Data Center Cooling – What You Need to Watch Out For
Data center operators have become significantly more knowledgeable about Data Center Cooling over the past decade, but many installations still have problems that either inhibit them from running at maximum capacity or cause them to waste energy.
The ultimate goal in data center cooling is to have increased control over cooling temperature set points for IT air intake, while using a lower volume of air to be delivered to the data hall.
Here is a list of the common issues thermal management experts observe when visiting data centers:
This occurs when cold air leaks from the plenum under the raised floor into nearby spaces or into support columns. They are somewhat common and can result in everything from losing pressure in the chamber to letting in air that is warm, humid or unfiltered. To mitigate the risk of phantom leakage occurring, inspect the surroundings and support column for holes and make sure they are all sealed.
Placement of perforated tiles is often done wrong. They should not be used in hot aisles or whitespace areas, as this wastes cooling capacity. It is also important to limit the number used on the intake side of the racks, as too many can create problems. One symptom of mal-located perforated tiles is lower-than-usual air temperature at the top of the IT racks.
Raised Floor Openings Left Unsealed
Any holes that are left unsealed from the building process will result in loss of cooling, meaning wasted resources. Therefore it is important to check under electrical gear for unsealed openings.
Poor Rack Sealing
Sealing empty spaces is something that should be common sense in data center cooling, but is often overlooked. An operator who cares about efficiency will put blanking panels in empty rack spaces and should also seal any openings on the side of or under cabinets.
Improperly Calibrated Temperature and Humidity Sensors
Some sensors may need to be calibrated after shipment, and all sensors may need to be recalibrated over time. Incorrect sensors result in poor cooling management and units that work against each other. Operators should check climate control sensors periodically (at least every six months) and recalibrate if necessary.
CRACs Fighting for Humidity Control
Returning air to adjacent CRACs at different temperatures is another way to cause two units to work against each other. This results in different humidity measurements, meaning that one unit will be humidifying the air and the other will be dehumidifying. Resolving this issue takes an understanding of the Psychrometric chart and thoughtful setting of humidity control points.
When more cooling is running than necessary without a way to keep redundant units off, too many units end up running in a low-efficiency state. This tends to occur when some racks are difficult to keep cool while underfloor cooling temperature is high. Most operators respond to this by bringing additional cooling units online, but it is better to actually run fewer CRACs at higher capacity.
Empty Cabinet Spaces
Empty cabinet spaces skew airflow balance and either recirculate exhaust into the cold aisle or take away cold air from where it is needed. This causes the system to overcool to compensate
Poor Rack Layout
Optimal layout makes use of long hot-aisle/cold-aisle configuration, in which main CRACs are placed at the end of each row.
Lack of Cooling Management
Improvements in a system’s cooling management result in minimized stranded capacity and lower operating cost. Simple things like blanking panels can add up to savings over time but are often overlooked. It is even possible to defer expansions or new builds by managing cooling properly.