When the Short-Circuit Model Does Not Match the Inverter

Negative-Sequence Behaviour of Inverter-Based Resources, Why the Fixed-Angle Converter Model Breaks Down, and What That Means for Every Protection Setting Derived From It

1. Executive Summary

Short-circuit programs represent inverter-based resources with a converter-interfaced source model: a controlled current source whose positive-sequence output follows a prescribed voltage-dependent characteristic, and whose negative-sequence output is a fixed proportional relationship to negative-sequence voltage with a fixed phase angle. The model is convenient, it converges, and it lets a phasor-domain program produce a fault study for a plant that contains no synchronous machines.

It is also, for many units in service, an approximation with weak physical foundations — and the ways it fails are precisely the ways that matter for protection.

Electromagnetic transient simulation of real inverters using field settings shows three behaviours that the phasor model does not reproduce. Converters produce negative-sequence current during balanced three-phase faults, where the model predicts essentially none. The phase relationship between negative-sequence current and negative-sequence voltage is not fixed; it moves substantially from cycle to cycle and can sweep across most of the unit circle within the first several cycles of a fault. And the effective ratio between negative-sequence current and negative-sequence voltage bears little resemblance to the coefficient entered in the model, differing by an order of magnitude between balanced and unbalanced fault types.

The consequence is not that fault duty numbers are wrong. Interrupting duty at a transmission bus is set by the surrounding network, and a current-limited converter contributes little to it either way. The consequence is that a family of protection elements — negative-sequence directional, phase selection, and the polarising logic behind them — make their decision using exactly the quantity the model gets wrong, during exactly the window in which the model is least valid.

This paper explains why converters behave this way, what the behaviour does and does not break, how the standards framework has responded, and what a defensible study workflow looks like when the phasor model alone cannot answer the question.

The finding stated plainly

A converter is not a sequence-domain source. It is a closed-loop controller producing three phase currents, and the sequence decomposition of its output is a consequence of its control dynamics rather than a property designed into it.



Where the converter has no dedicated negative-sequence regulator, negative-sequence current is not controlled at all — it is a residue of the positive-sequence control loop operating on an unbalanced input. Modelling an uncontrolled residue as a fixed-angle source is where the error enters.

2. What the Short-Circuit Program Assumescccc

The converter-interfaced source model in a commercial short-circuit program is, in essence, three assumptions.

• Positive-sequence output follows a defined characteristic — typically a reactive current injection proportional to the positive-sequence voltage depression, subject to a total current limit and a priority rule between real and reactive components.
• Negative-sequence output is proportional to negative-sequence voltage through a gain, commonly called a K factor, with the current constrained to a fixed phase relationship to that voltage. This is mathematically equivalent to representing the converter as a fixed admittance in the negative-sequence network.
• Zero-sequence output is zero, because the converters sit behind a delta winding, so any zero-sequence contribution at the point of interconnection comes from the transformer grounding rather than from the inverters.

Each assumption is defensible as an engineering idealisation. The first is a reasonable description of what a grid-following converter is instructed to do in the positive sequence. The third is simply correct. The second is the problem — not because the form of the relationship is unreasonable, but because it presumes that the converter has a negative-sequence controller enforcing it. Many do not.



The model also carries an implicit assumption that is rarely examined: that the fault response is a steady state. A phasor short-circuit program computes one answer per fault. The converter, meanwhile, is a control system with a settling transient that is comparable in duration to the protection operating time.

3. What the Converter Actually Is

A grid-following converter measures terminal voltage, extracts a reference angle using a phase-locked loop, transforms measured currents into a synchronous reference frame, compares them against references produced by outer control loops, and drives the modulator to close the error. It is a sampled, closed-loop, nonlinear system with limiters.

That architecture has consequences that are invisible in the sequence domain.

3.1 The Reference Frame Can Be Wrong

The phase-locked loop tracks positive-sequence angle. During an unbalanced fault the terminal voltage contains a negative-sequence component that appears, in the positive-sequence reference frame, as a second-harmonic oscillation. Unless the loop includes a sequence separation stage, its output angle oscillates. Every current reference expressed in that frame inherits the oscillation, and the currents actually injected into the network are therefore modulated at twice fundamental frequency.

Decomposed into sequences, that modulation appears as negative-sequence current with a time-varying angle.

During a severe voltage depression the loop can lose lock entirely. The reference frame then drifts relative to the network, and the relationship between commanded and delivered current becomes arbitrary until lock is re-established.

3.2 There May Be No Negative-Sequence Controller

Regulating negative-sequence current requires a dedicated control path: sequence separation of the measured quantities, a second current regulator operating in the negative-sequence frame, and a reference generator that computes the required negative-sequence injection. Converters designed before unbalanced-fault injection was a requirement frequently have none of this. Their negative-sequence output is whatever the plant impedance and the positive-sequence loop produce, which is a function of hardware and tuning rather than of any setting an engineer can enter into a study.

3.3 The Limiter Is Nonlinear

Current limiting is the defining behaviour of a converter under fault. When the limiter saturates, the relationship between reference and output becomes nonlinear, and the priority logic — which component of current is reduced first — determines what the network sees. A saturated converter is not describable by a linear sequence-domain admittance, which is what the phasor model assumes it is.

3.4 The DC Side Talks Back

Unbalanced fault conditions produce a power oscillation at twice fundamental frequency on the DC link. That ripple propagates into the voltage control loop and back into current references. The AC-side sequence content and the DC-side ripple are coupled, and neither exists in the phasor model.

4. Negative-Sequence Current in a Balanced Fault

The observation that provokes the most disbelief is negative-sequence current during a three-phase fault. In the idealised sequence framework, a balanced fault produces no negative-sequence voltage and therefore, under the model, no negative-sequence current. Measured behaviour disagrees, and there are several sound reasons why.

• The fault is not perfectly balanced. Real three-phase faults involve arc resistance that differs per phase, pole scatter in the initiating switching, and untransposed line sections. A small residual unbalance is normal.
• The network behind the converter is not perfectly balanced. Untransposed lines, unequal collector circuits, and asymmetric transformer impedances all produce standing negative-sequence voltage even in normal operation.
• The control transient itself is unbalanced. A step change in terminal voltage excites the phase-locked loop and the current regulators. During settling, the three phase currents are not a balanced set, regardless of how balanced the fault is. Sequence decomposition of an unbalanced transient necessarily yields negative-sequence content.
• Sequence extraction has a settling time. Filters used to separate sequence components inside the controller — and inside the measuring instrument — need roughly a cycle to converge. During that interval the extracted quantities are not the true sequence components of the waveform, and any control action based on them is acting on an error.
• Hardware tolerances. Gate timing, sensor gain and offset errors, and device characteristics differ between phases at the level that matters here.



The practical significance is not that a large negative-sequence current appears. It is that a non-trivial negative-sequence current appears while negative-sequence voltage is very small — which means the ratio between them, the quantity the model treats as a fixed gain, becomes very large and entirely disconnected from the configured value.

5. Why the Angle Wanders

If negative-sequence current is a controlled quantity produced by a dedicated regulator, its angle relative to negative-sequence voltage is a setting. If it is an uncontrolled residue, its angle is set by whatever produced it — and different mechanisms produce it at different times in the fault.

In the first cycle, the dominant contributors are the sequence extraction transient and the initial control excursion, whose phase has no defined relationship to the fault voltage. Over the next few cycles, the phase-locked loop settles or oscillates, current regulators reach their limits, and the limiter priority logic begins to govern. Later still, if the plant controller acts, the operating point moves again. Each of these regimes produces a different angular relationship, so the measured angle migrates as the fault progresses rather than converging to a constant.

When a converter has no negative-sequence regulator, the observed relationship is closer to that of a voltage-controlled current source with dynamic and nonlinear behaviour than to a fixed impedance. That is the correct engineering characterisation, and it is fundamentally incompatible with a fixed-angle sequence model.

Why an engineer should care about an angle

Magnitude errors change margins. Angle errors change decisions.

A negative-sequence directional element does not compare magnitudes. It compares the angle between negative-sequence current and negative-sequence voltage against an expected range, and declares the fault forward or reverse on that basis. If the angle is not where the model says it is, the element does not become less accurate — it declares the wrong direction.

6. The Effective K2 Is Not the Configured K2

The gain relating negative-sequence current to negative-sequence voltage is entered into the short-circuit model as a number. It is worth examining what that number means when the converter is not enforcing it.

For an unbalanced fault, negative-sequence voltage at the terminals is substantial, and the converter’s negative-sequence current — controlled or not — is of a comparable per-unit order. The effective ratio is within an order of magnitude of a typical configured value, and the model is at least dimensionally sensible even if the angle is wrong.

For a balanced fault, negative-sequence voltage is very small while the converter still produces measurable negative-sequence current from the mechanisms in Section 4. The effective ratio is therefore very large, and it varies as the transient decays. There is no single coefficient that describes both cases, which means a model calibrated against one fault type will misrepresent the other.



This is worth stating explicitly because it is a common source of false comfort: a model tuned to match an unbalanced fault case, validated against that case, and then applied across the full fault study is a model that has been validated only where it happens to work.

7. The First Three Cycles Problem

Protection is fast. A transmission line protection scheme with communication typically operates in one to two cycles, and a distance element without communication in one to three. Breaker operating time adds a further two to three cycles.

The converter behaviour described above is at its most erratic during precisely that interval. The first cycle contains the sequence extraction transient. The second and third contain the control settling and the limiter engaging. Whatever quasi-steady behaviour the converter eventually reaches, it reaches after the relay has already decided.

A phasor short-circuit program produces one answer, which implicitly represents a settled condition. Protection settings derived from that answer describe a condition that does not exist at the moment the protection acts. The mismatch is not a modelling refinement — it is the operating window.

The design consequence

Any protection scheme whose correct operation depends on a sequence-quantity angle at an inverter-dominated terminal should be evaluated against a time-domain simulation of the first several cycles, not against a single phasor result.

Where that evaluation cannot be performed, the scheme should be designed so that it does not depend on the quantity at all.

8. What This Breaks in Protection

9. What It Does Not Break

Overstating the problem is as damaging as ignoring it, and several things are genuinely unaffected.

• Equipment interrupting and momentary duty at transmission buses. A converter contributes a small multiple of its rated current, briefly. The duty that sizes breakers and switchgear is set by the surrounding network, and adding a converter changes it marginally.
• Line current differential protection. It compares currents at the two terminals and does not rely on sequence angle relationships to determine direction. It is the most robust choice for lines terminating at inverter-based plants, which is why it is increasingly specified as the primary scheme.
• Zero-sequence quantities derived from the interconnection transformer. The grounding source is a transformer, and a transformer behaves like a transformer. These quantities are well represented in conventional tools, provided the model attributes them to the transformer rather than to the plant.
• Steady-state load flow and reactive capability studies. The phenomena discussed here are fault-transient behaviours and do not affect steady-state analysis.



The correct conclusion is narrow and specific: the converter model is inadequate for the sequence-angle-dependent protection questions, and adequate for most of what else a short-circuit study is used for. Discarding the phasor study entirely would be an overreaction; relying on it for directional and phase-selection decisions is the error.

10. The Standards Response

The industry has been aware of this class of problem for some years, and the response has taken two forms: requiring converters to behave in a way that the sequence model can represent, and requiring models that can represent behaviour the sequence model cannot.

10.1 Requiring the Behaviour

The interconnection and interoperability standard for inverter-based resources connected to transmission systems addresses this directly. It requires negative-sequence current injection during unbalanced faults in addition to the positive-sequence reactive injection, and — critically for protection — it constrains the angle. For full-converter resources the negative-sequence current is required to lead the negative-sequence voltage by ninety to one hundred degrees, a window chosen to emulate the behaviour of a synchronous machine and thereby to keep conventional protection elements operating as intended. A wider window is permitted for doubly-fed machines, whose topology makes the tighter requirement impractical.

The standard requires the capability and the angular behaviour but does not prescribe a single gain; the setpoint is left to project-specific evaluation and agreement. The practical implication is that the gain in the study model should be the gain agreed and configured for the plant, and that agreement should exist before the protection study is performed.

This is the resolution of the apparent paradox in the measured behaviour. A converter whose negative-sequence angle sweeps freely is not exhibiting a law of nature; it is exhibiting the absence of negative-sequence control. The standard exists to require that control. Whether a specific unit implements it, and whether it is enabled and configured on the specific project, is a verifiable fact and should be verified rather than assumed.

10.2 Requiring the Model

The modelling standards have moved in parallel. The consolidated dynamic model verification and validation standard now requires verified electromagnetic transient models for inverter-based resources, covering the converters, collector system, plant controller, transformers, and the protective functions and limiters on both the AC and DC sides — and requires the electromagnetic transient and positive-sequence models to be cross-validated against each other under large-signal disturbances. Model updates are required after changes that alter dynamic response, including firmware, control mode, and settings changes.

Read alongside the ride-through performance requirements and the disturbance monitoring obligations that support them, the direction of travel is clear: behaviour during the first cycles of a fault is now a regulated, modelled, measured, and evidenced property of the plant rather than a vendor internal matter.

11. Grid-Forming Changes the Answer Again

Grid-forming converters do not resolve this problem; they change its shape. A grid-forming unit regulates voltage magnitude and angle behind a virtual impedance rather than injecting a controlled current, and its negative-sequence behaviour depends on the inner voltage control and the virtual impedance design. Some implementations actively suppress negative-sequence content, some leave it uncontrolled, and some are explicitly designed to present a defined negative-sequence characteristic.

The consequence for protection is that the negative-sequence characteristic of a grid-forming plant is a design parameter of the control implementation rather than a property of the technology class. It cannot be assumed from the datasheet, and it cannot be inferred from experience with grid-following units. Where grid-forming capability is being procured — increasingly common at weak interconnections and where islanding is a requirement — the negative-sequence behaviour and its compatibility with the protection scheme belongs in the technical evaluation at procurement, not in commissioning.

12. A Defensible Study Workflow

The workflow below reflects what the analysis actually requires rather than what the tools conveniently produce.

1. Use the phasor short-circuit study for what it is good at: equipment duty, coordination of time-overcurrent elements against non-converter sources, and the overall fault current distribution across the network.
2. Identify every protection element and scheme whose correct operation depends on a sequence-quantity angle at a terminal where an inverter-based plant is a significant source. That set — not the whole scheme — is the scope of the additional analysis.
3. Obtain the manufacturer’s validated electromagnetic transient model for the specific converter, with the firmware version and the control configuration that will actually be deployed. A generic model of the technology class will not answer the question, because the question is about the manufacturer’s control implementation.
4. Simulate the governing fault set in the time domain: balanced and unbalanced faults, at the terminal and at remote locations, at strong and weak system conditions, with the plant at high and low output and, for storage, in both charging and discharging states.
5. Extract the quantities the protection actually uses — sequence magnitudes and angles as a function of time from fault inception — over the first several cycles, at the relay location, using filtering representative of the relay’s own measurement.
6. Evaluate the elements against those time-varying quantities rather than against a single phasor result. Where the tools support it, replay the simulated waveforms into relay models or physical relays to observe the actual decision rather than inferring it.
7. Where an element cannot be shown to operate correctly, change the scheme rather than the setting. Current differential as the primary scheme, communication-assisted logic with weak-infeed and echo provisions, and reduced reliance on negative-sequence direction at the inverter terminal are the standard remedies.
8. Document the converter configuration the study assumed — negative-sequence injection enabled or not, gain, angle, priority logic, current limit, ride-through settings — as a controlled record, and re-verify after commissioning.
   

13. Procurement, Settings Control, and Validation

Three practices convert this from a recurring surprise into a managed engineering item, and all three sit upstream of the study.

13.1 Specify the Behaviour, Not Just the Rating

The converter specification should state the required unbalanced-fault behaviour explicitly: negative-sequence injection capability, the required angular relationship, the gain range and its adjustability, the current limit and the priority logic between sequence and between real and reactive components, and the ride-through configuration. A unit that cannot meet the required negative-sequence behaviour is a unit that will force protection scheme changes, and that is a procurement finding, not a commissioning finding.

13.2 Specify the Model as a Deliverable

Validated positive-sequence and electromagnetic transient models, in the required formats, with the right to use them for the studies the project must perform and to provide them to the transmission planner, should be contractual. Factory test evidence supporting the model should accompany it. Obtaining these after the purchase order is a negotiation; obtaining them as a deliverable is a line item.

13.3 Control the Settings

The negative-sequence configuration is a setting, and settings change during commissioning. A plant whose protection study assumed negative-sequence injection enabled with a specified angle, and whose converters were commissioned with that function disabled to resolve a tuning problem, has a protection scheme validated against a plant that does not exist. Maintain a controlled register of converter and plant controller settings, treat any change to fault-response configuration as a trigger for re-evaluating the protection study, and re-verify the model after commissioning — which is in any case now a modelling obligation rather than a good practice.

14. Reading This Result Correctly

Findings of this kind circulate quickly and are easy to over-read. Four qualifications keep the conclusion honest.

It is a finding about a unit, not about all inverters

Negative-sequence behaviour is a property of a manufacturer’s control implementation and its configuration. A result from one unit at one firmware revision with one set of field settings describes that unit. Units designed against the current interconnection performance requirements, with negative-sequence regulation implemented and enabled, behave differently — that is the entire point of the requirement.

It does not mean the phasor study is worthless

It means the phasor study is being asked a question outside its validity. Equipment duty, load flow, and coordination against conventional sources remain properly within scope. The sequence-angle-dependent protection questions are not.

It does not mean electromagnetic transient simulation is automatically right

An unvalidated transient model, or a validated model run with a configuration that does not match the plant, produces confident wrong answers faster than a phasor tool does. The value comes from a model verified against measured response, run at the configuration actually deployed.

The measurement window matters

Sequence quantities extracted with a one-cycle filter, a half-cycle filter, or the relay’s own filtering produce different numbers during a transient. Comparing results across studies, or against a relay’s behaviour, requires knowing which filtering produced them. Much apparent disagreement between analyses is disagreement about measurement, not about physics.

15. Keentel Electrical Power Engineering Services

Keentel Engineering is an electrical power systems engineering firm. Work of exactly this kind — where the answer depends on whether the model represents the equipment — is the centre of our practice.

15.1 Power System Studies

• Short-circuit and fault analysis with inverter-based resources represented as current-limited sources, including the sequence behaviour actually configured rather than a library default.
• Protective device coordination and selectivity studies, and selective coordination where required for emergency, legally required standby, and critical operations power systems.
• Protection scheme design and settings for lines and equipment terminating at inverter-dominated plants, including current differential, communication-assisted schemes, weak-infeed and echo logic, and phase-selection strategy where conventional logic cannot be relied on.
• Arc-flash incident energy analysis and labelling, aligned to the final configuration and the settings actually installed.
• Load flow, motor starting, reactive capability, and voltage regulation studies.
• Harmonic, flicker, and power quality studies, grid strength assessment, and transient stability analysis.

15.2 Electromagnetic Transient Modelling and Model Validation

• Electromagnetic transient studies for control interaction, weak-grid stability, ride-through verification, insulation coordination, switching and overvoltage transients, and fault-response behaviour at the timescales protection operates on.
• Positive-sequence and electromagnetic transient model development, cross-validation between platforms, and submittal packages prepared to the transmission planner’s format and level of detail.
• Model verification and validation against staged test data and disturbance records, including step-response test design, data capture specification, and parameter calibration.
• Baseline model and settings management across commissioning, with re-verification after firmware, control mode, or settings changes.

15.3 Interconnection and Compliance

• Point-of-interconnection engineering, substation and collector design, and interconnection facility design for generation, storage, and large-load projects.
• Interconnection application support, study-phase technical packages, and coordination with the utility, transmission provider, and system operator.
• NERC compliance support including ride-through design evaluation, disturbance monitoring architecture, model verification obligations, facility ratings, and relay loadability.

15.4 Design and Owner’s Engineer Services

• Substation and transmission design, medium- and low-voltage distribution design, and grounding, bonding, and lightning protection design.
• Design review of EPC and vendor submittals, QA/QC of third-party study packages, equipment specification including behavioural and model deliverables, and commissioning and test program support.

Keentel Engineering holds a Florida Certificate of Authorization and maintains offices in Tampa, Austin, Sacramento, and Baltimore, supporting projects across the interconnections.

16. Frequently Asked Questions

Notice and Disclaimer

This document is original technical content prepared by Keentel Engineering LLC for general professional information. It is not project-specific engineering advice and does not constitute a study, a protection design, a settings recommendation, or a compliance determination for any facility. Protection design and settings must be developed from project-specific analysis using validated models of the equipment actually installed.



Descriptions of converter behaviour in this document are general engineering discussion of mechanisms observed in inverter-based resources. Behaviour varies by manufacturer, product, firmware revision, and configuration, and no statement here should be taken as characterising any particular product, manufacturer, or installation.

Standards and regulatory references are provided by subject for orientation and are current to the date of publication to the best of our knowledge. The current published edition of each standard, and the requirements adopted by the applicable authority, utility, or system operator, govern and should be verified directly for any project decision.

Keentel Engineering LLC is an independent engineering consultancy. Reference to any standard, industry organisation, software tool, or equipment category in this document does not imply affiliation with, endorsement by, or sponsorship from any such organisation or manufacturer.