Renewable Modelling · Large Voltage Deviations

Modelling IBGs for Large Voltage Deviations

System faults throw the voltage down hard, and how a plant behaves in those few seconds decides whether the grid rides through or cascades. Synchronous machines answer with physics — a big fault current, an internal voltage, synchronising torque. An inverter-based generator (IBG) answers with control, within a current ceiling, and can just as easily disconnect. This guide works through the fault-response phenomena — out-of-step, short-circuit current, short-term voltage recovery — the studies they drive, the counter-intuitive way fault ride-through helps or hurts transient stability depending on where the IBG sits, the 2016 Blue Cut fire that lost ~1200 MW of PV, and which functions you actually need to model for a large-voltage-deviation study. It continues the companion guides on the characteristics of IBG and frequency-stability modelling.

Reading time ≈ 36 min · Fault response, real cases & the functions that matter

The frequency-stability guide dealt with the active-power balance, where the main issue is a sudden mismatch between generation and load and the frequency moves. This guide deals with large voltage deviations, where the trigger is normally a fault and the main concerns are voltage, current, fault ride-through, protection and post-fault recovery. The two overlap — a fault can also disturb frequency if generation or load trips with it — but here the focus is the voltage-driven part of the disturbance. When a fault throws the voltage down, a synchronous machine answers with physics: a large fault current, an internal voltage behind reactance, synchronising torque. An inverter-based generator (IBG) answers with control, inside a current ceiling, and can just as easily disconnect or momentarily stop injecting. So the fault response of an IBG is a control-and-protection question, not a physical one, and getting a large-voltage-deviation study right means representing exactly the functions that decide whether the plant rides through or drops out.

What a large voltage deviation is

A large voltage deviation is a severe departure of bus voltage from its normal operating range, usually caused by a short circuit, fault clearing, line trip, transformer trip or major switching event. It is different from a small voltage variation because it can activate current limits, protection, ride-through logic, phase-locked-loop (PLL) problems, motor stalling and converter mode changes — effects that a small variation never reaches. Because those mechanisms are fast and non-linear, the study cannot be a simple steady-state check; it has to follow the event through time.

The study time scale

Large-voltage-deviation studies normally cover the first few cycles to several seconds after a disturbance. That window contains the fault current, the protection operation, the converter current limiting, the low- and high-voltage ride-through (LVRT / HVRT) behaviour, the induction-motor stalling and re-acceleration, and the early post-fault voltage recovery. Slower phenomena — long-term voltage control, tap-changer action — belong to the long-term voltage-stability guide, not here.

A fault event, step by step

1. A fault occurs and the voltage collapses near the fault.
2. Fault current flows from synchronous machines, grid sources and IBGs.
3. Protection detects the fault and the breakers clear it.
4. IBGs either ride through, current-limit, enter momentary cessation, or trip.
5. Motors and other loads attempt to recover.
6. Voltage either recovers or stays depressed.
7. The system remains stable, loses synchronism, trips generation, or cascades.

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)

OOSOut-of-step (loss of synchronism)

FRTFault ride-through

LVRT / HVRTLow- / high-voltage ride-through

PLLPhase-locked loop

PCCPoint of common coupling

POIPoint of interconnection

CCTCritical clearing time

X/RReactance-to-resistance ratio of the fault path

IMInduction motor

PVPhotovoltaic

BESSBattery energy storage system

DFIGDoubly-fed induction generator

WTGWind-turbine generator

SVCStatic var compensator

STATCOMStatic synchronous compensator

PODPower oscillation damping

ROCOFRate of change of frequency

MPPTMaximum power point tracking

IGBTInsulated-gate bipolar transistor

PRC-024-2NERC standard on generator frequency / voltage relay settings

Key idea

In a fault a machine gives a large fault current, an internal voltage and synchronising torque; an IBG gives a controlled current near its ceiling (an example figure is \(\approx 1\) pu RMS), no inherent voltage source, and can trip or momentarily cease — so fault response is a control and protection question, not physics.
Out-of-step is a synchronous-machine phenomenon (full-converter IBGs have no grid-coupled rotor angle), but high IBG penetration still erodes the inertia, reactive reserves and voltage regulators that keep machines in step — and IBGs cut the short-circuit current that protection relies on.
Fault ride-through is not automatically good for stability: keeping IBGs connected helps transient stability when they sit near a load sink, but can hurt it near a source. Location and the active-power behaviour are decisive.
Model the fault-response functions — current limit and priority, LVRT/HVRT, the PLL, dynamic reactive-current support, and the trip / momentary-cessation protections. The 2016 Blue Cut fire lost ~1200 MW of PV through exactly those paths.

Key terms used on this page

01Out-of-step (OOS)

Loss of synchronism: one group of synchronous machines no longer rotates in step with another, their rotor angles separating past recovery.

02Electrical centre

The network point that sees a near-zero voltage — like a short circuit — when two areas are about \(180^\circ\) apart.

03Short-circuit current

The large current a source supplies into a fault; an ac component plus a dc offset that decays at a rate set by the X/R ratio.

04X/R ratio

Reactance to resistance of the fault path; a higher X/R means a slower-decaying dc offset and a higher peak asymmetrical current.

05Fault ride-through (LVRT / HVRT)

Staying connected through a low (LVRT) or high (HVRT) voltage excursion instead of tripping, within a defined voltage–time envelope.

06Momentary cessation

The inverter temporarily stops (or greatly reduces) current injection while staying electrically connected, then resumes after a delay.

07Current limit & priority

The converter’s maximum current, shared between active and reactive components; the priority rule decides which is served first.

08Dynamic reactive-current support

Fast injection (or absorption) of reactive current during a voltage deviation to help the voltage recover.

09PLL

Phase-locked loop; it estimates the grid voltage angle and must keep tracking through a fault or the converter current can be wrong or unstable.

10Short-term voltage stability

Whether voltage recovers within a few seconds of a fault, shaped by reactive support, motor stalling and whether IBGs stay connected.

11Induction-motor stalling

Motors failing to re-accelerate after a dip, drawing heavy reactive current and holding the voltage down.

12Transient stability

A synchronous machine’s ability to stay in step after a large disturbance (large-disturbance rotor-angle stability).

13CCT

Critical clearing time: the longest a fault can persist and the system still stay stable; faster clearing generally improves stability.

14Net load

Load minus IBG output; when IBGs trip or cease, net load jumps and depresses the voltage.

15PRC-024-2

A NERC reliability standard on generator frequency and voltage protective-relay settings; relevant because those settings decide whether generation stays connected.

16Blue Cut fire

A 2016 Southern California event in which transmission faults led to the loss of ~1200 MW of PV — a modelling lesson, not a universal rule.

Symbols used on this page

\(\Delta T_e\) — change in electrical torque after a disturbance; \(\Delta\delta,\ \Delta\omega\) — rotor-angle and rotor-speed deviations (Section 5).
\(K_s\) — synchronising-torque coefficient (in phase with \(\Delta\delta\)); \(K_d\) — damping-torque coefficient (in phase with \(\Delta\omega\)).
“pu” — per unit, a quantity expressed as a fraction of its rated value.

Section 1

Large voltage deviations: what happens

Large voltage deviations are triggered by system faults — three-phase, single-line-to-ground and other short circuits. Unless a large fraction of the generation or load changes with them, frequency can be set aside; the concern is the voltage excursion and what it sets off. Severe faults can push machines out of step and cascade to a wide-area blackout, which is exactly why planners and operators study the response to large voltage deviations so carefully.

Three phenomena travel with a fault: the out-of-step condition, the short-circuit current, and the short-term dynamic voltage response after the fault clears. The sections below take each in turn — first how a synchronous machine and an IBG differ in a fault, then the three phenomena, then the ride-through and current-limit behaviour, and finally how they map to the studies and the IBG functions each one needs.

Section 2

Synchronous machines versus IBGs in a fault

Start with the machine, because it is the behaviour every protection scheme was built around. A synchronous machine naturally contributes a high fault current, because it has a rotating magnetic field and stored electromagnetic and kinetic energy. Its current contains an ac component at system frequency and a decaying dc offset, and its internal voltage behind reactance together with its rotor-angle dynamics are central to transient stability. Feed a fault and the machine simply delivers — often several times rated current in the first cycles — from physics, not from a control loop.

An IBG does not behave like this. During a fault a grid-following IBG normally behaves like a controlled current source: it injects current according to its converter controls and current limits, and it cannot inject unlimited current. What it injects — and whether the plant even stays connected — is decided by the converter current ceiling, the PLL tracking, the voltage measurement, the fault-ride-through logic, the active/reactive current priority and the protection settings. Nothing here is automatic physics; it is all design and configuration. (A Type-3 doubly-fed induction generator, or DFIG, is a partial exception: its stator is directly connected to the grid, so its fault contribution sits between a full converter and a synchronous machine.)

Why IBG fault current is limited

Converter semiconductor devices — typically insulated-gate bipolar transistors (IGBTs) — have strict thermal and overcurrent limits. To protect them, IBGs usually limit fault current to around rated current or a modest overload, unlike synchronous machines that may initially contribute several times rated current. The exact ceiling is vendor- and design-dependent; do not assume a single value applies to every technology.

Some approximate figures help build intuition, provided they are read as examples and not universal constants. Measurements on a large Type-4 (full-converter) PV plant give a peak fault current that generally does not exceed rated by more than about 30%, decaying quickly as the control holds the current near rated, so the RMS fault contribution is roughly 1 pu. These are practical example values, not fixed rules: the actual IBG fault current depends on the converter rating, the control mode, the voltage level, the current priority, the grid code and the vendor protection — always check vendor data and the grid-code requirement for the specific plant.

The waveform differs too. A synchronous-machine fault current commonly shows an ac component plus a dc offset whose size depends on the fault-inception angle and the network X/R ratio, decaying naturally with the circuit time constant. A converter fault current is more controlled: it is shaped by the current loop and the limiter and may not follow the same natural decay pattern at all. That single difference — natural decay versus commanded shape — is why so much fault-response modelling for IBGs is really controller-and-protection modelling.

A note on grid-forming converters

Grid-forming converters may respond differently from the grid-following converters described above, because they behave more like a controlled voltage source and can present a stiffer voltage during a disturbance. But they still have current limits and protection, so their large-disturbance behaviour must be modelled with the correct control mode and vendor data — grid-forming control does not by itself remove the current ceiling or automatically solve fault recovery.

Section 3

Short-circuit current, and why IBGs reduce it

Short-circuit current contribution is the current a source supplies during a fault. Protection engineers use it to check relay operation, breaker duty, the depth of the voltage depression and the adequacy of the fault level. A short circuit forces the voltage between conductors to (near) zero and draws a large current with two parts: an ac component and a dc component that decays at a rate set by the X/R ratio of the faulted circuit. Line, transformer and generator protections lean, at least implicitly, on the contrast between the fault current and the pre-fault load current, and the quantity used for relay settings is usually an RMS value.

When synchronous machines are replaced by IBGs, the total fault current tends to reduce — most sharply when the IBGs sit near the fault, because an inverter limits its current to about its rated value (oversizing buys only a little more). Lower fault current can make overcurrent protection less sensitive: the relay may see less current, a different phase angle, a shorter duration and less dc offset, so distance, differential, directional and voltage-based schemes may all be affected depending on the network, and protection may need redesign. IBGs can inject additional reactive current to support the voltage during the fault and its recovery if the operator requires it, but that dynamic support is limited and is not comparable to the fault current a synchronous machine delivers near the fault.

A controllable current source — with a coordination catch

Because the contribution is set by power electronics, an IBG can program its fault current and response time, which can even help protection coordination. The inverter interface uses either a voltage-control or a current-control scheme, with the dc-link capacitor holding the dc voltage near constant through transients: a voltage-control scheme gives a higher initial current overshoot, a current-control scheme rises more slowly and settles back — but in either case the fault current cannot appear in the first instant, because it waits on the PLL to lock and the control to act. A single small unit contributes little; but the total from many small units can shift the fault level enough to spoil overcurrent coordination, cause wrong fuse operation or hamper fault detection. In islanded LV/MV situations the IBG fault contribution can be significant for coordination, depending on the size of the isolated grid.

Section 4

Out-of-step and the electrical centre

Out-of-step (OOS) means one group of synchronous machines no longer rotates in synchronism with another group. Their rotor angles separate so far that the electrical torque can no longer pull them back together — it is a rotor-angle stability problem. When two buses (including a machine’s internal, fictitious bus behind its transient reactance) reach an angle difference of about \(180^\circ\), there is a point in the network where the voltage is theoretically zero: the electrical centre.

The electrical centre is the point where the voltage becomes very low, or theoretically zero, when two equivalent voltage sources are approximately \(180^\circ\) apart. At that point an out-of-step condition can look electrically similar to a fault — but it repeats and moves as the machines swing, which a real fault does not. In practice OOS usually involves more than two machines, and it matters more as the network grows larger and the voltage level higher. Table 1 sets the two apart. This table helps distinguish a real short circuit from an out-of-step electrical-centre condition: both can create very low voltage at a point in the network, but their duration, movement, repetition and protection response differ.

Table 1 — A short circuit versus the out-of-step condition at the electrical centre (after CIGRE Table 3.4).
Property	Short-circuit fault	Out-of-step (OOS)
Duration of the near-zero voltage	Tens of ms to a few hundred ms	Instantaneous (each pass through the electrical centre)
Movement of the near-zero-voltage point	Fixed at the fault location	Moves through the network as the machines swing
Repetition of the zero-voltage condition	Generally once	A few times per second until the OOS relay operates
Protection response	Cleared by fault protection (overcurrent, distance, differential)	Detected by out-of-step / blocking relays; must not be tripped as a fault

The practical message is that OOS protection and fault protection must be coordinated: a swinging electrical centre should not be treated blindly as an ordinary permanent fault, or the wrong protection may operate.

Full-converter IBGs are not subject to classical OOS. PV has no rotating mass behind its inverter, and Type-4 WTGs and converter-driven machines connect non-synchronously, so they have no directly grid-coupled rotor angle and cannot go out of step in the synchronous-machine sense. But that does not make them irrelevant. High IBG penetration can still affect OOS risk by changing the dispatch, reducing the synchronous inertia, reducing the synchronising torque, changing the power flows, reducing the fault current and altering the voltage support — it erodes the very reserves and regulators that keep the remaining machines in step. Those are the differences set out in the characteristics guide, and they are why requirements such as IBG fault ride-through have been imposed.

Section 5

Transient stability, CCT and fault location

Transient stability is the ability of synchronous machines to remain in synchronism after a large disturbance such as a short circuit; the key question is whether the rotor angles separate and then recover, or continue to diverge. IBGs affect it indirectly — through voltage support, active-power behaviour, fault ride-through, and the amount and location of synchronous generation they displace. Rotor-angle stability rests on two torque components: a synchronising torque in phase with the angle deviation, and a damping torque in phase with the speed deviation. Decomposing the electrical torque:

\[ \Delta T_e = K_s\,\Delta\delta + K_d\,\Delta\omega \]

\(\Delta T_e\): change in electrical torque after a disturbance
\(K_s\): synchronising-torque coefficient (in phase with the angle deviation)
\(K_d\): damping-torque coefficient (in phase with the speed deviation)
\(\Delta\delta,\ \Delta\omega\): rotor-angle and rotor-speed deviations

Both matter: too little \(K_s\) lets the angle run away on the first swing; too little \(K_d\) leaves a growing or poorly damped oscillation. The system inertia and damping set how the power swing decays, so anything that changes them — including IBGs displacing machines — changes rotor-angle stability.

Critical clearing time (CCT)

The critical clearing time is the longest a fault can remain before it is cleared while the system still stays stable. Faster clearing generally improves transient stability, because the machines have less time to accelerate away from synchronism. CCT is not a fixed number: it depends on the system condition — the dispatch, the fault type and location, the network topology and the inertia at that moment. Because relay clearing times are essentially fixed (the shortest extra-high-voltage, EHV, clearing is around 3 cycles), operators often lean on other indices such as the area exchange and the maximum power transfer.

The amplitude of the post-fault power swing depends on the loads, their type and location, and on nearby IBGs — residential PV in particular. The critical case is PV that disconnects (or momentarily ceases) right after the fault, which is equivalent to a jump in net load. Where that jump lands is decisive, and it is genuinely counter-intuitive, so take it step by step:

IBG near a load sink. Ride-through keeps local generation and voltage support online. Losing the PV would make the load pull more power over the line, so the transmission flow rises and the swing can grow — therefore here LVRT (staying connected) improves transient stability and reduces the import.
IBG near a generation source. Losing the PV makes the nearby net load draw more of the station’s own output, so the transmission flow away from the station falls. Keeping the PV connected can then worsen transient stability, and PV without LVRT near a source can actually improve it, because the plant’s active-power recovery changes the accelerating power differently.
The effect is system-specific. It must be studied for the actual network and fault, not assumed from a rule of thumb.

So the IBG’s active-power behaviour following a fault, and its network location, are the two key factors in a transient-stability study. Because PV is spread across the network and it is the units near the fault that dominate, the influence of the IBG location is, in effect, the influence of the fault location.

Section 6

Fault ride-through: LVRT, HVRT and momentary cessation

LVRT means the IBG must remain connected during a defined low-voltage event instead of tripping immediately. The LVRT curve normally gives voltage thresholds and the allowed duration at each; if the voltage stays below the curve for too long, the plant may trip. But LVRT is not only a protection curve — it also interacts with the current limits, the reactive-current injection, the active-current reduction and the post-fault recovery, so it cannot be modelled as a threshold alone.

HVRT means the IBG must remain connected during a defined high-voltage event. HVRT matters after fault clearing, load rejection, capacitor switching or weak-grid over-voltage, and the model should include the thresholds, the delays and the reactive-current behaviour where required. FRT — fault ride-through — is the umbrella term: LVRT and HVRT are the low- and high-voltage parts of the broader ride-through requirement, and the model must represent both the stay-connected envelope and the behaviour while inside the envelope.

Momentary cessation, and how it differs from a trip or a block

Momentary cessation means the inverter temporarily stops injecting current while remaining electrically connected. It is not the same as a full trip, but to the grid it can still look like a sudden loss of active and reactive current — and if many IBGs enter momentary cessation together, voltage and frequency recovery can be severely worsened. It helps to separate four distinct behaviours:

Current limiting — the inverter keeps injecting current, but within its ceiling.
Momentary cessation — the inverter stops or greatly reduces current temporarily, then returns after conditions recover.
Trip — the inverter disconnects and normally requires a reconnection delay.
Blocking — the gate pulses are blocked to protect the converter; depending on the design this may behave like cessation or like a trip.

This distinction is essential: a model that captures the ride-through envelope but treats cessation as a trip (or ignores it) will get the recovery badly wrong.

Section 7

Current limits, reactive support and P/Q priority

Dynamic reactive-current support means the IBG injects or absorbs reactive current rapidly during voltage deviations to support the recovery. During a voltage dip many grid codes require reactive-current injection; during an over-voltage, reactive-current absorption may be required. Most inverters inject up to about 1 pu reactive current shortly after a fault to help the voltage recover, and this generally supersedes the pre-fault control mode — so whether the inverter was on voltage, power-factor or reactive-power control makes little difference to its fault contribution. But the support is limited by the total current rating: more reactive current means less active current.

Priority, and the total current limit

During a fault the active and reactive current commands compete for the same converter current capacity. Reactive-current priority gives voltage support first and reduces the active current if necessary; active-current priority tries to maintain active power first and may provide less voltage support. The selected priority can change the study result, so it must be set deliberately. And the converter cannot independently provide any requested active current and any requested reactive current: the vector sum must stay within the maximum current limit \(\left(\sqrt{i_d^2 + i_q^2} \le I_{max}\right)\). Any LVRT/HVRT reactive-current requirement therefore has to be read together with the current limiter — the limiter, not the requirement, decides what the plant can actually deliver in a deep dip.

The PLL is the other fault-critical block. The PLL estimates the grid voltage angle; during a deep voltage dip, an unbalanced fault, or a weak-grid condition, that angle may be distorted or hard to measure. If the PLL loses accurate tracking, the active- and reactive-current injection can be wrong, delayed or unstable — which is one of the main reasons EMT or detailed vendor models are needed for severe faults.

Why very low voltage is hard for an RMS current-source model

Active power is roughly the product of voltage and active current. At very low voltage, delivering any meaningful active power would require an unrealistically high current — which the current limit forbids. A pure current-source RMS model that tries to inject fixed active and reactive current into a near-zero-voltage bus can therefore run into numerical convergence difficulty, because active and reactive current can no longer be injected independently. Deep voltage dips need EMT (for example in EMTP®), or a carefully limited RMS model that reduces the current commands as the voltage falls.

Section 8

Short-term voltage stability and motor behaviour

Short-term voltage stability is the ability of the voltage to recover within seconds after a fault. It is strongly affected by dynamic reactive support, motor stalling, load behaviour, protection action and whether the IBGs remain connected. The dominant load-side mechanism is the induction motor (IM). During a voltage dip an induction motor slows down, because its electrical torque falls; if the voltage does not recover quickly enough, the motor may stall. A stalled motor draws a high reactive current, which depresses the voltage further and can delay or prevent recovery — classically summer air-conditioning load. Drive-fed or inverter-interfaced motors, common in domestic appliances, are much less prone to this.

After the fault clears, motors need voltage and reactive power to re-accelerate. If IBGs, battery energy storage systems (BESS), STATCOMs, SVCs or generators provide enough reactive support, the recovery is faster; if not, motors may stay stalled and hold the voltage low until slower thermal protection disconnects them. IBGs act in the same time frame, so they matter here. An IBG without dynamic reactive support may worsen the post-fault recovery, because it does not provide reactive current when the motor load needs it most; worse, cutting an IBG’s active power raises the net load and lowers the load-bus voltage. Conversely, a properly controlled IBG can improve recovery if it stays connected and supports the voltage — though dynamic reactive support can also overshoot and lift the voltage too far, which is why HVRT matters as much as LVRT.

A single-phase PV example on a power-system simulator shows the failure mode. After a fault the faulted line is removed and later restored, and the network looks stable at 30 seconds — yet the single-phase PVs on one line-to-line voltage fail to restart and stay disconnected, because they do not meet the FRT requirement. Their loss raises the net load and drops the load-bus voltage. Sitting at 70% of rated voltage is abnormal and cannot be sustained: on a P–V (nose) curve, resistive load can operate at the lower voltage, but constant-power load cannot and must be disconnected. A Western Electricity Coordinating Council (WECC) study of an 800 MW PV plant, with a three-phase fault at the 230 kV point of interconnection, put six scenarios side by side (Table 2), and the results (Table 3) are instructive: with PV voltage control the transient voltage at the PCC is higher, better damped and recovers faster (with some overshoot); induction-motor stalling delays the recovery at the load bus, where PV voltage control has little effect.

Table 2 — The six WECC voltage / reactive-control scenarios (after CIGRE Table 3.5).
Scenario	Source	Power factor	Q & V control	Motor stalling	Current priority
1	Solar PV	±0.95	Yes	No	Q priority
2	Solar PV	±0.95	Yes	Yes	Q priority
3	Solar PV	1.0	No	No	P priority
4	Solar PV	1.0	No	Yes	P priority
5	Thermal	N/A	N/A	No	N/A
6	Thermal	N/A	N/A	Yes	N/A

Table 3 — Summary of the short-term dynamic voltage response (after CIGRE Table 3.6).
Scenario	Transient voltage at generator terminals	Post-fault steady-state voltage	Slow recovery on adjacent load buses
1 — PV, voltage control, motors do not stall	High	Normal	No
2 — PV, voltage control, motors stall	High	Normal	Yes
3 — PV, no V/Q control, motors do not stall	Normal	Low	No
4 — PV, no V/Q control, motors stall	Normal	Low	Yes
5 — Thermal, motors do not stall	Low	Normal	No
6 — Thermal, motors stall	Low	Normal	No

The broad lessons: IBGs can help transient voltage performance when they regulate voltage — better damping and faster recovery from low inertia and fast control; without LVRT, losing a large amount of distributed PV on a fault is a real concern; and high distributed-PV penetration can raise the steady-state voltage under normal conditions. For large-system studies, very detailed generator models are not necessary.

Section 9

Balanced versus unbalanced faults

The fault type changes what the study can see. Three-phase (balanced) faults are useful for worst-case voltage depression and for transient-stability screening: they collapse all three phases together and give the deepest symmetrical dip. Single-line-to-ground and other unbalanced faults are essential when negative-sequence control, phase-specific protection, PLL behaviour under unbalance, or unbalanced voltage recovery matters — and they are the far more common fault in real networks. EMT is normally stronger than RMS for detailed unbalanced converter behaviour, because the phase quantities and the converter’s sequence-dependent response are represented directly rather than as positive-sequence phasors.

Sequence components, briefly

An unbalanced fault is usually analysed with symmetrical components, and the protections in the functions table below are named after them, so it is worth stating what they are:

Positive sequence — a balanced three-phase set in the normal phase order; the component present in normal balanced operation.
Negative sequence — a balanced set in reverse phase order; it appears during unbalanced faults and is a strong signature of unbalance.
Zero sequence — three phases in phase with one another; its presence depends strongly on the system grounding and the transformer connections.

Section 10

Real-world: the 2016 Blue Cut fire

On 16 August 2016 a fire in Southern California drove faults on a transmission corridor of three 500 kV lines (SCE) and two 287 kV lines (LADWP): thirteen 500 kV faults and two 287 kV faults over the day. Four caused significant PV loss; the largest was nearly 1200 MW (a line-to-line fault cleared in 2.49 cycles). No PV was de-energised by the fault directly — the plants ceased or tripped in response to it — and the Western Interconnection frequency dipped only to 59.867 Hz. Taken step by step:

1. Transmission faults caused voltage disturbances along the corridor.
2. Many PV plants did not ride through as expected.
3. Some tripped on false low-frequency behaviour: the fault-distorted voltage waveform made the inverter PLL read a frequency below 57 Hz and trip instantaneously (per the PRC-024-2 curve), even though the real frequency never came near 57 Hz — about 700 MW.
4. Some entered momentary cessation, stopping current when the voltage left the continuous operating range and resuming after a delay and a ramp (about two minutes back to full output here) — about 450 MW.
5. Some tripped on dc overcurrent after momentary cessation began (cause left for the manufacturers to investigate) — about 100 MW.
6. The combined loss of PV worsened the system disturbance.

Tripping is the more damaging path, removing the resource for about five minutes; a quickly restored momentary cessation produces a milder dip. The NERC/WECC task force recommended adding a short time delay to inverter frequency tripping so the inverter rides through the distorted-waveform period; restoring momentary-cessation output within no more than five seconds; and revising PRC-024-2 so that “outside the curve” is a may-trip region (to protect equipment), not a must-trip one. PRC-024-2 is the NERC reliability standard on generator frequency and voltage protective-relay settings; it is relevant here because those settings can decide whether generation stays connected during a disturbance.

The engineering lesson is that the ride-through curve alone is not enough: PLL behaviour, protection logic, momentary cessation and current limits must all be represented. Equally, Blue Cut is an example of how protection and controls can dominate the IBG fault response — it should not be used as a universal quantitative rule for all PV plants, because modern requirements and vendor controls may differ.

Section 11

Phenomena → studies, and RMS versus EMT

The three phenomena map onto four kinds of study, each with its own indices:

Short-term voltage stability — whether the voltage recovers within a few seconds. Indices: the Critical Area Exchange (the largest area export that stays stable for a given fault) and the Loading Level (the largest load before collapse); the induction-motor fraction is a key input.
Transient (rotor-angle) stability — whether machines stay in step after a severe fault. Indices: the critical clearing time (CCT), the critical area exchange and the maximum power transfer. Operators lean on the exchange and transfer indices because relay clearing times are essentially fixed (the shortest EHV clearing is around 3 cycles).
Short-circuit fault-current provision — a grid-interconnection check that the breaker duty is met. The IBG fault level is often taken, as a conservative planning assumption, as up to about 150% (roughly 1.1–1.5 pu) of rated — higher than the ~1 pu RMS a measured plant delivers, because the check is a bounding one — and both the peak and the breaking current (at about 70–100 ms) are examined; a simulation-based approach grows in importance as penetration rises.
LVRT studies — in two flavours: to obtain better ride-through capability, and to assess the impact of having (or not having) LVRT on rotor-angle, voltage and frequency stability.

These overlap. A slow recovery from losing IBGs without ride-through, or from stalling motors, shortens the time to an OOS; and a lower short-circuit level can lengthen the fault duration, making the system more vulnerable to voltage instability. So the phenomena are best kept distinct by their end behaviour — OOS or voltage collapse — and their remedy — generator tripping or load shedding.

The choice of model type then follows the phenomenon. RMS may be acceptable for bulk transient-stability screening if the IBG model has validated LVRT/HVRT, current limits, active/reactive priority and protection behaviour. EMT is needed when the result depends on waveform detail, unbalanced faults, very low voltage, weak-grid PLL behaviour, vendor-specific protection, momentary cessation, harmonics or detailed converter controls. For the underlying comparison, see the EMT versus RMS guide.

RMS or EMT for a large-voltage-deviation study?

Use RMS when — bulk-system transient stability is the objective; the voltage does not collapse to near zero at the plant model; the model has validated current limits and FRT logic; and the protection is phasor-based and represented by settings and delays.
Use EMT when — severe or unbalanced faults are studied; the voltage is very low at the IBG terminals; the PLL or current-control behaviour decides the result; momentary cessation or converter blocking is important; vendor-specific LVRT/HVRT compliance is being tested; the protection depends on waveform or sub-cycle behaviour; or weak-grid interaction is possible.

Section 12

Which functions to model

Read Table 4 as a large-voltage-deviation modelling guide: a “Yes” means the function can change the fault current, the ride-through, the current injection, the voltage recovery, the protection operation or the post-fault stability. Compared with a frequency study, far more is needed here — the fault exercises the current limit, the PLL, the ride-through envelope and most of the protections at once.

Table 4 — Recommended IBG functionalities for a large-voltage-deviation study (after CIGRE Table 3.7).
Category	Functionality	Model?	Note
Control	DC source control	Yes, if the dc link is modelled	dc voltage rises during momentary cessation
Control	Current control (inner loop)	Yes, if the dc link is modelled; always for EMT	controls active / reactive current injection during and after the dip
Control	PLL	Yes	determines angle tracking during the fault; angle jumps at inception and clearing
Control	MPPT	No	sub-second fault; irrelevant unless active-power recovery or dc-side behaviour affects the result
Protection	Reduce max inverter current on dc overvoltage	Yes, if the dc link is modelled	can trip or block the converter
Protection	Limit inverter current rate of change after a fault	Yes	di/dt device protection; can disconnect the IBG
Protection	Current limit	Yes	caps total converter current; active/reactive priority and pre-fault current shape the response
Protection	DC overvoltage protection	Yes, if the dc link is modelled	can trip or block the converter
Protection	Over / under-voltage protection	Yes	may trip or block the unit on the voltage excursion
Protection	Over / under-frequency protection	Yes	a distorted waveform can cause a false frequency trip (see Blue Cut)
Protection	Detecting a balanced fault	N/A	positive-sequence overvoltage protection may be used
Protection	Detecting an unbalanced short-circuit fault	N/A	negative-sequence overvoltage protection may be used
Protection	Detecting a single-line-to-ground fault	N/A	zero-sequence overvoltage protection may be used
Protection	ROCOF tripping	Yes	an angle jump can look like a large ROCOF
Protection	Vector jump	Yes	a fault-driven phase jump can trip it
Protection	Transfer trip	No
Protection	Anti-islanding active detection	Yes	not required by all utilities
Capability	P(f) control (over / under frequency)	No	a frequency function; not decisive for the voltage-driven event
Capability	Voltage control by reactive power	Yes	pre-fault Q affects the fault-on / post-fault response
Capability	Voltage control by active power, P(V)	Yes	slow (tens of seconds to a minute); affects the pre-fault operating point
Capability	Synthetic inertia	Yes	affects damping and the power swing
Capability	ROCOF immunity	No	a frequency-ride-through setting; not the driver here
Capability	Fault ride-through (LV / HV)	Yes	the central capability; determines the ride-through envelope, modelled as under/over-voltage settings
Capability	Active behaviour during fast voltage variations	Yes	dynamic reactive-current support; interacts with the current limit
Capability	Power oscillation damping (POD)	Yes	mitigates the post-fault swing

The “Yes” entries are, in plain terms, the fault-response chain: the current control sets what is injected; the PLL sets the angle it is injected at; the current limit caps the total; the over/under-voltage and dc protections decide whether the unit trips or blocks; LVRT/HVRT sets the ride-through envelope; the dynamic reactive current supports the voltage; and momentary cessation removes the current temporarily. The main “No” entries are the slow or frequency-side functions: MPPT does not matter for a sub-second fault unless active-power recovery or dc-side behaviour affects the result; long-term irradiance variation is irrelevant to the first few seconds; and plant-level slow voltage control is less important than the fast current injection during the fault, except where its references set the pre-fault operating point. Two points bind the table together. First, the dc link: during momentary cessation the ac current is forced to zero while the dc source keeps feeding, so the dc voltage rises and can trip the dc-overvoltage protection — without that path modelled, the simulated active and reactive power differ from the measured response. Second, the ride-through envelope: LVRT/HVRT is best represented as under- and over-voltage settings with the right thresholds and delays, working hand in hand with the current limit and the active/reactive priority. The practical message: a large-voltage-deviation study needs the current limit, the PLL behaviour, LVRT/HVRT, the dynamic reactive current, momentary cessation and the trip logic — and does not need unnecessary detail unless it affects those behaviours.

Section 13

A study workflow, outputs and recovery criteria

Putting it together into a repeatable procedure:

A large-voltage-deviation study workflow

1. Define the fault type, location, duration and clearing sequence.
2. Initialise the pre-fault power flow and the IBG operating mode.
3. Define the synchronous-machine, load and motor models.
4. Include the IBG current limit, PLL, FRT envelope and protection.
5. Define the active/reactive current priority.
6. Simulate the fault and its clearing.
7. Check the voltage recovery, fault current, motor stalling, OOS risk and IBG trip / cessation.
8. Repeat for credible fault locations and clearing times.
9. Compare the RMS results with an EMT / vendor model for the critical cases.
10. Document the assumptions, settings and model-validity limits.

The signals worth plotting are the ones that reveal the fault response: the PCC/POI voltage magnitude (and the individual phase voltages in EMT); the converter current magnitude and its active and reactive components; the active and reactive power; the PLL angle or frequency and the dc-link voltage where available; the current-limit status; the LVRT/HVRT status and the momentary-cessation flag; the trip and blocking flags; the motor speed or motor reactive current; the generator rotor angles for transient stability; the protection-relay operating time; and the post-fault voltage-recovery time.

Judging voltage recovery

Voltage recovery should not be judged only by whether the voltage eventually returns. Check how low it goes, how long it stays low, whether motors stall, whether IBGs trip or cease, whether protection operates, and whether the recovery meets the grid-code or planning criteria. A voltage that “recovers” after several seconds of a deep depression — having stalled the motors and tripped the IBGs on the way — is not a passing result.

Common mistakes

The traps that most often catch a large-voltage-deviation study — each a theme from the sections above:

Ten traps to avoid

Treating IBG fault current like synchronous-machine fault current.
Ignoring the current limits during LVRT/HVRT.
Modelling the ride-through envelope but not momentary cessation.
Assuming that staying connected always improves transient stability.
Ignoring PLL behaviour during deep voltage dips.
Using RMS for very low-voltage converter behaviour without validation.
Ignoring motor stalling and load recovery.
Assuming PV or full-converter plants have classical rotor-angle out-of-step behaviour.
Ignoring the protection settings and delays.
Using the Blue Cut event as a universal numerical rule instead of a modelling lesson.

Key points

Under a fault, an IBG answers with control — and can just leave

Large voltage deviations are normally fault-driven events.
Synchronous machines respond with natural fault current and rotor-angle dynamics; IBGs respond through controls, current limits and protection.
Full-converter IBGs have no classical out-of-step, but they still affect transient stability indirectly — through inertia, reactive reserves, fault current and voltage support.
IBGs reduce the fault current and can change protection performance (magnitude, angle, duration, dc offset).
Short-term voltage recovery depends on reactive support, motor recovery and whether the IBGs stay connected.
LVRT/HVRT must be modelled with the current limits and the priority logic, not as a curve only.
Momentary cessation can remove current without a full trip and can strongly affect recovery.
PLL behaviour matters during deep faults and weak-grid conditions.
RMS can be used for validated bulk screening, but EMT / vendor models are needed for detailed severe-fault converter behaviour.

For the underlying differences and the inverter itself, see the characteristics of inverter-based generation and inverter-characteristics guides; for the ride-through and sequence behaviour, the plant-controller FRT and converter-protection 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; the NERC/WECC disturbance report documents the 2016 Blue Cut fire event; NERC PRC-024-2 is the frequency- and voltage-relay standard behind the tripping behaviour; and IEC 60909 is the reference for short-circuit-current definitions.

CIGRE/CIRED Joint Working Group, Modelling of Inverter-Based Generation for Power System Dynamic Studies. CIGRE Technical Brochure.
NERC and WECC, 1,200 MW Fault-Induced Solar Photovoltaic Resource Interruption Disturbance Report (Blue Cut Fire, 16 August 2016). North American Electric Reliability Corporation.
NERC Reliability Standard PRC-024-2 — Generator Frequency and Voltage Protective Relay Settings. North American Electric Reliability Corporation.
IEC 60909, Short-Circuit Currents in Three-Phase A.C. Systems. International Electrotechnical Commission.

Part 9 Reading now

Modelling IBGs for Large Voltage Deviations

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

Series progress 9 of 12

Part 10

Modelling IBGs for Long-Term Voltage Stability

The slower voltage-collapse mechanism over seconds to minutes: reactive balance, tap changers and P–V/Q–V limits.