Experimental Characterization of an Energy Efficient Chiller-less Data Center Test Facility with Warm Water Cooled Servers (SEMI-THERM 2012)

Overview

The paper examines a chiller-less test data centre built under a DOE-funded project. Typical data centres were taken to spend about 50% of IT energy on cooling. In the architecture, servers are cooled by a combination of warm-water cooled hardware and recirculated air, with heat rejected through a closed secondary liquid loop to an ambient-air dry cooler and no chiller. Equipment and whole-facility performance were characterised. This structure removes the need for chiller compressors and thereby takes out the main item of cooling energy. The dry cooler's performance depends on outdoor air temperature.

In a run of more than 22 hours, cooling energy averaged 3.5% of IT energy. The run conditions were an average ambient of 23.8 C and an average IT power of 13.14 kW. These figures indicate that in a liquid-cooling topology based on a dry cooler, the cooling load can fall to a few percent. The test conditions are stated alongside the result, which allows the figure to be read as a benchmark for similar conditions.

Because the source is an abstract-level record, method details, measurement uncertainty and results in other climates are not stated here. The results should therefore be treated as a benchmark valid for this single test facility and the stated conditions. The paper's authors are IBM, and the work was presented at a SEMI-THERM symposium; the results belong to a test facility. The record rests only on the abstract, so the full paper has not been assessed.

Key points

1. Typical data centres were taken to spend about 50% of IT energy on cooling.
2. Architecture: warm-water cooled servers plus recirculated air; closed secondary liquid loop to an ambient-air dry cooler; no chiller.
3. A 22+ hour run averaged cooling energy of 3.5% of IT energy.
4. Run conditions: average ambient 23.8 C and average IT power 13.14 kW.
5. The work was done at a test facility built under a DOE-funded project.

Using it in design

• Support in the energy model a dry-cooler-only liquid-cooling topology where cooling overhead is a few percent of IT load.
• Use this topology as a benchmark when users compare chiller-based designs.
• Show the benchmark figures with the note that they come from a single test facility and the stated conditions.

Who should read it

Mechanical and energy engineers evaluating warm-water cooling and chiller-less architectures.

Related standards

• ASHRAE W-classes

Publication details

Authors

Milnes P. David; Madhusudan Iyengar; Pritish Parida; Robert Simons; Mark Schultz; Michael Gaynes; Roger Schmidt; Timothy Chainer

Publisher

SEMI-THERM

Year

2012

Document number

SEMI-THERM 2012

Event

28th SEMI-THERM Symposium (2012)

Document language

English

Source type

Industry body

Access

Citation only

Licence

Copyright retained; citation only — the publisher’s terms apply to the original document

Citation