A grid-forming battery energy storage system behaves like a voltage source behind an impedance. It holds its own internal voltage and phase angle and lets current flow in response to the network, rather than measuring the network voltage and injecting a calculated current as a grid-following inverter does. That gives it inherently fast, voltage-supporting behaviour, and the ability to operate at very low system strength.
It also means some Generator Performance Standards (GPS) clauses, written with grid-following control in mind, do not map neatly onto how the plant works. The studies are the same; the way results are measured and argued is different. This article focuses on those differences. Our article on BESS project development in Australia covers the wider project context.
Where the Differences Show Up
| Clause | Grid-following expectation | Grid-forming reality | Consequence |
|---|---|---|---|
| S5.2.5.5A fault response | Reactive current set by a gain on voltage deviation | Current set by internal voltage and a virtual impedance | The gradient achieved varies with fault depth, grid strength and time |
| S5.2.5.5A recovery | Active power ramps back after clearance | Active power can overshoot briefly as the internal angle realigns | Overshoot above rating must be explained and bounded |
| S5.2.5.13 voltage control | Fast reactive power response to voltage steps | Inherent response plus a slower outer loop | Reference following at low SCR can be capped by current limits or inverter droop |
| S5.2.5.15 withstand SCR | A minimum SCR of 3.0 is common | Stable operation at very low SCR is possible | A negotiated withstand value well below 3.0, with commercial benefit |
| Benchmarking | RMS and EMT models enter fault mode on similar triggers | EMT entry often driven by current, RMS by a voltage threshold | Systematic mismatch near the fault mode threshold |
Fault Response: Virtual Impedance, Not a Gain
The automatic access standard in S5.2.5.5A asks for capacitive reactive current of at least 4% of maximum continuous current for each 1% drop in voltage below the initiating point, with a rise time of no more than 40 ms. A grid-following inverter delivers this through a configured gain.
A grid-forming inverter responds to a voltage drop because the difference between its internal voltage and the depressed network voltage drives current through its virtual impedance. When that current reaches the inverter's limit, a current limiting scheme, often a larger virtual impedance switched in during the fault, takes over. The result is a fast, natural response, but one whose "gradient" is not a setting:
- it depends on the fault depth, because current limiting engages for deep faults and not for shallow ones
- it depends on grid strength and X/R, because those change how much current a given voltage difference drives
- it can decay through a long fault, as the internal voltage and angle move towards the faulted network conditions
That last point makes the measurement instant decisive. Reactive current read early in the fault can look very different from reactive current read at the end of a long, breaker-fail-duration fault. Agree with the NSP and AEMO when the value is to be read, for example at typical primary clearance time, and state it in the GPS and the report.
Typical negotiated positions
Grid-forming plants frequently land on a negotiated S5.2.5.5A standard. The common elements are:
- a reactive current gradient below 4% per 1%, often expressed with different values for capacitive and inductive response
- wider initiating thresholds, reflecting when the current limiting scheme actually engages
- settling or recovery times somewhat longer than the automatic values
- a response described for all faults outside the initiating range, whether or not the fault mode triggers
The justification rests on the minimum access standard being met, the response being adequately controlled, and wide-area studies showing no adverse effect. The fact that a grid-forming response supports voltage from the first instant of the fault, rather than after a detection delay, is a legitimate part of the argument, but it does not replace the numbers.
Active Power Overshoot After Clearance
When a fault clears, the network voltage and angle jump back. A grid-forming inverter's internal angle has to realign, and during that realignment active power can briefly exceed its pre-fault level, sometimes above rating. Reviewers notice this. The report should show that the overshoot is brief, bounded, damped, and within the plant's equipment capability, and that it does not cause voltage or frequency problems nearby.
Voltage Control at Low System Strength
In the S5.2.5.13 step tests, a grid-forming plant may fail to follow a reactive power or power factor reference at minimum SCR because its current limit or internal droop caps the output. That is acceptable when the plant is genuinely at a limit. It is challenged when it happens far from any limit, for example at zero active and reactive power. In that case reviewers will ask for re-tuning. See our article on S5.2.5.13 voltage and reactive power control for the step test matrix.
Withstand SCR: An Advantage Worth Demonstrating
S5.2.5.15 sets a minimum access standard of a short-circuit ratio of 3.0, and the performance standard records the lower of 3.0 and the value the plant can actually withstand. Grid-forming plant can often demonstrate stable operation at a much lower SCR.
That matters commercially as well as technically. Under the system strength framework, the quantity a connecting plant is charged for scales with its withstand SCR, so a robust, lower withstand value reduces the system strength charge. Demonstrate the lowest value the plant can reliably support across all operating modes, using the methodology in AEMO's system strength impact assessment guidelines, and declare any settings changed specifically to achieve it.
One caveat: charging at very low SCR is a power transfer problem, not a control problem. A weak network cannot deliver rated power into a load, whatever the inverter does. AEMO's approach to withstand testing in charging mode has been evolving, so check the current guideline before designing the study.
Benchmarking Mismatch
AEMO requires PSS/E (RMS) and PSCAD (EMT) models to agree within tolerance. Grid-forming models have a characteristic mismatch: in the EMT model, entry into the fault current limiting mode is usually triggered by measured current, so the voltage at which it engages varies with SCR and X/R. In the RMS model, it is often triggered by a fixed voltage threshold. Faults with retained voltage near that threshold therefore show different behaviour in the two models.
Reviewers do not accept "both models settle to the same level" as a pass on its own. They want the root cause identified, the steps taken to improve alignment listed, and an explanation of why further improvement is not possible. Plotting the fault mode flags from both models on the same chart makes the cause visible. Our article on PSS/E and PSCAD modelling covers the wider benchmarking process.
OEM Statements Reviewers Expect
OEM Evidence for a Grid-Forming BESS Application
- Description of the grid-forming control structure, with functional block diagrams
- Fault current limiting method, entry and exit conditions, and current priority
- Positive and negative sequence behaviour during unbalanced faults
- Low-SCR operating capability, with project-specific verification rather than general claims
- Statement on fault counting or lockout logic for multiple faults
- Statement on whether vector shift or phase angle protection is fitted
- Firmware and plant controller versions matching the submitted models
- Confirmation of which behaviours appear only in the EMT model
Get a Grid-Forming Application Through Cleanly
Grid-forming batteries are increasingly what NSPs want to see in weak parts of the network. Getting them through the GPS depends on explaining clearly how the plant behaves and agreeing early how its performance will be measured.
ElectraGlobe runs PSCAD and PSS/E studies for grid-forming and grid-following BESS, prepares GPS proposals and works through AEMO and NSP review. See our steady-state and dynamic studies service, or contact us.
Frequently Asked Questions
Does a grid-forming BESS meet the S5.2.5.5A reactive current requirement?
It can, but its response comes from a virtual impedance rather than a configured gain, so the reactive current achieved varies with fault depth, grid strength and time. Many grid-forming plants negotiate a standard that describes their actual response, supported by evidence that the minimum standard is met and the response is adequately controlled.
Why does the measurement instant matter for a grid-forming inverter?
Reactive current from a grid-forming inverter can decay during a long fault as its internal voltage and angle move towards the faulted conditions. The percentage of maximum current measured depends on when it is read, so the sampling instant should be agreed and stated.
Can a grid-forming battery operate below an SCR of 3.0?
Often yes. The performance standard records the lower of 3.0 and the demonstrated withstand value, so a grid-forming plant that can operate stably at a lower SCR can have that value recorded, which also reduces its system strength charge.
Why do grid-forming PSS/E and PSCAD models disagree?
A common reason is that the EMT model enters its fault current limiting mode on measured current while the RMS model uses a fixed voltage threshold, so faults near that threshold behave differently. The difference should be explained and minimised, not simply accepted.