S5.2.5.10 is the clause of Schedule 5.2 that requires a generating system to have protection that disconnects it, or parts of it, when it is operating in a way that is unstable. For synchronous machines that traditionally meant pole-slip protection. For inverter-based plant the instability of concern is oscillation: sustained swings in active power, reactive power or voltage driven by control interactions, often at weak connection points.
For years many inverter plants proposed the automatic standard with little more than an OEM statement. That is changing. NSPs increasingly expect a defined oscillation detection function, with stated frequency bands, thresholds and actions, and they want to see it designed and tested.
Why Oscillations Became a Concern
Inverter plants are control systems connected to other control systems. When the grid is weak, or when several plants with fast controls sit electrically close together, their controls can interact. The results observed in the NEM and internationally include:
- sub-synchronous oscillations, typically in the range of a few hertz to tens of hertz, visible in active and reactive power
- voltage oscillations that cause flicker for nearby customers
- low-frequency oscillations involving plant controller and network dynamics
These can persist for long periods without tripping anything, because conventional voltage, frequency and overcurrent protection does not see them. The plant keeps running while degrading power quality and sometimes destabilising neighbours. S5.2.5.10 is the clause that gives NSPs a way to require plant to recognise the condition and act.
What a Detection Function Looks Like
Requirements vary between NSPs and continue to evolve, so treat the following as a description of the features commonly asked for rather than a fixed specification. Check the NSP's current requirements for your connection.
| Feature | Typical expectation |
|---|---|
| Measured quantities | Active power, reactive power and voltage, ideally at the connection point, using single-phase as well as positive sequence quantities |
| Frequency range | Coverage from around 0.1 Hz up to around 25 Hz, split into sub-bands with separate settings |
| Thresholds | At least two per quantity and band: a warning alarm and a disconnection threshold |
| Magnitude setting | Peak-to-peak oscillation amplitude as a percentage of rating, settable to small values |
| Time setting | Pick-up time per band, so short transients do not trigger action |
| Actions | Configurable: alarm only, run-back, run-back then trip, partial trip or full trip |
| Source discrimination | Ability to distinguish oscillations originating in the plant from those imposed by the grid, for example from the phase relationship between voltage and power |
| Blocking | Automatic blocking during frequency excursions outside the normal band, with timed re-enable |
| Override | Local and remote enable and disable |
| Data | All quantities, alarms and flags to the NSP's SCADA, with high-resolution records from before to after each event |
The source discrimination feature matters. A plant that trips because a neighbouring plant is oscillating has made the problem worse, not better. The intent is that a plant disconnects when it is the cause, and reports when it is a victim.
How It Is Proposed in the GPS
At the connection application stage the detailed design does not usually exist, so the proposal is a commitment with defined content. A typical negotiated standard records that:
- the plant will not cause instability that would adversely affect the power system, assessed against AEMO's power system stability guidelines
- voltage disturbances from sustained oscillation are limited by the plant's rapid voltage change and flicker limits under S5.2.5.2
- an instability detection system will alarm and, where agreed, disconnect the plant for oscillations in P, Q or V beyond agreed thresholds
- the detailed detection scheme, settings and tests will be agreed with the NSP during detailed design
Support it with an OEM statement listing the detection functions the plant controller or inverters actually provide. Reviewers distinguish between a function that exists in the product and a function that the model simply does not show.
Monitoring points
NSPs that require an oscillation detection function generally add matching monitoring points under the SCADA clause, S5.2.6: the detection alarms and trip flags, and often a flicker measurement on a short window. Plan the SCADA points list with this in mind.
Commissioning: Alarm First, Trip Later
Setting oscillation protection is genuinely uncertain. Set it too sensitive and the plant trips for harmless events; too insensitive and it never acts. A common approach is a staged introduction:
Introducing an Oscillation Detection Function
- Agree the detection scheme, bands and initial settings with the NSP during detailed design
- Test the function by secondary injection before energisation, so each band and threshold is proven
- Run in alarm-only mode for an agreed demonstration period after commissioning
- Review recorded events with the NSP and adjust thresholds
- Enable the agreed run-back or trip actions only once the settings are shown to be appropriate
- Keep the settings in the plant's settings register, the model documentation and the GPS consistent
The alarm-only period gives both parties evidence. It also protects the generator from nuisance tripping while the network around it is still being understood.
Studies That Support the Clause
S5.2.5.10 does not have a single dedicated study, but several studies feed it:
- Small-signal and damping checks under S5.2.5.13, which show the plant's own controls are well damped across the range of grid strength
- Withstand SCR tests under S5.2.5.15, which show where the plant's stability margin runs out
- Wide-area studies that include nearby inverter plant, where interaction risks show up
- Model quality, because a detection function cannot be tuned against a model that oscillates numerically
If the studies show any lightly damped mode, expect the NSP to focus its detection requirements on that frequency range.
Get the Detection Scheme Defined Early
The longer S5.2.5.10 is left as a vague commitment, the more likely it becomes a late requirement that the chosen plant controller cannot meet without a software change.
ElectraGlobe prepares the protection clauses of GPS proposals, works with OEMs to define oscillation detection schemes, and supports their testing at commissioning. See our AEMO GPS applications service, our explainer on common GPS application pitfalls, or contact us.
Frequently Asked Questions
What does S5.2.5.10 require?
It requires protection that disconnects a generating system, or parts of it, when it operates unstably. For inverter-based plant this is increasingly interpreted as an oscillation detection function that alarms and can disconnect the plant for sustained oscillations in active power, reactive power or voltage.
What frequency range should oscillation detection cover?
Requirements vary by NSP, but detection from around 0.1 Hz up to around 25 Hz, split into bands with separate thresholds and time settings, is a common expectation. Check the NSP's current requirements for the specific connection.
Why distinguish grid-originated from plant-originated oscillations?
If a plant trips because another plant is oscillating, it removes support from the network without fixing the cause. Discrimination lets the plant act when it is the source and simply report when it is responding to an oscillation imposed by the grid.
When is the detection scheme finalised?
Usually during detailed design. The GPS records the commitment and the agreed approach at application stage, and the detailed settings are agreed with the NSP and tested before and after energisation, often with an alarm-only demonstration period first.