Data centers need a dynamic system for managing cooling.
Every year, data centers use about 420 terawatt hours of electricity, which is about 3% of the world’s total electricity use. Cooling, which takes up about 40% of the energy used in a data center (about 168 TWh), is needed to keep important IT programs running. But when temperatures go above certain limits, bad things happen. Optimization of cooling is a hard, multivariable, and changing problem. The Solution uses machine learning and predictive control to meet service level agreements and environmental requirements in a way that is efficient and good for the environment. Machine learning is used to train an Influence Model, which is then used by predictive control software to decide which cooling units to send to hot spots to get rid of them and to distribute capacity across a set of cooling units to get the most out of them. By adding intelligence to standard operating strategies like machine rotation, Vigilent has solved the exploitation vs. exploration conflict of reinforcement learning in the demanding mission critical facility industry. The system uses open protocols like BACnet and Modbus, which make it easy to connect to other IT and facility management solutions.
Positive facts about Dynamic Cooling Management System
Takes ~ 85,000 control actions every day
100,000 coefficients that are learned in real-time
30%+ reduction in energy use and CO2 emissions
30%+ increase in HVAC life, eliminating its associated carbon footprint
Hot spot reduction up to 99%, reducing the risk of IT outages
Payback in less than 3 years
Increases HVAC capacity by 10-15%
Video about Dynamic Cooling Management System
https://www.youtube.com/embed/DNH5iFRhsFg
Dynamic Cooling Management System website : https://vigilent.com/products-and-services/dynamic-control/
