Free · Control modes

Name the model, then the number.

A statement about inertia, damping, fault current or stability is a statement about a simulation until someone says which simulation produced it. Two distinctions carry the weight, and two of the three codes on the hub page write one of them into what they ask for.

First distinction

What is the difference between an RMS and an EMT model?

An RMS, or phasor, model tracks fundamental-frequency magnitudes and is what most planning studies run. An EMT model solves the instantaneous waveform, so sub-cycle control action, behaviour at the converter current limit and the response to a phase jump appear in it at all. A result that shows up in one model class and not the other is a fact about the models before it is a fact about the plant.

Control interaction with the network is the one to keep straight, because it is not only an EMT question. It is also a frequency-domain one, assessed by setting the converter’s own impedance against the network’s — which is why Chile asks a grid-forming plant for an impedance-versus-frequency diagram, a Nichols plot and its control architecture in Laplace form alongside the EMT model. Switzerland asks for the same control in both classes at once: synthetic inertia must begin within 10 ms of a frequency disturbance and be demonstrated in simulation models, RMS and EMT.

Second distinction

Does a vendor’s generic model prove anything about my plant?

Not on its own. A generic library model describes a class of converter; a project-specific validated model describes the firmware you are buying, with your settings, checked against measurement. Chile lets a manufacturer validate an EMT model by any one of five methods, and they are five different grades of evidence: a factory test report on the converter during system events, a hardware-in-the-loop test, field measurements of the installed equipment after a real fault, a laboratory test, or a mapping of the converter’s actual firmware version, at installation level, to the version of the model.

Which grade you were given is a question a specification can ask, and “validated” on its own does not answer it. Chile then closes the loop somewhere else: any modification of an IBR’s control parameters needs the Coordinador’s authorisation before it is implemented, whichever mode the plant runs. Between the two articles, a firmware update that changes control behaviour after the study is a compliance event rather than a maintenance one.

What a specification can do without waiting for anyone is put model delivery in the supply contract: the grid-forming entry and the grid-following entry carry what to demand and when it is due.

What the codes ask

Do grid-forming converters need an EMT model?

Whether a study is required at all is the network operator’s call, not the control mode’s and not a vendor’s, so a specification demanding one should name the document behind it. Chile is such a document, and it asks more of the grid-forming case than of the grid-following one: the grid-forming model must let the operator verify the voltage controller, the fast current-injection controller, the frequency/power controller, the inertial response and the active phase-jump control, evaluated at the connection point against a weak grid. The grid-following list is two items and carries no weak-grid qualifier.

The asymmetry is the argument of this page written by a regulator rather than by us: the weak-grid case is exactly where a phasor model stops describing a grid-forming plant, so that is where the annex puts the demonstration. Before the plant enters operation the Coordinador must also validate the transitions — normal voltage control to fault voltage control and back across inception and clearance, and frequency/power control to inertial response — plus operating stability at phase jumps of up to 60 degrees, determining the maximum tolerated limit, island operation, and black start where it applies. Three phase-jump numbers in one document, and they are not the same duty: 25 degrees is what a grid-following plant must ride stably, 30 degrees is the grid-forming design minimum ridden with the limiter off, and 60 degrees is what the operator validates the grid-forming model against.

What the study owes you is a criterion, not a verdict. Finland writes one: the plant’s response to network oscillations must not amplify them, and where its own properties do not achieve that, the code says it shall be equipped with a separate damping control — so the finding that decides is whether the plant clears that bar unaided. Chile writes the other: the Coordinador determines the maximum phase-jump angle the plant actually tolerates, which is a number to hold a bidder to rather than a pass mark. A study that comes back without either has described your plant and not decided anything.

The number that hides its model

Is a converter’s inertia constant comparable to a synchronous machine’s?

Only if the current limit and the duration are quoted beside it. A machine’s inertia constant measures energy in a spinning mass. A grid-following converter synthesises the effect by measuring frequency, differentiating and dispatching, so its response lags by its own measurement chain. A grid-forming converter measures nothing first. All three are quoted in the same units, and a figure without a per-unit current limit and a duration beside it is a marketing number rather than a specification.

The grid-forming case is the one that surprises people: power leaves the terminals because the internal voltage phasor is held while the network’s moves, which is why Finland asks for the current to start changing within a few milliseconds of the step instead of for a response time.

Chile has already written the package that replaces the single number. Under a transitional article, before a grid-forming installation enters operation, it declares an inertia constant and an effective one derived from its own RoCoF response, overcurrent capability and peak current in per unit, maximum three-phase, single-phase and negative-sequence contribution at the connection point, the active damping power for a system frequency variation or oscillation between 0.05 and 1 Hz, and an equivalent damping factor. That list is what a specification asking for "virtual inertia" should be asking for instead, and a grid-following plant owes none of it.

Damping is where two of the codes differ usefully. Finland bands it 0.2 to 1 Hz interarea, 1 to 15 Hz weak-network grid-following, 15 to 45 Hz series-compensated. Switzerland bands the same phenomenon in four: 0.1 to 1 Hz inter-area, 1 to 2 Hz local, 2 to 15 Hz voltage oscillations and control interactions, 15 to 45 Hz subsynchronous and resonance. A damping study answering one banding has not answered the other, which is the practical form of the rule at the top of this page.

Where this sits