Grid-forming GFM

What it is (precise)

Grid-forming vs grid-following in one line: a GFM inverter sets voltage and frequency like a small power station; a GFL inverter injects current into a waveform something else is holding up. Mechanically, a grid-forming PCS regulates voltage behind a virtual or physical impedance and imposes frequency from an internal reference, so it presents to the network as a stiff AC source.

Real and reactive power then emerge from the angle and magnitude difference between that internal voltage and the grid, exactly as they do for a synchronous machine. The three dominant control families are droop control, virtual synchronous machine (VSM/VSG) algorithms, and virtual oscillator control (VOC); all produce the same external behavior of a voltage source whose angle resists change.

Because the response is inherent to the voltage-source physics rather than computed from a measurement chain, a GFM plant reacts to a phase jump or frequency event in sub-cycle time, typically well under 10 ms, versus the 100 ms and slower response loops of a measured, dispatched Frequency response from a GFL plant. The same property lets a GFM battery ride through disturbances, contribute fault current, deliver an inherent inertial response, and, when it is the only source on a de-energized feeder, perform a Black start and form the grid that grid-following resources then lock onto.

The critical physical difference from a synchronous machine is current headroom. A generator can push 5 to 7 per-unit fault current for cycles; an IGBT-based inverter is thermally limited to roughly 1.1 to 1.5 per unit, and only briefly. When a fault drags the GFM inverter into its current limit, the controller must saturate gracefully and re-form voltage as the fault clears. How a vendor handles current saturation is one of the sharpest quality differences between GFM implementations.

Why it matters in a real grid-scale project

As synchronous thermal generation retires, system strength and inertia fall, and renewable-rich nodes become electrically weak. Grid-following inverters depend on a clean voltage to track: below a short-circuit ratio of roughly 2 to 3 their phase-locked loops start to oscillate or trip, and below about 1.5 stable GFL operation is generally not bankable.

A GFM plant can operate at SCR near 1 or below because it does not track the grid, it helps make it. At many remote points of interconnection, specifying GFM is what makes the project connectable at all, and system operators increasingly write it into interconnection agreements.

Commercially, GFM extends eligibility to inertia markets, system-strength and stability-service contracts, and black-start agreements, revenue that a GFL plant cannot access. The trade-offs must be sized honestly: fault-current duty and inertial power consume PCS headroom, so the conversion equipment is often oversized or de-rated relative to a pure energy-shifting design, and the SOC operating band must hold energy in reserve to back the response in both directions.

Early flagship GFM batteries in South Australia (the Hornsdale expansion) and later large stability-contract projects in Great Britain (Blackhillock in Scotland, contracted by National Grid ESO — now NESO — for stability services) proved the model; the constraints they carried belong in your sizing model and revenue case from day one, not bolted on after layout.

Typical values and standards

Representative figures from utility-scale practice: frequency droop settings of 2 to 6 percent (a 5 percent droop moves the plant from zero to full power over a 5 percent frequency deviation); emulated inertia constants H of roughly 2 to 10 s, comparable to the 2 to 7 s of conventional generators; brief overcurrent capability of about 1.1 to 1.5 per unit; and RoCoF withstand of 1 to 2 Hz/s and beyond as demanded by modern grid codes. Treat all of these as tuning ranges set by the interconnection study and the applicable code, not fixed properties of the hardware.

On standards, keep the roles straight. IEEE 2800-2022 sets bulk-system interconnection requirements for inverter-based resources in North America and frames the performance envelope GFM must meet; IEEE 1547 and UL 1741 (with the SA/SB supplements) govern distribution-level interconnection and inverter grid-support functions.

NERC has published functional specifications recommending GFM capability for new BESS interconnections, and formal IEEE standards work on GFM definitions is underway. In Great Britain, grid code modification GC0137 defines a minimum grid-forming (GBGF) specification, and AEMO in Australia publishes a voluntary grid-forming inverter specification. None of these are safety documents.

The safety regime is fully independent of control mode: UL 9540 is the ESS product safety certification, UL 9540A is the fire test method that characterizes cell-to-cell thermal-runaway propagation, NFPA 855 is the installation standard, and NFPA 68/69 cover deflagration venting and explosion prevention. A GFM battery carries exactly the same safety obligations as a GFL one, and battery chemistry is likewise independent of the control mode; LFP dominates stationary BESS regardless. What chemistry and C-rate do bound is the power and energy headroom available to back the GFM response.

How it shows up in specs, studies and contracts

On the PCS datasheet, look past the phrase GFM-capable. Confirm which control method is implemented, that the specific operating mode is type-tested rather than merely available in firmware, the per-unit overcurrent magnitude and duration, the behavior in current saturation, and whether GFM operation narrows the P-Q capability chart or de-rates the MVA nameplate. Four-quadrant operation is assumed for GFM service, so check the reactive envelope at the temperatures and voltages the site will actually see, not just the nominal point.

In interconnection studies, GFM projects at weak points of interconnection almost always require EMT (electromagnetic transient) simulation, typically in PSCAD, on top of the usual RMS studies; a vendor that cannot deliver a validated, site-tunable EMT model of its GFM control will stall your queue position. Ask early, in writing, and get model delivery into the supply contract.

In commercial documents, GFM appears as stability or inertia service terms, black-start availability obligations with associated SOC reserve requirements, and warranty language: the extra cycling and held reserve from GFM duty must be reflected in the battery capacity warranty and augmentation plan, or the project bears an uncontracted degradation risk.

Sensible diligence questions for any GFM procurement: which grid code or specification was the control tested against, and by whom; what happens at the current limit during a close-in fault; how much continuous headroom and SOC band does the promised inertial and fault-current duty consume; can the mode be retuned or switched after commissioning, and who approves the settings; and does the Ride-through envelope hold in GFM mode at the site's actual short-circuit ratio.

Common pitfalls

The recurring traps are contractual, not electrical. GFM-capable on a datasheet without a type test behind it is a claim, not a capability, and firmware updates can silently change control behavior that an interconnection study already approved, so lock control versions into the compliance record.

Do not let anyone treat GFM as a version of grid-following with extra features; they are distinct control modes with different failure behavior, and a plant studied as GFL cannot simply be flipped to GFM without re-study. Finally, virtual inertia marketing figures are meaningless without the per-unit current limit and duration attached: inertia you cannot supply current for does not exist.