By Michael Young, Pr. Eng.
Part 7: Dealing with part load operations within a data centre: Things to consider when using a DX system.
 |
Michael Young is a trainer, coach and a pre-sales engineer in the HVAC industry. He graduated from the University of the Witwatersrand in the field of Mechanical Engineering (B.Sc Mech Eng) in 2008 and qualified as a Professional Engineer (Pr.Eng) in 2013. Michael is passionate about promoting knowledge and helping other young engineers grow within the industry through his training workshops and coaching sessions. |
When cooling units are required to operate lower than 25% of the design load, there is a chance that the unit may display a warning or even shut down on a critical alarm.
To prevent this from occurring, two actions can be taken:
The first action is to increase the heat load within the data centre by installing heater banks within the server rack. This option is often not viable as it can waste electrical energy while trying to prevent the cooling unit from triggering any alarms.
The second action that can be taken is to reach out to the manufacturer of the unit and see which parameters can be adjusted.
Often the requirement is to satisfy the supply air condition and in many cooling units, this setpoint is what controls the operations of the compressor. Since the load within the data centre is much lower than the design condition, either the airflow or the return air condition of the cooling unit must be adjusted to meet the lower cooling requirement.
So, let’s consider the two options that are available:
For option 1, we can operate the fans at a higher speed, and we can lower the return air condition to the unit to satisfy the required supply air temperature. This option will result in an increase in electrical consumption of the unit as well as a higher noise generation within the white space.
Option 2 is to reduce the airflow of the unit and increase the return air condition setpoint. This option will decrease the electrical consumption of the unit as well as noise levels.
While option 1 seems the most viable, the final decision requires an analysis of the entire system.
This situation once happened to me where the data centre was operating at low loads and the units kept triggering alarm warnings. My first initial thought was to choose option 2 whereby we decrease the airflow and increase the return air condition. However, in this situation, the servers were not populated in a uniform manner and decreasing the airflow of the units created hot spots within the data centre.
I also learnt that decreasing the airflow would result in low superheat alarms as insufficient heat transfer would occur over the coil, resulting in a much ‘cooler’ refrigerant that leaves the coil.
I had no choice but to employ option 2 which was to increase the airflow. The tricky part was to determine how much to increase the airflow to. So, I conducted a site audit where I measured the approximate airflow over each server, and this resulted in the approximate airflow for each rack.
I then consulted with the manufacturer and asked them to perform a unit analysis to predict what the return air condition should be at my measured airflow requirements. I also asked for the supplier to confirm if the modified return air condition and airflow rates would prevent the triggering of the ‘low super heat alarm’.
The good news was that I was able to use the supplier’s unit performance rating to readjust the unit’s setpoints which prevents the occurrence of alarms.
So, when faced with low load operations, an analysis of the entire data centre must be performed whereby the development of hot spots must be addressed, the return air condition must be selected, and operations of the compressor and expansion valve must be understood to ensure the units still operate in the correct manner. Wishing you a successful month ahead.
