A data center is a building, a dedicated space within a building or a group of buildings used to house computer systems and associated components, like servers or other telecommunications or data storage systems.

Data centers generate lot of heat and maintaining the right temperature is a key for their normal operation and avoiding downtimes, Serra-Holm said.

“High-performance computing applications such as artificial intelligence, cryptocurrency, augmented reality, and machine learning with newer, faster graphic processing units are driving graphics-intensive workloads and huge amounts of data,” Sera Holm said.

“Maintaining the ideal data center cooling temperature prevents problems such as downtime in operations, damage to equipment and increased power cost.”

Perstorp is Malmo, Sweden-based specialty chemicals maker.

Global data centers’ electricity consumption in 2022 accounted for 1.4% to 1.7% of global electricity demand and the energy use will only continue to grow, Sera Holm said, citing International Energy Agency data.

“Workloads handled by large data centers has resulted in a substantial increase in energy use in this segment over the past several years, growing by 20-40% annually.”

For example, she emphasized, data centers energy consumption in the United States is expected to increase fivefold by 2026, and more than tenfold by 2030, while 40-45% of the energy used in data centers is related to cooling.

While the immersion cooling is still a nascent technology and the demand for such fluids is currently small, it should exceed 250,000 metric tons by 2032, according to Kline & Co. consultancy.

The selection of cooling technology is primarily driven by rack power density, and average rack power density has been constantly increasing over time, Sera Holm said. According to a survey conducted by the Uptime Institute, average rack density has increased almost fourfold between 2011 and 2020.

Computer room air conditioner cooling works like a common room air conditioning, using a refrigerant and a compressor, and no large volumes of coolants are necessary. This type of cooling can cool servers of up to 15 Kilowatts power density, according to Perstorp.

Similar type of cooling is rear door heat exchanger. These systems are radiator-like doors attached to the back racks with coils and cooling fans, where water refrigerant is circulated in the coils.

“This method of heat dissipation is very efficient because it places heat removal close to the heat source. They can cool racks of up to 55 kW power density,” Sera Holm said.

Another type of cooling that uses much more liquids is direct liquid cooling, also known as liquid-to-chip or direct cold plate cooling.

“It is a cooling method that involves circulating liquid in a flat, rectangular component such as cold plate, with a chamber-like structure that is made of highly conductive material such as copper, which is attached to the server’s central processor units,” she said, adding that this type of cooling can cool server racks up to 100 kW power density.

Finally, in the immersion cooling the racks are in direct contact with a cooling liquid which removes the heat through convection.

“The cooling liquid is pumped to a heat exchanger, then cooled and recirculated. The liquid needs to be thermally conductive, electrically non-conductive or dielectric, non-toxic, non-corrosive,” Ser Holm said.

The immersion cooling can be single-phase and two-phase, and can be used for server racks up to up to 250 kW power density.

With slight variations, both technologies involve immersion of the servers and electrical components in the fluid. In two-phase cooling servers are immersed in in a low-boiling point fluid which undergoes a phase change, from liquid to gas, as it absorbs heat from servers. In single phase, the servers are installed vertically in a coolant bath of a dielectric fluid and the heat is transferred to the coolant through direct contact with server components. The coolant remains in the liquid phase and is cooled via a heat exchanger in a cooling distribution unit. Typical cooling fluids include mineral oils and synthetic fluids.

“The adoption of the immersion cooling technology is still very limited, but it is significantly growing,” Sera Holm said. The pundit predicts that high and ultra-high rack power density will require adoption of immersion cooling.

“While rack power density has almost doubled over the last decade, and the trend is expected to continue, vulnerabilities in the data center’s cooling infrastructures is among the main root causes for downtime,” she said.

The company found that for enterprises with revenue models that depend solely on the data centers’ ability to deliver IT and networking services to customers – such as telecommunications service providers and e-commerce companies – downtime can be particularly costly. In 2023, downtime’s cost has soared due to the growing complexity of data center operations, with studies estimating downtime cost at around U.S. $9,000 per minute, Perstorp found. 

Sera-holm said that immersion cooling is a preferred technology due to multiple factors.

“Both single-phase and two-phase cooling offer ease of location for data centers thanks to their plug and play configuration and lack of dependence on traditional infrastructure. Both techniques also perform well against each other where retrofitting compatibility is concerned, as many severs are already compatible with immersion. Single-phase immersion cooling solutions require significantly less maintenance and are easier to access and repair.”

Kline found that less than 1% of data centers globally have deployed immersion cooling in 2023.

“However, immersion cooling adoption is likely to accelerate in the coming years under multiple drivers such as sustainability-driven design goals, governmental regulation, rising power costs, technological demands for cooling cutting-edge central and graphic processor units. As well as the need for more efficient cooling systems,” Sera Holm said in conclusion.

She reiterated that single-phase immersion cooling is likely to be the preferred technology in the future.

“Multiple fluids are already commercially available on the market and the fluid selection will depend on cost, performance and sustainability considerations.”