Written by Evan Goldstein
One of the topics presented at this year’s ASHRAE conference in Chicago was metrics for benchmarking data center HVAC system performance. Telecom providers are beginning to take notice of these metrics. For example, Verizon is now incorporating thermal efficiency metrics into its design standards.
The most popular metric is Power Usage Effectiveness (PUE), which is simply the total power consumed by the site divided by the power consumed only by IT equipment. A PUE of 3.0 would be considered extremely inefficient, and a PUE of 1.2 is about as good as it gets.
The two other critical measures are Thermal Conformance (the percentage of data center equipment observed to be at design temperature) and Thermal Resilience (similar to Thermal Conformance but with redundant systems shut down). These three metrics; efficiency, conformance, and resilience, can be traded off against one another but often gaining performance in one area is traded for a reduction in another. A common graphic representation of these tradeoffs is shown in the triangular image shown here.
Depending on the site owner, one field will take precedence over another. For instance, a cloud computing data center having one or two servers down due to overtemperature could be preferable to spending millions extra in cooling costs over the course of the year due to the inefficiency of having perfect temperature control.
For financial institutions, where a split-second delay can result in huge losses in a financial market, efficiency takes a back seat to conformance and resilience.
In the case of our small-cell clients, much of the equipment has a wider range of acceptable temperatures than highly sensitive computing equipment, and efficiency must be considered. This is best done by actually calculating loads rather than simply installing a 5-ton unit in every case (tons being a measure of cooling capacity).
It may not seem to be such a big deal to put a 5-ton unit in a room with a 3-ton load, a conservative but highly inefficient design. But in the aggregate, say looking over 100 projects, if we had not calculated our loads and simply thrown a 3-ton unit in every room, that would have resulted in 500 tons worth of equipment, a data center-sized capacity.
However, because we calculate our loads and size units properly, we will have installed 300 tons with confidence in our thermal conformance and can save the site owner substantial sums due to lower equipment costs and increased efficiency.