Renewable Modelling · Frequency Stability

Modelling IBGs for Frequency Stability

Trying to model every function of an inverter-based generator in every study is slow and pointless. The trick is to match the model to the phenomenon: for a frequency-stability study, a handful of functions decide the result and the rest can be left out. This guide works through how IBGs actually behave under frequency deviations — ROCOF and inertia, over- and under-frequency transients, the 50.2 Hz reconnection trap — why so many operators still fall back on a negative-load model, how small islands differ from large interconnected grids, and, from that, which functions you genuinely need to represent for a frequency study in EMTP® and RMS tools. It builds on the companion guides to the characteristics of inverter-based generation and the inverter itself.

Reading time ≈ 32 min · Behaviour, real cases & the functions that matter

Now that the RMS / phasor and EMT modelling tools have been introduced, the series turns to applying them, study by study. This page starts with frequency stability. Frequency and voltage are the two halves of power-system stability, and they turn on different quantities: frequency on the active-power balance, voltage on the reactive-power balance (the subject of the guides that follow). The central question here: after a sudden imbalance between generation and demand, does the system frequency stay within acceptable limits, and do the inverter-based generators (IBGs) support or worsen the response?

What frequency stability means

Frequency stability is the ability of the power system to maintain or recover an acceptable frequency after a sudden active-power imbalance. If generation is suddenly less than load, frequency falls; if generation is suddenly greater than load, frequency rises. The study checks whether the decline or rise is arrested before protection, load shedding, generator trips or cascading disconnection occur. Frequency is controlled mainly by the balance between active-power generation and active-power demand — reactive power mainly affects voltage magnitude, whereas an active-power imbalance affects system speed and frequency. That is why a frequency study concentrates on inertia, ROCOF, primary response, fast frequency response, under-frequency load shedding and over-frequency generation reduction.

A frequency event, step by step

1. A generator, interconnector, load block or large plant trips.
2. Active power becomes unbalanced.
3. Frequency starts to move.
4. Inertia sets the initial rate of change of frequency (ROCOF).
5. Fast frequency response or synthetic inertia may act.
6. Primary frequency response changes active power (governor / droop).
7. Secondary and tertiary control restore frequency and reserves.
8. Protection or load shedding may operate if limits are exceeded.

Abbreviations used on this page

IBGInverter-based generation (or generator)

RMSRoot-mean-square (phasor) simulation

EMTElectromagnetic transient (time domain)

EMTP®Electromagnetic Transients Program (an EMT tool)

ROCOFRate of change of frequency

FFRFast frequency response

FSMFrequency-sensitive mode

LFSM-OLimited frequency-sensitive mode – over-frequency

LFSM-ULimited frequency-sensitive mode – under-frequency

P(f)Active power as a function of frequency (droop)

UFLSUnder-frequency load shedding

MPPTMaximum power point tracking

PVPhotovoltaic

WTGWind-turbine generator

BESSBattery energy storage system

RESRenewable energy sources

PLLPhase-locked loop

FRTFault ride-through

PODPower oscillation damping

AGCAutomatic generation control (secondary control)

HVDCHigh-voltage direct current

TSO / DSOTransmission / distribution system operator

CAISOCalifornia Independent System Operator

PCC / POIPoint of common coupling / interconnection

Key idea

Frequency stability is an active-power balance problem: inertia sets the initial ROCOF, and the frequency nadir depends on the inertia, the disturbance size, the speed of the response, the headroom and the protection.
IBGs do not naturally provide synchronous inertia — the converter decouples the stored energy from the grid — but they can provide controlled synthetic inertia or fast frequency response if energy or headroom is available, and PV at maximum power has no upward reserve at all.
Over- and under-frequency are not symmetric: reducing generation on over-frequency (LFSM-O) is easy, but increasing it on under-frequency needs headroom, storage or a fast primary source — and after a wind turbine releases rotor energy its recovery can cause a second dip.
Because frequency controls and relays act in tens of milliseconds to minutes, RMS models are usually enough — but only if the over/under-frequency and ROCOF protection, the reconnection settings, and the reserve / headroom logic are in the model, or the study will miss events like the 50.2 Hz cascade.

Key terms used on this page

01Frequency stability

Keeping frequency within limits after an active-power imbalance between generation and load.

02Inertia

Kinetic energy stored in rotating masses; released or absorbed immediately when frequency changes, slowing the initial movement.

03ROCOF

Rate of change of frequency (Hz/s); its initial value is set by the system inertia. Lower inertia means higher ROCOF for the same imbalance.

04Frequency nadir

The lowest frequency reached after a generation-loss event, before primary control recovers it.

05Synthetic inertia

A controlled active-power response that imitates part of the effect of inertia, usually driven by measured ROCOF; not identical to true inertia.

06Fast frequency response (FFR)

A fast controlled active-power response after a frequency event; broader than synthetic inertia, and not necessarily inertia-like.

07P(f) control / droop

Active power changed as a function of frequency; droop sets how much power changes per unit of frequency deviation.

08LFSM-O / LFSM-U

Limited frequency-sensitive mode: reduce output above an over-frequency threshold, or raise it below an under-frequency threshold.

09Headroom / curtailment

Operating below the available maximum (curtailment) to keep reserve (headroom) for raising output when frequency falls.

10Deadband / ramp rate

A frequency band around nominal where the control does not act; the ramp rate limits how fast the active power may change.

11ROCOF immunity

The requirement to stay connected through an acceptable rate of change of frequency, rather than nuisance-tripping.

12Negative-load model

Representing embedded generation as a fixed reduction in load, with no dynamic response, protection or trip behaviour of its own.

Symbols used on this page

\(f_0\) — nominal frequency (50 Hz in the UK / Europe); \(\Delta P\) — the active-power imbalance; \(H\) — the inertia constant (seconds); \(S\) — the rated apparent power.
\(P = \dfrac{V_1 V_2}{X}\sin\delta_{12}\) — the power-angle relation (Section 12): \(V_1, V_2\) bus voltages, \(X\) reactance, \(\delta_{12}\) the angle between them.
\(P_{ref}\) — the active-power reference; \(K_{HFRT},\ f_{start}\) — the over-frequency droop gain and its start frequency (Section 9).

Section 1

The frequency-stability phenomenon

Frequency stability is the system’s ability to cope with an imbalance between generation and load, and it splits by timescale. Short-term behaviour, in the range of a few seconds and below, is governed by ROCOF, inertia (including synthetic inertia) and fast frequency response. Long-term behaviour, up to several minutes, is about frequency recovery. Frequency control itself is simply a generating unit’s ability — whether directly or inverter-connected — to adjust its active power to hold the system frequency.

Islanded systems behave very differently from large interconnected ones: a containment or a stability issue can arise even in the normal operation of an autonomous microgrid, or during grid restoration after a blackout. And the phenomenon is gaining importance precisely because of IBGs: as the penetration of renewable energy sources (RES), including IBGs, rises, system inertia tends to fall, giving larger frequency excursions and a higher ROCOF unless the missing inertia is offset by fast response. A higher ROCOF drives a lower nadir, which risks tripping under-frequency load shedding (UFLS) relays and settling at a lower frequency — and that is what drives the new frequency-response requirements placed on IBGs.

Section 2

Inertia, ROCOF, nadir and recovery

Four quantities describe every frequency event, and keeping them straight is half the battle. Inertia is the kinetic energy stored in rotating masses. In a synchronous-machine-dominated system that stored energy is released or absorbed immediately when frequency changes, which slows the initial frequency movement. Most grid-following IBGs do not naturally provide grid-visible inertia, because the rotating or dc-side energy is decoupled from the grid by the converter; any inertial-like response must be deliberately created by controls, and it must respect the converter rating and the available energy.

ROCOF is the rate of change of frequency (Hz/s) — how fast frequency is moving immediately after a disturbance. Low inertia gives a higher ROCOF for the same power imbalance. Crucially, ROCOF is not the same as the frequency nadir: ROCOF is the initial slope; the nadir is the lowest frequency reached. The nadir depends on the size of the disturbance, the system inertia, the speed of the frequency response, the load damping, the protection action, and the available headroom from responsive resources.

The link between the imbalance, the inertia and the initial ROCOF is the swing equation, which in its simplest form says the initial rate of change of frequency is:

\[ \left.\frac{df}{dt}\right|_{0} = \frac{f_0\,\Delta P}{2\,H\,S} \]

\(f_0\): nominal frequency (Hz)
\(\Delta P\): the active-power imbalance (generation minus load) at the instant of the event
\(H\): the system inertia constant (seconds)
\(S\): the rated apparent power the inertia constant is referred to

Read it as three plain facts: a larger power imbalance causes a faster frequency movement; more inertia slows the movement; and the frequency response then reduces the imbalance and arrests the fall or rise. No numbers are needed to use it.

After the nadir, frequency may recover to a temporary quasi-steady value. Primary control arrests the fall but may not restore frequency to nominal — it leaves a residual offset. Secondary control (AGC) or a dispatch action is then normally needed to restore the frequency to nominal and to rebuild the reserves for the next event.

Section 3

Match functions to phenomena

Before the frequency detail, a change of viewpoint that governs the whole modelling approach. Rather than asking “what can the inverter do?”, ask “what does this phenomenon need?” Considering every function in every study is not just wasteful; it is often infeasible. So the phenomena that dynamic studies investigate are listed first (Table 1), each mapped to the study types it drives; the rest of this page takes the first row — frequency deviations — and works out which functions matter.

Table 1 — Power-system phenomena and the study types they drive (after CIGRE Table 3.1).
Phenomenon	Key aspects	Type of studies
Response to frequency deviations	Device protection, system support, plant-level control, synthetic inertia, frequency response and regulation	Frequency stability; transient stability
Response to large voltage deviations	Device protection, FRT capability, grid support, (synthetic inertia)	Short-term voltage stability; transient stability; fault-current provision; low/high-voltage ride-through
Response to long-term voltage deviations	V/Q control, permanent limits, plant-level control	Long-term voltage stability
Small-disturbance analysis	Model simplifications for small-signal work	Small-disturbance stability
Unintentional islanding	Islanding detection	Unintentional islanding
Other phenomena	Low- and high-frequency controller interaction, switching transients	Various impact-analysis studies

Section 4

Why the negative-load model is not enough

Despite decades of impact studies, the CIGRE/CIRED survey found that over 30% of utilities and system operators still use no IBG model at all — transmission and distribution system operators (TSOs and DSOs) alike — representing embedded generation instead as a simple reduction in load, a negative load. A negative-load model treats the IBG as a fixed reduction in demand. It may be acceptable for simple steady-state or very basic frequency screening, but it cannot represent active frequency response, trip settings, ROCOF behaviour, ramp limits, headroom, synthetic inertia, recovery, reconnection or protection interactions — so using it for a high-IBG frequency study can be dangerously optimistic or pessimistic, depending on the event.

The reasons operators fall back on it form a familiar list of information gaps:

missing vendor control information;
uncertain protection settings;
unknown frequency-response capability;
unknown headroom or curtailment;
unknown aggregation behaviour of many small devices;
a lack of validated dynamic models.

The justification is fading fast. As IBGs displace synchronous machines and grid codes demand active functions, a negative load — which has no frequency response, no voltage control and no protection behaviour — is no longer adequate. Modelling IBGs correctly, and differently from machines, has become essential.

Section 5

Frequency versus rotor-angle stability — and RMS versus EMT

Frequency-stability studies are classified by the disturbance that provokes them, on a longer horizon than the short-term dynamics above: a large-imbalance case — losing a big unit, studied over up to several minutes — and a slow-perturbation case — small changes and slow plant dynamics, over up to tens of minutes. Rotor-angle (transient) stability can also involve frequency: a fault accelerates all machines, and after it clears a plant may trip, raising tie-line flow, worsening angle stability and dropping frequency — so the power swing overlaps the frequency response.

RMS or EMT for a frequency study?

Most system-wide frequency-stability studies are performed in RMS, because the dominant behaviour is the active-power balance over hundreds of milliseconds to minutes — comfortably within the reach of a phasor model, and far slower than the switching-level detail EMT resolves. EMT is not normally required for the whole network; it becomes important only if the frequency event is coupled with weak-grid converter instability, detailed islanding detection, fast protection, waveform-based ROCOF or vector-jump behaviour, or vendor-specific converter control. For where the two methods sit against each other, see EMT versus RMS on a weak grid.

Section 6

How IBGs respond to frequency

What an IBG can actually do in a frequency event depends on its primary source and its reserve. In practical categories:

PV without storage — can reduce active power during over-frequency, but cannot increase it during under-frequency unless it is curtailed below its maximum-power point.
PV with curtailment / headroom — can increase output up to the available irradiance and the converter limit.
Battery storage (BESS) — can inject or absorb active power quickly, if the state of charge and the converter rating allow.
Wind (WTG) with de-loading — can provide upward response using its headroom.
Wind synthetic inertia — can temporarily extract kinetic energy from the rotor, but the rotor-speed recovery must be modelled.
Micro-turbine / fuel cell / organic Rankine cycle (ORC) — response is set by slower primary-energy dynamics.

Two ideas underpin all upward response. Headroom means the plant is deliberately operating below its available maximum active power, so it can raise output when frequency falls; without headroom or storage, a non-dispatchable renewable plant cannot provide sustained upward frequency response. Curtailment means reducing output below the available renewable power — it looks wasteful in normal operation, but it is what creates the reserve that frequency support draws on. In modelling terms, an IBG’s frequency control has four parts — tripping (over/under-frequency and ROCOF protection), a continuous droop / frequency-sensitive mode (FSM), an emergency LFSM-O / LFSM-U on large deviations, and continuous synthetic inertia acting on measured ROCOF — and what can actually participate depends on the primary source’s controllability and limits, the storage capacity and state of charge, the controller structure and settings, and the protection settings.

Section 7

Synthetic inertia, fast frequency response and P(f)

These terms are constantly confused, so pin them down. Synthetic inertia is not physical inertia: it is a controlled active-power response designed to imitate part of the effect of inertia, usually responding to ROCOF or frequency deviation and injecting extra active power for a short period. It is delayed by measurement, filtering, detection and converter control, so it cannot be exactly identical to synchronous inertia. Fast frequency response (FFR) is a broader term: any active-power response delivered quickly after a frequency event, triggered by ROCOF, frequency deviation, a dead-band crossing or a control signal, and delivered by a BESS, curtailed PV, wind headroom or other fast controllable resource. Set out side by side:

Inertia — the natural, immediate response of a rotating mass.
Synthetic inertia — a controlled short-term response, often ROCOF-based.
Fast frequency response — a fast controlled active-power response, not necessarily inertia-like.
Primary frequency response — droop-based response to the frequency deviation, over seconds.

The recovery problem — and the plant limits

After a wind turbine releases kinetic energy, its rotor speed may need to recover; during recovery the active power can dip below the pre-event value, and if many turbines recover together this can cause a second frequency dip — so the recovery strategy must be modelled, not just the release. The plant limits differ by technology. For a BESS, the response is fast but not unlimited: the model should include the power rating, the energy capacity, the state of charge, the active-power limits, the response delay, the ramp rate and any reserve-restoration logic. For PV, the available active power changes with irradiance: at maximum power it has no upward reserve; if curtailed, the difference between available and actual output is the upward reserve; and even for over-frequency, the ramp and minimum-power behaviour should be represented.

The continuous control is P(f): active power changed as a function of frequency. On under-frequency the plant increases active power if it has headroom or storage; on over-frequency it reduces active power to help arrest a high frequency. Three settings shape it. Droop defines how much active power changes for a given frequency deviation. The dead-band is the small frequency range around nominal where the controller deliberately does not respond, to avoid unnecessary active-power movement during normal small variations. The ramp rate limits how quickly the active power may change — even if the control requests a fast response, the plant may not be allowed, or able, to change output instantly.

One asymmetry is worth stating plainly, because it shapes every result: over-frequency and under-frequency are not symmetric for many IBGs. Reducing generation during over-frequency is usually easy; increasing generation during under-frequency requires stored energy, headroom, or a primary source that can raise power quickly.

Section 8

Rising ROCOF and false islanding

The initial ROCOF after an imbalance is purely a function of the system inertia at that instant. As IBGs displace machines, the inertia of a small network or an islanded area can fall sharply, so the same generation–demand imbalance now produces a higher ROCOF. Synthetic inertia can help limit it, but today it is limited in extent and not as fast as a machine’s inherent, immediate response — so the frequency can drop significantly in the first seconds, before primary control begins to act, deepening the nadir and increasing the amount of load shedding.

ROCOF immunity versus ROCOF protection

ROCOF immunity means the IBG should stay connected during specified rates of frequency change. ROCOF protection means the device may trip if the ROCOF exceeds a threshold. These two must be coordinated — otherwise an IBG may trip during exactly the frequency event where the system needs its support. And there is a subtler trap: a high ROCOF or a vector jump during a real system disturbance can look like islanding to some protection algorithms. If the settings are too sensitive, IBGs may trip unnecessarily, deepening the frequency event — the same tension examined on the islanding page. Many small systems rely on ROCOF to detect islanding, so when a unit trips and the ROCOF is now higher than it used to be, those systems are liable to falsely detect an island and disconnect — which is why both ROCOF protection and ROCOF immunity belong in a frequency-stability model.

Section 9

Over- and under-frequency, LFSM-O and the 50.2 Hz trap

Over-frequency follows generation exceeding load — a tie-line trip of an exporting area, the formation of an island, or the loss of a large load or an exporting HVDC link. Under-frequency follows the mirror causes — a tie-line trip of an importing area, an island, or the loss of a large unit or an importing HVDC link; fast primary response, delivered in the first seconds, may be needed if the nadir is reached before governor action becomes effective.

The main over-frequency tool is LFSM-O — Limited Frequency-Sensitive Mode, Over-frequency — a grid-code function that reduces generation when frequency rises above a defined threshold, helping bring the high frequency back down by reducing active-power injection. Its behaviour is set by a start frequency, a droop / slope, a dead-band, a minimum stable output, a ramp-rate limit, and a reconnection / restoration behaviour. Its mirror, LFSM-U (under-frequency), increases active power during low frequency — but only if the plant has headroom, storage or fast primary-energy capability. In an active-current form, the over-frequency reduction can be written:

\[ I_{pHFRT} = I_{pini} + K_{HFRT}\left( f - f_{start} \right) \]

\(I_{pini}\): the active current at the start of the over-frequency response
\(f\): the measured system frequency
\(f_{start}\): the frequency at which the reduction begins
\(K_{HFRT}\): the droop gain — negative, so the active current falls as the frequency rises

Sign convention: because \(K_{HFRT}\) is negative, the active power (current) reduces as the frequency increases above \(f_{start}\) — over-frequency droop applied to generation. Always check the sign convention of the tool you use.

The 50.2 Hz problem, and controlled reconnection

The 50.2 Hz problem is the historical risk that many distributed generators disconnect — or later reconnect — together around a similar frequency threshold near 50.2 Hz. If many small devices act simultaneously, their combined response can create a large system disturbance: simultaneous tripping worsens the frequency and voltage event, and simultaneous reconnection creates a sudden power step. Modern grid codes therefore require staggered response, ride-through and controlled reconnection. Automatic reconnection must be modelled carefully: if many IBGs reconnect at nearly the same frequency or after the same time delay, they create a sudden active-power increase; staggering, ramping and randomised delay reduce the risk. The direct lesson for modelling is that the over/under-frequency protection and the automatic-reconnection behaviour of distributed generation must be represented, or the study will simply not see the event.

Section 10

Grid size changes everything

The impact of IBGs in a small isolated island is fundamentally different from their impact in a large interconnected grid — in the type of source, the achievable penetration and the critical deviation that must be studied. Table 2 sets the two side by side.

Table 2 — Small isolated grid versus large network, and the frequency control that dominates each (after CIGRE Table 3.2).
Aspect	Small isolated grid	Large network
Main generation	Diesel engine, gas turbine, small hydro	Thermal (coal, oil, gas), hydro, nuclear
Likely IBG penetration	Can exceed 50% (a possible high-penetration condition, not a rule)	Not likely over 50%
System inertia	Small with a high IBG share, \(H \lt 3\) s	Inertia constant \(H \approx 3\)–\(6\) s
Critical load / generation change	Over seconds to a dozen seconds	Over minutes to a dozen minutes
Key frequency control	Governor / primary frequency control	AGC / secondary frequency control

Why the two need different model detail:

A small isolated grid has low inertia and is highly sensitive to a single plant’s output changes, with steeper frequency ramps — so primary control and the fast IBG response are critical, and the individual plant models matter. In a small island of a few MW, one wind farm, PV plant or BESS can be a large percentage of total generation; its ramp limits, protection, frequency response and recovery behaviour can materially change the nadir and the ROCOF, so aggregate negative-load modelling is usually not enough. A steep 70–80% change in IBG output within 10 seconds can swing the whole island’s frequency, and not every unit has AGC — so primary control is the key to the analysis.
A large interconnected grid has larger inertia and diversity, so the average frequency moves more slowly and ramps from solar/wind forecast error matter more than single trips; secondary control, reserve scheduling and inter-area support dominate. But it is not immune: one plant may be small, yet many similar IBGs acting together can create a system-level effect, so the aggregation must preserve the combined response, protection settings, ramping and reconnection behaviour. Because large thermal units have low ramp limits (about 1–5% per minute), secondary frequency control (AGC) is the key, against a slow trend of a few per cent per hour.

Section 11

Real-world lessons

The CAISO solar eclipse. This example is useful because it shows that frequency security is not only about sudden generator trips: predictable renewable ramps, reserve scheduling and headroom management also matter. During the eclipse of 21 August 2017 across the United States, the California Independent System Operator (CAISO) saw about 6 GW of utility-scale solar and a further 1.46 GW behind the meter ramp down at roughly 48 MW per minute and, on recovery, up at as much as 150 MW per minute. Because the eclipse was predictable, the operator needed enough upward and downward reserve and ramping capability, pre-procured regulation reserves and used its real-time imbalance market to ride it out — and BESS or curtailed resources can provide exactly that headroom and reserve if it is planned.

Headroom is not the whole story. In a CAISO study of losing two nuclear units (about 2.6 GW), the modelled frequency response fell below the operator’s obligation and looked optimistic against real events — partly because the case carried a large modelled headroom. The lesson is that the amount of headroom is a good indicator of frequency response but not the only one: a lot of headroom concentrated on a few governor-responsive units can deliver less frequency-control capability than a smaller headroom spread across many units.

Restoration and the 50.2 Hz effect. During grid restoration the system is weak and lightly connected, so frequency is more sensitive and the protection and reconnection behaviour of distributed generation becomes important — uncontrolled reconnection can disturb the restoration. A European grid-restoration study started at 51.5 Hz to leave a margin above the 49 Hz load-shedding threshold, but the system then has to bring frequency close to nominal to synchronise with its neighbours — and under the older rules distributed generation reconnected at 50.2 Hz, so crossing that value can trip and reconnect distributed generation and destabilise the restoration. All three cases point the same way: the over/under-frequency protection and the automatic reconnection belong in the model.

Section 12

Which functions to model for a frequency study

Now the payoff. First a word on the primary source, because it is easy to over-simplify: the inverter can change current quickly, but the primary source decides whether energy is actually available. For frequency studies the primary source cannot always be neglected — the short-term inverter response, the medium-term energy recovery and the long-term reserve restoration may all depend on PV irradiance, wind rotor speed, BESS state of charge or a turbine/fuel-cell ramp limit. Two related modelling calls follow. MPPT can often be ignored during a very short frequency event if irradiance is constant and the plant is not providing frequency response; it matters when the study includes active-power reserve, headroom, solar ramping, reconnection or recovery over seconds to minutes — and even then a simplified available-power and active-power-reference model is usually enough, not switching-level detail. dc-link dynamics may be unnecessary for most RMS frequency studies, but can matter if the response uses fast active-power injection, if converter limits are reached, or if a BESS / PV dc-side control or a momentary-cessation / recovery affects the active power.

One protection mechanism deserves its own note: the vector jump, a passive anti-islanding scheme that trips on a step in voltage phase. Across a reactance \(X\) between two buses, the transferred active power obeys the familiar power-angle relation:

\[ P = \frac{V_1\,V_2}{X}\,\sin\delta_{12} \]

\(P\): active power transferred between the two buses
\(V_1,\ V_2\): the two bus voltage magnitudes
\(X\): reactance between the buses
\(\delta_{12}\): the angle difference between the two bus voltages

If the active-power flow changes without a change in voltage magnitude, the angle \(\delta_{12}\) must step. A generator trip — or the disconnection of an IBG — produces exactly such a step, which can cause an unwanted vector-jump trip and cascade a further frequency drop. That is why vector jump and anti-islanding matter for a frequency study.

Read Table 3 as a frequency-stability modelling guide, not a universal IBG checklist: a “Yes” means the function can change the ROCOF, the nadir, the recovery, the protection operation or the active-power balance; a “No” means it is normally not decisive for the frequency phenomenon unless a special project condition makes it relevant. The ranking is for general studies; a specific study with special controls or system conditions may differ.

Table 3 — Recommended IBG functionalities for a frequency-deviation study, ranked by necessity (after CIGRE Table 3.3).
Category	Functionality	Model for frequency?	Note
Control	DC source control	No
Control	Current control	No	much faster than the system frequency; enters via active-power / current limits and delays
Control	PLL	No	not the limiting mechanism for ordinary frequency studies
Control	MPPT	Yes — small island only	Only for long-term small-island runs; otherwise model as a change in \(P_{ref}\)
Protection	Reduce max inverter current on dc over-voltage	No
Protection	Limit inverter current rate of change after a fault	No
Protection	Current limit	Yes	Shapes the achievable active-power increase during a dip
Protection	DC over-voltage protection	No
Protection	Over / under-voltage protection	No
Protection	Over / under-frequency protection	Yes	Include automatic reconnection
Protection	Detecting a balanced fault	No	Positive-sequence over-voltage protection may be used
Protection	Detecting an unbalanced short-circuit fault	No	Negative-sequence over-voltage protection may be used
Protection	Detecting a single-line-to-ground fault	No	Zero-sequence over-voltage protection may be used
Protection	ROCOF tripping	Yes
Protection	Vector jump	Yes	Passive anti-islanding on a phase step
Protection	Transfer trip	No	Yes if a large block of IBG is tripped
Protection	Anti-islanding active detection	Yes	Not required by all utilities
Capability	P(f) control (over / under frequency)	Yes	Droop on active power
Capability	Voltage control by reactive power	Yes	Pre-fault Q affects the dynamic response
Capability	Voltage control by active power, P(V)	No
Capability	Synthetic inertia	Yes	ROCOF-driven; a form of fast frequency response
Capability	ROCOF immunity	Yes	Frequency ride-through; model when its setting differs from ROCOF protection
Capability	Fault ride-through (LV / HV)	No	but its outcome (tripped or connected) can change the active-power balance
Capability	Active behaviour during fast voltage variations	No
Capability	Power oscillation damping (POD)	Yes	Only if implemented and relevant to the studied frequency / common-mode oscillation

Why the fast controls come out and the slow ones stay in. Current control (No): in bulk RMS frequency studies the inner current loop is much faster than the system-frequency dynamics and can be represented by active-power / current limits and response delays — but if the study involves weak-grid converter instability or fast control interaction, that assumption must be checked with EMT or vendor-model validation. PLL (No): PLL dynamics are usually not the limiting mechanism, but they can matter during weak-grid faults, large voltage disturbances, islanding detection or fast frequency/angle transients, where a detailed EMT model may be required. MPPT (Yes, small island): it matters when the plant’s available active power, curtailment point or headroom affects the response — often a simplified available-power / active-power-reference model, not switching-level detail. Current limit (Yes): when frequency falls, the extra active power an IBG can inject is bounded not only by headroom but by the maximum current, and different limiting logic gives a different response. And a word on ride-through: although this page is about frequency, a real disturbance may include voltage dips, swells or faults — if IBGs trip on voltage protection before or during a frequency event, the active-power balance changes, so the LVRT / HVRT assumptions can still affect the frequency result.

A frequency-stability study workflow

1. Define the disturbance: generation loss, load loss, interconnector trip, renewable ramp or a restoration case.
2. Define the initial dispatch, inertia and reserves.
3. Identify the IBG share, the headroom, the BESS state of charge and the available renewable power.
4. Model the active-power frequency response, droop, dead-band, ramp limits and delays.
5. Model the frequency protection, ROCOF immunity, anti-islanding and reconnection.
6. Include UFLS and the generator governor response where relevant.
7. Simulate the ROCOF, the nadir, the settling frequency and the recovery.
8. Check whether any IBG trips, enters the current limit or changes mode.
9. Repeat for high / low renewable output, low inertia and credible contingencies.
10. Validate critical cases against EMT or vendor models where fast control / protection may decide the result.

The signals worth plotting are the ones that reveal the response: the system frequency, the ROCOF and the nadir; the active power from the synchronous generators and from the PV / wind / BESS; the IBG frequency-response output; the BESS state of charge and the wind rotor-speed recovery where relevant; the protection-trip and reconnection flags; the UFLS stages; the reserve / headroom usage; and the frequency-response delay and ramp behaviour — measured at the PCC / POI.

Common mistakes

The traps that most often catch a frequency-stability study — each a theme from the sections above:

Ten traps to avoid

Treating an IBG as negative load at high penetration.
Calling synthetic inertia “real inertia”.
Forgetting that PV at maximum power has no upward reserve.
Ignoring headroom, curtailment and BESS state of charge.
Ignoring wind rotor-speed recovery after synthetic inertia.
Ignoring ROCOF or vector-jump nuisance trips.
Assuming over-frequency and under-frequency response are symmetric.
Ignoring the reconnection behaviour after trips.
Using RMS for a case where fast protection or weak-grid converter control decides the answer.
Modelling fast inverter detail while omitting the plant-level active-power control and reserve logic.

Key points

Model the phenomenon, not the whole inverter

Frequency stability is an active-power balance problem.
Inertia slows the initial frequency movement; lower inertia raises the ROCOF.
The frequency nadir depends on the inertia, the disturbance size, the response speed, the headroom and the protection.
IBGs give no natural synchronous inertia, but can provide controlled FFR or synthetic inertia if energy or headroom is available.
PV at maximum power can reduce output on over-frequency but cannot raise it on under-frequency without headroom or storage.
BESS responds fast but is limited by its rating and state of charge; wind can give synthetic inertia, but the recovery must be modelled.
ROCOF / vector-jump settings can cause unwanted tripping if not coordinated with immunity.
LFSM-O and controlled reconnection are essential to avoid over-frequency and mass-reconnection problems.
RMS is normally suitable for bulk frequency studies, but EMT / vendor validation is needed where fast control, protection or weak-grid behaviour decides the result.

For the underlying behaviours and the inverter itself, see the guides to the characteristics of inverter-based generation and the inverter’s topology, control and protection; for the voltage-stability counterparts, the large-voltage-deviation and long-term voltage-stability guides; and for the model-type choice, EMT versus RMS.

References

The CIGRE/CIRED joint working-group brochure on inverter-based generation is the primary reference for this page; CIGRE TB 450 informs the wind-integration background; the California ISO report documents the 2017 solar-eclipse operation; and the EMTP® documentation is the reference for the RMS and EMT model framing.

CIGRE/CIRED Joint Working Group, Modelling of Inverter-Based Generation for Power System Dynamic Studies. CIGRE Technical Brochure.
CIGRE Working Group, Grid Integration of Wind Generation. CIGRE Technical Brochure 450.
California ISO, Solar Eclipse Readiness and System Operation — 21 August 2017. CAISO.
EMTP®, Renewable and Inverter-Based Generation Modelling — RMS and EMT Documentation. Powersys / EMTP®.

Part 8 Reading now

Modelling IBGs for Frequency Stability

Applying the models to frequency stability: inertia, ROCOF, synthetic inertia and fast frequency response.

Series progress 8 of 12

Part 9

Modelling IBGs for Large Voltage Deviations

Fault-driven events: fault current, ride-through, current limits, momentary cessation and protection.