About 2 months ago, we ticked over only two years remaining until the slated closure of Yallourn Power Station in Victoria

… which we showed here on 15th August was still slated for 1st July 2028.

Understandably* there’s been a bit of discussion about this looming closure

* I say ‘understandably’ for a few different reasons, including:

(a)  its size, and the diminishing number of coal units remaining (especially just in VIC);

(b)  the time that has elapsed since the prior domino to fall at prior domino to fall (at Liddell);

(c)  Memories of the disruption caused by the sudden closure of Hazelwood in VIC

(d) and also due to:

i.   the marked difference in investability in VRE and storage

ii.  and heightened sense of challenge with 82% by 2030

… since those earlier dominoes.

(e)  plus more…

Questions in Mainstream Media

On Friday 7th August 2026, Ryan Cropp and Angela MacDonald-Smith wrote ‘‘Where will the energy come from?’: Fresh doubts over Victorian power’ in the AFR

… and that seemed to trigger some follow-on coverage there.

Also on Friday 7th August 2026 in the Australian, Perry Williams wrote ‘Victoria and South Australia face electricity shortage risk; NSW predicted to extend life of coal’

… and similarly there has been follow-on coverage there as well.

There’s also been other coverage in other media outlets.

Stemming from various pieces of analysis

Those articles above referenced some analysis and reporting from various parties.

Analysis/Commentary by Us

For us here at Global-Roam Pty Ltd, the above mainstream media coverage drew from some articles on WattClarity®, amongst other sources:

1)  Covering two sequential weekly updates of AEMO’s MT PASA modelling results:

(a)  On Wednesday 5th August, we wrote ‘AEMO’s MT PASA run flags shortfalls forecast for VIC and SA (but the devil is in the details)’, with reference to MN144738 on Tuesday 4th August (and results on Monday 3rd August)

…hence being published as latest results at the time of those two media articles above.

(b)  Which was followed on Tuesday 11th August by ‘A week later, AEMO MT PASA run shows only shortfalls in South Australia’.

2)  Also on Wednesday 5th August, we wrote the article ‘‘What’s Yallourn ever done for us?’ Part 1’, within which we asked two questions (that resonated) …

Q1 was ‘where is the Energy going to come from (in just 2 years) to replace all that Energy that Yallourn has been producing this winter?’

… more on this, below

Q2  was ‘where the Capacity is going to come from as well’?

These questions were referenced in the AFR article.

By Rystad

In his article in the Australian, Perry Williams referenced questions Rystad has asked in a client note – more with respect to NSW than VIC in this paragraph:

‘“Given the challenges of getting wind farms to financial close, it is unlikely the pace of wind installations will materialise,” Rystad said in a client note. “Thus the more likely outcome over the next five-to-ten years will be an extension to the coal plants, particularly in NSW, as the volume of wind generation is simply not being built to replace the retiring volumes.”’

In other words, the Rystad comments more focused on Eraring than Yallourn.

Others

In these two articles, and others that followed, there were many others whose views were referenced in one of more articles.

What’s in the ESOO?

Of course it is worth noting that the 2026 ESOO was released 7 days ago … so on Saturday we summarised in ‘What’s noted in the 2026 ESOO, about the looming closure of Yallourn Power Station?’.

A closer look at July 2026

In this earlier article, we zoom in from a 10-year trend view to focus just on calendar 2026 up until 5th August 2026 (the day at which the article was published), and July 2026 particular stands out as a period with high utilisation of Yallourn, hence a useful time range to do some ‘back of the envelope’ numbers.

Repeating the two questions we posed:

Q1)  ‘where is the Energy going to come from (in just 2 years) to replace all that Energy that Yallourn  … [given that] in July 2026, the station produced roughly 959,000 MWh on an ‘as generated’ basis … so roughly 863,100MWh delivered onto the transmission grid (after ‘auxiliaries’ used onsite).’?

Q2) ‘where the Capacity is going to come from … [given that] through July 2026, the output range from Yallourn station as a whole was:

i.  as low as 928MW (just at the start of the month, we see above … but commonly no lower than ~1,200MW for much of the month)

ii.  as high as 1,423MW

Essentially the answer to this question will probably lie in:

1)  a combination of both:

(a)  increased contributions from existing plant,

(b)  and also contributions from new.

2)  but also some uncertainty about:

(a)  whether these contributions will get all the way there

(b)  and what the spot price will be as a result.

It’s almost 4 weeks since those two questions were posed in that ‘‘What’s Yallourn ever done for us?’ Part 1’ article.

1)  In the intervening weeks I’ve tried to squeeze an hour here-and-there to explore this thread;

2)  But am heading away tomorrow for a while (where thoughts of Yallourn might be further away);

3)  So posting this now, in anticipation that this might also attract great comments from some readers!

So, without further ado, and emphasising this is an incomplete answer, here’s as far as we’ve managed to date …

Side 1/2 = Back-of-the-envelope calculated possible contributions from others (existing)

Two readers provided comments on my earlier article that speak to the possible contributions to the gap in July 2028 from existing projects, should that month work out to be otherwise identical in every other way (minus Yallourn):

1)  In his comment here, AndrewJ made the comment:

‘We’ve got 2.7GW of gas that ran at 6% capacity factor in July with a daily peak of 28% on July 8 so IMO plenty of capacity for Yallourn to shut, likely with a bit of a price spike but not as bad as 2016.

[SNIP]’

2)  In his comment here, Allan O’Neil noted:

‘[SNIP] … power systems inevitably have substantial “latent energy capability”: flexibility to produce more or less energy to meet varying net demand requirements. So looking at an individual plant’s actual levels of production over a period and assuming this could only be replaced by new supply is completely wrong.

Looked at another way, in 2025/26 there was about 4.4 TWh of available energy not produced by the Loy Yang A & B stations (which ran at 80% utilisation of their available capacity after outages), and 3 TWh of offloaded or curtailed large-scale wind and solar energy which was not supplied. Not all of this >7 TWh of “latent” energy would have been available at the right times to “replace” Yallourn, but a large chunk of it would have. And that’s before thinking about net exports to other regions, offloaded / curtailed wind and solar in those regions, or energy available from gas plant.’

In order to explore this side of the story (i.e. supply from existing units … the ‘latent energy capability’ that Allan wrote about) I extracted dispatch interval data for the 8,928 dispatch intervals through July 2026, in order to explore…

A very simplistic view

Adding everything together we get a view like this, which seems to suggest that Yallourn could have been removed through July 2026 with no ill effects … but let’s be very clear that this is an enormously overly-simplistic high-level view – for reasons you’ll see below:

In this case, over the month of July:

1)  This very, very (overly) simplistic method can be clearly seen to significantly overstate the latent energy capability … we’ll break it down below.

2)  In this case we see:

(a)  that Yallourn produced 958,572MWh as generated, based on Targets

i.  which we should really reduce by perhaps 10% because of auxiliary energy use

ii.  but is sufficient for our ‘back of the envelope’ comparison here.

(b)  The ‘every other available existing option supplying into VIC’ bucket suggests a maximum of 4,260,608MWh latent energy (or a 444% coverage).

But read on for why that’s grossly over-optimistic…

Supply-side units (Generators, and Battery Injections)

Let’s start with ‘spare availability’ for each individual Scheduled and Semi-Scheduled units in Victoria (i.e. calculated as ‘Availability – Target’ ).  There’s a series of time-series trends below that help illustrate this ‘latent energy capability’.

We’ll use the Nelson Review categorisation in terms of Bulk Energy (a.k.a. VRE), Shaping and Firming.

‘Bulk Energy’

Remember that we coined the term ‘Anytime/Anywhere Energy’ to describe this in the GRC2018 and GenInsights21, in (both) of which we wrote about ‘the Schism’.  Focused firstly on Large Solar in Victoria, we see the following:

In this chart we can clearly see that there was very little curtailment of Large Solar in Victoria through July 2026 – regardless of whether it was due to Network Curtailment or Economic Curtailment.

Compared to the Yallourn Production through July (958,572MWh as generated):

1)  Curtailed Solar was just 5,843MWh

2)  Or 1% of what Yallourn produced … so not much additional there to contribute, across the whole fleet.

The numbers are larger for Wind, albeit still not stellar (from this perspective):

With this trend:

1)  We can clearly see:

(a)  three periods during the month when both occurred simultaneously:

i.  Yallourn output dropped; and

ii.  Curtailment of Wind climbed to some sizeable numbers (sometimes more than Yallourn instantaneous production)

… which will have been due to the same factors (oversupply and low prices)

(b)  But, in between those three periods were periods spanning many days with very little curtailment.

2)  In bulk energy terms through the month:

(a)  Compared to the Yallourn Production through July (958,572MWh as generated):

(b)  Curtailed Wind was just 164,022MWh, or 17% of what Yallourn produced.

(c)  To me this provides a real world, real life example of the depths of storage required to run a high renewables grid.

Shaping

In this category we have the results for batteries, but a strong caveat for every reader that these results are wildly unrealistic, in terms of aggregate battery capability, given the shallowness of the storage:

With this trend:

1)  If we were to make two massive simplifying assumptions:

(a)  assume infinite depth of storage in these batteries; and

(b)  some (otherwise curtailed) energy source to charge them with…

2)  Then we could do the comparison:

(a)  Compared to the Yallourn Production through July (958,572MWh as generated):

(b)  Unused discharge capability was 1,018,581MWh, or 106% of what Yallourn produced.

3)  For the analysts amongst our readership, understand that the ‘BESS Spare’ calculation of Availability – Target includes many instances of negative Target (i.e. when BESS were charging in aggregate).  So:

(a)  Whilst the numbers might appear extra large as a result

(b)  Actually relying on them (i.e. assuming discharge) would mean even less energy in state of charge.

Firming

Let’s walk through the types of firming currently available…

The comparison to sibling coal-fired station Loy Yang A was as above:

1)  Doing the comparison:

(a)  Compared to the Yallourn Production through July (958,572MWh as generated):

(b)  Unused discharge capability was 177,982MWh, or 19% of what Yallourn produced.

2)  But there is also some caveat here relating to events of Wednesday 8th July 2026 (which would have to be explored in a follow-on article).

Here’s sibling Loy Yang B:

1)  Doing the comparison:

(a)  Compared to the Yallourn Production through July (958,572MWh as generated):

(b)  Unused discharge capability was 64,618MWh, or 7% of what Yallourn produced.

2)  But again the caveat here relating to events of Wednesday 8th July 2026 (which would have to be explored in a follow-on article).

Here’s the trend for hydro:

1)  Doing the comparison:

(a)  Compared to the Yallourn Production through July (958,572MWh as generated):

(b)  Unused discharge capability was 937,040MWh, or 98% of what Yallourn produced.

2)  But again the caveat here that this is wildly over-optimistic, given the energy limits for these units.

So we’re left with gas-fired units, as options for supply in Victoria if we ‘magic-ed away’ Yallourn’s contributions in July 2026:

1)  Doing the comparison:

(a)  Compared to the Yallourn Production through July (958,572MWh as generated):

(b)  Unused discharge capability was 1,142,124MWh, or 119% of what Yallourn produced.

(c)  Which was essentially the point AndrewJ made in his comment here

2)  But there might be some less obvious caveats here:

(a)  in terms of the deliverability of what would be a massive increase in gas fired generation over the month

… something Paul Simshauser and Joel Gilmore have looked at, such as in their paper ‘Solving for Y’ from 2024.

(b)  and the events of Wednesday 8th July 2026 are also illustrative.

Spare (and very optimistic/simplistic!) capacity to draw more from Neighbouring regions

To complete the picture, in terms of what existing sources might have stepped up to fill the ‘sans Yallourn’ gap in July 2026, we wonder what might have been able to be imported from elsewhere.

Now, to do this analysis:

1)  A proper accounting would mean doing much as the above for individual sources in the neighbouring regions and comparing that to interconnector limitations to see what was actually deliverable;

2)  But I thought it simpler to start with just the interconnector limitations (i.e. taking no consideration to whether that neighbouring region actually had spare capacity to share), so being wildly over-optimistic:.  This looks at:

(a)  spare ‘Import’ capability :

i.  over V-SA (which included PEC Stage 1 at the time)

ii.  over Murraylink

iii. and VIC1-NSW1

(b)  and ‘Export’ capability from TAS to VIC over Basslink (given the numbering convention of interconnectors)

So here we go, starting with Basslink (now corrected … thanks to Tom for pointing out that double-up):

In other words, even if Tasmania had an infinite amount of spare energy, in the extreme:

1)  It could only supply another 63,933MWh to VIC

2)  Or 7% of what Yallourn produced over the month.

Moving onto the smaller of the links to SA (Murraylink) we see the following:

In other words, even if SA had an infinite amount of spare energy, in the extreme:

1)  It could only supply another 35,267 MWh to VIC

2)  Or 4 % of what Yallourn produced over the month.

Moving onto the larger of the links to SA (V-SA, including PEC stage 1 at this point) we see the following:

In other words, even if SA had an infinite amount of spare energy, in the extreme:

1)  It could only supply another 233,712 MWh to VIC

2)  Or 24 % of what Yallourn produced over the month.

Moving onto the VIC1-NSW1 interconnector, we see the following:

In other words, even if SA had an infinite amount of spare energy, in the extreme:

1)  It could only supply another 417,427 MWh to VIC

2)  Or 44 % of what Yallourn produced over the month.

Returning to that very simplistic view

So, with the above context in mind, we take another look at the seemingly large surplus of ‘latent energy’ capability:

In this case, over the month of July:

1)  Repeating the high-level figures:

(a)  that Yallourn produced 958,572MWh as generated, based on Targets

(b)  The ‘every other available existing option’ bucket suggests a maximum of 4,260,608MWh latent energy (or a 444% coverage).

2)  But this 444% capability breaks down to:

(a)  1% Large Solar and 17% Wind … which would have been useful in this case, if it could be stored

i.  something we did not investigate in this ‘back of the envelope

ii.  though that fact that it was spilled does seem to suggest it could not be stored, with the existing asset fleet.

(b)  An apparent 106% in battery capability

i.  but this should be discounted towards close to zero from an Additional Energy because of very limited depth of storage

ii.  remembering that batteries currently run very close to dry each daily cycle.

(c)  In terms of firming:

i.  26% from the 6 x Loy Yang units (notwithstanding that these numbers don’t work for Wednesday 8th July 2026)

ii.  98% for Hydro (a gross over-simplification of what’s actually possible)

iii.  119% for Gas (with some questions about deliverability of gas for a sustained period in winter, given network size and competing winter usage)

(d)  Not forgetting, but heavily discounting an absolute maximum additional contribution of imports:

i.  7% from TAS

ii.  28% from SA (both interconnectors together)

iii.  44% from NSW

… these percentages needing to be very significantly discounted due to real world implications.

So from the above readers will understand that the gas units would have been running very heavily through this period:

1)  with spot prices much higher, as a result

2)  and some questions about the deliverability of this hypothetical outcome for all dispatch intervals.

Side 2/2 = Back-of-the-envelope calculated possible contributions from others (under development)

One  readers provided comments on my earlier article that speak to the possible contributions to the gap in July 2028 from new projects, should that month work out to be otherwise identical in every other way (minus Yallourn):

… in this comment here Tytowatt noted:

‘There is also rooftop solar, which has YoY growth of +30GWh/month in winter in VIC. So that “only” leaves 800GWh to replace in 2028. This is obviously variable as both May and June recorded VIC’s first ever YoY declines in rooftop solar (July is back on trend at +33GWh/month YoY).

VIC grid batteries will be at 10GWh by mid-2028, which gives ~500MW overnight average discharge. So “only” 700-900MW of capacity to replace. That’s assuming they can charge each day: +30GWh/month is +60GWh/month by July 2028 = +2GWh/day of extra rooftop solar. So “only” another 8GWh/day of new July generation needed to fill those batteries.

I expect Yallourn will be subsidised to stay open for a few more years as that will be cheaper than running gas. Perhaps operating seasonally like Playford B. Marinus is maybe 2030, Snowy 2 maybe 2029. Some solar hybrids in the North of the state seem like the best option.

Corop (230MW/704MWh) seems like the most likely to happen – might take OX2 a few months to sort financing out after taking over the project in July. Barwon (250MW/500MWh) and Elaine (125MW/250MWh) would also help, but Elgin so far hasn’t closed any of its projects. Only other one I know of that could be running around 2028 is Derby (95MW/210MWh). Ah, there’s Meadow Ck (332MW/1,000MWh) that got approvals in June. Cooba solar farm has approvals, but not much news in over a year from that one. Hazelwood North not expected to start until 2028 and operations in 2031, so too far out to consider.’

As noted here in this recap, on p73/124 of the 2026 ESOO there was the following passage:

‘Maintaining reliability to 2030

Despite the volume of retirements of large coal generators in the near term, expected USE remains within the reliability standard immediately following these closures. There is an additional 2.3 GW of anticipated and committed generation expected to be operational prior to the closure of Yallourn W (1.6 GW) in Victoria, 6.3 GW in New South Wales prior to the closure of Eraring (2.9 GW), and a further 3.8 GW in Queensland prior to the closure of Gladstone (1.7 GW). Much of this is VRE generation that will offset the loss of energy generation from the coal generators that have supplied much of the NEM’s electricity over their operating lives. With these new generation resources, supplemented by storage solutions, reliability gaps do not emerge within the T-3 horizon of the RRO (2029-30). ’

So let’s use the above as inspiration for this consideration…

Three bundles to consider

Let’s clarify up-front that there are three bundles of ‘new’ here to consider:

Bundle #1)  New projects in the Consumer Energy Resource Space

Given the rate of growth recently:

1)   in rooftop PV (recently boosted with the expansion of SRES from 100kW to 1MW) and also

2)   in residential batteries (boosted by Bowen’s bonanza), this needs to be considered.

That’s something perhaps for later…

Bundle #2)  Transmission Developments

If there were network developments that might assist in increasing the transfer capacity into Victoria by July 2028, these would be worth noting here….

1)  Without going into details, my vague understanding is that the two that would potentially make most difference (Marinus Link and VNI West) to import capability are not expected until some years after winter 2028.

2)  But, if I have missed something, I’m sure our readership will let me know …

Bundle #3)  New projects in the Large Scale Space

This is where the AEMO’s note about the ‘2.3GW of anticipated and committed generation … much of this is VRE’ comes in … and what we will start to explore below:

AEMO’s generation information page

From the AEMO’s Generation information subsite I have downloaded the latest version of the ‘Generation Information’ spreadsheet:

1)  This one was dated 31st July 2026;

2)  Note that this version will probably not have been the version AEMO used as input to its ESOO modelling:

(a)  it is likely that the AEMO used an earlier version of this development schedule (given the time to run the model, analyse the results and write the document, etc).

(b)  So the 2.3GW number might (or might not quite) match what I am able to pull out of the (assumed to be newer) version of the spreadsheet … but I figured it’s best to use the most up-to-date numbers.

Taking the spreadsheet and zooming in to projects that the spreadsheet expects will be contributing in winter 2028 (i.e. the time of Yallourn closure) yields the following list:

Projects marked as ‘In Commissioning’

I’ve again split them into three columns, to match the the Nelson Review categorisation in terms of Bulk Energy (a.k.a. VRE), Shaping and Firming:

So that means:

1)  400MW of Shaping, in the form of 2 x batteries

2)  962MW of Bulk Energy across 5 units (405MW solar and 557MW wind)

Projects marked as ‘Committed’

I’ve split them into three columns, to match the the Nelson Review categorisation in terms of Bulk Energy (a.k.a. VRE), Shaping and Firming:

So that means:

1)  nn MW of Shaping, in the form of 8 x batteries

2)  nn MW of Bulk Energy across 2 units

Projects marked as ‘Committed*’

I did a quick scan of the 2026 ESOO and did not easily see what AEMO means by the asterisk in the ‘Committed*’ status – but am assuming it means not-quite-as-firm as ‘Committed’ (without the asterisk) but better than ‘Anticipated’.

These two projects are here:

So that means:

1)  250 MW of Shaping

2)  ?? MW of Bulk Energy

Projects marked as ‘Anticipated’

Remember that (as documented in the article ‘*Why* the ‘switch-a-roo with the ESOO’?’) a change was made for the ESOO 2023 onwards such that the AEMO also considered projects marked as ‘Anticipated’ in the ESOO modelling, albeit with a 6 month delay to anticipated commercial operations compared with that

So that means:

1)  1,590 MW of Shaping (of those with sizes specified)

2)  969 MW of Bulk Energy (of those with sizes specified)

So understanding what contributions these projects might have produced in a hypothetical ‘July 2028 is just like July 2026 except for Yallourn’ scenario – that’s something for a follow-on article.

Zooming further inwards, within July 2026

Of course we can extend this ‘back of the envelope’ analysis further, to zoom into a more focused time-range (say Wednesday 8th July 2026) to perform the same analysis but to understand what might have happened:

1)  Without Yallourn

2)  But with:

(a)  Increased contributions where existing units could contribute; and also

(b)  New contributions from those which will be there in July 2028.

But that’s fuel for another article later (pardon the pun).