Electronics Guide

Voltage Source Behind an Impedance

The defining model of a grid-forming inverter is a controllable voltage source behind an impedance. The controller commands an internal voltage of a chosen magnitude and angle, and the current that flows is set by the difference between that internal voltage and the network voltage, divided by the coupling impedance. The impedance may be physical, consisting of the filter inductance and the step-up transformer, or it may be synthesized in software as a virtual impedance, or both. The converter does not command current directly; current becomes an outcome.

This is the same relationship that governs a synchronous machine, and the consequences follow the same familiar rules. Real power flow across a predominantly inductive coupling depends chiefly on the angle difference between the internal voltage and the network voltage, while reactive power flow depends chiefly on the magnitude difference. A grid-forming converter therefore controls real power by adjusting the frequency, and hence the accumulated angle, of its internal voltage, and controls reactive power by adjusting the internal voltage magnitude. It never needs to be told what the grid angle is, because it asserts an angle of its own.

Because the internal voltage is maintained continuously, the converter responds to a network disturbance in the instant the disturbance occurs. A phase jump on the network produces an immediate change in the angle across the coupling impedance and therefore an immediate change in real power, with no measurement, filtering, or control-loop delay in the path. A drop in network voltage produces an immediate reactive current contribution. These instantaneous responses are the properties that grid codes now single out as the signature of genuine grid-forming behavior, because they cannot be reproduced by any controller that must first estimate the grid state.

Grid-Following Behavior by Contrast

A grid-following inverter estimates the grid voltage angle with a phase-locked loop, transforms its measurements into a reference frame aligned with that angle, and regulates direct-axis and quadrature-axis current to deliver commanded real and reactive power. The scheme is straightforward, decouples the two power components cleanly, and limits current by construction, since current is the controlled variable. It has served the industry well and remains appropriate for the great majority of distribution-connected resources.

Its weaknesses appear where the grid is not stiff. On a weak network, characterized by a low short-circuit ratio between the available fault current at the point of connection and the converter rating, the converter's own current injection moves the voltage that the phase-locked loop is trying to track. The estimator and the plant become coupled, and control loops that were comfortably stable on a strong bus can oscillate or lose synchronism entirely. Subsynchronous control interactions observed in wind-rich regions of the United States, and the more general problem of converter-driven stability, trace back to this coupling.

The behavior of grid-following resources during grid disturbances has also produced recorded reliability events. The 2016 Blue Cut Fire disturbance in southern California saw roughly 1,200 megawatts of photovoltaic generation reduce output following a transmission fault, largely because inverter frequency estimation misread the distorted voltage waveform and because units entered momentary cessation rather than riding through. Similar large-scale reductions in inverter output followed transmission faults near Odessa, Texas, in 2021 and again in 2022, the second event being the larger of the two. These events motivated the ride-through requirements now embedded in interconnection standards, and they sharpened the argument that converters should be able to hold up a voltage rather than merely react to one.

Synchronization Without a Phase-Locked Loop

A grid-forming converter still has to stay in step with everything else on the network, but it does so through the physics of power transfer rather than through an estimator. If the converter's internal frequency exceeds the network frequency, its angle advances relative to the network, real power export rises, and a droop or swing-equation term slows the internal frequency back down. This is exactly the self-synchronizing mechanism of a synchronous generator, and it is stable within the same limits: the restoring action persists only while the power angle remains below the point of maximum transfer.

Most practical products retain a phase-locked loop even so, though not in the primary control path. It is used for initial synchronization before the interconnection contactor closes, for protection and monitoring functions, for grid-code reporting, and sometimes for supervisory logic that decides when to switch operating modes. The important architectural point is that the loop that sets the internal voltage angle during normal operation does not depend on an angle measurement, so estimator dynamics cannot destabilize it.

What Grid-Forming Is Not

Several capabilities are frequently confused with grid-forming control, and specifications have had to draw the lines explicitly. Fast frequency response is a droop-like power adjustment triggered by a measured frequency deviation; a grid-following converter can provide it, and providing it does not make the converter grid-forming. Synthetic inertia based on a measured rate of change of frequency is likewise a measurement-driven function, delayed by the filtering that any rate-of-change estimate requires, and it is not equivalent to the instantaneous power exchange that a voltage source delivers. Dynamic reactive current injection during a voltage dip is likewise a current-mode function. IEEE Std 2800-2022 requires it of transmission-connected inverter-based resources; IEEE Std 1547-2018 permits dynamic voltage support at distribution level but neither mandates nor specifies it in detail.

Nor does grid-forming control require islanded operation, although it enables it. A grid-forming converter can and usually does operate connected to a live transmission system, where it holds an internal voltage that happens to be almost in step with everything around it. Conversely, the ability to run a standalone load, which any uninterruptible power supply possesses, does not by itself constitute the transmission-grade grid-forming capability that system operators specify, because that specification also demands defined behavior during faults, defined power sharing with other sources, and defined damping of network oscillations.

Frequency and Voltage Droop

Droop control is the oldest and simplest grid-forming law. The controller reduces its internal frequency in proportion to measured real power output and reduces its internal voltage magnitude in proportion to measured reactive power output. The frequency droop coefficient is conventionally expressed as the percentage frequency change from no load to full load: a five percent droop on a sixty-hertz system corresponds to three hertz across the full power range, and typical settings for inverter-based resources fall between two and five percent. Voltage droop coefficients of two to five percent are common.

The scheme shares load among parallel sources without any communication between them. Each unit slides along its own characteristic until all units settle at a common frequency, at which point each carries a share of the load set by its droop coefficient and its no-load setpoint. Because frequency is a global variable in a synchronized network, real power sharing through frequency droop is inherently accurate. Reactive power sharing through voltage droop is not, because voltage differs from bus to bus by the drop across the intervening impedance, and units electrically closer to the load will take more than their share unless virtual impedance or a secondary controller corrects the imbalance.

Droop necessarily leaves a steady-state offset, since the frequency deviation is what carries the power-sharing information. A secondary controller, resident in the plant controller or in a microgrid controller, slowly shifts the no-load setpoints to restore nominal frequency and voltage over seconds without disturbing the sharing established by the primary layer. The pairing of a fast, local, communication-free primary layer with a slower, coordinated secondary layer is the standard hierarchical arrangement.

One practical detail dominates droop tuning: the power measurement must be filtered. Instantaneous power computed from three-phase quantities carries a component at twice the fundamental frequency whenever the system is unbalanced or distorted, and feeding that ripple into the frequency reference would modulate the internal voltage angle. A low-pass filter removes it but introduces lag, and that lag is not a nuisance to be minimized. It is what gives the controller its dynamic character.

Virtual Synchronous Machine Control

Virtual synchronous machine control, also called virtual synchronous generator or synchronverter control, takes the emulation further by implementing the swing equation directly. The controller integrates the difference between a power setpoint and measured electrical power, scaled by a virtual inertia constant, to produce a virtual rotor speed, and integrates that speed to produce the internal voltage angle. A damping term proportional to the speed deviation stands in for the damper windings of a real machine, and a virtual excitation loop regulates the internal voltage magnitude against terminal voltage and reactive power. Emulated inertia constants are typically chosen in the range of a few seconds, comparable to the two to nine seconds typical of real turbogenerators and hydro units.

The appeal is that the resulting device presents the network with dynamics the industry already knows how to study. Existing stability tools, protection philosophies, and operator intuition transfer. The virtual inertia contribution is genuine in the sense that power is exchanged in proportion to frequency acceleration, with no measurement delay, because it emerges from the angle difference across the coupling impedance rather than from a computed derivative.

The appeal is also the trap. A converter has no rotor, so every joule of emulated inertial response comes out of the energy resource behind the direct-current bus and through semiconductors with a thermal time constant measured in milliseconds. Emulating a large inertia constant commits the converter to large, fast power excursions that it may not be able to deliver. Designers therefore treat the virtual inertia constant as a contract with the energy source and the thermal design, not as a free parameter, and many products deliberately choose modest emulated inertia with strong damping.

It is worth recognizing that droop control with a filtered power measurement and virtual synchronous machine control are mathematically close relatives. The first-order lag in the droop power filter plays the role of the inertia term, and the droop gain plays the role of damping. Choosing one family over the other is largely a question of which parameters the designer wishes to expose and how the behavior is to be documented for a grid operator, rather than a question of fundamentally different capability.

Virtual Oscillator Control

Virtual oscillator control abandons the machine analogy. The controller implements the differential equations of a nonlinear oscillator, typically of the Van der Pol type, directly in the voltage reference, so that the inverter output behaves like a coupled oscillator in a network of oscillators. Such networks synchronize from arbitrary initial conditions, and the synchronization is a property of the oscillator dynamics rather than of a linearization around an operating point. The practical consequence is a controller with very fast synchronization and provable stability over a wide region rather than only near equilibrium.

Plain virtual oscillator control has an inconvenient property for a grid asset: its power sharing is determined by the oscillator parameters, and it does not accept a dispatch setpoint naturally. Dispatchable virtual oscillator control resolves this by reformulating the dynamics so that the equilibrium corresponds to specified real and reactive power setpoints while retaining the wide-area synchronization guarantees. Near its equilibrium, dispatchable virtual oscillator control reduces to droop behavior, which makes it compatible with existing grid-code expressions of frequency and voltage response while offering better behavior during large transients. The method is well represented in laboratory and demonstration work and is beginning to appear in commercial control platforms.

Matching Control and the Direct-Current Side

Matching control observes that the direct-current link capacitor of a voltage-source converter obeys an equation with the same structure as the swing equation, with capacitor voltage playing the role of rotor speed. The controller therefore derives the internal voltage frequency directly from the measured direct-current link voltage. The elegance is that the emulated inertia is physically backed: when the alternating-current side demands more power than the source supplies, the link voltage falls, the internal frequency falls with it, and the power demand is reduced automatically. Nothing pretends to have energy that is not present.

The method makes the link capacitor and the source dynamics part of the grid-forming behavior, which is both its strength and its limitation. The available inertia is set by the capacitance and the allowable link voltage excursion, which is small in most designs, so matching control is usually combined with droop or swing dynamics rather than used alone. Its real contribution is the reminder that grid-forming control is a whole-converter problem in which the direct-current side cannot be treated as an ideal source.

Inner Loops and Virtual Impedance

Beneath whichever outer law sets the internal voltage, most implementations use a cascaded inner structure: an alternating-voltage regulator across the filter capacitor, and inside it a current regulator on the converter-side inductor current. The inner current loop is retained precisely because it provides a place to impose a hard limit on the semiconductor current, and its bandwidth is set an order of magnitude above the voltage loop so that the two do not interact. Designs that omit the inner loops and drive the modulator straight from the voltage reference exist, and they respond faster, but they leave the designer with no clean point at which to enforce a current constraint.

Virtual impedance is the other essential ingredient. The controller subtracts a term proportional to output current, and often to its derivative, from the voltage reference, which makes the converter appear to have a series impedance that no physical component provides. The technique serves several purposes at once. A virtual inductance makes the coupling predominantly inductive, which restores the clean decoupling between angle and real power that droop control assumes; this matters greatly on low-voltage networks, where line resistance is comparable to or larger than reactance and the usual pairings break down. A virtual resistance at selected frequencies damps the resonance of the output filter and can suppress interaction with other converters. A virtual impedance that grows with current becomes a current-limiting mechanism, discussed below. Because the impedance is synthesized rather than physical, it dissipates no power and can be scheduled with operating point, frequency, or sequence component.

Why Limiting Is the Central Difficulty

A synchronous generator tolerates severe overcurrent because its windings and magnetic circuit have enormous thermal mass; subtransient fault contributions of five to seven times rated current for several cycles are routine, and utility protection schemes were designed around exactly that. A semiconductor bridge has no such margin. The junction of an insulated-gate bipolar transistor or a silicon carbide device reaches its limit in microseconds to milliseconds, and short-circuit withstand times are quoted in microseconds: on the order of ten for a silicon insulated-gate bipolar transistor, and a few for a silicon carbide device. Practical converters are rated for continuous output near their nameplate current and can be designed for perhaps 1.1 to 1.5 per unit for a fraction of a second, with the upper end of that range bought by oversizing the bridge, the filter, and the cooling system.

This creates a direct conflict with voltage-source behavior. A voltage source behind a small impedance, connected to a nearby short circuit, delivers whatever current the impedance permits, which is far more than the bridge can survive. The controller must therefore intervene during faults. The entire engineering difficulty of grid-forming control lies in intervening in a way that protects the hardware without destroying the voltage-source character that justified the design, and without leaving the converter unable to return to normal operation when the fault clears.

Current Saturation and Loss of Synchronism

The most direct approach saturates the current reference at the output of the inner voltage loop. The bridge is protected, but the moment saturation begins the voltage loop no longer controls anything: the converter has become a current source of fixed magnitude, and the outer law that sets the internal angle continues to run against a power measurement it can no longer influence.

The result is a well-documented transient instability. Under a fault, the electrical power that a current-limited converter can export falls, often far below the power setpoint. The swing-equation or droop integrator sees a persistent power deficit and accelerates the internal angle without bound. If the fault lasts long enough, the angle advances past the point of no return, and when the fault clears the converter cannot recover synchronism. It then either pole-slips repeatedly or trips, which is precisely the outcome the grid-forming resource was installed to prevent. The condition is analogous to the loss of synchronism of a generator whose fault clearing time exceeds its critical clearing time, but the converter's margin is far smaller because its current limit is far tighter.

Mitigations all work by re-establishing a feedback path to the angle while current is saturated. Freezing or clamping the integrator that produces the internal angle prevents unbounded acceleration. Feeding the difference between the commanded and delivered current back into the angle dynamics gives the outer loop something to act on. Some schemes explicitly regulate the power angle during saturation, and some reduce the internal voltage magnitude so that the unsaturated current naturally falls within the limit. Verifying the recovery behavior across a range of fault types, durations, and grid strengths is a core part of grid-forming validation, and it is a common failure mode in early product designs.

Virtual Impedance Limiting

An alternative keeps the converter in voltage-source mode throughout by increasing the virtual impedance as current approaches the limit. Threshold virtual impedance schemes leave the nominal impedance unchanged in normal operation and add a steeply rising term once measured current exceeds a chosen fraction of the rating. Because the voltage reference is reduced smoothly rather than clipped, the internal voltage remains meaningful, the outer law continues to see a power that responds to its angle, and the loss-of-synchronism mechanism described above is largely avoided.

The price is a softer limit. Current overshoots during the first instants of a fault before the impedance builds up, so the scheme must be tuned against the actual withstand capability of the bridge, and it is often paired with a hard saturation as a last line of defense. The design also has to avoid an impedance so large that the converter effectively disconnects itself from the faulted network, contributing nothing to the fault current that downstream protection needs in order to operate.

Mode Switching and Its Hazards

Many commercial designs simply switch to grid-following current control when a fault is detected, then switch back afterward. The approach uses proven current-mode fault logic and is easy to certify against existing ride-through requirements. It also has real hazards. The transition itself is a discontinuity that can excite transients; the returning transition must re-establish the internal angle without a jump; the phase-locked loop that the current mode relies on may be unreliable during precisely the depressed and distorted voltage conditions that triggered the switch; and, most importantly, during the fault the grid loses the voltage reference that the resource was procured to provide. In a network with a high proportion of converter-interfaced generation, a fleet that all abandons grid-forming behavior at the same moment reproduces the weakness the fleet was meant to eliminate. System operators have therefore begun to scrutinize mode-switching behavior in their specifications, and several require the internal voltage source to persist through the disturbance.

Unbalanced Faults and Negative-Sequence Response

The majority of transmission and distribution faults are unbalanced, most commonly single-line-to-ground, and unbalance produces negative-sequence voltage that a naive three-phase controller handles badly. The negative-sequence component produces a second-harmonic ripple in real power and, if the converter presents a low negative-sequence impedance, a large negative-sequence current that consumes the available current headroom without contributing to voltage support.

IEEE Std 2800-2022 requires transmission-connected inverter-based resources to provide negative-sequence current injection during unbalanced faults, because that injection helps unbalance-sensitive protection relays detect and clear the fault correctly. Meeting the requirement in a grid-forming converter means controlling the negative-sequence internal voltage and negative-sequence virtual impedance explicitly, usually through separate positive- and negative-sequence control paths built on a decoupled sequence extraction stage. The current limit must then be enforced on the combined phase currents rather than on either sequence separately, and the designer must decide how to allocate scarce headroom between positive-sequence voltage support and negative-sequence injection. That allocation is a policy choice as much as an engineering one, and grid codes increasingly specify it.

Control Cannot Manufacture Energy

Every grid-forming response draws real energy. An inertial response to a falling frequency delivers power for as long as the frequency falls; a droop response holds an offset in power for as long as the frequency offset persists; and a phase-jump response delivers a large power pulse within milliseconds. None of this is available unless the resource behind the direct-current bus can supply it and the converter has been sized to pass it.

Battery energy storage is the natural host. A battery can source and sink power in either direction within milliseconds, is not committed to an energy production schedule, and can be sized in energy independently of the converter's power rating. This is why nearly every early grid-forming demonstration and nearly every emerging grid-code requirement is framed around storage, and why storage inverters were the first products to be marketed as grid-forming.

Photovoltaic plants can operate in grid-forming mode only by curtailing below the maximum power point so that upward headroom exists, which sacrifices revenue continuously in exchange for a capability used occasionally. Hybrid plants that place a battery on a shared direct-current bus avoid the trade-off and are becoming the common answer. Wind turbines can extract kinetic energy from the rotor for a brief overproduction, but the extraction is followed by a recovery period of reduced output as the rotor is brought back to speed, and full-converter type-four machines are far better suited to grid-forming control than doubly-fed designs. Voltage-source-converter high-voltage direct-current terminals are excellent grid-forming devices, since the direct-current link is backed by the far end of the link, and grid-forming control of offshore wind connections has become an active area of specification.

Sizing the Margins

Three margins define what a grid-forming plant can promise. Current headroom, expressed as the ratio of the maximum sustained current to the rating, determines the fault contribution and the size of the phase-jump response. Energy headroom, expressed in seconds at rated power, determines how long an inertial or droop response can be sustained. Thermal margin determines how often such responses can be repeated without exceeding the junction temperature limits that govern the power module's fatigue life.

These margins are not free. Oversizing the bridge, the output filter, and the cooling system to gain fault-current headroom adds cost to every unit, and reserving state of charge to guarantee a bidirectional inertial response reduces the energy available for arbitrage or capacity services. Rational procurement therefore treats grid-forming capability as a specified, verified, and separately compensated service. The stability procurements run by system operators in Great Britain and the grid-forming demonstration funding provided in Australia both exist because the capability has a cost that energy markets alone do not recover.

Grid Strength and an Inverted Constraint

Grid-following converters become difficult on weak grids and behave well on strong ones. Grid-forming converters invert the constraint. A voltage source connected through a small impedance to another voltage source produces a large current for a small angle or magnitude error, so on a very stiff bus a grid-forming controller with aggressive gains can exhibit poorly damped power oscillations or excessive circulating current. Conversely, grid-forming control operates comfortably at short-circuit ratios at which grid-following control fails, including networks with no synchronous machines at all.

The practical implications are twofold. First, a network needs a deliberate mix rather than a single technology, and the appropriate proportion of grid-forming capacity is a planning study rather than a rule of thumb. Second, a grid-forming product intended for general use must remain stable across the full range of connection strengths it may encounter, which usually means adaptive virtual impedance or gain scheduling, and which must be demonstrated across that range during type testing rather than at a single nominal impedance.

Many Units on One Network

A utility-scale plant contains dozens to hundreds of inverters behind a medium-voltage collector system and a step-up transformer. Grid-forming control applied at the unit level makes each of those inverters a voltage source sharing a network of collector cable impedances, which raises questions of circulating current, reactive sharing accuracy, and interaction at the resonances of the collector system. Virtual impedance and careful attention to the sequence and harmonic impedance the units present to one another are the usual tools. Plant-level control adds a further layer, coordinating the units and delivering the setpoints and secondary restoration that the point of interconnection requires.

Interaction between plants is the harder problem. Converter-driven oscillations in the range from a few hertz to several hundred hertz have been observed on real networks, and they arise from the interaction of control loops with network impedance rather than from any single device being defective. Impedance-based analysis, in which each converter is characterized by its small-signal terminal impedance across frequency and the aggregate is checked for stability, has become the standard analytical framework. It also underpins a growing expectation that manufacturers supply validated electromagnetic-transient models rather than positive-sequence phasor models, since phasor models cannot represent these phenomena at all.

IEEE 2800 and Its Grid-Forming Amendment

IEEE Std 2800-2022 established uniform interconnection and performance requirements for inverter-based resources connected to transmission and sub-transmission systems, a domain that IEEE Std 1547 explicitly excludes. It is written in technology-neutral terms and does not mandate grid-forming control, but several of its requirements, including voltage and frequency ride-through, dynamic reactive current injection, negative-sequence current injection during unbalanced faults, and the supply of validated electromagnetic-transient models, define the environment in which a grid-forming product must operate. IEEE Std 2800.2-2026, the companion recommended practice covering test and verification procedures for inverter-based resources interconnecting with bulk power systems, was published in 2026 and provides the plant-level conformity assessment framework.

Two further projects address grid-forming directly. Project P2800a is an amendment to IEEE Std 2800-2022 titled "Reduce Barriers for IBRs with Grid-Forming Equipment," authorized in December 2025, which revisits clauses in the base standard whose current-mode assumptions unintentionally penalize converters that hold an internal voltage source. Project P2800.1 is a recommended practice on the functional capabilities and performance of grid-forming equipment in inverter-based resources, intended to specify minimum grid-forming functional and performance capabilities together with verification test procedures. Both projects carry authorization through December 2029, so readers should treat them as works in progress and check their status before citing requirements from them.

The UNIFI Specifications

The Universal Interoperability for Grid-Forming Inverters consortium, co-led by the National Renewable Energy Laboratory, the University of Texas at Austin, and the Electric Power Research Institute with United States Department of Energy support, publishes the most widely cited vendor-neutral specification for grid-forming behavior. The UNIFI specifications state requirements at both the power-system level and the inverter level, with the explicit aim of allowing grid-forming resources from different manufacturers to operate together at any scale without bespoke integration studies. Version 1 appeared in 2022, Version 2 in 2024, and Version 3 in January 2026. Because the specifications are freely available and were written before any binding standard existed, they have functioned as the common technical vocabulary for the field.

Great Britain: Grid Code Modification GC0137

Great Britain moved first among national grid codes. Modification GC0137, "Minimum Specification Required for Provision of GB Grid Forming Capability," was approved by the regulator on 31 January 2022 and added a non-mandatory grid-forming specification to the GB Grid Code, having begun life under the name of virtual synchronous machine capability. The specification is notable for describing behavior in terms an operator can test rather than in terms of a control algorithm: it addresses internal voltage source behavior, active power response to a phase jump, reactive power response to a voltage jump, damping of network oscillations, fast fault current contribution, and inertial power contribution, while deliberately leaving the magnitude of several responses to be set commercially rather than prescribed. The system operator has procured the resulting capability through dedicated stability tenders rather than relying on energy market revenue.

Australia: The AEMO Voluntary Specification

The Australian Energy Market Operator published its Voluntary Specification for Grid-forming Inverters in May 2023, defining a set of core capabilities that a device must demonstrate to be classified as grid-forming, together with additional capabilities that are desirable but not universal. A companion Core Requirements Test framework followed in January 2024, giving a simulation-based procedure for determining whether a given piece of equipment and control mode meets the core capabilities. Australia's interest is not theoretical: South Australia has run for extended periods at very high instantaneous shares of non-synchronous generation, and system strength in the remote parts of its network is genuinely scarce.

Europe: The Network Code on Requirements for Generators

The European approach is moving from voluntary to binding. ENTSO-E published a Phase II technical report on grid-forming capability of power park modules on 4 November 2025, supporting an amendment to the Network Code on Requirements for Generators. The draft amendment would introduce grid-forming obligations for non-synchronous power park modules and electricity storage modules in the larger connection categories, with the effect of making the capability mandatory for new plants above roughly one megawatt, subject to the type thresholds that vary by synchronous area. ENTSO-E has indicated that a non-binding implementation guidance document will follow once the final code is adopted. This would be the first mandatory grid-forming requirement applied across a continental interconnection, and engineers designing equipment for the European market should track the code's progress rather than assume the current draft text.

North American Guidance

In North America, the North American Electric Reliability Corporation has led with guidance rather than mandatory requirements for grid-forming specifically. Its white paper "Grid Forming Technology: Bulk Power System Reliability Considerations," published in December 2021, framed the reliability case, and "Grid Forming Functional Specifications for BPS-Connected Battery Energy Storage Systems," published in September 2023, offered a functional specification and a simple test system that planners can use to verify grid-forming characteristics for interconnecting resources. Separately, reliability standard PRC-029-1 imposes frequency and voltage ride-through obligations on inverter-based resources generally, in direct response to the recorded disturbances. The Federal Energy Regulatory Commission approved it in July 2025, its ride-through curves follow those of IEEE Std 2800-2022, and it takes effect on 1 October 2026. Regional entities have gone further in places: the Midcontinent Independent System Operator has developed grid-forming performance requirements for standalone battery energy storage, and several interconnection queues now request grid-forming capability where system strength studies justify it.