Strengthening Data Center Operations Using Grid-Forming Battery Energy Storage as a Line-Interactive Uninterruptible Power Supply

The rapid growth of hyperscale data centers, driven by rising demand for artificial intelligence workloads, presents new challenges for grid reliability, resiliency, and clean-energy goals. These facilities impose significantly higher and more complex demands on power systems, with AI-related workloads causing fast, millisecond-scale load fluctuations known as pulsating loads. Such behavior can strain the bulk power system, complicating energy management, interconnection, and local generation planning. To address this problem, this paper investigated the use of energy storage systems to manage these high-frequency load dynamics while maintaining continuous data center operation. Through a detailed analysis of AI load characteristics, this paper presents an electromagnetic transient study evaluating the performance of a grid-forming battery energy storage system for data center applications in both grid-connected and islanded modes, including transitions between these operating states. The proposed approach treats the battery energy storage system as a medium-voltage, line-interactive uninterruptible power supply while simultaneously smoothing AI-driven load transients at the point of interconnection. Results show that the proposed architecture can significantly attenuate fast power fluctuations, improve voltage and power quality at the grid interface, and enhance overall data center reliability.