As the number of data centers, cryptocurrency mining operations and other large loads has grown on the North American electric grid, the chance for “large amounts of voltage-sensitive load loss” also has increased, according to an incident review released Jan. 8 by NERC.
The review covered an incident last year in the Eastern Interconnection that, in the ERO’s analysis, illustrates the potential dangers of simultaneous loss of large loads. Details about the incident, such as the location and the utilities involved, was not included in the report, a common practice in NERC’s Lessons Learned reports and other incident reviews.
The incident began around 7 p.m. ET on July 10, 2024, when a lightning arrestor on a 230-kV transmission line failed. This led to a permanent fault that locked out the transmission line. In the following 82 seconds, the line’s auto-reclosing control launched three auto-reclose attempts at each end of the line, resulting in six system faults with voltage depression that the protection system detected and cleared. Fault durations ranged from 42 to 66 milliseconds.
While the six faults were occurring, the same local area experienced 1,500 MW of load reduction. All of the affected load was “data-center type load,” the ERO said, of which there is “a high concentration” in the area of the disturbance, and “was disconnected on the customer side by customer protection and controls.”
The load loss caused frequency to rise as high as 60.047 Hz, settling back to 60 Hz in about four minutes. Voltage rose to 1.07/unit at the highest level, indicating a voltage 7% higher than the base voltage of 230 kV; the system fell back to normal operating values within a few minutes after operators removed shunt capacitor banks in the area.
Following the disturbance, grid operators held discussions with data center owners to determine the cause of their load reductions. They found that in response to the initial disruption, the data centers transferred their loads to their backup power systems.
“Data center loads are sensitive to voltage disturbances,” NERC noted in the report. “The data center protections and controls are designed to avoid equipment outages for voltage disturbances. … To ride through voltage disturbances on the electric grid, data centers employ uninterruptible power supply (UPS) systems that will instantaneously take over providing power to the data center equipment when a grid disturbance occurs.”
Data centers may employ different UPS designs with differing characteristics, the ERO continued. A static centralized UPS uses power electronics upon the detection of a grid disturbance to switch load to a battery bank that can provide power to operate either until the disturbance is cleared or until a backup generator can be started. These systems will transfer the load back to the grid automatically if the voltage returns to normal quickly.
By contrast, a dynamic/diesel rotary UPS, or DRUPS, uses a flywheel both to provide uninterruptible power and to start a diesel generator in the event of a grid disturbance. In this case the load typically must be transferred back to the grid manually no matter how quickly normal voltage is restored.
The data center owners also identified another protection scheme that affects the response of data center loads to voltage disturbances. This scheme takes effect if a certain number of voltage disturbances are detected within a set period of time; if the condition is met, the center’s load is transferred to the backup system and must be reconnected to the grid manually. The typical triggering threshold is three voltage disturbances within a minute.
While no significant operating issues were encountered as a result of the incident, NERC noted that “the potential exists for issues in future incidents if the load is not reconnected in a controlled manner.” If more data centers had gone offline at the same time or tried to reconnect simultaneously when the disturbance was over, it could have presented challenges to balancing authorities and transmission operators.
“This incident has highlighted potential reliability risks to the [grid] with respect to the voltage ride-through characteristics of large data center loads,” NERC said. “Similar incidents have occurred in other interconnections with cryptocurrency mining loads as well as oil/gas loads. While these loads are different than the data center loads in this incident, they present the same challenges to the operators and planners of the BES [bulk electric system]. Understanding the changing dynamic nature of load is critical to the future operation of the BES.”
NERC’s latest Long-Term Reliability Assessment, released Dec. 17, 2024, identified data centers and industrial applications as a rapidly growing sector that could cause reliability issues, especially when coupled with the move from traditional generation resources to renewable energy. (See NERC Warns Challenges ‘Mounting’ in Coming Decade.)