The Hurt Factor: A Different Perspective of Free Cooling, Paid Cooling, & Local Electric Utility Rates for Data Centers

Powered by Tableau

The opposite of Free cooling is technically "Paid" cooling. The map above shows how much it hurts - in US Dollars/kW - for "Paid Cooling" for various locations in the US using local electric utility rates. 

So here's how this effort came about: a couple of weeks ago, I was interested to know how many hours of free cooling can be extracted by data centers based on geographic locations. I had gathered data on zip codes for major concentrations of data centers around the continental US and I ran the Air Side free cooling calculations for each zip code using The Green Grid's free cooling calculator. I used a modest 72 F dry bulb and 52 F dew point as set points (yes .... you can go higher and you should).

I decided to try something new. I subtracted the number of free cooling hours from the total number of hours per year (~8760). This gave me the "Paid" hours of cooling- as one would have to spend energy to operate the mechanical compression systems in order to maintain temperature setpoints. 

But there was still something missing. I still wanted to know what the "hurt" factor is for operating a data center in an environment that doesn't make use of free cooling. To be able to answer that, one has to look at the local electric utility rates. Some utilities in northern locations charge more than utilities in the deep south. So while the southern data centers may not benefit from free cooling, they are getting a excellent break in electric utility rates.  Therefore the product of paid cooling hours and the local utility rates should give a decent idea of how much it "hurts" to operate the data center in non-desirable geographic locations and non-desirable electric utility rate jurisdictions. 

I gathered information on the local electric utility rates for each zip code. For this exercise, I used the Industrial rates in $/kWh as provided in the rate structure of each utility. Those that I could not find, I used the Department of Energy's 2012 rate structure. Once I had the paid cooling hours and the electric utility rates, I multiplied the two numbers to give me $/kW and I charted the results using Tableau 7.0 software.


Some of the observations that caught my attention:

• The least hurt factors occur in the Pacific Northwest and the Rocky Mountain States. Hence it makes sense why major data centers are concentrating in those States.
• California and Florida have approximately the same hurt factor. “Ouch”! Considering California benefits from so many hours of free cooling.
• Texas ranks a lot better than either California or Florida because their utilities rates are very competitive.
• The Mid-Atlantic States suffer from higher rates and "less" free cooling hours.
• Virginia benefits from very competitive utility rates and hence evens-out the less number of free cooling hours (compared with northern latitudes).

Some readers asked questions about specific locations, which I thought were very worthy of mentioning. So I'm adding images of a few more locations in New Jersey where rates are cheaper than New York. I'm also adding maps with more detail of the NY/NJ and California areas. 

Learn About Tableau

Learn About Tableau

I could go on with observations but in summary, I like the results that I charted. I think it's a unique way to look at data and I plan to fine tune it with more accurate electric utility rates and higher temperature setpoints. I also want to look at locking out humidification and investigate water side cooling in the settings in the Green Grid's free cooling calculator-something I did not have the luxury to modify during this exercise.  email if you would like me to share the utility rates separately with you.