In recent years, national policies have been tightening restrictions on the PUE (Power Usage Effectiveness) of data centres, and PUE is a commonly accepted and adopted indicator for measuring the energy efficiency of data centre infrastructure in data centres at home and abroad.PUE = total data centre power consumption / IT equipment power consumption, and is used to calculate how much of the total power provided to the data centre is actually used for IT equipment. The closer the PUE value is to 1 the more energy efficient it is. A higher PUE value indicates that more power is consumed by supporting infrastructure such as cooling and power supply.
This is not the first time that policies have restricted high PUE IDCs in the context of green dual carbon. As data centres in many locations have completed access to energy consumption monitoring, the direction of low carbon data centre operations is becoming more and more certain. HVAC equipment is critical to energy consumption, and opportunities for advanced cooling investment are highlighted.
Modern data centres are responsible for large volumes of data computing and storage, which places increasing demands on the stability of system performance. In order to provide greater energy efficiency to meet the needs of business development, the cooling of servers and even server rooms becomes critical. To solve the problem of “high heat density data centre cooling”, the most popular technology in the industry is liquid cooling, which not only cools down quickly and consumes less energy, but is also the most environmentally friendly.
The energy consumption generated by air conditioning in the server room accounts for 40% of the total energy consumption of the data centre. The most effective way to reduce energy consumption is to start from the cooling side, with advanced technologies such as indirect evaporative cooling, liquid cooling and seawater cooling, which can significantly reduce the PUE level, looking ahead to the next five years will usher in greater opportunities.
Liquid cooling technology, as a recognised tool to reduce PUE, has become the focus of industry technology. While many data centres still use the traditional method of cooling units and fans to cool computer hardware, the advantages of liquid cooling technology over air cooling are clear, including
More heat removal: the same volume of liquid removes 3000 times more heat than the same volume of air.
Faster temperature transfer: liquid conducts 25 times more heat than air.
Better noise quality: liquid-cooled systems are 10 to 15 dBA quieter than air-cooled systems at the same level of heat dissipation.
Energy saving: liquid-cooled systems are approximately 30% more energy efficient than air-cooled systems.
Liquid cooling is divided into the following three main types.
(1) Indirect contact type: the cooling liquid does not directly contact with the heating device, the main heating device is fixed on the cold plate, relying on the liquid flowing through the cold plate in the designed liquid cooling heat sink, through the circulation of heat away from the cooling method. According to the presence or absence of phase change is divided into: single-phase indirect liquid cooling; two-phase indirect liquid cooling.
(2) Direct contact: a cooling method in which the heat sink component (single board & chip) is directly submerged in a non-conductive liquid and the heat sink component releases heat through evaporation.
This is most notably submerged liquid cooling, which, as the name suggests, involves submerging the heat generating components in a cooling liquid and relying on the flow of the liquid to circulate the heat generated by the operation of the IT equipment. Submerged liquid cooling is typically a direct contact type of liquid cooling, and is currently the more popular method of submerging servers in specially designed enclosures. The submerged liquid cooling system structure is divided into two parts, the boiling tank and the cooling tower.
Of these, the boiling tank is the key to the whole system and consists of the part being dissipated (the server), the boiling tank, the fluorinated liquid, the condenser The evaporator, the cooling fan, the cooling pump and the control system make up the system. The main function is to finally release the system heat into the natural environment.
The electronic equipment is submerged in a fluorinated liquid tank that “reaches a certain temperature and undergoes a phase change”. The fluorinated liquid is heated and boils into steam, which rises and condenses into liquid when it meets the condenser at the top of the case, and the condenser is connected to an external liquid cooling source for heat transfer to the outside world. Through the circulation of fluorinated liquid inside the chassis and the external circulation of glycol water solution in the condenser (double circulation system), the temperature control of the server and chip cooling is achieved to ensure the stability and life of the server.
(3) Spray cooling: The main feature is that non-corrosive coolant is sprayed directly onto the surface of the heat generating equipment or the extended surface in contact with it through the spray plate on the server chassis. The absorbed heat is transferred and exchanged with the external environment of the large cooling source, thus achieving the target temperature of the control system.
The latest document issued by four departments, including the Development and Reform Commission and the Ministry of Industry and Information Technology, requires large and ultra-large data centres to reduce their PUE to below 1.3 by 2025, while liquid cooling can reduce the PUE to around 1.1, which is fully capable of meeting the policy requirements. In addition, liquid cooling can solve the problem of high-density heat dissipation and save construction costs, while also having the advantages of quietness and low noise, and easy siting, which can better adapt to future development trends.
According to foreign research agency ResearchAndMarkets, the global liquid-cooled data centre market will reach US$4.55 billion by 2023, with a compound annual growth rate of 27.7%. According to the 2020 Global Data Centre Survey Report released by the Uptime Institute, 71% of data centres worldwide have an average power density of less than 10 kW/rack, and only 16% of data centres have an average power density of more than 20 kW/rack. Currently, liquid-cooled data centres are mainly limited by cost, while it is estimated that at a single rack density of 20 kW, the cost of building liquid-cooled and air-cooled data centres can be equalised. As data centre power density rises in the future, liquid-cooled data centres are expected to become more widespread.