考虑多时段设备耦合的数据中心能效优化方法

Optimization Method for Energy Efficiency of Data Center Considering Multi-period Equipment Coupling

大数据时代,数据中心的规模越来越大,其高能耗问题也日益突出。当前,中国数据中心总能耗已约占社会总耗电量的2.7%,其能效优化研究已成为近年来关注的热点。然而,当前研究忽略了信息技术(IT)设备和空调设备的时段间启停待机关系和爬坡约束,其给出的单时段最优策略在实际运行时可能出现相邻时段间IT设备频繁启停,导致IT设备损耗过大或算力需求增长过快时无法第一时间平衡算力供需等问题。因此,数据中心能效优化建模必须考虑时段耦合性。基于此,综合考虑了数据中心IT设备和空调设备的能耗特性与换热耦合关系,以及IT设备的时段间启停待机关系和爬坡约束,建立了基于多时段设备耦合的数据中心能效优化模型。针对时段耦合约束及IT设备数量规模大导致的大量整数变量带来的计算效率难题,分别提出前瞻解耦算法和IT设备聚合算法进行求解。最后,通过一个10万台IT设备的大规模数据中心仿真算例,验证了该方法的实用性和高效性。

In the era of big data,the scale of data centers is getting bigger and bigger,and the problem of high energy consumption is becoming more and more prominent.At present,the total energy consumption of data centers in China accounts for about 2.7%of the total electricity consumption of the society,and the research on energy efficiency optimization has become a hot topic in recent years.However,the current research ignores the inter-period start-stop-standby relationship and ramping constraints of information technology(IT)equipment and air-conditioning equipment,and the single-period optimal strategy given by it may cause frequent start-up of IT equipment between adjacent periods during actual operation.Due to excessive IT equipment loss or rapid growth in computing power demand,it is impossible to balance the computing power supply and demand immediately.Therefore,the energy efficiency optimization modeling of data centers must consider the time period coupling.Based on this,by comprehensively considering the energy consumption characteristics and heat exchange coupling relationship between IT equipment and air-conditioning equipment in data centers,as well as the inter-period start-stop-standby relationship and ramping constraints of IT equipment,an energy efficiency optimization model of data centers based on multi-period equipment coupling is established.To solve the computational efficiency problems caused by time coupling constraints and a large number of integer variables produced by the large number of IT equipment,a forward-looking decoupling algorithm and an IT equipment aggregation algorithm are separately proposed.Finally,a large-scale data center simulation example of 100000 IT equipment is used to verify the practicability and high-efficiency of the proposed method.