Two hundred milliseconds. That’s roughly how long a weak part of the grid has to arrest a fault before protection systems start tripping things out in sympathy, cascading into the kind of event AEMO doesn’t like writing incident reports about. A spinning coal or gas turbine does this almost by accident, because physics gives it inertia whether anyone asked for it or not. A battery doesn’t have that luxury. Everything a battery does to support the grid, it has to be told to do by its inverter. And for a long time, most inverters in Australia were only ever told to follow, not to lead.
That distinction is the whole story of grid-forming inverters, and why engineers who’d happily ignore most grid jargon have started paying close attention to this one.
Following versus forming, in plain terms
Picture two people trying to keep a shared rhythm. One listens to the beat and claps along a fraction of a second after they hear it. The other sets the beat themselves, and everyone else claps to them. Grid-following inverters are the first kind. They measure the voltage waveform the grid is already producing, usually set by a large synchronous generator somewhere, and they synchronise their output to match it. That’s been perfectly adequate for decades because there was always a spinning machine somewhere nearby setting the beat.
Grid-forming inverters are the second kind. They don’t wait to measure an existing waveform; they establish one, with voltage and frequency characteristics that other generation can lock onto. Push a grid-forming battery system and it will respond to a disturbance in a few milliseconds, instinctively, the way a spinning turbine’s rotor does under Newton’s laws rather than under software instruction. That’s the part that matters: it’s not just faster control, it’s a different category of control, closer to physical inertia than to a feedback loop.
Why this has gone from niche to urgent
I’ve been tracking AEMO’s engineering framework documents on this for a couple of years now, and the tone has shifted from “watch this space” to something closer to “we need this built.” The reason is straightforward once you look at where coal is actually heading. Our own coverage of the pace mismatch between coal exits and replacement builds has tracked how fast synchronous capacity is leaving parts of the NEM, and every unit that retires takes its inertia with it, not just its megawatts.
South Australia is the live case study. It already runs instantaneous renewable penetration above 100 per cent of local demand on some days, something AEMO’s own operational data confirms, and it’s had to manage system strength deliberately rather than by accident, because the number of spinning machines left on the system at any given moment can get genuinely thin. Victoria and NSW are heading the same direction as their own coal fleets wind down, a point our look at the NSW coal exit outrunning its REZ build covers in more detail than I can here.
A worked example, because the units matter
Let’s be careful with the number here, because “100 megawatts” and “100 megawatt-hours” get swapped around constantly in coverage of this technology and they are not interchangeable. A 100 MW/200 MWh battery, a common configuration for a two-hour system, can only deliver 100 megawatts of power at any instant no matter how much stored energy sits behind it. Grid-forming capability doesn’t change that power ceiling. What it changes is what happens in the first 20 to 100 milliseconds of a disturbance, well before the battery’s four-hour or two-hour energy duration becomes relevant at all.
Think of it as two separate jobs with two separate units. The energy capacity, measured in megawatt-hours, is about how long the battery can sustain output; that’s the number that matters for evening peak shaving. The grid-forming behaviour is about milliseconds of voltage and frequency response, measured in power terms, and it barely touches the state of charge. A battery can be nearly empty and still perform its grid-forming duty, because that duty draws almost nothing from the stored energy pool. Conflating the two is a common mistake in press coverage of these projects, and honestly, in some marketing material too.
Where it’s actually being built
The Hornsdale Power Reserve in South Australia was Australia’s first big proving ground for advanced inverter control, and Neoen’s subsequent builds have continued to push the envelope; our profile of Neoen’s expansion from battery pioneer into wind developer touches on how that technical credibility has carried into newer projects. The Waratah Super Battery in NSW is the project I’d point to as the clearest statement of intent, because it was explicitly procured with system strength services in mind, not just energy arbitrage; we covered its slow start and overdue payoff in Waratah Super Battery: the shock absorber finally earning its keep. Akaysha Energy, which has no retail brand and doesn’t need one, has built its whole pitch around exactly this kind of ancillary service revenue, something we unpacked in our profile of Akaysha’s model.
None of this is cheap to retrofit. Grid-forming control requires more sophisticated inverter hardware and control software than grid-following, and manufacturers are still converging on common standards for how to specify and test it. The AEMC and AEMO have been working through exactly this problem via system strength frameworks and the broader Integrated System Plan, which our explainer on what the ISP actually says goes through in detail.
It isn’t a free lunch, and I’d push back on some of the enthusiasm
Here’s my mildly unfashionable view on this: grid-forming inverters get spoken about in some circles as though they solve the inertia problem outright, full stop, nothing more to discuss. They don’t. They provide a function analogous to inertia, synthetically, through fast control loops and sometimes supercapacitor buffering, but the physics underneath is still a chemical battery with finite power electronics ratings. Stack enough grid-forming batteries into a region and you genuinely can displace the need for synchronous condensers or minimum-synchronous-generation rules. But “enough” is doing a lot of work in that sentence, and AEMO’s own system strength gap studies, published periodically for each region, are the document to read before anyone tells you a given REZ is fully solved on this front. I’d rather see that number quoted honestly, even when it’s not flattering to a given project’s media release.
There’s also a cost question nobody loves discussing out loud. Retrofitting grid-forming capability, or procuring it as a service through contracts like those under the Capacity Investment Scheme, adds expense that ultimately lands somewhere in the transmission or market cost stack. Our ongoing look at who really pays for the transmission build-out is relevant background here, because system strength procurement is quietly becoming its own line item, not a free add-on to battery deployment.
Pumped hydro and spinning machines haven’t gone away
It’s worth remembering that grid-forming inverters aren’t the only way to get this job done. Synchronous condensers, essentially spinning machines with no turbine attached, do the same physical trick the old-fashioned way, and several are already running or planned in weak parts of the network. Pumped hydro, when the pump-turbine itself is a synchronous machine, offers genuine rotational inertia on top of its storage function, something our comparison of pumped hydro versus batteries for firming touches on. Snowy 2.0, whatever one thinks of its cost trajectory, will bring real inertia to a part of the grid that badly needs it, assuming it eventually gets finished.
My own hunch, for what it’s worth, is that the endgame looks like a mix: grid-forming batteries doing the fast, localised work near weak REZ nodes, with a smaller number of synchronous condensers and pumped hydro units providing bulk system strength across wider regions. Nobody’s going to solve this with one technology alone, and anyone pitching it that way is selling something.
What to watch over the next couple of years
Keep an eye on AEMO’s updated system strength requirements as each REZ reaches financial close, and on how many of the Capacity Investment Scheme’s battery tenders specify grid-forming control as a condition of contract rather than a bonus feature. That’s the signal that tells you whether this capability is becoming standard practice or staying a boutique add-on for flagship projects. The Clean Energy Regulator’s generation data and AEMO’s system strength guidance are the primary sources worth bookmarking if you want to track this without relying on press releases. The AEMC’s rule change work on essential system services, available through its own public register, is the other document I check before writing anything definitive on this topic.
I came off a long ride out past the Barossa a few months back thinking about exactly this, oddly enough, somewhere near a wind farm substation with a transmission tower humming overhead. The grid doesn’t care whether its stability comes from a spinning rotor or a very fast piece of silicon. It only cares whether the voltage holds for the next 200 milliseconds. That’s the bar grid-forming inverters have to clear, and increasingly, they are.
– Anjali Rao, Grid & Storage Correspondent
Image: Thennicke (CC BY-SA 4.0) via Wikimedia Commons