Data center infrastructures are at a tipping point. High-density servers provide more computing power with less energy consumption than traditional servers, but they require increased cooling. According to American Power Conversion Corporation (APC). The BlueGene systems put enormous demands on the cooling capacities of the data center which it is installed in. The airflow for a single BG/P rack is over 4500 CFM and well above the airflow required for the typical IT infrastructure. The system's air movers cannot be completely relied upon alone to guarantee the airflow demands of rack because the variable designs of the raised floor data center environment and thus the need exists to measure the total rack airflow, sometimes on ...
Page 1 of 8
Apparatus and Method for Measuring Volumetric Airflow Rates for a BlueGene Rack or Rack Assembly With Vertical Airflow
The nature of the airflow in the BlueGene rack system is that it is actually a function of both the speed of the fans in the rack, which produce a suction on the airflow, and the airflow supply capability of the raised floor in the room which the rack is installed. The design of the airflow requires that the raise floor be able to supply 4500 CFM per rack. Unless there is only one air supply to the raised floor and only one floor exit location, it is very difficult to determine the airflow through a rack system by measuring how much airflow is coming up out of the floor prior to installing the rack. With multiple rack installation, addition or removal of a rack will effect the distribution of airflow to other systems within the room. The air handling system operating settings can change depending on the number of racks installed which changes the airflow supply dynamics of the raise floor. For example, a given amount of air is supplied to cool an initial number of racks and the airflow from one 2' x 2' tile will be different from the reading from the same location when racks are added or removed from the room even if the overall air flow supply is ratio to the number of racks in the room. Thus it is not possible to reliably correlate airflow supplied from the raise floor to the rack airflow.
The rack airflow must be measured at the exhaust to determine throughput. The standard method of measuring airflow is to connect the system to a flow bench, a device with a metering element, such as a venturi meter or orifice plate. The largest flow bench available in Rochester has a 2000 CFM capacity, only half of the BGP's projected airflow rate. Vertical airflow racks cannot be attached to the flowbench chamber. Alternate, less accurate methods of measuring airflow rate include using hot wire anemometers, and differential pressure to estimate airflow rates.
The BlueGene program needed a repeatable, accurate method of measuring airflow.
A capture hood (flow measuring hood) is an instrument used to measure volumetric airflows from supply or exhaust and grilles from standard HVAC systems. The Alnor version of the capture hood is called a Balometer. Their Balometer is a capture hood uses thermal anemometry to measure volumetric flow. The Alnor version was used with a foamcore duct to measure airflow rates on BG/L. However, two Stortridge Balometers with 2500 CFM capacity each would be needed to measure the flow rate of a BGP rack.
The capture hood (red image below) is popular with contractors and facility personnel because of its lightweight construction and ability to document and store readings. However many capture hoods air have limited airflow volumetric ranges which can be measured accuracy or measured...
