Optimal Energy Management of Internet Data Center With Distributed Energy Resources

Abstract

Currently, an increasing number of Internet data centers (IDCs) are trying to apply distributed energy resources (DERs), such as renewable energy, battery energy storage systems (BESS), and conventional generators (CG). However, uncertain renewable energy presents significant challenges to the safe and stable operation of IDCs. A two-stage optimal operation model based on distributionally robust optimization (DRO) is proposed for an IDC with DERs to lower the total cost of the IDC, including operating cost, carbon emission cost, and re-dispatch cost under the worst-case probability distribution of renewable energy. An ambiguity set is formed by combining norm-1 and norm-inf under the given confidence level to capture the possible probability distributions of uncertain renewable energy. The proposed model can be turned into a tractable mixed-integer linear programming problem and efficiently solved by the column-and-constraint generation algorithm. Experimental results show the benefits of DERs integration and workload dispatch. Comparative analysis further demonstrates the superiority and scalability of the proposed model.