Topology
Data Center Layout by NERC in 2024 (from )
Load Increase
Figure ES-1. Total U.S. data center electricity use from 2014 through 2028. (from )
Large Loads Applying to Connect :
- Data Centers: Large volume of requests across the continent, concentrated in the mid-Atlantic, Upper Midwest, Texas and California. Size: 10s-100s of MW
- Cryptocurrency Mining: Large volume of requests across the continent, heavily concentrated in Texas, Southeast and Northeast. Size: 10s-100s of MW
- Electrification: Transportation, buildings, residential. Size: smaller individually, but aggregate can be 100s of MW
- Commercial and Industrial: Electric Arc Furnaces, Oil and Gas, Ammonia Green Hydrogen. Size: Can be 100s of MW to GW
Data center growth :
- ERCOT is forecasting explosive demand growth, driven by data center projects, Bitcoin operations, development in oil- and natural-gas-producing areas, and industrial facilities.
- PJM is experiencing large growth in data centers that are in turn driving higher demand forecasts. Loudoun County, Virginia, in the PJM assessment area, is home to “Data Center Alley,” the largest concentration of data centers in the world. Electrification in transportation, heating, and industrial sectors is also spurring demand growth.
Concerns
Data Centers and Large Industrial Load Growth in large load parcels like data centers and industrial facilities pose various challenges for system planners and operators, in addition to fueling rapid demand growth discussed elsewhere in this report. Other types of large industrial loads include smelters, manufacturing centers, hydrogen electrolyzers, and future electrified mass transit or shipping charging stations. Adding large parcels of load on the system can add new uncertainties to peak and hourly load forecasting. For example, data centers have longer operating hours and require more heating and cooling than other commercial buildings. In Texas, crypto mining facilities have connected in recent years that scale their operations (and thus electricity demand) depending on electricity prices. The behavior of large data centers during normal grid faults is also an emerging concern because customer-initiated automatic disconnecting of sizeable load can cause operating issues. Sudden and unexpected disconnecting load during a grid fault is analogous to well-known IBR performance issues and poses similar reliability risks. Planners and operators need to consider the characteristics of these loads as they begin to proliferate and support NERC RSTC’s Large Loads Task Force to share best practices.
Forecasting Considerations :
- There is no formal forecasting model that is used today to forecast large loads, in particular data centers
- Data informing forecasting is obtained by utilities / ISOs directly from customers and developers seeking to interconnect
- No standardization in data center consumption requirements, interconnection queue, modeling, etc.
Large Data Center Loads – Key Considerations :
- Power Quality: Harmonics, Voltage fluctuations/flicker
- Transient Stability: Voltage recovery, frequency
- Small Signal Stability: Forced oscillations at low frequencies
- Resonance Stability: Exciting torsional modes of nearby units
- Other: Ramping rate, ride-through requirements
Modeling
Statistical power models This paper presents two types of statistical power models that relate CPU usage of Google’s Power Distribution Units (PDUs, commonly referred to as power domains) to their power consumption.
Component based electrical models In this paper, a component based electrical energy consumption modelling approach is presented to identify the losses of different components as well as their interactions to the total electrical energy consumption of the data center.
News
FERC rejects a co-location proposal Chairman Willie Phillips dissented from the decision on the “first of its kind” co-location proposal, saying it could harm national security and grid reliability.
Voltage-sensitive load loss Specifically, when considering data centers and cryptocurrency mining facilities, entities should be aware of the potential for large amounts of voltage-sensitive load loss during normally cleared faults on the BES.
Incident Details:
At approximately 7:00 p.m. Eastern on July 10, 2024, a lightning arrestor failed on a 230 kV transmission line in the Eastern Interconnection, resulting in a permanent fault that eventually “locked out” the transmission line. The auto-reclosing control on the transmission line was configured for three auto-reclose attempts staggered at each end of the line. This configuration resulted in 6 successive system faults in an 82-second period. The protection system detected these faults and cleared them properly. The shortest fault duration was the initial fault at 42 milliseconds, and the longest fault duration was 66 milliseconds. The voltage magnitudes during the fault ranged from .25 to .40 per unit in the load-loss area.
Coincident with this six-fault disturbance, the same local area saw an approximate 1,500 MW of load reduction. None of this load was disconnected from the system by utility equipment; rather, the load was disconnected on the customer side by customer protection and controls. It was determined that the 1,500 MW of load reduction was exclusively data center-type load. The area where the disturbance occurred has a high concentration of data center loads.
Frequency and voltage rose due to the load loss. Frequency rose to a high of 60.047 Hz and settled back to 60.0 Hz in approximately 4 minutes. At the highest level, voltage rose to 1.07 per unit. Operators removed shunt capacitor banks in the local area to return voltages to normal operating values.
Change Log
2025-01-20
- Add item of Component based electrical models
- Rename section title “Other Related Articles” to “News”
- Add a news item of NERC’s incident review about voltage-sensitive load loss
- Improve the item of Large Loads Applying to Connect
- Rename section title “Layout” to “Topology”
- Improve the section of “Load Increase”