verizon-svlg

Verizon Tries Out Alternative Cooling Methods

5 comments

verizon-svlg

A diagram of a new airflow pattern being tested by Verizon in an effort to improve its cooling efficiency. (Image: Verizon)

PALO ALTO, Calif. - Verizon Terremark is experimenting with a novel way to push hot and cold air around data center racks. Ben Stewart, senior vice president of facility engineering at the company, talked up the concept Tuesday at the 2013 Data Center Efficiency Summit.

Stewart calls the new approach virtual containment. Think of the concept as a new and potentially energy-saving way to push hot and cold air among racks in a data center.

Instead of having a downflow computer-room air handler (CRAH) on a raised floor running at a set speed and possibly not delivering enough cool air to a rack – or too much of it – Verizon is interested in finding a happy medium. Stewart opted for the term “Goldilocks zone” in his explanation.

“We can actually tune it to deliver exactly the right amount of cold air to this aisle and to this cabinet,” he said.

Researchers tried putting temperature sensors in cold aisles, and whenever it gets too cold, an alarm went off and an employee adjusted the floor tiles to deliver just the right temperature. But such a manual process might not scale well. The company developed a system to “dynamically balance the floor to achieve the Goldilocks zone, without the guy having to come out there,” Stewart said, resulting in savings of over $1 million a year with an eight-month payback period.

Then the researchers looked at adding barriers to form hot and cold aisles, but the result was too complicated. The system in tests now is simpler, with cold air being pushed over several adjacent aisles and hot air being pulled down through the cabinets and sent back to the CRAH underneath the cabinets.

“All we have to worry about is cold air sneaking back down there,” Stewart said. “It’s something we’re playing with right now, just kind of on the drawing board.”

The experiment currently spans around 2,000 square feet, Stewart said after his presentation.

Broader Focus on Efficiency and Sustainability

The airflow experiments are part of a broader focus on data center energy management at Verizon. The company already lays claim to six solar power sites that generate 5.4 megawatts of electricity, and 12 fuel-cell sites that make 9.6 megawatts, according to figures Stewart shared in his presentation. The figures put the company ahead of some other companies in the Fortune 500.

But increasing reliance on green energy sources isn’t the only way to reduce environmental impact. Energy efficiency can also move the needle – thus Verizon’s recent work in the area.

The company has begun working more with free cooling, and its Terremark data center unit is a long-time proponent of using flywheels for uninterruptible power systems.

“Flywheels are said to be more energy efficient” than batteries in a data center’s uninterruptible power supply, Stewart said. “Where you really get your energy savings is you don’t have batteries. … We don’t have to keep the flywheels at 77 degrees. We don’t have to maintain specific humidities to maintain the flywheel.”

Implementing free cooling with outside air at data centers could be another way for Verizon to consume less energy in its operations.

“As the market starts to accept that we can raise temperatures in the data center -and as we all know, we certainly can – then we have more sites where we can actually use free cooling,” Stewart said.

About the Author

Jordan Novet is a freelance writer based in San Francisco. He has written for newspapers and magazines and most recently covered cloud computing and big data at GigaOM.

Add Your Comments

  • (will not be published)

5 Comments

  1. Wondering why the heat isn't vented up since hot air rises... I get the cold air sinking from above into the environment but it still seems like there is a fair amount of electricity being used to rework physics if hot air is forced to go down. I always thought (still do) that it was never an issue of handling cold air, it was how you deal with heat rejection since computers turn electricity into heat. NetApp had an interesting design that I never saw catch on broadly but is still my personal 'coolness' benchmark for a facility level or entire data room 'containment' system. It focused on segregation of air type by temperature and physics and made the data room the plenum and put cold air just on on the rack fronts. The tour was pretty toasty, but the design was simple and slick at the same time.

  2. PGT

    The fundamental difference with TMRK and NetApp facilities would be that NetApp retains 100% control over what goes in and how it goes in whereas Terremark is a multi-use facility where they've got to accommodate a multitude of variables as a shared-use facility. I've been in both the NetApp facility in RTP and Terremark's east coast facilities.

  3. MJ

    Could someone just remind me does hot air rise or fall ?? Oh yeah sorry. Why would you add more energy to an inefficient process by sucking the air down ? How is that more efficient than having the cold air 'pulled up' by the rising heat and then having a chimney to take the heat away ? Its more fans waiting to go wrong and that need servicing. If you want a plenum put it above and even just chimney above the racks and leave them topless without containment at the ends, but this is just adding more components to go wrong compared to letting physics do it stuff. Do they have cables under the floor ? How do people work there ? Or do they just let hot air rise as it does when they lift tiles ? And having guys manually reset floor vents in real time. Really ??? You can see how a million dollars was saved when they stopped it. The guy would never be right for a dynamic load.

  4. Mark, Maybe they are making use of fluid dynamics, in concrete the Venturi Effect that it produced by running fluid (hot Air) under the raised floor by towards de CRAH creates a fall in static pressure at each floor opening to maintain total pressure of the system constant that coupled with the fact that hot air has to start with a lower static pressure than cold air due to its lower density, it should be able to create enough suction to keep ambient pressure the cool air flowing as depicted.

  5. I too am wondering how pushing air against its natural inclination saves energy. If there is something I 'am missing I would like to know more.