当前位置:网站首页>Low carbon data center construction ideas and future trends
Low carbon data center construction ideas and future trends
2022-06-10 19:09:00 【H3C-Navigator】
In recent years , With IT The rapid development of Technology , The demand for data processing capability is getting higher and higher , The demand for infrastructure services such as data centers has increased dramatically . At the same time, green development 、 The concept of energy conservation and environmental protection has gradually penetrated and developed in many industries , Become an important way to improve the comprehensive benefits of development . In China, carbon dioxide emissions will strive to 2030 Reach the peak of carbon by 2000 、 stay 2060 Under the vision of achieving carbon neutrality and energy efficiency dual control by , green 、 Energy saving 、 Low carbon data center development has become the focus . Relevant research shows that , The cooling system of the data center is the most power consuming part of the entire data center , Of total data center energy consumption 30%~50%. This paper analyzes the current situation of HVAC system in data center , Discuss the development path of low carbon data center , by “ Double carbon ” Provide suggestions on the planning and construction of the data center under the target .
Data center low carbon policy
At present , China has issued a series of policies to promote the green development of data centers .2020 year 12 month , National Development and Reform Commission 、 Central network information office 、 Ministry of industry and information technology 、 Jointly issued by the national energy administration and other four ministries and commissions 《 Guiding opinions on accelerating the construction of collaborative innovation system of national integrated big data center 》 It is proposed that , To 2025 year , The eastern and Western data centers achieve structural balance , large 、 Very large data centers run PUE No more than 1.3.2021 year 7 month , Issued by the ministry of industry and information technology 《 Three year action plan for new data center development (2021—2023 year )》, Make it clear to 2023 year , New large and above data centers PUE Down to 1.3 following . Since the energy consumption of HVAC system accounts for a relatively high proportion of the total energy consumption of the data center , To make the data center PUE Compliance with policy limits , Reducing the energy consumption of HVAC system is the focus of the current data center .
Combined with the data center refrigeration system, this paper describes the green data center 、 Low carbon 、 modularization 、 The development direction of intelligent technology .
Status quo of data center refrigeration system
At present, China's data center refrigeration system is mainly divided into the following types : Water cooling 、 The wind is cold 、 Indirect evaporation 、 Magnetic levitation phase change refrigeration and liquid cooling .
Water cooling system is the most commonly used cooling system in the data center at present , The comprehensive heat of the room is transferred to the outdoor cooling tower through the end air conditioner through primary and secondary heat exchange , The heat is transferred to the outdoor through the fan , More heat exchange losses lead to high energy consumption in system operation , And the water system pipeline is complex , There are many valves , Extremely unfavorable to operation and maintenance . At present, more and more users begin to pay attention to the energy-saving problems caused by the high energy consumption in the data center , The operating water temperature of the refrigerator in the water cooling system is directly related to the refrigeration energy consumption , Relevant research shows that , The average outlet temperature of chilled water increases every 1℃,COP Ascension approx 3.8%, The temperature of chilled water in the new data center is also gradually planned to 18/24 degree , The existing water-cooled data center can be optimized and transformed according to the actual situation , Save energy consumption in operation , With the water chiller and cooling tower 、 Improvement of plate material exchange process and heat exchange efficiency , At present, the average annual operation of the water cooling system is PUE Some areas can be lower than 1.3.
The air cooling system is mainly in the following four forms :
System 1. Air cooled chiller + Terminal precision air conditioner ;
System 2. Direct expansion precision air conditioner + Outdoor unit ;
System 3. Indirect evaporative cooling air conditioning ;

chart 1 Indirect evaporation unit and heat exchange principle
System 4. Magnetic levitation phase change refrigeration air conditioning ;

chart 2 Magnetic levitation phase change refrigeration system
According to previous data center planning and design experience , When the construction of data center is restricted by the impact of water resources , Will consider the system 1 And system 2 Air cooled form , However, high energy consumption has always been the focus of attention , The two air-cooled systems operate annually PUE Even as high as 2.0 above , Not in line with the current society “ Double carbon ” Data center construction requirements . With the increasing demand for energy conservation , Indirect evaporative cooling system 、 Magnetic levitation phase change refrigeration system has gradually replaced the traditional form of air conditioning , In terms of system form , By reducing the number of heat exchange links 、 Add energy-saving modules , Increased heat exchange efficiency , Extended free cooling service time , It solves the problem of traditional air cooling system PUE Too high a problem ; On system architecture , The refrigeration system is becoming more and more modular , Users can flexibly configure refrigeration power according to their own construction needs , On demand refrigeration , Reduced inefficient energy consumption . These refrigeration systems are gradually applied on a large scale with the construction of the national low-carbon data center .
Data center low carbon route development
Now ,“ Carbon reaches its peak 、 Carbon neutralization ” It has become a green energy-saving modern data center 、 Important guiding principles for ecological civilization planning , It poses more severe challenges to the development of the data center .
l Location of data center
Energy efficiency level and location of data center 、 technology 、 Management and other aspects are closely related . The location of the data center should consider the climate and regional factors as much as possible , Superior geographical location is a prerequisite for solving the high energy consumption of the data center . chart 3 yes GB50176 Five zones divided according to domestic climate parameters , At present, the data center is mainly clustered in , Annual average temperature in Guiyang 15.3 degree , The annual average temperature in Zhangbei and Wulanchabu areas is less than 10℃. Secondly, it is equipped with efficient power supply and distribution equipment 、 refrigeration unit 、 Intelligent control system, etc , Create powerful conditions for extremely low-power data centers .
In addition, the location of the data center can give priority to areas rich in renewable energy , Realize local consumption of renewable energy . Take Zhangbei area as an example , Relevant data show that the average annual utilization hours of wind energy in Zhangbei are 2400 hours , The average annual sunshine hours of solar energy are about 3000 Hours . During the Beijing East Olympic Games , Zhangbei renewable energy has realized all competition venues for the first time in history 100% Green power supply , For data centers , Not IT The priority of green power supply can be increased for the distribution of key facilities , Access to green energy , Green data center . Wind power 、 Photovoltaic 、 Energy storage 、 Pumping and storage and other forms of energy will be greatly developed in the future , In the future, the site selection of data center should include green and sustainable energy utilization into the measurement index .

l Data center liquid cooling
“ Low carbon ” The construction requirements of the centralized data center show that the traditional air-cooled direct expansion 、 High energy consumption refrigeration systems such as air-cooled chillers can no longer meet the current policy and market requirements , With cloud computing 、5G、 Big data network construction , The power of a single cabinet can reach 20kw、30kw, More and more attention has been paid to the cooling of server chips , Compared to air with low thermal conductivity , The cooling capacity of liquid is that of air 1000~3500 times , Liquid cooling has the advantage of ultra-high heat conduction , It has become one of the refrigeration systems in the data center .
There are three main forms of liquid cooling : Submerged 、 Cold plate and back plate ;
Submerged liquid cooling can be divided into single-phase submerged type and two-phase submerged type according to whether there is phase change in heat exchange , In the single-phase immersion type, the device and the refrigerant always maintain the liquid state during the heat exchange process , Use the temperature difference of the refrigerant itself for convection or increase the power to strengthen convection ; Phase change refrigerant is used in two-phase immersion type , The refrigerant is vaporized after heating , Condense at the top of the cabinet and then flow back to the cabinet .

chart 4 Schematic diagram of submerged liquid cooling
In the cold plate liquid cooling, the working liquid is separated from the cooled object , The working fluid is not in direct contact with electronic devices , The heat of the cooled object is transferred to the refrigerant through high-efficiency heat transfer components such as liquid cooling plate . Some scholars have conducted energy-saving analysis on the adaptability of cold plate liquid cooling in different thermal zones in China , It is found that the plate liquid system PUE All are lower than 1.13, It has remarkable energy saving effect . The first batch of industrial standards for liquid cooling series in China, which was formulated by Xinhua Group III under Ziguang Co., Ltd., has been officially released , meanwhile , Xinhua III participated in the industry ( such as CCSA、ODCC And the Electronics Association ) Almost all liquid cooling standards are formulated . At present, we can provide customers with customized cold plate liquid cooling server .
The back plate liquid cooling also separates the working liquid from the cooled object , The working fluid is not in direct contact with electronic devices , Transfer heat to the cooling medium through the server fan .

chart 5 Schematic diagram of cold plate liquid cooling

chart 6 Schematic diagram of back plate liquid cooling

surface 1 Advantages and disadvantages of liquid cooling system
l Data center modularity
In recent years, the pace of data center construction and delivery has accelerated , modularization 、 Low granularity has become one of the mainstream demands of data center construction . The modular machine room has high flexibility , Users can purchase the data center by stages according to their own construction or lease 、 The construction of the , Each sub unit is relatively independent , It can adapt to a variety of scenarios , In the modular deployment mode ,UPS、IT equipment 、 Air conditioner 、 distribution 、 fire 、 lighting 、 Subsystem modules such as generic cabling can be stacked and deployed in the main production area of the machine room .
Modularity indicates high product integration to a certain extent , From the perspective of refrigeration, heat exchange and electrical transmission , More efficient conversion , Lower PUE, Meet the construction requirements of low-carbon data center , This is also the indirect evaporation of the current data center 、 Liquid cooling 、 The main reason why the form of high-voltage DC is favored .
l Intelligent data center
Data center intelligence means transforming the data center “ The brain ”, take AI Technology applied in data center , Build a neural network model through big data , Depth adaptive logic control , Establish energy consumption and IT load 、 Climatic conditions 、 Machine learning models such as the number of equipment running , Linkage control the operation mode of each subsystem , Accurately infer and configure the optimal control logic of the data center , Adjust parameters in real time , Lower data center PUE.
At present, the control unit area division in the data center is quite obvious , The control logic is relatively simple , Take the refrigeration system as an example , In order to make full use of natural cold source , Reduce the energy consumption of system operation , Set up the unit “ Dry condition ” Pattern , When the outdoor wet bulb temperature 、 When the water temperature under the cooling tower reaches a certain value , The cold mechanism stops , The end return water transfers heat to the cooling side through the plate exchanger . At present, the water cooling system of most domestic data centers is dry 、 The temperature and humidity during wet condition switching are at a fixed adjustment value , The temperature and humidity are actually disturbed by many factors , Outdoor parameters can be comprehensively considered , Modeling based on the measured data of the data center's perennial operation , To realize the dynamic intelligent control of the refrigeration system , Reasonably optimize the setting parameters , Maximize the use of natural cold sources .

chart 7 Xinhua 3 micro module data center
Sum up , With the country's low-carbon 、 The requirements of energy conservation are becoming more and more strict and people's awareness of environmental protection is improving , The green energy-saving rating of data centers is becoming more and more regular , In particular, high energy consumption in HVAC has always been the focus of social attention , This paper analyzes the main refrigeration systems used in domestic data centers , Let readers have an intuitive understanding of the energy consumption and development of the existing refrigeration system .
future , Under the new target, the construction of data center will start from site selection planning , Green solutions 、 Low energy technology products will play an important role , modularization + Intelligent data center will become the mainstream , Equipped with advanced liquid cooling system and efficient power transmission system , Intelligent tuning control will cover every corner of the data center , Realize green high technology to fill the entire life cycle of the data center . At the time , We will usher in the real era of low-carbon data center .
Data center is a big energy consumer , The transition to low carbon and energy conservation is an inevitable trend , Various low-carbon energy-saving technologies focusing on the cooling system of the data center will “ Double carbon ” Play a complementary role in the process , Drive the data center to gradually realize “ Carbon neutralization ” The goal is .
边栏推荐
- 腾讯云数据库TDSQL-大咖论道 | 基础软件的过去、现在、未来
- JS Touch
- [kuangbin]专题十二 基础DP1
- Adobe Premiere基础-导入导出,合并素材,源文件编译,脱机(二)
- 直播预告 | 社交新纪元,共探元宇宙社交新体验
- 基于ssm在线订餐系统设计与实现.rar(项目源码)
- Ruijie x32pro brush openwrt enable wireless 160MHz
- JS Standard
- What is Bi? Talk about the definition and function of Bi
- Cross domain error: when allowcredentials is true, allowedorigins cannot contain the special value "*“
猜你喜欢

基于SSM流量计量云系统的设计与实现.rar(论文+项目源码)

端午“沉浸式云旅游”怎么玩?即构助力“直播+”新场景落地

【代理】10分钟掌握正向代理和反向代理的本质区别

Nodejs basic architecture analysis parsing engine directory plug-in installation core module

Adobe Premiere Foundation (the last step of video subtitle adding) (6)

AEC: analysis of echo generation causes and echo cancellation principle

Adobe Premiere基础(动画制作-弹性动画)(八)

libcurl 7.61.0 VS2013 编译教程

直播预告 | 解构OLAP!新型多维分析架构范式全公开!Apache Doris 将带来五个重磅议题!

Design and implementation of online ordering system based on SSM Rar (project source code)
随机推荐
Screen output of DB2 stored procedure, output parameters, and return result set
How to correctly understand the real-time nature of Bi?
[kuangbin]专题二十二 区间DP
HelloWorld example of TestNG and how to run it from the command line
AEC: analysis of echo generation causes and echo cancellation principle
数据治理经典6大痛点?这本书教你解决
Huawei cloud hcde Cloud Road phase II: how does Huawei cloud help small and medium-sized manufacturing enterprises' digital transformation?
Metadata management, the basic construction of enterprises in the digital era
【杂谈】恭喜自己获得CSDN专家称号,努力终会换来结果
最长上升子序列(LIS)洛谷
《Single Image Haze Removal Using Dark Channel Prior》去雾代码实现分析
In the era of data processing, data analysis has become the basic construction
Live broadcast preview | a new era of social interaction, exploring new social experiences in the universe
第161章 SQL函数 YEAR
VS从txt文件读取中文汉字产生乱码的解决办法(超简单)
OPENCV 检测人脸 不依赖于任何第三方库
数据库防火墙的性能和高可用性分析
Some summary about YUV format
5. golang generics and reflection
Google Earth engine (GEE) -- Copernicus atmosphere monitoring (CAMs) global aerosol AOI near real-time observation data set