A better way to exhibit the gains achievable through energy efficiency might be to illustrate savings made possible by specific technologies

In Response To "Data Centers are Adopting Green Initiatives but are Wary of Vendors' Marketing Messages"
Dr. Sidd Mukherjee
March 19, 2008

The article surveys green initiatives in datacenters. Some important points are made:

a) almost a fifth of those organizations with green initiatives do not include the data center
b) of those who did, 44% suggested cooling a center to cut energy use
c) most respondents were unconvinced of vendors claims of environmental friendliness.

This is disconcerting, and suggests that vendors and clients are not communicating well.

A better way to exhibit the gains achievable through energy efficiency might be to illustrate savings made possible by specific technologies.

A substantial share (30% and increasing) of the power used in a modern datacenter is used for cooling. Electricity rates are increasing throughout the world, and it is clear that carbon taxes will hasten the rise in the near future. There are two components to electricity tariffs, the Kilowatt-Hour (KWH)charge, and the demand charge, measured by the peak demand over a certain time period in KiloWatts (KW). The first is the total energy used, while the second is based on the maximum rate of energy use. For an illustration of these tariffs, one may visit the ConEd site in Ref [1]

The Kilowatt-Hour charge is based on the total energy consumed during the billing period, and is obviously reduced by more efficient AC. A more efficient air conditioner will eat fewer KWH. But the second demand charge is not as easy to minimize. This is due to time sensitive pricing, which makes electricity during peak hours much more (as much as a factor of two) more expensive than during off peak hours. At the same time, the power drawn by any air conditioner will increase with the external temperature, thus in the afternoon (when the demand tariffs are the highest) the air conditioning will draw the most electricity. This is the part that is not addressed by merely improving efficiency.

This variable pricing gives rise to opportunities for arbitrage. One such technology is that used in Thermal Storage Systems. These systems use cheaper off peak power to store ice, chilled water, or a suitable phase change material such as Glauber's salt at night and use it up during the day to meet cooling needs. Another beneficial feature is that the chillers and ancillary equipment need not be sized for peak load, since a large fraction of the cooling during high demand periods is delivered by the previously stored ice. This reduces the initial cost, and thus any associated financing charges.

The leading candidates for Thermal Storage media are chilled water and ice. Although phase change materials promise better specific storage capacity, the technology of phase change materials is not as mature as the water or ice based systems. Chilled water based system require larger storage tanks but achieve a better chiller efficiency and lower chiller cost than ice based systems. Ice storage systems can be retrofitted to some existing installations.

Thermal Storage Systems may be designed for full or part storage. Full storage systems displace all of the peak demand by generating a sufficient quantity of chilled medium during off peak hours to meet all of the cooling load during peak hours. The system pulls no power for the chiller during peak hours. Partial storage systems do not eliminate chiller load but they reduce it substantially during peak, meeting a part of cooling demand with previously chilled medium.

Substantial savings are possible as may be seen in the references below ,which document a reduction in demand by 3MW and cost savings of almost a quarter of a million US$ per year at the Dallas Veterans Affairs Medical Center. The cost savings pays for the project in seven years at current energy prices. Given the projected rise in electric costs, the payback period will probably decrease. Several other projects are detailed in the references as well.

References

Back to the Data Center Study