Source-linked AI summary
Comparative Analysis of Large Load Interconnection Requirements Across U.S. Transmission Entities
Rafi Zahedi, Rahul Anilkumar, Amin Zamani
TL;DR
Large-load interconnection requirements remain fragmented even as large electrical loads create technical and operational challenges for the grid. This paper compares requirements across eleven U.S. transmission entities and finds substantial differences in coverage and maturity, while identifying data centers as potential grid resources.
Problem
Large-load grid interconnection presents technical and operational challenges, and requirements across entities remain fragmented and variable.
Source: §1 · p. 2Method
Results
LIPA addresses all seven technical categories at least partially, while Dominion and Duke Energy provide detailed requirements across nearly all categories.
Source: §3 · p. 11Takeaways & Limitations
Data centers’ backup generators and battery storage can be treated as dispatchable grid energy sources rather than only as loads.
Source: §3 · p. 11
Abstract
from arXiv · showhide
The rapid growth of large loads, especially data centers driven by artificial intelligence, is creating unprecedented demands on the North American electric grid. Data centers now account for nearly 80% of large load interconnection requests in the Western Interconnection, and U.S. electricity consumption from data centers is projected to reach 325-580 TWh by 2028. Despite this urgency, technical interconnection requirements for large loads have not yet matured and still contain several gaps. This paper presents a comparative analysis of large load interconnection requirements across eleven U.S. entities, including AEP, Ameren, CAISO, Dominion, Duke Energy, ERCOT, LIPA, MISO, PJM, Southern Company, and SPP, under seven technical categories: large load definition, load data and modeling, fault ride-through, load curtailment and shedding, dispatchability, power quality, and reactive power control. The analysis reveals significant variation in both the scope and specificity of requirements. LIPA addresses all seven technical categories at least partially, while Dominion and Duke Energy provide detailed coverage across nearly all categories. The analysis also finds that data center applicants can be viewed as potential grid resources, rather than only as consumers. The findings provide a structured baseline from which utilities, regulators, ISOs/RTOs, and developers can identify gaps, benchmark against peers, and accelerate the development of interconnection standards.
1. INTRODUCTION
Rapid growth in large electrical loads is creating technical and operational challenges for grid interconnection, while requirements remain fragmented across U.S. entities. This paper compares requirements across eleven entities to identify convergence, divergence, and gaps.
1. INTRODUCTION
- Data centers accounted for nearly 80% of large-load interconnection requests across the Western Interconnection.U.S. data-center electricity consumption is projected to reach 325–580 TWh by 2028.Source: §1 · p. 2
- A 230 kV line fault in July 2024 triggered the loss of approximately 1,500 MW of voltage-sensitive data-center load in the Eastern Interconnection.The load transferred to backup generation, causing frequency overshoot and high voltages.Source: §1 · p. 2
- Large-load interconnection requirements remain fragmented, with definitions, ride-through expectations, curtailment, dispatchability, power quality, and reactive power requirements varying by entity.The paper describes grid interconnection as a key issue for data centers and other large loads.Source: §1 · p. 2
- The paper compares technical point-of-interconnection requirements across eleven U.S. transmission entities using public handbooks, guidance, and proposed standards.Its cross-entity analysis assesses requirements maturity and identifies areas of convergence, divergence, and gaps.Source: §1 · p. 2
2. COMPARATIVE FINDINGS OF U.S. TRANSMISSION ENTITIES
The analysis compares published requirements across eleven U.S. transmission entities in seven technical categories. Large-load classifications differ across entities, and planners need accurate load-composition data and suitable studies to assess impacts.
2. COMPARATIVE FINDINGS OF U.S. TRANSMISSION ENTITIES
- The comparison covers eleven entities and seven technical categories, using side-by-side requirements to identify convergence and divergence.Sources include utility handbooks, ISO/RTO guidance, and proposed standards.Source: §2 · p. 2
2.1 Large load definition
- Definitions vary: some entities use a demand threshold, while SPP, Southern Company, and Duke Energy also consider interconnection voltage.The threshold comparison is presented in Fig. 1.Source: §2.1 · p. 3, Figure 1 · p. 3
- 75 MW or greater qualifies as a large load under every surveyed entity; the most common threshold is 50 MW.SPP’s 10 MW threshold applies only to loads connected at 69 kV, while LIPA’s next-lowest threshold is 20 MW.Source: §2.1 · p. 3, Figure 1 · p. 3
2.2 Load data and modeling
- Accurate load-composition data and appropriate simulation tools and study processes are needed to predict facility behavior and assess impacts.The study processes define scope, timeline, and energization acceptance criteria.Source: §2.2 · p. 3
Load composition data
The section describes load-composition data collection and study requirements, then defines fault ride-through as a facility’s obligation and capability to remain connected during voltage and frequency disturbances.
Load composition data
- A NERC Level 3 Essential Action Alert asks entities to collect nine pieces of modeling data from large-load developers.The paper compares data requested in entities’ published documents.Source: Load composition data · p. 3
- All entities require steady-state and short-circuit studies as a baseline; dynamic/transient stability and EMT studies are added as system conditions warrant.Table II compares study types and processes across entities.Source: Load composition data · p. 4, Table II · p. 4
2.3 Fault ride-through
- Fault ride-through is a facility’s capability and obligation to remain connected at the POI through voltage and frequency disturbances.The paper treats voltage and frequency ride-through separately in the following subsections.Source: §2.3 · p. 4
Voltage ride-through
Five entities publish explicit numeric voltage ride-through envelopes, but their required ride-through times diverge outside the shared normal operating interval. MISO’s proposal also specifies active-power boundaries by voltage region.
Voltage ride-through
- Five entities—ERCOT, SPP, LIPA, MISO, and CAISO—publish numeric VRT envelopes; AEP references ITIC/CBEMA and requires ride-through of two consecutive excursions outside 0.92–1.05 pu.The numeric envelopes use time–voltage bands.Source: Voltage ride-through · p. 5, Table III · p. 6
- MISO’s proposal permits no power reduction at 0.90–1.20 pu, proportional reduction at 0.50–0.90 pu, and flexible reduction below 0.50 pu.The framework also requires withstanding three voltage disturbances within 10 seconds and phase-angle changes up to 25 degrees.Source: Voltage ride-through · p. 5, Figure 2 · p. 5
- ERCOT requires continuous operation at 0.90–1.10 pu and at least 90% recovery of pre-disturbance consumption within 2 seconds after voltage exceeds 0.90 pu.During disturbances, current may not exceed 150% of normal maximum for more than 0.5 seconds.Source: Voltage ride-through · p. 5
- SPP requires six fault-clearing attempts within 90 seconds, while Dominion requires ride-through of at least six automatic reclosing attempts of 50–70 ms each.Duke Energy and Southern Company require VRT settings for review and retain authority to mandate adjustments.Source: Voltage ride-through · p. 5
- CAISO requires up to 6.0 s of ride-through at 0.70–0.90 pu, roughly 3× the 2.0 s required by ERCOT, SPP, MISO, and LIPA.In the 1.10–1.20 pu swell band, LIPA and CAISO require 2.0 s, 2–4× more than MISO and SPP.Source: Voltage ride-through · p. 5, Figure 3 · p. 6
Frequency ride-through
FRT requirements differ across entities, while curtailment and shedding provisions vary in coverage and can treat data-center backup resources as grid resources. Dispatchability requirements govern demand changes and oscillatory behavior.
Frequency ride-through
- Five entities publish explicit numeric FRT envelopes; ERCOT requires the largest band, while the other four have identical envelopes.ERCOT also requires continued grid-current consumption and active power within 10% of pre-disturbance levels.Source: Frequency ride-through · p. 6
- MISO’s proposed FRT framework pairs frequency bands with active-power boundaries, including 299-second connection and ±10% power maintenance in extended bands.It also requires ride-through for RoCoF within ±5.0 Hz/s.Source: Frequency ride-through · pp. 6–7, Figure 4 · p. 7
2.4 Load curtailment and load shedding
- Southern Company and CAISO span manual load shed, UVLS, and UFLS; LIPA adds a quantitative UFLS demand-reduction obligation below 59.5 Hz.LIPA requires reduction within 0.30 seconds by at least the lesser of 17% of maximum rated demand or 50% of pre-event demand.Source: §2.4 · pp. 7–8
- Dominion asks data-center applicants whether backup generation or BESS can participate in demand-side management, framing facilities as potential grid resources.The passage identifies this as a way to view data centers as more than consumers.Source: §2.4 · p. 8
2.5 Dispatchability
- Dispatchability requirements govern how quickly and variably a large load may change demand and how it limits oscillatory behavior to preserve grid stability.The supplied passage describes the scope of dispatchability requirements, not entity-specific provisions.Source: §2.5 · p. 8
Fluctuation and ramp-rate
Ramp-rate and fluctuation requirements vary in specificity, while load-oscillation controls are explicit for only some entities. Power-quality coverage is uneven, with harmonics most widely addressed and voltage unbalance least addressed.
Fluctuation and ramp-rate
- Southern Company provides the most detailed fluctuation framework; MISO and LIPA have minor constraints, while three entities lack published numerical fluctuation limits.Southern Company, Duke Energy, MISO, SPP, LIPA, and CAISO have explicit ramp-rate limits.Source: Fluctuation and ramp-rate · p. 8
- Southern Company and LIPA publish FFT-based active-power oscillation limits, while Duke Energy requires minimizing oscillation in its cited inter-area frequency range.LIPA caps adjacent bins at 10 MW, all bins at 20 MW in 0.1–5 Hz, and the 5–55 Hz range at 3.5 MW.Source: Fluctuation and ramp-rate · p. 8
2.6 Power quality
- Power-quality coverage is uneven across harmonics, flicker, and voltage unbalance, with harmonics most widely addressed and voltage unbalance least.The passage identifies these three categories but does not specify entity-level coverage here.Source: §2.6 · p. 9
Harmonic distortion
Harmonic-distortion requirements share a common reference standard, but entity-specific additions vary.
Harmonic distortion
- Six of eleven entities address harmonic distortion, and IEEE 519 is their common standard.Table VIII compares harmonic-distortion requirements across entities.Source: Harmonic distortion · p. 9, Table VIII · p. 9
- Entities vary in what requirements they add beyond IEEE 519.The comparison is presented in Table VIII.Source: Harmonic distortion · p. 9, Table VIII · p. 9
- The supplied comparison identifies six entities addressing harmonics, rather than universal coverage across the eleven surveyed entities.The passage gives the count but does not name the six entities.Source: Harmonic distortion · p. 9
Voltage flicker
Six entities address flicker using IEEE 1453 as a common reference; Southern Company and LIPA set stricter individual emission limits above 35 kV.
Voltage flicker
- Six entities address flicker, with IEEE 1453 as the common reference and Pst = 0.8 and Plt = 0.6 as the most common benchmarks.Table IX compares voltage-flicker requirements.Source: Voltage flicker · p. 10, Table IX · p. 10
- For service above 35 kV, Southern Company and LIPA set individual emission limits of Pst = 0.35 and Plt = 0.25.These are stricter than the most common benchmarks reported for flicker.Source: Voltage flicker · p. 10
- The total cumulative flicker tolerated from all sources at the location remains 0.8/0.6, distinct from the stricter individual limits.The passage distinguishes individual emissions from the system-wide cumulative allowance.Source: Voltage flicker · p. 10
Voltage unbalance
Voltage unbalance is the least commonly specified category, with only Duke Energy and Southern Company providing explicit numeric limits. Reactive-power requirements vary: 0.95 is the most common power-factor threshold, while LIPA and CAISO specify other benchmarks.
Voltage unbalance
- Only Duke Energy and Southern Company provide explicit numeric voltage-unbalance limits, making it the least commonly specified category.Table X compares voltage-unbalance requirements.Source: Voltage unbalance · p. 10, Table X · p. 10
2.7 Reactive power control
- The 0.95 power-factor threshold is most common; LIPA requires at least 0.98, while CAISO specifies a 0.97 lagging to 0.99 leading band.Table XI summarizes reactive-power and power-factor requirements.Source: §2.7 · p. 10, Table XI · p. 10
3. CONCLUSION
The eleven-entity comparison spans seven technical categories and finds uneven maturity in large-load interconnection requirements. It also identifies data centers’ backup generation and batteries as potential grid resources, not only loads.
3. CONCLUSION
- LIPA addresses all seven categories at least partially; Dominion and Duke provide detailed coverage across nearly all, while ERCOT, MISO, Southern Company, and SPP also show strong coverage.The categories include load definition, modeling, ride-through, curtailment and shedding, dispatchability, power quality, and reactive-power control.Source: §3 · p. 11
- MISO’s new Large Load Working Group framework fills many previous gaps, particularly in fault ride-through and post-fault recovery.Source: §3 · p. 11
- Southern Company and LIPA alone publish full quantitative dispatchability frameworks, but their differing limits and measurement methods leave a grid-stability blind spot.Source: §3 · p. 11
- Data centers’ backup generators and increasingly utility-scale batteries can be treated as dispatchable grid energy sources rather than only as loads.Source: §3 · p. 11