I continue to see an on-going debate about defining PUE, including comparisons of one data center to another. What is particularly interesting are PUE comparisons that present PUE numbers accurate to the second significant digit. Also, the numbers are typically presented out of context, without crucial information that is absolutely required to accurately calculate PUE. Here are just a few of those data points required to accurately report PUE:
Duration of measurement period
Climate in which the facility is located. This is important for determining:
power use of vapor compression cooling equipment
solar loads
building envelope heat transmission
outside air dry-bulb and wet-build temperatures
Cooling system topology (e.g., air- or water-cooled)
Cooling equipment COP, EER, and part-load efficiency profile
Electrical system topology
UPS part-load efficiency profile
Building envelope attributes (wall and roof area, insulation values, heat capacities of components, moisture permeability, etc.)
Amount of outside air for ventilation and building pressurization
Cooling supply air temperatures and temperature rise across equipment
etc.
Unfortunately, there is not a simple "point and shoot" method of reporting PUE. Comparing PUEs of different data centers is even more difficult. The fact that most people are missing is that estimating, measuring or comparing PUE must be done looking from the facility perspective, since all of the non-technology power usage comes from the components in the facility.
The good news is that this type of energy estimation method has been around for years and has well-established guidelines. One of these standards that is used in the U.S. (and in many other countries) is the ASHRAE 90.1 energy standard. While not specifically written for data center facilities, it contains all of the methodology for establishing a baseline minimum energy performance that is easily applied to data centers. (Any data center facility going after a LEED certification and even many local building codes require reporting the energy use of the facility using the ASHRAE standard - so the use of this standard is common).
So why am I touting this guideline as the "gold standard" in determining and comparing PUE? Here's why:
- It has a rigorous process of validation that has on-going maintenance with public input.
- There are very specific rules and a standard approach.
- There is also a way to level-set building energy usage based on specific building type and size, cooling infrastructure, equipment and occupancy schedules and climate.
- When a building is compared to the minimum allowable standard, it is already normalized so it can be compared to other buildings.
Like many other industry standards, it is consensus built so certainly it is not perfect, but it is very thorough and defensible.
So next time someone makes a claim of a certain PUE value, make sure you get the details before it is used in a comparison.
