A great site I found in 2016 was energyliteracy.com. The creator Saul Griffith painstakingly combed through Department of Energy data to produce an interactive graph of all energy production to power consumption in the US.
In this post, I will take a look at data center power consumption and discuss relative impacts to consider vs. other industries.
If you are not familiar with a British thermal unit (BTU), it is defined as the amount of heat required to raise the temperature of one pound of water by one degree Fahrenheit. The metric unit equivalent is the joule, where one BTU is about 1055 joules. Griffith further refines the measurements into quads which are quadrillion BTUs. If you are overloaded on the measuring systems, just know that this is a lot of energy!
The first thing I took a look at was the total energy produced vs. the waste in the United States. The comparison of production to waste gives us a good look at the net efficiency of the energy systems in the US.
Looking specifically at electricity production portion of the chart we can immediately see a large amount of inefficiency with 38.49 quads produced and only about 7.8 quads being used. The remaining 30.69 quads are wasted! Using a ratio of quads produced divided by the quads consumed (38.49/7.8), we get 4.93. We can compare this to Power Usage Effectiveness like ratios for data centers and see this is indeed dismal. Imagine a data center with a dismal PUE of 5!
After being astounded by the poor efficiency of electrical generation and consumption let's now look specifically at data centers in comparison to other industry sectors.
According to energyliteracy.com data centers consume a total of 0.029 quads with 0.016 quads of waste and 0.013 quads of consumption. Performing a similar energy usage ratio comparison (0.029/0.013) gives us 2.23.
If we scan across other industrial consumers of electricity (click on the electricity column specifically), we can see that there are consumers that are larger than the data center industry. With single-family detached (3.563 quads, usage ratio 1.6), manufacturing (2.534 quads, usage ratio 1.73), and office (0.833 quads, usage ratio 1.56) being listed as much higher consumers. The usage ratios show that the electrical usage of these industrial sectors are also more efficient that data centers.
Now let's consider a computerized world and the effects of adding devices controlled by computer vs. the world without computerization. Are we gaining in efficiency? Even if data centers and the associated computer infrastructure are not super efficient, the overall effect of adding computers to an energy consuming device will typically make the device more efficient. See Koomey's law if you are interested in trends in energy efficiency.
If you agree that a computerized world is more energy efficient, then we can start considering the associated effects of adding computer-controlled devices which rely on data center infrastructure to create a net positive efficiency affect. Data centers should compete and become more efficient than the energy sectors that the technology services. Using less overall energy is a first positive gain by adding computer devices. The transition becomes even more efficient with the computer controlled infrastructure being as minimally impactful as possible.
The change in efficiency by computerizing industries with the highest energy consumption represents targeted opportunities to reduce the energy intensity of a given industry by applying computerized methods. Innovations can specifically be focused to provide a better social outcome through increased efficiency with initial impacts targeted at the most inefficient industries - ideally those that are not as efficient as the data center industry.
Is the United States thus becoming more energy efficient? Taking a look at the annual report from Energy Information Administration we see some positive trends.
We also get exposed to a much more complicated view of energy in the United States and see dynamics from economic growth, fuel source changes, and in the EIA report, we can see references to improved technologies for oil and gas production.
We can potentially see data center effects on the energy production industry where one example of computerized oil production not only allows for greater yields but even can make fuel sources that were out of reach attainable. Careful consideration must be used in the application of technology to ensure there is a net benefit without loss of efficiency or added detrimental environmental impacts.
The compounding effects of data centers and their associated potential for computerizing industries should be considered necessary as the United States population continues to grow and the demand for energy either remains flat or grows relative to population and economic drivers. Increasing energy efficiency through application of computerized systems and data centers can produce compounding effects and provide overall societal benefits. Benefits would potentially include reduction in energy demands and increases in efficiency that would continue to follow Moore's Law like improvements over time in industried that have traditionally not been advanced through application of computerization.
Resources
- "U.S. Energy Data". http://energyliteracy.com/
- "Explaining All Of U.S. Energy Use With A Genius Chart And Cheeseburgers". 2016. Thinkprogress. https://thinkprogress.org/explaining-energy-use-with-genius-chart-cheeseburgers-6850dbd8dcf4#.9498ot85b
- "More On Efficiency Trends In Computing, From My Forthcoming Book". 2017. Jonathan G. Koomey, Ph.D.. http://www.koomey.com/post/14466436072
- " Processing & Refining Crude Oil | What We Do | About The Refinery ". Pascagoula.Chevron.Com. http://pascagoula.chevron.com/abouttherefinery/whatwedo/processingandrefining.aspx
- "United States Data Center Energy Usage Report". https://eta.lbl.gov/publications/united-states-data-center-energy-usag
- http://www.eia.gov/outlooks/aeo/pdf/0383(2017).pdf
- Fischer, Barry, Kathleen Davis, and Steven Yates. 2013. "America’S Energy Distribution: The Top 1% Of Homes Consume 4 Times More Electricity Than Average (And Why It Matters) - Opower". Opower. https://blog.opower.com/2013/03/americas-energy-distribution-the-top-1-of-homes-consume-4-times-more-electricity-than-average-and-why-it-matters/
"Power Usage Effectiveness". 2017. En.Wikipedia.Org. https://en.wikipedia.org/wiki/Power_usage_effectiveness