Here, Iceotope outlines the five different forms of liquid cooling making it a tasty alternative to air cooling for data centres.
According to research company MarketsandMarkets, the data centre liquid cooling market size is expected to grow from USD 1.2 billion in 2019 to USD 3.2 billion by 2024.
Factors driving forecast growth include increasing need for energy-efficient cooling solutions, growing demand for compact and noise-free solutions, need for lower operating costs and need for better (microprocessor) overclocking potential.
Liquid cooling use has until recently been primarily centred on niche applications such as high-performance computing requirements (HPC) and the use of power-hungry CPUs and GPUs for artificial intelligence (AI), machine learning (ML), advanced robotics and the like. However, AI, for example is anticipated to become mainstream.
In addition, one of the worst kept secrets is the power requirements of next generation CPUs hitting the IT marketplace in the next 24 months. Coupled with the data processing needs of all AI-enabled applications, 400W – 450W microprocessors will be a considerable challenge for the cooling resources of many legacy data centres.
To meet these requirements, air is not effective as a cooling medium. By contrast, liquid cooling is increasingly being looked at as the only way to ensure IT equipment can be reliably operated.
Liquids have a much greater capacity to capture heat by unit volume; they remove heat more efficiently and allow chips to work harder (i.e. increased clock speed).
Additionally, heat is rejected to the atmosphere either via dry coolers or, in hotter environments, cooling towers. Where the infrastructure exists, liquid cooling also enables data centre heat to be reused in other applications such as district heating schemes.
Liquid cooling comes in a range of five flavours
The question for those new to liquid cooling (the technology itself is not new, having been around since the days of mainframe computing), is what are the different types of liquid cooling available on the market today?
According a helpful white paper by Iceotope partners, Schneider Electric, there are two basic forms of liquid cooling, direct to chip (also called conductive or cold plate) and immersive. From these two categories come a total of five main liquid cooling methods:
Direct to chip single-phase liquid cooling
Liquid coolant is taken directly to the hotter components (CPUs or GPUs) using a cold plate directly on the chip within the server. Other electronic components on the server board are not in direct physical contact with the liquid coolant, although some designs include cold plates around memory modules.
Fans are still required to provide airflow through the server to remove residual heat. Dedicated gamers are quite familiar with cold plate technology for direct cooling of microprocessors.
Water or a dielectric liquid can be used as the coolant to the cold plates. The use of water does introduce a downtime risk if there is leakage, but leak prevention systems (LPS) can be used to keep the water loop at slight vacuum to help mitigate this.
Fluid manifolds are installed at the back of the rack to distribute fluid to the IT equipment, the interface between server and the manifold is typically achieved via a non-spill, non-drip coupling to ensure cleanliness and safety of the installation.
Direct to chip two-phase liquid cooling
This method is just like the direct to chip single-phase method, except that the fluid utilised is two-phase, it changes from one state to another – i.e. from liquid to gas in taking away the heat. Two-phase is consider more effective than single-phase (in terms of heat rejection) but requires additional system controls to ensure proper operation.
Chassis-based liquid cooling (single phase)
The server boards are immersed in a single-phase dielectric liquid, and since all the electronic components are in contact with the coolant, all sources of heat are removed.
All server fans can be removed since the electronics are placed in an environment which is inherently slow to react to any external changes in temperature, enabling near silent operation.
In addition, the servers are within a sealed module (compatible with standard data centre racks), mitigating the risk of leaks and rendering the boards and components immune to the influence of humidity and pollutants.
Tub/open bath, single-phase liquid cooling
In this method IT equipment is completely submerged in fluid. Unlike traditional IT racks, where servers are horizontally stacked from the bottom to the top, tub immersed servers are pulled out on a vertical plane.
In some systems, this method incorporates centralized power supplies to provide power to all the servers within the tub.
The heat within the dielectric fluid is transferred to a water loop via heat exchanger using a pump or natural convection. This method typically uses oil-based dielectric as the fluid. Note that the heat exchanger may be installed inside or outside the tub.
Tub/ open bath, two-phase liquid cooling
As with the single-phase tub method, the IT is completely submerged in the fluid. The difference here is the use of two-phase dielectric coolant which changes from one state to another – i.e. from liquid to gas in removing the heat. Because of this phase change, an engineered fluid must be used in the tubs.
Choosing the best fit
As mentioned at the top of this article, many applications requiring highly dense data processing are suited to liquid cooling. However, for many data centre managers there is also the continuous pressure to deliver energy efficiency and lowered costs. Across the board, liquid cooling is perceived as the optimal cooling solution.
Direct to chip and immersive liquid cooling offer considerable benefits to data centre owners compared to air cooling. For retrofit sites, rack-based solutions like direct to chip and chassis immersive liquid cooling provide the easiest transition.
For new sites, and those in harsh environments, immersive liquid cooling is the optimum approach because it can capture all the heat as well as isolate the IT from the surrounding atmosphere.
