Research On Data Center Participation In Power Grid Operation And Planning Considering Workload Scheduling

Under the goals of carbon peaking and carbon neutrality,new energy will usher in leapfrog development.Its randomness and volatility bring great challenges to the secure and economic operation of the power grid.There is an urgent need to exploit the regulation potential of demand side resources and improve the power grid’s ability to absorb and regulate new energy.Data centers are kind of demand response resources with great potential due to its huge load volume and rapid growth.With the help of workload scheduling,data centers can transfer power load in temporal and spatial dimensions.Making full use of the load characteristics of data centers and integrating data centers into the power grid operation and planning process in the way of demand response can improve the security,reliability and economy of the power system and help achieve the goals of carbon peaking and carbon neutrality.At present,there are still many deficiencies in the research on demand response of data centers.First,there is a lack of quantitative analysis of the load regulation capacity of data centers,and the regulation potential of data centers is not fully exploited.Second,there is a lack of the workload scheduling strategy for data centers to participate in the distribution network market,and the supply and demand balance of distribution network does not make full use of the regulation potential of data centers.Third,the load regulation capacity of data centers is not fully taken into account in the transmission expansion planning,which is not suitable for the trend of high proportion of new energy integration.In view of the above problems,this paper carries out the following research work.Firstly,a data center load modeling and regulation potential evaluation method based on workload scheduling is proposed.A workload model for data center processing and the data center power consumption model are established.Considering geographical load balancing,batch workload scheduling,and thermal storage operation utilizing thermal inertia of buildings,the load regulation potential of data centers is evaluated,and the calculation method of baseline load,load cutting and increasing capacity for data centers is proposed.The example results show that data centers can obtain the maximum load regulation potential when the three methods are combined.Factors such as arriving workloads,outdoor temperature,and response duration can all affect the regulation capacity for data centers.Secondly,a workload scheduling strategy for multiple data centers in the peer-to-peer(P2P)energy trading scenario in the distribution network is proposed.The operation framework of multiple data centers is established,in which the operator aims to maximize the total profits of all data centers in the data network and optimize the workload scheduling strategy.A P2 P market optimization model including data centers is established,and the alternating direction method of multipliers is used for distributed clearing to protect the privacy of prosumers.A bilevel optimization model of multiple data centers in P2 P energy trading scenario is constructed.The upper problem is the operation optimization of multiple data centers,and the lower problem is the clearing of P2 P markets.An optimization calculation method based on differential evolution and metamodel is proposed.The example results show that data centers can obtain more profits by using bilevel optimization.It can also improve the resource utilization rates of all markets and the welfare of the whole society.Finally,a stochastic transmission expansion planning method considering the spatial load transfer capacity of data centers is proposed.In the scenario of large-scale grid-connected wind power,taking into account uncertain factors such as conventional loads,data center workloads,wind power output randomness and equipment availability,a stochastic transmission expansion planning model with data centers is established based on Monte Carlo simulation.Using Benders decomposition,the model is divided into the master problem of grid investment and operation and the subproblem of data center operation.The grid and data centers interact with limited information,and optimal solution can be obtained by self-optimization and continuous iteration.The example results show that the uncertainty of workloads,the proportion of data center loads and bandwidth cost coefficient can affect the results of transmission expansion planning.Considering the spatial load transfer of data centers can reduce the investment cost of transmission line,promote the consumption of renewable energy,and improve the economy of power system operation.