Most data center operations need to know the amount of air flow from a raised floor tile or an overhead supply...
duct grill, and a flow hood is the best way to gauge it.
In a data center, the flow hood works on an overhead grill just as it does in any other room. The hood must fit the size of the duct for a proper seal. If the hood opening is bigger than the airflow source, air will leak out, invalidating the measurement. In that case, use a smaller hood that fits the grill, or a mask to simulate an extended duct surface (See Figure 1).
Raised access panels are different, and a common feature in data center designs. The under-floor pressure in a server room is low compared with the pressure in air supply ducts. Placing the hood over a floor tile creates back pressure that reduces airflow. If you use a standard hood to measure data center airflow from the floor, the recorded value will be lower than the actual flow when the tile vents directly into the room. You can calculate the corrections, but they differ for each tile and hood type, and require complex equations. Instead, obtain a hood designed for raised floor usage that incorporates electronic measurement and correction (see Figure 2). This type of hood can be used for either floor tiles or duct grills.
Handheld anemometers, a way to measure air speed, can be useful in knowledgeable hands. For data center airflow readings, use adapters that act like air hoods. Results from a handheld anemometer must go through even more complex conversions and corrections than flow hood measurements do. In the data center, anemometers take velocity and temperature and humidity measurements in the air stream well, but are not very useful for determining airflow volumes.
You might also measure air conditioner discharge and return air, under-floor or in-duct velocity or static pressure, air intake to or discharge from a piece of computing equipment, or ventilation air from outside. These measurements are all fairly difficult and rarely useful, unless the data center experiences a problem indicated by the common airflow measurements and needs engineering help to diagnose it.
Related Q&A from Robert McFarlane
Our latest firewall/VPN firmware upgrade left CPU usage at 100%. A malfunctioning DHCP-Server means people aren't getting IPs. I have to pull the ... Continue Reading
Do battery cooling cabinets save money over cooling batteries within the whole data center? Continue Reading
We're setting up a 3,700-square-foot server area with 175 server racks and in-row cooling. It's a greenfield project. How do we estimate power ... Continue Reading
Have a question for an expert?
Please add a title for your question
Get answers from a TechTarget expert on whatever's puzzling you.