How much new dispatchable generation does Australia need?

There is no market consensus of the amount of dispatchable power needed over the next decade or how much new dispatchable power is likely to be required.

The work has either not been done by the Australian Energy Market Operator, or has not been published. Most existing academic studies are too theoretical to be of practical use but they do indicate that you don’t need much new dispatchable power until the variable renewables fraction (the percentage of wind and solar) gets above 70 per cent.

Secondly, the newly introduced Reliabilty Obligation is so constrained in its assumption as to not be much use. For instance, it doesn’t include almost certain transmission augmentations and equally won’t include any thermal closures that have not been formally announced (i.e. within the three-year window). For 10-year planning purposes, this is far too stringent.

A plan that takes into account likely thermal closures and the optimal way to replace them  – similar, perhaps to the California Integrated Resource Procurement modelling – would, in this analyst’s opinion, greatly assist the industry and stakeholders to gauge likely requirements.

A presentation on the 2019 California Plan is found at this link.

It goes well beyond AEMO’s own Integrated System Plan in dealing not just with transmission but a grid-wide low cost resource plan with official status. I’d encourage AEMO to include such a plan as a supplement to the Reliabilty Obligation.

I’d encourage anyone to look at the combination of the two models (low cost resource procurement and separate reliability model) as an example of what should be available to stakeholders.

In my opinion, something like the CPUC IRP is in line with the intent of the Finkel Report Recommendation which called for a system plan. The ISP, good as it is, is but a subset of what’s required.

Less haste, more speed

As with most of the past decade, the overwhelming issue for electricity stakeholders in Australia is what will replace the ageing coal generation fleet.

Recent news about Callide B, not-so private signals from Yallourn owners, comments by Vales Point management, significant coal supply issues at Mt Piper, and yet another confirmation that Eraring is going to close in 2032, mean it’s worth looking again at the replacement task.

Even Loy Yang B management stated on a recent Energy Insiders podcast that they couldn’t guarantee the future of LYB beyond 2030. Of the above list, only Liddell is factored into the reliability obligation.

However, we would strictly see the next decade timetable as follows:

Only slightly less attention grabbing than the 20 per cent is the NSW situation. We assume for this purpose Mt Piper will somehow magically find new coal.

NSW is my state, and I don’t really like bashing it, so just let me say that I think NSW is doing next to nothing in the face of this evidence. Or what they are doing is inadequate.

It is true that Snowy 2 will provide 2000MW of power, but it won’t provide the equivalent amount of energy. It will, in fact, be a net consumer of energy.

By common consent (1) there is next-to-zero spare transmission capacity in NSW; (2) It’s the hardest state to get a new solar or wind project done; (3) there is no state incentive for new generation, i.e. no renewable target; (4) The State is a net importer of energy and the transmission lines regularly reach capacity, but expanding them is a 7 year process; (5) Big generators such as AGL and ORG that might invest are on a capital strike because the federal govt keeps kicking them in the guts.

Indicative of just how far off the pace NSW is, John Barilaro, Leader of the NSW Nationals, the governing junior party, wants to build a nuclear plant. Essentially this indicates the leadership is still in denial.

Again, as a NSW resident and loyal citizen and as an optimist at heart I can fortunately still subscribe to “where there’s a will there’s a way.”

Discussion – how much dispatchable power is needed?

ITK considers there is overwhelming evidence that the lowest cost fuel for new energy is wind and solar and that the solar can be at utility scale or distributed. Even factoring in the transmission costs the low cost position of VRE is beyond doubt.

Equally, it’s clear that some quantity of dispatchable or firming power is required.

However, there is very little accepted industry consensus on is how much dispatchable power is required. There are various studies, based on a variety of approaches, well discussed at Paul Graham, CSIRO LCOE study. One useful graph, and it’s from 2017, so already out of date, is from the CSIRO Low-Emissions-Technology-Roadmap  page 47.

But this is far from a year-by-year model that takes into account actual coal retirements and actual transmission constraints, and accounts for all the other dispatchable options besides batteries.

The modelling methodology was conservative, in that it took the worst weekly wind and solar resource availability over nine years. But there were other simplifying assumptions (zero ramp rate). However, if we assume the chart is accurate it shows that until VRE gets to 50 per cent, from its current level of 18 per cent, virtually nothing is needed.

As an important aside, that same report noted:

“Building additional renewable generation capacity beyond the amount at which renewable generation must be sometimes curtailed, rather than installing additional battery storage, allows for a lower system cost. The amount of additional renewable generation capacity required to minimise total system costs results in effective capacity factors for wind and large-scale solar PV decreasing by 83% and 62% of their average values at low penetration respectively”

This is the essence of the planning problem for Australia. How do solar and wind operators get paid for lowering system costs when the optimal average capacity factor is only a fraction of individual unit capacity factors?

But it may be that other firming technologies or dispatchable generation technologies result in lower “spillage”.  Ask yourself which “expert speaker” at the AFR Summit actually understands the issue and whether the CSIRO is correct or not. I might say that the system-wide modelling in California, which could serve as a good example for Australia, also indicates the lower system cost of having excess solar.

In addition to the CSIRO and other academic studies, there are various AEMO pieces of work from the Statement of Opportunities, medium term adequacy report, and now the Reliablity trigger.

Of those the things the market has to pay most attention to is the brand new Reliability Obligation, which came into force in July 2019. A key component of the RRO is the determination of whether there are forecast reliability gaps in the future.

The new rules require AEMO to include a ‘Reliability Forecast’ in its annual Electricity Statement of Opportunities (ESOO) setting out the reliability forecasts for a 10 year period (the last five years of which are only required to be indicative in nature).

AEMO is in the process of changing (naturally) the reliability standard of 0.0002% unserved energy. The changed standard will effectively give greater weight to tail events. In the 2019 ESOO there was no material breach of the standard as measured by AEMO, even if in NSW, that was only because Liddell now has only 1 unit closing in 2022. This is old news, but the point is there is no forecast reliability gap.

Still, my complaint would be that the ESOO doesn’t contain, within its 10-year forecast, an explicit estimate of the amount of dispatchable power that’s actually required.

Also, and again as expected, the AEMO standard is conservative. For instance, the 2019 ESOO doesn’t allow for any of the transmission augmentation currently in planning, yet almost certain to go ahead. On the other hand, nor does it allow for closures other than Liddell.

Replacing Vales Point will take more than 3 years

To summarise, the system is more or less OK according to AEMO. There is no official or consensus estimate of how much new dispatchable power will be needed over the next decade. Snowy 2 will be effective at reducing outage risk when Humelink is built, but that won’t be until the mid 2020s. Snowy could today get rid of the black-out risk in Victoria, but there isn’t enough transmission. How bad is that?

AEMO’s reliability obligation is, by legal necessity, almost certainly unrealistic about the medium term.

If Vales Point closure is announced in 2026, and three years later Eraring closure is formally announced, what’s the plan? Is a plan needed? What about if Yallourn closes because it’s unprofitable?

Can we rely on the energy market which doesn’t allow scarcity pricing, and the Reliabiltiy Obligation to produce new investment within three years? I’d argue that in some cases, even the approval for a gas station takes a year or more.

As well as that, at the moment there is every indication that NSW transmission will be clogged up right into the mid 2020s. Visible progress on developing REZs is zero. Perhaps something is going on beneath the surface.