Metrics and Economizer Mode Data Center Cooling

This audio was created using Microsoft Azure Speech Services

So far in this series about economizer-based cooling strategies, I’ve looked at when not to use economizer mode for data center cooling and the cost benefit considerations of economizer mode. Now we turn our attention to the other metrics involved. To recap, these are: annual water consumption, total cooling energy, annual hours on full economizer mode, annual hours on partial economizer mode and annualized power usage effectiveness (PUE).

In terms of annual water consumption, economizer modes used with cooling towers are subject to the most water consumption compared to some other types of economizer modes due to the evaporation of water in the cooling tower. This is because the evaporation process occurs continuously all year round. The evaporative assist component of the other economizer modes consumes water to a much lesser extent since it only uses the evaporative assist process during the hotter periods of the year.

Therefore at the lower extreme, air conditioner bypass with evaporative assist via direct fresh air (based on analysis of a 1MW data center at 50% IT load, located in St Louis, MO) will have an annual water consumption of 1000 liters, while at the other extreme, water economizer using chiller bypass via heat exchanger will consume a whopping 26 million liters of water. This is not to say that evaporative assist is a panacea for all ills because air conditioner bypass via air heat exchanger with evaporative assist comes in second highest, consuming almost five million liters of water annually.

Evaporative assist is most effective in dry climates such as Las Vegas and Dubai. The cost of an evaporative cooler must be balanced against its effectiveness in climates that are more humid. It is possible to spend more on evaporative cooling than is saved on cooling system energy.

In terms of total cooling energy, the economizer mode with the largest energy cost is the CRAC compressor bypass via second coil. This is due mainly to the energy penalty of having distributed cooling systems. The economizer mode with the lowest energy consumption is the air conditioner bypass via air heat exchanger. The air conditioner bypass via air heat wheel consumes only slightly more energy.

Turning to annual hours on full economizer mode, the chiller bypass via heat exchanger economizer mode provides the least at 4,705 hours while the air conditioner bypass via air heat exchanger economizer mode provides 7,074 full economizer mode hours.

Partial economizer mode operation occurs when the economizer mode alone is unable to cool the data center due to the outdoor conditions and requires assistance from the compressor. This is a very important attribute for any economizer mode since very few locations around the world allow for operating 100% of the time in economizer mode. In some locations, partial economizer mode hours are much greater than full economizer mode hours, resulting in more energy savings derived from partial operation. The air conditioner bypass via direct fresh air economizer mode provides zero partial economizer mode hours due to the humidity conditions required in the data center.

Finally, power usage effectiveness (PUE) is the proportion of total energy used for the entire data center compared to the total energy used by the IT equipment. This annualized estimate is based on a common power infrastructure. The CRAC compressor bypass via second coil economizer mode provides the highest, or worst, annual PUE of 1.39. The air conditioner bypass via air heat exchanger economizer mode provides the lowest PUE of 1.25, in use at Schneider Electric’s St. Louis data center.

If you’d like more information, John Niemann, John Bean and Victor Avelar’s excellent white paper “Economizer Modes of Data Center Cooling” is well worth a read and available as a free download from the APC website.

Tags: , , ,