Enhancing Grid Resilience for Giga-Watt Scale Data Centers Using High Voltage Circuit Breaker Operated Braking Resistors

Overview

As hyperscale and co-located data centres scale, the grid sees large, voltage-sensitive IT loads with plant sizes ranging from 500 MW to 2 GW. A sudden loss of these loads as they switch to on-site UPS during a grid voltage excursion causes a grid frequency rise from generation and load imbalance, and a voltage rise because less power is flowing through the network. In the July 2024 incident about 1,500 MW of voltage-sensitive demand disconnected after six voltage depressions from automatic reclosing on a 230 kV line; frequency peaked at 60.047 Hz and voltage reached 1.07 pu before shunt capacitors were withdrawn.

The proposed remedy is braking resistors switched by high-voltage circuit breakers at the data-centre transmission substation, temporarily absorbing the lost load so that the imbalance is staggered in time. Simulations on a test bed with a gigawatt-scale data-centre cluster on a 345 kV network, for different synchronous/inverter-based generation mixes, show that brief resistor insertion limits the rate of change of frequency while governor response and capacitor switching act. Resistor insertion for 0.25-0.85 seconds sufficiently reduced rate of change of frequency and gave time for primary governor response and capacitor switching.

Illustrative scenario: five 200 MW data centres with a 1:1 IT to non-IT split give a maximum load drop of 500 MW at the substation when ANSI 27/59 relays transfer IT loads to UPS. For a 500 MW IT-load loss the largest single-stage braking resistor tested was 250 MW (half the IT load); in the IBR-rich case a 370 MW multi-stage bank (130, 130 and 110 MW) reduced the maximum step change from 250 MW to 130 MW. A precedent cited is the Bonneville Power Administration 1400 MW braking resistor at Chief Joseph substation.

The finding is that multi-stage braking resistors are more effective than a single stage in inverter-rich systems. Limits: it is a conceptual application; the authors list thermal adequacy, cycling, stress testing in regions where IBRs operate in grid-following mode and other items as future work. It is a preprint.

Key points

1. Data-centre plant sizes are cited as ranging between 500 MW and 2 GW.
2. Resistor insertion for 0.25-0.85 seconds sufficiently reduced rate of change of frequency and gave time for primary governor response and capacitor switching.
3. In the July 2024 incident about 1,500 MW of voltage-sensitive demand disconnected after six voltage depressions from automatic reclosing on a 230 kV line; frequency peaked at 60.047 Hz and voltage reached 1.07 pu.
4. Illustrative scenario: five 200 MW data centres with a 1:1 IT to non-IT split give a maximum load drop of 500 MW at the substation when ANSI 27/59 relays transfer IT loads to UPS.
5. For a 500 MW IT-load loss the largest single-stage braking resistor tested was 250 MW; in the IBR-rich case a 370 MW multi-stage bank (130, 130 and 110 MW) reduced the maximum step change from 250 MW to 130 MW.
6. Precedent cited: the Bonneville Power Administration 1400 MW braking resistor at Chief Joseph substation.

Using it in design

• In gigawatt-scale data-centre interconnection studies, add the load-loss scenario from IT load transferring to UPS in a voltage dip (for example 5 × 200 MW with a 1:1 split giving 500 MW) to the grid frequency and voltage analysis.
• In inverter-rich grids evaluate multi-stage braking resistor banks (such as 130/130/110 MW) against a single stage; insertion time was in the 0.25-0.85 s range.
• Coordinate ANSI 27/59 relay settings and UPS transfer thresholds with grid resilience at the point of connection.

Who should read it

Grid engineers and data-centre designers planning gigawatt-scale data-centre interconnection and grid resilience.

Document contents

1. I. Introduction (A. Research motivation, B. Manuscript contribution)
2. II. Non-Voltage Sensitive Load Ride Through
3. III. Test Bed System Description
4. IV. Simulation Results and Discussion (A. Single machine system with single-stage resistive braking, B. Multi-machine system with single-stage resistive braking, C. IBR rich system with multi-stage resistive braking, D. System stability evaluation through eigenvalue trajectories, E. Investigating risk of motor stalling due to grid induced events)
5. V. Conclusion and Future Work

Read the document

Publication details

Authors

Soham Ghosh; Mohammad Ashraf Hossain Sadi

Publisher

arXiv preprint (Black & Veatch / University of Central Missouri)

Year

2025

Document number

arXiv:2512.21295

Document language

English

Source type

Scientific

Access

Open licence — downloadable

DOI

10.48550/arXiv.2512.21295

Licence

CC BY 4.0 — share with attribution

Licence evidence

arxiv.org

Citation