Flexible Data Centres as Grid-Interactive Energy Resources: A Multi-Objective Optimisation Framework for Demand Response, Renewable Integration, and Energy Storage

Abstract

The rapid expansion of cloud computing and artificial intelligence is increasing electricity demand from data centres while intensifying local grid congestion and connection delays. Nevertheless, data centres also possess controllable resources, including delay-tolerant computing workloads, battery energy storage, onsite renewable generation and flexible cooling, that can enable them to operate as grid-interactive energy resources. This paper develops a multi-objective linear optimisation framework for coordinating these resources. The formulation simultaneously minimises electricity expenditure, operational carbon emissions and peak grid demand while preserving computational service requirements, battery operating limits and power balance. A transparent 24-hour illustrative case study is used to demonstrate the framework. Four operating strategies are compared: conventional operation, workload shifting, battery-only flexibility and coordinated optimisation. Under the stated synthetic assumptions, the coordinated strategy reduces peak grid demand, daily energy expenditure and operational emissions relative to the conventional case while maintaining full use of available onsite renewable generation. The analysis also shows that workload scheduling and storage are complementary: scheduling relocates delay-tolerant demand toward low-price and low-carbon periods, whereas storage provides fast temporal balancing and peak support. The proposed formulation offers a reproducible foundation for larger studies involving multiple data-centre sites, uncertainty, ancillary-service markets, thermal dynamics and carbon-aware computing. The results should be interpreted as an internally consistent methodological demonstration rather than measurements from a commercial facility.