Keeping the Lights On — Smarter Fault Management for Data Center Power Systems

Data centers demand near-perfect power availability, yet medium-voltage distribution systems remain vulnerable to faults that can cascade into costly, potentially dangerous outages. This paper presents a hybrid industrial FLISR architecture that combines the speed of decentralized, peer-to-peer relay communication with the oversight of a centralized controller — enabling automated fault detection, isolation, and service restoration in under 100 milliseconds. Validated through hardware-in-the-loop simulation across ten distinct fault scenarios, the approach demonstrates how mission-critical facilities can significantly reduce backup generator dependency, lower arc-flash risk, and maintain uptime even when communication failures or breaker malfunctions occur.

Key topics covered:

• How Tier I–IV redundancy models shape MV power system design requirements
• Trade-offs between disconnect, VFI, and ring main unit topologies
• Centralized vs. decentralized vs. hybrid FLISR architectures compared
• Zone selective interlocking for faster, safer cable fault clearing
• Automated source loss detection and restoration within a 3-second window
• System performance validated across cable, bus, and source-loss scenarios
• How HIL simulation supports both engineering validation and operator training