Wide-Area EMT Studies · Model Adequacy

Model Adequacy for Wide-Area EMT Studies

The nine-part IBG modelling series ended by choosing between RMS, EMT and impedance-based analysis. This guide — the first of three on wide-area EMT practice — asks the next question: once EMT is chosen, is the model itself adequate for the study? A wide-area EMT case can hold hundreds of models — inverters, HVDC, lines, cables, transformers, protection and loads — and a single inadequate one, in the wrong place, turns an expensive simulation into a confident wrong answer. It covers vendor-specific, site-specific and generic models, aggregation, the network, protection, DER and model visibility. It follows CIGRE Technical Brochure 881, with APS engineering interpretation, for EMTP® and other EMT tools.

Reading time ≈ 25 min · Vendor vs generic, aggregation, network, protection & visibility

The nine-part Modelling Inverter-Based Generation series worked from what makes an inverter-based generator (IBG) different, through the inverter and the RMS / phasor and EMT modelling methods, to their application in frequency, voltage, small-signal, islanding and control-interaction studies. Its final guide — control interactions, SSCI and model selection — explained when RMS is sufficient and when EMT, impedance-based analysis or vendor-model validation becomes necessary. This page goes one step further: if EMT is selected, the model itself must still be adequate for the question being asked. A wide-area EMT case can hold hundreds of models — inverters, HVDC, lines, cables, transformers, protection and loads — and a single inadequate one, in the wrong place, turns an expensive simulation into a confident wrong answer.

Three connected guides on wide-area EMT practice

This is the first of three further guides that follow the series. This page asks whether a model is adequate: vendor-specific, site-specific, generic, aggregated, and the network, protection, DER and visibility around it. The next guide asks how a model is accepted and validated: single-machine infinite-bus (SMIB) tests, weak-grid checks, field validation, disturbance playback and hardware-in-the-loop. The following guide asks how large-scale cases are built and run: load-flow initialisation, network equivalencing, co-simulation, hybrid PDT–EMT and real-time simulation.

By the end of this page, the reader should understand

what model adequacy means, and why a model that runs is not necessarily correct;
when vendor-specific and site-specific EMT models are essential, and when generic models are acceptable;
why generic models are riskier in the local study area, and why site settings make the same OEM product behave differently;
how aggregation affects wind, PV and battery-plant representation;
which network details matter in wide-area EMT, and why protection models can decide the result;
why DER and load modelling matter at high IBG penetration;
and why black-boxing, model visibility and source-code escrow matter.

What model adequacy means

Model adequacy means the model contains enough physical, control, protection and network detail to answer the specific study question with credible accuracy. An adequate model is not always the most detailed one: it is the simplest model that still preserves the behaviour that can change the study conclusion. Add detail that cannot change the answer and you pay in run time and fragility for nothing; drop detail that can, and the result is quietly wrong.

Adequacy depends on the question

There is no single “adequate” model — only a model adequate for a stated study. The required detail depends on:

the study phenomenon (fault ride-through, SSCI, harmonics, weak-grid control, frequency, voltage recovery, protection);
the local system strength, and the distance from the contingency;
the converter technology, and the control / protection functions that may operate;
the required output; and
the purpose — planning screening, compliance, connection approval or incident investigation.

A note on IBG and IBR

The APS series uses IBG (inverter-based generation). CIGRE’s brochures use IBR (inverter-based resource). On this page, IBR appears where the source material uses it; it is closely aligned with IBG, but slightly broader, because it can include generation, storage and other converter-interfaced resources such as HVDC and FACTS. Read the two as the same idea for the purposes of adequacy.

Abbreviations used on this page

EMTElectromagnetic transient (time-domain) simulation

PDTPhasor-domain transient (the RMS / phasor method)

RMSRoot-mean-square (phasor) simulation

EMTP®Electromagnetic Transients Program (an EMT tool)

IBG / IBRInverter-based generation / resource

OEMOriginal equipment manufacturer (the vendor)

IPIntellectual property

SCRShort-circuit ratio (system strength)

X/RReactance-to-resistance ratio

PoC / PCCPoint of connection / common coupling

DFIGDoubly-fed induction generator (Type-3 wind)

WTGWind-turbine generator

PVPhotovoltaic

BESSBattery energy storage system

HVDCHigh-voltage direct current

FACTSFlexible AC Transmission System device

SVC / STATCOMStatic var compensator / static synchronous compensator

LCC / VSCLine-commutated / voltage-source converter

PLLPhase-locked loop

PWM / IGBTPulse-width modulation / insulated-gate bipolar transistor

LVRT / HVRTLow- / high-voltage ride-through

FRTFault ride-through

ROCOFRate of change of frequency

SSCISub-synchronous control interaction

DERDistributed energy resource

CMLDThe WECC composite dynamic load model

Key idea

Choosing EMT only pays off if the models are adequate. A vendor-specific, site-specific model — the OEM’s control code, tuned to the actual site — is essential wherever an IBG’s specific performance or its interaction with neighbours is the question, above all at low system strength and near the contingency.
Generic models have a place — planning, screening, distant plant, legacy plant with no data — but used outside it they throw away the EMT advantage: they can fail to initialise, look falsely stable, or hide an instability while appearing detailed.
Check adequacy under the series’ three headings — Control, Protection and Capability — and around the inverter: sensible aggregation, the right network / cable / transformer detail, the protection that actually decides the outcome, and dynamic load / DER behaviour.
Detailed vendor models come black-boxed to protect the OEM’s IP — so the study engineer needs enough output visibility and validation evidence to trust and diagnose the result, and enough governance (source-code escrow) to keep it usable for decades.

Check adequacy under three headings

The series read every plant through three repeated categories; adequacy is best checked the same way:

Control — PLL, current control, plant controller, voltage / reactive control, active-power recovery, damping functions.
Protection — LVRT / HVRT, over/under-voltage, over/under-frequency, ROCOF, momentary cessation, converter blocking, plant and network protection.
Capability — current limit, reactive range, active-power headroom, dc-link behaviour, overload capability, the ride-through envelope.

If a function under any heading can change the study conclusion, the model must represent it — whatever the plant’s label.

Key terms used on this page

01Model adequacy

Whether a model has enough physical, control, protection and network detail to answer the specific study question credibly.

02Vendor-specific model

A model that represents the actual OEM product and its control / protection philosophy, often embedding the real control code.

03Site-specific model

A model parameterised for the actual project — grid-code settings, controller tuning, layout, cable / transformer data, protection, plant controller.

04Generic model

A less detailed model of a control strategy or plant type, not tailored to a product or site; useful within, but not beyond, its intended purpose.

05System strength (SCR)

How stiffly the grid holds voltage at a node; a low short-circuit ratio (weak grid) makes voltage and angle very sensitive to injected current.

06Aggregation

Representing many similar units as fewer scaled equivalents to cut computation, without losing the behaviour relevant to the study.

07Collector system

The medium-voltage network gathering the units to the plant substation; represented by an equivalent that preserves losses, impedance and (where needed) resonance.

08Node-breaker vs bus-branch

Whether the model keeps the full substation switching topology (breakers) or collapses each substation to buses and branches.

09Bergeron model

A travelling-wave line model, usually adequate for stability-focused wide-area EMT up to the inverter control bandwidth (~200–300 Hz).

10Average-switch model

An average-value converter model that drops explicit switching but keeps the controls and dc-link dynamics — far faster, adequate for stability.

11Black-boxing

Compiling or encrypting the sensitive parts of a model so it runs correctly but hides the manufacturer’s intellectual property.

12Source-code escrow

Holding a model’s source code with a trusted third party, to be released only if the OEM can no longer support it.

13CMLD

The WECC composite load model: an aggregate of static, electronic and motor load plus distributed generation at a feeder.

14Prime mover

The energy-source side of an IBG (turbine aerodynamics, pitch, drive train); often simplified for wide-area stability studies.

15Local study area

The part of the network near the contingency, where model fidelity matters most.

16DER

Distributed energy resource: small generation, storage and controllable load on the distribution network, aggregated for transmission studies.

Section 1

Why model adequacy decides the study

Reaching for an EMT tool is a decision to spend time and computing effort for accuracy the phasor-domain transient (PDT) — the RMS / phasor — method cannot deliver. Follow the logic one step at a time:

The RMS / PDT method strips away the fast waveform and converter-control detail, keeping the fundamental-frequency behaviour.
EMT can represent that detail — the switching, the inner control loops, the protection, the current limits.
But EMT only helps if the model actually includes the relevant controls, protection, limits and site settings.
A generic or incomplete EMT model can be worse than a well-understood RMS model, because it looks detailed while hiding the behaviour that matters.

That is the whole case for this page. Put a model into a wide-area EMT case that is less capable than the phenomenon demands, and you get the cost of EMT with the blind spots of PDT. The damage is worst when the inadequate model sits in the local study area, near where the contingency is applied — a generic model far from the disturbance does less harm than the same model at the heart of the event.

Two ideas run through everything below. First, match the model to the question: the required detail depends on the phenomenon, the location and, above all, the system strength. Second, a model that runs is not a model that is right: adequacy is judged against grid-code requirements, vendor documentation, commissioning and disturbance records, or a benchmark case — not against whether the simulation completed. The sections below work outward from the inverter.

Section 2

Vendor-specific and site-specific models

For a wide-area EMT study whose purpose is the real performance of an inverter-based generator (IBG) — or a cluster of them — the detailed, manufacturer-supplied model is usually the only adequate choice. Two words carry the requirement, and they are not the same thing:

Vendor-specific means the model represents the actual OEM product and its control and protection philosophy — ideally the real control code — because the converter control is the most complex and most proprietary part of the machine (with the prime-mover controls proprietary too), differing by manufacturer, product vintage and regional configuration.
Site-specific means the model is parameterised for the actual project: the grid-code settings, the controller tuning, the plant layout, the transformer and cable data, the protection settings, the plant controller and the local network conditions.

The point that catches people out: a vendor-specific model without site-specific settings is not enough. It may represent the right product but the wrong plant. When a vendor model is validated against a laboratory or field test it can match the real equipment closely — that match is the whole point — but only for the site it was tuned to.

Why the same product behaves differently at different sites

The same turbine, inverter or battery model can respond differently from one site to the next because:

the grid-code settings and the fault-ride-through requirements differ;
the PLL and current-controller tuning may have been changed;
the system strength at the PoC / PCC is different;
nearby converters, cables, filters, series capacitors or STATCOMs change the interaction risk;
special controllers may be enabled for weak or series-compensated grids;
and the plant-level controller settings differ.

A vendor model run with the wrong site parameters can be unreliably optimistic or pessimistic — and either way it drives a wrong decision.

The local-study-area rule

Generic models are lowest risk far from the event and highest risk in the local study area. If the contingency is applied near the plant, the local IBG model should normally be vendor-specific and site-specific. Without those models, long-cable and offshore projects can show instabilities a generic model never reveals, adverse interactions between neighbouring IBGs go unseen, real design constraints (an allowance for dc-link voltage fluctuation, say) are simply absent, and a plant’s technical performance requirement can be set too high or too low. The one caution is cost: vendor models carry a heavier computational burden, so the number of vendor-model runs has to be planned.

Section 3

When generic EMT models are acceptable

A generic model represents a control strategy and a set of control loops, not a particular product or site. Because the actual control logic is absent, its response can diverge sharply from a vendor, site-specific model — and the divergence is worst exactly where it matters. At low system strength a generic model may fail to initialise, or swing the other way and return a response that is too stable, hiding a real instability. There is also a maturity gap: OEMs routinely develop generic PDT models but generally not generic EMT models, and at the time of writing there is no internationally agreed generic EMT model. Table 1 draws the line.

Table 1 — Where a generic EMT model is, and is not, normally acceptable.
Generic EMT models may be acceptable for…	…but are normally not acceptable for
Early long-term planning	Connection compliance
Initial screening studies	Weak-grid operation
Future plants with no OEM / site data yet	SSCI / control-interaction studies
Legacy plant with no model, if tuned to recordings	LVRT / HVRT performance assessment
Distant plants outside the local study area	Project-specific protection behaviour
Academic or sensitivity studies	Multi-vendor plant interaction
	Incident investigation
	Local plant near the disturbance
	Technical performance guarantees

Generic is not bad — but it must be bounded

Generic models are genuinely useful when the question is broad and the exact OEM and site details are unknown — a long-term planning scenario whose future generation has no fixed location, type or make. Their limitation is not that they are generic. The limitation is using them outside their validated range, or using them to make project-specific conclusions. Keep a generic model inside its purpose and it is the right tool; ask it to certify a specific plant on a weak grid and it is the wrong one.

Section 4

Legacy plant with no model: a workflow

An existing plant with incomplete documentation is the classic case for a carefully bounded generic model. Work it in order:

Building a model for a legacy plant

1. Request the OEM’s vendor-specific EMT model of the plant.
2. If the exact model is unavailable, request one from the same OEM / product family, re-parameterised to the site.
3. Only if no better option exists, use a generic model of the correct technology (for example Type-3 or Type-4 wind).
4. Collect disturbance recordings across a wide range of operating conditions.
5. Collect ratings, transformer data, cable / collector data and protection settings.
6. Tune the model across multiple operating points, not one.
7. Check the LVRT / HVRT envelope, the active- and reactive-power recovery, the fault-current limit and the plant controller.
8. Document the uncertainty, and avoid overclaiming what the model can support.

Do not overfit one recording

Do not tune a generic model to match one event perfectly if it becomes wrong elsewhere. A generic model has few adjustable parameters, so “perfecting” a single operating point often degrades another. A credible model behaves reasonably over a range of operating points and disturbances — a good fit everywhere beats a perfect fit at one point.

Section 5

What a detailed IBG model contains

A detailed EMT model of an inverter-based generator can include the aerodynamics, the pitch control, the mechanical drive train, the electrical generator, the measurements, the converter control and protection, and the semiconductor switching scheme (the PWM of the IGBTs). A wind-turbine generator (WTG) model may carry all of these; a solar-inverter model needs only the electrical source, the measurements and the converter control. Which of these must actually be represented — and to what depth — is the model-adequacy question in miniature, and it is decided by the study type. Table 2 gives the shape of it for a wind-turbine generator.

Table 2 — Which wind-turbine sub-systems a model must represent, by study type, in PDT and EMT tools (after CIGRE TB 881, Table 2-1). “PDT” = the model is capable in the phasor domain; “EMT” = EMT representation is required; “–” = generally not required; “*” = case-by-case, confirm with the OEM.
Component	Transient stability	Sub-synchronous interaction	High-frequency transient	Harmonics
Aerodynamics	PDT, EMT	EMT*	–	EMT*
Pitch controller	PDT, EMT	EMT*	–	EMT*
Mechanical drive train	PDT, EMT	EMT	–	EMT*
Torsional damping	PDT, EMT	EMT	–	EMT*
Electrical generator	PDT, EMT	EMT	EMT	EMT
Dynamic braking chopper	PDT, EMT	EMT	–	EMT
DC link	PDT, EMT	EMT	EMT*	EMT
IGBT switches & PWM	–	–	EMT	EMT
Inner-loop converter control	EMT	EMT	EMT*	EMT
Outer-loop converter control	PDT, EMT	EMT	EMT*	EMT
Phase-locked loop	EMT	EMT	EMT*	EMT
Low- / high-voltage ride-through	PDT, EMT	EMT	EMT / –	–
Protection	PDT, EMT	EMT	EMT	–

Two rows carry the headline. The inner-loop converter control and the PLL need an EMT representation even for transient-stability studies, because their fast behaviour is what a phasor model cannot see; the slower aerodynamic and pitch dynamics are often marked case-by-case, and the switching itself only matters for high-frequency and harmonic work. Two further points guide the depth:

The prime mover can often be simplified. For wide-area EMT stability studies the aerodynamic, pitch and mechanical drive-train dynamics can often be simplified or neglected if they do not affect the time window or phenomenon being studied — a design-dependent judgement, best made with the OEM and confirmed against measured response. But for wind-recovery, torsional-interaction, frequency-response or drive-train-damping studies, those dynamics may matter and must be kept.
Switching is usually averaged. Representing every IGBT switching event is expensive, and prohibitive when tens or hundreds of inverters are in one case, so most vendors use an average-switch (average-value) model that removes the explicit semiconductor switching but keeps the converter controls, the dc-link behaviour and the dynamic response needed for stability studies — comparable accuracy for a network fault, roughly an order of magnitude faster. Explicit switching is still required where switching harmonics, semiconductor stress, detailed PWM interaction or filter resonance is the study target.

For any wide-area stability study, the one thing that must be right is the internal control system. For a Type-3 turbine that means the rotor-side converter control (its inner current loop, the active/reactive-power controllers, the power filters that keep the reference from jumping, and the drive-train torsional damping that injects power to damp the shaft resonance), the grid-side converter control (the PLL, the dc-link voltage and current control) and the ride-through logic — together with the turbine protection, including the dc choppers. That is the part of the model that dominates the dynamic performance, and the part a generic model most often gets wrong.

Section 6

Model aggregation

Aggregation means replacing many similar inverters or turbines with one or a smaller number of equivalent models. It cuts computation time — an inverter-based plant can be hundreds of small, identical units, and modelling each multiplies the run time for almost no benefit — but it must preserve the behaviour relevant to the study. A single-technology plant of a hundred identical turbines can collapse to one aggregate unit on one equivalent feeder; a plant mixing two technologies needs one aggregate per technology.

What aggregation must not combine

Aggregation should not combine units with materially different behaviour:

different technologies (full-converter versus doubly-fed, i.e. DFIG, PV versus wind versus BESS);
different OEMs;
different control modes;
different protection settings;
different feeder electrical characteristics; or
different plant sections with materially different voltage exposure.

The rule in one line: aggregate identical units with a similar electrical location and settings; do not aggregate dissimilar units just because they share a plant.

\[ S_{agg} = N\,S_{unit}, \qquad i_{src} = (N-1)\,i_{unit} \]

\(N\): number of identical units represented by the aggregate
\(S_{agg},\ S_{unit}\): the aggregate and single-unit apparent-power ratings
\(i_{unit}\): the current of one representative modelled unit
\(i_{src}\): the current injected by a parallel current source standing in for the other \(N-1\) units

One representative unit is modelled in detail; its rating and its unit transformer are scaled up by \(N\), and a current source carrying \((N-1)\) copies of its current feeds the main transformer. Driving the “other” units through a separate transformer can be more numerically stable than injecting everything into one, though it needs care if other plant shares the medium-voltage busbar.

The collector-system equivalent deserves special attention: it should preserve the active and reactive losses, the impedance, the voltage drop and the resonance behaviour relevant to the study. In the simplest case one equivalent series impedance connects the medium-voltage collector bus to the aggregate unit transformer, chosen so it dissipates the same active and reactive power as the real collector network; more elaborate methods account for the voltage spread across the feeders. But for a harmonic or resonance study a simple equivalent collector impedance may not be enough — the frequency-dependent behaviour of the real feeders can matter. A complete aggregate model carries the medium-voltage network equivalent, the explicit large substation transformer(s), the turbine and plant-controller control systems, the scaled unit transformers, any harmonic filter and shunt reactive devices, the connection to the PCC, and the associated turbine- and plant-level protection.

Section 7

The network around the inverters

An EMT study usually represents only part of the network in detail and reduces the rest to an equivalent (the subject of the large-scale-EMT guide). Within the detailed part, four choices decide adequacy.

Node-breaker versus bus-branch

Node-breaker modelling keeps the actual substation switching topology — every circuit breaker and isolator. Bus-branch modelling simplifies the substation into buses and branches. Bus-branch is faster and often enough, and can even speed EMT up. Node-breaker is the safer choice when the switching actions, breaker states, intertrips or “impossible operations” matter — because bus-branch discards topology, a user can perform a switching that the real substation cannot allow (taking a line out while keeping its reactor in service) and silently invalidate the result. The governing rule: the model should not allow an operation that cannot happen physically.

The passive elements need matching judgement:

Lines and cables. Bergeron models are normally adequate for many stability-focused wide-area EMT studies (up to the inverter control bandwidth, ~200–300 Hz). Frequency-dependent or geometrical models may be needed for harmonics, resonance, cable-dominated networks, long cables, offshore wind, temporary over-voltages and the interaction between stability and power quality. Cables need special attention, because their frequency dependence differs from overhead lines — and it is the long-cable, offshore cases where inadequate models most often mislead.
Transformer saturation. It looks like equipment detail, but it can shape an IBG’s response after a fault or an energisation. Residual core flux and the voltage recovery on fault clearance can drive an inrush, a voltage dip and a false ride-through or protection operation. Model saturation when the transformer is large relative to the local fault level or the nearby converter capacity, or when energisation and inrush affect the study.

HVDC and FACTS are not background detail

HVDC links, STATCOMs and SVCs are not scenery in a high-IBG EMT study. They are controlled power-electronic devices that can interact with IBG controls, so their vendor-specific, site-specific models can be just as important as the wind, PV or BESS models. Line-commutated (LCC) HVDC additionally needs its large harmonic and reactive filters and their controls represented, is theoretically more susceptible to low system strength, and brings commutation-related behaviour that can decide a result — precisely where accurate modelling matters most.

Section 8

Protection decides the result

Protection is not an optional add-on. In many EMT studies the final result is decided by whether a protection function trips, blocks, rides through, recloses or initiates a runback — and EMT is well suited to it, because the voltage and current waveforms the relays act on are already in the solution and the time step is short enough to resolve them. Adequacy means representing the protection at every level that can change the outcome:

Component-level — converter overcurrent, dc-link over-voltage, LVRT / HVRT, momentary cessation, converter blocking.
Plant-level — voltage / frequency protection, ROCOF and vector-jump, anti-islanding, plant-controller logic.
Network — distance and differential protection, and the fault-detection schemes on lines and transformers.
Synchronous-machine — pole-slip, under-excitation (loss-of-field) and volts-per-hertz protection, which matter more as fewer machines carry the system.
System-level and emergency schemes — under-frequency and under-voltage load shedding (UFLS / UVLS), special-protection schemes, generation runback.

The inclusion test mirrors the series: if a protection function under Control, Protection or Capability can trip, block or ride through in a way that changes the study conclusion, it belongs in the model — and system-level schemes belong whenever their operation would push the total generation or load change past the largest single credible contingency, regardless of ownership boundaries.

Section 9

Distributed resources and load

At high DER penetration, load is no longer passive. Distributed PV, batteries, electronic loads and motor loads can trip, ride through, stall, recover, reconnect or interact with voltage and frequency protection — the same behaviours the frequency, voltage-stability and islanding guides traced for transmission-connected plant, now on the distribution side. A static polynomial load is not adequate where these matter.

CMLD and large motor drives

CMLD is the WECC composite load model. It represents a distribution feeder as a mixture of static load, electronic load, motor load and distributed generation. It is useful because voltage recovery and frequency response often depend on the combined behaviour of load and DER, not only on transmission-connected generation. Large converter-fed motor drives should be treated like other converter-interfaced devices when their rating is large enough to influence voltage, system strength or nearby converter interaction: a small drive can sit inside an aggregate load model, but a large one, or one close to other converters, may need a vendor model of its own.

Section 10

Visibility, IP and black-boxing

The best vendor models embed the real control — and sometimes hardware — representation, which is exactly what makes them commercially sensitive. A model is software: easy to copy, exposed to cyber risk, and capable of leaking intellectual property (IP) built over decades. The common answer is black-boxing: the sensitive parts are compiled or encrypted into a binary the EMT tool calls at run time, so the model runs correctly but its internals are hidden. A black-box model can be technically necessary and commercially reasonable — but the study engineer still needs enough visibility to trust and diagnose the result.

Output channels a black-box model should still expose

PCC voltage and current; active and reactive current; active and reactive power;
PLL angle and frequency; dc-link voltage;
current-limit status; LVRT / HVRT status; protection flags;
momentary-cessation, converter-block and trip status;
plant-controller references and control-mode status.

In a low-strength, inter-plant-oscillation case, these internal references and the PLL output are exactly what the engineer needs to find the source of an instability.

Trust but verify. The user does not need the OEM’s source code to be public, but they do need enough documentation, output channels and validation evidence to confirm the model is behaving credibly. And for the long term, source-code escrow is a risk-management measure: a trusted third party holds the model source code so the model can be maintained if the OEM disappears, stops supporting the region, or the original compiler and software environment become obsolete — a real risk over a plant life that can exceed thirty years.

Section 11

A model-adequacy decision, in practice

Two practical aids close the page: a decision workflow to pick the right model for each device, and a checklist to confirm a supplied model is usable before you trust it.

Model-adequacy decision workflow

1. Define the study question.
2. Identify the local study area.
3. Identify the phenomena: FRT, SSCI, harmonics, weak-grid control, frequency, voltage recovery, protection.
4. Identify which Control, Protection and Capability functions can change the answer.
5. Classify each IBG / HVDC / FACTS / DER / load model as vendor-specific, site-specific, generic or equivalent.
6. Check that generic models are used only where they are acceptable (Table 1).
7. Check the aggregation rules.
8. Check the network, transformer, cable and protection detail.
9. Check the visibility and output channels.
10. Check the validation evidence and the known limitations.

Before trusting a supplied model, confirm…

the vendor and model version; the site-specific parameter set;
a valid software version and compiler requirements; the supported time step; the required initialisation method;
the valid SCR / X/R range; the supported control modes; the plant controller included;
LVRT / HVRT included; the current limiter included; protection flags available; dc-link behaviour where relevant;
the aggregation basis documented; the collector network represented correctly; the network-support devices included;
validation evidence available; and the known limitations documented.

Common mistakes

The adequacy traps that most often undermine a wide-area EMT study:

Ten traps to avoid

Assuming a vendor-specific model is correct without site-specific settings.
Using a generic model in the local study area for a project-specific conclusion.
Treating a model that runs as an adequate model.
Aggregating different technologies or control settings into one equivalent.
Leaving out the plant controller or the protection logic.
Ignoring the collector-system and cable frequency dependence.
Using bus-branch topology where switching / intertrip logic needs node-breaker detail.
Ignoring transformer saturation or inrush where it can trigger ride-through or protection.
Trusting a black-box model without output channels and validation evidence.
Overfitting a generic legacy model to a single disturbance recording.

Key points

The model is the study

An EMT study is only as good as its models; an inadequate model near the contingency can give a confident wrong answer.
Vendor-specific is not enough without site-specific settings — the right product can still be the wrong plant.
Generic models fit planning, screening, distant and legacy plant, but not compliance, weak grids, SSCI, protection behaviour or local plant near the disturbance.
Local-study-area models need higher fidelity; represent Control, Protection and Capability functions that can change the answer.
Aggregate only similar units, with a proper collector equivalent; do not merge dissimilar plant.
Match the network, cable, transformer-saturation and HVDC / FACTS detail to the question, and keep topology physically valid.
Protection and DER / load behaviour can decide the result — model them where they matter.
A black-box model still needs output channels, validation evidence and escrow governance.

The main takeaway: EMT accuracy is not created by the solver alone. It is created by the right model detail in the right location — vendor-specific and site-specific models near the disturbance, appropriate generic models only where the question allows, realistic aggregation, adequate network and protection representation, and enough visibility to validate the result. A model that runs is only the starting point; a model that is adequate is one that can be trusted for the specific decision.

Once the model is judged adequate in principle, the next question is whether it has actually been accepted and validated. The next guide explains SMIB acceptance tests, weak-grid checks, field validation, disturbance playback and hardware-in-the-loop; the third guide covers how the large-scale case is built and run. This page is based on CIGRE Technical Brochure 881, with APS engineering interpretation.

References

CIGRE Technical Brochure 881 is the primary reference for this page; the CIGRE/CIRED brochure on inverter-based generation gives the wider modelling context; the WECC composite load model documents the aggregate DER / load representation; and IEC 61400-27-2 is the interfacing standard for wind-turbine models.

CIGRE Working Group, Electromagnetic Transient Simulation Models for Large-Scale System Impact Studies in Power Systems Having a High Penetration of Inverter-Connected Generation. CIGRE Technical Brochure 881.
CIGRE/CIRED Joint Working Group, Modelling of Inverter-Based Generation for Power System Dynamic Studies. CIGRE Technical Brochure.
Western Electricity Coordinating Council (WECC), Modelling and Validation Working Group, Composite Load Model (CMLD) Specification.
IEC 61400-27-2, Wind Energy Generation Systems — Part 27-2: Electrical Simulation Models — Model Interface. International Electrotechnical Commission.

Part 5 Reading now

Model Adequacy for Wide-Area EMT Studies

When a generic EMT model is adequate and when only a vendor-specific, site-specific model will do — the adequacy question for wide-area studies.

Series progress 5 of 12

Part 6

Acceptance Testing and Validation of EMT Models

How an EMT model earns trust: SMIB acceptance tests, weak-grid checks, staged field tests and playback validation against real disturbances.