Demand Response Strategy Selection And Power Cost Optimization For Datacenters In Smart Grid Environment

With the widespread application of cloud computing and cloud storage,the power consumption of datacenters has grown rapidly,and the cost of power consumption has gradually increased.Largescale datacenters are flexible and adjustable,which have the potential to participate in demand response programs as power loads of the smart grid,which has been widely concerned.Most of the existing researches focused on the energy-saving management of specific energy-consuming equipment in the datacenters,but there were relatively less researches on the application of demand response programs in the datacenters.At the same time,the datacenter is not only a demand-side user in the smart grid in order to balance the operating pressure of the power grid,but also we need to focus on its own operations including information processing,data analysis,and data storage.At present,relevant research rarely regards datacenter users as participants in demand response programs,but only as service requesters.In order to reduce operating costs,while the datacenter participates in the smart grid demand response programs to obtain the corresponding electricity deduction amount,it might have to postpone the users’ tasks during the demand response period at the same time.Thus,this paper regarded the datacenter as a power load in the smart grid system,studied and analyzed its demand response management strategies,and establishes corresponding models with respect to different demand response programs.On this basis,combined with the characteristics of datacenter energy consumption,a datacenter workload dynamic scheduling algorithm was proposed.By actively and dynamically adjusting the power consumption of the datacenter in demand response periods,it can better participate in certain demand response programs.All kinds of demand response programs were evaluated after comprehensively considering the power scheduling capability and electricity charge deduction of the datacenter.In addition,considering the impact of datacenter participation in demand response programs on its user services,this paper constructed a modeling of penalty to reasonably quantify the loss of revenue that the datacenter gave to the users as compensation.In order to compare the difference in the net benefits of different datacenters participating in demand response,and to provide guidance for datacenter optimization of operating costs,four types of datacenters with different characteristics were set,and the solution of the net income obtained by the datacenter participating in the demand response programs were obtained by using GA(Genetic Algorithm)-based algorithms.