Research On Optimal Scheduling Of Electrical And Thermal Systems In Data Center Microgrid

With the development of 5G,AI,IoTs,cloud computing and other information technology,the Internet industry ushered in a new round of development and challenges.As the basic infrastructure,the data centers have surged in the number and size in the past decade.The proportion of the demand in data centers for electricity is increasing as well.The tromendous power consumption have brought enormous operational costs to the data center enterprises and induced the impact on the environment.How to reduce operating costs,deal with the environmental impact is becoming the next focus for the data center enterprises.Data center,as primary load,requires stable energy supply.The conventional units,energy storage system and renewable power will be equipped inside with data centers connected to the grid.Thereby this connection constitutes a small data center distribution network,i.e."data center microgrid".What's more,the duration time fo the workloads are different according to the characteristics of each workload,the duration time of some is relatively long,such as hours.For these workloads which have longer duration period,the processing time can be adjusted within a certain range and may be divided into pieces and be processed seperately.This special demand characteristics provides great potentials for data centers to utilize renewable resources,changing electricity prices to its full extent.Since the power could be transferred into heat by IT servers in the data center,there lies great opportunities in the waste heat recovery of data center.The main works are shown as follows:First,this article has described the work pattern of data center and devices within.The characteristics of two types of workloads,delay-sensitive workloads and delay-tolerant workloads,in the data center have been analyzed.And the research on waste heat recovery has been introduced.Second,an energy management model on electrical and thermal systems in the data center microgrid has been brought up in this paper.By temporal workload allocation,this model could achieve the joint optimization of the electrical and thermal systems in the microgrid.A two-stage stochastic programming problem has been introduced concerning the uncertainty of the stochastic parameters in the system,including the output of the renewable power,electricity price,and electrical and thermal demand.The objective is the minimization of the operation cost and the carbon emission fee.Several cases have tested the effectiveness of the proposed model.Finally,an energy management model on electrical and thermal systems in the geographically distributed data center microgrid has been further analyzed.By spatial and temporal workload allocation,the model could dispatch all the devices and workloads allocation of achieve the cooperation of thermal and power system.Based on the former model mentioned above,this model is a two-stage stochastic programming model,the objective of which is the minimization of operation cost,data migration cost,load-shedding penalty and service requirement cost.The model has been verified after cases studied.Then,the spatio-temporal workload allocation module,waste heat recovery module,and the service requirement module have been further analyzed and the affect of them has been evaluated.As the given results shows,the proposed model could reduce the overall cost and utilization of the fossil resources.