex, because
they use special-purpose equipment with a key goal of
ensuring reliability Generally, identifying energy-efficiency
opportunities and accurately characterizing savings is best
suited for engineering and technical experts who specialize
in data center facilities and IT equipment. Program managers
should evaluate whether their traditional technical services
team has sufficient data center expertise to support the
evaluation of cooling, power delivery, and conditioning
systems and, if so, whether using their staff in this capacity
is effective. Some examples of how this issue has been
addressed include:
•	The process evaluation of PG&E's 2006 - 2008 High-Tech
Program recommended that PG&E expand its efforts to
educate customers and utility account managers about
IT energy-efficiency measures and technologies for data
centers. The evaluation noted that PG&E had difficulties
performing comprehensive energy audits for data centers,
and that its technical support activities instead focused on
specific measures.
•	If a utility cannot rely on in-house expertise, it should select
qualified technical service contractors who are familiar with
specific data centers technologies (e.g., cooling and power
systems) and limit each of these contractors to evaluating
the specific systems within their area of expertise.22 The
evaluation of Silicon Valley Powers Fiscal Year 2008-2009
program recommended that the utility work with local
consultants and engineers to identify opportunities.
•	Another method of providing technical support services
is to contract with an energy service provider (ESP). ESPs
can implement portions ofthe program, such astechnical
As part of a regulatory requirement to provide a portion of
its energy-efficiency program portfolio through independent
contractors, SCEselected an energy service provider (ESP)
to run a data center retrofit service.
The ESP performs marketing and outreach activities in
partnership with SCE account representatives, provides
technical assessments of customer facilities, and manages
the incentive application process for customers.
"It is advantageous for utilities to partner with ESPs that
have traction and success in the data center market.
Relationships with industry trade allies, engaging in
customer-specific data center technical situations, and
accelerating the program time-to-market are valuable
examples that ESPs can offer utilities looking to enter the IT
and data center efficiency segment."
-Geoff Overland, WECC
reviews, engineering support, and turnkey implementation
services. For comprehensive program delivery, ESPs
generally consider either a cost-based service contract or a
performance-based arrangement that makes payments for
energy-efficiency accomplishments.
¦	Long Lead Times: Data center deployments often take a
longer time to complete than other types of energy-efficiency
engagements. All projects, whether related to IT equipment
or its supporting infrastructure, require careful planning
and execution. A data center often takes two years or more
to build, and end-to-end project cycles (from planning to
commissioning and operation) are typically five years or
longer. For projects that aim to recapture capacity (such as
airflow management upgrades) or increase utilization rates
(such as virtualization and consolidation of servers), savings
can also be masked over short periods by growth in IT
workloads. In simple terms, reclaimed capacity will often be
allocated or used up quickly by new equipment. Customized
incentive retrofit and new construction projects have long
periods between application approval and project completion.
This extended implementation period can complicate program
planning; utility incentive funds for retrofit projects may not
be paid (and the accomplishments cannot be accrued) in the
program year they were committed. A program approach that
captures a time-dependent energy savings stream, based on
projected load growth, may solve this dilemma, although no
known new construction programs have this feature. Most
importantly, efficiency program evaluators should be informed
ofthe long project lead times associated with this market.
¦	Production Cycles: Data centers are not static environments.
Servers and many other types of IT equipment have production
cycles of about a year, due to frequent technological upgrades.
These product cycles are unlike product categories such as
"Technical service contractors that typically conduct commercial HVAC audits may not be able to assess complex data center cooling systems..
11
-------
HVAC equipment food service equipment, and residential
appliances, which generally advance only over multiyear
timeframes. Technological advances can cause data center
equipmentto become antiquated with relative frequency.
This phenomenon suggests that IT equipment incentives
should reward customers who purchase premium-efficiency
equipment that compares to currently available standard
models, rather than to their existing installed equipment.
¦ Freeridership: Industry leaders (often utility-scale data
centers) are more likelyto undertake energy-efficiency
projects (particularly in new construction) in the absence
of utility program offerings and involvement. These industry
leaders are generally companies from the high technology,
communications, financial services, and government sectors.
Proper targeting from the onset can help decrease issues
surrounding freeridership.
• Many industry leaders undertake energy-efficiency retrofits
absent utility efforts because the most recent technologies
and high-operational efficiencies are required to remain
competitive. Program designs that set maximum incentive
payments, preclude certain rate classes or facility loads, or
limit participation by operators of large, sophisticated data
centers can mitigate freeridership concerns.
• Server virtualization may be common practice in utility-
scale data centers, but is less common for localized data
centers. The evaluation of BC Hydros Data Center Program
concluded that customers (i.e., utility-scale data centers)
who are committed to pursuing server virtualization will
undertake these projects regardless of the additional
incentive dollars.
Implementation Strategies
A program development and launch strategy can feature phased
deployment, thereby encouraging early accomplishments,
while laying the foundation for a broader portfolio. The program
delivery roadmap presented in Table 7 focuses on delivering
high-value program elements, while building internal expertise
and customer and vendor engagement. The elements contained
in the Early Stage Program column could be delivered in the
program cycle's first year, with Mid-Stage activities added in the
second or third years. Advanced Programs require sophisticated
program development work, but could be added to address local
markets.
Table 7. Sample Data Center Program Implementation Timeline
Activity/Program Element
Planning
Early Stage
Program
Mid-Stage
Program
Advanced
Programs
Assess market baseline
V



Prepare work papers for prescriptive rebate measures and
submitto regulators for approval
V



Hold internal stakeholder training (for account representatives,
program managers, etc.)
y



Meet with program evaluatorsto review the design and
implementation plan
y



identify potential vendor partners
y



Prepare standard calculation models for selected measures
y



identify technical support contractors and issue an RFPfor
technical support and/or program management and delivery
services
y



Offer prescriptive rebate program

y


Hold customer and vendor training events

y
y

Participate in vendor-sponsored outreach activities

y
y

Conduct energy assessment services

y
y
y
Offer customized incentives for selected measures and
technologies

y


Expand customized incentive measure eligibility


y

Monitor program results

y
y
y
Consider new construction program


y
y
Consider offering server room retro-commissioning program


y
y
Consider upstream/midstream rebate program for prescriptive
measures (e.g., energy-efficient servers)


y
y
-------
Keys to Success:
BC Hydro believes the key to their program's success was a
committed internal team of program and account managers
who worked well with industry stakeholders and who had
technical expertise and team building and interpersonal
skills.
WECC recommends that utility program managers should
focus on one measure, or a small subset of measures, to
get the program started. By entering the market with a
narrow focus, utilities can avoid risking too much, expedite
implementation, and ultimately grow more successfully.
SUMMARY
Energy efficiency programs can achieve sizeable energy savings
and advance the adoption of efficient products and practices in
data centers. While market barriers exist, they are understood by
the industry and can be addressed with proven program designs
and effective emerging strategies that focus on key measures
and systems. Tapping into data center industry interest in
managing costs and recapturing under-utilized infrastructure
is an effective way to engage data center management.
Energy-efficiency measures can be employed without reducing
reliability, and in some cases, can improve reliability, or help
recapture capacity. Efficiency program administrators can
partner with their customers and the vendor community to
deliver cost-effective program results starting with a modest
program portfolio that can be extended and expanded based on
early success.
13
-------
RESOURCES
Federal Resources
EPA and DOE offer the following tools and resources to help
efficiency programs, customers, and implementers:
¦	Energystar.gov/products - includes information on qualified office
equipment and enterprise servers. Market share information for
existing ENERGY STAR qualified products is available at energystar.
gov/usd.
¦	Energystar.gov/datacenters - includes information on guiding
principles for measuring energy efficiency in data centers, the
ENERGY STAR rating for data centers, and ENERGY STAR product
development efforts.
¦	Energystar.gov/lowcarbonit - includes information on the ENERGY
STAR Low Carbon IT campaign, a nationwide effort to assist and
recognize organizations for reducing IT energy consumption.
¦	Energystar.gov/ia/products/power_mgt/downloads/
DataCenter-Top12-Brochure-Final.pdf- EPA's Top 12 Ways to
Decrease the Energy Consumption of Your Data Center provides more
information about energy-efficiency opportunities in data centers.
¦	www1.eere.energy.gov/manufacturing/datacenters/software.
html - includes a Data Center Profiler Software Tool Suite that allows
companies to identify and evaluate efficiency opportunities in data
centers. DOE also offers training and certification for data center
practitioners.
Ratepayer-Funded Programs Promoting Efficient Data
Centers
Consortium for Energy Efficiency: Data Center Program Offering
Summary Table (2010/2011 Data Center Programs), www.ceel.
org/com/dcs/dcs-main.php3
Arizona Public Service: www.aps.com/main/services/
SolutionsForBusiness/it.html
Austin Energy: www.austinenergy.com/energy%20Efficiency/
Programs/Rebates/Commercial/Commercial%20Energy/
dataCenter.htm
Duke Energy: www.duke-energy.com/north-carolina-business/
smart-saver/smart-saver-custom-incentive-program.asp
Pacific Gas and Electric Company: www.pge.com/hightech/
Xcel Energy: www.xcelenergy.com/Save_Money_&_Energy/
For_Your_Business/Customized_Solutions/Data_Center_
Efficiency_-_CO_-_MN
INDUSTRY INITIATIVES AND GROUPS
C-Net: www.cnet-training.com/
Climate Savers Computing Initiative: www.
climatesaverscomputing.org/
Consortium for Energy Efficiency (CEE): www.cee1.org/
Datacenter Dynamics: www.datacenterdynamics.com/
The Green Grid: www.thegreengrid.org/
Uptime Institute: www.uptimeinstitute.com/
Evaluation Resources
California Measurement and Advisory Council searchable
database: www.calmac.org/search.asp
Consortium for Energy Efficiency Evaluation Clearinghouse:
www.cee1.org/search/search.php
ACKNOWLEDGEMENTS
The authors of this document, Mark Bramfitt (Bramfitt
Consulting), Allison Bard and Robert Huang (The Cadmus
Group), and Maureen McNamara (U.S. Environmental
Protection Agency) would like to thankthe following individuals
who provided insights to inform the development, or were
instrumental in reviewing this guide:
David Rogers, British Colombia Hydro
Geoff Overland, Wisconsin Energy Conservation Corporation
Jason Erwin, Consortium for Energy Efficiency
Jason Weiner, Duke Energy
Jennifer Stephenson, U.S. Environmental Protection Agency
Kyle Addiss, Southern California Edison
Lee Cooper, Pacific Gas and Electric Company
Michael Zatz, U.S. Environmental Protection Agency
Robert Jericho, Commonwealth Edison
Robert Meyers, U.S. Environmental Protection Agency
Ted Brown, Seattle City Light
Zachary Suttile, Willdan Energy Solutions

LEARN MORE AT
energystar.gov
ENERGY STAR, a U.S. Environmental Protection Agency program, helps us all
save money and protect our environment through energy efficient products and
practices. For more information, visit www.energystar.gov
14
-------