Peter Matte Provides Insight on Achieving Better Airflow and More Efficient Cooling Within the Data Center

By Robert Lindsey

Peter Matte is a Senior Vice President with Ascent, a leading provider of solutions for the design, engineering, construction, and operation of data centers and mission-critical facilities. Matte will be a noted speaker at the Critical Facilities Summit, September 24th through 25th, at the Hilton Anatole in Dallas, Texas. He will be hosting a session, "Airflow: Theory vs. Practice," on September 24th. We recently talked with Matte about how data center facility professionals can achieve better, more efficient airflow and cooling in their mission-critical environments.

Q: What do you think are the differences between airflow theory and the actual methods used to set up airflow and cooling for IT server loads in data center facilities?
Peter Matte: In the past ten years, I've been out in the field, helping to implement various energy conservation projects in data centers. The main thing I've seen is that what we call "airflow theory" doesn't always work in practical, "real world" scenarios. A manufacturer may claim that their cooling unit will deliver a certain cubic-feet-per-minute (cfm) volume of air over a certain area, and promise that you will achieve a certain temperature, humidity, dewpoint, etc. But in practice, that doesn't mean the unit is going to deliver airflow to the servers in the way that you want it to. When you're designing a new data center, or planning out growth in your current data center, it helps to have some practical experience in behind-the-scenes engineering of airflow management and cooling systems.

Q: Can you give us an example of how airflow theory differs from actual airflow results in the data center?
Matte: It's usually a matter of inaccurate measurements of where the air is flowing, and how well the air is cooling your IT load. There are many "real world" examples, but here's a possible scenario:

Let's say an IT team wants to bring in 6 new blade servers, requiring 8 kWs per server, and they want to locate the server load in a certain quadrant in the data center. The Facilities Manager looks at the Building Management System and says, "Well, the static pressure under the floor is .07 inches of water, and we're delivering about 350-400 cfm. I think if you spread that server load out into three cabinets, in theory, that will service the IT load, and we'll be able to provide enough cooling for it."

But that assessment is often incredibly inaccurate, due to a multitude of reasons. For example, the area where you're installing the servers may not even be close to the area where the sensors for the BMS were registering the static pressure. You often see this in older data centers, where static pressure and CFM airflow are at a certain level in one area of the room, but are sometimes 2 to 3 times that amount in another area of the exact same room. It's usually because the centrifugal fans used in older cooling units are taking a "scattershot" approach to airflow. They're just throwing out random plumes of air, which causes pockets of negative air pressure and eddies to develop in the room.

Q: You've talked about taking a practical approach to thermodynamics. Can you explain how facilities managers should take such an approach?
Matte: You've got to use what I call "sneaker technology," which means you've got to put on your sneakers, walk around your data center, and take a series of airflow and cooling measurements above and underneath the floor before you implement any kind of new IT installation. You can't just rely on your BMS, DCIM, or other monitoring technologies to give you accurate measurements, because the airflow and temperature readings they provide may not be as accurate as you'd like.

There are some very inexpensive tools, such as an anemometer, that you can use to take measurements throughout your data center, and you need to measure temperatures, air pressure, etc., in several locations, not just one. After you take these measurements, you can implement various strategies, such as putting blanking panels and dams in certain places under the floor, to direct the cold air where you want it to go. In my upcoming presentation at CFS, I'll talk quite a bit about the measurement tools you can use, the places where you should take readings, and airflow management strategies.

The best part is, you don't have to hire any consultants to do this for you! You can take these measurements and implement airflow strategies to improve your PUE and reduce your TCO all by yourself. These are just practical things that you can do on your own, at minimal or no cost.

Q: What are the biggest takeaways that attendees will receive from attending your presentation, "Airflow: Theory vs. Practice," at the Critical Facilities Summit?
Matte: I think the biggest takeaways they will receive are:

• A practical approach to thermodynamics
• An understanding of data center math (e.g. true deliverable velocity, temperature, pressure compared to nominal load), and what these metrics and measurements will mean to their facility
• A cost-effective approach to achieving a target PUE
• An accurate methodology for determining appropriate cooling levels per IT load

To hear more from Peter Matte and other presenters, join us at the Critical Facilities Summit on September 24th through 25th at the Hilton Anatole in Dallas, Texas. For full agenda details, visit www.CriticalFacilitiesSummit.com..