AI Has Data Centers Heating Up. Better Systems Can Optimize Cooling While Lowering Costs

Data centers are mission-critical facilities that keep key parts of our society running, but they are also major energy consumers. A U.S. Department of Energy report found that while data centers consumed approximately 4.4% of total U.S. electricity in 2023, they are predicted to consume 6.7% to 12% by 2028. 

This figure not only represents the major environmental impact of these facilities, but also a significant cost burden for operators. As a result, more data center owners are transitioning to primary-only variable chilled water systems for their cooling needs. 

Cooling accounts for nearly 40% of a data center’s energy consumption, and these systems can lead to significantly more energy savings compared to primary-secondary chilled water systems, which utilize two separate water loops for cooling. 

“With a primary-secondary, you're keeping a fixed flow rate through your chillers at all times, which means you've got a fixed amount of input power that's required to run your pump,” said Peter Sumner, general manager at critical cooling specialists Airedale by Modine. “With a primary loop system, each of the chillers will be working on a variable flow rate, which means that you can slow down your pumps, resulting in major energy savings.” 

While the benefits of these systems are clear, there are also some challenges to managing primary-only variable chilled water systems that data center owners will need to address, he said. For example, when the pumps have been turned down and a system is working with lower loads, it becomes more difficult to control water temperatures and flow distribution. 

Because of these issues, Sumner said, operators who choose a primary-only system may need more than a traditional building management system to keep things running smoothly. He explained that a traditional BMS is event-driven, meaning that something happens and then the BMS makes a decision about how to handle the situation. Instead, operators need a system that can quickly make incremental adjustments.

“You need a system that is preemptively changing set points to keep on top of the load at all times,” he said. 

Airedale, Sumner said, not only produces all of the indoor and outdoor cooling equipment that mission-critical facilities need, it also can link the equipment together into one system that can go beyond the work of a traditional BMS to keep everything operational. 

“A BMS provider is an expert switching things on and off in a building, but they're not a cooling expert — we are,” Sumner said. 

The company’s PLC-based controllers can be added to each chiller, creating an additional layer of control that automatically adjusts and optimizes chiller flow, allows for networked redundant intelligence and measures system temperatures. The systems prioritize free cooling, which is dynamically managed to ensure cooling is maintained at the lowest possible energy input.

He said that as the use of AI continues to rise and data centers become even greater energy consumers, coolant distribution units will be working overtime. This has inspired Airedale to optimize its cooling systems. 

The firm has created a digital twin — a replication of a data hall — based on a customer’s live operational data center. This is helping the company improve its data center equipment and systems. 

“We’ve already optimized the system as a whole,” Sumner said. “Now, we're trying to find small, incremental efficiency increases to help operators drive down costs.”  

This article was produced in collaboration between Airedale by Modine and Studio B. Bisnow news staff was not involved in the production of this content.

Studio B is Bisnow’s in-house content and design studio. To learn more about how Studio B can help your team, reach out to studio@bisnow.com.