Reliability vs. Resilience: Engineering the US Power Grid for Load Growth, Extreme Weather and Inverter-Based Resources

November 11, 2026 | Blog

Part A — Reliability vs. Resilience at a Glance

For most of the last century, US utilities had a clear job description: keep the lights on under every credible single contingency, and keep the number and length of customer interruptions within accepted limits. That is reliability, and the industry measures it well. NERC planning standards test the bulk system against defined contingencies. Resource adequacy studies set a loss-of-load target, and IEEE 1366 gives distribution utilities a common way to count interruptions.

The grid is now being asked to do something different. Load is growing again after fifteen flat years, led by data centers, manufacturing and electrification. Inverter-based resources (IBRs) such as solar, wind and battery storage are replacing synchronous machines, changing how the system behaves in the first second after a fault. Hurricanes, winter storms, wildfire and heat waves are producing outages that last days or weeks rather than minutes or hours. Physical and cyber attacks are no longer hypothetical. None of these sits comfortably inside a traditional N-1 study.

Resilience is the word the industry uses for that wider problem. It is not a replacement for reliability. It covers the high-impact, low-probability (HILP) events that reliability criteria were never designed to cover, and it asks a different question: not "how often do customers go out?" but "when something big goes wrong, how much do we lose, how fast do we recover, and what do we learn?"

This guide is Keentel Engineering's practical view of how US planners, operators and large customers can engineer for both. It explains the definitions, metrics and standards, works through the calculations with real formulas and numbers, reviews what recent US disturbances have taught us, and sets out the hardening, operational and planning measures that deliver the most resilience per dollar.

In plain terms: reliability is how well the grid performs on normal days and for expected failures. Resilience is how well it holds up, and how fast it comes back, on the worst days. A grid can score well on one and poorly on the other, so both need their own targets, metrics and investment cases.

Reliability and resilience compared

Five things every planner should take away

• Resilience starts where reliability statistics stop. IEEE 1366 normalizes away major event days so that utilities can track day-to-day performance. Those excluded days are exactly where resilience lives. A utility can show steady SAIDI improvement while its storm performance gets worse.
• Depth and duration both matter, but not equally. In Keentel's model in Part B, reducing the number of customers who lose power by 10% cuts customer-hours lost by 10%. Restoring 10% faster cuts it by only 7.6% unless the degraded-state period also shrinks. Hardening and restoration capability are complements, not substitutes.
• The cheapest resilience is usually not the most visible. Reclosers, fault location, isolation and service restoration (FLISR) and targeted vegetation work often deliver avoided customer-minutes at a fraction of the cost of undergrounding. Undergrounding still wins in specific places: wildfire ignition zones, storm-surge-free urban cores and repeatedly damaged segments.
• The new grid fails in new ways. Several recent US disturbances involved IBR plants reducing output or tripping for normally cleared faults. Large data center loads have also dropped more than 1,000 MW within seconds. These are reliability problems with resilience consequences, and they require electromagnetic transient (EMT) modeling, ride-through requirements and better monitoring.
• Resilience is organizational as much as electrical. Black start, spares, mutual assistance, operator training, gas-electric coordination and clear decision rights decide how long an outage lasts once the hardware has done what it can.
  
  

The illustrative system used in this guide

All numbers in the worked examples are illustrative and were computed for this guide. They are meant to show method and order of magnitude, not to represent any particular utility.

Figure 1. Left: resilience curves for a 100,000-customer storm zone under three investment cases; the shaded area above each curve is customer-hours lost. Right: illustrative levelized cost per avoided customer-minute for six common hardening options (7% discount rate).

Part B — Engineering Guide

1. Definitions that hold up in a rate case

1.1 Reliability: adequacy plus operating reliability

NERC defines bulk power system reliability through two concepts. Adequacy is the ability of the system to supply aggregate demand and energy at all times, accounting for scheduled and reasonably expected unscheduled outages. Operating reliability is the ability to withstand sudden disturbances such as short circuits or the unanticipated loss of system elements. Adequacy is studied with probabilistic models of generation and load. Operating reliability is studied with deterministic contingency analysis under TPL-001 and in real time by Reliability Coordinators and Transmission Operators.

At the distribution level, reliability is described by IEEE 1366 indices: how often the average customer is interrupted and for how long. These are performance outcomes, measured after the fact, and they are what most state commissions use for performance-based regulation.

1.2 Resilience: a capability, not a statistic

The most widely used US definition comes from FERC's 2018 grid resilience proceeding. It describes resilience as the ability to withstand and reduce the magnitude or duration of disruptive events, including the capability to anticipate, absorb, adapt to and rapidly recover from such events. The National Infrastructure Advisory Council (NIAC) framework, which NERC also references, breaks the same idea into four parts:

• Robustness: the ability to absorb shocks and keep operating.
• Resourcefulness: the ability to detect and manage a crisis as it unfolds.
• Rapid recovery: the ability to restore service quickly and in a controlled way, given the extent of damage.
• Adaptability: the ability to learn from each event and improve.

The practical difference is that reliability can be met by satisfying a criterion, while resilience has to be built and exercised. A transmission system can pass every TPL-001 contingency and still collapse when a hurricane takes out 200 structures, because no planning criterion asks it to survive that.

1.3 The resilience trapezoid

Engineers describe a resilience event with a performance curve that has five phases:

• Pre-disturbance (anticipate): forecasting, pre-staging crews, reconfiguring, committing extra units, de-energizing in wildfire conditions.
• Degradation (absorb): performance falls as elements fail. Robustness limits the slope and depth.
• Degraded state (respond): the system sits at or near its lowest point while damage is assessed. Resourcefulness, including situational awareness, switching and islanding, shortens this phase.
• Restoration (recover): performance climbs as crews repair and re-energize. Crew numbers, spares, access and black start capability set the slope.
• Post-restoration (adapt): the system returns to normal, ideally stronger, through lessons learned, standards changes and targeted investment.

The area between the normal performance level and the curve is the event's total loss. Measured in customer-hours, element-days or MVA-days, it is the closest thing the industry has to a single resilience metric.

2. Measuring reliability

2.1 IEEE 1366 indices

The core distribution indices are defined in IEEE 1366:

where Ni is the number of customers interrupted by event i, ri is the restoration time in minutes, and NT is the total number of customers served. MAIFI counts momentary interruptions, which matter more as reclosing and automation increase.



Worked example 1 — annual indices. A utility serves 250,000 customers. Over a year it records 300,000 customer interruptions and 36,000,000 customer-minutes of interruption (CMI).

• SAIFI = 300,000 / 250,000 = 1.2 interruptions per customer
• SAIDI = 36,000,000 / 250,000 = 144 minutes per customer
• CAIDI = 144 / 1.2 = 120 minutes per interruption

2.2 Major event days and the 2.5β method

IEEE 1366 separates major event days (MEDs) from normal days so that a single hurricane does not swamp a year's trend. The threshold uses the natural logarithm of five years of daily SAIDI values, excluding days with zero SAIDI:

where α is the mean and β the standard deviation of ln(daily SAIDI). Any day whose SAIDI exceeds TMED is a major event day.

Worked example 2 — MED threshold. Keentel generated five years of synthetic daily SAIDI for a mid-sized utility, log-normally distributed with seven embedded storm days. The log statistics are α = −1.421 and β = 1.020, so:

• TMED = exp(−1.421 + 2.5 × 1.020) = 3.09 minutes per day
• 18 days in five years exceed the threshold, 5 of them in the most recent year
• The most recent year's SAIDI is 171.8 minutes including MEDs and 138.5 minutes excluding them

The 33 minutes removed by the MED rule are a fifth of the total. That share is the part of customer experience that normalized reliability reporting does not show, and it is growing in many service territories. Keentel recommends that utilities report both values side by side, plus the customer-minutes on MEDs as a separate resilience indicator.

2.3 Resource adequacy metrics

Bulk system adequacy uses probabilistic metrics:

Loss of load expectation (LOLE) is the expected number of hours or days per year with insufficient supply. Expected unserved energy (EUE) is the expected energy not delivered. The common "0.1 days per year" (one day in ten years) target is an industry convention adopted by many regions, not a NERC standard, and planners increasingly recognize that LOLE alone hides the size and duration of shortfalls. EUE, normalized EUE, and the hourly pattern of risk now matter as much as the headline LOLE, especially in winter. Effective load-carrying capability (ELCC) is the matching tool for valuing variable and energy-limited resources.

3. Measuring resilience

3.1 There is no standard metric yet



Unlike reliability, resilience has no IEEE or NERC index that every utility reports. That is partly because resilience depends on the threat: the right metric for a hurricane differs from the right metric for a cyber intrusion or a cold snap. Surveys of international utilities show that a little more than half distinguish formally between reliability and resilience, and fewer than half have resilience-specific frameworks. In practice, US utilities and regulators use a combination of the measures below.

NERC's transmission resilience analysis uses Transmission Availability Data System (TADS) outage data to build outage, restore and performance curves for every large event. For Hurricane Helene in 2024, NERC reported 431 automatic transmission outages, the largest event in its 2015–2024 record, yet the industry restored 95% of outaged elements within about a week. That combination is a good illustration: very large size and nadir, but strong recovery, made possible by tens of thousands of mutual-assistance personnel.

3.2 Resilience loss as an area



For a single event, the loss is the integral of lost performance over time:

For a simple trapezoid with depth D (fraction of customers out at the nadir), degradation time td, degraded-state time ts and restoration time tr, the integral becomes:

A 75 kVA single-phase pad-mount would see 80 kVA, about 107% of nameplate, at the short winter or evening peak. IEEE C57.91 thermal models normally allow short peaks above nameplate when the daily cycle and ambient are favorable, and many utilities permit peak loading of roughly 120–140% for residential units. The design therefore uses 25 transformers of 75 kVA, with a planned upgrade path to 100 kVA where EV adoption is high.

Worked example 3 — three investment cases for a storm zone. A 100,000-customer coastal zone is hit by a major hurricane. Keentel compared three cases:

Hardening cuts loss by 62%. Adding automation and microgrids takes the total reduction to 77%, mainly by shortening the degraded state and the restoration.

3.3 Hardening versus faster restoration

The trapezoid formula shows why both levers are needed. Starting from the baseline case:

• 10% less depth (hardening that keeps 10% more of the affected customers in service) reduces loss by 10.0%, because depth multiplies the whole area.
• 10% faster restoration (tr shortened by 10%) reduces loss by only 7.6%, because the degradation and degraded-state parts are unchanged.
• 10% faster restoration plus a 10% shorter degraded state reduces loss by 9.5%.

NERC's "rerunning history" analysis of Hurricane Ida transmission outages reached a similar conclusion with real data: removing 10% of outages reduced the nadir and MVA-days lost by about 10%, while faster restoration reduced MVA-days lost and time to 95% restoration but left the nadir unchanged. The practical message is that the outages that both happen early and take longest to restore are the most valuable ones to prevent.

3.4 Putting a dollar value on resilience



Unserved energy converts customer-hours into an economic loss:

Worked example 4 — societal cost of the storm. With an average load of 1.6 kW per customer (a blended residential and small commercial mix), the three cases in Worked example 3 give:

Value of lost load (VoLL) is the most uncertain input in any resilience case. US interruption cost studies, including the Interruption Cost Estimate (ICE) Calculator maintained by Lawrence Berkeley National Laboratory, show per-kWh costs that differ by orders of magnitude between residential, small commercial and large industrial customers. They also show that most survey data covers outages of up to about 16 hours, not multi-day events. Long outages have costs that rise faster than linearly: spoiled inventory, lost production, displaced households, and health and safety impacts. Keentel always presents resilience benefits as a range and states the VoLL basis explicitly. A benefit-cost ratio that only works at the top of the VoLL range should be flagged as such.

3.5 Risk: probability times consequence



Resilience planning ranks threats by expected annual loss:

Worked example 5 — a utility threat register (illustrative).

Expected annual loss is a starting point, not the answer. For true HILP events, regulators and boards are often more concerned with the tail than the mean: a 1% chance of a $5 billion event has the same expected loss as a 50% chance of a $100 million event, but society treats them very differently. Keentel therefore pairs expected-loss ranking with scenario stress tests, such as a "30-day event" in which fuel, crews, spares and communications are all constrained at once.

4. What is different about the new grid

4.1 Load growth and very large loads

After roughly fifteen years of flat demand, US load forecasts are rising sharply. Lawrence Berkeley National Laboratory's December 2024 report to Congress estimated that data centers used about 4.4% of US electricity in 2023 and could reach between 6.7% and 12% by 2028. Transmission planners in several regions now see interconnection requests for single campuses of 300 MW to more than 1 GW, often with in-service dates shorter than the time it takes to build a 345 kV line.

Large loads raise reliability and resilience issues that legacy planning assumptions did not cover:

• Voltage-sensitive load loss. Data center power systems are designed to protect the IT load. When a transmission fault depresses voltage, uninterruptible power supply (UPS) systems and transfer schemes can move the whole facility to batteries or on-site generation in a few cycles. NERC has documented an event in which roughly 1,500 MW of data center load in the Eastern Interconnection dropped off almost simultaneously for a normally cleared transmission fault. To the grid, that is a large generation surplus: frequency and voltage rise, and the load returns on its own schedule, not the operator's.
• Ride-through. Unlike generators, most loads have no ride-through obligation. Regions are now developing large-load interconnection requirements that address voltage and frequency ride-through, ramp rates on return, and data sharing.
• Visibility. Many large loads are invisible to operators beyond a revenue meter. Telemetry of real and reactive power, status of on-site generation and UPS mode, and phasor measurement unit (PMU) data at the point of interconnection should be standard.
• Protection coordination. Facility protection must coordinate with utility reclosing and with the load's own transfer schemes, so that a utility reclose does not catch a large motor or power-electronic load out of phase.
• Emergency programs. Large loads that are on backup generation during an emergency may not provide the load relief that underfrequency load shedding (UFLS) and manual load-shed plans assume. Conversely, a flexible large load can be one of the most effective emergency resources a Balancing Authority has, if it is contracted and tested.
• Fast power swings. AI training workloads can swing by hundreds of megawatts in seconds, which can excite oscillations and stress voltage control. On-site battery storage and rack-level energy buffering are the usual mitigations.



Worked example 6 — frequency effect of a large-load trip. A 1,500 MW campus drops to backup power during a fault. The steady-state frequency deviation follows the interconnection's frequency response characteristic β, and the initial rate of change of frequency (RoCoF) follows its stored kinetic energy:

Neither number is dangerous by itself. The risk comes from combinations: several large loads in the same electrical area, a simultaneous IBR response to the same fault, low-inertia hours, or the load returning all at once during recovery. These effects can only be assessed properly with dynamic and, near the point of interconnection, EMT models of the load's power electronics and controls.

4.2 Inverter-based resources and new stability problems

IBRs connect through power electronics. Their behavior during and after a fault is set by control software, not by the physics of a rotating machine. That brings capabilities synchronous machines lack, such as very fast active and reactive power control. It also brings failure modes that the industry did not anticipate.

Since 2016, NERC has published a series of disturbance reports in which solar PV and wind plants reduced output or tripped for faults that were cleared normally by transmission protection. In a 2021 series of Southern California events, several hundred megawatts of solar PV output was lost in each of four incidents. In the March 2022 Texas Panhandle events, wind plants reduced output for two separate phase-to-phase faults on 345 kV lines. Causes identified across these reports include:

• Plant controller interactions and dynamic active power reduction after the fault.
• AC overvoltage tripping on fault recovery.
• Instantaneous inverter protection operating on measured frequency or phase jumps.
• Sub-synchronous oscillation protection.
• Auxiliary power and pitch-system failures in wind turbines.
• Settings that did not match the plant's actual capability, or EMT models that did not match the as-built plant.

FERC Order No. 901 (2023) directed NERC to develop standards on IBR data sharing, model validation, planning and operational studies, and performance requirements. The resulting work includes PRC-029-1, the IBR frequency and voltage ride-through standard approved by FERC in 2025, PRC-030-1 for unexpected IBR event mitigation, and updates to modeling, monitoring and planning standards. IEEE 2800-2022 sets uniform technical minimum capabilities for transmission-connected IBRs, and many US transmission owners have adopted it, in whole or in part, in their interconnection requirements.

The classical stability categories of rotor angle, voltage and frequency have been extended to include resonance stability (sub-synchronous resonance and torsional interactions) and converter-driven stability (control interactions between converters and with a weak grid). Converter-driven problems typically appear in weak areas with a low short-circuit ratio. They can show up as oscillations from a few hertz to several hundred hertz, and positive-sequence RMS models often cannot reproduce them.

4.3 Extreme weather and interdependence

Weather is the dominant cause of large US outages. Recent events show different failure patterns:

• Hurricanes destroy distribution poles and transmission structures and flood substations. In Puerto Rico after Hurricanes Irma and Maria in 2017, the transmission and distribution systems were so badly damaged that full restoration took about eleven months. Hurricane Ida in 2021 took all eight transmission lines serving New Orleans out of service. Hurricane Helene in 2024 produced the largest transmission outage event in NERC's ten-year record.
• Extreme cold affects generation and fuel supply more than wires. In Winter Storm Uri (February 2021), ERCOT shed up to about 20,000 MW of load and came within minutes of a system-wide collapse. More than 200 people died and about 4.5 million homes and businesses lost power. In Winter Storm Elliott (December 2022), the FERC–NERC inquiry found about 90.5 GW of unplanned generation outages at the peak, roughly 13% of projected available capacity in the Eastern Interconnection. About 55% of outages, derates and failures to start were caused by freezing and fuel problems. Natural gas units made up the largest share, and many units froze at temperatures above their documented minimum operating temperature.
• Wildfire is both a hazard to the grid and, in parts of the West, a hazard the grid can cause. It leads to de-energization (public safety power shutoffs, or PSPS), smoke-related insulator flashovers, damage to wooden poles, and air-intake problems at generators and synchronous condensers.
• Heat waves combine high load with reduced line ratings, derated thermal generation and reduced transformer capacity.
• Flooding damages substations and control houses. Raising control equipment and building flood walls are often more cost-effective than relocating a station.

Interdependence makes these events worse. In some US regions, natural-gas-fired plants make up more than 80% of winter peak capacity, and gas production and pipeline compression depend on electricity. Telecommunications, water treatment and fuel distribution all need power to restore power. NERC's cold weather standard EOP-012 now requires generator owners to implement freeze protection measures and cold weather preparedness plans, and TPL-008-1 adds benchmark extreme heat and cold events to transmission planning assessments.

4.4 Physical, cyber and space-weather threats

• Physical attacks. In December 2022, gunfire damaged two distribution substations in Moore County, North Carolina, leaving roughly 40,000–45,000 customers without power for several days. The event led FERC to direct a review of CIP-014, the physical security standard for critical transmission stations. Resilience measures include perimeter protection, ballistic barriers for transformers, spare transformer strategies and mobile substations.
• Cyber threats. The NERC CIP standards cover bulk electric system cyber assets, and CIP-013 addresses supply chain risk. As more control moves to software, including inverter firmware, distributed energy resource management systems (DERMS) and cloud-hosted tools, the boundary of what counts as critical keeps expanding.
• Geomagnetic disturbances. The March 1989 geomagnetic storm collapsed the Hydro-Québec 735 kV system in about 90 seconds and left about six million people without power for roughly nine hours. TPL-007 requires US planners to assess GMD vulnerability and develop corrective action plans.

4.5 Transmission and distribution coordination

Most new resources are connected at the distribution level, while the bulk system is still operated as if distribution were a passive load. That gap affects resilience in several ways:

• Visibility. Rooftop PV, behind-the-meter storage and electric vehicle charging change net load in ways the transmission operator cannot see directly. Distribution-connected PV tripping for transmission faults has added to the generation lost in several Western events.
• Data exchange. The IEC Common Information Model (CIM) gives transmission and distribution models a common format, which makes it practical to exchange network models and DER aggregations.
• Speed. Real-time contingency analysis typically runs on state estimator snapshots every few minutes. Oscillations, fast ramps and IBR control interactions happen in seconds or less. PMU data at 30 or 60 samples per second, and linear state estimators built on it, close part of that gap.
• Power quality. Harmonics and supraharmonics from power electronics, and resonances between capacitor banks and cable capacitance, are now routine planning concerns. IEEE 519 limits still apply at the point of common coupling.
• Coordinated emergency response. UFLS, manual load shedding and restoration plans need to account for feeders that now export power, and for DER that will reconnect automatically after an outage under IEEE 1547-2018 settings.

5. Getting more from existing wires: grid-enhancing technologies

Building new transmission takes seven to ten years in much of the US. Grid-enhancing technologies (GETs) increase the capability of existing assets in months and add operational flexibility during emergencies. The three main families are dynamic line ratings, advanced power flow control and topology optimization.

5.1 Dynamic and ambient-adjusted line ratings

Overhead line ratings are traditionally static. They are calculated for conservative weather, typically a hot summer afternoon with a light crosswind and full sun. FERC Order No. 881 requires transmission providers to use ambient-adjusted ratings (AARs) that update with forecast air temperature, plus seasonal ratings. Dynamic line ratings (DLR) go further by also using measured or forecast wind and solar conditions.

The rating comes from the IEEE 738 steady-state heat balance: the current at which heat gained from resistive losses and the sun equals heat lost by convection and radiation, at the conductor's maximum operating temperature:

Worked example 7 — DLR on a 230 kV line. Keentel applied the IEEE 738 method to a 230 kV line with 795 kcmil 26/7 ACSR "Drake" conductor (28.1 mm diameter), a 100 °C maximum operating temperature, emissivity 0.7, absorptivity 0.9 and 300 m elevation.

Two lessons follow. First, DLR usually adds 10–35% capacity, and the gains are often largest during windy storms and cold fronts, which is exactly when generation patterns shift and transmission is under stress. Second, the last row shows that a static rating is not always conservative. On a hot, still afternoon, the line's real capability can fall well below its static rating. DLR is a safety tool as well as a capacity tool, and that matters in heat waves.

5.2 Power flow control and topology optimization

Advanced power flow controllers, including modular static synchronous series compensators (SSSCs) and phase-shifting transformers, push flow away from overloaded paths onto underused ones. Topology optimization software finds switching configurations, such as opening a 138 kV breaker or splitting a bus, that relieve congestion or post-contingency overloads. Both give operators extra tools when a storm has removed several elements and the remaining network is stressed.

6. Hardening the infrastructure

6.1 Transmission structures and conductors

• Structure loading. Many older lines were designed to the minimum loading districts of the National Electrical Safety Code (NESC). Rebuilding critical lines with steel or concrete poles designed for extreme wind and ice, adding anti-cascade structures at intervals, and replacing aging wood are the most common transmission hardening measures. The Gulf Coast experience after 2020–2021 shows how structure failures dominate restoration time.
• Advanced conductors. Reconductoring with high-temperature, low-sag (HTLS) or other advanced conductors can roughly double a corridor's thermal capacity on existing structures, usually far faster than building a new line. Each project needs a structural check, a sag and clearance check at the higher operating temperature, and a review of terminal equipment ratings.
• Insulation and shielding. Upgrading insulation in contaminated or smoke-prone areas and adding or improving shield wires and line surge arresters reduces flashovers.
• Substations. Typical measures are raising control equipment above design flood elevation, building flood walls, hardening control-house roofs and doors against wind, and seismic qualification of equipment to IEEE 693 in seismic zones. Mobile transformers and mobile substations at 115–230 kV are valuable spares.

6.2 Distribution hardening

• Pole replacement and upgrades to higher strength classes on backbone feeders.
• Vegetation management on a shorter cycle, and removal of hazard trees outside the right-of-way, which cause many storm outages.
• Covered (tree-wire) conductors reduce faults from vegetation contact and wildfire ignition risk at a fraction of the cost of undergrounding. They need suitable hardware and lightning protection.
• Undergrounding removes exposure to wind, ice and vegetation, but costs several times more per mile, takes longer to repair when it fails, and is vulnerable to flooding and storm surge. It is most justified in high-consequence wildfire areas, dense urban cores and the last spans that repeatedly fail.
• Sectionalizing and automation. Reclosers, automated switches and FLISR shrink the number of customers affected by each fault and restore unfaulted sections automatically in seconds to minutes.

6.3 Wildfire-specific measures

In wildfire-prone areas, the grid has to avoid starting fires as well as survive them. Mitigation includes fast-trip and sensitive ground fault settings during high-risk conditions, disabling reclosing on high-risk feeders, covered conductors, undergrounding the highest-risk segments, weather stations and fire cameras, and PSPS. PSPS is effective at preventing ignitions but is itself a planned outage, so reducing PSPS scope through sectionalizing, microgrids and hardening is a resilience goal in its own right.

7. Monitoring, protection and automation

• Advanced metering infrastructure (AMI) reports outages and voltage at the customer level, which speeds damage assessment and shows which customers are still out after crews think a feeder has been restored.
• Wide-area monitoring, protection and control (WAMPAC) uses synchronized phasor measurements under IEEE C37.118 to detect oscillations, islanding and voltage instability in real time. After the 2025 Iberian Peninsula blackout, the European expert panel recommended much wider PMU coverage. US operators that already have it use it to locate oscillation sources in IBR-rich areas.
• Distribution fault anticipation (DFA) and other waveform analytics detect pre-failure signatures such as failing clamps, arcing in vegetation contact and incipient cable faults, before they become outages or ignition sources.
• Remedial action schemes (RAS) extend the capability of the system for specific contingencies, but they add complexity. Several large disturbances have involved a RAS that was armed differently from what the operators or the real-time model assumed. Every RAS needs clear arming procedures, model representation and periodic review.
• Protection for low-fault-current systems. As IBRs replace synchronous machines, fault currents fall and become more controlled. That affects distance, overcurrent and directional elements. Protection studies need IBR fault models that reflect the actual inverter controls, not generic voltage-behind-impedance sources.

8. Local resources: microgrids, storage and flexible demand

8.1 Microgrids for critical loads

A microgrid with local generation, storage and controls can keep a hospital, water plant, emergency operations center or community shelter energized when the bulk grid is down. Relevant US standards include IEEE 1547-2018 (DER interconnection, including intentional islanding) and IEEE 2030.7 and 2030.8 (microgrid controller specification and testing). The design questions are how long the microgrid must island, which loads it serves, how it resynchronizes, and how its protection works with much lower fault current in island mode.

Worked example 8 — sizing storage for a 72-hour critical load. A 2 MW critical facility must be served for 72 hours without the grid. Keentel modeled an hourly energy balance with 95% one-way battery efficiency and a 20% minimum state-of-charge reserve:

Solar cuts the storage requirement by about 78% on clear days, but only by about 28% under storm skies. A practical design therefore combines storage for fast response and grid-forming control, PV for daytime energy, and a dispatchable backup generator with fuel for several days. The storage provides the stable island voltage and frequency, and the generator covers the long tail.

8.2 Battery storage on the bulk system

Utility-scale battery energy storage systems (BESS) now provide fast frequency response, voltage support and, with grid-forming inverters, black start capability. In several recent US disturbances, battery response contributed to faster frequency recovery. For resilience, the key specifications are the state of charge held in reserve during forecast extreme events, grid-forming capability, ride-through to IEEE 2800 and PRC-029, and whether the plant can energize a transmission path from black.

8.3 Demand response and virtual power plants

Demand response, distributed storage and flexible loads aggregated into virtual power plants (VPPs) can shave peaks during extreme heat or cold and reduce the load that must be restored after an outage. Advanced distribution management systems (ADMS) and DERMS coordinate these resources with feeder switching and voltage control. For resilience, the question is whether the resource will be available and controllable during the event. Behind-the-meter batteries may be fully discharged by the customer, and communications may be down.

9. Modeling the new grid: EMT and model quality

Positive-sequence RMS stability programs remain the workhorse for large-system planning, but they cannot represent fast inverter controls, phase-locked loop dynamics, protection functions inside the inverter, or sub-synchronous interactions. Many US transmission owners and ISOs now require an EMT model for every new IBR plant, and increasingly for large power-electronic loads.

A practical EMT model requirement covers the following:

• Real-code representation of the inverter and plant controller, with the same firmware version as the equipment that will be installed.
• All protection functions that can trip the plant or reduce its output, including instantaneous over-voltage, frequency and phase-jump protection.
• Correct representation of plant-level controls, collector system, main power transformer and reactive devices.
• Validated performance across a defined range of short-circuit ratios, including weak-grid conditions.
• Compatibility with the transmission owner's simulation tool versions, plus documentation, parameter lists and benchmark results.
• A model verification process before the model is accepted, and an as-built parameter update once commissioning settings are final.

Keentel's experience is that a large share of first-submission EMT models fail at least one of these checks. The most common problems are generic controls in place of real code, missing protection functions, and settings that differ from the commissioned plant.

10. Data, analytics and AI

Machine learning is useful in resilience work where there is plenty of data and a clear prediction target:

• Supervised learning for outage prediction before a storm (which feeders, how many customers, how many crews), vegetation risk from LiDAR and satellite imagery, and equipment failure probability.
• Unsupervised learning for anomaly detection in PMU, AMI and DFA data.
• Reinforcement learning for switching and restoration sequencing, still mainly in research and pilot use.
• Large language models for searching procedures, summarizing event logs and drafting reports. Operational data and critical infrastructure details should not be shared with public AI services, and outputs must be checked by an engineer.

AI does not replace physics-based studies. It is best used to focus scarce engineering attention, such as which feeders to harden first or which oscillation to investigate, and to speed up the operator's picture of the system during an event.

11. Operational and organizational resilience

Hardware sets how much damage an event causes. Organizations decide how long it lasts.

• Black start and restoration. NERC EOP-005 requires Transmission Operators to keep restoration plans built around black start resources, and to test them. Winter Storm Elliott showed that black start units share the same fuel and freezing risks as the rest of the fleet. Diversity of black start fuel, cold weather testing, and battery or hydro black start options all reduce that common-mode risk. Restoration plans should also cover energizing long cables and IBR-rich areas, operating FACTS devices at low short-circuit levels, and synchronizing islands.
• Situational awareness. Operators need to see the system's true state, including RAS arming status, DER output and large-load status. Several disturbances in recent years were made worse because real-time contingency analysis was not set up to alarm on a condition the operators did not know about.
• Training. Operator training simulators should include multi-entity restoration, islanding, IBR-driven oscillations and large-load events, not only the classic loss-of-generation drills.
• Spares and supply chain. Lead times for large power transformers, high-voltage breakers and some protection equipment now run to years. Shared spare programs, standardized specifications and mobile substations reduce exposure. Supply chain cyber risk is covered by CIP-013.
• Mutual assistance. The restoration speed after Hurricane Helene relied on tens of thousands of crew members from across North America. Pre-arranged mutual assistance agreements, staging sites and logistics plans are core resilience assets.
• Workforce. Protection engineers, EMT modelers and experienced system operators are scarce. Documented procedures, cross-training and partnerships with universities and consultants protect institutional knowledge.
• Gas-electric coordination. Operators should know which plants have firm fuel, dual-fuel capability and on-site storage, and should coordinate with pipelines before forecast cold events.

12. A risk-based framework for choosing investments

12.1 The process

Keentel uses a seven-step framework that fits naturally into an integrated resource plan, a transmission planning cycle or a state resilience filing:

• Identify threats relevant to the service area: hurricane, ice, extreme cold, heat, wildfire, flood, earthquake, physical and cyber attack, GMD.
• Model hazard intensity for each threat, such as wind speed return periods, ice thickness, flood depth or fire weather index.
• Assess vulnerability with fragility curves for poles, structures, substations and generation, and with system models showing how damage turns into customer outages.
• Estimate consequences in customer-hours, critical load lost, unserved energy and societal cost, using a stated VoLL range.
• Evaluate options for how each reduces probability, depth, duration or consequence.
• Rank and select by cost per unit of risk reduced, subject to constraints such as critical facilities, equity and regulatory direction.
• Measure and adapt by tracking performance in real events and updating the models.

12.2 Ranking options by cost per avoided customer-minute

A simple, transparent ranking divides each option's levelized annual cost by the customer-minutes it avoids each year:

Worked example 9 — ranking six options (7% discount rate, illustrative costs).

Automation and vegetation work usually rank first. That does not mean microgrids and undergrounding are poor investments. Customer-minutes treat every customer equally, while a microgrid serving a hospital or water plant protects far more societal value per minute, and undergrounding in a wildfire zone avoids ignition risk that CMI does not capture. Keentel therefore uses CMI ranking as a first screen, then applies weighting for critical customers, wildfire risk reduction and equity, and reports both results.

12.3 Planning beyond N-1

US planning standards already include some resilience scenarios:

• TPL-001 requires assessment of extreme events, such as the loss of all lines on a common corridor or an entire substation. Mitigation is not mandatory, but the cost of actions to reduce the consequences must be evaluated.
• TPL-007 covers benchmark geomagnetic disturbances.
• TPL-008-1 adds benchmark extreme heat and cold events, with corrective action plans where performance criteria are not met.
• FERC Order No. 1920 requires long-term regional transmission planning over at least 20 years with multiple scenarios, including extreme weather.

A resilience-aware planning study adds scenarios built from real events in the region, such as the loss of a coastal transmission corridor, a multi-day cold snap with gas curtailment, or a wildfire outage of a double-circuit 230 kV path. It then tests both the deterministic performance and the restoration path back to normal.



13. US standards map for reliability and resilience

Three lessons stand out:

• The loop roughly halves the subdivision's outage minutes by turning repair time into switching time.
• Automation on a single-source loop adds little once manual switching is reasonably fast, because upstream outages now dominate. (Restoration within five minutes would also count as momentary under IEEE 1366, so those minutes would leave SAIDI entirely.)
• The second source is where the big gain lies. It only works if the second feeder is truly independent — a different substation bus or substation, a different route — and has capacity to pick up the load.

A fair comparison must also consider that a radial design would rarely put all 200 homes on one lateral. Split into four radial laterals, the cable-fault contribution falls to about 6.1 minutes — still twice the loop's 3.1 minutes, and with a six-hour outage for every affected customer rather than 90 minutes.

9.3 Frequency vs. duration

A single-source loop does not reduce how often customers are interrupted by cable faults (SAIFI) by much: the faulted half still loses power until switching. In the model, SAIFI falls only from 0.32 to 0.29. The loop's value is in duration. Customers notice both, but long outages drive complaints, spoiled food and frozen pipes.

10. Automation and FLISR

Fault location, isolation and service restoration (FLISR) uses automated switches or fault interrupters, sensors and communications — managed locally or by an advanced distribution management system (ADMS) — to isolate a faulted section and restore the healthy sections automatically, typically within one to five minutes.

For underground systems, automation lives in pad-mounted switchgear at the main branch points: motor-operated load-interrupter switches and vacuum fault interrupters with electronic controls (IEEE C37.60), voltage and current sensors, and radio or fiber communications. Individual residential loop-feed transformers are rarely automated; instead, automated switchgear divides a large development into zones, and each zone is a small manual loop.

Design considerations for automated loops:

• Capacity checks before transfer. The scheme must confirm the alternate source can carry the transferred load, using real-time or seasonal limits.
• Protection settings after transfer. Fault-interrupter settings that coordinated in the normal configuration may not coordinate after transfer. Use adaptive setting groups.
• Cybersecurity and communications reliability. Utility distribution automation is outside NERC CIP but subject to utility security programs; campus owners need their own controls.
• Arc-flash. Remote operation removes workers from the switching location, a safety gain in itself.

11. Grounding and the neutral

US distribution is predominantly four-wire multigrounded wye: the neutral is carried with the phase conductors and grounded at the substation, at every transformer and at intervals along the line (NESC Rule 96 sets minimum grounding for multigrounded systems). In URD, the cable's concentric neutral serves as the neutral conductor and part of the grounding system, and a ground rod or ring is installed at each pad-mount.

Practical points:

• Neutral sizing. Single-phase URD loops normally use a full neutral (equal ampacity to the phase), because the neutral carries the full load current back. Three-phase cables may use a one-third neutral where the load is balanced.
• Neutral corrosion. Older bare concentric neutrals in corrosive soils lost strands over time and raised fault and touch voltages. Jacketed cable is now standard.
• Fault return. A ground fault on a URD loop returns mainly through the concentric neutral, which keeps fault current high enough to blow the fuse quickly. Broken neutrals are a recognized cause of slow-clearing faults and elevated touch voltages; cable testing should include neutral checks.
• Customer side. The service neutral bonds to the customer's grounding electrode system under NEC Article 250; stray-voltage concerns on farms and pools trace back to neutral-to-earth voltage on multigrounded systems.

12. Arc-flash and switching safety

Operating load-break elbows, replacing bayonet fuses and working in secondary compartments are among the most hazardous routine tasks in URD work.

• Primary compartment. Utility workers follow NESC Part 4, Rule 410, which requires arc-flash assessment and provides tables for typical distribution tasks; non-utility workers on customer-owned systems follow NFPA 70E. IEEE 1584-2018 calculates incident energy for three-phase equipment from 208 V to 15 kV; it does not model single-phase arcs, so single-phase 7.2 kV and 120/240 V tasks need the NESC or NFPA 70E table methods or engineering judgment.
• Secondary compartment. A fault on the 240 V side of a 75 kVA unit draws about 14.7 kA, but on the primary this is only about 14,700 × 240 / 7,200 ≈ 490 A — a current the bayonet fuse may take seconds to clear. Long clearing times mean high incident energy at low voltage. Many utilities require de-energizing the transformer, or the highest PPE category, for secondary compartment work.
• Switching orders. Every loop reconfiguration should follow a written, verified switching order with lockout/tagout, voltage testing and grounding of the isolated section before cable work begins.
• Elbow operation. Load-break elbows interrupt load, not fault current. Closing an elbow into a faulted section is limited by its fault-close rating; isolate and test the section before re-energizing.

13. Electrification and DER

13.1 EVs and heat pumps

Electrification raises both peak demand and coincidence. Practical responses in loop design:

• design transformer pads, secondary pedestals and conduit for the next transformer size;
• limit homes per transformer (six to eight rather than ten to twelve) for new all-electric developments;
• size the loop for the future reconfigured load (Worked example 2's 1,800 kVA case);
• work with the utility on managed charging programs, which can shift EV load and protect transformers far more cheaply than larger equipment.

13.2 Rooftop solar and batteries

Residential solar under IEEE 1547-2018 and UL 1741 SB can raise secondary voltage at midday and reverse power flow through the pad-mount transformer.

• Voltage. On long secondaries with many systems, voltage can exceed the 126 V upper limit of ANSI C84.1 Range A. Smart-inverter volt-var and volt-watt functions help; short secondaries and adequately sized transformers help more.
• Loop reconfiguration. With the open point moved, the same DER is connected farther from the source, so voltage rise and hosting capacity change. Hosting capacity studies should cover the reconfigured state.
• Protection. Small inverter-based DER contributes little fault current, so fuse coordination is rarely affected. However, inverters must cease to energize on loss of the utility source, which matters for crew safety when a loop section is isolated.
• Batteries and backup. Home batteries with backup capability isolate from the grid during outages through transfer equipment; they reduce the customer impact of outages but do not change utility reliability indices.

14. Resilience: flood, wind and wildfire

Underground distribution removes most exposure to wind, ice and vegetation. In a hurricane or ice storm, URD subdivisions often keep power while overhead areas do not — until the overhead feeder that supplies them fails. Undergrounding the development does not undergird the feeder, which is another argument for two-source loops.

Underground systems have their own hazards:

• Flooding. Pad-mounted transformers and switchgear in floodplains should be placed above the design flood elevation or specified as submersible. Saltwater from storm surge damages equipment that survives fresh-water flooding.
• Dig-ins. Third-party excavation damage is a leading cause of URD faults. Warning tape, accurate as-builts and "call before you dig" (811) participation reduce it.
• Wildfire. Underground cable eliminates ignition risk along those segments; pad-mounted equipment still needs defensible space and fire-resistant enclosures in high-risk areas.
• Aging cable. First-generation polyethylene URD cable installed in the 1970s and 1980s has high failure rates from water treeing. Diagnostic testing — very-low-frequency (VLF) tan delta under IEEE 400.2 and partial discharge under IEEE 400.3 — prioritizes replacement or rejuvenation.

15. Costs (indicative)

Installed costs vary widely by region, soil, labor, trenching method and the post-2021 rise in transformer and switchgear prices. The ranges below are indicative for 2026 planning only.

16. A practical roadmap

For a utility, ISO or large customer starting a resilience program, Keentel recommends the following sequence:

• Year 0–1: Baseline. Report IEEE 1366 indices with and without MEDs. Build resilience curves for the last five large events. Create a threat register with expected annual loss. Inventory critical customers and backup power.
• Year 0–1: Quick wins. Expand reclosers and FLISR, tighten vegetation programs, deploy ambient-adjusted ratings, review RAS documentation and real-time model alarms, and confirm black start fuel and cold weather readiness.
• Year 1–3: Studies. Run extreme-event and TPL-008 assessments. Complete EMT studies in weak, IBR-rich areas. Set large-load interconnection requirements. Prioritize hardening using risk-based ranking.
• Year 1–5: Build. Harden priority transmission corridors and substations, build microgrids for critical facilities, add grid-forming storage, deploy PMUs and DFA, and run targeted undergrounding where wildfire or repeated damage justifies it.
• Every year: Learn. Measure performance in each real event against the resilience curves, update fragility and cost assumptions, and feed the lessons into the next plan.
  

Part C — Three Anonymous US Case Studies

The following case studies are composites based on the kinds of projects Keentel Engineering supports. Names, locations and some figures have been changed or generalized to protect client confidentiality.

Case study 1 — Gulf Coast transmission hardening after back-to-back hurricanes

Situation

A vertically integrated utility on the Gulf Coast had been hit by four named storms in two seasons. Its 115 kV and 230 kV network into a coastal load pocket had lost hundreds of structures, and in the worst event every transmission path into the pocket was out at the same time. The state commission asked for a resilience plan with a quantified benefit case before approving accelerated cost recovery.

What Keentel did

• Built outage, restore and performance curves for each storm from the utility's outage records, using the same element-days and MVA-days approach NERC uses in its TADS analyses.
• Ran "rerunning history" scenarios: removing the outages that specific hardening projects would have prevented, and shortening restore times for projects that improve access or add mobile equipment.
• Developed fragility curves for wood H-frame, lattice steel and steel monopole structures against design wind speed, and combined them with regional hurricane return periods to estimate expected annual loss for each corridor.
• Ranked 37 candidate projects by cost per MVA-day avoided, then applied a second screen for paths that close the load pocket to the rest of the system.

Outcome

The first-phase portfolio replaced 140 miles of wood structures on three corridors with steel monopoles designed for higher wind loading, added anti-cascade structures, raised control equipment at four flood-exposed substations, and bought two 230/115 kV mobile transformers. Modeled MVA-days lost in a repeat of the worst storm fell by about 45%. Most of the benefit came from preventing the early, long-duration outages on the paths into the load pocket, which also cut the nadir the most.

Lesson

Hardening is most valuable where an outage both happens early and lasts long. Ranking projects by their effect on the resilience curve, not just on outage counts, targets those elements directly.

Case study 2 — Western distribution utility balancing wildfire risk and outage impact

Situation

A distribution utility in a high wildfire-risk area of the West relied heavily on public safety power shutoffs. PSPS events had prevented ignitions, but they left tens of thousands of customers, including a regional hospital and two water treatment plants, without power for up to three days at a time. The utility needed a mitigation plan that reduced both ignition risk and PSPS impact.

What Keentel did

• Combined the utility's ignition risk model with a consequence model of customer-hours lost to PSPS and to storm outages on each of 118 circuits.
• Evaluated covered conductor, targeted undergrounding, sectionalizing devices, fast-trip settings, and microgrids for critical facilities, using levelized cost per avoided customer-minute and per unit of ignition risk reduced.
• Designed protection settings for high-risk days: sensitive ground fault detection, reclosing blocked on high-risk sections, and fast curves coordinated with downstream fuses.
• Specified two microgrids, one at the hospital (3.5 MW) and one at the larger water plant (2 MW). Each combines grid-forming battery storage, PV and existing diesel generation, with protection designed for the low fault current available in island mode.

Outcome

The plan placed covered conductor on about 60% of high-risk overhead miles and undergrounded only the 8% of miles with the highest combined ignition and egress risk. Added sectionalizing reduced the average number of customers in each PSPS de-energization zone by about 60%. With the microgrids in service, both critical facilities can stay energized through a 72-hour PSPS event without relying only on diesel.

Lesson

Covered conductor and sectionalizing addressed most of the risk at a fraction of the cost of system-wide undergrounding. Microgrids are justified by the value of the specific loads they protect, not by a general CMI calculation.

Case study 3 — Interconnecting a 1.2 GW data center campus at 345 kV

Situation

A developer proposed a 1.2 GW AI data center campus to be built in three phases at a new 345 kV substation in a region that already had significant wind and solar. The transmission owner was concerned about voltage-sensitive load loss, sudden load swings and the campus's effect on regional UFLS and restoration plans.

What Keentel did

• Built a detailed EMT model of the campus's medium-voltage distribution, UPS systems, on-site BESS and transfer logic, using vendor-specific control models for the UPS and battery inverters.
• Simulated normally cleared faults at 345 kV and 138 kV across a range of system strengths. The original design moved the full campus to UPS batteries for any voltage dip below 0.8 per unit lasting more than about 50 ms, which would have removed up to 1.2 GW from the grid for a single nearby fault.
• Worked with the developer's electrical designers to change the UPS input voltage tolerance and transfer logic, so the campus rides through faults that the transmission system clears normally while still protecting the IT load.
• Set requirements for controlled ramp-back after any transfer, telemetry to the Transmission Operator (real and reactive power, UPS and generator status, PMU data at the point of interconnection), and coordination of facility protection with utility reclosing.
• Studied the campus's on-site BESS as a resource that could smooth training-workload power swings and, under a separate agreement, provide emergency load reduction.

Outcome

With the revised settings, simulated load loss for the most severe normally cleared fault fell from 1.2 GW to under 100 MW, which is within the region's existing operating margins. The transmission owner adopted the study's ride-through, telemetry and ramp-rate requirements in its new large-load interconnection guideline.

Lesson

Large loads can be designed to support grid reliability, but only if ride-through and control requirements are set early, before UPS and transfer equipment is specified and ordered.

Part D — 20-Question Technical FAQ

How Keentel Engineering Helps

Keentel Engineering supports utilities, ISOs, developers and large-load customers across the full reliability and resilience lifecycle, at 4 kV to 765 kV:

• Resilience planning: threat registers, fragility modeling, resilience curves, expected-annual-loss and VoLL-based benefit-cost analysis for rate cases and state resilience plans.
• Transmission planning studies: TPL-001 steady-state, short-circuit and stability assessments including extreme events; TPL-007 GMD and TPL-008 extreme temperature assessments; long-term scenario studies.
• EMT modeling and studies: IBR and large-load EMT model review, verification and as-built updates; weak-grid, control interaction and sub-synchronous studies; disturbance reproduction.
• Large-load interconnection: data center and industrial load studies, ride-through and transfer-logic review, telemetry and protection coordination requirements.
• IBR compliance: IEEE 2800 and PRC-029 ride-through assessment, PRC-024, PRC-028, PRC-030 and MOD-026/027 support.
• Protection and automation: relay coordination, low-fault-current protection, FLISR design, RAS review and wildfire-season protection settings.
• Grid-enhancing technologies: IEEE 738 ratings, ambient-adjusted and dynamic line rating implementation, reconductoring studies.
• Microgrids and storage: critical-facility microgrid design, grid-forming BESS specification, black start studies and IEEE 1547 and 2030.7/2030.8 compliance.
• Hardening engineering: transmission structure and substation hardening, flood elevation, mobile substation planning.

Why Clients Choose Keentel

• Senior engineering depth in power system studies, protection and EMT modeling, led by IEEE Senior Member engineers.
• US practice focus: NERC standards, FERC orders, ISO and RTO interconnection processes and state commission requirements.
• Clear, defensible results: calculations and assumptions documented so they hold up in a regulatory filing or a design review.
• Practical recommendations ranked by value and cost, not only by technical merit.
• Responsive delivery for interconnection deadlines, compliance dates and post-event reviews.

Contact Keentel Engineering

Planning a resilience program, a large-load interconnection or an IBR compliance study? Talk to our engineers.

• Phone: (813) 389-7871
• Email: contact@keentelengineering.com
• Book a 15-minute call: calendly.com/keentel-engineering/15min
• Web: keentelengineering.com

References

• IEEE Std 1366-2022, IEEE Guide for Electric Power Distribution Reliability Indices.
• IEEE Std 2800-2022, IEEE Standard for Interconnection and Interoperability of Inverter-Based Resources Interconnecting with Associated Transmission Electric Power Systems.
• IEEE Std 1547-2018, Interconnection and Interoperability of Distributed Energy Resources with Associated Electric Power Systems Interfaces.
• IEEE Std 2030.7-2017 and IEEE Std 2030.8-2018, microgrid controller specification and testing.
• IEEE Std 738-2023, Calculating the Current-Temperature Relationship of Bare Overhead Conductors.
• IEEE Std 693-2018, Seismic Design of Substations; IEEE Std 519-2022, Harmonic Control; IEEE Std C37.118.1, Synchrophasor Measurements.
• NERC Reliability Standards TPL-001, TPL-007, TPL-008, EOP-005, EOP-012, CIP-013, CIP-014, PRC-024, PRC-028, PRC-029 and PRC-030 (nerc.com).
• NERC, State of Reliability report, 2025, including transmission resilience analysis of Hurricane Helene.
• NERC, Odessa Disturbance report (2021) and 2021 California Solar PV Disturbances report (2022).
• NERC, Panhandle Wind Disturbance Report, 2022.
• NERC, Incident Review: Considering Simultaneous Voltage-Sensitive Load Reductions, 2025.
• FERC and NERC, The February 2021 Cold Weather Outages in Texas and the South Central United States, November 2021.
• FERC and NERC, Inquiry into Bulk-Power System Operations During December 2022 Winter Storm Elliott, October 2023.
• FERC Order No. 881 (2021), Order No. 901 (2023) and Order No. 1920 (2024).
• National Infrastructure Advisory Council, Critical Infrastructure Resilience Final Report and Recommendations, 2009.
• Shehabi, A. et al., 2024 United States Data Center Energy Usage Report, Lawrence Berkeley National Laboratory, December 2024.
• Lawrence Berkeley National Laboratory, Interruption Cost Estimate (ICE) Calculator, icecalculator.com.
• ENTSO-E Expert Panel, Grid Incident in Spain and Portugal on 28 April 2025, Final Report, 2026.
• WECC and NERC, review of the November 13, 2025 Wyoming disturbance, 2026.
• US Department of Energy, Grid Deployment Office, Puerto Rico grid recovery and modernization resources.

Disclaimer



This article is for general technical information only and does not constitute engineering advice for any specific project. Worked examples use illustrative assumptions and were prepared for teaching purposes. Standards, regulatory requirements and their effective dates change; always confirm current requirements with the applicable standards body, regulator, Reliability Coordinator and transmission owner. Project decisions should be based on site-specific studies performed by qualified, licensed engineers. Public event information is summarized from publicly available NERC, FERC, WECC, ENTSO-E and US government reports.