WA’s SWIS grid: an island with no way to call for help

There’s no button in Perth marked ‘call for help.’ If a generator trips on the South West Interconnected System on a stinking February afternoon, there’s no interconnector to lean on, no neighbouring state to lend a few hundred megawatts while things get sorted. The SWIS is generally described as one of the largest electricity systems in the world that isn’t connected to anything else. Everything it needs (capacity, reserve, frequency control) it has to grow itself.

I’ve been through the AEMO WEM documents enough times now that I keep a mental map of the SWIS running from Kalbarri down through Perth to Albany, with a long skinny leg out east to Kalgoorlie. It covers roughly 2.3 million people and most of WA’s economic activity, but it sits entirely apart from the National Electricity Market that the rest of this site spends most of its time writing about. Different rules, different operator function, different history. It’s worth treating as its own animal rather than a smaller cousin of the NEM.

An island grid, not a metaphor #

When people say ‘islanded grid’ about the SWIS they mean it literally, not as a figure of speech. There is genuinely no physical link to the eastern states’ grid, and building one across the Nullarbor has never stacked up on distance or cost. That isolation shapes almost every technical decision AEMO and Western Power make down there. Frequency control, for instance, has to be managed entirely within the SWIS’s own inertia and its own fleet of grid-forming and grid-following assets: a subject I’ve gone through in more detail in our piece on grid-forming inverters, because WA has had to think about synthetic inertia earlier and harder than most of the NEM.

AEMO does operate the SWIS, so there’s a shared institutional thread with the NEM, but it does so under an entirely separate rulebook: the Wholesale Electricity Market, or WEM, governed by the Electricity Industry Act 2004 (WA) rather than the National Electricity Law. Let’s be careful with that distinction, because it trips people up constantly in the trade press: the WEM is not a state-based version of the NEM, it’s a different market design that predates a lot of the NEM’s current architecture.

The Reserve Capacity Mechanism, not the CIS #

The clearest proof of that difference is the capacity market. On the east coast, firming is increasingly being bought through the Commonwealth’s Capacity Investment Scheme, which I’ve written about at length in how the Capacity Investment Scheme actually works. WA never needed a federal scheme to invent capacity payments, because the WEM has run its own obligation-based Reserve Capacity Mechanism for close to two decades. Retailers and large users have to hold enough certified capacity to cover their share of expected peak demand, with penalties if the system falls short. It’s a different animal to the CIS’s underwrite-and-tender model, and arguably a more mature one, having survived several redesigns already.

The other structural change worth flagging is what WA calls the New Energy Rules, which moved the WEM from a clunky day-ahead-style structure to real-time dispatch in October 2023 – bringing it philosophically closer to how the NEM already runs its five-minute settlement, a process I’ve laid out in how the NEM dispatches power every five minutes. It was overdue. The old WEM design was creaking under the weight of rooftop solar variability well before that reform landed.

Rooftop solar’s minimum demand problem, but sharper #

Every state has a version of the midday duck curve now, but the SWIS gets it in a more concentrated form because the grid is smaller and there’s nowhere for the surplus to go. AEMO’s WEM Electricity Statement of Opportunities has flagged minimum operational demand risk for several years running, as rooftop solar output on a mild, sunny weekend can push underlying demand on the grid down towards levels that used to only occur overnight. That’s a genuinely different problem to peak demand, and it’s the one I think doesn’t get enough attention outside WA. We’ve covered the general mechanics of this in rooftop solar grid stability: the honest read, but the SWIS is arguably the sharpest version of that story in the country, precisely because it can’t export the problem to a neighbouring region the way, say, South Australia occasionally can down an interconnector.

Worked simply: if a coal or gas unit needs to stay synchronised to hold the system stable, but underlying demand has fallen below what that unit can technically run down to, something has to give – either the rooftop solar gets curtailed, or the thermal unit runs uneconomically, or storage soaks up the difference. WA has been dealing with that three-way trade-off in real time for longer than most of the NEM.

The coal exit, on WA’s own timetable #

The WA government’s own position, set out a few years back, has state-owned coal generation, the Muja and Collie power stations, winding down this decade, with the last units expected off well before 2030. That timeline has shifted before and I’d treat any specific year with some caution until the next Statement of Opportunities confirms it, but the direction hasn’t changed: Synergy, the state-owned gentailer, isn’t planning to run coal much beyond the next few years.

What replaces it is the more interesting question. Unlike the NEM, where coal retirements are creating genuine reliability anxiety I’ve written about in is Australia closing coal faster than it can replace it, WA’s smaller system means the replacement fleet is more visible and more concentrated: utility batteries, gas peaking plant at sites like Pinjar and Kwinana, and a growing wind and solar build feeding into the same corridor Western Power already owns.

Batteries doing the heavy lifting #

Synergy’s big battery build at Collie, and the earlier Kwinana battery, are the clearest sign of where the state’s money is going. On Synergy’s own published figures the Collie battery is sized around 500 megawatts, with roughly four hours of duration behind it: call it in the order of 2,000 megawatt-hours of stored energy, not a permanent 500 MW power station. That distinction matters more in WA than almost anywhere else, because there’s no interconnector to borrow firming capacity from if the battery runs flat during a prolonged low-solar, high-demand stretch. I’ve gone through the general firming trade-offs between batteries and gas peakers in gas peakers versus big batteries: who firms the grid, and the SWIS is arguably the purest test case in the country of whether four-hour batteries are actually enough, given there’s no fallback grid to lean on.

Two state-owned utilities, one awkward split #

Western Power owns and runs the poles and wires; Synergy owns most of the generation and does the retailing. Both are state-owned, which is unusual by NEM standards where privatised gentailers like AGL and Origin dominate the conversation I’ve written about in AGL Energy: coal’s biggest landlord trying to check out early. The split between network ownership and generation ownership in WA mirrors the broader national principle covered in transmission versus distribution: who owns the poles and wires, but with the added wrinkle that both halves answer, ultimately, to the same shareholder: the WA government. That’s not necessarily a bad model. I’d argue it’s given WA more room to move on things like curtailment and battery scheduling than jurisdictions where those decisions sit across three or four separate private balance sheets. But it does mean the state carries more of the financial risk directly, rather than spreading it through a market of competing generators.

A quieter transition, still worth watching #

I rode out to Collie a few years back on a long weekend loop past Wellington Dam, mostly for the scenery, but you can’t miss the power station stacks from the road and it’s a decent prompt to think about what happens to a town built around a coal fleet once the fleet stops running. WA’s version of the coal exit gets a fraction of the national media attention that Eraring or Yallourn get, probably because the SWIS is smaller and less familiar to east-coast readers, but the underlying problem is the same one I’ve covered in NSW’s energy transition – a grid trying to replace synchronous generation with inverter-based supply on a timetable that keeps testing everyone’s patience.

The difference in WA is that there’s no interconnector to paper over a slow patch. Whatever the SWIS builds, it builds for itself. That’s either the most honest test case in the country for the next stage of the transition, or the one place where a bad year would show up fastest. Worth watching either way.

– Anjali Rao, Grid & Storage Correspondent

Photo by Raisa Milova on Unsplash