Raven Jones, P.E., C.E.M.
Wildan Energy Solutions
Data center cooling is the most important and often the least understood task of all critical facility processes. Unlike comfort cooling, each critical
facility is a unique space requiring precise heat removal. Most current research focuses on new infrastructure technology, rather than optimizing existing
facility systems. This detailed case study seeks to close this knowledge gap by addressing how to optimize and create versatility in a facility's existing
HVAC system with the utilization of high granularity cooling performance data and artificial intelligence optimization. The greatest resource for optimizing
HVAC units in a data center is the performance data that can be acquired from its cooling system. This paper analyzes pre- and post-installation data that
demonstrate a decrease in energy usage by as much as 30% by utilizing data from distributed sensors, and input into a machine learning algorithm which
controls fan and cooling set points on existing equipment. We give special attention to the implementation of machine learning, which reaches beyond building
controls, leveraging a facility's data to make its systems smarter in ways not possible or practical before the introduction of artificial intelligence.
Energy, utility and operational savings are attained by not only controlling, but optimizing cooling units to deliver cool air only to where it is needed
by the most efficient equipment.