1. Creating Energy-Efficient Data Centers
Paul Scheihing
U.S. Department of Energy
Office of Energy Efficiency and Renewable Energy
Industrial Technologies Program
Data Center Facilities and Engineering Conference
Washington, DC
May 18, 2007

2. Why Data Centers? Highly energy-intensive and rapidly growing
Consume 10 to 100 times more energy per square foot than a typical office building
Large potential impact on electricity supply and distribution
Used about 45 billion kWh in 2005
At current rates, power requirements could double in 5 years.

3. Potential Benefits of Improved Data Center Energy Efficiency
Save 20 billion kWh per year by 2015
Worth $2 billion, ≈ annual electricity use in 1.8 million American homes
Potentially defer need to build 2,300 MW of new generating capacity
And avoid 3.4 million metric tons of carbon emissions (like taking 675,000 cars off the road)
Extend life and capacity of existing data center infrastructures
Assume ITP initiative will reduce electricity consumption at data centers by 20 billion kWh per year (1)
This is equal to the amount of electricity consumed by 1.8 million homes (2)
This initiative will reduce energy costs for the industry by $2 billion per year (3)
This initiative will reduce GHG emissions by 3.39 million metric tons of carbon per year (4)
This reduction would be equal to taking 675,000 cars off the road (5)
Notes:
(1) 20 billion kWh savings in 2011 (2015?) was decided by Paul
(2) based on average household electricity consumption of 11,000 kWh per year.  This value is from:
(3) based on electricity cost of $0.10 per kWh in 2011 (2015?).  Average electricity price for commercial facilities was $ per kWh according to Table ES1 in 2005 according to EIA’s Electric Power Annual 2005.
(4) based on average carbon emissions for electricity in the U.S. according to data from Tables A8 and A18 in EIA’s Annual Energy Outlook 2007.
(5) based on average carbon emissions of 5 metric tons of carbon per car per year according to:

4. Building Existing Knowledge Base
R&D Roadmap by Lawrence Berkeley National Lab (LBNL) identifies and prioritizes data center opportunities and research.
With funding from PG&E and CEC, LBNL conducted benchmark studies of 22 data centers:
Found wide variation in performance (total power/IT power)
Identified best practices
DOE will greatly expand current knowledge base.
Ratio of Total Data Center Power to IT Equipment Power
Total Power/IT Power

5. Energy Efficiency Opportunities
Power Distribution & Conversions
Server Load/ Computing Operations
Cooling Equipment

6. Data Center Energy Use Typical Data Center Energy End Use 100 Units
Power Conversions
& Distribution
35 Units
Cooling Equipment
Server Load /Computing Operations
Server Load/ Computing Operations
Cooling equipment (air conditioners, fans, humidifiers, pumps)
Power distribution equipment (transformers, uninterruptible power supplies, AC/DC inverters, wiring)
33 Units Delivered

7. Cooling & Power Conversions
Data Center Cooling and Power Conversion Performance Varies
Cooling & Power Conversions
Cooling & Power Conversions
Server Load /Computing Operations
Server Load /Computing Operations
Typical Practice
Better Practice

8. Typical Energy Flow/Use
Electricity Generation & Transmission Losses
Delivered Power
Power Conversion & Distribution
Cooling Equipment
Server Load/ Computing Operations
Fuel Burned at Power Plant

9. Typical Energy Flow/Use
Power Conversion & Distribution
Cooling Equipment
Server Load/ Computing Operations
Electricity Generation & Transmission Losses
…ultimately reducing fuel burned at the power plant
Reducing power demand and losses
Lowering power conversion losses
Delivered Electricity
Will reduce cooling needs
Reducing server power requirements
Fuel Burned at Power Plant

10. Energy Efficiency Opportunities
Better air management Move to liquid cooling Optimized chilled-water plants Use of free cooling
On-site generation CHP applications Waste heat for cooling Use of renewable energy Fuel cells
Power Conversion & Distribution
Server Load/ Computing Operations
Cooling Equipment
Load management Server innovation
Power Conversion: Potential for 10 to 30% improvement
Server Load/Computing: Potential for 30 to 50% improvement (check EPA report)
Load management
Virtualization
Sleep modes
Load shifting
Server innovation
Chip design
Efficient power supplies
Semiconductor materials
Cooling: Potential for 30 to 50% improvement (check EPA report)
Alternative Power Generation:
On-site generation (eliminates transmission losses)
CHP applications
Use of waste heat for cooling
Use of renewable energy (could couple with DC power distribution systems) PV
Fuel cells
High voltage distribution Use of DC power Highly efficient UPS systems Efficient redundancy strategies
Dale’s revision:
Improve “in-the-box” power supply efficiency
Improve efficiency of software applications
Improve resource utilization (e.g. virtualization)
Reduce idle power (power management)
Hardware innovation (e.g. more efficient computations per watt)
Alternative Power Generation

11. Opportunity Potential
Comparison of Projected Electricity Use, All Scenarios, 2007 to 2011
140
2006 Baseline 58.7
120
Business as usual
100
Current trends 80
Improved operational management
Annual Energy Use (Billion kWh/year) 60
Best practice 40
State of the art 20
2007
2008
2009
2010
2011

12. What Is Needed
Assistance in identifying the best opportunities for savings at each data center
Outside validation to help convince management that addressing opportunities is feasible and cost-effective

13. DOE Data Center Team Industrial Technologies Program
Building Technologies Program
Hydrogen, Fuel Cells, & Infrastructure Technologies
Federal Energy Management Program (FEMP)
DOE National Laboratories

14. DOE Data Center Program Objectives
Provide systems approach
Build tools, expertise, and strategy
Raise awareness of the opportunity
Recognize industry leaders

15. Save Energy Now: Industry Assessments
200 completed
Natural gas savings = 52 trillion Btu/yr
≈ 725,000 U.S. homes
Carbon dioxide avoided = 3.3 million metric tons/year
Cost savings opportunity = $475 million per year
Savings implemented or planned = $256 million (154 plants)
2 – 4 years
Modify steam turbine operation
Use oxygen for combustion
Change process steam use
> 4 years
Install CHP system
< 9 months
Improve insulation
Implement steam trap program
Clean heat transfer surfaces
9 mo. – 2 years
Heat feed water with boiler blowdown
Lower excess oxygen
Flue gas heat recovery
Carbon dioxide avoided = 3.3 million metric tons/year (7% of total US greenhouse gas emission growth, 2004 – 2005)
Estimated Payback Periods for Recommended Actions

16. Program Strategy Build on Save Energy Now model
DOE deployed software tools, training curriculum, and qualified experts to train and work with staff at large U.S. plants.
65% of recommended actions now completed, in progress, or planned.
With industry input, develop appropriate tools, training, and qualified experts to improve data centers.
Conduct pilots, promote and facilitate industry implementation.
Tool kit include: best-practice guides and case studies, benchmarking, and sub-system assessment protocols/tools

17. 2007 Move Forward Plan
Build strong liaisons and partnerships with industry
Develop robust new energy assessment program
Develop tools and info on best practices
Sub-system assessment protocol and analysis tool
Assessment framework and energy profiling tool
Conduct pilot assessments at data centers
Provide awareness training
Screen for industrial demonstrations
Provide Federal procurement specifications
ITP has a number of initiatives to impact energy efficiency in the U.S. industrial sector.
ITP is developing new R&D Technology Platforms to impact a broader industry sector in utilize Supply Chain to motivate energy savings.
ITP is working with American National Standards Institute (ANSI) to develop energy efficiency certification
ITP is developing an initiative linking carbon emissions reduction to energy efficiency
Emphasis placed on deployment of new technologies and processes.

18. Stakeholders EPA States Utilities Industry Organizations
e.g., Green Grid, ASHRAE, AFCOM, 7x24, SVLG
Equipment suppliers
Research organizations
Consultants

19. How Can Industry Participate?
Register on web site to get regular updates
Participate in Peer Review of products, protocols and best practices
Sign up for Technical Working Groups on web site
Conduct Self Benchmarking and report results
Use tools from LBNL site and download protocol at :
Apply for Data Center Assessments (solicitation coming in Fall)

20. Self-benchmarking Guide Other Reports (demonstrations)
Web-based Resources
Good starting point for those seeking efficiency measures
Best Practices
Benchmark data
Self-benchmarking Guide
Case Studies
Other Reports (demonstrations)
Design Guidance

21. end
Contact: Paul Scheihing