Is PUE for me? Top 3 System Selection Considerations for PUE

What You Need to Know about Data Center Cooling and PUE

Power usage effectiveness (PUE) is a metric used to determine the energy efficiency of a data center. But what should you base your system selection on to meet your thermal management goals in the first place?

Junk In, Junk Out

There is a saying among data analysts, “junk in, junk out”, which is an expression used to convey the relationship between input and output. This is a truth in data analysis, and system selection in data centers is no exception.  How do we define “junk”? What does your organization consider to be worthy of decision making? Now to really shine the spotlight – does your organization unknowingly consider “junk” worthy of decision making? The “junk in, junk out” axiom is not nearly as simple as it seems on the surface.  Three considerations stick out when trying to avoid “Junk in, junk out” in system selection for a data center: site location, site operation, and organizational decision making.

#1 Data Center Site LocationDense IOT City - Thermal Management Trends

Site location has a tremendous impact on system selection.  A data center in Los Angeles, CA will be able to take advantage of wet bulb temperature in a way a data center in Atlanta, GA would never be able to. Conversely, a data center in Minneapolis, MN will be able to take advantage of dry bulb temperature in a way a data center in Dallas, TX would never be able to.  So, when comparing solutions from various system providers, be sure the energy use information you are being provided is targeted as close as possible to the climate of your site location.  I always break down performance into increments aligned to ASHRAE bin data to avoid “junk out” situations with energy analysis. For example, a solution provider may indicate free-cooling capability at 55°F dry bulb, but this data is worthless without knowing how many hours a year the area is below 55°F.   In this instance, your PUE calculations will be skewed.

Water usage is another point of consideration, as it may have a cost or perceived sustainability impact in your area. Depending on the type of power plant the additional energy used on site to avoid water usage directly impacts the total amount of water used at the power plant. Cooling towers, with evaporative and adiabatic solutions, can use water much more efficiently than a power plant. This means your system type utilizing water may have a lower NET water use than another system type.

#2 Energy Usage at Data Center Site

Equally important is site operation.  When looking at energy data, the hotter the return air the better. What temperature will your system operate at? Will it be the 95°F assumed in the data analysis you were provided? If not, we have a “junk in” situation.  Certain free-cooling methods rely on a temperature differential between indoor and outdoor temperatures to function. If your data assumes a 95°F return but your room operates at 85°F, it will have a dramatic impact on the actual performance in your facility.

#3 Data Center Costs & Decision Making

Last, but not least, is organizational decision making.  Does your organization look to minimize upfront cost? Life cycle cost? Payback? Very rarely will a solution minimize all three of these. Financials don’t escape “junk in, junk out” either. Upfront costs and life cycle costs require scope leveling. Apples to apples operational costs can be very challenging to nail down and making bad assumptions can cause a sub-optimal decision to be made, which will affect your PUE.  Energy costs can be calculated fairly easily if site-specific weather data and actual site operation data are used.

Maintenance has a few more wrinkles, however, as there are many factors to consider. Will you be locked into a service provider based on your system selection? What are the differences in preventative maintenance requirements? What’s the impact of equipment reliability on system uptime? The list goes on and on.

There is no one size fits all solution for comparing system types. Every site is different and every organization manages decisions differently. Instead of viewing it as a bottleneck, view it as an opportunity to make sure the right decision is made for each specific site, which will improve organization-wide efficiency much more than a top down roll-out.

Through a consultative approach, Data Aire is able to provide site and application-specific environmental solutions to help you meet your power usage goals.  To learn more about the unique applications Data Aire supports, please watch this video.