Why Grid-Following Inverters Become Unstable Without a Grid Fault

Part A — Weak-Grid Instability Without a Fault at a Glance

When engineers think about power system instability, they usually picture a short circuit: a lightning strike on a 345 kV line, a breaker failure, a cascade of trips. Grid-following inverters can become unstable in a quieter way. On a weak transmission system, a solar, wind or battery plant can oscillate or lose control during completely normal operation — a routine line outage for maintenance, a capacitor bank switching, a small change in power setpoint, a modest phase jump from a distant event, or simply an increase in output as the sun rises. No fault is required. The cause is the interaction between the inverter's own control loops and the impedance of the network it is connected to.

The mechanism in one paragraph

A grid-following inverter measures the voltage at its terminals, uses a phase-locked loop (PLL) to find that voltage's angle and frequency, and injects current aligned with it. On a strong grid, the injected current barely changes the measured voltage. On a weak grid, the network impedance is large, so every change in injected current changes the measured voltage — which changes the PLL output, which changes the current again. The inverter is no longer following an independent grid voltage; it is partly following itself. Whether that feedback loop settles or grows depends on network impedance, PLL and current-controller tuning, operating point, plant controller settings and the behavior of nearby converters.

What Keentel's analysis shows

Keentel Engineering ran a parameter scan of a grid-following inverter plant on a Thevenin grid at short-circuit ratios (SCR) from 1.2 to 5, with PLL bandwidths from 5 Hz to 40 Hz. The only disturbance was a 5° grid phase step — no fault, no voltage dip.

When locked, the d-axis aligns with the voltage vector, so d-axis current carries active power and q-axis current carries reactive power. The PLL's bandwidth — roughly its natural frequency — sets how quickly it tracks angle changes:

2.3 Why the measured voltage is not independent

The voltage at the point of measurement is the grid's Thevenin voltage plus the drop across the network impedance caused by the inverter's own current:

With a small Zgrid, VPCC is essentially Egrid, an independent reference. With a large Zgrid, VPCC depends strongly on Iinv. The PLL is then tracking a voltage that its own controller is moving. This closed loop through the network is the root of weak-grid control interaction.

3. Grid strength metrics

3.1 Short-circuit ratio

Consider three plants in one area:

Each plant appears comfortably strong on its own, but WSCR = (900 × 150 + 750 × 200 + 600 × 100) ÷ 450² = 1.70. Together, the three plants are in a weak-grid condition that none of their individual SCRs reveals. ERCOT uses WSCR in its screening and has identified areas where WSCR falls toward 1.5, the level at which detailed EMT studies and mitigation are expected.

Composite SCR (CSCR) is calculated from short-circuit capacity with all nearby IBR contributions removed, divided by the total IBR rating in the area. Both metrics capture the same reality: grid strength is shared.

3.4 Typical screening interpretation

3.5 Why one SCR value is never enough

Keentel's stability map illustrates the last point clearly: at the same SCR, PLL bandwidth alone changes the outcome from stable to unstable.

4. Two different weak-grid limits

4.1 The static power-transfer limit

Even ignoring control dynamics, a weak network limits how much power a current source can push through it. For a plant at unity power factor connected through a Thevenin impedance, PCC voltage collapses as power increases. In Keentel's calculation with X/R = 10, the maximum steady-state power at unity power factor is approximately:

so a plant needs an SCR of about 1.8 to export its rating at unity power factor. With enough reactive support to hold PCC voltage at 1.0 pu, the limit rises to approximately the SCR itself:

Near these limits, the angle across the network approaches 90°, the voltage sensitivity to power becomes very high, and small disturbances can push the operating point past the nose of the PV curve. That is a loss of steady-state equilibrium — an instability that requires no fault.

4.2 Worked example: angle and voltage sensitivity

For a 100 MW plant at SCR 1.5 with X/R = 10, the network reactance on the plant base is about 0.66 pu. At rated output with voltage held at 1.0 pu:

At SCR 5, the same plant operates at about 11°. The voltage sensitivity to active power, dV/dP, is several times larger in the weak case. A PLL tracking angle and a voltage controller regulating magnitude both operate with much higher effective loop gain.

4.3 The dynamic (control interaction) limit

The second limit is dynamic: the closed loop formed by the PLL, current controller, voltage controller, plant controller and network impedance can become unstable at operating points well inside the static limit. This is the limit shown in the top panel of Figure 1, and it is the main subject of this guide.

5. Mechanisms of instability without a fault

5.1 Frequency mapping between frames

Oscillations observed in the dq frame map to two side-band frequencies in phase quantities. An oscillation at fdq in the dq frame of a 60 Hz system appears at:

For example, a 15 Hz PLL-driven oscillation in the dq frame appears as components at about 45 Hz and 75 Hz in phase voltages and currents. Recognizing these side-bands in PMU, DFR or plant controller data is often the first clue that an event involves converter control interaction.

5.2 Triggers that are not faults

6. Impedance-based stability analysis



6.1 The concept

Small-signal stability of an inverter connected to a grid can be analyzed by treating the inverter as a current source in parallel with an output impedance Zinv(s), and the grid as a voltage source behind Zgrid(s). The current flowing into the grid is:

The system is stable if the inverter is stable when connected to an ideal grid and the minor-loop gain Zgrid/Zinv satisfies the Nyquist criterion. In practice, engineers look for frequencies where:

• the magnitudes of Zinv and Zgrid cross, and
• the phase difference between them approaches or exceeds 180°, which happens when Zinv has a negative real part (negative damping).

6.2 Why the PLL creates negative resistance

At low frequencies, the PLL makes the inverter's q-axis impedance behave like a negative resistance whose magnitude grows with PLL bandwidth and active power output. Across a frequency band near the PLL bandwidth, the inverter therefore acts as an energy source for small perturbations. In a strong grid, the grid impedance is small and its magnitude does not cross the inverter impedance in that band. In a weak grid, the grid impedance magnitude rises and crosses in exactly the band where the inverter impedance is negatively damped. That is the frequency-domain view of the instability shown in Figure 1.

6.3 Practical frequency scanning

Engineers apply impedance analysis using frequency scans:

1. Inverter scan: inject small perturbations at a series of frequencies into an EMT model of the inverter or plant (with the vendor's actual control code) and measure the response to obtain Zinv(f) or its dq matrix form.
2. Network scan: obtain Zgrid(f) at the POI from an EMT or frequency-domain network model, including lines, cables, transformers, capacitor banks and filters.
3. Stability assessment: apply Nyquist or generalized Nyquist criteria and identify frequencies and operating conditions with low margin.
4. Confirm in time domain: verify critical cases with full EMT simulation.

Frequency scans are efficient for screening many network conditions and are increasingly requested by US transmission providers for low-strength interconnections.

7. Plant-level interactions

7.1 The plant controller

The plant power controller (PPC) regulates voltage or reactive power at the POI by sending setpoints to many inverters over a communication network. Its loop includes:

• measurement filters at the POI,
• PI or droop control,
• communication delays of tens to hundreds of milliseconds,
• inverter reactive current response,
• and the network's voltage sensitivity to reactive power.

The effective loop gain is proportional to the network's voltage sensitivity, approximately:

so the same PPC gain that is well damped at SCR 5 can be underdamped or unstable at SCR 1.5. Plant controller instability typically appears as slow voltage and reactive power oscillations, around 0.5 to 3 Hz.

7.2 Multiple plants regulating the same area

When several plants each regulate voltage at nearby buses with high gain and little droop, they can fight each other — one raising voltage while another lowers it — producing sustained oscillation. Coordinated voltage droop, appropriate gains and clearly assigned control responsibilities prevent this. The analysis must include all plants in the area, which is why WSCR-based screening and multi-plant EMT studies matter.

7.3 Collector system resonance

Large solar and wind plants contain long 34.5 kV collector cables whose capacitance resonates with transformer and network inductance, typically at several hundred hertz to a few kilohertz. If an inverter's output impedance is negatively damped near that resonance, harmonic instability can appear: growing distortion at a specific harmonic, often without any fundamental-frequency problem. These interactions require detailed collector models and vendor-specific inverter impedance data.

8. Model accuracy and validation

8.1 Why generic models are not enough

US planning studies use positive-sequence generic models — the WECC second-generation renewable models REGC, REEC and REPC — in PSS®E and PSLF. These are excellent for system-wide studies of active and reactive power response, voltage recovery and frequency response. They do not represent:

• the PLL and current controller in enough detail to capture fast control interaction,
• the converter's actual output impedance at frequencies above a few hertz,
• protection and fault-ride-through logic as implemented in firmware,
• or harmonics and resonances.

A positive-sequence study can show a plant as perfectly stable at an operating point where its actual control code would oscillate.

8.2 EMT models

Electromagnetic transient (EMT) models in PSCAD or similar tools simulate instantaneous three-phase waveforms with microsecond time steps, and vendor-specific models can include the inverter's real control code. NERC guidance recommends EMT models for IBRs in low-strength areas and identifies model quality expectations: correct initialization, representation of all relevant controls and protections, alignment with as-built settings and benchmarking against test data.

8.3 Model validation requirements

8.4 Settings drift

A model is only as good as its parameter set. Inverter firmware updates, parameter changes during commissioning, plant controller retuning and the addition of new equipment can all invalidate a validated model. Change management — tracking every firmware version and parameter change and re-verifying stability when they change — is as important as the original study.

9. US requirements relevant to weak-grid stability

10. Mitigation options

10.1 Worked example: how much synchronous condenser capacity?

Suppose a 300 MW solar plant has an SCR of 1.6 at its 230 kV POI under a critical N-1 outage, and EMT studies show the vendor's standard tuning requires an SCR of at least 2.0. The required short-circuit capacity is 2.0 × 300 = 600 MVA, compared with 1.6 × 300 = 480 MVA available. The deficit is 120 MVA of short-circuit contribution. A synchronous condenser contributes roughly its rating divided by its subtransient reactance through its step-up transformer:

A 50 MVA condenser with X″d = 0.15 pu and a 0.10 pu step-up transformer contributes about 50 ÷ 0.25 = 200 MVA, more than enough in this simplified estimate. In practice, the contribution at the POI depends on where the condenser is connected, and the EMT study must confirm that the plant is stable with the condenser — and with the condenser out of service, if that is a credible condition.

11. A practical study workflow

1. Screen grid strength with SCR, ESCR, WSCR or CSCR for normal and N-1 conditions, including planned outages and future projects.
2. Request vendor models early: EMT model with exact firmware and parameters, plant controller model, inverter impedance data where available.
3. Check model quality: initialization, flat runs, response to small steps, comparison with positive-sequence models.
4. Build an adequate network model: enough of the transmission system, nearby IBRs, capacitor banks, series compensation and HVDC.
5. Run small-disturbance cases: phase steps, voltage steps, power setpoint changes, capacitor switching and line switching across SCR conditions.
6. Run large-disturbance cases: faults and ride-through, to confirm that tuning chosen for small-signal stability does not compromise fault response.
7. Perform frequency scans where multiple converters or resonances are present.
8. Identify and test mitigation: tuning, equipment, operating limits.
9. Document settings and models that become part of the interconnection agreement.
10. Validate after commissioning using high-resolution field data and update models as required.

12. Monitoring and event analysis

Even with careful studies, oscillations sometimes appear in operation. Early detection limits consequences:

When an oscillation is detected, operators can often stabilize the system quickly by reducing plant output, changing plant controller mode or restoring an outaged line, while engineers analyze the root cause.



13. Grid-following versus grid-forming: an engineering comparison

Grid-forming capability is increasingly attractive for storage plants and hybrid projects in weak areas, but it is not a universal fix. Grid-forming plants must also be studied for their interaction with the network, protection and other converters.

14. Communicating results to transmission providers

A weak-grid stability study report that transmission providers can accept quickly typically includes:

1. Grid strength assessment with all metrics and the contingencies considered.
2. Model description, including firmware versions, parameter sets and model quality checks.
3. Study cases, with disturbances, operating conditions and network configurations.
4. Results, with plots and quantitative damping or oscillation metrics.
5. Sensitivities, particularly to SCR, dispatch, PLL and PPC parameters.
6. Mitigation and final settings.
7. Commissioning and validation plan.

Clear, quantitative reporting reduces the cycles of comments and responses that can delay interconnection agreements.

15. Worked example: contingency SCR at a 230 kV POI

Grid strength must be checked for the conditions the plant will actually see, not only the all-lines-in-service case. Consider a 250 MW solar-plus-storage plant connected to a 230 kV switching station with three outgoing lines.

The all-lines-in-service SCR of 4.0 suggests a moderately strong connection. But under the credible condition of the strongest line out with a nearby generator decommitted, ESCR falls to about 1.7 — inside the range where Keentel's stability map shows sustained oscillation for PLL bandwidths of 15 Hz and above. Under the N-2 condition, ESCR of 1.16 approaches the static limit of the plant at unity power factor (about 0.55 × 1.32 ≈ 0.73 pu of rating), so the plant could not export full power stably even with perfect controls.

The engineering conclusions are:

1. The plant must be tuned and verified for ESCR down to about 1.7, not 4.0.
2. An operating limit or automatic runback is needed for the N-2 condition, or that condition must be shown to be outside the planning criteria.
3. The interconnection study must state which contingencies were considered, because the transmission provider will ask.

16. Worked example: plant controller loop gain versus SCR

A plant power controller regulates POI voltage with an integral gain Ki (pu reactive power per pu voltage error per second). The voltage change caused by a reactive power change on the plant base is approximately:

so the effective integral loop gain seen by the controller is:

With a communication and processing delay Td, the loop is a delayed integrator. Its phase margin is approximately:

Suppose the controller has Ki = 20 pu/pu/s and Td = 150 ms (communication, measurement filtering and inverter response combined).

The same gain gives a comfortable 56° phase margin at SCR 5, a marginal 4° at SCR 2, and instability at SCR 1.5 — a slow oscillation near 2 Hz, exactly the kind of event described in Case Study 1. The cure is not mysterious: scale the gain with expected SCR, add voltage droop (which reduces the effective integral action and shares regulation among plants), and reduce communication delay where possible. A simple sizing rule is to choose Ki so that the phase margin is at least 45° at the minimum credible SCR:

For SCRmin = 1.7 and Td = 0.15 s, Ki should not exceed about 8.9 pu/pu/s. This simplified calculation is a screening tool; the final gain must be verified with the vendor's plant controller model in EMT and positive-sequence simulation.



17. Worked example: how PLL bandwidth sets the vulnerable frequency band

A PLL tuned to bandwidth fPLL with damping ratio ζ has PI gains approximately

The PLL introduces negative damping into the inverter's impedance around its own bandwidth, and the size of that negative damping scales with active power output. A 5 Hz PLL restricts the problem band to near-fundamental frequencies where the grid impedance is small and largely inductive. A 30 Hz PLL extends negative damping into a band where weak-grid impedance is large. That is why reducing PLL bandwidth is usually the first tuning lever.

The trade-off is that a slow PLL tracks phase jumps and frequency changes more slowly, which affects fault recovery, phase-jump ride-through and fast frequency response. IEEE 2800-2022 sets requirements for ride-through of phase-angle jumps and for performance during and after disturbances, so tuning must balance small-signal stability against large-disturbance performance. Keentel's practice is to scan PLL bandwidth across the full credible SCR range and confirm ride-through cases at the chosen setting, rather than optimizing for one condition.

18. Common misconceptions corrected

Weak-grid stability is often summarized in simplified ways that are directionally correct but incomplete. Keentel's review of the common explanations highlights the following points.

19. Small-signal modeling in practice

19.1 State-space linearization



For research and detailed analysis, an inverter and network can be represented in state-space form and linearized around an operating p

The eigenvalues λ = σ ± jω of A show the frequency and damping of each mode. The damping ratio is:

Planning practice typically looks for damping ratios of at least 3 to 5% on all oscillatory modes under credible conditions, with some utilities specifying their own criteria. Participation factors identify which states — PLL, current controller, plant controller or network — dominate each mode, pointing directly to the most effective mitigation.

19.2 Limits of linear analysis

Linearized models are exact only for small perturbations. Real inverters include current limits, rate limits, mode switching (for example entering and leaving fault-ride-through logic) and protection. A mode that is lightly damped in a linear model may settle into a sustained oscillation bounded by a limiter, or a large disturbance may push the plant into a mode switch that a linear model cannot represent. Linear analysis is therefore a screening and insight tool; EMT simulation remains the reference.

19.3 Measuring damping from simulation or field data

When an oscillation is observed in an EMT run or in PMU data, the damping ratio can be estimated from successive peaks using the logarithmic decrement:

For example, if successive peaks of a 2 Hz reactive power oscillation are 12, 10.8 and 9.7 Mvar, δlog ≈ 0.105 and ζ ≈ 1.7% — well below a 5% criterion even though the oscillation is decaying. A decaying oscillation is not necessarily an acceptable one.

20. Data to request from inverter and plant controller vendors

Requesting these items early in the interconnection process avoids study delays when the transmission provider identifies a low-strength condition.

20A. Commissioning and field verification in weak areas

Studies predict stability; commissioning proves it. In low-strength areas Keentel recommends a staged commissioning approach that tests the very controls most likely to interact with the network.

Neutral-side CTs are often used for generator differential protection; terminal-side CTs are commonly used for directional power. The protection engineer must confirm which CT feeds the 32 element, that its polarity is correct for the chosen power convention and that its accuracy is appropriate.

16.1 Loss of potential

If a VT fuse blows or a VT circuit is opened, the measured voltage on one or more phases collapses, and the calculated three-phase power can be wrong in magnitude or sign. Reverse power elements should be supervised by loss-of-potential (LOP, device 60) logic, which blocks voltage-dependent functions when a VT problem is detected. Without this supervision, a VT fuse failure on a lightly loaded unit can cause a false reverse power trip.

17. Worked example 2: 100 MVA combustion turbine at 13.8 kV

Because motoring power is large, a pickup of half the motoring power — about 7% of MW rating, 6 MW — gives a large margin above measurement error:

With rated current of 4,184 A and full reactive output of about 60 Mvar, even a 0.5° angle error gives a measurement error of only about 0.5 MW, under 10% of the setting. The time delay is chosen from the manufacturer's allowable motoring duration and plant operating philosophy, typically 5 to 10 seconds.

The contrast with the steam unit is the core lesson: the same protection function needs completely different design attention depending on the prime mover. A combustion turbine 32 setting is easy to make secure; a condensing steam turbine 32 setting requires a measurement accuracy review.

18. Phase-angle compensation in numerical relays

Many numerical generator relays allow the engineer to enter a CT phase-angle correction. If the CT's phase displacement at low current is known from test data or the manufacturer's accuracy curves, the relay rotates the measured current phasor by the correction angle before calculating power:

The correction should be based on test data at the current levels seen during motoring, not at rated current, because CT phase errors are typically larger at low current. Overcompensation is as harmful as no compensation, so corrections should be verified during primary testing by comparing relay power with revenue metering at minimum stable load and high reactive output.

19. Motoring duration and turbine damage

Manufacturers often specify how long a unit can safely motor. For condensing steam turbines, allowable times depend on condenser vacuum, exhaust hood cooling sprays and blade design, and may range from several seconds to a few minutes. For units with non-condensing or back-pressure turbines, windage heating can be more severe. Factors that change allowable motoring time include:

• Condenser vacuum: poor vacuum increases windage losses and blade temperature.
• Exhaust hood sprays: spray water cools the last stages and extends allowable time, but can cause erosion if used excessively.
• Unit size: longer last-stage blades experience higher windage heating.
• Steam remaining in the turbine: extraction lines and reheaters can feed steam back into the turbine after a trip.

Reverse power time delays should be well inside the manufacturer's allowable motoring time, with margin for breaker failure clearing if the primary trip path fails.

20. Reading event records

When a generator trips on reverse power, the relay's event report and the plant's digital fault recorder (DFR) or sequence-of-events recorder (SER) show what happened. An engineer reviewing the event should check:

Each step should be compared directly with simulation results using the same disturbance. Agreement in rise time, overshoot, settling time and oscillation frequency builds confidence that the model will predict behavior under conditions that cannot safely be tested in the field, such as low-SCR outages. Disagreement is an early warning that the model, the settings or the network representation needs correction — far better discovered at commissioning than during a summer peak with a key line out of service.

Commissioning tests in weak areas must be coordinated closely with the transmission operator. Test windows are often chosen when system strength is representative, and operators may require the ability to stop a test immediately if oscillations appear. A clear test plan with stop criteria, communication protocols and pre-agreed fallback settings is part of good engineering practice.

Ongoing performance monitoring

After commissioning, periodic review of plant event records against the validated model keeps the model honest. Keentel recommends reviewing any event in which the plant experiences a phase jump, voltage step or line switching above a defined threshold, and comparing recorded response with model playback. Gradual divergence can reveal component aging, undocumented setting changes or network changes that have altered system strength.

21. Key takeaways

1. Instability does not require a fault. Normal operating changes can trigger control interaction in weak grids.
2. The PLL makes the inverter follow a voltage it partly creates. That feedback through the network impedance is the root cause.
3. There are two weak-grid limits: a static power-transfer limit and a dynamic control interaction limit.
4. SCR is a screening tool, not a verdict. ESCR, WSCR, frequency-dependent impedance and inverter design all matter.
5. Model accuracy is decisive. Generic positive-sequence models cannot show fast control interaction; validated vendor EMT models can.
6. Mitigation has many forms: tuning, plant controller coordination, grid-forming technology, synchronous condensers and reinforcement.

22. How Keentel Engineering helps

Keentel Engineering performs grid strength, EMT and interconnection studies for solar, wind, storage and hybrid developers, independent power producers, utilities and transmission owners, from 4 kV to 765 kV.

Why clients choose Keentel Engineering

• Control-aware power system engineering: we understand the inverter's control loops as well as the network they connect to.
• EMT depth: detailed PSCAD modeling of plants, collector systems and transmission networks.
• Interconnection experience: practical knowledge of ERCOT, MISO, SPP, PJM, CAISO and utility study requirements.
• Vendor-neutral perspective: experience across inverter manufacturers and control architectures.
• Licensed professional engineering: a Florida-registered firm with offices in Tampa, Austin, Sacramento and Baltimore, supporting projects nationwide.

Interconnecting a solar, wind or storage project in a weak area? Talk to Keentel Engineering at (813) 389-7871, contact@keentelengineering.com, or book a 15-minute scoping call at calendly.com/keentel-engineering/15min.

Part C — Case Studies

The following are illustrative scenarios based on conditions typical of US IBR interconnections. They show how Keentel Engineering approaches weak-grid stability problems; site details are generalised and figures are rounded.

Case Study 1 — 200 MW solar plant at 138 kV: oscillation during a planned outage

Situation. A 200 MW solar plant connected at 138 kV operates normally for a year. During a planned outage of a nearby 138 kV line for maintenance, operators observe sustained 2 Hz oscillations in plant reactive power and POI voltage around midday. No fault occurred.

Analysis. Keentel's review shows that the SCR at the POI falls from about 3.2 with all lines in service to about 1.7 during the outage. PMU data show the oscillation in the plant controller's voltage loop. EMT studies with the vendor and PPC models reproduce the oscillation and confirm that the PPC voltage control gain, tuned for normal conditions, has very low damping at SCR 1.7.

Engineered solution. Keentel retunes the PPC voltage control with additional droop and reduced gain, verifies stability across the SCR range including outage conditions, and develops an operating procedure that limits plant output during specific planned outages until the new settings are commissioned.

Outcome. The plant operates stably through subsequent outages, and the transmission operator incorporates the outage-dependent operating guidance into its procedures.

Case Study 2 — 345 kV wind and storage cluster: WSCR and multi-plant interaction

Situation. A developer proposes a 250 MW battery plant near two existing wind farms at 345 kV. Each plant's individual SCR exceeds 3, but the transmission provider's WSCR screening for the area returns about 1.6.

Analysis. Keentel builds a multi-plant EMT model including both wind farms with their vendor models, the proposed battery and the surrounding 345 kV network. Small phase steps produce a poorly damped 12 Hz oscillation in the dq frame (visible as about 48 Hz and 72 Hz components in phase currents) involving the wind turbine converters and the battery's PLL. Frequency scans identify negative damping in the battery's impedance near the interaction frequency.

Engineered solution. Keentel evaluates the battery vendor's grid-forming option and a retuned grid-following option. The grid-forming configuration provides well-damped response across all studied conditions and improves damping for the existing wind farms.

Outcome. The battery is interconnected with grid-forming controls, the area's oscillation risk is reduced, and the study supports the transmission provider's broader weak-grid planning.

Case Study 3 — 150 MW solar plant at 115 kV: harmonic instability in the collector system

Situation. During commissioning of a 150 MW solar plant connected at 115 kV, voltage distortion at about the 23rd harmonic grows whenever output exceeds 60% of rating. Fundamental-frequency voltage and power are normal.

Analysis. Keentel's harmonic and EMT analysis shows a parallel resonance near 1.4 kHz formed by the long 34.5 kV collector cables and the main power transformer. The inverter's output impedance has negative damping near that frequency at higher output, driven by its current controller and PWM delay.

Engineered solution. Keentel works with the inverter vendor to enable active damping in the current controller and evaluates a small passive damping filter on the 34.5 kV bus as backup. Revised settings are verified with EMT simulation and field measurements.

Outcome. The harmonic growth disappears at all output levels, and voltage distortion at the POI meets IEEE 519 limits.

Part D — Technical FAQ: Weak-Grid Stability of Grid-Following Inverters

References and Further Reading

Links were current at publication (October 2026).



• IEEE Std 2800-2022, Standard for Interconnection and Interoperability of Inverter-Based Resources Interconnecting with Associated Transmission Electric Power Systems; IEEE P2800.2. IEEE.
• NERC Reliability Guideline, Integrating Inverter-Based Resources into Low Short Circuit Strength Systems. North American Electric Reliability Corporation.
• NERC Reliability Guideline, Electromagnetic Transient Modeling for BPS-Connected Inverter-Based Resources. North American Electric Reliability Corporation.
• NERC Reliability Standards PRC-029, TPL-001, MOD-026 and MOD-027; NERC disturbance reports on inverter-based resource performance. North American Electric Reliability Corporation.
• IEEE/CIGRE, Definition and Classification of Power System Stability — Revisited and Extended (2021). IEEE Transactions on Power Systems.
• J. Sun, Impedance-Based Stability Criterion for Grid-Connected Inverters. IEEE Transactions on Power Electronics.
• IEEE Std 519-2022, Standard for Harmonic Control in Electric Power Systems. IEEE.
• FERC Order No. 2023, Improvements to Generator Interconnection Procedures and Agreements. Federal Energy Regulatory Commission.
  

Disclaimer

This article is general technical information for educational purposes. It is not engineering advice for any specific project and does not create a professional relationship. The stability map and power-transfer limits come from Keentel Engineering's simplified single-plant models with illustrative parameters, not field data; actual stability limits depend on the specific inverter firmware, plant design, network and operating conditions and must be verified with validated vendor models. Project designs must follow current standards, transmission provider and ISO/RTO requirements and project-specific engineering analysis, sealed by a licensed professional engineer where required.



Case studies are illustrative scenarios based on typical project conditions, not records of specific client projects. Figures are rounded and indicative.

IEEE, NERC, FERC, ERCOT, WECC, PSS®E, PSLF, PSCAD™ and other names are the property of their respective owners. Keentel Engineering is not affiliated with or endorsed by these organisations or any software or equipment vendor.