vertiv, formerly emerson power network, has announced that it has developed liebert icom autotuning, which is machine learning software that will enhance the reliability, efficiency, and lifespan of data center cooling systems.

vertiv is the latest company to use machine learning to autmoate data center cooling.
(wikimedia commons)

autotuning is a new feature of vertiv’s liebert icom-s thermal management control and will be available for liebert cooling systems throughout north america, after its unveiling at the air conditioning, heating, refrigeration exposition (ahr expo) in las vegas at the end of january.

the purpose of the control is to use machine learning and advanced algorithms to regulate liebert direct expansion (dx) compressors or chilled water valves to balance and stabilize system components such as chillers, crac units, and others. instead of manually tuning each individual component, the liebert icom autotuning will bring all the systems under one control.

according to the press release, the autotuning feature will harmonize “compressors, fans and condensers to eliminate costly short cycling” and for chilled water systems it will eliminate “rapid fluctuations in valve positions, called ‘valve hunting,’ to balance fan speeds, water temperature and flow rates.”

“the introduction of liebert icom autotuning marks a new level of controls available to operators of data centers and other mission critical operations” said john peter valiulis, vice president, north america thermal management marketing for vertiv. “it’s part of wave of technology we are introducing to enhance energy efficiency and provide the highest level of protection for mission critical operations.”

as was noted in an article on datacenterknowledge.com, vertiv is not the first company to introduce machine learning into its data center infrastructure.

in 2014, google announced that it was using machine learning to analyze the reams of data in its server farms to improve usage efficiency. last july, salesforce purchased coolan, a three-year-old startup that produced software for improving data center management and hardware buying decisions. also in july, romonet, a london-based software designer, filed patents for machine learning capabilities for its cloud-based platform that analyzes data center costs.

to see the specifications for the new liebert icom-s autotuning, click https://www.vertivco.com/globalassets/documents/brochures/liebert-icom-autotuning-ds-en-na-sl-18850_124828_0.pdf.