More than any other industry, electric supply is the most important attribute necessary for siting and operating a data center or mission-critical facility. The following article takes a look at how utilities can affect the location and operation of data centers by understanding specific power requirements and costs, and reviews incentives offered by utilities in each state for data centers.
Simply stated, data centers require a lot of power. That said, there is no “typical load.” Rather, a small data center, especially one housed within a standard office operation, could require less than 1 megawatt, while some of the larger data centers easily exceed 100 megawatts. The greatest amount of market activity is 5–20 megawatts.
From a utility perspective, data centers have desirable load factors. Many data center operators will plan for a 90 percent load factor, although utility professionals frequently report that the actual number is more typically 80–85 percent. That said, BLS/Sugarloaf have clients that have achieved over 95 percent. The voltage characteristics will be dependent on the utility’s policy, and, in most cases, are determined by the requested capacity and available service in the area.
Most large facilities (over 10 megawatts) will be fed via the utility’s transmission or subtransmission systems. Service at this level will then require the data center operator to design, construct, and maintain a substation to provide distribution voltage (13.2 kilovolts) to the facility.
Due to the cost of a substation ($3–$7 million), users will seek opportunities to convert this cost from a capital cost to an operation cost, with the utility or a third party providing the substation.
Power is the lifeline of the data center. Without power, the data center ceases to function. Employees and customers can’t access information. Transactions stop. Financial trades aren’t possible. Planes stop flying. Health records can’t be accessed. System interruptions are measured by the cost per second of downtime. In financial service operations, these costs can be millions of dollars per second, with downtime having a tremendous impact on customer satisfaction as well as creating embarrassing media coverage, as we have seen recently with the New York Stock Exchange and United Airlines.
This impact is why many data centers are considered “mission-critical.” Keeping the data centers up and running is job number one for the CIO and site manager.
Careers end when data centers go down. For data center operations the key measure is “uptime,” with success measured in “9s” (the percentage of time the data center is available is indicated by the number of nines; for example, 99.998 percent of the time would be considered “four 9s”). New data centers being built today are generally targeting three 9s or better, which works out to be less than nine hours of total downtime per year.
Adequate and reliable power is necessary to avoid such downtime. For an electric utility, data centers can be challenging customers. Data center operators are very demanding, requiring information on power quality factors, outage frequencies, causes, and what has been or will be done to avoid future events. Electric utilities measure uptime or reliability using three indexes:
Power represents 60–70 percent of the total operational cost (TOC) of a data center. If able, a user will chase lower-cost power to reduce the overall operating cost of the facility. This effort, coupled with optimizing the data center design for energy conservation, reduces the TOC, as well as the facility’s carbon footprint. As a result, data center operators are beginning to consider alternative sources of energy to reduce cost. These include renewables, fuel cells, and cogeneration. Enterprise users, which typically have fewer business limitations on where they can locate, will make power cost one of the most heavily weighted selection factors.
Many nonfinancial services enterprise users will decline to consider a location where the average cost will be above $0.07 per kilowatt-hour. Often, a rate less than $0.05 per kilowatt-hour is the target, and multiple data center operations have been able to secure prices below $0.04, particularly in areas using hydro generation. Colocation providers, on the other hand, tend to locate in close proximity to their customers and along robust fiber routes.
Thus, we see a lot of data center activity in large metropolitan markets such as New York/New Jersey, Chicago, and Dallas. In the last few years, secondary markets have seen strong growth in data center development, including Columbus, Ohio, and Indianapolis, Indiana. Colocation providers are less sensitive to national price differentials, but they work hard to ensure the lowest cost possible within their target geography. However, the price per kilowatt-hour is not the only material cost to be considered.
Due to the large infrastructure requirements of a data center, another critical factor is the total cost required to provide electric service, and whether any of those costs are offset by credits or refunds via the utility’s tariffs or service policies. It is necessary to analyze tariffs and rates for any hidden costs that may make a region less competitive, such as duplicate service cost, demand ratchets, revenue credit, security deposits, and start-up costs that negatively impact data center operations.
These costs can have a significant impact on the financial analysis. The greater the cost borne by the customer, the greater the chance the customer will choose an alternate location.
See below for PDF of full article.