Liquid-cooling has enthralled the data center industry, and rightly so!
Liquid-cooling is highly efficient and enables continual scalability as digital infrastructure demands grow.
However, what we don’t hear about as much is that most liquid-cooled data centers are actually hybrid cooled.
What is hybrid cooling?
Hybrid cooling combines both liquid and air-cooling systems within one data center.
Hybrid cooling leverages the strengths of both technologies, creating a customised solution tailored to your specific requirements and limitations.
Why is hybrid cooling required?
Residual heat from liquid-cooled racks
Liquid-cooled workloads produce a small amount of residual heat that requires an air-cooled load or heat repurpose. Liquid-cooling systems typically have less residual heat than entirely air-cooled systems, in particular immersion cooling. However, air-cooling or heat repurpose is still required.
Lower to medium density racks
Not all rack densities require liquid-cooling.
Data centers often have data rows of varying densities. For example, some will be made up of high power HPC racks completing complex tasks that require direct-to-chip cooling, while others will be <10kW handling networking, storage and other more basic functions.
For lower to medium density racks, air-cooled systems are a simpler, lower risk and a lower cost approach.
What are the benefits of hybrid cooling?
Optimal efficiency and lower CAPEX
Using indirect evaporative cooling units as perimeter airside cooling removes the requirement for chillers, meaning that the primary cooling loop for liquid-cooling can operate at higher temperatures, giving a lot more flexibility to the heat rejection system selection and increasing the opportunities for heat recovery. Not only does this minimise mechanical cooling, it may then be possible to handle all the liquid-cooled heat rejection via cooling towers or adiabatic dry coolers, significantly improving the operational efficiency of the data center. Carefully considering the heat rejection system for the airside cooling allows for radically reduced CAPEX, reducing the amount of chillers, pumping systems and other supporting infrastructure on site.
Increased flexibility and scalability
For full flexibility between air-cooled loads and liquid-cooled loads, a common heat rejection system, such as air-cooled chillers, with sufficient ‘tap-off’ points for fan wall units and CDUs can be considered. Should the day one load be fully air-cooled, CDUs can be introduced at a later stage so long as the TCS loop has been installed ahead of time to lessen the disruption to data center operations. This provides a smoother transition into a more liquid-cooled system, whilst also allowing for future scalability as rack densities increase.
