Harnessing dynamic carbon intensity for energy-data co-optimization in internet data centers

Internet data centers (IDCs) possess inherent spatiotemporal regulation capabilities, positioning them as flexible resources within the power grid to enhance the integration of renewables and reduce emissions. However, conventional dispatch methods for grids with IDCs primarily emphasize economic optimization, often neglecting the carbon intensity fluctuations induced by variable renewables and data loads. This study addresses the gap by proposing a spatiotemporally coupled two-stage energy-data co-optimization method that incorporates dynamic carbon intensity. The method leverages the complementary capabilities of electrical energy storage and ice-storage air-conditioner, the spatiotemporal flexibility of interactive and batch data, and a dynamic carbon intensity feedback mechanism to jointly optimize operating costs, renewable utilization, and carbon emissions. The proposed method is validated on the IEEE 14-bus power system. Results show that energy-data co-optimization reduces IDC operating costs by 19.778 %, lowers the renewable curtailment rate from 5.996 % to 1.210 %, and reduces the carbon emission responsibility from 40.896 tCO2 to 11.448 tCO2, respectively. Furthermore, the carbon-aware load migration facilitates the transfer of data loads to regions with abundant low-carbon electricity, achieving cross-regional complementarity between wind and photovoltaic generation, with the complementarity metric increased from 0.449 to 0.465.