Coordinating Power Grid Frequency Regulation Service with Data Center Load Flexibility

TL;DR

This work tackles the problem of grid stability and carbon emissions from frequency regulation by enabling data centers, especially GPU-rich ones, to participate in regulation markets. It introduces Exogenous Carbon as a grid-side carbon-impact metric and EcoCenter, a framework for maximizing regulation provision through coordinated GPU power modulation, core allocation, and multi-GPU scheduling. The approach is validated with a high-fidelity grid model and real workload traces, showing that grid-side carbon savings from regulation can outweigh the data center's operational and embodied carbon, often by large margins, while delivering high-quality regulation at 2-second granularity. The results demonstrate a practical path to reduce fossil-fuel reliance, increase renewable penetration, and lower total cost of ownership for data centers when coupled with intelligent grid-aware control.

Abstract

AI/ML data center growth have led to higher energy consumption and carbon emissions. The shift to renewable energy and growing data center energy demands can destabilize the power grid. Power grids rely on frequency regulation reserves, typically fossil-fueled power plants, to stabilize and balance the supply and demand of electricity. This paper sheds light on the hidden carbon emissions of frequency regulation service. Our work explores how modern GPU data centers can coordinate with power grids to reduce the need for fossil-fueled frequency regulation reserves. We first introduce a novel metric, Exogenous Carbon, to quantify grid-side carbon emission reductions resulting from data center participation in regulation service. We additionally introduce EcoCenter, a framework to maximize the amount of frequency regulation provision that GPU data centers can provide, and thus, reduce the amount of frequency regulation reserves necessary. We demonstrate that data center participation in frequency regulation can result in Exogenous carbon savings that oftentimes outweigh Operational carbon emissions.

Figures (14)

• Figure 1: The growth of renewable solar and wind energy in the US renewable_data and its intermittent nature is increasingly destabilizing electrical power grids FreqRegNews.
• Figure 2: Various types of demand response (DR) services work across different timescales. Our work targets frequency regulation service, the most challenging DR service.
• Figure 3: Power grids must balance the power generation of renewable resources and end-user power consumption by relying on fossil-fuel power plants for regulation reserves. This work explores how data centers can contribute to grid balancing to reduce reliance on traditional regulation reserves and quantify the grid-side carbon footprint benefits .
• Figure 4: The "hidden emissions" of grid regulation reserves includes unaccounted emissions of performing regulation service (grid only reports CI of generation), regulation reserves cause power plants to run at lower efficiency, and requires more grid resources to provide regulation capacity. Data center regulation service helps alleviate these emissions.
• Figure 5: EcoCenter overview. Red boxes show the inputs to the framework and the green boxes show the output.